an experimental model for traumatic axonal injury … · an experimental model for traumatic axonal...

AN EXPERIMENTAL MODEL FOR TRAUMATIC

AXONAL INJURY BASED ON CYTOSKELETAL

EVOLUTION

by

Adam John Fournier

A dissertation submitted to The Johns Hopkins University in conformity with the

requirements for the degree of Doctor of Philosophy

Baltimore, Maryland

May, 2014

© Adam John Fournier 2014

All rights reserved

ii

Abstract

Traumatic brain injury (TBI) and spinal cord injury (SCI) are debilitating causes of

traumatic death and disability to millions of people worldwide. These injuries occur from

damage to the brain and spinal cord resulting from external mechanical stimuli, including

rapid linear/rotational acceleration and/or deceleration, blast waves, crush, impact, or

penetration by a projectile. A primary pathology of TBI and SCI is traumatic axonal

injury (TAI) where rapidly applied loads trigger a progressive series of changes in the

cytoskeletal network that provides neural cells with structure and stability. These

changes gradually evolve from cytoskeletal alterations to a delayed axonal disconnection,

a process potentially amenable for therapeutic intervention.

The goal of this work is to implement experimental models for traumatic axonal

injury that provide quantitative measures for assessing changes in neurological tissues

and to connect these across multiple length scales. To better understand TAI, we have

developed a new experimental platform to apply controlled loads on isolated CNS axons.

We apply focal compression to neural axons where the applied load is predicted using a

ABSTRACT

iii

validated finite element model of the system. The experimental and finite element

models have led to the development of threshold criteria, governing the cellular response

of the axons to the applied load, as continued growth, degeneration (TAI), or regrowth.

An approach to assess the temporal evolution of the cytoskeleton during the TAI

response of the cell was developed using confocal microscopy and transmission electron

microscopy. The ability to visualize the live cell in situ and in-vitro response was

accomplished through confocal microscopy where fluorescently tagged microtubules and

neurofilaments were continuously imaged prior to, during, and immediately following

focal compression. Comparisons between unloaded and loaded live cells demonstrate

both spatial and temporal changes for cytoskeletal populations within the imaged volume.

Transmission electron microscopy connected the changes observed through confocal

imaging with alterations in the ultrastructural composition of microtubules and

neurofilaments within neural axons. These metrics provide a pathway for connecting

changes in cytoskeletal spatial distributions to previously observed changes in measured

intensity using confocal microscopy with the same loading platform in situ and in vitro,

and may be critical in understanding mechanical failure and degeneration of the

cytoskeletal system for neural axons undergoing TAI.

Our experimental framework can be applied to developing new connections with

existing analytical and computational models for predicting TBI and SCI at smaller

length scales. This could manifest itself in the form of new standards and protocols for

ABSTRACT

iv

protection against TAI, and for improvement of protective materials and restraint

systems.

Primary Reader: Professor K.T. Ramesh

Secondary Readers: Professor Arun Venkatesan

Professor Sean Sun

v

Acknowledgements

This thesis would not have been possible without the backing and assistance of

numerous individuals.

I would like to genuinely thank my advisor, Professor K.T. Ramesh, for being a

source of encouragement, direction, and for pressing me to expand my ways of thinking.

I would also like to express my thanks to my thesis readers, Prof. Arun Venkatesan and

Prof. Sean Sun, for taking the time to serve on my thesis committee and providing

feedback on my work. I am also grateful to Prof. Venkatesan for providing me with

access to his lab group, equipment, and animal cell line for much of the work I completed

and for providing me with sound advice on a number of occasions.

I would like to thank Dr. Suneil Hosmane for introducing me to the work he had done

on the axon injury microcompression platform and sharing his insights on fluorescent

labeling approaches for cytoskeletal constituents. I would also like to thank Dr. Rika

Wright-Carlsen for introducing me to Dr. Hosmane and for the direction and guidance

she provided during my first years at Hopkins.

ACKNOWLEDGEMENTS

vi

To everyone at the Johns Hopkins Medical Institutes microscope facility, thank you

for your patience with my questions and for providing valuable guidance on confocal and

TEM equipment. To the faculty at JHU and the number of colleagues who have helped

me throughout my time here, I would like to thank you. To the amazing administrative

staff in the Mechanical Engineering Department at JHU, I would like to express my

deepest thanks for saving me over the past few years.

To all my labmates, both previous and present, you have made my time as a graduate

student an amazing experience. Thank you for sharing your insights, knowledge, and

time as we shared our successes and worked through our troubles over many late nights

and weekends. Thank you also for reminding me to smile because crazy people like

company.

To my work, the U.S. Army Aberdeen Test Center, thank you for supporting my

academic pursuits and for providing me with an opportunity to better serve the Army and

our Country.

To my family and friends, I appreciate everyone’s backing over my time at JHU. To

my parents, Donald and Helen Fournier, thank you for your continued support and

encouragement to go further. To my sister, Meghan Fournier, thank you for being there

and for keeping in touch over the years.

To my wife, Joahnna Fournier, thank you for your support, encouragement, and love

during my time at JHU. I’m sorry I took so long, but I look forward to spending time

with you and Clara in the years to come.

vii

Dedication

This thesis is dedicated to my wife Joahnna, our daughter Clara, our families, and

Cinco.

viii

Contents

Abstract ii

Acknowledgements v

List of Tables xiv

List of Figures xv

1 Introduction 1

1.1 Significance of Problem.............................................................................. 1

1.2 Objectives and Structure of Thesis ............................................................. 3

1.3 Contributions to Research Field ................................................................. 6

2 Background 9

2.1 Traumatic Brain Injury and Spinal Cord Injury.......................................... 9

2.2 Central Nervous System Anatomy and Structure ....................................... 13

2.2.1 Central Nervous System ........................................................ 13

CONTENTS

ix

2.2.2 Neural Cells and Tissues ........................................................ 15

2.2.3 Cytoskeletal System ............................................................... 17

2.3 Definition of Traumatic Axonal Injury ....................................................... 21

2.3.1 Experimental Techniques to Investigate TAI ........................ 22

2.3.1.1 Biomechanics of TAI ............................................................. 22

2.3.1.2 Pathobiology of TAI .............................................................. 27

2.4 Summary of Current TAI Research ............................................................ 35

3 Axon Injury Microcompression Platform 37

3.1 Introduction ................................................................................................. 37

3.2 AIM Platform Description .......................................................................... 38

3.3 Axon Integration into AIM Platform .......................................................... 41

3.4 Finite Element Model of AIM Platform ..................................................... 42

3.4.1 Idealization of AIM Finite Element Model ........................... 42

3.4.2 PDMS Material Parameters & Constitutive Equations .......... 43

3.4.3 Boundary Conditions for Finite Element Model ................... 45

3.5 Platform Optimization ................................................................................ 45

3.5.1 Design Characterization ......................................................... 45

3.5.2 Applied Load ......................................................................... 47

3.6 Validation of AIM Finite Element Model .................................................. 48

3.6.1 Validation of Input Pressure-Displacement

Relationship ........................................................................... 49

3.6.2 Validation of Contact Pressure Predictions ........................... 50

CONTENTS

x

3.7 Thresholds for TAI Response ..................................................................... 53

3.7.1 Classification of Axonal Response ........................................ 53

3.7.2 Contact Pressure and Axonal Response ................................. 53

3.7.3 Confirmation of Regrowth Response..................................... 56

4 Visualization of Cytoskeletal Deformation 59

4.1 Introduction ................................................................................................. 59

4.2 In-Vitro and In Situ Cytoskeletal Deformation ........................................... 60

4.2.1 Cell Isolation, Cytoskeleton and Membrane

Labeling ................................................................................. 60

4.2.2 Immunohistochemical Labeling ............................................. 63

4.2.3 Experimental Procedures ....................................................... 64

4.3 Visualizing Structure .................................................................................. 66

4.3.1 Confocal Imaging Setup ........................................................ 66

4.3.2 Image Acquisition .................................................................. 68

4.3.3 Post-Processing of Image Data .............................................. 69

4.4 Cytoskeletal Population Response to TAI .................................................. 71

4.4.1 Changes in Axonal Cytoskeleton with Mechanical

Loading .................................................................................. 72

4.4.2 Colocation & Effects of Fluorescent Labeling ...................... 73

4.4.3 Evolution of Spatial Distribution of Microtubules &

Neurofilaments ....................................................................... 75

4.4.4 Microtubules & Neurofilament Response to Load ................ 79

CONTENTS

xi

4.4.4.1 Temporal Evolution of Microtubules & Neurofilaments....... 79

4.4.5 Limitations of Study .............................................................. 83

4.5 Summary of Results .................................................................................... 84

5 TEM Observations of Cytoskeletal Evolution in CNS Axons 86

5.1 Introduction ................................................................................................. 86

5.2 Experimental Approach .............................................................................. 87

5.2.1 Cell Culture and Isolation ...................................................... 87

5.2.2 Experimental Protocols .......................................................... 88

5.2.3 Fixation, Labelling, and Embedding for TEM....................... 89

5.2.4 Transmission Electron Microscopy ....................................... 91

5.2.4.1 Imaging Setup ........................................................................ 92

5.2.4.2 Image Acquisition .................................................................. 92

5.2.4.3 Post-Processing of Image Data .............................................. 93

5.3 Quantification of TEM Data ....................................................................... 95

5.3.1 TEM Quantification for Microtubules ................................... 96

5.3.2 TEM Quantification for Neurofilaments ............................... 97

5.3.2 Statistical Analysis of TEM Data .......................................... 98

5.4 Cytoskeletal Component Level Response to TAI....................................... 99

5.4.1 Structural Changes in Cytoskeleton ....................................... 100

5.4.1.1 Morphological Assessment of Microtubules ......................... 100

5.4.1.2 Morphological Assessment of Neurofilaments ...................... 102

5.4.2 Changes in Cytoskeletal Spatial Distribution ........................ 106

CONTENTS

xii

5.4.2.1 Quantitative Assessment of Microtubules ............................. 106

5.4.2.2 Quantitative Assessment of Neurofilaments .......................... 113

5.4.3 Connecting TEM Measures to Confocal Results ................... 119

5.4.4 Changes in Cytoskeletal Temporal Distribution .................... 119

5.4.4.1 Quantitative Comparisons with the Literature ....................... 119

5.4.4.2 Load Response of Cytoskeleton............................................. 123

5.4.4.3 Temporal Evolution of Cytoskeleton ..................................... 123

5.4.5 Numerical Approach to Cytoskeletal Metrics ........................ 124

5.4.6 Spacing Mechanism for Neurofilament Sidearms ................. 127

5.4.7 Limitations of Study .............................................................. 128

5.5 Summary of TEM Study of Cytoskeletal Evolution ................................... 129

6 Discussion and Future Directions 131

6.1 Major Contributions & Future Directions .................................................. 132

6.1.1 Axonal Injury Micro-Compression Platform

Development & Threshold Validation for TAI ..................... 132

6.1.2 In-Vitro and In Situ Visualization and Quantification

of Cytoskeletal Deformation under Load .............................. 137

6.1.3 Cytoskeleton Quantification and Temporal Evolution

Under Focal Axon Compression ............................................ 140

6.2 Clinical and TAI Research Implications ..................................................... 142

6.2.1 Relating Structural Evolution and Loss of Neural

Cognition with TAI ................................................................ 143

CONTENTS

xiii

6.2.2 Applications to Therapeutic Intervention .............................. 144

6.2.3 Prevention and Mitigation of TBI and SCI ............................ 145

6.3 Summary Suggestions for Future Work ..................................................... 146

6.4 Conclusions ................................................................................................. 149

Appendix A: PDMS Stretch-Stress Response to Uniaxial Load 151

Appendix B: MATLAB Code for Confocal Analysis 153

Appendix C: MATLAB Code for Analysis of Neurofilament Metrics 156

Bibliography 168

Vita 188

xiv

List of Tables

2.1 Mechanisms of TBI and SCI. (Adapted from [16]) .................................... 13 2.2 Experimental Techniques for Investigating TAI ........................................ 25 2.3 Experimental Markers for Detecting and Characterizing TAI (Adapted

from [65]) .................................................................................................... 30 3.1 AIM platform geometry. (Adapted from [51]) ........................................... 40 3.2 Mesh density sensitivity. The aspect ratios of elements in the

compression pad and glass substrate remain constant as mesh density is increased. (Adapted from [51]) ................................................................... 43

3.3 Coefficients for Equation 3.3 relating contact pressure, input pressure, and membrane thickness. (Adapted from [51]) .......................................... 48

4.1 Confocal imaging plane resolution by magnification ................................. 67 5.1 TEM grid, slot, and specimen identification for Control and Crushed

axons ........................................................................................................... 93 5.2 Power law fitting coefficients (Equation 5.4) for microtubule measures

of Control and Crushed axons and the 95% confidence intervals for those coefficients ........................................................................................ 108

5.3 Quantification of microtubules and neurofilaments following focal compression ................................................................................................ 113 5.4 Linear fitting coefficients (Equation 5.8) for neurofilament measures of Control and Crushed axons and the 95% confidence intervals for

those coefficients ........................................................................................ 115 A.1 Stretch-Stress data from the literature and uniaxial tests of Sylgard 184 [110] ............................................................................................................ 151

xv

List of Figures

1.1 Flow chart outlining our framework for studying the cytoskeletal response process of traumatic axonal injury. Each step presents a significant increase in the degree to which we can connect the cellular level TAI response to changes in the cytoskeleton of neural axons ........... 5

2.1 Mechanics classifications for TBI and SCI mechanisms include

penetration, inertial loading, contact, and blast. Here an individual illustrates potential combinations of inertial and contact loading mechanisms from an elevated position with forward momentum. Note the absence of a helmet or other protective equipment. [14] ...................... 11

2.2 Blast injury mechanism for TBI and SCI where the shock waves from an explosive device can result in injuries to the brain and spinal cord parenchyma. Here the focus is on the primary mechanism where blast-induced neurotrauma (primary injury), without contact, is occurring due to the blast wave itself. Secondary (penetration) and tertiary (impact) mechanisms and their associated neurotrauma from flying debris and coup-contrecoup motion are also shown. (Image adapted from [15]) ........ 11

2.3 Loading across multiple length scales, from the macroscale to the nanometer level, connects the biomechanics of TAI (traumatic axonal injury) to the pathological response that culminates in TBI and SCI. Loading is applied at the macroscale in the form of contact, penetration, and noncontact (through linear and rotational acceleration/deceleration) methods spanning a duration of tens of milliseconds to a few seconds. These loads translate to changes at multiple length scales of neural tissues, their cells, and the underlying cytoskeletal structure. At the smaller length scales, the complex applied load is understood in the context of simple mechanical loading (tension, compression, and shear). These simplified mechanical loads lead to the pathobiological response

LIST OF FIGURES

xvi

observed for TAI and culminate in TBI and SCI at larger length and time scales ................................................................................................... 12

2.4 Central nervous system (CNS) components of brain and spinal cord. At the periphery of the spinal cord the peripheral nervous system (PNS), a subdivision of the nervous system containing all nerves and ganglia outside the CNS, interact to form a communication relay between the two divisions of the nervous system. (Images adapted from [17, 18]) ...... 14

2.5 White and grey matter of the CNS. A) Transverse section of spinal cord. B) Frontal section of the brain. (Image adapted from [19]) ............. 15

2.6 Cells of the CNS. Glial cells include astrocytes, radial glial cells, oligodendrocytes, and ependymal cells in the CNS. These cells provide homeostasis, maintenance, and protection for neurons. Neurons, myelinated by oligodendrocytes, compose the white matter tracts of the CNS. Capillaries and other parts of the cerebral vasculature deliver oxygenated blood, glucose and other nutrients to the brain by the arteries and the veins carry deoxygenated blood back to the heart, removing carbon dioxide, lactic acid, and other metabolic products. (Image adapted from [21]) .......................................................................... 16

2.7 Primary parts of neuron for information flow. Dendrites receive electrical impulses and pass the information through the cell body and down the axon where it passes the signal onto another neuron or to an effector cell (muscle, gland, or organ cell capable of responding to a stimulus) [20]. (Image adapted from [22]) ................................................ 17

2.8 Cytoskeletal substructure of neuron. Microtubules and neurofilaments are two primary constituents of the neuron. A) Microtubules forms as 24nm wide hollow cylinder and function to provide transport and structural rigidity to the axon. B) Neurofilaments have a rod domain core from which sidearm extensions splay outward and function to regulate axon diameter and provide mechanical strength and stability. (Image adapted from [23]) .......................................................................... 18

2.9 Transmission electron microscopy of microtubules. Microtubules (examples shown by arrows) exist as linear rod-like hollow cylinders that function to provide intracellular transport and structural rigidity to cells. Scale bar = 50nm. (Image adapted from [24]) .................................. 19

2.10 Cryoelectron microscopic images and illustrations of microtubule growth and shrinkage. A) α-tubulin and β-tubulin are assembled as linear protofilaments. B) Depolymerization of tubulin protofilaments. Note the dissembling protofilaments are highly curved while the assembling protofilaments are very straight and fan like. Scale bars = 25nm. (Image adapted from [29]) ............................................................... 19

2.11 Electron micrographs of neurofilaments. Neurofilament project sidearm extensions from neurofilament core (inset, arrows). Scale bars = 200nm. (Image adapted from [34]) .......................................................................... 20

LIST OF FIGURES

xvii

2.12 Illustrative example of the neurofilament substructure, a major structural component of the axonal cytoskeleton. The rod domains of NF subtype proteins run parallel in the core of the NF. Whereas NF-light (white) has only a rod domain, the sidearm domains of NF-medium (red) and NF-heavy (blue) stand out from the NF core creating a physical spacing between neighboring NFs [35] ................................................................... 21

2.13 3-D Cell Shearing Device (3-D CSD) components. (a) A schematic representation of the 3-D CSD. The device can be mounted on a confocal microscope to obtain 3-D images before, during, and after mechanical deformation. A closed-loop proportional-integral-derivative controller system (PID controller) with feedback from a digital variable reluctance transducer (DVRT) governs a linear actuator, inducing motion of the cell chamber top plate (not to scale). (b) The cell chamber consists of a top plate with polyethylene filters to interface with the 3-D cell cultures. The top plate is mounted above the cell reservoirs and connected to the linear actuator to impart high rate deformation. (c) The horizontal motion of the linear actuator drives the displacement of the cell chamber top plate, inducing shear deformation in the Sylgard mold and matrix (with either cells or microbeads) (not to scale) [55] ................. 23

2.14 Dynamic mechanical stretch of isolated cortical axons. A, B) Schematic illustration of axonal stretch injury model. An axon-only region of the elastic membrane overlaps with a 2 by 15mm slit at the bottom of an airtight chamber. A controlled air pulse deflects the elastic membrane downward, thus inducing a tensile elongation exclusively to the axons. C) Phase-contrast imaging of the axon-only region (formed by a silicone stamp that creates microchannels permitting only axon outgrowth across the channels). D) Fluorescence microscopic confirmation that the neurites in the microchannels were axons demonstrated by immunoreactivity to neurofilament protein (NF, green), while immunoreactivity to microtubule-associated protein 2 (MAP2, red), a specific marker for the dendrites, was found exclusively outside the channels. Scale bar = 100μm. (Image adapted from [54]) ......................... 26

2.15 Temporal evolution of secondary axotomy and its effects across multiple length scales of neural tissue shown in Figure 2.3. Following loading, changes at the cytoskeletal level propagate up to the cell and tissue levels where they are manifested in the forms of nodal blebs and damage tissue parenchyma. (Images adapted from [20, 54]) ................................... 28

2.16 Ion channels located along the axon membrane in neurons open as a result of mechanical loading, stretch is shown here, leading to an ionic influx. A) The mechanism shown starts with mechanosensitive sodium channels activation in response to an applied strain on the axon membrane. In response to the influx of sodium, sodium/calcium exchangers reverse direction (B) and the voltage gated calcium channels

LIST OF FIGURES

xviii

are activated (C). These mechanisms, collectively, are thought to contribute to the pathological influx of calcium in the axon. (Image adapted from [68]) ...................................................................................... 31

2.17 Computer-assisted analysis of NF sidearms in control (A, C, E) and injured (B, D, F) axons. Electron micrographs were taken of uninjured (A) and injured (B) axons. Enlarged, color-enhanced axonal fields reveal the detail of control (C) and injured (D) axons. C) Wide interfilament spacing (arrows) and prominent sidearms (arrowhead) are observed in control axons. D) NF compaction and reduction of NF sidearm height are observed in injured axons. E) Further enlargement of an individual sidearm for control axon (green arrow). F) Individual sidearm (green arrow) for injured axons reveal reduced height in concert with the compaction of the adjacent NFs (arrows) which lie in close proximity. Scale bars = 500nm for A, B. Scale bars = 50nm for C, D. Scale bars = 5nm for E, F. (Image is adapted from [101]) ......................... 33

2.18 Transmission electron microscopy (TEM) of longitudinal sections for injured axons demonstrated ultrastructural alterations of the axonal microtubule lattice (arrows). Microtubules were identified as dark 24nm wide filamentous structures that traversed the main axis of an axon. (A) Microtubules appear unchanged in straight areas of the axon. B-C) In areas where nodal blebs are present, microtubules appear to lose continuity at the peaks and display conspicuous free ends that appear frayed (similar to a shorting microtubule undergoing catastrophic depolymerization) (asterisk). Scale bar = 500nm. (Image is adapted from [54]) ............................................................................................................ 34

2.19 Understanding the development and evolution of TAI mechanisms are an important step towards the advancement of preventative measures and treatment options for TBI and SCI ....................................................... 35

3.1 The AIM platform was assembled from two independently fabricated

templates. The neuronal template (A), which fluidically defined where neurons and medium resided, was constructed as follows: Beginning with a bare silicon wafer, (A Top) an initial resist layer was processed to yield two linear arrays of microchannels. (A Middle) Afterwards, a thicker resist layer was deposited to first define the compartment bases and the compression injury pad clearance height. (A Bottom) This process was repeated once again to complete the compartment height and define the compression injury pad geometry. (B) A separate controller template defined which areas of the final device were controlled (or deformed) by compressed air and was fabricated using a single resist layer. (C) A representative AIM device was filled with dye to visually show both the neuronal (blue/green) and control (red) layers. (D) The device cross-section depicts the relative dimensions of the

LIST OF FIGURES

xix

compression injury pad relative to the compartment and microchannels. Schematics A and B are not to scale. [51] .................................................. 40

3.2 The AIM platform was assembled from two independently fabricated templates. The neuronal template (A), which fluidically defined where neurons and medium resided, was constructed as follows: Beginning with a bare silicon wafer, (A Top) an initial resist layer was processed to yield two linear arrays of microchannels. (A Middle) Afterwards, a thicker resist layer was deposited to first define the compartment bases and the compression injury pad clearance height. (A Bottom) This process was repeated once again to complete the compartment height and define the compression injury pad geometry. (B) A separate controller template defined which areas of the final device were controlled (or deformed) by compressed air and was fabricated using a single resist layer. (C) A representative AIM device was filled with dye to visually show both the neuronal (blue/green) and control (red) layers. (D) The device cross-section depicts the relative dimensions of the compression injury pad relative to the compartment and microchannels. Schematics A and B are not to scale. [51] .................................................. 41

3.3 (Left) The assembled AIM device consists of a PDMS control layer, membrane, and glass slide. (Right) A subsection of the finite element model depicts variable mesh sizes for the compression pad, PDMS membrane, and glass substrate. Scale bar = 100μm ................................... 43

3.4 Mooney-Rivlin fit of experimentally obtained uniaxial data for Sylgard 184. (Adapted from [51]) ............................................................................ 44

3.5 A) AIM devices were filled with Fluorescein isothiocyanate (FITC) dye and were imaged under confocal microscopy to characterize compression pad deflection. (B, C) Representative image stacks demonstrated the near linear relationship of the normalized input to compression pad deflection (D) and close correlation to FEM modeling. For the same normalized input as the device images in (B), the corresponding FE results are shown. As membrane thickness between device layers can vary slightly between batches, FEMs were developed to quantitatively assess the relationship between input pressure, membrane thickness, and contact pressure at the glass substrate, an estimate of axonal injury. [51] .................................................................... 46

3.6 (A) A 3-D schematic of the AIM platform under pressure application and microchannel deflection. (B) The same AIM device shown in Figure 3.5 was imaged at the microchannel interface to determine the extent of microchannel deflection. Under loads greater than that required to bring the compression pad in contact with the glass substrate (> 68.5kPa), a > 3μm gap could be seen. This was sufficient to allow unperturbed axon outgrowth (diameter < 2μm) before, during, and after injury. [51] ............ 47

LIST OF FIGURES

xx

3.7 Percentage of compression pad deflection as a function of input pressure for a membrane thickness of 55μm ............................................................. 49

3.8 Symmetric half of AIM platform mounted to an acrylic base. Scale bar = 200μm ......................................................................................................... 50

3.9 Experimental setup for load cell measurements of contact pressure applied by AIM platform. (Left) Multiaxial translation stage for alignment with the compression pad of the AIM platform. (Middle) Sensing post extension from load cell. (Right) Machined 40μm wide post at the tip of the load cell sensing post extension ................................. 51

3.10 Alignment of PDMS compression pad with sensing post from load cell. Scale bars = 200μm ..................................................................................... 51

3.11 Contact force measurement comparison between FE model prediction and load cell experiments for a specified geometry across a range of pressure. The FE model appears to over predict the contact force at higher input pressures. This is attributed to out-of-plane bending caused by symmetrically cutting the AIM to allow for visual alignment of the PDMS compression pad with the load cell sensing post extension. ........... 52

3.12 Representative images of axons that continued to grow, degenerate, or regrow as a function of injury level. (A) Under mild injuries (~25kPa), axons generally continued to grow from left to right as evidenced by progressing growth cones (red triangles). (B) At medium levels of injury (~68kPa), more axons began to undergo degeneration as shown by axoplasm disruption and nodal swellings. (C) Under severe compression (~192kPa), which led to rapid transection, a fraction of axons were able to regrow. These axons were often seen retracting, stuttering, and pausing prior to reformation of the growth cone. Axons (green) were false colored to enhance contrast and allow clear visualization. Scale bar 20μm. [51] .................................................................................................. 54

3.13 Individual axons were binned into one of three following categories: continued growth, degenerating, or regrowing and separated into ranges of injuries: Mild (< 55 kPa), Medium (55–95 kPa), and Severe (> 95 kPa). Quantification of control (uninjured) axons was also done to determine baseline levels of continued growth, degeneration, and regrowth. All experimental conditions were completed in triplicate. Statistical analysis was performed by Tukey pairwise 1-way ANOVA: ** = p-value < 0.01, *** = p-value < 0.001. Error bars on graphs correspond to standard errors. [51] ............................................................. 55

3.14 A tau-labeled (microtubule marker; red) axon was subjected to severe (235kPa) compression injury and images were collected every 30mins for 8hr post injury. Immediately after injury, the distal segment of the transected axon underwent classic axonal degeneration as evident by nodal swellings, while the axon tip (white arrows) first retracted (~30mins), then began to reform a growth cone (~1hr 30mins). After

LIST OF FIGURES

xxi

reformation, the axon was able to extend past the site of injury. Dotted white lines demarcate the region of the compression pad. Scale bar 25μm. [51]. ................................................................................................. 56

3.15 Growth rates for Uninjured (Control) axons and those with a Regrowth response to focal compression. Growth rates increased by 40% for Regrowth by comparison to Control axons. *p-value < 0.05, unpaired 1-way Welch’s t-test. Error bars on graphs correspond to standard errors. (Adapted from [51]) .................................................................................... 57

4.1 The AIM testing platform isolates individual neural axons and provides

a controlled microfluidic environment for applying focal compression. A) Illustrative example of an AIM platform inside a glass well. B) Inset of AIM platform depicting microchannels connecting to injury compartment, where loading is applied through a compression pad. C) Idealization of axon loading environment and orientation within the AIM. D) Inset of single axon growth through microchannel into testing environment. E) Inset of individual axon under compression pad i) prior to, ii) during, and iii) immediately following applied load [51]. Scale bar = 10μm for D. Scale bars = 20μm for E. [113] ..................................... 62

4.2 Process flow chart of cell preparation, labeling, and incubation for confocal imaging. Cells were obtained from E17 rat pups. Dissociated neurons were electroporated and the nucleofected cells were plated in AIM platforms. For cytoskeletal labeling, pCMV-AC-GFP plasmids with a C-terminal TurboGFP, encoding a neurofilament-GFP fusion or a microtubule-GFP fusion gene with a cytomegalovirus (CMV) promoter, were used. Following an incubation period of 7-10 day, cell membranes were labeling with CMPTx red cell tracker and taken to experimentation. ......................................................................................... 63

4.3 Confocal imaging setup for focal compression experiments. A stage-cover provided a temperature controlled environment for imaging with access for the pressure control lines for regulating microfluidic compression pads of the AIM. .................................................................... 65

4.4 Confocal imaging subvolume prior to and during focal compression. A-B) Illustrative example of the subvolume of axon prior to (A) and under loading (B). Here the axon membrane is shown in red and the cytoskeletal components of microtubules and neurofilaments are shown in green and blue respectively. C-D) Volumetric data set for single axons prior to (C) and during focal compression (D). E-F) Fluorescent images take using 488nm (green) and 562nm (red) wavelengths. 562nm intensity represents axons membrane and the 488nm represents the cytoskeletal constituent (microtubules or neurofilaments) of interest. Scale bars = 2μm ......................................................................................... 67

LIST OF FIGURES

xxii

4.5 Confocal intensity subvolume selection process. Acquired Z-stacks are processed to create 3D volumetric sets of intensity data to be analyzed. A) Full range 3D confocal intensity data set taken from a single experiment. B) Subset of volumetric data segmented focusing on the volume of axon underneath the compression pad. C) Intensity thresholds are applied to remove background noise. In this example, surfaces are temporarily created to visualize the volume containing the axon membrane (D) and the neurofilaments (E). Scale bar = 10μm for A. Scale bars = 3μm for B-E ........................................................................... 68

4.6 Illustrative and confocal examples of the subvolume of axon, with labeled cytoskeletal components, under the compression pad. A) The spatial and temporal information from initial unloaded (Ω0) and under load (Ω) states are connected through the intensity, Φ, which maps the spatial description from Ω0 to Ω. Here the axon membrane is shown in red and the cytoskeletal components of microtubules and neurofilaments are shown in green and blue respectively. B) Volumetric Z-stack of initial unloaded (Ω0) and under load (Ω) states. Scale bars = 10μm .......... 70

4.7 Confocal imaging confirms transfected cytoskeletal constituents were correctly labeled using a secondary labeling process celled immunohistochemistry. A-B) For microtubules, transfected tau protein (green) and immunohistochemical labeled Beta-III tubulin (red) have similar expression patterns and, using a triaxial planar view, are observed to colocalize in the same spatial domain (yellow). C) Nontransfected, immunohistochemically labeled microtubules appear to exhibit a similar expression pattern as the transfected microtubules with immunohistochemical labeling. D-E) For neurofilaments, transfected NF-Medium (green) and immunohistochemical labeled NF-Medium (red) have the same expression pattern, and using a triaxial planar view, are observed to colocalize in the same spatial domain (yellow). F) Nontransfected immunohistochemically labeled neurofilaments appear to exhibit a similar expression pattern as transfected neurofilaments with immunohistochemical labeling. Scale bars = 2μm for A-F. [113] ............. 74

4.8 Representative examples of intensity profiles (y-axis) for microtubules plotted along the volume of axon underneath the compression pad (x-axis). The two panels represent Control and Loaded cells where a single axon is represented before loading at time t0 (blue) and while under load at time t (red). For the Control population, no load was applied at time t. A) No observable change in ΦA for Control between time t0 and time t states. B) Overall magnitude of ΦA does not change during loading, though its distribution within the subvolume appears to. [113] ................. 76

4.9 Representative examples of intensity profiles (y-axis) for neurofilaments plotted along the volume of axon underneath the compression pad (x-axis). The two panels represent Control and Loaded cells where a single

LIST OF FIGURES

xxiii

axon is represented before loading at time t0 (blue) and while under load at time t (red). For the Control population, no load was applied at time t. A) No observable change in ΦA for Control between time t0 and time t states. B) Decreased magnitude of ΦA is measured across the entire subvolume during loading. [113] ...................................................... 77

4.10 Results for One-way ANOVA with Tukey’s multiple comparison test of percent change in Φ� for Control axons and for axons at time t<1min and t=5min. The number of samples and the population group are shown along the x-axis and the corresponding percent change in Φ� is plotted along the y-axis. No significant observable percent change in Φ� was observed between Control and Loaded axons at times t<1min and t=5min for Microtubules. Loaded axons for neurofilaments exhibited a statistical significance decrease of 24% and 30% at t<1min and t=5min respectively compared to Control axons. * p<0.001[113] .......................... 78

4.11 Percent change in cytoskeletal density as a function of time comparing across multiple studies. The time between loading and quantification is plotted along the x-axis and varies from t<1min to t=72hrs. The percent change in cytoskeletal density is plotted along the y-axis. Data from our study taken during loading is shown at times t<1min and t=5min. A) Microtubule density did not exhibit a measurable change between the unloaded and loaded states during initial loading. Changes in density became apparent at the 5min, for our study, and appear to fall in line with other studies which have observed decreases in microtubule density, or the proteins associated with microtubule expression, at greater time intervals following deformation [45, 49]. B) Neurofilament density exhibits an immediate decrease of 24% in expression during loading and continues to decrease to 30% within 5min. Other studies have exhibited decreases in the magnitude of the density change as time increases, a trend we observe in the current study [49, 116]. [113] ........... 81

4.12 Aggregate data from Figures 4.10A-B comparing percent changes of cytoskeletal density over time following TAI. Neurofilament density (blue square) decreases at a faster rate than the microtubule density (red circle) [45, 49, 116]..................................................................................... 82

5.1 Transmission electron microscope used for imaging cytoskeletal

constituents of neural axons ........................................................................ 92 5.2 Focal compression of isolated primary hippocampal axons. A)

Schematic illustration of axon loading environment and orientation within the AIM. The axon only region overlaps with a 20μm thick compression pad above the testing chamber. Microfluidics are used to control the compression pad and localize loading to the area underneath the pad (blue box) (Chapter 3). B) A series of panoramic TEM images reconstructing the entire area of the axon under the compression pad at

LIST OF FIGURES

xxiv

higher magnification. C) A single TEM image for quantifying number, density, and spacing of the cytoskeletal structures. Scale bar = 500nm. [118] ............................................................................................................ 94

5.3 Neurofilament TEM images prior to (left) and after (right) background subtraction. No Au nanoparticles were found outside the axon indicating the labeling technique was successful. Following background subtraction images were exported as TIFFs for quantification ................... 95

5.4 Spacing along unit length (L) of the axon for microtubule measures at L/4, L/2, and 3L/4 ....................................................................................... 97

5.5 Axonal segmentation into thirds. Diameter measurements were made at the midpoint of each area, along the length of the axon for neurofilament quantification .............................................................................................. 98

5.6 TEM images for microtubules of (A-B) no load (Control) and (C-D) loaded (Crushed) axons. Microtubules are indicated by black arrows (B, D). Axon diameter, number of microtubules, and spacing between microtubules were measured for each image along unit axon length, L, at L/4, L/2, and 3L/4. A) In Control axons, microtubules are oriented along the principal axis of the axon. C) In Crushed axons, microtubules appear disorganized and misaligned. B,D) Inset of Control and Crushed axons showing diameter (D) and spacing (SMT) measurements for microtubules. Scale bars = 100nm. [118] ................................................... 101

5.7 Degenerative response associated with axons undergoing TAI. Nodal blebs (arrow head) of Crushed axon show mitochondria in each bleb. Inset of Crush axon bleb showing microtubule breakage, rupture, and depolymerization. Scale bars = 100nm. [118] ........................................... 102

5.8 TEM images for neurofilaments of (A-C) no load (Control) and (D-F) loaded (Crushed) axons. 6nm Au nanoparticles, outlined in yellow, were used to measure areal density and spacing between neurofilaments for each image. A-C) In Control axons, Au nanoparticles are regularly spaced along the length of the axon and across the axon diameter. D-F) In Crushed axons, Au-nanoparticles appear more heterogeneous in their distribution and spacing. B-C, E-F) Inset of Control and Crushed axons showing areal density and spacing distribution for nanoparticles. There is a greater number and areal density in Control axons (C) than in Crushed axons (F). Scale bars = 100nm. [118] ......................................... 104

5.9 TEM images for neurofilaments of (A-C) no load (Control) and (D-F) loaded (Crushed) axons. 6nm Au nanoparticles, outlined in yellow, were used to measure areal density and spacing between neurofilaments for each image. B, E) Yellow circles highlight the 6nm Au nanoparticles (black dots) used to assess quantity, density distribution, and spacing of neurofilaments using antibody labeling. The aggregation of Au nanoparticles appears towards the midline of the axon with fewer nanoparticles at the axon membrane for Crushed than Control groups. C,

LIST OF FIGURES

xxv

F) Inset showing 6nm Au nanoparticles (arrows). Scale bar = 100nm for A-F. [113] ................................................................................................... 105

5.10 Raw data plots of all cytoskeletal metrics (𝑁𝑀𝑇, 𝜌𝐿𝑀𝑇, 𝑆𝑀𝑇, 𝑁𝑁𝐹, 𝜌𝐴𝑁𝐹 , and 𝑆𝑁𝐹) as functions of axon diameter. A-F) All plots are expansions of the data summarized in Figures 5.11-5.16. 95% confidence intervals are shown (dashed lines) for each of the fit curves. A, C, E) Power law fits are used to estimate the relationship between axon diameter and microtubule cytoskeletal measures. B, D, F) Linear fits approximate the relationship between axon caliber and neurofilament metrics. Only 𝑆𝑁𝐹 for Crushed axons appear to remain constant, at approximately 70nm, following loading ........................................................................................ 110

5.11 The number of (𝑁𝑀𝑇) microtubules for no load (Control) and loaded (Crushed) axons. Axon diameter bins are given along the X-axis and error bars are standard error mean for all plots. A strong dependency on axon caliber is observed the number of microtubules in Control and Crushed axons. 𝑁𝑀𝑇 is significantly lower for Crushed across all axon diameters. The observed differences in 𝑁𝑀𝑇 are more apparent at larger axon diameters. Adapted from [118]. *(p<0.05) ........................................ 111

5.12 The linear density (𝜌𝐿𝑀𝑇) of microtubules for no load (Control) and loaded (Crushed) axons. Axon diameter bins are given along the X-axis and error bars are standard error mean for all plots. A strong dependency on axon caliber is observed for all linear density measures of microtubules in Control and Crushed axons. 𝜌𝐿𝑀𝑇 is lower for Crushed than Control in nearly all axon diameters. The observed differences in 𝜌𝐿𝑀𝑇 are more apparent at larger axon diameters. Adapted from [118]. *(p<0.05) .................................................................................................... 111

5.13 The spacing between (𝑆𝑀𝑇) microtubules for no load (Control) and loaded (Crushed) axons. Axon diameter bins are given along the X-axis and error bars are standard error mean for all plots. A strong dependency on axon caliber is observed for all spacing measurements between microtubules in Control and Crushed axons. 𝑆𝑀𝑇 appears larger for Crushed than Control in nearly all axon diameters. These observed differences 𝑆𝑀𝑇 are more apparent at larger axon diameters. Adapted from [118]. *(p<0.05) ................................................................................. 112

5.14 The number of (𝑁𝑁𝐹) Au-nanoparticles for no load (Control) and loaded (Crushed) axons. Axon diameter bins are given along the X-axis and error bars are standard error mean for all plots. A strong dependency on axon caliber is observed for the number of neurofilaments for both Control and Crushed axons. The 𝑁𝑁𝐹 observed is significantly lower for Crushed across all comparable axon diameters. These observed differences in 𝑁𝑁𝐹 are more apparent at larger axon diameters. Adapted from [118]. * (p<0.05) ................................................................................ 117

LIST OF FIGURES

xxvi

5.15 The areal density (𝜌𝐴𝑁𝐹) of Au-nanoparticles for no load (Control) and loaded (Crushed) axons. Axon diameter bins are given along the X-axis and error bars are standard error mean for all plots. A strong dependency on axon caliber is observed for all areal density measures. 𝜌𝐴𝑁𝐹 is significantly lower for Crushed than Control all comparable axon diameters. These observed differences in 𝜌𝐴𝑁𝐹 are more apparent at larger axon diameters. Adapted from [118]. * (p<0.05) ............................ 118

5.16 The spacing between (𝑆𝑁𝐹) Au-nanoparticles for no load (Control) and loaded (Crushed) axons. Axon diameter bins are given along the X-axis and error bars are standard error mean for all plots. The strong dependency on axon caliber observed for the Control group does not appear for the Crushed group. 𝑆𝑁𝐹 appears larger for Crushed than Control in nearly all axon diameters and remains approximately 70nm for all Crushed axons regardless of axon diameter. These observed differences in 𝑆𝑁𝐹 are more apparent at larger axon diameters. Adapted from [118]. * (p<0.05) ............................................................................... 118

5.17 Temporal description of the percent change from Control for the mean number, density, and spacing of microtubules. Data from the current study is plotted with reported Literature values at known time points and error bars are standard error mean for all plots. A) Mean number of microtubules appears to decrease immediately following loading and may require as long as 7 days to return to Control values. B) Microtubule linear density decreases appear to peak at 15min following injury before returning to Control values. C) Spacing for microtubules appears to increase following loading and return to Control values at longer time periods. [118] ........................................................................... 121

5.18 Temporal description of the percent change from Control for the mean number, density, and spacing of neurofilaments. Data from the current study is plotted with reported Literature values at known time points and error bars are standard error mean for all plots. A) The mean number of neurofilaments decreases following loading; however the return to Control values appears to require less time than microtubules by comparison (Figure 5.17A). B) Areal density for neurofilaments appears to decreases immediately following loading and may increase above the Control values at 4hrs. In addition to the lack of quantifiable data from the Literature, the differences in percent change from Control and unclear trend may be a function of a conformational shift in the neurofilament sidearm that could affect the quantification methods used in this study. C) Spacing for neurofilaments appears to increase following loading and return to Control values at longer time periods. [118] ............................................................................................................ 122

5.19 The distribution of neurofilaments in Control group relating the spacing between neurofilaments (𝑆𝑁𝐹) with the areal density (𝜌𝐴𝑁𝐹). A-B) For

LIST OF FIGURES

xxvii

Control axons, the distribution is close to homogeneous; where 𝜌𝐴𝑁𝐹 and 𝑆𝑁𝐹 are strong correlated with the number of neurofilaments and axon caliber. C) Schematic illustrating the neurofilament distribution changes with respect to 𝑆𝑁𝐹 and 𝜌𝐴𝑁𝐹 as axon caliber increases. D) With increasing axon caliber, 𝑆𝑁𝐹 slowly decreases and the 𝜌𝐴𝑁𝐹 increases. For Crushed axons, the neurofilament distribution is heterogeneous, meaning the numerical relationship between 𝑆𝑁𝐹 and 𝜌𝐴𝑁𝐹 breaks down and no longer applies. ................................................................................. 126

6.1 AIM platform with 1µm notch heights integrated at the base of the

compression pad. A) The notch allows for controlled compression at known strain levels where the notch height represents the portion of the axon not to be compressed. B) A laser scan profile shown measure notch heights at the base of the modified compression pad ................................. 135

6.2 Laser scan profile of the notch controlled devices. Pairs of lines, by color, represent the base of the compression pad and the notches formed to control strain during loading. Here the average notch height is 1.0µm (black arrow). .............................................................................................. 136

1

Chapter 1

Introduction

1.1 Significance of Problem

Traumatic brain injury (TBI) and spinal cord injury (SCI) are debilitating causes of

traumatic death and disability worldwide. Over 10 million deaths and hospitalizations

associated with these injuries occur each year with over 1.7 million of these occurring in

the United States [1, 2]. Of the injuries in the United States, approximately 52,000

individuals die as a direct consequence [3]. On a larger stage, it is estimated that 57

million people worldwide have experienced such injuries [1]. These injuries are not only

incapacitating for the individuals who sustain the injuries, but also create an economic

burden on their families, caregivers, and the socio-economic system. Socio-economic

loss of productivity and health costs are attributable to TBI and SCI. A study conducted

in 2007 to determine the cost-of-illness in Spain, considering the perspective of society,

using a 1-year time horizon and including a wide scope of related costs (medical,

CHAPTER 1. INTRODUCTION

2

adaptation, material, administrative, costs of police, firefighters and roadside assistance,

productivity losses due to institutionalization and sick leave, as well as an estimate of

productivity losses of careers, and productivity losses due to death) found TBI and SCI to

cost Spain between $1.7-5.8 billion annually [4]. A similar analysis for loss of

productivity and health costs in the United States found lifetime direct and indirect

medical costs of over $76.5 billion in 2010 [5, 6].

TBI and SCI occur from damage to the brain and spinal cord resulting from an

external mechanical force, including rapid linear/rotational acceleration and/or

deceleration, blast waves, crush, impact, or penetration by a projectile. This type of

loading is commonly sustained through falls, exposure to blasts, collisions in sporting

events, vehicular accidents, and assaults. Recent media attention on sports related

concussions and service members exposed to blast events has led to increased public

awareness and funding for the research community to improve the clinical and

pathophysiological understanding of these injuries. A study conducted over Operation

Iraqi Freedom and Operation Enduring Freedom observed that the wounding patterns

seen in Iraq and Afghanistan resemble the patterns from previous conflicts (Korea,

Vietnam, and WWII), with some notable exceptions: a greater proportion of head and

neck wounds, and a lower proportion of thoracic wounds [7]. These researchers

indicated an explosive mechanism accounted for 78% of injuries, which is the highest

proportion seen in any large-scale conflict [7]. Additionally, within leagues for soccer,

football, hockey, and other sports participants at all skill and age levels are becoming

more aware of TBI and SCI due to the increased media attention on the impairment of


3

cognitive, physical and psychosocial functions associated with these injuries. The outcry

from participants, fans, and from parents of players has resulted in new protocols for

managing concussions. However, there is still a great deal that is unknown regarding

these injuries.

Due to the overwhelming consequences of TBI and SCI, there has been an ongoing

effort to improve our understanding of the complex nature of the loading, the neural

tissue, and the pathophysiological response, with the end goal of developing preventative

and therapeutic measures. In doing so, the pathological classifications of primary and

secondary injuries, as well as focal versus diffuse injuries, have also led to the improved

understanding of the multiple pathological pathways that result in TBI and SCI.

1.2 Objectives and Structure of Thesis

This thesis focuses on understanding the mechanics of traumatic axonal injury (TAI),

more commonly called diffuse axonal injury (DAI) in humans [8]. Traumatic axonal

injury is a common pathology connecting TBI and SCI, and is considered a progressive

event gradually evolving from focal axonal alterations to delayed axonal disconnection, a

process potentially amenable for therapeutic intervention. This temporal response is

commonly referred to as a secondary injury and develops over the course of minutes,

days, and even months, providing a window for intervention and treatment. While there

have been significant steps made towards improving the understanding of the subcellular

processes of the events leading to TAI, effective treatments have yet to be developed [6,


4

8]. In terms of prevention, an improved understanding of the mechanics associated with

the pathological response of TAI is needed. Several animal models for TAI have been

developed, primarily examining the response of neural tissues to various forms of

loading. The usefulness of these models relies on the ability of researchers to control the

mechanical load and induce a TAI response in the neural tissue. By controlling the load

and having a firm understanding of the boundary conditions, the observations made can

enable insights into the cellular and subcellular changes associated with TAI. These

observations might enable therapeutic targeting of subcellular processes to mitigate or

prevent the cascade of TAI events culminating in TBI and SCI.

A key component of this thesis is defining the cellular and subcellular changes that

occur following a controlled load that culminates in TAI. To do this, the features

regulating the mechanical response of the cell need to be visualized. The mechanical

response is governed by the cytoskeleton of the neural axons, which provides structural

integrity, rigidity, shape, and transport within the cell. Rupture, failure, and degeneration

of cytoskeletal components leads to cellular (and higher) length scale degeneration.

Figure 1.1 shows our framework for studying the cytoskeletal physical response process

of traumatic axonal injury.


5

Figure 1.1: Flow chart outlining our framework for studying the cytoskeletal response process of traumatic axonal injury. Each step presents a significant increase in the degree to which we can connect the cellular level TAI response to changes in the cytoskeleton of neural axons.

The following chapter gives an overview of the current state of traumatic brain injury,

spinal cord injury, and traumatic axonal injury research. A general background of the

mechanisms for TBI and SCI is presented. This is followed by an overview of the central

nervous system where details are presented from the system level down to the subcellular

level. A definition for TAI is presented, and a discussion is provided of the experimental

models covering both the biomechanics and pathophysiology of TAI. An experimental

model for inducing TAI is developed in Chapter 3. This model utilizes a focal

compression platform enabling in vitro and in situ visualization of neural axons. A finite

element model is developed for quantifying applied loads and the validation of that

model using both imaging and integrating instrumentation into the testing platform is

presented. Chapter 4 presents changes in fluorescently labeled cytoskeletal constituents

captured in situ and under load utilizing the same focal compression platform. The

temporal aspects of the cytoskeletal changes in neural axons are discussed. Next, the

structural basis underlying the observed changes in fluorescence during axon loading is

explored through the use of TEM (Chapter 5). Chapter 5 qualitatively and quantitatively


6

details the cytoskeletal distributions for both those axons undergoing TAI and control

axons. Finally the implications and connections of this work to improving therapeutic

interventions of TAI are discussed (Chapter 6).

1.3 Contributions to Research Field

The research effort presented provides a basis for improving the fidelity of analytical

approaches concerned with the temporal evolution of TAI. Insights into traumatic brain

injury and spinal cord injury require a firm comprehension of mechanics (from several

viewpoints) as much as an understanding of the process is required from clinical,

pathological, and biomedical perspectives. Damage to the brain and spinal cord

propagates from a mechanical assault on the central nervous system, but the TAI

development is influenced by structural changes observed at the cellular and subcellular

level. It is understood that the loads applied at the macroscopic level of the individual are

translated across length scales to the level of the cell in combinations of simple

mechanical loads (tension, shear, and compression). It is important to understand how

these simplified mechanical loads at the cellular level translate to the structural changes

observed at the subcellular level of the cytoskeleton. Changes in the cytoskeleton

propagate up to the cellular, tissue, and organ levels in the form of TBI and SCI driven by

TAI. Understanding this requires the ability to couple the temporal changes in the

cytoskeleton across length scales, from the macroscale down to the nanoscale, and back.


7

To approach this problem, a novel focal compression platform is developed that is

capable of isolating neural cells and their axons. The platform is transparent, enabling

continuous visualization of in vitro and in situ loading events. A finite element model is

developed to optimize the platform and to ascertain applied loads to neural cells.

Analysis of the loading corridors applied defines TAI responses for the neural cells. This

information is fed into an experimental approach to quantify changes in cytoskeletal

expression within the loading corridors associated with TAI, and provides insights into

the temporal evolution of specific cytoskeletal populations in the axon.

Unlike currently implemented experimental approaches to TAI, our approach enables

quantification of isolated neural axons prior to, during, and immediately following a

controlled load. Existing methods for exploring the mechanics of subcellular populations

under controlled load utilize methods that damage the cells, confound interpretation of

cytoskeletal changes, and are limited in temporal and spatial resolution. Improving the

understanding of cytoskeletal evolution under load requires a methodology to capture

information as it is applied in situ. We use confocal imaging to obtain continuous

visualization of the cellular and subcellular response of the cell during loading, a direct

improvement over existing visualization methods. Finally, transmission electron

microscopy is utilized to quantify cytoskeletal changes within axons, under the same

loading conditions, in order to develop insights as to how cytoskeletal components

change immediately following loading and how this connects to the confocal imaging

results.


8

This research provides a framework for improving the fidelity of models of TAI by

providing quantification of cytoskeletal changes less than 1min after loading. This

information can be used by researchers and clinicians to target therapeutic treatments and

prevention measures for TAI. The ability to effectively relate the temporal evolution of

the cytoskeleton within a cell to the pathological condition of TAI is an invaluable tool in

the development of first response measures to TBI and SCI.

9

Chapter 2

Background

2.1 Traumatic Brain Injury and Spinal Cord Injury

A single definition for traumatic brain injury (TBI) and spinal cord injury (SCI) is

extremely difficult because of the complex clinical, pathological, and cellular/molecular

features associated with these processes. A classification system has been suggested

using pathological, clinical, or mechanistic classifications to support translational and

targeted approaches for communicating TBI and SCI research [9]. Pathological

classifications can be anatomical, describing injury as focal or diffuse, or

pathophysiological, based on primary and secondary injuries. A number of clinical

classifications have been developed including the Glasgow Coma Scale (GCS) for

clinical diagnosis of TBI, though limitations in its applicability to pediatric assessment

and poor performance in mild TBI (mTBI) discrimination are known [10].

CHAPTER 2. BACKGROUND

10

Mechanistic classifications of TBI and SCI describe impact, inertial loading,

penetrating, and blast injuries applied to the head and spine that result in damage to the

brain and spinal cord. These mechanisms can occur individually or as a combination.

Impact injuries require the body to make contact with an object, with the contact force

transmitted to the brain or spinal cord. Experimental studies in nonhuman primates have

demonstrated that acute subdural hemorrhages secondary to torn bridging veins are

produced by rapid linear acceleration (and deceleration) [11, 12]. More recent research

has suggested that we must also include rotational acceleration and deceleration [13].

Inertial forces causing injury do not require contact, but instead the brain moves within

the cranial cavity causing damage. An example of potential impact and inertial loading is

shown in Figure 2.1. Penetrating injuries produce damage when an object passes through

the protective covering of the skull or vertebral column resulting in direct parenchymal

damage to the underlying tissue. Blast injuries are the least well understood currently

and are primarily seen in military or terrorist situations where the shock waves from an

explosive device can result in injuries to the brain and spinal cord parenchyma (Figure

2.2) [15]. An outline of the mechanisms associated with TBI and SCI are given in Table

2.1 and the pathological description for traumatic axonal injury is given in Section 2.3.

Figure 2.3 illustrates how loads applied at the macroscale scale (in events such as vehicle

accidents, sport injuries, falls, or blasts) are translated smaller length scales where

observable changes in the subcellular structure lead to pathological responses that

manifest as TBI and SCI.


11

Figure 2.1: Mechanics classifications for TBI and SCI mechanisms include penetration, inertial loading, contact, and blast. Here an individual illustrates potential combinations of inertial and contact loading mechanisms from an elevated position with forward momentum. Note the absence of a helmet or other protective equipment. [14].

Figure 2.2: Blast injury mechanism for TBI and SCI where the shock waves from an explosive device can result in injuries to the brain and spinal cord parenchyma. Here the focus is on the primary mechanism where blast-induced neurotrauma (primary injury), without contact, is occurring due to the blast wave itself. Secondary (penetration) and tertiary (impact) mechanisms and their associated neurotrauma from flying debris and coup-contrecoup motion are also shown. (Image adapted from [15]).


12

Figu

re 2

.3:

Lo

adin

g ac

ross

mul

tiple

len

gth

scal

es,

from

the

mac

rosc

ale

to t

he n

anom

eter

lev

el,

conn

ects

the

bi

omec

hani

cs o

f TAI

(tra

umat

ic a

xona

l inj

ury)

to t

he p

atho

logi

cal r

espo

nse

that

cul

min

ates

in T

BI a

nd S

CI.

Loa

ding

is

appl

ied

at

the

mac

rosc

ale

in

the

form

of

co

ntac

t, pe

netra

tion,

an

d no

ncon

tact

(th

roug

h lin

ear

and

rota

tiona

l ac

cele

ratio

n/de

cele

ratio

n) m

etho

ds s

pann

ing

a du

ratio

n of

tens

of m

illise

cond

s to

a fe

w s

econ

ds.

Thes

e lo

ads

trans

late

to

cha

nges

at m

ultip

le le

ngth

sca

les

of n

eura

l tis

sues

, the

ir ce

lls, a

nd th

e un

derly

ing

cyto

skel

etal

stru

ctur

e. A

t the

sm

alle

r le

ngth

sca

les,

the

com

plex

app

lied

load

is u

nder

stoo

d in

the

cont

ext o

f sim

ple

mec

hani

cal l

oadi

ng (t

ensi

on, c

ompr

essi

on,

and

shea

r).

Thes

e si

mpl

ified

mec

hani

cal l

oads

lead

to th

e pa

thob

iolo

gica

l res

pons

e ob

serv

ed fo

r TA

I and

cul

min

ate

in

TBI a

nd S

CI a

t lar

ger l

engt

h an

d tim

e sc

ales

.


13

Table 2.1: Mechanisms of TBI and SCI. (Adapted from [16]) Mechanism Main Pathology Impact Vascular (hemorrhages)

Traumatic axonal injury Inertial loading – Linear & Rotational Traumatic axonal injury Penetrating Local tissue necrosis Blast Traumatic axonal injury Brain swelling

2.2 Central Nervous System Anatomy and Structure

The structure of the central nervous system (CNS), at multiple length scales, plays a

critical role in the development of TBI and SCI. To better understand the CNS it is

useful to approach this complex system from the larger scale size, brain and spinal cord

(organ level), and move toward the primary functional components of the CNS, neurons

(cellular level), before moving into the subcellular regime.

2.2.1 Central Nervous System

The central nervous system (CNS) is one of two major divisions of the nervous

system and consists of the brain and spinal cord (Figure 2.4). The CNS processes

information to and from the peripheral nervous system (PNS) and is the main network of

coordination and control for the entire body. The spinal cord extends various types of

nerve fibers from the brain and acts as a switching relay terminal for the PNS. The CNS

divisions are protectively encased in rigid structures of the skull and vertebral column.

Brain and spinal cord tissues are composed of grey and white matter (Figure 2.5).

The grey matter primarily contains the cell bodies of neurons and associated processes


14

such as dendrites. The white matter predominantly consists of myelinated neural axons

and function to connect various neuronal cell bodies (grey matter) to each other. The

myelin acts as an insulator, increasing the transmission speed along axons, and gives the

white matter its color. In living tissue, grey matter actually has a very light grey color

with yellowish or pinkish hues, which come from capillary blood vessels and neuronal

cell bodies [20].

Figure 2.4: Central nervous system (CNS) components of brain and spinal cord. At the periphery of the spinal cord the peripheral nervous system (PNS), a subdivision of the nervous system containing all nerves and ganglia outside the CNS, interact to form a communication relay between the two divisions of the nervous system. (Images adapted from [17, 18]).


15

Figure 2.5: White and grey matter of the CNS. A) Transverse section of spinal cord. B) Frontal section of the brain. (Image adapted from [19]).

2.2.2 Neural Cells and Tissues

The brain and spinal cord are heterogeneous and consist of glial cells, neurons, and

parts of the cerebral vasculature (Figure 2.6). The neurons are the primary functional

component of the CNS and transmit information through electrical impulses along long

slender projections called axons (Figure 2.7). Axons function as transmission lines for the

CNS and, as bundles, form nerve fibers. Axons range in lengths from a few microns to up

to one meter and have variable diameters of a few hundred nanometers to a few microns.

Non-neuronal, glial cells function as the glue of the nervous system providing

homeostasis, support, and protection of neurons. Glial cells include astrocytes, radial

glial cells, oligodendrocytes, and ependymal cells in the CNS. For perspective, more

than 10 billion neurons constitute less than one tenth of brain cells while the remainder

consist of glial cells.


16

Figure 2.6: Cells of the CNS. Glial cells include astrocytes, radial glial cells, oligodendrocytes, and ependymal cells in the CNS. These cells provide homeostasis, maintenance, and protection for neurons. Neurons, myelinated by oligodendrocytes, compose the white matter tracts of the CNS. Capillaries and other parts of the cerebral vasculature deliver oxygenated blood, glucose and other nutrients to the brain by the arteries and the veins carry deoxygenated blood back to the heart, removing carbon dioxide, lactic acid, and other metabolic products. (Image adapted from [21]).


17

Figure 2.7: Primary parts of neuron for information flow. Dendrites receive electrical impulses and pass the information through the cell body and down the axon where it passes the signal onto another neuron or to an effector cell (muscle, gland, or organ cell capable of responding to a stimulus) [20]. (Image adapted from [22]).

2.2.3 Cytoskeletal System

The cytoskeleton is an elaborate network of proteins, having a variety of

configurations, whose purpose is to provide functional and structural stability to the cell.

Major constituents of the cytoskeleton in CNS axons include neurofilaments (NFs) and

microtubules (MTs) (Figure 2.8).

Microtubules are hollow cylindrical structures assembled from dimers of α-tubulin

and β-tubulin as linear protofilaments (Figure 2.9). They have an approximate molecular

weight of 50kDa and form as 24nm wide hollow cylinders [25, 26]. Microtubule

configurations are dynamic, and they can change length by actively polymerizing and


18

depolymerizing over a variable range of rates, with steady state rates approximating 2.0-

2.5μm/min [26-28] (Figure 2.10). Microtubules are also polarized: the positive end is

always more active than the negative end of the tubule. Microtubule functions include

intracellular transport and structural rigidity [25-28]. Microtubules are often associated

with molecular motors, which are proteins designed to travel along the microtubule,

usually to help transport organelles intracellularly. These molecular motors include

kinesins, which move towards the positive end of the microtubule, and dyneins, which

move towards the negative end of the microtubule.

Figure 2.8: Cytoskeletal substructure of neuron. Microtubules and neurofilaments are two primary constituents of the neuron. A) Microtubules forms as 24nm wide hollow cylinder and function to provide transport and structural rigidity to the axon. B) Neurofilaments have a rod domain core from which sidearm extensions splay outward and function to regulate axon diameter and provide mechanical strength and stability. (Image adapted from [23]).


19

Figure 2.9: Transmission electron microscopy of microtubules. Microtubules (examples shown by arrows) exist as linear rod-like hollow cylinders that function to provide intracellular transport and structural rigidity to cells. Scale bar = 50nm. (Image adapted from [24]).

Figure 2.10: Cryoelectron microscopic images and illustrations of microtubule growth and shrinkage. A) α-tubulin and β-tubulin are assembled as linear protofilaments. B) Depolymerization of tubulin protofilaments. Note the dissembling protofilaments are highly curved while the assembling protofilaments are very straight and fan like. Scale bars = 25nm. (Image adapted from [29]).


20

Neurofilaments provide mechanical strength and stability, determine axon diameter,

and may be involved in transport of intracellular components [30-33]. Neurofilaments

are heteropolymers that consist of three polarized sidearm subunits: NF-light (~70kDa),

NF-medium (~150kDa), and NF-heavy (~200kDa). The subunits attach at a rod domain

core measuring approximately 12nm in diameter, and the polarized sidearms function to

space NFs from each other at regular intervals across the axon diameter [32, 33]. Figure

2.11 shows an electron micrograph where the thicker neurofilament core (horizontal

orientation) has thin sidearm projections branching outwards.

Figure 2.11: Electron micrographs of neurofilaments. Neurofilament project sidearm extensions from neurofilament core (inset, arrows). Scale bars = 200nm. (Image adapted from [34]).

Although each subtype NF protein is composed of similar rod domains, the structure

of NF-medium and NF-heavy also includes “sidearm” domains of differing lengths [35]

(Figure 2.12). These sidearms may be phosphorylated, the extent of which is thought to

influence the diameter of axons [36-38]. Neurofilaments interact with other cytoskeletal

elements, notably microtubules for transport along the axon, and dysfunction of

neurofilaments leads to inability of the cell to withstand mechanical force.


21

Figure 2.12: Illustrative example of the neurofilament substructure, a major structural component of the axonal cytoskeleton. The rod domains of NF subtype proteins run parallel in the core of the NF. Whereas NF-light (white) has only a rod domain, the sidearm domains of NF-medium (red) and NF-heavy (blue) stand out from the NF core creating a physical spacing between neighboring NFs [35].

2.3 Definition of Traumatic Axonal Injury

Traumatic axonal injury (TAI), a common pathology for both TBI and SCI, is

characterized by focal or multifocal damage to the axons of neural cells comprising the

white matter tracts in the CNS [8, 35, 39]. Focal axonal damage can result from

mechanical forces applied to the white matter of the CNS over short time durations,

spanning tens of milliseconds to a few seconds [12, 35, 40-51]. These loads are

understood through several potential mechanisms for injury including contact,

penetration, and noncontact (in the form of accelerative and decelerative) loading at the

mesoscale.

At smaller length scales, the loading is understood as a combination of stretch,

compression, and shear applied to neural cells and their substructure [8, 45-48, 51-58].

These loads affect the organization and distribution of the substructure of neural cells and

translate to pathological changes that are highly dependent on the mechanism of loading


22

(Figure 2.3). Applied loads lead to a series of progressive changes in neural axons

indicating the onset of TAI.

2.3.1 Experimental Techniques to Investigate TAI

2.3.1.1 Biomechanics of TAI

A variety of animal models for TAI have been utilized to examine the effects of

dynamic shear, tensile, and compressive strains at both cellular and subcellular levels [8,

40, 45, 47-49, 51, 53-56, 59-63] (Table 2.2). In vitro models of axonal injury include

models based on transection, shear, compression, hydrostatic pressure, hydrodynamic

changes, acceleration, and cell stretch [64]. Several of these in vitro injury models aim to

deform neural cells in a controlled manner so as to mimic tissue deformation. Such

models have been used to study several aspects of axonal injury including the immediate

post-injury rise in intracellular calcium levels, the electrophysiological responses of

neurons, changes in ionic homeostasis, neurofilament alterations, membrane permeability

changes, and axonal swelling [65-69]. Although these models are limited by being

generally one-dimensional, they provide insights into axonal changes in response to

injury and are useful alternatives to in vivo models in certain studies. Morrison et al.

developed an in vitro model of two-dimensional stretch injury in which a device was used

to stretch tissue slice cultures in a biaxial manner [70]. A model permitting three-

dimensional deformation of neural cultures has also been established by LaPlaca et al.

[55]. One important aspect to keep in mind is the type of mechanical insult, such as


23

uniaxial or biaxial stretch, can influence the acute mechanisms of axonal injury [71]. It is

critical therefore to understand that the primary aim of in vitro models is to mimic the

structural consequences of the injury, rather than mimicking the mechanical force levels

observed at larger length scales.

Figure 2.13: 3-D Cell Shearing Device (3-D CSD) components. (a) A schematic representation of the 3-D CSD. The device can be mounted on a confocal microscope to obtain 3-D images before, during, and after mechanical deformation. A closed-loop proportional-integral-derivative controller system (PID controller) with feedback from a digital variable reluctance transducer (DVRT) governs a linear actuator, inducing motion of the cell chamber top plate (not to scale). (b) The cell chamber consists of a top plate with polyethylene filters to interface with the 3-D cell cultures. The top plate is mounted above the cell reservoirs and connected to the linear actuator to impart high rate deformation. (c) The horizontal motion of the linear actuator drives the displacement of the cell chamber top plate, inducing shear deformation in the Sylgard mold and matrix (with either cells or microbeads) (not to scale) [55].


24

To apply dynamic shear, LaPlaca et al. developed a device that incorporates the

features of three dimensional (3D) cell cultures under prescribed conditions of simple

shear strain (Figure 2.13) [55]. The device works by generating a shear strain between

the top plate, driven by a linear actuator, and the cell reservoir, where neurons had been

integrated into a scaffolding gel from E17 rats. This approach allowed LaPlaca et al. to

connect neuronal viability with applied shear strain rates in a 3D cell culture and

provided evidence of cellular thresholds for linear shear strain fields.

Tensile loading of neural axons has been accomplished by several researchers under a

variety of experimental models [8, 45, 47-49, 53, 54]. Tang-Schomer et al. grew primary

cortical neurons from E18 rats on a micropatterned cell culture platform (Figure 2.14)

[54]. The platform confined axonal growth to a specific region where loading was to be

applied. The platform was placed in a sealed device and subjected to dynamic

mechanical stretch from a controlled air pulse. Other researchers have used optic nerve

stretch to investigate the effects of tensile loading on adult mice and guinea-pig axons

and their associated substructures [8, 45, 47-49, 53]. These studies have provided insight

into the evolution of the cytoskeletal structure within neural axons following load.


25

Tabl

e 2.

2:

Expe

rimen

tal t

echn

ique

s for

inve

stig

atin

g TA

I. St

udy

Ana

tom

ical

Len

gth

Scal

e of

Exp

erim

ent

Loa

ding

Met

hodo

logy

V

isua

lizat

ion

Rag

hupa

thi a

nd M

argu

lies

(200

2) [5

9]

Org

an

Rap

id, i

nerti

al (n

onim

pact

) he

ad ro

tatio

n Li

ght m

icro

scop

y, im

mun

oblo

tting

Pettu

s and

Pov

lisho

ck

(199

6) [5

6]

Tiss

ue

Flui

d pe

rcus

sive

inju

ry

Tran

smis

sion

ele

ctro

n m

icro

scop

y

Saat

man

et a

l. (1

998)

[61]

Ti

ssue

La

tera

l flu

id p

ercu

ssiv

e in

jury

Li

ght m

icro

scop

y, im

mun

osta

inin

g La

Plac

a et

al.

2005

[55]

Ti

ssue

Sh

ear s

train

of 3

D n

eura

l cel

l cu

lture

C

onfo

cal I

mag

ing

Max

wel

l and

Gra

ham

(1

997)

[45]

C

ell

Opt

ic n

erve

stre

tch

Tran

smis

sion

ele

ctro

n m

icro

scop

y

Jafa

ri et

al.

(199

7) [4

7]

Cel

l O

ptic

ner

ve st

retc

h Tr

ansm

issi

on e

lect

ron

mic

rosc

opy

Jafa

ri et

al.

(199

8) [4

8]

Cel

l O

ptic

ner

ve st

retc

h Tr

ansm

issi

on e

lect

ron

mic

rosc

opy

Serb

est e

t al.

(200

7) [4

9]

Cel

l O

ptic

ner

ve st

retc

h Im

mun

oblo

tting

M

axw

ell e

t al.

(200

3) [5

3]

Cel

l O

ptic

ner

ve st

retc

h Tr

ansm

issi

on e

lect

ron

mic

rosc

opy

Tang

-Sch

omer

et a

l. (2

003)

[54]

C

ell

In v

itro

dyna

mic

stre

tch

Fluo

resc

ence

mic

rosc

opy,

tra

nsm

issi

on e

lect

ron

mic

rosc

opy

Saat

man

et a

l. (2

003)

[60]

C

ell

Opt

ic n

erve

stre

tch

Ligh

t mic

rosc

opy,

im

mun

ohis

toch

emis

try

Raj

agop

alan

et a

l. (2

010)

[6

3]

Cel

l Em

bryo

mot

or n

euro

n st

retc

h Fl

uore

scen

ce a

nd li

ght m

icro

scop

y


26

Figure 2.14: Dynamic mechanical stretch of isolated cortical axons. A, B) Schematic illustration of axonal stretch injury model. An axon-only region of the elastic membrane overlaps with a 2 by 15mm slit at the bottom of an airtight chamber. A controlled air pulse deflects the elastic membrane downward, thus inducing a tensile elongation exclusively to the axons. C) Phase-contrast imaging of the axon-only region (formed by a silicone stamp that creates microchannels permitting only axon outgrowth across the channels). D) Fluorescence microscopic confirmation that the neurites in the microchannels were axons demonstrated by immunoreactivity to neurofilament protein (NF, green), while immunoreactivity to microtubule-associated protein 2 (MAP2, red), a specific marker for the dendrites, was found exclusively outside the channels. Scale bar = 100μm. (Image adapted from [54]).

Fluid-percussive injury (FPI) has been utilized to examine the effects of a moderate

TBI in an attempt to understand changes in cytoskeletal configuration following load [56,

61, 62]. Researchers used adult male cats, under anesthesia, with a steel tube inserted

into the skull to apply FPI through a column of saline impacted by a pendulum. While

FPI (and other weight drop methods) can apply compressive loads to bulk tissue, it is

known local tension exists at the boundaries of the impacted regions. This limitation

inspired the loading methodology applied for the current research.


27

The use of simple mechanical platforms to apply controlled loads to neural cells is a

useful approach to developing insights for TBI and SCI. Researchers have expanded the

understanding of these injuries using tensile loading platforms (Table 2.2). We now

move to address how mechanical loading in the form of transverse compression (applied

exclusively to neural axons) connects to changes at the cellular and subcellular levels.

Some possible explanations for why the literature has not answered this form of loading

include the difficulty of isolating neural axons, developing a controlled testing platform

for applying known loads, and providing a qualitative and quantitative approach for

characterizing observed changes. Additionally, while studies have provided insight into

the temporal aspects of the cytoskeletal structure following TAI there is a general lack of

understanding of the evolution of these changes during and immediately following

loading. This compounds existing difficulties for exploring mechanics of subcellular

populations under controlled load which include methods that kill the cells, confound

interpretation of cytoskeletal changes, and are limited in temporal and spatial resolution.

2.3.1.2 Pathobiology of TAI

Complex loading associated with TBI and SCI at the mesoscale translates to smaller

length scales in the form of combinations of stretch, compression, and shear as

mechanisms for TAI (Figure 2.3). It is therefore necessary to understand how these

forms of loading translate into the morphological and functional damage associated with

TAI. While the majority of axons may not undergo immediate disruption (primary

axotomy) at the time of injury, the loading leads to a temporal response culminating in a


28

progressive loss of neural connectivity leading to physical and cognitive disability

(secondary axotomy) [52, 54, 56-58] (Figure 2.3).

Morphological damage is usually apparent by the formation of nodal blebs, and

swellings at the site of injury. These blebs are associated with secondary axotomy where

the axon is not transected, yet a sequence of events is initiated leading to degeneration of

the axon. Secondary axotomy has been linked to the disruption of microtubule and

neurofilament networks in white matter tracts of the CNS (as a result of mechanical

perturbation) and has been associated with the formation of nodal blebs, or swellings,

along the length of the axon at the site of injury [45-50, 52-54, 56, 57] (Figure 2.15).

Figure 2.15: Temporal evolution of secondary axotomy and its effects across multiple length scales of neural tissue shown in Figure 2.3. Following loading, changes at the cytoskeletal level propagate up to the cell and tissue levels where they are manifested in the forms of nodal blebs and damage tissue parenchyma. (Images adapted from [20, 54]).


29

Secondary axotomy can be viewed at two levels for TAI: (i) the level of the cell and

the associated influx of ions, and (ii) the level of the cytoskeleton and changes to its

distribution.

For secondary axotomy at the cellular level, initial swelling and further damage of

injured axons may result from a change in ionic homeostasis. These changes have been

detected and characterize TAI through the use of experimental markers (Table 2.3). In

models of white matter anoxia, a pathologic influx of sodium (Na+) through sodium

channels in axons has been well characterized; experimental models using dynamic

stretch of axons in vitro have also resulted in sodium influx [79-85]. The sodium influx

induced in these models was found to modulate a substantial increase in the intracellular

calcium (Ca2+) concentration in injured axons (Figure 2.16) [68]. Using a model of optic

nerve stretch, Maxwell and colleagues have also reported evidence of calcium influx into

axons following stretch injury [86, 87]. As a result of these ionic changes, osmotic

swelling of axons may occur shortly following injury due to sodium influx, while the

increase in intracellular calcium may induce the deleterious activation of proteases

leading to additional cytoskeletal damage [9, 88-91]. Although these processes are

thought to play important roles in the degeneration of injured axons, swelling alone is not

considered the principal feature of traumatically injured axons. Rather, researchers have

shown an accumulation of axonal transport proteins in swollen regions of axons is the

clear signature of catastrophic damage [92-97]. The most commonly used markers of this

accumulation are the fast transport β-amyloid precursor protein (β-APP) and the slow

transport neurofilament (NF) proteins.


30

30

Tabl

e 2.

3:

Expe

rimen

tal m

arke

rs fo

r det

ectin

g an

d ch

arac

teriz

ing

TAI.

(Ada

pted

from

[65]

).

App

roac

h fo

r de

tect

ing

TA

I M

arke

rs

Path

olog

ical

Info

rmat

ion

Stud

y

Neu

roim

agin

g D

TI/D

WI

Impa

ired

dire

ctio

nalit

y an

d in

tegr

ity o

f ax

ons,

type

s of e

dem

a [7

3, 7

4]

Seru

m/C

SF A

naly

sis

S-10

0 G

lial d

estru

ctio

n [7

5, 7

6]

N

SE

Neu

rona

l inj

ury

CTP

M

icro

tubu

le d

isso

lutio

n

Imm

unoh

isto

chem

istry

Β

-APP

Im

paire

d ax

onal

tran

spor

t [4

9, 5

9-61

, 77,

78]

NF-

68/N

F-20

0 N

euro

filam

ent s

truct

ural

dam

age

El

ectro

n m

icro

scop

y U

ltras

truct

ural

cha

nges

A

xona

l ret

ract

ion

bulb

s, ax

olem

ma

and

mye

lin s

heat

h in

terr

uptio

n, m

icro

tubu

le

loss

, N

F co

mpa

ctio

n,

axon

al

and

mito

chon

dria

l sw

ellin

g

[35,

45,

47,

48,

53,

54,

56,

78]

CTP

= c

leav

ed T

au p

rote

in, D

TI =

diff

usio

n te

nsor

imag

ing,

DW

I = d

iffus

ion

wei

ghte

d im

agin

g, N

SE =

Neu

ron

spec

ific

enol

ase,

S-1

00 =

S-1

00 p

rote

in fa

mily


31

Figure 2.16: Ion channels located along the axon membrane in neurons open as a result of mechanical loading, stretch is shown here, leading to an ionic influx. A) The mechanism shown starts with mechanosensitive sodium channels activation in response to an applied strain on the axon membrane. In response to the influx of sodium, sodium/calcium exchangers reverse direction (B) and the voltage gated calcium channels are activated (C). These mechanisms, collectively, are thought to contribute to the pathological influx of calcium in the axon. (Image adapted from [68]).

At the subcellular level, one of the primary sources of impaired axonal transport is

thought to be damage to the axonal cytoskeleton, which is primarily composed of

microtubules and neurofilament proteins. It was initially proposed by Povlishock and

others that trauma induces compaction of NFs due to proteolysis of the sidearms resulting

in impaired transport with subsequent swelling [8, 98, 99]. Other researchers have

supported this hypothesis, showing that the rod domains of NF-heavy proteins are

exposed in axons, potentially due to loss of the sidearms following trauma [99]. In

addition, following inertial brain trauma in the pig, the accumulation of dephosphorylated


32

NF sidearm domains in swollen regions of damaged axons appears in the absence of the

rod regions, which also suggests that sidearm cleavage had occurred [100].

Alternatively, dephosphorylation of the NF sidearms following trauma may also

contribute to the compaction of the axonal NFs [101]. A modification to Povlishock’s

proposal came in a follow on study by Okonkwo et al. where, instead of sidearm loss,

NFs saw a reduction in height of the sidearm length extension (Figure 2.17) [101].

Coupled with the changes in NF structure, the disassembly of microtubules in axons

may contribute to the impairment of axonal transport following trauma. Axonal

microtubules are the primary conduits for fast axonal transport and researchers have

shown they may break down due to high calcium concentrations [45]. Tang-Schomer et

al. has shown mechanical rupture of cytoskeleton in nodal blebs, where microtubule

breakage is implicated in the accumulation of proteins in these regions following load

(Figure 2.18) [54]. Other researchers have shown a decrease in the number of

microtubules in axons shortly following injury in an in vivo optic nerve stretch model [8,

45].


33

Figure 2.17: Computer-assisted analysis of NF sidearms in control (A, C, E) and injured (B, D, F) axons. Electron micrographs were taken of uninjured (A) and injured (B) axons. Enlarged, color-enhanced axonal fields reveal the detail of control (C) and injured (D) axons. C) Wide interfilament spacing (arrows) and prominent sidearms (arrowhead) are observed in control axons. D) NF compaction and reduction of NF sidearm height are observed in injured axons. E) Further enlargement of an individual sidearm for control axon (green arrow). F) Individual sidearm (green arrow) for injured axons reveal reduced height in concert with the compaction of the adjacent NFs (arrows) which lie in close proximity. Scale bars = 500nm for A, B. Scale bars = 50nm for C, D. Scale bars = 5nm for E, F. (Image is adapted from [101]).


34

Figure 2.18: Transmission electron microscopy (TEM) of longitudinal sections for injured axons demonstrated ultrastructural alterations of the axonal microtubule lattice (arrows). Microtubules were identified as dark 24nm wide filamentous structures that traversed the main axis of an axon. (A) Microtubules appear unchanged in straight areas of the axon. B-C) In areas where nodal blebs are present, microtubules appear to lose continuity at the peaks and display conspicuous free ends that appear frayed (similar to a shorting microtubule undergoing catastrophic depolymerization) (asterisk). Scale bar = 500nm. (Image is adapted from [54]).

These cytoskeletal changes are augmented by direct damage to the axon membrane,

or “axolemma,” at the most severe mechanical levels of trauma [56, 102, 103]. The first

in vivo evidence of this axolemmal damage was demonstrated by the permeability of a

protein tracer into damaged axons. Interestingly, however, dynamic stretch of axons in

vitro did not induce axolemma permeability to small proteins unless primary axotomy

had occurred [69]. In addition, with ion channel blockade, calcium did not freely diffuse

into these axons, until primary axotomy occurred with a greater than 75% increase in

length [80]. Taken together, the literature suggests that changes in axolemmal

permeability reflect the most extreme circumstance of traumatic axonal trauma that only


35

occurs in severe injury. These circumstances may include immediate disconnection of

axons due to tissue tears or a physical wrenching of the tissue across obstructions such as

the bone structure encasing it.

2.4 Summary of Current TAI Research

Traumatic axonal injury research has encompassed several different areas as outlined

in Figure 2.19 and described within this chapter. The ultimate objective of each of these

research thrusts is to provide insight in the development of prevention measures and

therapeutic interventions for TBI and SCI. Before this end can be met, insight into the

temporal evolution of TAI needs to be further developed at the cellular and cytoskeletal

levels. An overview of the current research efforts into the evolution of neural axons and

their cytoskeletal structure has been presented. Researchers studying TBI and SCI have

utilized several animal models under a variety of loading methodologies (shear, stretch,

and compression). Focal compression is an area where little insight has been gained

regarding TAI from experimental models.

Figure 2.19: Understanding the development and evolution of TAI mechanisms are an important step towards the advancement of preventative measures and treatment options for TBI and SCI.


36

The subsequent chapters of this thesis will focus on the development of an

experimental framework for providing insights into the temporal evolution of TAI. An

experimental setup for exploring cellular level threshold corridors governing the CNS

axon response to focal compression will be developed. The subsequent changes in

cytoskeletal expression, for microtubules and neurofilaments, within the axon will be

defined for TAI. This approach makes use of confocal imaging to provide some insight

into the temporal evolution of the cytoskeleton in situ and in-vitro. A second approach

utilizing transmission electron microscopy (TEM) is presented to provide greater spatial

information regarding the cytoskeletal distribution and quantification following TAI

loads. Finally a mechanism for unifying changes in the cytoskeleton to cellular level

response for TAI will be proposed that can connect to the mesoscale conditions

pathologies of TBI and SCI.

37

Chapter 3

Axon Injury Microcompression Platform

3.1 Introduction

Under mechanical loading to the CNS, neural cells are subjected to a complex state of

loading. A simplified platform for applying known focal loads is required to obtain

insights regarding the TAI response of neural axons. Such a platform is described in this

chapter. Understanding how controlled inputs to the platform translate into applied loads

on the cell is critical for connecting the injury response to threshold levels. This chapter

also presents a finite element model (FEM) constructed for the axon injury

microcompression (AIM) platform. The model provides applied loads for given input

fluidic pressures to the AIM platform. The predicted loads are compared against

measurements taken by instrumentation integrated into the testing platform. The

correlation between applied loads and the injury response of the neural axons is

CHAPTER 3. AXON INJURY MICROCOMPRESSION PLATFORM

38

discussed. The majority of the work presented in Chapter 3 is from a previous publication

[51].

3.2 AIM Platform Description

The AIM platform (Figure 3.1) is an amalgamation of two distinct constructs: (A) a

microfabricated chamber with three compartments for neuronal cell bodies, proximal, and

distal axons; and (B) a valve-based elastomeric compression injury pad system for

inducing graded injury to micron-scale segments of single axons [51]. The

microfabricated chamber was constructed from two layers of polydimethylsiloxane

(PDMS) (Sylgard 184; Dow Corning, Midland, MI, USA) following well established

replica molding protocols [104].

Briefly, the master template for the first layer (for neurons and medium) was created

using a three layer microfabrication process. Silicon wafers (WRS Materials, San Jose,

CA, USA) were processed with 6.25-10mm thick SU-8 3005 (Microchem, Newton, MA,

USA) to define two parallel linear arrays of ~125 microchannels each (Figure 3.1A Top).

The process was immediately repeated with a 20mm SU-8 3025 layer to define the partial

height of each of the three compartments (Figure 3.1A Middle) and the separation

distance between the glass substrate and the injury pad. Finally, a 30mm SU-8 3025 layer

was deposited to define the height of the compression injury pad (Figure 3.1A Bottom)

and complete the composite height of all compartments. Once completed, the mold was


39

spun coat with an approximately 45–65mm thick layer of PDMS prepolymer and fully

cured at 80°C for 20min (Figure 3.2B).

The master for the second (control) layer was patterned with SU-8 3050 to define four

injury controllers (Figure 3.1B) connected by 50mm wide control lines to independently

addressable access ports (D = 1 mm). This master was used to create a ~5mm thick

PDMS cast (Figure 3.2A1), cut into individual devices, and punched with sharpened

gauge #23 needles (Figure 3.2A2; McMaster-Carr, Santa Fe Springs, CA, USA) to create

access ports. Individual devices and the PDMS-coated wafer were introduced into an

oxygen plasma cleaner (Harrick Plasma, Ithaca, NY, USA) and surface treated (30 Watts;

1.5mins). Injury controllers were visually aligned to compression pad features, brought

into intimate contact, and baked overnight at 80°C to fuse device layers (Figure 3.2C).

Afterwards, composite devices were removed and neuronal layer access ports were

created using 3mm biopsy punch tools (Figure 3.2D; Huot Instruments, Menomonee

Falls, WI, USA). Devices were sterilized by ethanol sonication, autoclaved, and sealed to

50mm #1 glass bottom petri dishes (Wilco Wells, Amsterdam, The Netherlands) prior to

use (Figure 3.2E).

Details pertaining to photoresist (soft/hard/post exposure) bake, exposure, and

development times can be found in the manufacturer’s technical sheet (Microchem,

Newton, MA, USA). A complete protocol for the microfluidic device fabrication is given

by Hosmane [105]. All PDMS protocols involved standard 10:1 base to cross-linker

ratios by mass, and detailed devices geometries are listed in Table 3.1.


40

Table 3.1: AIM platform geometry. (Adapted from [51]) Feature Figure Length (L) Width (W) Height (H) Resist Microchannels 3.1A Top 500μm 10μm 6.25-10μm 3005 Pad clearance 3.1A Middle 8.0-10.0mm 0.25-1.0mm 20μm 3025 Pad height 3.1A Bottom 1.8mm 30μm 30μm 3025 Control pad 3.1B 2.0mm 1.0mm 100μm 3050

Figure 3.1: The AIM platform was assembled from two independently fabricated templates. The neuronal template (A), which fluidically defined where neurons and medium resided, was constructed as follows: Beginning with a bare silicon wafer, (A Top) an initial resist layer was processed to yield two linear arrays of microchannels. (A Middle) Afterwards, a thicker resist layer was deposited to first define the compartment bases and the compression injury pad clearance height. (A Bottom) This process was repeated once again to complete the compartment height and define the compression injury pad geometry. (B) A separate controller template defined which areas of the final device were controlled (or deformed) by compressed air and was fabricated using a single resist layer. (C) A representative AIM device was filled with dye to visually show both the neuronal (blue/green) and control (red) layers. (D) The device cross-section depicts the relative dimensions of the compression injury pad relative to the compartment and microchannels. Schematics A and B are not to scale. [51]


41

Figure 3.2: A cross-sectional view of device construction. The AIM device was assembled by (A1) first pouring a thick layer of PDMS over the control template and baking until fully cured. (A2) The cured PDMS on the control wafer was removed and access ports were punched. (B) At the same time, the neuronal template was spin-coated with a thin-film of uncured PDMS followed by a complete bake. (C) Both the control devices and the coated neuronal template were oxygen plasma treated, aligned, bonded, and baked overnight to facilitate fusion between adjacent layers. (D) Composite devices were then removed and fluidic access ports to the neuronal layer were punched. (E) Devices were cleaned, bonded to glass-bottom Petri dishes, and seeded with neurons prior to injury experiments. Figure not to scale. [51]

3.3 Axon Integration into AIM Platform

Primary hippocampal neurons were derived from embryonic day 17 (E17) pups [106].

Prior to cell seeding, devices were coated overnight at 4°C with 200μg mL-1 PDL (Sigma,

St. Louis, MO, USA), washed 3x the next day with tissue culture grade H20, filled with

neurobasal media, and placed in a standard humidified cell culture incubator set to 37°C

and 5% CO2 (Thermo Scientific, Boston, MA, USA) for 15–30mins [107]. Primary

neurons were loaded into the somal compartment at a low density (< 100 cells per mm2;

150–450 neurons/device). After 6–8 days in culture, axons could be seen extending into


42

the middle and distal chamber of the device in sparse numbers to allow tracking of

individual processes for subsequent experiments. Media was added every 3 to 4 days to

maintain neuronal viability. For experiments in which cells were labeled with a

fluorescent protein, dissociated neurons were nucleofected (Amaxa, Gaithersburg, MD,

USA) with a plasmid encoding the tau-TdTomato gene as per the manufacturer’s

instructions. Efficiency of labeling was greater than 50%.

3.4 Finite Element Model of AIM Platform

A computational model of the AIM platform was constructed to assess the magnitude

of the compressive load applied by the PDMS compression pad to the axon and glass

substrate.

3.4.1 Idealization of AIM Finite Element Model

The system was idealized in a two dimensional (2D) model. Since the stiffness of the

axon is several orders of magnitude smaller than that of the PDMS compression pad and

glass substrate, the resistance of the axon itself to the applied loads was assumed to be

negligible [108]. The 2D plane strain FEM of the device was constructed in the Abaqus

6.9-EF/Standard commercial software package (Dassault Systems Simulia Corp.,

Providence, RI, USA). The model is composed of three assembled pieces: the glass

substrate, the PDMS membrane, and the PDMS control pad. Device dimensions were

taken from three-dimensional (3D) reconstructed confocal images of the device. A


43

section of the FEM is shown in Figure 3.3. The types of elements for the PDMS

membrane and control pad were 3-node linear triangle (CPE3H) and 4-node bilinear

quadrilateral hybrid (CPE4RH) elements. The total number of elements of the PDMS

membrane and PDMS control pad were 611 and 269 respectively. The glass substrate

used 13578 4-node bilinear quadrilateral elements (CPE4R). Mesh sensitivity studies

were conducted to ensure consistent contact force results. As the mesh density was

increased, the total contact force varied less than < 1% (Table 3.2).

Table 3.2: Mesh density sensitivity. The aspect ratios of elements in the compression pad and glass substrate remain constant as mesh density is increased. (Adapted from [51])

Element Length (μm) 10 5 3.3 2.5 Total Contact Force (mN) 6.02 6.06 6.04 6.08

Figure 3.3: (Left) The assembled AIM device consists of a PDMS control layer, membrane, and glass slide. (Right) A subsection of the finite element model depicts variable mesh sizes for the compression pad, PDMS membrane, and glass substrate. Scale bar = 100μm

3.4.2 PDMS Material Parameters and Constitutive Equations

The material properties for the structures in the FEM were derived from both

experimental measurements and theoretical values. The glass substrate was modeled as

linear elastic using properties from the literature [109]. Given the nonlinear behavior of


44

PDMS, a hyperelastic Mooney-Rivlin model was chosen to model its response. The

stress-strain relationship for a hyperelastic material is

𝝈 = 2𝐽−1 �𝐼3𝜕𝑊𝜕𝐼3

𝑰 + �𝜕𝑊𝜕𝐼1

+ 𝐼1𝜕𝑊𝜕𝐼2�𝑩 − 𝜕𝑊

𝜕𝐼2𝑩2� (3.1)

where σ is the Cauchy stress, W is the strain energy function, B is the left Cauchy-Green

deformation tensor, J is the volume ratio, and I1, I2, and I3 are first, second, and third

invariants of B, respectively.

For a Mooney-Rivlin material, the strain energy function, W, is

𝑊 = 𝐶1(𝐼1 − 3) + 𝐶2(𝐼2 − 3) (3.2)

where C1 and C2 are material constants. The material constants were determined to be

C1 = 254kPa and C2 = 146kPa by fitting the model to tension data from the literature and

to experimentally obtained compression data of PDMS samples using a Dynamic

Mechanical Analyzer (DMA) [110] (Figure 3.4). A complete data summary for this

provided in Appendix A.

Figure 3.4: Mooney-Rivlin fit of experimentally obtained uniaxial data for Sylgard 184. (Adapted from [51])


45

3.4.3 Boundary Conditions for Finite Element Model

The boundary conditions for the FEM were as follows. The base of the glass substrate

was fixed, and the surfaces in contact between the glass substrate, the PDMS membrane,

and the PDMS control pad were tied to restrict relative motion between the surfaces. A

hard contact constraint was defined between the PDMS compression pad and the surface

of the glass, which allowed for frictionless sliding between the PDMS membrane and the

glass substrate. Pressure loads were applied uniformly across the top surface of the

PDMS membrane within the control pad area in a single static ramp input.

For each applied pressure load, the contact force between the PDMS compression pad

and glass substrate was computed. The contact force is defined as the total force applied

by the compression pad to the glass substrate. For all experiments conducted in this

study, the applied pressure load was large enough to ensure that the compression pad

fully contacted the glass substrate. The contact pressure was computed by dividing the

total force by the contact surface area of 0.054mm2. This contact pressure provides a

quantifiable measure of the applied load to the axon.

3.5 Platform Optimization

3.5.1 Design Characterization

As the membrane thickness can fluctuate from one device batch to another, an

experimental and computational approach was taken to normalize device performance.

Optical imaging was utilized to measure the membrane thickness and provide data for


46

computational modeling as seen by Figure 3.5. Both the experimental and FEM-

generated cross sections of a representative device under varying pressure loads are

shown.

The AIM device geometry was optimized to prevent non-specific pinching at the

microchannel interface during compression pad deflection under normal operating

pressures (Figure 3.6). Longitudinal (side) image reconstructions demonstrate flat

compression pad profiles during pad deflection. Input pressure was normalized to the

pressure required for the compression pad to fully contact the glass substrate, and is

referred to as percent deflected.

Figure 3.5: A) AIM devices were filled with Fluorescein isothiocyanate (FITC) dye and were imaged under confocal microscopy to characterize compression pad deflection. (B, C) Representative image stacks demonstrated the near linear relationship of the normalized input to compression pad deflection (D) and close correlation to FEM modeling. For the same normalized input as the device images in (B), the corresponding FE results are shown. As membrane thickness between device layers can vary slightly between batches, FEMs were developed to quantitatively assess the relationship between input pressure, membrane thickness, and contact pressure at the glass substrate, an estimate of axonal injury. [51]


47

Figure 3.6: (A) A 3-D schematic of the AIM platform under pressure application and microchannel deflection. (B) The same AIM device shown in Figure 3.5 was imaged at the microchannel interface to determine the extent of microchannel deflection. Under loads greater than that required to bring the compression pad in contact with the glass substrate (> 68.5kPa), a > 3μm gap could be seen. This was sufficient to allow unperturbed axon outgrowth (diameter < 2μm) before, during, and after injury. [51]

3.5.2 Applied Load

For each fabricated injury device, our objective was to quantify the contact pressure

between the compression pad and the system of the axon and glass substrate. The applied

contact pressure is dependent on the input fluidic pressure from the control network and


48

the thickness of the PDMS membrane. A parametric finite element analysis was

conducted, and a linear relationship was determined between the contact pressure, input

pressure, and the PDMS membrane thickness using a Trust-region algorithm from

MATLAB v7.9.0.529 (R2009b) (The MathWorks Inc., Natick, MA, USA) having 95%

confidence bounds:

𝑓(𝑥,𝑦) = 𝑝00 + 𝑝10𝑥 + 𝑝01𝑦 (3)

where f is the contact pressure (kPa), x is the input pressure (kPa), y is the membrane

thickness (mm), and pij are the fitting coefficients (Table 3.3). This relationship was used

to tune the input pressures to achieve desired magnitudes of applied loading for each

AIM platform.

Table 3.3: Coefficients for Equation 3.3 relating contact pressure, input pressure, and membrane thickness. (Adapted from [51])

Coefficient Estimated Lower Limit for a 95% Confidence Bound

Estimated Upper Limit for a 95% Confidence Bound

p00 147.20 138.80 155.60 p10 4.06 4.02 4.10 p01 -6.47 -6.63 -6.31

3.6 Validation of AIM Finite Element Model

To validate the FEM solution, the computational results were compared to

experimental measurements of compression pad displacement and contact pressure for a

specified input pressure and device geometry.


49

3.6.1 Validation of Input Pressure-Displacement Relationship

A finite element model was created with the same geometrical parameters as the

experimental device. The deflection of the compression pad was measured from confocal

images of the experimental device for a specified input pressure. The same input

pressure was applied to the FE model, and the resulting deflection of the compression pad

was computed. The computationally determined compression pad deflections showed

good agreement with the experimentally measured deflection, demonstrating the accuracy

of the finite element solution up to the point of contact between the PDMS compression

pad and the glass base (Figure 3.5). A graph showing the applied input pressure vs.

percent deflection for a membrane thickness of 55mm is shown in Figure 3.7.

Figure 3.7: Percentage of compression pad deflection as a function of input pressure for a membrane thickness of 55μm.


50

3.6.2 Validation of Contact Pressure Predictions

In order to accept the contact pressure outputs from the AIM finite element model, a

series of experiments integrating instrumentation directly into the testing platform were

necessary.

The AIM device was cut into two symmetric halves and bonded to an acrylic base,

instead of the glass bottom dish, having a 200μm gap cut out (Figure 3.8). A 40μm wide

mounting pillar attached to a load cell (Transducer Techniques, Temecula, CA, USA)

was visually aligned with the compression pad of the AIM device (Figures 3.9-3.10).

Using known input fluidic pressures, contact pressure was measured between the PDMS

compression pad and the transducer tip.

Figure 3.8: Symmetric half of AIM platform mounted to an acrylic base. Scale bar = 200μm.


51

Figure 3.9: Experimental setup for load cell measurements of contact pressure applied by AIM platform. (Left) Multiaxial translation stage for alignment with the compression pad of the AIM platform. (Middle) Sensing post extension from load cell. (Right) Machined 40μm wide post at the tip of the load cell sensing post extension.

Figure 3.10: Alignment of PDMS compression pad with sensing post from load cell. Scale bars = 200μm.

Experimentally measured contact pressure for specified geometry and input fluidic

pressures were compared against finite element model predictions for contact pressure


52

(Figure 3.11). From the data it appears the experimental measures are approximately

equal to those predicted by the finite element model. Limitations in the experiment such

as out-of-plane bending could prevent the compression pad from uniformly compressing

on the load cell. The out-of-plane bending was caused by cutting the AIM symmetrically

to allow for visual alignment of the compression pad with the load cell and would

account for the lower contact force measurements at the higher input pressures. The 2D

plane strain finite element model assumption would, in contrast, reflect a uniform

displacement of the compression pad onto the contact surface.

Figure 3.11: Contact force measurement comparison between FE model prediction and load cell experiments for a specified geometry across a range of pressure. The FE model appears to over predict the contact force at higher input pressures. This is attributed to out-of-plane bending caused by symmetrically cutting the AIM to allow for visual alignment of the PDMS compression pad with the load cell sensing post extension.


53

3.7 Thresholds for TAI Response

3.7.1 Classification of Axonal Response

Axonal response to applied focal compression was binned into one of three potential

outcomes per experimental condition. Individual axons and growth cones were examined

frame-by-frame and classified as have a healthy (continued growth), degenerative (TAI),

or regrowth response. Examples of these classifications over the time course of several

hours following load application (time = 0min) are shown in Figure 3.12. Due to the fine

temporal resolution (~1.5mins) between acquired images, axon morphology could be

precisely tracked to correctly bin each individual axonal response (Figure 3.13).

3.7.2 Contact Pressure and Axonal Response

Using the established relationship between geometric parameters of the device and

input fluidic pressures, the applied contact pressures were tracked with each axonal

response. Overall, the percentage of degenerating axons rose as the injury level increased

until ~95kPa. However, beyond this threshold, a significant fraction of injured axons

began to regrow after injury (~46%; Figure 3.13).

For lower applied pressures (< 55kPa), the vast majority of axons remained healthy

and continued to grow even after focal compression (Figure 3.12A; Figure 3.13 Left). As

the applied pressure was increased (55–95 kPa), more axons began to degenerate, as seen

by a combination of nodal axonal swellings and thinning of the axon membrane [111].

Both the proximal and distal segments underwent axonal swelling and rapid degeneration


54

(Figure 3.12B; Figure 3.13 Middle). However, above 95kPa, complete transection, or

rapid severing of the axon was seen in all cases (Figure 3.12C).

Beyond this injury threshold, approximately 46% of axons were able to

spontaneously regrow as evidenced by the reformation of the growth cone after an initial

presence of dystrophic axonal end-bulbs. Axons that eventually regrew initially retracted,

stuttered, paused, and then within 1 to 6 h post injury, were able to reform a growth cone

and continue to extend past the point of injury (Figure 3.12C; Figure 3.1 Right).

Figure 3.12: Representative images of axons that continued to grow, degenerate, or regrow as a function of injury level. (A) Under mild injuries (~25kPa), axons generally continued to grow from left to right as evidenced by progressing growth cones (red triangles). (B) At medium levels of injury (~68kPa), more axons began to undergo degeneration as shown by axoplasm disruption and nodal swellings. (C) Under severe compression (~192kPa), which led to rapid transection, a fraction of axons were able to regrow. These axons were often seen retracting, stuttering, and pausing prior to reformation of the growth cone. Axons (green) were false colored to enhance contrast and allow clear visualization. Scale bar 20μm. [51]


55

Figu

re 3

.13:

I

ndiv

idua

l ax

ons

wer

e bi

nned

int

o on

e of

thr

ee f

ollo

win

g ca

tego

ries:

con

tinue

d gr

owth

, de

gene

ratin

g, o

r reg

row

ing

and

sepa

rate

d in

to ra

nges

of i

njur

ies:

Mild

(< 5

5 kP

a), M

ediu

m (5

5–95

kPa

), an

d Se

vere

(> 9

5 kP

a). Q

uant

ifica

tion

of c

ontro

l (un

inju

red)

axo

ns w

as a

lso

done

to d

eter

min

e ba

selin

e le

vels

of

cont

inue

d gr

owth

, de

gene

ratio

n, a

nd r

egro

wth

. Al

l ex

perim

enta

l co

nditi

ons

wer

e co

mpl

eted

in

tripl

icat

e.

Stat

istic

al a

naly

sis

was

per

form

ed b

y Tu

key

pairw

ise

1-w

ay A

NO

VA:

** =

p-v

alue

< 0

.01,

***

= p

-val

ue <

0.

001.

Err

or b

ars

on g

raph

s co

rresp

ond

to s

tand

ard

erro

rs. [

51]


56

3.7.3 Confirmation of Regrowth Response

To further verify the presence of regrowing axons, low density primary hippocampal

neurons were transfected with a tau fusion protein (red), a microtubule-associated protein

that specifically localizes to axons as opposed to dendrites. Time-lapse imaging in the

setting of severe compression injury (235kPa) confirmed axonal regrowth following

transection (Figure 3.14).

Figure 3.14: A tau-labeled (microtubule marker; red) axon was subjected to severe (235kPa) compression injury and images were collected every 30mins for 8hr post injury. Immediately after injury, the distal segment of the transected axon underwent classic axonal degeneration as evident by nodal swellings, while the axon tip (white arrows) first retracted (~30mins), then began to reform a growth cone (~1hr 30mins). After reformation, the axon was able to extend past the site of injury. Dotted white lines demarcate the region of the compression pad. Scale bar 25μm. [51]


57

Axon growth rates were quantified for regrowing and control (uninjured) axons by

measuring axon displacement at three time intervals and averaging the calculated rates.

Only individual axons that were not growing atop other axons were used for

quantification, and therefore only represent a subset of the overall regrowing population.

Regrowing axons extended approximately 40% faster (22.0 ± 5.8mm/hr), on average,

than uninjured axons (15.8 ± 4.0mm/hr) within the 12–16hr imaging window (Figure

3.15). The literature suggests local protein availability, due to accumulation in the nodal

blebs and the consequent degradation of the axon, is increased for axons undergoing

regrowth as a response to axotomy [112]. The increased protein availability may increase

the rate at which the growth cones progress for axons undergoing regrowth.

Figure 3.15: Growth rates for Uninjured (Control) axons and those with a Regrowth response to focal compression. Growth rates increased by 40% for Regrowth by comparison to Control axons. *p-value < 0.05, unpaired 1-way Welch’s t-test. Error bars on graphs correspond to standard errors. (Adapted from [51])

The degenerating response corridor of neural axonal to focal compression using the

AIM platform highlighted a major avenue for further research in understanding the

mechanics of TAI. Currently it is difficult to obtain insight into the mechanisms


58

occurring within live cells during mechanical loading because this complex environment

is dynamic and evolving. This is a particular challenge from a subcellular mechanics

perspective, where temporal and spatial information on the evolving cytoskeletal

structures is required under load.

This chapter outlined work done investigating the cellular level response to focal

compression. The remaining chapters will further investigate the degenerating response

corridor obtained from this research, but will move to smaller, subcellular, length scales.

This work will also investigate the temporal aspects of TAI by examining changes to the

cytoskeletal structure while under and immediately following loading.

59

Chapter 4

Visualization of Cytoskeletal Deformation

4.1 Introduction

At the subcellular level, mechanical loading of the CNS leads to changes in the

cytoskeleton within neural axons. To develop an understanding of the response for the

cytoskeleton to complex general loading, it is necessary to first understand the

cytoskeletal response under simple loading conditions. The AIM platform described in

Chapter 3 provides a controlled method for applying one loading condition in the form of

focal compression. As discussed in Chapter 2, the cytoskeletal constituents governing the

structural integrity of the cell include microtubules and neurofilaments. These

constituents and their response to focal compression will be explored in this chapter.

Quantification and visualization of 3D live subcellular cytoskeletal populations is

demonstrated prior to, during, and following mechanical loading of axons. This

methodology allows, for the first time, a continuous and quantitative 3D spatial and

CHAPTER 4. VISUALIZATION OF CYTOSKELETAL DEFORMATION

60

temporal visualization of evolving cytoskeletal substructure in situ and under load, thus

dramatically improving the understanding of in-vitro cellular mechanics. The majority of

the work presented in Chapter 4 is from a publication that is, at present, under review

[113].

4.2 In-Vitro and In Situ Cytoskeletal Deformation

The AIM platform described in Chapter 3 was utilized to apply the same focal

compression loads used previously, and in this chapter we will focus exclusively on the

degenerative loading corridor (Figure 4.1). The aim of the present chapter will now

move from the cellular response explored in Chapter 3 to the subcellular length scales of

the cytoskeleton. This will improve the understanding of cytoskeletal evolution under

TAI conditions for microtubules and neurofilaments.

4.2.1 Cell Isolation, Cytoskeleton and Membrane Labeling

Primary hippocampal neurons were isolated from E17 Sprague Dawley rat pups

(Charles River, Wilmington, MA, USA). Dissociated neurons were nucleofected

(Amaxa, Gaithersburg, MD, USA) and plated in AIM platforms as described in Chapter

3. For cytoskeletal labeling, pCMV-AC-GFP plasmids with a C-terminal TurboGFP

(Origene, Rockville, MD, USA), encoding a neurofilament-GFP fusion or a microtubule-

GFP fusion gene with a cytomegalovirus (CMV) promoter, were nucleofected.

The nucleofection protocol is as follows:


61

1. Centrifuge the required cell number (4-5x106 cells) at 80xg for 5min.

2. Remove excess liquid and resuspend the cell pallet carefully in 100µl of

nucleofector solution. Avoid leaving the cells in rat nucleofector solution for

more than 15min.

3. Slowly combine 100µl of cell suspension with 4µg of the cytoskeletal DNA label.

4. Transfer the cell/DNA suspension into a certified cuvette. Ensure the sample

covers the bottom of the cuvette without any air bubbles.

5. Select the appropriate nucleofector program for neural cells (Program 003).

6. Insert the cuvette into the loading port of the nucleofector and start the

nucleofection (you should see the cuvette rotate and hear a buzz).

7. Remove the cuvette.

8. Add 500µl of pre-equilibrated neural culture medium (neural basal 4+).

Neurofilaments were labeled with a combination of NF-light, NF-medium, and NF-

heavy protein-GFP fusions, while microtubules were labeled via microtubule-associated

tau protein-GFP fusion. Efficiency of labeling was greater than 50%.

Figure 4.2 outlines the process flow where nucleofected neurons were loaded into the

somal compartment at a density of 25x106 cells/mL. After 6–8 days in culture, individual

axons could be seen extending into the middle and distal chamber of the device, allowing

tracking of individual processes for subsequent experiments (Figure 4.1D). Media was

added every 3-4 days to maintain neuronal viability. Prior to experimentation, CMPTx

red cell tracker was added to label the membranes of neural axons as per the

manufacturer’s instructions (Gibco Life Technologies; Grand Island, NY, USA).


62

Figu

re 4

.1:

Th

e AI

M t

estin

g pl

atfo

rm i

sola

tes

indi

vidu

al n

eura

l ax

ons

and

prov

ides

a c

ontro

lled

mic

roflu

idic

en

viro

nmen

t for

app

lyin

g fo

cal c

ompr

essi

on.

A) Il

lust

rativ

e ex

ampl

e of

an

AIM

pla

tform

insi

de a

gla

ss w

ell.

B) I

nset

of

AIM

pla

tform

dep

ictin

g m

icro

chan

nels

con

nect

ing

to i

njur

y co

mpa

rtmen

t, w

here

loa

ding

is

appl

ied

thro

ugh

a co

mpr

essi

on p

ad.

C) I

deal

izat

ion

of a

xon

load

ing

envi

ronm

ent a

nd o

rient

atio

n w

ithin

the

AIM

. D

) Ins

et o

f sin

gle

axon

gr

owth

thro

ugh

mic

roch

anne

l int

o te

stin

g en

viro

nmen

t. E

) Ins

et o

f ind

ivid

ual a

xon

unde

r co

mpr

essi

on p

ad i)

prio

r to,

ii)

dur

ing,

and

iii)

imm

edia

tely

follo

win

g ap

plie

d lo

ad [5

1].

Scal

e ba

r = 1

0μm

for D

. Sca

le b

ars

= 20

μm fo

r E. [

113]


63

Figure 4.2: Process flow chart of cell preparation, labeling, and incubation for confocal imaging. Cells were obtained from E17 rat pups. Dissociated neurons were electroporated and the nucleofected cells were plated in AIM platforms. For cytoskeletal labeling, pCMV-AC-GFP plasmids with a C-terminal TurboGFP, encoding a neurofilament-GFP fusion or a microtubule-GFP fusion gene with a cytomegalovirus (CMV) promoter, were used. Following an incubation period of 7-10 day, cell membranes were labeling with CMPTx red cell tracker and taken to experimentation.

4.2.2 Immunohistochemical Labeling

To ensure cytoskeletal constituents were labeled correctly, a second approach using

immunohistochemical labeling was implemented to co-label microtubules and

neurofilaments. By using an immunohistochemical approach we could label the

microtubules and neurofilaments using a different method than was originally

implemented. We did not use immunohistochemical labeling in the live cell imaging

because it requires fixing the cells.


64

The co-labeling protocol is as follows. Nucleofected cells were washed with

phosphate-buffered saline (PBS) and fixed for 20mins at room temperature with 4%

paraformaldehyde (PFA). Cells were washed in PBS and incubated in blocking solution

containing 0.25% Triton-X and 5% normal donkey serum for 1hr. Primary antibodies

included chicken NF-medium (1:1000, Aves labs Inc; Tigard, OR, USA) and mouse

Beta-III tubulin (1:1000, Promega; Madison, WI, USA) diluted in blocking solution.

Samples were stored overnight at 4°C. Cultures were washed three times in PBS and

incubated with Alexa Fluor555-conjugated donkey anti-chicken and donkey anti-mouse

(1:250, Invitrogen) separately for 2hrs at room temperature. Fixed samples were

visualized under fluorescence and colocalization data was obtained.

4.2.3 Experimental Procedures

Prior to imaging, four independent tubing lines (O.D. = 1.52mm, I.D. = 0.51mm;

Cole Parmer; IL) connecting to gauge #21 blunt needles (McMaster-Carr; Santa Fe

Springs, CA, USA) were connected to each of the four control fluidic ports of the AIM.

The control gas (CO2) pressure was delimited with a Proportion Air electronic regulator

(Equilibar; Fletcher, NC, USA) to apply the desired level of pressure between the glass

substrate and injury pad. The desired pressure level was chosen using a validated finite

element model, discussed in Chapter 3, where geometric parameters for each platform

provide the input fluidic pressure required to obtain the injury response observed in TAI

[51].


65

AIM devices were moved to a pre-warmed (37ºC) live-cell AxioObserver.Z1

confocal microscope with an enhanced 3i Marianis/Yokogawa spinning disk and having a

stage-cover removed to allow easy access for the control lines (Zeiss; Oberkochen,

Germany) (Figure 4.3). Individual axons located underneath the injury pad were

identified and their coordinates were saved within Slidebook imaging software

(Intelligent Imaging Innovations, Inc; Denver, CO, USA).

Figure 4.3: Confocal imaging setup for focal compression experiments. A stage-cover provided a temperature controlled environment for imaging with access for the pressure control lines for regulating microfluidic compression pads of the AIM.

For each experiment, pressure application was performed by first adjusting the

electronic pressure regulator, then turning the stopcock ninety degrees to allow pressure

translation to the injury pad (<1s), holding the pressure (20-35s), and finally relieving the

pressure by turning the stopcock back to the start position. Images were taken

immediately before, during, and after loading. In all experiments, the pressure applied to


66

the control network resulted in intimate contact between the injury pad and glass

substrate/axons, which was seen by a loss and subsequent recovery of contrast for the

compression pad feature (Figure 4.1E).

Sham controls for confocal imaging were prepared and data was recorded in the same

manner as those undergoing dynamic compression, though no loading was applied.

4.3 Visualizing Structure

Confocal imaging and Z-stack processing were conducted at the Johns Hopkins

Medical Institutes Microscope Facility.

4.3.1 Confocal Imaging Setup

Fluorescence, bright-field, and phase-contrast images were captured at 63x (0.6 NA;

DIC, c488ht, c562mp) or 100x (1.4 NA; DIC, c488ht, c562mp) magnification with a

Zeiss live-cell inverted microscope (AxioObserver; Zeiss, Oberkochen, Germany) using

Zeiss Axiovision software. An idealization of the initial unloaded and loaded axon

subvolume, as well as examples of the confocal imaging stacks for each of the load

states, are shown in Figure 4.4. In Figure 4.4E-F the 488nm (green) and 562nm (red)

fluorecence reveal the expression of cytoskeletal constituents and cell membrane

respectively. Imaging plane resolution for the confocal microscope is shown in Table 4.1.


67

Figure 4.4: Confocal imaging subvolume prior to and during focal compression. A-B) Illustrative example of the subvolume of axon prior to (A) and under loading (B). Here the axon membrane is shown in red and the cytoskeletal components of microtubules and neurofilaments are shown in green and blue respectively. C-D) Volumetric data set for single axons prior to (C) and during focal compression (D). E-F) Fluorescent images take using 488nm (green) and 562nm (red) wavelengths. 562nm intensity represents axons membrane and the 488nm represents the cytoskeletal constituent (microtubules or neurofilaments) of interest. Scale bars = 2μm.

Table 4.1: Confocal imaging plane resolution by magnification Magnification X, Y-Plane Z-Plane 63X 260nm/pixel 270nm/pixel 100X 163.8nm/pixel 340nm/pixel


68

4.3.2 Image Acquisition

Imaging data was taken at known time points prior to, during, and immediately

following load application. When obtaining Z-stacks with fluorescence microscopy,

sample exposure time was limited to less than 200ms to minimize phototoxicity.

Acquired Z-stacks were processed in Imaris (Bitplane AG; Zurich, Switzerland) where a

subvolume of the Z-stack was selected representing the spatial coordinates of the axon

under the compression pad (Figure 4.5). The subvolume was exported into MATLAB

(Mathworks; Natick, MA, USA) for analysis. The MATLAB code written for confocal

imaging analysis is provided in Appendix B.

Figure 4.5: Confocal intensity subvolume selection process. Acquired Z-stacks are processed to create 3D volumetric sets of intensity data to be analyzed. A) Full range 3D confocal intensity data set taken from a single experiment. B) Subset of volumetric data segmented focusing on the volume of axon underneath the compression pad. C) Intensity thresholds are applied to remove background noise. In this example, surfaces are temporarily created to visualize the volume containing the axon membrane (D) and the neurofilaments (E). Scale bar = 10μm for A. Scale bars = 3μm for B-E.


69

4.3.3 Post-Processing of Image Data

For microscopy image acquisition, detected photons were converted to intensity

values at each pixel. Fluorescent microscopy provides data in the form of spatial

coordinates and the associated intensity Φ(x,y,z,t) at these coordinates. Φ can be used to

determine the local concentration of fluorophores and is correlated to the density of

specific cytoskeletal constituents’ protein fluorescence at that given point in time. The

fluorescence represents the labeled parts of the cell where the 562nm wavelength (red)

indicates the axon membrane and 488nm wavelength (green) indicates the cytoskeletal

constituent of interest.

From a continuum mechanics framework, we understand the experiment through an

Eulerian description. For the kinematics applied, we take the spatial (or current)

description of Φ and characterize the changes with respect to spatial coordinates x1, x2,

x3 (or x,y,z as given above), and time t as Φ(x,t). This approach centers attention to a

point in space where we observe what happens at that point in space as time changes. The

data sets we obtain through confocal imaging also provide spatial and temporal

information on the axon prior to load (at time t0) which is considered the initial

configuration (Ω0) of the axon. This configuration is linked to the same spatial

configuration (Ω) of the axon, at time t, during or following load (Figure 4.6A-B). Ω0

and Ω are connected by the intensity Φ(x,t) and the derivatives of Φ with respect to time

and space, �̇� and ∇𝚽 respectfully [114].


70

Figure 4.6: Illustrative and confocal examples of the subvolume of axon, with labeled cytoskeletal components, under the compression pad. A) The spatial and temporal information from initial unloaded (Ω0) and under load (Ω) states are connected through the intensity, Φ, which maps the spatial description from Ω0 to Ω. Here the axon membrane is shown in red and the cytoskeletal components of microtubules and neurofilaments are shown in green and blue respectively. B) Volumetric Z-stack of initial unloaded (Ω0) and under load (Ω) states. Scale bars = 10μm.

Cytoskeletal intensity, Φ, was integrated over the axons’ cross-section and plotted

along the length of the axon under compression as ΦA at t=1min.

𝚽𝑨(𝑦, 𝑡) = 1𝐴𝑥𝑧

∫𝚽(x, y, z, t)𝑑𝑥 𝑑𝑧 (4.1)


71

Here Axz represents the measured cross sectional area of the axon membrane in the

XZ-plane. Φ per unit area, ΦA, was calculated for the unloaded (time t0) and loaded

(time t) states and comparisons were made for sham controls.

To compare intensity data across multiple axons, cumulative intensity per unit

subvolume was calculated as 𝚽� where,

𝚽� (𝑦, 𝑡) = 1𝐴𝑥𝑧𝐿𝑦

∫𝚽(x, z, t)|𝑦 𝑑𝑥 𝑑𝑧 = ∑ 𝚽𝑛𝑖=1

𝑈𝑛𝑖𝑡 𝑆𝑢𝑏𝑣𝑜𝑙𝑢𝑚𝑒∝ 𝜌𝑐𝑦𝑡𝑜𝑠𝑘𝑒𝑙𝑒𝑡𝑜𝑛 (4.2)

Here unit subvolume represents the volume of the axon under the compression pad

and was calculated using Axz and the slice thickness of the y-axis (Ly), along the length

of the axon. 𝚽� is representative of the local concentration of fluorophores at a given

point in time within the axon membrane and scales with the density of a specified

cytoskeletal constituent’s protein fluorescence (𝜌𝑐𝑦𝑡𝑜𝑠𝑘𝑒𝑙𝑒𝑡𝑜𝑛). 𝚽� was averaged over the

entire subvolume and the percent change in 𝚽� between time t0 and time t was calculated

for t<1min and t=5min.

4.4 Cytoskeletal Population Response to TAI

TAI (in response to applied mechanical load) results in nodal bleb formation and

systematic degeneration of cytoskeletal structures along the axon. Using a controlled

mechanical environment, relevant applied loading rates and loads, and real time imaging,

it was observed that neurofilament expression decreases markedly upon initial loading,

while microtubule expression was not significantly changed for the same time period.

These changes persisted within the first 5 minutes of loading. The findings suggest that


72

immediate localized compaction and alterations to neurofilaments may serve as a trigger

for further secondary damage to the axon, representing a key insight into the temporal

aspects of cytoskeletal degeneration at the component level.

4.4.1 Changes in Axonal Cytoskeleton with Mechanical Loading

In the AIM platform, individual axons passed through microchannels into a loading

compartment featuring the compression pad (Figure 4.1A-E) [51]. In Chapter 3 it was

shown that only axons, and not dendrites or other neuronal components, pass through the

microchannels into the compression compartment by virtue of optimization of the

geometry of the platform (Figure 4.1B, D).

Time lapse imaging of the axons prior to, during, and immediately following

controlled mechanical loading confirmed contact between compression pad and glass

base and that the axons remained attached to the glass base (Figure 4.1E). Control input

fluidic pressures to achieve contact were estimated using the validated FEM described in

Chapter 3 and were confirmed visually for all experiments (Figure 4.1E ii) [51]. Live

imaging was used to estimate the time scale for dynamic compression to be on the order

of 1.5ms for axons with a range of diameters from 0.25-2.0μm as observed through

confocal data. Under dynamic compression, axons were consistently found to display

multiple regions of nodal blebs where the compression occurred within 1min of load

application, indicating that the injury response of TAI was achieved (Figure 4.1E iii)

[51].


73

4.4.2 Colocation & Effects of Fluorescent Labeling

A secondary method for labeling proteins, known as immunohistochemistry (Section

4.2.2), was utilized to confirm fluorescent labeling of microtubules and neurofilaments

using the transfection process described in section 4.2.1 was accomplished. By labeling

the microtubules or neurofilaments using both methods, the fluorescent should be

expressed in the same spatial domain. Our results indicated (i) the transfected and the

immunohistochemically labeled constituents result in detectable fluorescence expression

within axons (Fig. 4.7A, D), (ii) the transfected neurofilaments and microtubules

colocalize with the immunohistochemically labeled neurofilaments and microtubules for

the same cells (Figure 4.7B, E), (iii) nontransfected cells appear to have the same

expression pattern as transfected cells that have been immunohistochemically labeled

(Figure 4.7C, F). This set of results confirms the transfection process used in our

experiments, appropriately labeled the microtubules and neurofilament of specified

axons. Finally, while the expression of both cytoskeletal components was observed

throughout the axon, areas of increased fluorescence intensity, likely representing regions

of accumulation of cytoskeletal elements, could be observed. This implies an increased

density of the specified cytoskeletal constituent wherever the increased fluorescent

intensity is observed.


74

Figure 4.7: Confocal imaging confirms transfected cytoskeletal constituents were correctly labeled using a secondary labeling process celled immunohistochemistry. A-B) For microtubules, transfected tau protein (green) and immunohistochemical labeled Beta-III tubulin (red) have similar expression patterns and, using a triaxial planar view, are observed to colocalize in the same spatial domain (yellow). C) Nontransfected, immunohistochemically labeled microtubules appear to exhibit a similar expression pattern as the transfected microtubules with immunohistochemical labeling. D-E) For neurofilaments, transfected NF-Medium (green) and immunohistochemical labeled NF-Medium (red) have the same expression pattern, and using a triaxial planar view, are observed to colocalize in the same spatial domain (yellow). F) Nontransfected immunohistochemically labeled neurofilaments appear to exhibit a similar expression pattern as transfected neurofilaments with immunohistochemical labeling. Scale bars = 2μm for A-F. [113]

The question of how observed changes in intensity during and immediately following

loading relate to changes in cytoskeletal structure is of primary concern. The use of a

CMV promoter for inserting GFP-tagged proteins is known to result in overexpression of

fusion proteins and therefore the measured absolute values of intensity are likely larger


75

than in untagged systems. Given these concerns, the percent change in 𝚽� (rather than

absolute values) is used to ascertain changes in cytoskeletal populations under load.

There is also concern that the overexpression of these proteins may potentially alter the

relevant aspects of the cytoskeleton. This alteration of spatial and transport properties of

the cytoskeleton may be important, but the question is difficult to answer because of

limitations in the techniques used for this work. The literature and communications with

the manufacturer have indicated this can occur with some cytoskeletal systems, though it

is typically not observed [115]. To address this, spatial distribution and intensity were

measured for non-transfected axons with immunolabeling and compared against

transfected with immunolabeling (Figure 4.7C, F). These results indicate no observable

difference in spatial distribution and intensity between the two groups.

4.4.3 Evolution of the Spatial Distribution of Microtubules &

Neurofilaments

The ΦA changes for intensity along the length of the axon for sham (Control) and

loaded groups are given in Figures 4.9 - 4.10. The data indicates there is no observable

change in ΦA of the sham control specimens for both microtubule and neurofilaments

between time t0 and time t=1min (Figure 4.8A, 4.9A). This means the microtubule and

neurofilament protein levels do not significantly change for the sham control groups.

Following compression, the change in magnitude of ΦA for microtubules appears

negligible at time t=1min (Figure 4.8B). In contrast, neurofilament expression


76

significantly decreases between time t0 and time t=1min along the length of the

compressed axon volume (Figure 4.9B).

Figure 4.8: Representative examples of intensity profiles (y-axis) for microtubules plotted along the volume of axon underneath the compression pad (x-axis). The two panels represent Control and Loaded cells where a single axon is represented before loading at time t0 (blue) and while under load at time t (red). For the Control population, no load was applied at time t. A) No observable change in ΦA for Control between time t0 and time t states. B) Overall magnitude of ΦA does not change during loading, though its distribution within the subvolume appears to. [113]


77

Figure 4.9: Representative examples of intensity profiles (y-axis) for neurofilaments plotted along the volume of axon underneath the compression pad (x-axis). The two panels represent Control and Loaded cells where a single axon is represented before loading at time t0 (blue) and while under load at time t (red). For the Control population, no load was applied at time t. A) No observable change in ΦA for Control between time t0 and time t states. B) Decreased magnitude of ΦA is measured across the entire subvolume during loading. [113]

A One-way analysis of variance (ANOVA) comparing the percent change in

volumetric intensity measurement of 𝚽� between time t0 and time t was completed using


78

GraphPad Prism 5 (GraphPad Software Inc., La Jolla, CA, USA) and is shown in Figure

4.10. The results indicate no significant change in the microtubule signal at t<1min (3%

+/- 2.9; n.s.) or at t=5min (-6% +/- 2.7; n.s.) whereas percent change in 𝚽� for the

neurofilaments significantly decreases at t<1min (-24% +/- 5.8; p<0.001) and t=5min (-

30% +/- 9.3; p<0.001) (Figure 4.10).

Figure 4.10: Results for One-way ANOVA with Tukey’s multiple comparison test of percent change in 𝚽� for Control axons and for axons at time t<1min and t=5min. The number of samples and the population group are shown along the x-axis and the corresponding percent change in 𝚽� is plotted along the y-axis. No significant observable percent change in 𝚽� was observed between Control and Loaded axons at times t<1min and t=5min for Microtubules. Loaded axons for neurofilaments exhibited a statistical significance decrease of 24% and 30% at t<1min and t=5min respectively compared to Control axons. * p<0.001[113]


79

4.4.4 Microtubule & Neurofilament Response to Load

Understanding the mechanism for the rapid decrease in Φ for neurofilaments

observed in our study is critical. Neurofilament dispersion and compaction following

loading has been observed by previous researchers [47-50, 99, 101]. In one study where

impact acceleration was used, Povlishock et al. found evidence of neurofilament

compaction and sidearm alteration from 5min-24hrs post injury [99]. In a second study,

Okonkwo et al. found alterations of the neurofilament sidearm length following fluid

percussive injury [101]. Both of these studies appear to indicate that while an alteration

in the sidearm length of the neurofilament may occur soon after injury, the sidearms are

still present, though at a reduced length. These alterations in sidearm length represent a

conformational shape change for the protein structure. The decrease in measured Φ

detected may be directly linked to modifications in the neurofilament sidearm protein

folding, resulting in decreased fluorophore expression.

4.4.4.1 Temporal Evolution of Microtubules & Neurofilaments

The live-cell results were compared with those of other studies where fixed neural

cytoskeletal components were measured through TEM and immunoblotting (Figure

4.11A-B). Our findings are in agreement with the long-term trends in microtubule and

neurofilament densities observed by previous studies employing TEM and

immunoblotting methods at known time points following loading (Figure 4.11A-B) [45,

49, 116]. As time between when the load is applied and when quantification occurs is


80

increased, so do the percent decrease in cytoskeletal densities for both cytoskeletal

components.

In Figure 4.11A, our data indicated decreases in microtubule density at 5min;

however these results were not significant (Figure 4.10). The eventual decrease in

microtubule density appears to follow the long term trends established by other studies

where decreases in microtubule density, or the proteins associated with microtubule

expression, have been observed at greater time intervals following deformation (Figure

4.11A) [45, 49]. Interestingly, Maxwell and Graham noted an 86% decrease in

microtubule density at 15mins following loading [45]. This appears to be a greater

decrease in density than was observed by Serbest et al. at 30mins and appears to be an

outlier for Figure 4.11A [45, 49]. We believe this may be due to Maxwell and Graham

using quantification data from the internode (mylinated) region of the axon. Our data is

only from unmyelinated axons and the data from Serbest et al. does not separate out

regions of the axon into myelinated and unmyelinated sections.

Figure 4.11B shows neurofilament density exhibiting an immediate decrease of 24%

in expression during loading and this continues to decrease to 30% within 5min. By

comparison, previous researchers have also observed decreases in the density of

neurofilaments and that the magnitude of these appears to increase as time increases

(Figure 4.11B) [49, 116].

To compare microtubule and neurofilament measurements directly, the time history

of cytoskeletal densities were plotted for each on the same graph (Figure 4.12). While

the slope were not significantly different, neurofilament density is observed as decreasing


81

at a faster rate than the microtubule density following loading (Figure 4.12) [45, 49, 116].

This would indicate dissimilarities in the response of the cytoskeletal constituents.

Figure 4.11: Percent change in cytoskeletal density as a function of time comparing across multiple studies. The time between loading and quantification is plotted along the x-axis and varies from t<1min to t=72hrs. The percent change in cytoskeletal density is plotted along the y-axis. Data from our study taken during loading is shown at times t<1min and t=5min. A) Microtubule density did not exhibit a measurable change between the unloaded and loaded states during initial loading. Changes in density became apparent at the 5min, for our study, and appear to fall in line with other studies which have observed decreases in microtubule density, or the proteins associated with microtubule expression, at greater time intervals following deformation [45, 49]. B) Neurofilament density exhibits an immediate decrease of 24% in expression during loading and continues to decrease to 30% within 5min. Other studies have exhibited decreases in the magnitude of the density change as time increases, a trend we observe in the current study [49, 116]. [113]


82

Figure 4.12: Aggregate data from Figures 4.10A-B comparing percent changes of cytoskeletal density over time following TAI. Neurofilament density (blue square) decreases at a faster rate than the microtubule density (red circle) [45, 49, 116].

Differences between the mechanical response of microtubules and neurofilaments, as

well as their distribution within the axon, lead to variations in cytoskeletal expression

over the short term response of the neural axon. For the short term, <5min time frame,

local intracellular mechanisms may govern the whole cell response to the applied load.

While the cytoskeleton is in a continuously dynamic and evolving state, it is known that

microtubules are the stiffest component of the axonal cytoskeleton and may not

degenerate in response to applied load as readily as less rigid components of the

cytoskeleton like the neurofilaments [117]. Thus the simplest cytoskeletal structures

within the axon may change their conformational shape at a faster rate before observable

changes can be detected for the more rigid constructs (Figure 4.12). Additionally the


83

microtubules are oriented along the axial direction of the axon where the neurofilament

sidearms are extended transversely across the across diameter. This would mean the

sidearms’ orientation is more likely influenced by the load direction than the

microtubules, which are oriented orthogonally to the load direction.

4.4.5 Limitations of Study

In addition to the overexpression observed by usage of the CMV promoter, potential

limitations covering areas of biology, imaging, and image processing exist for the study.

Biological limitations include the usage of rats instead of humans. Further E17 rat

pups are not necessarily representative of the adults. These are standard limitations that

exist in many animal models and are accepted by the research community. Additionally

a major limitation is the use of unmyelinated axons where many axons in the CNS are

myelinated by oligodendrocytes. Unmyelinated axons were used to simplify the loading

by ensuring the myelination could not distribute the applied load over the axon.

Imaging limitations include only one cytoskeletal plane could be taken at a specific

point in time and the plane resolutions of the confocal microscope were larger than the

size of the cytoskeletal constituents we are trying to image. This means that while we

can image the population we cannot observe the individual cytoskeletal constituents at

this point using the confocal imaging setup utilized in this experiment. However the

main point of the experiment was to define the average behavior of a cytoskeletal

population in focal compression and therefore the resolution to observe individual

cytoskeletal constituents is not needed.


84

Post-processing of the confocal Z-stack, by applying threshold limits to remove

background intensity noise, may have removed part of the intensity data. The noise

associated with the intensity was uniformly distributed throughout the entire subvolume

and should not affect the analysis. Additionally large standard deviations exist in the data

set, though this is typical for biological specimens.

4.5 Summary of Results

The confocal data suggests temporal aspects of developing structural damage within

neural axons may need to be considered when attempting to accurately model TAI.

Currently there is a great degree of emphasis on microtubules as the key component to

understanding axonal degeneration. Microtubule breakage and deformation has been

posited as leading to a physical blockade, impeding normal axonal transport and leading

to undulations and further axon degeneration. A conformational shift in the protein

structure of neurofilaments may be a very early event in response to applied loading. The

conformational shift could precede the microtubule disruption and breakdown observed

at the later time stages of TAI induced through compressive loads.

The work presented in Chapter 4 suggests changes in cytoskeletal density can be

captured in situ and under load, leading to improved insight into in-vitro cellular

mechanics. The results indicate neurofilament expression decreases dramatically while

under load and changes in microtubule expression are not readily observed until later

times. This data suggests temporal aspects for cytoskeletal changes of neurofilament and


85

microtubule expression, not previously observed, may be critical in understanding

mechanical failure and degeneration of the cytoskeletal system for axons undergoing

TAI. Chapter 5 qualitatively and quantitatively details the cytoskeletal distributions

within the compressed volumes for axons undergoing TAI and control axons using TEM.

This quantification will link the structural basis of the cytoskeleton underlying the

observed changes in fluorescence during axon loading we observed in this chapter.

86

Chapter 5

TEM Observations of Cytoskeletal

Evolution in CNS Axons

5.1 Introduction

The hallmarks for TAI are nodal bleb formation and systematic degeneration of

cytoskeletal structures along the axon. In Chapter 3 an understanding of the injury

response of neural axons to simple loading conditions (focal compression) was obtained.

In Chapter 4 this understanding was expanded to capture the evolution of specific

cytoskeletal constituents’ populations within neural axons undergoing TAI. In this

chapter, we utilize transmission electron microscopy to quantify these observations at

higher resolution. The observations are quantified through metrics that describe the

spatial distribution of microtubules and neurofilaments within Control and Crushed

axons. This information is combined with data from the literature to improve our

CHAPTER 5. TEM OBSERVATIONS OF CYTOSKELETAL EVOLUTION IN CNS AXONS

87

understanding of the temporal evolution of the cytoskeleton following TAI. The majority

of the work presented in this chapter is from a publication that is, at present, under review

[118].

5.2 Experimental Approach

The AIM platform described in Chapters 3 and 4 was utilized to apply the same focal

compression loads used previously; again focusing exclusively on the degenerative

loading corridor (Figure 3.12B - 3.13). We will now move from the cytoskeletal

population response examined at hundreds of nanometers in the previous chapter, to

analyzing the evolution of individual constituents at length scales on the order of tens of

nanometers.

5.2.1 Cell Culture and Isolation

Primary hippocampal neurons were isolated from E17 Sprague Dawley rat pups

(Charles River, Wilmington, MA, USA) as previously described in Chapter 4. In the

work described in this chapter, the cells were not nucleofected or fluorescently labeled.

Dissociated neurons were loaded into the somal compartment at a density of 25x106

cells/mL. After a period of 6–8 days in culture, axons could be observed extending into

the middle and distal chambers of the AIM device in sparse numbers to allow tracking of

individual processes for subsequent experiments. Media was added every 3 to 4 days to

maintain neuronal viability.


88

5.2.2 Experimental Protocols

As reported in previous chapters, dynamic compression was applied exclusively to

the axons of primary hippocampal neurons. Four independent tubing lines (O.D. =

1.52mm, I.D. = 0.51mm; Cole Parmer; IL) attaching to gauge #21 blunt needles

(McMaster-Carr; Santa Fe Springs, CA, USA) were coupled to each of the four control

fluidic ports of the AIM prior to loading. The control gas (CO2) pressure was

manipulated with a Proportion Air electronic regulator (Equilibar; Fletcher, NC, USA) to

apply the desired level of pressure between the glass substrate and injury pad.

Using the previously described finite element model from Chapter 3, geometric

parameters for each platform provided the input fluidic pressure required to obtain the

injury response observed in TAI. Pressure application was performed by adjusting the

electronic pressure regulator, turning the stopcock ninety degrees to allow pressure

translation to the injury pad (<1s), holding the pressure (<5s), and relieving the pressure

by turning the stopcock back to the start position. Visual confirmation of contact

between the compression pad and glass substrate was obtained for all groups during load

application. Fix was immediately added through access ports following loading and

tubing connectors were removed. Following a period of 15min in fix, AIM platforms

were peeled from the glass bottom petri dishes and additional fix was added to the dish.

Sham controls were prepared and data was recorded in the same manner as those

undergoing dynamic compression, but no loading was applied.


89

5.2.3 Fixation, Labelling, and Embedding for TEM

Different approaches were used for preparing microtubule and neurofilament TEM

specimens. The choice of approach required identifying what each group necessitates in

terms of morphology and labeling. In our study microtubules were relatively easy to

identify and quantify so morphology was of paramount concern. Neurofilaments by

comparison were harder to identify and required immuno electron microscopy with gold

(Au) labeling using 2º NF-medium antibodies to confirm presence within processed TEM

sections. Previous research regarding immunogold labeling using antibody specific

binding has been shown to be effective in detecting the presence of a specified molecule

of interest [119-121].

One inherent limitation of using immunogold labeling is that the observed Au

nanoparticles are approximately 15-30nm from the primary binding sites of the antibody

[122]. However, the spacing of the nanoparticles from their binding sites should be

approximately constant and the number density of the nanoparticles is equal to the

number density of the binding sites. As labeling was of primary importance for the

neurofilaments, the quality of morphology determination was sacrificed in favor of

quantification of labeling in this TEM characterization.

Microtubule TEM samples were fixed with a mixture of 2% paraformaldehyde, 2.5%

glutaraldehyde, 0.1M sodium cacodylate (SC), and 1% sucrose for 1hr. Cells were

washed with a 0.1M SC, 3mM calcium chloride (CaCl2), and 3% sucrose buffer in three

10min rinses and stored in 1% osmium plus 0.8% potassium ferricyanide on ice and in

the dark for 1hr, followed with three 0.1M maleate buffer rinses, 5min each. Cells were


90

stained with 2% aqueous uranyl acetate (0.22μm filtered, 1hr, dark) in 0.1M maleate

buffer. Cells were dehydrated in a graded series of ethanol washes: 30%, 50%, 70%,

90% and 100% with 10min between washes.

Neurofilament TEM samples were fixed with a mixture of 6% paraformaldehyde,

0.5M SC, and 1.0M CaCl2 for 1hr, washed with a 0.1M SC buffer in three 10min rinses

and blocked with 1% bovine serum albumin (BSA) at 4°C for 1hr. BSA was removed

and a mixture of the primary NF-medium antibody (Enzo Life Sciences, Farmingdale,

NY, USA) and 0.02% saponin in 0.1M SC buffer (1:100) was added, and cells were

incubated overnight at 4ºC. Cells were incubated back to room temperature and washed

with 0.1M SC buffer six times in 10min rinses. A secondary antibody mixture of 6nm

Au particles (Jackson Immuno Research Laboratories, Inc., West Grove, PA, USA) and

0.01% saponin in 0.1M SC buffer (1:200) was added, and the cells were incubated at 4ºC

for 4hrs. Cells were incubated back to room temperature for 10min, rinsed with 0.1M SC

buffer six times with 10min between rinses, and then 1% glutaraldehyde was added in

0.1M SC for 1hr. Cells were rinsed three times with ddH2O, with 10mins between

rinses. 0.5% osmium tetraoxide in 0.1M SC buffer was added and cells were placed on

ice and in the dark for 30min. Cells were rinsed with ddH2O 3 times, with 10mins

between rinses, followed by dehydration in a graded series of ethanol washes: 30%, 50%,

70%, 90% and 100% with 10min between washes.

Following the graded dehydration by ethanol, cells for both microtubule and

neurofilament samples were embedded in a resin of Epon using DMP-30 as a catalyst and

incubated overnight at room temperature. Cells go through a series of Epon + DMP-30


91

changes and were subjected to vacuum in an attempt to thoroughly embed cells in the

resin. The cells were incubated overnight at 65°C. The cells were separated from the

dishes through a process of thermal shocking, by placing samples into liquid N2, and

embedded specimens were retrieved for post-embedding processing.

Samples were cut down to specified grids and went through a triple staining process

with 1% tannic acid (aqueous) (Mallinckrodt Pharmaceuticals, St. Louis, MO, USA),

filtered using 0.22μm filter, for 10min by flotation using formvar coated slot grids (Ted

Pella, Inc., Redding, CA, USA) before being rinsed with ddH2O for one minute. Grids

were then stained with aqueous 2% uranyl acetate (Polysciences, Inc., Warrington, PA,

USA) for 20min and rinsed with ddH2O. Finally grids were stained for 1min on 0.04%

lead citrate (aqueous-filtered) and rinsed with ddH2O. Longitudinal and transverse TEM

slices (60-80nm thickness) were obtained using a microtome diamond blade and serial

sections were taken to address concerns of embedding and processing 3D embedded

substructures [123].

5.2.4 Transmission Electron Microscopy

Imaging using transmission electron microscopy was conducted at the Johns Hopkins

Medical Institutes Microscope Facility.


92

5.2.4.1 Imaging Setup

TEM imaging was conducted using a Philips/FEI BioTwin CM120 Transmission

Electron Microscope (Figure 5.1). The Philips has two high-resolution cooled digital

cameras, a Gatan Orius (4 Mpixel, 8-bit) for low noise and an AMT XR80 (8Mpixel, 16-

bit). The Philips has a 20-120 kV operating voltage and features low dose levels (for

minimizing beam damage), a liquid nitrogen cooled anti-contamination device, and a

goniometer stage.

Figure 5.1: Transmission electron microscope used for imaging cytoskeletal constituents of neural axons.

5.2.4.2 Image Acquisition

TEM imaging data were obtained for Control and Crushed sections at both low and

high magnification. When obtaining high resolution images, the magnification was kept

constant (46,000x) to ensure images taken across a region of interest could be stitched

together for analysis (Figure 5.2B). Sample exposure time and operating voltage (20-

120kV) of the Philips were controlled by adjusting the beam intensity at higher

magnifications to prevent beam damage from over exposure. TEM grid identification of


93

the samples used for quantification is given in Table 5.1. Images were saved as TIFFs

and exported into Adobe Photoshop CS6 (Adobe Systems Incorporated, San Jose, CA,

USA) for post-processing.

Table 5.1: TEM grid, slot, and specimen identification for Control and Crushed axons.

Holder Number Slot Number Axon ID Load Status Blue 102813 C2 1-5 Control NF 69393 B1 1-15 Control NF Blue 102813 A6 1-4 Crushed NF Blue 102813 A7 1-11 Crushed NF 75056 H3 1-6 Control MT 75059 A1 1-4 Control MT 75059 A2 1-5 Control MT 75057 G1 1-3 Crushed MT 75057 G3 1-3 Crushed MT 75057 G4 1-6 Crushed MT 75057 H2 1-3 Crushed MT 75057 H3 1-4 Crushed MT 75058 M2 1 Crushed MT

5.2.4.3 Post-Processing of TEM Image Data

Areas outside the axon membrane in TIFF images were removed from the

neurofilament data using Adobe Photoshop CS6 (Adobe Systems Incorporated, San Jose,

CA, USA) as shown in Figure 5.3. This was done to improve the quality of the

automated image processing quantification conducted using MATLAB for the

neurofilament data. Manual quantification of microtubule data did not require a

background subtraction. With the background noise removed, neurofilament TIFF

images were exported to MATLAB (Mathworks; Natick, MA, USA) for analysis. The

MATLAB code written for neurofilament image analysis is given in Appendix C.


94

Figure 5.2: Focal compression of isolated primary hippocampal axons. A) Schematic illustration of axon loading environment and orientation within the AIM. The axon only region overlaps with a 20μm thick compression pad above the testing chamber. Microfluidics are used to control the compression pad and localize loading to the area underneath the pad (blue box) (Chapter 3). B) A series of panoramic TEM images reconstructing the entire area of the axon under the compression pad at higher magnification. C) A single TEM image for quantifying number, density, and spacing of the cytoskeletal structures. Scale bar = 500nm. [118]


95

Figure 5.3: Neurofilament TEM images prior to (left) and after (right) background subtraction. No Au nanoparticles were found outside the axon indicating the labeling technique was successful. Following background subtraction images were exported as TIFFs for quantification.

5.3 Quantification of TEM Data

Metrics for TEM data quantification utilized a series of longitudinal sections taken at

the same magnification (x46,000) and examined the entire segment under the

compression pad (Figure 5.2B-C). Previous researchers have used transverse (cross-

section) examination of axons as a method for examining changes in cytoskeletal

structure following TAI [47, 124]. However, given the loading methodology employed in

this work and the heterogeneous distribution of cytoskeletal components observed by

previous researchers in cross sections of TEM axon samples, we found it more useful to


96

examine the longitudinal section of the entire length of axons under the compression pad

[47, 124].

Quantification of TEM images requires metrics for assessing quantity, spacing, and

density. Previous researchers indicate a relationship between the axon diameter and

cytoskeletal constituents within them. Therefore, our TEM images were segmented to

assess required measures at known diameters while limiting variation in the structure for

the chosen magnification [47, 48].

5.3.1 TEM Quantification for Microtubules

For microtubules, the length of the axon (L) for a given TEM image was measured

and the unit length was divided into quarters. Axon diameter (D) and the number of MTs

(NMT) were measured at L/4, L/2, and 3L/4 (Figure 5.4). Microtubule linear density

across the diameter of the axon (𝜌𝐿𝑀𝑇) was calculated using by the number of MTs (NMT)

and the axon diameter (D)

𝜌𝐿𝑀𝑇 = 𝑁𝑀𝑇𝐷

(5.1)

The average spacing for microtubules (SMT), at a given D, was related to the linear

density and corrects for the thickness (t) of the microtubules

𝑆𝑀𝑇 =𝐷 − 𝑁𝑀𝑇 ∙ 𝑡

𝑁𝑀𝑇=

1𝜌𝐿𝑀𝑇

− 𝑡 (5.2)


97

Figure 5.4: Spacing along unit length (L) of the axon for microtubule measures at L/4, L/2, and 3L/4.

5.3.2 TEM Quantification for Neurofilaments

Neurofilament measurements were made by dividing the axonal length into thirds for

a given TEM image and measuring the diameter at the midpoint of each of these thirds

(Figure 5.5). For each of the thirds the areal density (𝜌𝐴𝑁𝐹) of Au-nanoparticles was

computed by measuring the number of 6nm Au-nanoparticles (NNF) and the area of the

axon (A) using the MATLAB (Mathworks; Natick MA, USA) code provided in

Appendix C,

𝜌𝐴𝑁𝐹 = 𝑁𝑁𝐹𝐴

(5.3)


98

Average spacing for NF (SNF) was measured as the distance between a given Au-

nanoparticle and its nearest Au-nanoparticle neighbor. The spacing was measured for all

Au-nanoparticles in a given sample and assumes a proportional binding relationship

where the observed Au-nanoparticles requires the presence of the NF-medium binding

sites and infers the existence of NFs.

Figure 5.5: Axonal segmentation into thirds. Diameter measurements were made at the midpoint of each area, along the length of the axon for neurofilament quantification.

5.3.3 Statistical Analysis of TEM Data

Collected data was divided into four groups: (1) microtubule Control cells, (2)

neurofilament Control cells, (3) microtubule Crushed cells, (4) neurofilament Crushed

cells. To evaluate the relationship between axon caliber and cytoskeletal measures, the

data was further separated into bins covering the ranges of axon diameters used: <0.5μm,

0.5-1.0μm, 1.0-1.5μm, and 1.5-2.0μm. To determine the significance between Control


99

and Crushed groups, for a specific axon diameter bin, Bonferroni’s multiple comparison

tests were used. The tests determine the significance (p < 0.05) of the pairing.

5.4 Cytoskeletal Component Level Response to TAI

Using a previously described controlled mechanical environment with relevant

applied loading rates and loads to induce at TAI response in CNS neural axons, we find

appreciable changes in cytoskeletal spatial distributions within neural axons fixed less

than 1min after loading. These changes include decreases in the number and density of

microtubules and neurofilaments, as well as increases in their spacing, following loading.

Another unique result from our study came from examining the dependency that

microtubule and neurofilament measures have on axon caliber. This dependency

continues to exist immediately following load for most measures; the exception is

neurofilament spacing, though the slope of these measures is modified in all cases.

Neurofilament spacing for Crushed axons appears constant across all axon diameters

indicating phosphorylated sidearm mechanism governing the spacing is modified. Our

findings suggest conformational changes in the neurofilament structure may serve as a

trigger for further secondary damage to the axon, representing a key insight into the

temporal aspects of cytoskeletal degeneration at the component level.


100

5.4.1 Structural Changes in Cytoskeleton

5.4.1.1 Morphological Assessment of Microtubules

Microtubules were identifiable as long rod-like structures ~24nm in diameter within

the axon (Figure 5.6). For the Control group, microtubules were observed along the

primary axis of the axon and regularly distributed across the axon diameter (Figure 5.6

A-B). For the Crushed group, microtubules appeared to be unchanged in areas where no

nodal blebs or swellings were evident (Figure 5.6C lower center). In areas where

swellings were present, the microtubules (arrows) on average aligned along the axis of

the axon; however the distribution was frayed and disorganized (Figure 5.6C upper right,

D). Microtubules of the Crushed group exhibited breaking points where the ends

appeared to be undergoing depolymerization (arrows) within nodal blebs (arrow heads)

(Figure 5.7). Spatially, the nodal blebs appeared restricted to the compressed axon

volume and the portions of the axon immediately adjacent to this volume. Another

interesting observation was the presence of mitochondria (dark elliptically shaped

structures) within the nodal blebs (arrow heads) (Figure 5.7). Mitochondria are organelles

that synthesize adenosine triphosphate (ATP), a molecule that functions as a universal

energy-transfer molecule. The mitochondrial presence in the nodal blebs suggests that

either nodal bleb activity is a primary site for metabolic activity, or perhaps, that the

mitochondria organelles were caught in the breakdown of the transport mechanism of the

microtubules, and the axon is collapsing around them.


101

Figure 5.6: TEM images for microtubules of (A-B) no load (Control) and (C-D) loaded (Crushed) axons. Microtubules are indicated by black arrows (B, D). Axon diameter, number of microtubules, and spacing between microtubules were measured for each image along unit axon length, L, at L/4, L/2, and 3L/4. A) In Control axons, microtubules are oriented along the principal axis of the axon. C) In Crushed axons, microtubules appear disorganized and misaligned. B,D) Inset of Control and Crushed axons showing diameter (D) and spacing (SMT) measurements for microtubules. Scale bars = 100nm. [118]


102

Figure 5.7: Degenerative response associated with axons undergoing TAI. Nodal blebs (arrow head) of Crushed axon show mitochondria in each bleb. Inset of Crush axon bleb showing microtubule breakage, rupture, and depolymerization. Scale bars = 100nm. [118]

5.4.1.2 Morphological Assessment of Neurofilaments

Neurofilaments, where localization was confirmed by Au labeling using 2° NF-

medium antibodies, were identified within treated axons. Examples of Control and

Curshed groups with Au nanoparticles are shown in Figures 5.8 - 5.9. In these examples,

the 6nm Au nanoparticles are highlighted with yellow circles (Figure 5.8-5.9B, E). Insets

of these examples at higher magnification shown the nanoparticles as small black dots

(arrows) distributed within the axons (Figures 5.8-5.9C, F). In the Control group, Au

nanoparticles were spaced regularly along the length of the axon and across the axon

diameter indicating neurofilaments were distributed uniformly within the axon (Figure

5.8A-C). The number of and areal density of the Au nanoparticles for the Crushed group


103

appeared lower than the Control group and the spacing of Au nanoparticles was

heterogeneous within the axon membrane in areas with and without nodal blebs,

indicating neurofilament distribution had been modified (Figure 5.8D-F). In many cases,

the number of nanoparticles at the membranes were observed less for the Crushed group

and appear to aggregate along the midline of the axon, away from the membrane, as

compared to the Control group (Figure 5.9B, E). This suggests the neurofilament

sidearms, which govern the spacing of the neurofilaments between the membrane and

each other, may be modified by the loading.


104

Figure 5.8: TEM images for neurofilaments of (A-C) no load (Control) and (D-F) loaded (Crushed) axons. 6nm Au nanoparticles, outlined in yellow, were used to measure areal density and spacing between neurofilaments for each image. A-C) In Control axons, Au nanoparticles are regularly spaced along the length of the axon and across the axon diameter. D-F) In Crushed axons, Au-nanoparticles appear more heterogeneous in their distribution and spacing. B-C, E-F) Inset of Control and Crushed axons showing areal density and spacing distribution for nanoparticles. There is a greater number and areal density in Control axons (C) than in Crushed axons (F). Scale bars = 100nm. [118]


105

Figure 5.9: TEM images for neurofilaments of (A-C) no load (Control) and (D-F) loaded (Crushed) axons. 6nm Au nanoparticles, outlined in yellow, were used to measure areal density and spacing between neurofilaments for each image. B, E) Yellow circles highlight the 6nm Au nanoparticles (black dots) used to assess quantity, density distribution, and spacing of neurofilaments using antibody labeling. The aggregation of Au nanoparticles appears towards the midline of the axon with fewer nanoparticles at the axon membrane for Crushed than Control groups. C, F) Inset showing 6nm Au nanoparticles (arrows). Scale bar = 100nm for A-F. [113]


106

5.4.2 Changes in Cytoskeletal Spatial Distribution

One goal of this work is to find measurable changes in cytoskeletal distribution and to

link these quantifiable metrics to TAI. Previous researchers have indicated the number,

spacing, and density of microtubules and neurofilaments in control axons to be measures

with a strong dependency on axon caliber [45, 47, 48, 53]. Our Control group data

supports the literature as we also observe a strong dependency of axon caliber on 𝑁𝑀𝑇

and 𝑁𝑁𝐹. Additionally we observe a higher rate of increase in 𝑁𝑁𝐹 than 𝑁𝑀𝑇 for increases

in axon diameter of Controls, a trend supported by previous researchers [48]. Decreased

𝜌𝐿𝑀𝑇 and increased 𝑆𝑀𝑇 has been linked with increasing axon caliber by previous

researchers and is reflected in our data set [45, 47, 48, 53]. Researchers also report

increases in 𝑆𝑁𝐹 and decreases in 𝜌𝐴𝑁𝐹 with increasing axon diameter, findings we

observe for Control axons [47, 48, 53]. Following loading, Crushed axons exhibited

significant changes in the measures used, though the axon caliber dependency persisted

for most.

5.4.2.1 Quantitative Assessment of Microtubules

The number of (𝑁𝑀𝑇), linear density (𝜌𝐿𝑀𝑇), and spacing between (𝑆𝑀𝑇) microtubules

were the quantitative measures taken from axons fixed <1min after loading and compared

for Control and Crushed groups (Figures 5.10 - 5.13). Previous researchers have

indicated the number, spacing, and density of microtubules control axons to be measures

with a strong dependency on axon caliber [45, 47, 48, 53]. The raw data for these


107

measures are shown in Figure 5.10A, C, E with 95% confidence intervals. A power law

fit was used in the form of

𝑦(𝐷) = 𝑎𝐷𝑚 (5.4)

where y represents the cytoskeletal measure, D is axon diameter, and a, m are fitting

parameters for the power law. From Figure 5.10, the smallest diameters measured are

approximately 100nm and no data was taken for axons of a smaller caliber than this.

Therefore an assumption was made for the appropriate form of the power law fit because

axons of diameter 0nm should have not microtubules within them.

Table 5.2 provides the coefficients for the power law fits, in the form of Equation 5.4,

and for the 95% confidence intervals of the metrics shown in Figures 5.10A, C, E (dash

lines). Using these fits, a 1000nm caliber axon example for Control is estimated to have

𝑁𝑀𝑇 = 10 MTs, a 𝜌𝐿𝑀𝑇= 0.0105µm-1, and 𝑆𝑀𝑇 = 80nm. The same diameter axon for

Crushed is estimated to have 𝑁𝑀𝑇 = 7 MTs, a 𝜌𝐿𝑀𝑇= 0.008µm-1, and 𝑆𝑀𝑇 = 160nm.

Note the power law fits work well for the majority of the data population, but may be less

reliable at the outliers (for axons where 𝑁𝑀𝑇< 3 or 𝑁𝑀𝑇 >15).

The data in Figures 5.10A, C, E appears in bands which can be connected to the

whole number values of 𝑁𝑀𝑇 in Figure 5.10A. Given the dependence of 𝑁𝑀𝑇 on axon

diameter, Equation 5.4 takes the form,

𝑁𝑀𝑇 = 𝑎𝐷𝑚 (5.5)

Equations 5.1-5.2 can be rewritten by substituting Equation 5.5 as,

𝜌𝐿𝑀𝑇 =𝑁𝑀𝑇𝐷

= 𝑎𝐷𝑚−1 (5.6)


108

Tabl

e 5.

2:

Pow

er la

w fi

tting

coe

ffic

ient

s (Eq

uatio

n 5.

4) fo

r mic

rotu

bule

mea

sure

s of C

ontro

l and

Cru

shed

axo

ns

and

the

95%

con

fiden

ce in

terv

als f

or th

ose

coef

ficie

nts.

Cyt

oske

leta

l Mea

sure

men

t L

oad

Stat

us

a (L

ower

/Upp

er) 9

5%

Con

fiden

ce In

terv

al

m

(Low

er/U

pper

) 95%

C

onfid

ence

Inte

rval

M

T N

umbe

r 𝑵𝑴𝑴

Con

trol

0.21

73

(0.1

479/

0.28

67)

0.55

89

(0.5

105/

0.60

73)

C

rush

ed

0.58

29

(0.4

279/

0.73

80)

0.35

23

(0.3

085/

0.39

62)

MT

Line

ar D

ensi

ty 𝝆

𝑳 𝑴𝑴 (μ

m-1

) C

ontro

l 0.

1640

(0

.121

9/0.

2062

) -0

.397

3 (-

0.43

88/-0

.355

8)

C

rush

ed

0.22

81

(0.1

611/

0.29

51)

-0.4

838

(-0.

5366

/-0.4

309)

M

T Sp

acin

g 𝑺𝑴𝑴 (n

m)

Con

trol

0.72

82

(0.2

830/

1.17

3)

0.67

86

(0.5

866/

0.77

07)

C

rush

ed

0.17

04

(0.0

7747

/0.2

632)

0.

9932

(0

.910

5/1.

076)


109

𝑆𝑀𝑇 =𝐷 −𝑁𝑀𝑇 ∙ 𝑡

𝑁𝑀𝑇=

𝐷𝑁𝑀𝑇

− 𝑡 =1

𝜌𝐿𝑀𝑇

− 𝑡 = 𝑎−1𝐷1−𝑚 − 𝑡 (5.7)

For the power law fit of Equation 5.4, Equation 5.6 indicates m for 𝜌𝐿𝑀𝑇 should be ≅ m-1

for 𝑁𝑀𝑇, which according to Table 5.2 it is for both Control and Crushed axons.

When the raw data is binned by axon caliber into 500nm demarcations, the contrasts

between Control and Crushed groups for microtubule measures are easily comparable

(Figures 5.11 - 5.13). 𝑁𝑀𝑇 increased with axon diameter in the Control group, a trend

observed by previous researchers (Figure 5.11) [48]. Similarly the decrease in 𝜌𝐿𝑀𝑇 and

increase in 𝑆𝑀𝑇 for Controls has been indicated by previous researchers as the diameter of

the axon increased (Figures 5.12-5.13) [45, 47, 48, 53]. The 𝑁𝑀𝑇 for the Control group

was significantly higher than the Crushed group across all axon diameters (p<0.05)

(Figure 5.11). This could indicate the microtubules are being moved out of the middle

plane (where TEM images are taken from) or that the microtubules are in a state of

retraction or depolymerization. The 𝜌𝐿𝑀𝑇 for the Control group was higher than the

Crushed group, though the results is statistically significant only for 0.5-1.5μm diameter

axons (Figure 5.12). This makes sense as the diameter of a crushed axon is likely greater

than that of a control axon and, with a decrease in 𝑁𝑀𝑇, the linear density would also

decrease. The 𝑆𝑀𝑇 for Crushed axons was significantly higher than the Control groups at

axon diameters of 0.5μm and above indicating the microtubules may be spreading out in

response to the applied load (Figure 5.13). A common observation was that the

differences between Control and Crushed groups are magnified at the larger axonal

diameters for all measures taken.


110

Figure 5.10: Raw data plots of all cytoskeletal metrics (𝑁𝑀𝑇, 𝜌𝐿𝑀𝑇, 𝑆𝑀𝑇, 𝑁𝑁𝐹, 𝜌𝐴𝑁𝐹 , and 𝑆𝑁𝐹) as functions of axon diameter. A-F) All plots are expansions of the data summarized in Figures 5.11-5.16. 95% confidence intervals are shown (dashed lines) for each of the fit curves. A, C, E) Power law fits are used to estimate the relationship between axon diameter and microtubule cytoskeletal measures. B, D, F) Linear fits approximate the relationship between axon caliber and neurofilament metrics. Only 𝑆𝑁𝐹for Crushed axons appear to remain constant, at approximately 70nm, following loading.


111

Figure 5.11: The number of (𝑁𝑀𝑇) microtubules for no load (Control) and loaded (Crushed) axons. Axon diameter bins are given along the X-axis and error bars are standard error mean for all plots. A strong dependency on axon caliber is observed the number of microtubules in Control and Crushed axons. 𝑁𝑀𝑇 is significantly lower for Crushed across all axon diameters. The observed differences in 𝑁𝑀𝑇 are more apparent at larger axon diameters. Adapted from [118]. *(p<0.05)

Figure 5.12: The linear density (𝜌𝐿𝑀𝑇) of microtubules for no load (Control) and loaded (Crushed) axons. Axon diameter bins are given along the X-axis and error bars are standard error mean for all plots. A strong dependency on axon caliber is observed for all linear density measures of microtubules in Control and Crushed axons. 𝜌𝐿𝑀𝑇 is lower for Crushed than Control in nearly all axon diameters. The observed differences in 𝜌𝐿𝑀𝑇

are more apparent at larger axon diameters. Adapted from [118]. *(p<0.05)


112

Figure 5.13: The spacing between (𝑆𝑀𝑇) microtubules for no load (Control) and loaded (Crushed) axons. Axon diameter bins are given along the X-axis and error bars are standard error mean for all plots. A strong dependency on axon caliber is observed for all spacing measurements between microtubules in Control and Crushed axons. 𝑆𝑀𝑇appears larger for Crushed than Control in nearly all axon diameters. These observed differences 𝑆𝑀𝑇 are more apparent at larger axon diameters. Adapted from [118]. *(p<0.05)

The mean number of (𝑁�𝑀𝑇), linear density (�̅�𝐿𝑀𝑇), and spacing between (𝑆�̅�𝑇)

microtubules is presented for Control and Crushed groups for all axon diameters (Table

5.3). Our microtubule data agree with previous researchers where 𝑁�𝑀𝑇 and �̅�𝐿𝑀𝑇 was

decreased for Crushed axons than Control axons for the same axon caliber. Also, the

increased 𝑆�̅�𝑇 observed by researchers for Crushed axons are supported by our data [47,

48, 53].


113

Table 5.3: Quantification of microtubules and neurofilaments following focal compression.

Cytoskeletal Measurement Control ± SEM

Crushed ± SEM

% Change from Control ± SEM

Mean MT Number 𝑵�𝑴𝑴 9 ± 2 6 ± 1 -33 ± 10 Mean MT Linear Density 𝝆�𝑳𝑴𝑴 (μm-1) 11 ± 2 8 ± 3 -27 ± 13 Mean MT Spacing 𝑺�𝑴𝑴 (nm) 75 ± 15 189 ± 27 152 ± 62 Mean NF Number 𝑵�𝑵𝑵 91 ± 36 40 ± 13 -56 ± 22 Mean NF Linear Density 𝝆�𝑨𝑵𝑵 (μm-2) 46 ± 14 25 ± 6 -47 ± 21 Mean NF Spacing 𝑺�𝑵𝑵 (nm) 55 ± 8 70 ± 3 28 ± 20

Mean number (𝑁�𝑀𝑇), linear density (�̅�𝐿𝑀𝑇), and spacing (𝑆�̅�𝑇) of microtubules and mean number (𝑁�𝑁𝐹), areal density (�̅�𝐴𝑁𝐹), and spacing (𝑆�̅�𝐹) of neurofilaments from axons in all bin sizes. Standard error mean (SEM) is given for all values. For microtubules (n=4) and neurofilaments (n=3) for Control and Crushed populations. [118]

5.4.2.2 Quantitative Assessment of Neurofilaments

The number of (𝑁𝑁𝐹), areal density (𝜌𝐴𝑁𝐹), and spacing between (𝑆𝑁𝐹) Au-

nanoparticles were the quantitative measures taken and compared for Control and

Crushed neurofilament groups (Figures 5.10, 5.14 - 5.16). Previous researchers have

indicated the number, spacing, and areal density of neurofilaments in control axons to be

measures with a strong dependency on axon caliber [45, 47, 48, 53]. Note that no

Control axons were found having an axon diameter 1.5-2.0μm for this assessment.

The raw data is shown in Figures 5.10B, D, F with 95% confidence intervals. The

data in Figures 5.10B, D, F does not exhibit the same power law data bands observed for

microtubules in Figures 5.10A, C, E, but the populations can be distinguished using a

linear fit in the following form


114

𝑦(𝐷) = 𝑝1𝐷 + 𝑝2 (5.8)

where y represents the cytoskeletal measure, D is axon diameter, and 𝑝1 and 𝑝2 are linear

fitting parameters. No data was taken for axons of a caliber <100nm; therefore our linear

fit is limited in applicability to the data range shown. Additionally, the linear fits work

well for the majority of the data population, but may become less reliable at the outliers

of the data set.

Table 5.4 provides the coefficients for the linear fits for each of the metrics shown in

Figures 5.10B, D, F. These fits estimate 𝑁𝑁𝐹, 𝜌𝐴𝑁𝐹 , and 𝑆𝑁𝐹 for Control and Crushed

axons of a specified axon diameter. An example of a Control axon with a diameter of

1000nm is estimated to have 𝑁𝑁𝐹 = 130 NFs, a 𝜌𝐴𝑁𝐹= 66µm-2, and 𝑆𝑁𝐹 = 53nm. The

same diameter axon for Crushed is estimated to have 𝑁𝑁𝐹 = 55 NFs, a 𝜌𝐴𝑁𝐹= 32µm-1,

and 𝑆𝑁𝐹 = 67nm. Note the power law fits work well for the majority of the data

population, but may be less reliable at the outliers (for axons where 𝑁𝑁𝐹< 10 or 𝑁𝑁𝐹

>100).


115

Tabl

e 5.

4:

Line

ar fi

tting

coe

ffic

ient

s (E

quat

ion

5.8)

for n

euro

filam

ent m

easu

res

of C

ontro

l and

Cru

shed

axo

ns a

nd

the

95%

con

fiden

ce in

terv

als f

or th

ose

coef

ficie

nts.

Cyt

oske

leta

l Mea

sure

men

t L

oad

Stat

us

𝒑 𝟏

(Low

er/U

pper

) 95%

C

onfid

ence

Inte

rval

𝒑 𝟐

(L

ower

/Upp

er) 9

5%

Con

fiden

ce In

terv

al

NF

Num

ber 𝑵𝑵𝑵

Con

trol

0.13

57

(0.1

070/

0.16

45)

-5.7

27

(-17

.34/

5.88

3)

C

rush

ed

0.05

205

(0.0

4571

/0.0

5839

) 2.

204

(-1.

934/

6.34

3)

NF

Are

al D

ensi

ty 𝝆

𝑨 𝑵𝑵 (μ

m-2

) C

ontro

l 0.

0595

(0

.037

7/0.

0812

9)

6.11

5 (-

2.68

9/14

.92)

Cru

shed

0.

0266

3 (0

.026

63/0

.030

99)

5.65

9 (-

2.81

9/8.

5)

NF

Spac

ing

𝑺𝑵𝑵 (n

m)

Con

trol

-0.0

1985

(-

0.05

361/

0.01

392)

72

.89

(59.

25/8

6.52

)

Cru

shed

-0

.004

766

(-0.

0196

6/0.

0101

3)

72.2

2 (6

2.5/

81.9

5)


116

The raw data is binned by axon caliber into 500nm demarcations to contrast and

compare between Control and Crushed groups for neurofilaments measures (Figures 5.14

- 5.16). The number of neurofilaments (as inferred from 𝑁𝑁𝐹) increased with axon

diameter in the Control group and agrees with observations from the literature (Figure

5.14) [48]. Similarly the increase in 𝜌𝐴𝑁𝐹 and decrease in the 𝑆𝑁𝐹 for Controls as the

diameter of the axon increased supports previous researchers (Figures 5.15 - 5.16) [45,

47, 48, 53]. The 𝑁𝑁𝐹 and 𝜌𝐴𝑁𝐹 were significantly lower in the Crushed group across all

comparable axon diameters (p<0.05) (Figures 5.14 - 5.15). This may indicate the NF-

medium sidearm antibody, used for Au nanoparticle labeling, is no longer accessible and

has been modified by the loading. While 𝑆𝑁𝐹 appears higher for Crushed groups, there

was not a statistically significant difference between axons of comparable axon diameters

(Figure 5.16). It is interesting to note that 𝑆𝑁𝐹 did not significantly change as axon

diameters increased for Crushed axons unlike the Control (Figure 5.16). As expected, the

differences between Control and Crushed neurofilament groups were magnified at the

larger axon diameters for all comparable measures taken. The mean number of (𝑁�𝑁𝐹),

areal density (�̅�𝐴𝑁𝐹), and spacing between (𝑆�̅�𝐹) Au-nanoparticles were computed for

Control and Crushed neurofilament groups for all diameters and are also shown in Table

5.3.

When comparing the percent change from Control, measures taken from loaded axons

in the literature indicate increased measurements of 𝑁�𝑁𝐹 and �̅�𝐴𝑁𝐹 and decreases in 𝑆�̅�𝐹

[45, 47, 48, 53]. This contradicts the measurements we observe where 𝑁�𝑁𝐹 and �̅�𝐴𝑁𝐹


117

decrease and 𝑆�̅�𝐹 increases for this study. This difficulty in comparing neurofilament

trends directly may be due to variations in loading as well as temporal differences

between our studies. Our study has a focal compressive loading methodology whereas

much of the literature uses stretch techniques that are known to produce tensile loading

[45, 47, 48, 53]. Temporal differences may arise because our work focuses on the

immediate response of the cytoskeleton to loading (fixation time <1min from loading),

whereas previous researchers have quantified their data at later time periods with fixation

times ranging between 15min-7days after loading.

Figure 5.14: The number of (𝑁𝑁𝐹) Au-nanoparticles for no load (Control) and loaded (Crushed) axons. Axon diameter bins are given along the X-axis and error bars are standard error mean for all plots. A strong dependency on axon caliber is observed for the number of neurofilaments for both Control and Crushed axons. The 𝑁𝑁𝐹 observed is significantly lower for Crushed across all comparable axon diameters. These observed differences in 𝑁𝑁𝐹 are more apparent at larger axon diameters. Adapted from [118]. * (p<0.05)


118

Figure 5.15: The areal density (𝜌𝐴𝑁𝐹) of Au-nanoparticles for no load (Control) and loaded (Crushed) axons. Axon diameter bins are given along the X-axis and error bars are standard error mean for all plots. A strong dependency on axon caliber is observed for all areal density measures. 𝜌𝐴𝑁𝐹 is significantly lower for Crushed than Control all comparable axon diameters. These observed differences in 𝜌𝐴𝑁𝐹 are more apparent at larger axon diameters. Adapted from [118]. * (p<0.05)

Figure 5.16: The spacing between (𝑆𝑁𝐹) Au-nanoparticles for no load (Control) and loaded (Crushed) axons. Axon diameter bins are given along the X-axis and error bars are standard error mean for all plots. The strong dependency on axon caliber observed for the Control group does not appear for the Crushed group. 𝑆𝑁𝐹 appears larger for Crushed than Control in nearly all axon diameters and remains approximately 70nm for all Crushed axons regardless of axon diameter. These observed differences in 𝑆𝑁𝐹 are more apparent at larger axon diameters. Adapted from [118]. * (p<0.05)


119

5.4.3 Connecting TEM Measures to Confocal Results

In Chapter 4, the confocal results indicate a 24% decrease in the protein levels

associated with neurofilaments and no observable changes in microtubule protein levels

while under load. Figures 5.10A, C, E shows that 𝑁𝑀𝑇, 𝜌𝐿𝑀𝑇, and 𝑆𝑀𝑇 change with

applied loads, however the 95% confidence intervals for the Control and Crushed fits

cannot be separated. 𝑁𝑁𝐹, 𝜌𝐴𝑁𝐹, and 𝑆𝑁𝐹 also change with applied loads, though the 95%

confidence intervals of the Control and Crushed fits are better separated, particularly at

the larger axon calibers (Figure 5.10B, D, F). The TEM quantification supports the

confocal measurements where microtubules did not exhibit a measurable change during,

and immediately following, applied loadings while the neurofilament proteins exhibited a

statistically significant 24% decrease. This indicates the in situ response of cytoskeletal

populations can be captured during loading and can be connected to quantifiable metrics

covering the spatial distribution of individual cytoskeletal constituents.

5.4.4 Changes in Cytoskeletal Temporal Distribution

5.4.4.1 Quantitative Comparisons with the Literature

The percent change from Control was computed for 𝑁�𝑀𝑇, �̅�𝐿𝑀𝑇, 𝑆�̅�𝑇, 𝑁�𝑁𝐹, �̅�𝐴𝑁𝐹, and

𝑆�̅�𝐹 and is presented in Table 5.3. To compare our results with previous researchers the

percent change from Control is plotted as a function of time for all measures (Figure

5.17-5.18). The time between loading and fixation is plotted along the x-axis for each


120

subplot and varies from t=1min to t=7days. The percent change from Control for each of

the measures is plotted along the y-axis.

Our results show that while 𝑁�𝑀𝑇 and 𝑁�𝑁𝐹 decreased immediately following loading,

𝑁�𝑁𝐹 experienced a larger percent decrease than 𝑁�𝑀𝑇 (Figure 5.17A, 5.18A). This

indicated a greater rate of change for 𝑁�𝑁𝐹 than 𝑁�𝑀𝑇 over the same time period following

the applied load. Interestingly 𝑁�𝑁𝐹 appeared to return to Control values in advance of

𝑁�𝑀𝑇, which continued to decrease for 15min, indicating a delayed response of the

microtubules to the applied load. As with 𝑁�𝑀𝑇, �̅�𝐿𝑀𝑇 decreased for 15min following

loading before returning to Control values at 7days (Figure 5.17B). By comparison, �̅�𝐴𝑁𝐹

decreased immediately following loading and increases above the Control values at 4hrs

(Figure 5.18B). This particular measure was unclear in terms of its behavior due to a lack

of quantifiable data from the literature. Both 𝑆�̅�𝑇 and 𝑆�̅�𝐹 appeared to increase following

loading and return to Control values at longer time periods (𝑆�̅�𝐹 in advance of 𝑆�̅�𝑇)

(Figure 5.17C, 5.18C).


121

Figu

re 5

.17:

Te

mpo

ral d

escr

iptio

n of

the

perc

ent c

hang

e fro

m C

ontro

l for

the

mea

n nu

mbe

r, de

nsity

, and

spa

cing

of

mic

rotu

bule

s.

Dat

a fro

m th

e cu

rrent

stu

dy is

plo

tted

with

rep

orte

d Li

tera

ture

val

ues

at k

now

n tim

e po

ints

and

erro

r ba

rs a

re s

tand

ard

erro

r m

ean

for

all

plot

s.

A) M

ean

num

ber

of m

icro

tubu

les

appe

ars

to d

ecre

ase

imm

edia

tely

fo

llow

ing

load

ing

and

may

req

uire

as

long

as

7 da

ys t

o re

turn

to

Con

trol

valu

es.

B)

Mic

rotu

bule

lin

ear

dens

ity

decr

ease

s ap

pear

to p

eak

at 1

5min

follo

win

g in

jury

bef

ore

retu

rnin

g to

Con

trol v

alue

s.

C)

Spac

ing

for

mic

rotu

bule

s ap

pear

s to

incr

ease

follo

win

g lo

adin

g an

d re

turn

to C

ontro

l val

ues

at lo

nger

tim

e pe

riods

. [11

8]


122

Figu

re 5

.18:

Te

mpo

ral d

escr

iptio

n of

the

perc

ent c

hang

e fro

m C

ontro

l for

the

mea

n nu

mbe

r, de

nsity

, and

spa

cing

of

neur

ofila

men

ts.

Dat

a fro

m th

e cu

rrent

stu

dy is

plo

tted

with

rep

orte

d Li

tera

ture

val

ues

at k

now

n tim

e po

ints

and

erro

r ba

rs a

re s

tand

ard

erro

r m

ean

for

all

plot

s.

A) T

he m

ean

num

ber

of n

euro

filam

ents

dec

reas

es f

ollo

win

g lo

adin

g;

how

ever

the

retu

rn to

Con

trol v

alue

s ap

pear

s to

requ

ire le

ss ti

me

than

mic

rotu

bule

s by

com

paris

on (F

igur

e 5.

17A)

. B)

Ar

eal d

ensi

ty f

or n

euro

filam

ents

app

ears

to

decr

ease

s im

med

iate

ly f

ollo

win

g lo

adin

g an

d m

ay in

crea

se a

bove

the

C

ontro

l va

lues

at

4hrs

. I

n ad

ditio

n to

the

lac

k of

qua

ntifi

able

dat

a fro

m t

he L

itera

ture

, th

e di

ffere

nces

in

perc

ent

chan

ge fr

om C

ontro

l and

unc

lear

tren

d m

ay b

e a

func

tion

of a

con

form

atio

nal s

hift

in th

e ne

urof

ilam

ent s

idea

rm th

at

coul

d af

fect

the

qua

ntifi

catio

n m

etho

ds u

sed

in t

his

stud

y.

C)

Spac

ing

for

neur

ofila

men

ts a

ppea

rs t

o in

crea

se

follo

win

g lo

adin

g an

d re

turn

to C

ontro

l val

ues

at lo

nger

tim

e pe

riods

. [11

8]


123

5.4.4.2 Load Response of Cytoskeleton

Variances in the cytoskeletal degradation response between microtubules and

neurofilaments may arise from differences in the load response over the short term

response for the axon. For shorter time frames where fixation occurs within 1min of

loading, local intracellular mechanisms may govern the whole cell response.

Microtubules, being the most robust component of the cytoskeleton, can reasonably be

assumed to not readily degenerate in response to an applied load as compared to less rigid

cytoskeletal components like neurofilaments [117]. Across all diameters, the transverse

stiffness of the axon can be assumed as the same as the neurofilaments provide no

measurable increase in transverse stiffness and the microtubules are oriented

perpendicular to the load direction [117]. Therefore the simplest cytoskeletal structures

within the axon may fail at a faster rate before observable changes can be detected for the

more rigid constructs. While there is a great deal of evidence to support microtubule

disruption as a primary mechanism for nodal bleb formation and eventual axon

degeneration, our data indicates changes in neurofilaments are preceding measurable

spatial changes in microtubules.

5.4.4.3 Temporal Evolution of Cytoskeleton

Data from this study provides measurable insights into cytoskeletal changes <1min

after loading. Previous literature has focused on quantifying the spatial distributions of

microtubules and neurofilaments as early as 15min after loading. Our data indicates

metrics for neurofilament number and density are changing at higher rates than


124

equivalent metrics for the microtubules for the same time period. The literature indicates

microtubules reach their greatest difference from Controls, for number and density, at a

later time than neurofilaments [45, 47, 48, 53]. This supports results presented in Chapter

4 using confocal imaging where neurofilaments exhibited a significant 24% decrease in

measurable intensity while under load; yet microtubule changes were negligible.

Ahmadzadeh et al. has provided insight into temporal aspects of our microtubule data

over short time durations [125]. Ahmadzadeh was able to link the viscoelasticity of tau

proteins to the strain-rate dependent behavior of microtubules, where strain rates above

22-44s-1 led to breakage of the microtubules; though the microtubules and tau were still

present immediately after loading [125]. This connects to our confocal data which used

tau for labeling microtubules, and where no measurable change in fluorescence was

detected for the microtubules. Similarly the TEM images indicate that microtubules may

become disorganized and rupture following applied loads, but they are still present within

the axon for short time durations.

5.4.5 Numerical Approach to Cytoskeletal Metrics

From a different perspective, scaling arguments for cytoskeletal metrics (𝑁𝑀𝑇, 𝜌𝐿𝑀𝑇,

𝑆𝑀𝑇, 𝑁𝑁𝐹, 𝜌𝐴𝑁𝐹, and 𝑆𝑁𝐹) with axon caliber hold well for Control axons, and continue to

hold for Crushed axons; except for 𝑆𝑁𝐹 (Figure 5.10). A numerical approach to

understanding these metrics would show 𝜌𝐴𝑁𝐹 scales with the number of Au

nanoparticles within a specified area and, if expanded, is shown to be inversely


125

proportional to the square of the average spacing between Au nanoparticles (𝑆𝑁𝐹 )

(Figure 5.19) where.

𝜌𝐴𝑁𝐹 = 𝑁𝑁𝐹𝐴

= #𝐴∝

1𝑆𝑁𝐹2

(5.9)

From Figure 5.10F, we can observe a decrease in 𝑆𝑁𝐹 for Control axons as axon diameter

increases. For Figure 5.18, we can visually confirm the relationship between 𝑆𝑁𝐹 and

𝜌𝐴𝑁𝐹 where,

𝜌𝐴𝑁𝐹 = #𝐴

≅ 4

𝐴1 + 𝐴2 + 𝐴3 + 𝐴4=

44𝑆𝑁𝐹2

=1

𝑆𝑁𝐹2 (5.10)

If we rearrange Equation 5.9, the average spacing is shown to be inversely proportional to

the square root of the areal density,

𝑆𝑁𝐹 ∝ �1

𝜌𝐴𝑁𝐹 (5.11)

This indicates the incremental decreases we observe for the average spacing between

Au nanoparticles as axon caliber increases lead to increases in the areal density of the

nanoparticles (Figure 5.10D, F). This numerical relationship for neurofilament

distribution with respect to axon diameter between 𝜌𝐴𝑁𝐹 and 𝑆𝑁𝐹 holds for the Control

group, but appears to break down for the Crushed group.


126

Figure 5.19: The distribution of neurofilaments in Control group relating the spacing between neurofilaments (𝑆𝑁𝐹) with the areal density (𝜌𝐴𝑁𝐹). A-B) For Control axons, the distribution is close to homogeneous; where 𝜌𝐴𝑁𝐹 and 𝑆𝑁𝐹 are strong correlated with the number of neurofilaments and axon caliber. C) Schematic illustrating the neurofilament distribution changes with respect to 𝑆𝑁𝐹 and 𝜌𝐴𝑁𝐹 as axon caliber increases. D) With increasing axon caliber, 𝑆𝑁𝐹 slowly decreases and the 𝜌𝐴𝑁𝐹 increases. For Crushed axons, the neurofilament distribution is heterogeneous, meaning the numerical relationship between 𝑆𝑁𝐹 and 𝜌𝐴𝑁𝐹 breaks down and no longer applies.


127

5.4.6 Spacing Mechanism for Neurofilament Sidearms

An interesting question that arises from the spatial data is the mechanism governing

𝑆𝑁𝐹 and why the established numerical relationship with axon caliber appears nullified in

Crushed axons. This spacing is mediated by phosphorylation of carboxy-terminal of NF-

medium and NF-heavy sidearms where increases in the number of neurofilaments and

carboxy-terminal phosphorylation increase axon caliber [126]. Previous research has

indicated a change in this phosphorylation state, as initiated by applied loads, may lead to

alterations in ionic concentrations [8, 99]. The altered homeostasis includes interactions

between phosphates and protein kinases and leads to dephosphorylating the carboxy-

terminal resulting in changes to the sidearm structure [8, 87, 99]. We hypothesize

changes in the protein folding for the neurofilament sidearm structure following loading

leads to local modifications in conformational shape and spatial distributions within the

axon.

Local modifications in conformation represent alterations in protein folding where

NF-medium antibodies are no longer accessible and result in a decrease of observed Au-

nanoparticles. We posit that within the Crushed axon, local modifications to

neurofilament sidearms result in a spacing mechanism where a baseline level of

approximately 70nm exists regardless of axon caliber. Remarkably the 70nm value we

observe in Figure 5.10F and 5.16 for Crush axon neurofilament spacing has been

observed by researchers at later time periods following loading [48, 53].


128

5.4.7 Limitations of Study

Potential limitations covering areas of biology, sample fixation technique and

labeling, and image processing exist for the study.

As mentioned in Chapter 4, we have biological limitation by the using of rats instead

of humans. Additionally the usage of E17 rat pups is not necessarily representative of the

more mature central nervous tissue found in human adults. These are standard limitations

that exist in many animal models and are accepted by the research community.

The TEM images required fixing the neural cells for both Control and Crushed

groups. While we were able to apply fix <1min after load, limiting the time frame for

potential cytoskeletal component level changes in distribution, the inability to obtain

TEM high-resolution images without fixing the neural cells limits the scope of our work

from replicating the in situ response of the cell. Additionally the usage of immunogold

labeling is known to have distances between approximately 15-30nm from the primary

binding sites of the antibody [122]. Given the spacing of the nanoparticles from their

binding sites should be comparable and the number density of the nanoparticles is equal

to the number density of the binding sites, we can accept this limitation as a means for

quantifying the neurofilament distribution.

TEM imaging limitations allow only a single 2D plane from the axon to be

analyzed. While axons are inherently 3D objects, their growth was limited to a 2D glass

substrate due to experimental constraints. To address concerns processing 3D embedded

substructures serial sections were analyzed. Additionally, during the image processing of


129

these samples, limited portions of the axon membrane may have been removed as part of

background removal prior to quantification.

5.5 Summary of TEM Study of Cytoskeletal

Evolution

In this chapter, we use electron microscopy to improve our understanding of the

changes observed through confocal imaging with alterations in the ultrastructural

composition of microtubules and neurofilaments within neural axons. Standard

transmission electron microscopy processing methods were used to identify microtubules

while neurofilaments identification required the use of antibody labeling utilizing gold

nanoparticles. The number of, density, and spacing of microtubules and neurofilaments

were quantified for specimens in sham Control and Crushed groups with fixation at

<1min following load. These metrics provide a pathway for connecting changes in

cytoskeletal spatial distributions to previously observed changes in measured intensity

using confocal microscopy with the same loading platform in situ and in vitro (Chapter 4)

and may be critical in understanding mechanical failure and degeneration of the

cytoskeletal system for neural axons undergoing TAI. The following summarizes some of

the main conclusions of this chapter:


130

• The axon caliber dependency known to exist for microtubule and neurofilaments

extends to axons undergoing TAI, with the exception of neurofilament spacing,

which appears to remain constant across all Crushed axon diameters.

• The spacing mechanism for neurofilaments appears to have a baseline of 70nm

across all axon calibers undergoing TAI and the scaling arguments for a

numerical approach appear to break down for this measure.

• The temporal evolution of cytoskeletal metrics used indicate changes in

neurofilament spatial distributions within axons undergoing TAI precede

microtubule changes in response to applied loads.

The ability to assess the temporal response of the cytoskeleton to traumatic axonal

injury has numerous applications in the TBI and SCI field. The techniques developed in

this work can be applied to develop new protocols and standard in the treatment and

prevention of TAI. Further research is needed to examine the early biochemical

pathways initiated by neurofilament disruption, and how these disruptions affect

microtubule functions over short time durations following loading (<15min). The

mechanics of the applied load likely influence the whole cell response resulting TAI.

Strain rate, for example, has been implicated by several research groups for governing

cytoskeletal responses to applied load. An approach varying the strain rate for focal

compression on neural axons might provide insights to rate dependent responses of

cytoskeletal constituents. In the following chapter the clinical and research implications

for this work will be further discussed and the future directions for this work are

proposed in detail.

131

Chapter 6

Discussion and Future Directions

In this work, an experimental framework has been developed to investigate one of the

most common pathologies for traumatic brain injury (TBI) and spinal cord injury (SCI):

traumatic axonal injury (TAI). The framework implements a series of multiscale

experiments spanning the cellular and subcellular length scales for neural axons

undergoing TAI. The axons are loaded transversely in the form of focal compression

using a microfluidic platform. Applied loads are known prior to testing using a validated

finite element model. The cellular level injury response of the axon to applied loads

provides a series of thresholds for obtaining a response of continued growth,

degeneration (TAI), or regrowth. By using the degeneration threshold corridor, the

evolution of the cytoskeletal structure could be explored using confocal microscopy and

transmission electron microscopy. As a result, this experimental approach can be applied

CHAPTER 6. CYTOSKELETAL EVOLUTION UNDER FOCAL AXON COMPRESSION

132

to quantify the evolution of the cytoskeletal structure in TAI and may lead to the

development of preventative measures and injury mitigation strategies for TBI and SCI.

In the following sections of this chapter, the major contributions of this work will be

discussed, and potential clinical and research implications will be proposed. There are

numerous potential directions for exploring TAI within the developed framework and

these will be discussed in Section 6.3. The majority of the work presented in Chapter 6 is

from both existing publications and others that are, at present, under review [51, 113,

118].

6.1 Major Contributions & Future Directions

6.1.1 Axon Injury Micro-Compression Platform Development &

Threshold Validation for TAI

Traumatic axonal injury is often investigated using carefully controlled mechanical

platforms to apply known loads on neural axons that will elicit a degenerative cellular

response concomitant with TAI. The majority of these experimental platforms, for

cellular and subcellular responses, use axial stretch to investigate TAI [45, 47-50, 127].

While stretch is often identified as the primary mechanism of injury in TAI, other modes

of deformation at the cellular level can lead to axonal damage. As an example, neural

axons that lie in close proximity to stiffer structures within the skull and spinal column,

such as the falx cerebri and blood vessels, may be pinched or compressed during loading.

As described in Section 2.3, the complex loading conditions inducing TBI and SCI at the


133

mesoscale are translated down the length scale, to the cellular and subcellular level, in the

form of simple mechanical loads (stretch, compression, and shear). One of the objectives

of this work was to understand the loading and boundary conditions applied to isolated

neural axons using a microfluidic platform. This information was used to develop load-

response corridors for the neural axons placed in a state of focal compression.

A finite element model of the microfluidic AIM platform was developed with the

goal of optimizing the geometry of the platform and to estimate the applied load between

the compression pad and the glass substrate. The model utilized uniaxial testing data to

determine the constants of a hyperelastic Mooney-Rivlin material model (Section 3.4.2).

The finite element model was validated for input pressure-displacement measurements

for the compression pad and for contact pressure using imaging and integrating

instrumentation into the testing platform (Section 3.6). The model determined the

experimentally applied loads on isolated neural axons for specified input fluidic

pressures. Using this information, loading corridors were developed that could associate

the applied load with the cellular response of the neural axon where continued growth,

degeneration, or regrowth were observed for the cells following injury.

Although the relevance of focal compression to TAI, as compared to axial stretch, has

been discussed, the cellular response to the applied load leaves no question that TAI had

occurred. Understanding the injury response in terms of a controlled mechanical load is

useful at the cellular level and lower length scales. At the larger length scales, clinical

findings suggest pure stretch, shear, and compression do not readily occur, but exist as

complex combinations [64]. As a result, understanding the cellular level response to


134

focal compression provides insight into linking existing stretch and shear mechanical

models of TAI, with the goal of understanding how the cell responds to combinations of

these loads at larger length scales.

There are a number of ways in which the cell line responses and the AIM platform

can be extended in future work. Primary rat embryonic CNS cells were chosen for these

experiments due to their reproducibility of isolation, their extensive characterization in

the field of neuroscience, and their ability to survive in culture for extended periods of

time. Using these cells, we demonstrated the ability to examine the relationship between

injury thresholds and axon response within our AIM platform. In the case of embryonic

CNS cells, the work suggests that there are unique thresholds that govern the balance

between axon survival, degeneration, or regrowth (Section 3.7.2). This work provides a

path for comparisons within the adult CNS system, which is less permissive to regrowth.

A future comparative study between the two systems, as well as with peripheral nervous

system neurons in which successful regeneration typically occurs, could enable a deeper

understanding of the specific processes responsible for inhibitory or successful regrowth

in the adult CNS.

Variations in the AIM geometry of the compression pad might allow for controlled

focal compression at specified displacements (Figure 6.1A-B). Currently the entire axon

cross section is transversely compressed during experimentation (100% strain); however

by modifying the geometry of the base of the compression pad to create a gap or notch,

the displacement of the pad on the axon could be varied and investigated (Figure 6.2).


135

Finally the mechanics of the applied load, such as strain and strain rate, likely

influence the whole cell response resulting in TAI. In-vitro studies have successfully

shown while the mechanical strain applied to a neural cell can induce cell death, the rate

of the applied strain is also tied to cell viability [128]. A variation in the strain rate for

applying these loads would be a potential direction to pursue with respect to the injury

response.

Figure 6.1: AIM platform with 1µm notch heights integrated at the base of the compression pad. A) The notch allows for controlled compression at known strain levels where the notch height represents the portion of the axon not to be compressed. B) A laser scan profile shown measure notch heights at the base of the modified compression pad.


136

Figu

re 6

.2:

La

ser

scan

pro

file

of t

he n

otch

con

trolle

d de

vice

s.

Pairs

of

lines

, by

col

or,

repr

esen

t th

e ba

se o

f th

e co

mpr

essi

on p

ad a

nd th

e no

tche

s fo

rmed

to c

ontro

l stra

in d

urin

g lo

adin

g.

Her

e th

e av

erag

e no

tch

heig

ht is

1.0

µm

(bla

ck a

rrow

).


137

6.1.2 In-Vitro and In Situ Visualization and Quantification of

Cytoskeletal Deformation under Load

Conventional approaches for visualizing the mechanics of subcellular populations

under controlled load utilize approaches such as transmission electron microscopy and

immunoblotting. These methods are limited because they are known to damage the cells

that confound interpretation of cytoskeletal changes, and are limited in temporal and

spatial resolution. These methods yield strong insights into alterations in cytoskeletal

components of neural axons undergoing TAI, yet the fixation methods lead to temporal

limitations that inhibit understanding cytoskeletal evolution under load. A methodology

to capture information as it is applied in situ is required.

In this work, quantification and visualization of three-dimensional (3D) live

subcellular populations during mechanical loading of axons using a confocal microscope

at high magnification was demonstrated. The mechanical forces were obtained using a

computational (finite element) model validated by integrating instrumentation into the

testing platform (Section 3.4). This methodology allowed, for the first time, a

continuous, quantitative 3D spatial, and temporal visualization of evolving cytoskeletal

substructure in situ and under load, thus dramatically improving the understanding of in-

vitro cellular mechanics.

Although focal compression is largely ignored by existing models for TAI, our work

has shown the AIM platform to be highly successful in inducing TAI and providing 3D

in-vitro and in situ responses of the axon cytoskeleton during and following loading. For


138

areas of the central nervous system where neural axons are in close proximity to stiffer

surroundings (i.e. blood vessels and falx cerebri), the potential for undergoing

compression during a TBI or SCI event is increased. It appears that the absence of

compression driven criteria may lead to gross under predictions of injury in

computational models of TBI and SCI where only tensile driven experimental models are

utilized.

Several groups have highlighted the need to develop non-invasive techniques to probe

temporal evolution of the cytoskeleton, in response to TAI [45, 49, 116, 129, 130]. Some

of these groups have made steps to quantify microtubule and neurofilament expression as

early as 15min following loading. However none have explored the spatial distribution

or expression during loading [116]. The in situ and in-vitro response of the axons to TAI

induced loads that we measure are in agreement with the long-term trends in microtubule

and neurofilament densities observed by previous studies employing TEM and

immunoblotting methods at known time points following loading (Section 4.5; Figure

4.11) [45, 49, 116]. As time between load application and quantification is increased, so

does the percent decrease in cytoskeletal densities for both microtubules and

neurofilaments. However, the rate of decrease was found to be greater in the

neurofilaments than for the microtubules (Figure 4.12). The underlying evidence

indicates specified cytoskeletal populations of neural axons can be observed using

confocal imaging; however technological limitations of the confocal microscope prevent

visualization of individual cytoskeletal constituents.


139

In future work, multiple avenues exist to pursue the discussion of TAI. The question

of longer temporal effects on cytoskeletal expression in the experimental platform could

be investigated to compare directly with some of the pre-existing experimental models

for TAI. Banik et al. and Serbest et al. have quantified changes in cytoskeletal density

72hrs post-injury [49, 116]. By extending our experiment to longer time intervals, more

direct comparisons of the temporal aspects of cytoskeletal expression changes can be

made with the literature. Another improvement in this work could be made by

fluorescently labeling both microtubules and neurofilaments within the same axon. In the

work conducted, we labelled microtubules and neurofilaments using a green fluorescent

protein in separate populations of the same cell type. Using different color fluorophores

for each cytoskeletal constituent (cyan or yellow) would enable visualization of both

microtubule and neurofilament expression within the same cell and provide more insight

into changes in cytoskeletal populations under load.

Improving the understanding of the effects of strain, strain rate, and the associated

injury response would help in the development of computational models that might

effectively incorporate these lower scale insights into larger scale constructs. Strain rate,

for example, has been implicated by several research groups for governing cytoskeletal

responses to applied load. An approach varying the strain rate for focal compression on

neural axons might provide insights to rate dependent responses of cytoskeletal

constituents.

There are a number of ways in which our experimental model can be extended to

areas outside of the scope of TAI described in this work. The use of confocal imaging


140

allows for continuous and live 3D visualization of cytoskeletal expression. This isolated

subcellular approach for understanding the evolution of the cytoskeleton under load may

be applied across a wide range of fields including cellular motility and growth, changes

in cancer cell morphology, and apoptosis. There is immense value in observing and

understanding the mechanisms of changes in cellular and subcellular processes related to

their environment.

6.1.3 Cytoskeleton Quantification and Temporal Evolution under

Focal Axon Compression

The use of TEM to quantify changes in cytoskeletal spatial distributions improves the

visual resolution over that of confocal imaging, but sacrifices the ability to observe in

situ. The alterations observed in cytoskeletal expression using confocal imaging

provided an insight that neurofilaments were undergoing changes ahead of microtubules

for the same applied load. Temporal aspects of cytoskeletal changes have been

previously explored using TEM, though quantification of the distributions are limited to

time frames of 15min post-injury or later [47, 48, 53, 116]. The need to understand how

changes in cytoskeletal intensity expression (as witnessed using confocal microscopy)

translate to structural modifications in the spatial distribution of the cytoskeleton drove

the usage of TEM.

Measurable parameters governing the spatial distribution of the cytoskeleton include

the number, the spacing, and the density. While differences in testing platforms, loading


141

and boundary conditions, and the choice of imaging planes for quantification exist

between this work and that conducted by the literature, a significant number of parallels

are also observed with our results. The observation of axon caliber dependencies on all

measures for the Control groups was expected and has been previously identified in the

literature [48]. Interestingly this relationship, now previously identified by the literature,

appears to extend to axons undergoing TAI for most measures in our experiment; with

the exception of neurofilament spacing. The changes observed in this work are compared

with those from the literature where the temporal evolution of microtubules and

neurofilaments undergoing TAI are shown (Figure 5.16). These plots indicate the

maximum difference, in terms of percentage change from Control, are occurring at later

time periods for microtubules than for neurofilaments. This is a unique result because it

provides a mechanical framework to address the response of the cytoskeleton to applied

loads. The neurofilaments, more specifically the neurofilament sidearms, are described

as soft as compared to the more robust microtubules. We posit after loading, the

neurofilament sidearms depolymerize and undergo a conformational change in the

folding of the protein substructure. This conformation shape change in the sidearms and

substructure of the neurofilaments leads to modifications in the uptake of Au

nanoparticles used to identify neurofilaments.

The quantification approach used in this work is limited in that the quantified

distributions of microtubules and neurofilaments are only located along axial slices of the

axon underneath the compression pad. This was chosen because the volume compressed

underneath the pad is expected to show a more accurate spacing distribution over a larger


142

area than if transverse cross-sectional slice of the axon have been quantified. Previous

researchers have used transverse examination of axons as a method for examining

changes in cytoskeletal structure following TAI [47, 124]. However, given the loading

methodology employed in this study and the heterogeneous distribution of cytoskeletal

components observed by previous researchers in cross sections of TEM axon samples, we

found it more useful to examine the longitudinal section of the entire length of axons

under the compression pad [47, 124]. A future path might consider extending the analysis

to capture the transverse cross-sectional slices, which are known to be 60-80nm

thickness, along the entire volume where the compression was applied. Careful

consideration should be made as the entire compressed volume may occupy as many as

500 TEM image slices for a single axon.

6.2 Clinical and TAI Research Implications

Our experimental model for TAI can be used as a platform for studying traumatic

brain injury and spinal cord injury. It can be applied to develop a better understanding of

how spatial changes in the cytoskeletal substructure within neural axons lead to physical

impairment and loss of cognitive functions. It can also be used to investigate treatment

options in the form of drug delivery or electro-chemical therapy for prevention,

mitigation, and rehabilitation of TAI. The following sections describe some of these

potential applications and some of the clinical implications of this work.


143

6.2.1 Relating Structural Evolution and Loss of Neural Cognition

with TAI

The CNS is a complex system, where its survival is connected to the changes and

adaptations it can make across multiple length scales in response to its environment.

Studies over the last two decades consider TAI as a progressive process invoked by the

mechanical forces causing injury which gradually evolve from focal axonal alteration to

ultimate disconnection. The clinical manifestations of neural cognitive disorders or loss

of pre-existing neurological capacities frequently do not appear until weeks or months

after the injury has occurred. One of the primary challenges facing clinicians is the ability

to accurately diagnose TAI in a damaged region of the brain since the structural signature

of the pathology is not visible through screening tools such as medical imaging until the

damage has manifested at larger scales.

One way to improve the clinical approach of determining whether TAI is occurring

would be to utilize a computational model. This model would be able to relate the

loading conditions with a material model of the axons that could incorporate the

cytoskeletal structure observed in this and other studies. Changes in the distribution of

the cytoskeleton could be related to the viability of the axons, which in turn would be

connected to damage parameters associated with the fiber tracts of the CNS. The

damaged tracts could be related to functional outcomes for the patients where specified

regions of the brain that are damaged might relate to cognitive impairment or disability.

Existing computational modeling frameworks go so far as to associate axonal strain with


144

functional damage of specific fiber tracts. I suggest that by connecting the axonal strain

to changes in the underlying cytoskeletal structure within the axons, a more accurate

assessment of injury may be understood.

6.2.2 Applications to Therapeutic Intervention

The pathobiological response of the axon to TAI has been the focus of researchers

over the past several decades. By targeting known biochemical responses of neural tissue

to TAI, clinicians assess the likelihood of injury as well as target therapeutic delivery

systems for inhibiting damage propagation within neural tracts. Büki and Povlishock, in

a review of pathobiological consequences for TAI, highlight that the damaging

progression of events for the majority of injured axons associated with TAI may be

potentially attenuated via rationally targeted therapies [52]. Büki and Povlishock point

out that several groups have demonstrated traumatically induced focal axolemmal

permeability leads to local influx of Ca2+ with the subsequent activation of the cysteine

proteases, calpain and caspase, that then play a pivotal role in the ensuing pathogenesis of

TAI via proteolytic digestion of brain spectrin, a major constituent of the subaxolemmal

cytoskeletal network [52]. In this pathological progression (where the local overloading

of Ca2+ combined with the activation of calpains initiates mitochondrial injury that results

in the release of cytochrome-c, with the activation of caspase) both the activated calpain

and caspases participate in the degradation of the local axonal cytoskeleton, including

neurofilaments and microtubules, causing local axonal failure and disconnection. Some

of the potential targets for therapeutic intervention, as highlighted in Section 2.3.1.2,


145

include calpain-mediated spectrin proteolysis, mitochondrial permeability transition-

induced mitochondrial damage, cytoskeletal alteration caused by calcium-accumulation

(activation of calcineurin) and the activation of caspase-3 [52].

Of the aforementioned targets, the most ideal therapy should focus on intra-axonal

events preceding the mitochondrial release of cytochrome-c [52]. This is based on the

fact that mitochondrial cytochrome-c release evokes severe disturbances in the electron-

transport leading to bulk generation of free radicals, in addition to the activation of

caspases, [131-133]. As therapeutic interventions should target the early phases of TAI,

the calcium-activated neutral proteases and perhaps the induction of mitochondrial injury

(MPT) would seem to represent the most rational targets for therapeutic intervention.

Additionally several experimental studies have demonstrated the beneficial effects of

calpain inhibition in Ca2+-induced pathology in various central nervous system disorders

including ischemic brain damage [134-137], spinal cord injury [138-140] and cerebral

contusion [141-143].

6.2.3 Prevention and Mitigation of TBI and SCI

A major application of our experimental framework is in the prevention and

mitigation of traumatic brain injury and spinal cord injury through the avoidance of

traumatic axonal injury. Experimental models feed many of the computational models

that can be used as tools in quantifying the effectiveness of protective equipment (bomb

suits, combat helmet, sporting equipment, vehicle safety harnesses, and car safety

features) against TBI and SCI. In order to be effective, computational models require


146

accurate experimental data that can capture the conditions the models are required to

predict. Our experimental model provides a framework for assessing the cytoskeletal

distribution prior to and immediately following TAI through focal axonal compression.

Our experimental data can provide insights into the temporal evolution of the

cytoskeleton following TAI and can be integrated into computational models where the

neural substructure is parameterized. Additionally this framework can be extended into

the improvement of existing mechanical thresholds for TAI, and can be used in the

development of new protocols and standard for injury prevention in sports, travel, and the

military.

6.3 Summary Suggestions for Future Work

This work has developed an experimental model and framework for investigating

traumatic axonal injury. There are a variety of ways in which this work can be extended

and improved upon to expand our understanding of how TAI propagates and the role the

cytoskeleton has in it leading up to TBI and SCI. Some of the possible extensions of this

work, discussed earlier in the chapter, are summarized in the following points:

• Characterization of strain response of axon for focal compression: The

experimental model can be extended to assess the effects of strain using the AIM

platform. Displacement can be controlled using patterned gaps incorporated into

the base of the compression pad for the microfluidic device. When control

fluidics are input to the modified platforms, the compression pad will make


147

contact with the glass base; however, transverse gaps of a controlled height will

prevent the axons underneath from complete compression. This will facilitate a

means for assessing the cellular and subcellular thresholds of the axons as a

function of applied displacement (in the form of transverse compression).

• Characterization of strain rate on the response of axon to focal compression: The

experimental model can be modified to assess the effects of strain rate by

modifying the control fluidics of the AIM platform. Input fluidic pressure rates

are controlled by the size and geometry of the tubing and channels leading to the

pressure bladder within the control layer. By expanding or constricting these

geometries, the flow rate of the input CO2 may be increased or decreased.

Results from these experiments can be used to assess the rate effects of loading in

focal compression for neural axons.

• Validation of injury model for axon maturity and peripheral nervous system

(PNS) axons: The experimental model can be repeated for adult neural axons

from the CNS. It is likely the thresholds developed using embryonic rat CNS

cells have unique thresholds that may vary as maturity of the host increases due

to an increase in cytoskeletal proliferation. In a similar fashion, dorsal root

ganglion (DRG) would provide an excellent means for investigating this injury

model for the PNS where the threshold may vary for the applied load.


148

• Labeling of multiple cytoskeletal constituents within the same axon:

Microtubules and neurofilaments of the axons can be labelled with different color

fluorophores to enable visualization, using fluorescent imaging, within the same

cell. This visualization would improve the ability to assess how the specified

cytoskeletal constituents evolve with respect to each other.

• Transverse serial sectioning along the axis of the axon for transmission electron

microscopy specimens: Continuous samples, at 60-80nm thickness, for TEM

specimens could be accomplished by reorienting the cutting planes for grid

specimens along the plane transverse to the compression pad. This would

significantly increase the amount of TEM images obtained for a single axon,

from approximately 12-500 per axon, in order to traverse the 30µm compression

pad. This would improve the understanding of how cytoskeletal constituents

deform under focal compression by adding a third dimension to their

deformation.

• Extension of experimental model to analytical model bridging cytoskeleton to

cell for traumatic axonal injury: In this work, we have demonstrated the ability

to obtain a TAI response using a controlled experimental platform applying focal

compression. However, the majority of modeling work for assessing traumatic

brain injury and spinal cord injury can only handle information from the axon

level or above. Of these models, none can connect focal compression with

changes in the cytoskeletal substructure. An extension of these models is


149

required to capture the full extent of axonal injury in the brain or spinal cord

undergoing complex loading conditions.

6.4 Conclusions

In order to be a viable tool in the study of traumatic brain injury and spinal cord

injury, experimental models for traumatic axonal injury must be able to provide

quantitative measures for assessing changes in behavior of neurological tissues and to

connect these across multiple length scales. The goal of this work was to address both of

these issues. To better understand TAI, we have developed a new experimental platform

to apply controlled loads on isolated CNS axons. This platform uses focal compression

through microfluidics where the applied load is predicted using a validated finite element

model of the system. The experimental and finite element models have led to the

development of threshold criteria, governing the cellular response of the axons to the

applied load, as continued growth, degeneration (TAI), or regrowth. A framework to

assess the temporal evolution of the cytoskeleton during the TAI response of the cell was

developed using confocal microscopy and transmission electron microscopy. The ability

to visualize the live cell in situ and in-vitro response was accomplished through confocal

microscopy where fluorescently tagged microtubules and neurofilaments were

continuously imaged prior to, during, and immediately following focal compression.

Comparisons between unloaded and loaded live cells demonstrate both spatial and

temporal changes for cytoskeletal populations within the imaged volume. Transmission


150

electron microscopy connected the changes observed through confocal imaging with

alterations in the ultrastructural composition of microtubules and neurofilaments within

neural axons. These metrics provide a pathway for connecting changes in cytoskeletal

spatial distributions to previously observed changes in measured intensity using confocal

microscopy with the same loading platform in situ and in vitro, and may be critical in

understanding mechanical failure and degeneration of the cytoskeletal system for neural

axons undergoing TAI. Our experimental framework can be applied to developing new

connections with existing analytical and computational models for predicting TBI and

SCI at smaller length scales. This could manifest itself in the form of new standards and

protocols for protection against TAI, and to improve protective materials and restraint

systems. In a clinical setting the work might be used for therapeutic targeting and

intervention of the biochemical cascades known to exist in the propagation of TAI. This

work offers a ground-breaking experimental structure for traumatic axonal injury and

provides a launch point from which future research can propagate.

151

Appendix A

Table A.1: Stretch-Stress data from the literature and uniaxial tests of Sylgard 184

shown in Figure 3.4 [110]

Stretch (λ) Cauchy Stress (σ) (Pa) 1.400 896000.0 1.375 819500.0 1.342 724500.0 1.308 628000.0 1.275 561000.0 1.242 496666.7 1.210 423645.8 1.183 355000.0 1.142 274000.0 1.108 221666.7 1.067 128000.0 1.033 72333.3 1.006 0.0 1.000 0.0 0.996 -2777.1 0.992 -8097.2 0.988 -15592.5 0.984 -25063.2 0.980 -35904.7

Appendix A

152

0.976 -47557.4 0.972 -59788.8 0.968 -72099.7 0.964 -84357.3 0.960 -96584.4 0.956 -108716.8 0.952 -120572.4 0.948 -132565.1 0.944 -144329.0 0.940 -155958.0 0.936 -167697.7 0.932 -179212.7 0.928 -190770.3 0.924 -202226.8 0.920 -213556.4 0.916 -224978.3 0.912 -236295.2 0.908 -247699.0 0.904 -258950.5 0.900 -270244.8 0.896 -281654.8 0.892 -293036.1 0.888 -304455.3 0.884 -315913.8 0.880 -327347.4 0.876 -338899.7 0.872 -350585.8

153

Appendix B

The following is the MATLAB (Mathworks; Natick, MA, USA) code developed for

analysis of confocal data stacks. The data stacks were segmented in Imaris (Bitplane

AG; Zurich, Switzerland) and exported as a TIFF stack.

An example of the file format used for analyzing the confocal data is given below.

“CytoskeletalImaging_tr_Update_Imrs” and “RedCellTrackerImaging_Updated_Imrs”

are functions that call the TIFF stacks. The fileName denoted the date_cytoskeletal

population_AIM platform number_input fluidic pressure at time of loading_sample

number_specific TIFF stack_all the files in the Z stack. The C0 and C1 nomenclature

refers to the stack of TIFF associated intensities from the cytoskeleton (c488ht) or the

axon membrane (c562mp) respectively. The lo parameter refer to the lower bound

applied to remove noise from the image intensity and the same lo value was applied to

unloaded and loaded image stacks to ensure a baseline between the two could be

Appendix B

154

matched. The high parameter was the ceiling value measured for the image intensity and

was not required to match between the two stacks.

%% 20120813_NF CytoskeletalImaging_tr_Update_Imrs('20120813_NF_1_0psi_1_C0_Z*',100,867); %c488ht RedCellTrackerImaging_Updated_Imrs('20120813_NF_1_0psi_1_C1_Z*',700,10332); %c562mp CytoskeletalImaging_tr_Update_Imrs('20120813_NF_1_8psi_1_C0_Z*',100,2307); %c488ht RedCellTrackerImaging_Updated_Imrs('20120813_NF_1_8psi_1_C1_Z*',700,6699); %c562mp “CytoskeletalImaging_tr_Update_Imrs” and “RedCellTrackerImaging_Updated_Imrs”

assign values of B and A that represent the intensity data for the entire subvolume of

intensity data after it has been processed using the lo and high parameter thresholds. The

cumulative intensity data is found using “Sum_Intensity_RedCellTracker_PostFilter” and

“Sum_Intensity_Cytoskeletal_PostFilter”.

function [A] = RedCellTrackerImaging_Updated_Imrs(fileName,lo,hi) % Create an array of filenames that make up the image sequence fileFolder = fullfile('Users','Adam','Documents','Image_Stacks','MATLAB_TIFF'); dirOutput = dir(fileName); fileNames = {dirOutput.name}; numFrames = numel(fileNames); I = imread(fileNames{1}); % Preallocate the array sequence = zeros([size(I) numFrames],class(I)); sequence(:,:,2) = I; % Create image sequence array zz=0; for p = 1:numFrames sequence(:,:,p) = imread(fileNames{p}); end

Appendix B

155

Sum_Intensity_RedCellTracker_Raw=sum(sequence(:)) % Build the intensity profiles for p = 1:numFrames

%Remap image from original data range to a standard uint16 image, [0 65535] % A = imadjust(sequence,[low_in; high_in],[low_out; high_out]) maps the values in sequence to new values in A such that values between low_in and high_in map to values between low_out and high_out. Values below low_in and above high_in are clipped; that is, values below low_in map to low_out, and those above high_in map to high_out. You can use an empty matrix ([]) for [low_in high_in] or for [low_out high_out] to specify the default of [0 1].

A(:,:,p) = imadjust(sequence(:,:,p),[lo/65023; hi/65023], [0;1]); end Sum_Intensity_RedCellTracker_PostFilter=sum(A(:))

function [B] = CytoskeletalImaging_tr_Update_Imrs(fileName,lo,hi) % Create an array of filenames that make up the image sequence fileFolder = fullfile('Users','Adam','Documents','Image_Stacks','MATLAB_TIFF'); dirOutput = dir(fileName); fileNames = {dirOutput.name}; numFrames = numel(fileNames); J = imread(fileNames{1}); % Preallocate the array sequence2 = zeros([size(J) numFrames],class(J)); sequence2(:,:,2) = J; % Create image sequence array for q = 1:numFrames sequence2(:,:,q) = imread(fileNames{q}); end Sum_Intensity_Cytoskeletal_Raw=sum(sequence2(:)) % Build the intensity profiles for q = 1:numFrames

%Remap image from original data range to a standard uint16 image, [0 65535] % B = imadjust(sequence2,[low_in; high_in],[low_out; high_out]) maps the values in sequence to new values in A such that values between low_in and high_in map to values between low_out and high_out. Values below low_in and above high_in are clipped; that is, values below low_in map to low_out, and those above high_in map to high_out. You can use an empty matrix ([]) for [low_in high_in] or for [low_out high_out] to specify the default of [0 1].

B(:,:,q) = imadjust(sequence2(:,:,q),[lo/65023; hi/65023], [0;1]); end Sum_Intensity_Cytoskeletal_PostFilter=sum(B(:))

156

Appendix C

The following is the MATLAB (Mathworks; Natick, MA, USA) code developed, Dr.

James D. Hogan, for analysis of neurofilament metrics (𝑁𝑁𝐹, 𝜌𝐴𝑁𝐹, and 𝑆𝑁𝐹).

%% NF Characterization clear all close all pack %Load a reference image to obtain a scale bar image=imread('test_082.tif'); figure imshow(image) close all scale=100/(1727-1622); %convert to nano meters %Name the file and input total number of pictures filename='AxonC1'; e_itt=1; numofpics=353; for i=1:1:numofpics %% Processing the image image=imread(['Pic (' num2str(i) ').tif']) ; %insert jpg/tif names to be loaded %

Appendix C

157

% figure % imshow(image) %convert to gray scale % image1=rgb2gray(image); image2=mat2gray(image); % figure % imshow(image2) image3=image2(1:3008,1:2464); %cutout the name bar on the bottom % figure % imshow(image3) J3 = imsharpen(image3); % figure % imshow(J3) % % pause(1) % close all %fn to adjust the images % J3 = imadjust(J3,stretchlim(J3),[]); % figure % imshow(J3) %re-scale image between 0 and 1. % J3=J3/max(max(J3)); % figure % imshow(J3) %Threshold: choose a threshold for the gold tracer particles. This will slightly change for different %images as the brightness varies for different axons/images. % J4=(J3 >0.98); J4=(J3<0.450); % figure % imshow(J4) % pause(2) % close all % clear particles touching border of images and those that are smaller than % X pixels in size elimbw=imclearborder(J4);

Appendix C

158

elimbw2=bwareaopen(elimbw, 15); %choose the threshold for the deletion of smaller objects. % figure % imshow(elimbw2) % Stats: compute the statistics of the images [B,L]=bwboundaries(elimbw2,'noholes'); numRegions=max(L(:)); % figure % imshow(L) %% Index the "blob" stats clearvars stats STATS stats=regionprops(L,L,'all'); %Various statistics % scale turns pix into m MajorAxisStats=[stats.MajorAxisLength]*scale; % Avg_MajorAxisStats=mean(MajorAxisStats); MinorAxisStats=[stats.MinorAxisLength]*scale; % Avg_MinorAxisStats=mean(MinorAxisStats); EccentricityStats=[stats.Eccentricity]; % Avg_EccentricityStats=mean(EccentricityStats); OrientationStats=[stats.Orientation]; % Avg_OrientationStats =mean(OrientationStats); PerimeterStats=[stats.Perimeter]*scale; % Avg_PerimeterStats =mean(PerimeterStats); AreaStats=[stats.Area]*scale.^2; % Avg_AreaStats =mean(AreaStats); WeightedCentroidStats=[stats.WeightedCentroid]; EquivDiam=[stats.EquivDiameter]*scale; Circularity=2*(3.14*([stats.Area]*scale.^2)).^0.5./([stats.Perimeter]*scale); clearvars STATS WeightCent clearvars temp for k=1:max(size(stats(:,1))); temp=stats(k,1); for ii=1:2; WeightCent(k,ii)= temp.WeightedCentroid(1,ii);

Appendix C

159

end end STATS(:,1)=WeightCent(:,1).'*scale; %in nm the x location STATS(:,2)=WeightCent(:,2).'*scale; %in nm the y location STATS(:,3)=MinorAxisStats.'; STATS(:,4)=MajorAxisStats.'; STATS(:,5)=EccentricityStats.'; STATS(:,6)=OrientationStats.'; STATS(:,7)=PerimeterStats.'; STATS(:,8)=AreaStats.'; STATS(:,9)=EquivDiam.'; STATS(:,10)=Circularity.'; %Colour intensity mean across the measured "blob" for k=1:max(size(stats(:,1))); % k=1000; clearvars temp pixlist temp=stats(k,1); % pixlist=[temp.PixelList]; pixlist=[temp.PixelIdxList]; STATS(k,11)=mean(J3(pixlist(:,:))); STATS(k,12)=std(J3(pixlist(:,:))); % figure % imshow(J3(pixlist(:,:))) % figure % imshow(J3) end STATS(:,13)=Solidity.'; STATS(:,14)=EulerNumber.';

Appendix C

160

% % figure % plot(STATS(:,4),STATS(:,4)./STATS(:,3),'.') % % % % % figure % plot(STATS(:,10),STATS(:,11),'.') % % % % % % % % figure % plot(STATS(:,11),STATS(:,4),'.') % % % % % figure % plot(STATS(:,13),STATS(:,4),'.') % % figure % plot(STATS(:,14),STATS(:,4),'.') % figure % plot(STATS(:,6),STATS(:,4)./STATS(:,3),'.') %% The following is used as criteria : these may slightly change dependending on how well the image processign identifies circular gold particles STATS(STATS(:,4)<5 | STATS(:,4)>15,:)=[]; %remove sizes smallr than 5 and larger than 15 nm STATS(STATS(:,14)<-1,:)=[]; %remove non-well connected blobls STATS(STATS(:,4)./STATS(:,3)>3,:)=[]; %filter for AR. remove high aspect ratio blobs STATS(STATS(:,10)<0.55,:)=[]; %circularity index. keep circular objects % STATS(STATS(:,11)>prctile(STATS(:,11),30),:)=[]; %mean colour intensity STATS(STATS(:,11)> 0.25,:)=[]; %mean colour intensity. this is the most important one. before setting this threshold check size vs intensity plot figure imshow(J3) hold on plot(round2(STATS(:,1)/scale,1),round2(STATS(:,2)/scale,1),'.','Markersize', 25) saveas(gcf,['J3Plotofdataoverimage', num2str(i),'fig']) close all

Appendix C

161

% %save %put in save(['stats' num2str(i) num2str(e_itt)], 'STATS'); if e_itt==1 CombinedSTATS(:,:)=STATS(:,:); save(['CombinedSTATS' num2str(filename) '.mat'], 'CombinedSTATS'); end if e_itt>1 prevlengthStats=length(CombinedSTATS(:,1)); CombinedSTATS(prevlengthStats+1:prevlengthStats+length(STATS(:,1)),:)=STATS(:,:); %put this fragment in the next line save(['CombinedSTATS' num2str(filename) '.mat'], 'CombinedSTATS'); end %calculate flaw density % Flawdens(e_itt,1)= ((length(STATS(:,1))/(length(image(:,1))*scale*length(image(1,:))*scale)*10^6)^0.5)^3; %area and number Flawdens(e_itt,1)= length(STATS(:,1))/(length(find(L(:,:)<1))*scale*scale); %areal density in #/nm^2 %Nearest Neighbor gold particle distribution for an image nearest=1; e_itttt=1; % STATS(STATS(:,4)<5,:)=[]; if length(STATS(:,1))>nearest data(:,1)=round2(STATS(:,1),0.01); data(:,2)=round2(STATS(:,2),0.01); for j=1:1:length(data(:,1)) x0=data(j,1); y0=data(j,2);

Appendix C

162

clearvars temp for k=1:1:length(data(:,1)) temp(k,1)= ((data(k,1)-x0).^2 + (data(k,2)- y0).^2)^0.5; end temp=sort(temp(:,1)); temp(find(temp(:,1)==0),:)=[]; % Color_ran=[rand, rand, rand]; % hold on % plot([1:length(temp)],temp,'Color', Color_ran) % set(gca,'FontSize',25,'FontName','times'); % xlabel({'Nearest Neighbour'},'FontSize',30,'FontName','Times New Roman'); % ylabel({'Distance'},'FontSize',30,'FontName','Times New Roman'); values(e_itttt,1:nearest)= temp(1:nearest,1); % values(j,2,i)=y0; % for ht=1:length(temp) % nnumber of nearest neighbours % values(j,2+ht,i)=temp(ht,1); %nearest % end clearvars temp x0 y0 e_itttt=e_itttt+1; end % VAL(i,1)=mean(values); % VAL(i,2)=median(values); clearvars data % clearvars data values % e_itt=1; end %area and number Flawdens(e_itt,2)= mean(values); %avg closet spacing e_itt=e_itt+1; clearvars -except CombinedSTATS filename ommit maxsize scale e_itt Flawdens i save Flawdens Flawdens end

Appendix C

163

%% Plots to consider figure smoothhist2D([STATS(:,9), STATS(:,11)],10,[900 900]) colormap('jet') set(gca,'FontSize',25,'FontName','times'); xlabel({'major (um)'},'FontSize',30,'FontName','Times New Roman'); ylabel({'mean colour'},'FontSize',30,'FontName','Times New Roman'); figure smoothhist2D([STATS(:,4), STATS(:,3)],10,[900 900]) colormap('jet') set(gca,'FontSize',25,'FontName','times'); xlabel({'major (nm)'},'FontSize',30,'FontName','Times New Roman'); ylabel({'minor (nm)'},'FontSize',30,'FontName','Times New Roman'); figure smoothhist2D([STATS(:,11), STATS(:,10)],10,[900 900]) colormap('jet') set(gca,'FontSize',25,'FontName','times'); xlabel({'Colour'},'FontSize',30,'FontName','Times New Roman'); ylabel({'Circularity'},'FontSize',30,'FontName','Times New Roman'); figure smoothhist2D([STATS(:,5), STATS(:,10)],10,[900 900]) colormap('jet') set(gca,'FontSize',25,'FontName','times'); xlabel({'Colour'},'FontSize',30,'FontName','Times New Roman'); ylabel({'Circularity'},'FontSize',30,'FontName','Times New Roman'); %% Cumualtive distribution of size clearvars temp % temp(:,1)=CombinedSTATS(:,4); %length % temp(:,1)=CombinedSTATS(:,4)./CombinedSTATS(:,3); %length % temp(:,1)=CombinedSTATS(:,6); %length temp(:,1)=CombinedSTATS(:,4); %length % temp(:,2)=CombinedSTATS(:,8)/sum(CombinedSTATS(:,8)); %mass temp=sortrows(temp,1); %sorted temp temp(1,4)=1;

Appendix C

164

temp(1,5)=length(temp(:,1)); for jh=2:length(temp(:,1)); temp(jh,4)=temp(jh-1,4)-temp(jh-1,2); % % mass temp(jh,5)=temp(jh-1,5)-1; % number end % hold on % Color_ran=[rand, rand, rand]; % % hold on % semilogx(temp(:,1),temp(:,5)./length(temp(:,5)),'MarkerSize',15,'Marker','.','LineStyle','none','Color', Color_ran) % set(gca,'FontSize',25,'FontName','times'); % xlabel({'major (um)'},'FontSize',30,'FontName','Times New Roman'); % ylabel({'% Number > major (um)'},'FontSize',30,'FontName','Times New Roman'); % box('on') % set(gca,'XScale','log') % saveas(gcf,'NumberAR','fig') % close all %delete repeating index Length=length(temp(:,1)); for t=1:Length; number= temp(t,1); for tt=t+1:Length if number == temp(tt,1); temp(tt,1)=0; end end end temp(find(temp(:,1)==0),:)=[];

Appendix C

165

hold on Color_ran=[rand, rand, rand]; % hold on plot(temp(:,1),temp(:,5)./Length,'MarkerSize',15,'Marker','.','LineStyle','none','Color', Color_ran) set(gca,'FontSize',25,'FontName','times'); xlabel({'major (um)'},'FontSize',30,'FontName','Times New Roman'); ylabel({'% Number > major (um)'},'FontSize',30,'FontName','Times New Roman'); box('on') % set(gca,'XScale','log') % saveas(gcf,'NumberAR','fig') close all %maybe you dont even want a fit! f = ezfit('1/2-1/2*a*erf((log(x)-b)/(2^0.5*c))'); % fits with log showfit(f); % % Wiebull f = ezfit('c*a*b*x^(b-1)*exp(-a*x^b)'); % showfit(f); f = ezfit('-c*exp(-a*x^b)'); % showfit(f); %exponential f = ezfit('a*exp(b*x)'); % showfit(f); f = ezfit('1/2-1/2*c*erf((x-a)/(2^0.5*b))'); % fits with a Gaussian showfit(f); %% Spatial Correlation %average spacing between closest neighbours clear all pack nearest=100; numpics=91; e_itt=1;

Appendix C

166

for i=1:numpics load(['stats1 (' num2str(i) ')']) % STATS(STATS(:,4)<5,:)=[]; if length(STATS(:,1))>nearest data(:,1)=round2(STATS(:,1),0.01); data(:,2)=round2(STATS(:,2),0.01); for j=1:1:length(data(:,1)) x0=data(j,1); y0=data(j,2); for k=1:1:length(data(:,1)) temp(k,1)= ((data(k,1)-x0).^2 + (data(k,2)- y0).^2)^0.5; end temp=sort(temp(:,1)); temp(find(temp(:,1)==0),:)=[]; % Color_ran=[rand, rand, rand]; % hold on % plot([1:length(temp)],temp,'Color', Color_ran) % set(gca,'FontSize',25,'FontName','times'); % xlabel({'Nearest Neighbour'},'FontSize',30,'FontName','Times New Roman'); % ylabel({'Distance'},'FontSize',30,'FontName','Times New Roman'); values(e_itt,1:nearest)= temp(1:nearest,1); % values(j,2,i)=y0; % for ht=1:length(temp) % nnumber of nearest neighbours % values(j,2+ht,i)=temp(ht,1); %nearest % end e_itt=e_itt+1; clearvars temp x0 y0 end

Appendix C

167

% VAL(i,1)=mean(values); % VAL(i,2)=median(values); clearvars data % clearvars data values % e_itt=1; end end hold on ecdf(values) %compute means and ignore the zeroes save values values load values clearvars AVG AVG(:,1)=1:nearest; AVG(:,2)=mean(values(:,:)); AVG(:,3)=median(values(:,:)); AVG(:,4)=prctile(values(:,:),10); AVG(:,5)=prctile(values(:,:),90); figure hold on plot(AVG(:,1),AVG(:,4),'.') hold on plot(AVG(:,1),AVG(:,3),'r.') hold on plot(AVG(:,1),AVG(:,5),'black.')

168

Bibliography

[1] Langlois, J.A., Rutland-Brown, W., & Wald, M.M. (2006). The epidemiology and

impact of traumatic brain injury: a brief overview. The Journal of head trauma

rehabilitation, 21(5), 375-378.

[2] Faul, M., Xu, L., Wald, M.M., Coronado, V.G. Traumatic brain injury in the

United States: emergency department visits, hospitalizations, and deaths. Atlanta

(GA): Centers for Disease Control and Prevention, National Center for Injury

Prevention and Control; 2010.

[3] Ling, G.S. (2011). Traumatic brain injury and spinal cord injury. Cecil Medicine.

24th ed. Philadelphia, PA: Saunders Elsevier.

[4] García-Altés, A., Pérez, K., Novoa, A., Suelves, J.M., Bernabeu, M., Vidal, J., ...

& Luque, J. C. G. (2012). Spinal cord injury and traumatic brain injury: a cost-of-

illness study. Neuroepidemiology, 39(2), 103-108.

BIBLIOGRAPHY

169

[5] Finkelstein, E., Corso, P., Miller, T., and Associates. The Incidence and Economic

Burden of Injuries in the United States. New York (NY): Oxford University

Press; 2006.

[6] Coronado, V.G., McGuire, L.C., Faul, M., Sugerman, D., Pearson, W. The

Epidemiology and Prevention of TBI (in press) 2012.

[7] Owens, B.D., Kragh Jr, J.F., Wenke, J.C., Macaitis, J., Wade, C.E., & Holcomb,

J.B. (2008). Combat wounds in operation Iraqi Freedom and operation Enduring

Freedom. Journal of Trauma and Acute Care Surgery, 64(2), 295-299.

[8] Maxwell, W.L., Povlishock, J.T., and Graham, D.L. (1997) A Mechanistic

Analysis of Nondisruptive Axonal Injury: A Review. Journal of Neurotrauma.

14(7), 419-440.

[9] Saatman, K.E., Duhaime, A.C., Bullock, R., Maas, A.I., Valadka, A., & Manley,

G.T. (2008). Classification of traumatic brain injury for targeted

therapies. Journal of neurotrauma, 25(7), 719-738.

[10] McCullagh, S., Oucherlony, D., Protzner, A., Blair, N., & Feinstein, A. (2001).

Prediction of neuropsychiatric outcome following mild trauma brain injury: an

examination of the Glasgow Coma Scale. Brain Injury, 15(6), 489-497.

[11] Gennarelli, T.A., Thibault, L.E., Adams, J.H., Graham, D.I., Thompson, C.J., &

Marcincin, R.P. (1982). Diffuse axonal injury and traumatic coma in the

primate. Annals of neurology, 12(6), 564-574.

BIBLIOGRAPHY

170

[12] Gennarelli, T.A., & Thibault, L.E. (1982). Biomechanics of acute subdural

hematoma. Journal of Trauma-Injury, Infection, and Critical Care, 22(8), 680-

686.

[13] Wright, R.M., Post, A., Hoshizaki, B., & Ramesh, K.T. (2013). A multiscale

computational approach to estimating axonal damage under inertial loading of the

head. Journal of neurotrauma, 30(2), 102-118.

[14] [Physics plays into every BMX jump]. (n.d.). [Photo of a fall during a BMX

jump]. www.i-am-bored.com/

[15] Cernak, I., & Noble-Haeusslein, L.J. (2010). Traumatic brain injury: an overview

of pathobiology with emphasis on military populations. Journal of Cerebral

Blood Flow & Metabolism, 30(2), 255-266.

[16] Silver, J.M., McAllister, T.W., & Yudofsky, S.C. (Eds.). (2011). Textbook of

traumatic brain injury. American Psychiatric Pub.

[17] [Brain and spinal cord]. (n.d.). [Illustration of brain and spinal cord].

http://www.moodylaw.com/personal-injury/catastrophic-injury

[18] [Spinal cord and Central Nervous System]. (n.d.). [Illustration of CNS].

http://paulandperkins.com/wp-content/uploads/2013/07/Brain-and-Spinal-Cord-

Injuries.jpg

[19] [Grey and white matter of brain and spinal cord]. (n.d.). [Anatomical drawings of

frontal and transverse slices of the brain and spinal cord with labeled white and

grey matter for each]. www.studyblue.com/

[20] Saladin, K.S., & Miller, L. (1998). Anatomy & physiology. McGraw-Hill.

BIBLIOGRAPHY

171

[21] [Neural cells of the CNS]. (n.d.). [Illustrations of neuroglia, neurons, and cerebral

vasculature]. www.studyblue.com/

[22] [Electrical impulse flow in neuron]. (n.d.). [Illustration of the electrical signal

flow through a neuron]. www.galleryhip.com/

[23] Fletcher, D.A., & Mullins, R.D. (2010). Cell mechanics and the cytoskeleton.

Nature, 463(7280), 485-492.

[24] [TEM of microtubules]. (n.d.). [Longitudinal slice of microtubules].

http://www.cytochemistry.net/cell-biology/tub8a.jpg

[25] Conde, C., & Cáceres, A. (2009) Microtubule assembly, organization and

dynamics in axons and dendrites. Nature Reviews Neuroscience, 10(5), 319-332.

[26] Desai, A., & Mitchison, T.J. (1997) Microtubule polymerization dynamics.

Annual review of cell and developmental biology, 13(1), 83-117.

[27] Mitchison, T., & Kirschner, M. (1984) Dynamic instability of microtubule

growth. Nature, 312(5991), 237-242.

[28] O'Brien, E.T., Salmon, E.D., Walker, R.A., & Erickson, H.P. (1990) Effects of

magnesium on the dynamic instability of individual microtubules.

Biochemistry, 29(28), 6648-6656.

[29] [Microtubule growth and shrinkage]. (n.d.) [Cryoelectron microscopic images and

illustrations of microtubule growth and shrinkage].

http://www.pha.jhu.edu/~ghzheng/old/webct/note2_4.htm

[30] Hoffman, P.N., Griffin, J.W., & Price, D.L. (1984). Control of axonal caliber by

neurofilament transport. The Journal of cell biology, 99(2), 705-714.

BIBLIOGRAPHY

172

[31] Hoffman, P.N., Cleveland, D.W., Griffin, J.W., Landes, P.W., Cowan, N.J., &

Price, D.L. (1987). Neurofilament gene expression: a major determinant of axonal

caliber. Proceedings of the National Academy of Sciences, 84(10), 3472-3476.

[32] Chang, R., Kwak, Y., & Gebremichael, Y. (2009) Structural properties of

neurofilament sidearms: sequence-based modeling of neurofilament

architecture. Journal of molecular biology, 391(3), 648-660.

[33] Mukhopadhyay, R., Kumar, S., & Hoh, J.H. (2004) Molecular mechanisms for

organizing the neuronal cytoskeleton. Bioessays, 26(9), 1017-1025.

[34] Rao, M.V., Campbell, J., Yuan, A., Kumar, A., Gotow, T., Uchiyama, Y., &

Nixon, R.A. (2003). The neurofilament middle molecular mass subunit carboxyl-

terminal tail domains is essential for the radial growth and cytoskeletal

architecture of axons but not for regulating neurofilament transport rate. The

Journal of cell biology, 163(5), 1021-1031.

[35] Smith, D.H., & Meaney, D.F. (2000). Axonal damage in traumatic brain injury.

The Neuroscientist, 6(6), 483-495.

[36] Hisanaga, S.I., & Hirokawa, N. (1990). Molecular architecture of the

neurofilament: II. Reassembly process of neurofilament L protein in

vitro. Journal of molecular biology, 211(4), 871-882.

[37] Carden, M.J., Trojanowski, J.Q., Schlaepfer, W.W., & Lee, V.M.Y. (1987). Two-

stage expression of neurofilament polypeptides during rat neurogenesis with early

establishment of adult phosphorylation patterns. J. Neurosci, 7(11), 3489-3504.

BIBLIOGRAPHY

173

[38] Lee, V.M., Otvos, L., Carden, M.J., Hollosi, M., Dietzschold, B., & Lazzarini,

R.A. (1988). Identification of the major multiphosphorylation site in mammalian

neurofilaments. Proceedings of the National Academy of Sciences, 85(6), 1998-

2002.

[39] Taber, K.H., and Hurley, R.A. (2007) Traumatic Axonal Injury: Atlas of Major

Pathways. J. Neuropsychiatry Clin. Neurosci.19, iv-104.

[40] LaPlaca, M.C., Simon, C.M., Prado, G.R., and Cullen, D.K. (2007) CNS injury

biomechanics and experimental models. Prog. Brain Res., 161, 13–26.

[41] Gennarelli, T.A. (1993). Cerebral concussion and diffuse brain injuries. Head

injury, 2, 108-124.

[42] Graham, D.I., Adams, J.H., Nicoll, J.A.R., Maxwell, W.L., & Gennarelli, T.A.

(1995) The nature, distribution and causes of traumatic brain injury. Brain

Pathology, 5(4), 397-406.

[43] Povlishock, J.T., Becker, D.P., Cheng, C.L.Y., & Vaughan, G.W. (1983) Axonal

change in minor head injury. Journal of Neuropathology & Experimental

Neurology, 42(3), 225-242.

[44] Povlishock, J.T. (1992) Traumatically induced axonal injury: pathogenesis and

pathobiological implications. Brain pathology (Zurich, Switzerland), 2(1), 1-12.

[45] Maxwell, W.L., & Graham, D.I. (1997) Loss of axonal microtubules and

neurofilaments after stretch-injury to guinea pig optic nerve fibers. Journal of

neurotrauma, 14(9), 603-614.

BIBLIOGRAPHY

174

[46] Johnson, V.E., Stewart, W., & Smith, D.H. (2012) Axonal pathology in traumatic

brain injury. Experimental Neurology, 246, 35-43.

[47] Jafari, S.S., Maxwell, W.L., Neilson, M., & Graham, D.I. (1997) Axonal

cytoskeletal changes after non-disruptive axonal injury. Journal of

neurocytology, 26(4), 201-221.

[48] Jafari, S.S., Nielson, M., Graham, D.I., & Maxwell, W.L. (1998) Axonal

cytoskeletal changes after nondisruptive axonal injury. II. Intermediate sized

axons. Journal of neurotrauma, 15(11), 955-966.

[49] Serbest, G., Burkhardt, M.F., Siman, R., Raghupathi, R., & Saatman, K.E. (2007)

Temporal profiles of cytoskeletal protein loss following traumatic axonal injury in

mice. Neurochemical research, 32(12), 2006-2014.

[50] Chung, R.S., Staal, J.A., McCormack, G.H., Dickson, T.C., Cozens, M.A.,

Chuckowree, J.A., & Vickers, J.C. (2005) Mild axonal stretch injury in vitro

induces a progressive series of neurofilament alterations ultimately leading to

delayed axotomy. Journal of neurotrauma, 22(10), 1081-1091.

[51] Hosmane, S., Fournier, A., Wright, R., Rajbhandari, L., Siddique, R., Yang, I.H.,

& Thakor, N. (2011) Valve-based microfluidic compression platform: single axon

injury and regrowth. Lab on a Chip, 11(22), 3888-3895.

[52] Büki, A., & Povlishock, J. T. (2006) All roads lead to disconnection–Traumatic

axonal injury revisited. Acta neurochirurgica, 148(2), 181-194.

BIBLIOGRAPHY

175

[53] Maxwell, W.L., Domleo, A., McColl, G., Jafari, S.S., & Graham, D.I. (2003)

Post-acute alterations in the axonal cytoskeleton after traumatic axonal injury.

Journal of neurotrauma, 20(2), 151-168.

[54] Tang-Schomer, M.D., Patel, A.R., Baas, P.W., & Smith, D.H. (2010) Mechanical

breaking of microtubules in axons during dynamic stretch injury underlies

delayed elasticity, microtubule disassembly, and axon degeneration. The FASEB

Journal, 24(5), 1401-1410.

[55] LaPlaca, M.C., Cullen, D.K., McLoughlin, J.J., & Cargill II, R.S. (2005). High

rate shear strain of three-dimensional neural cell cultures: a new in vitro traumatic

brain injury model. Journal of biomechanics, 38(5), 1093-1105.

[56] Pettus, E.H., & Povlishock, J.T. (1996). Characterization of a distinct set of intra-

axonal ultrastructural changes associated with traumatically induced alteration in

axolemmal permeability. Brain research, 722(1), 1-11.

[57] Christman, C.W., Grady, M.S., Walker, S.A., Holloway, K.L., & Povlishock, J.T.

(1994) Ultrastructural studies of diffuse axonal injury in humans. Journal of

neurotrauma, 11(2), 173-186.

[58] Koike, T., Yang, Y., Suzuki, K., & Zheng, X. (2008) Axon & dendrite

degeneration: Its mechanisms and protective experimental paradigms.

Neurochemistry international, 52(4), 751-760.

[59] Raghupathi, R., & Margulies, S.S. (2002). Traumatic axonal injury after closed

head injury in the neonatal pig. Journal of neurotrauma, 19(7), 843-853.

BIBLIOGRAPHY

176

[60] Saatman, K.E., Abai, B., Grosvenor, A., Vorwerk, C.K., Smith, D.H., & Meaney,

D.F. (2003). Traumatic axonal injury results in biphasic calpain activation and

retrograde transport impairment in mice. Journal of Cerebral Blood Flow &

Metabolism, 23(1), 34-42.

[61] Saatman, K.E., Graham, D.I., & McIntosh, T.K. (1998). The neuronal

cytoskeleton is at risk after mild and moderate brain injury. Journal of

neurotrauma, 15(12), 1047-1058.

[62] Thompson, H.J., Lifshitz, J., Marklund, N., Grady, M.S., Graham, D.I., Hovda,

D.A., & McIntosh, T.K. (2005). Lateral fluid percussion brain injury: a 15-year

review and evaluation. Journal of neurotrauma, 22(1), 42-75.

[63] Rajagopalan, J., Tofangchi, A., & Saif, M.T. (2010). Drosophila Neurons

Actively Regulate Axonal Tension In Vivo. Biophysical journal, 99(10), 3208-

3215.

[64] Morrison III, B., Saatman, K.E., Meaney, D.F., & McIntosh, T.K. (1998). In vitro

central nervous system models of mechanically induced trauma: a review. Journal

of neurotrauma, 15(11), 911-928.

[65] Kilinc, D., Gallo, G., & Barbee, K.A. (2008). Mechanically-induced membrane

poration causes axonal beading and localized cytoskeletal damage. Experimental

neurology, 212(2), 422-430.

[66] Tavalin, S.J., Ellis, E.F., & Satin, L.S. (1995). Mechanical perturbation of

cultured cortical neurons reveals a stretch-induced delayed depolarization.

Journal of neurophysiology, 74(6), 2767-2773.

BIBLIOGRAPHY

177

[67] Zhang, L., Rzigalinski, B.A., Ellis, E.F., & Satin, L.S. (1996). Reduction of

voltage-dependent Mg2+ blockade of NMDA current in mechanically injured

neurons. Science, 274(5294), 1921-1923.

[68] Wolf, J. A., Stys, P. K., Lusardi, T., Meaney, D., & Smith, D. H. (2001).

Traumatic axonal injury induces calcium influx modulated by tetrodotoxin-

sensitive sodium channels. The Journal of Neuroscience, 21(6), 1923-1930.

[69] Smith, D. H., Wolf, J. A., Lusardi, T. A., Lee, V. M. Y., & Meaney, D. F. (1999).

High tolerance and delayed elastic response of cultured axons to dynamic stretch

injury. The Journal of neuroscience, 19(11), 4263-4269.

[70] Morrison III, B., Cater, H. L., Benham, C. D., & Sundstrom, L. E. (2006). An in

vitro model of traumatic brain injury utilising two-dimensional stretch of

organotypic hippocampal slice cultures. Journal of neuroscience methods,150(2),

192-201

[71] Geddes-Klein, D. M., Schiffman, K. B., & Meaney, D. F. (2006). Mechanisms

and consequences of neuronal stretch injury in vitro differ with the model of

trauma. Journal of neurotrauma, 23(2), 193-204.

[72] Wang, H. C., & Ma, Y. B. (2010). Experimental models of traumatic axonal

injury. Journal of Clinical Neuroscience, 17(2), 157-162.

[73] Lipton, M. L., Gellella, E., Lo, C., Gold, T., Ardekani, B. A., Shifteh, K., ... &

Branch, C. A. (2008). Multifocal white matter ultrastructural abnormalities in

mild traumatic brain injury with cognitive disability: a voxel-wise analysis of

diffusion tensor imaging. Journal of neurotrauma, 25(11), 1335-1342.

BIBLIOGRAPHY

178

[74] MacDonald, C. L., Dikranian, K., Song, S. K., Bayly, P. V., Holtzman, D. M., &

Brody, D. L. (2007). Detection of traumatic axonal injury with diffusion tensor

imaging in a mouse model of traumatic brain injury. Experimental neurology,

205(1), 116-131.

[75] Begaz, T., Kyriacou, D. N., Segal, J., & Bazarian, J. J. (2006). Serum biochemical

markers for post-concussion syndrome in patients with mild traumatic brain

injury. Journal of neurotrauma, 23(8), 1201-1210.

[76] Lajtha A., Banik N., Ray S.K. Handbook of neurochemistry and molecular

neurobiology: brain and spinal cord trauma. 3rd ed. Springer; 2008; p. 343–61.

[77] Geddes, J. F., Whitwell, H. L., & Graham, D. I. (2000). Traumatic axonal injury:

practical issues for diagnosis in medicolegal cases. Neuropathology and applied

neurobiology, 26(2), 105-116.

[78] Abd-Elfattah Foda, M. A., & Marmarou, A. (1994). A new model of diffuse brain

injury in rats: Part II: Morphological characterization. Journal of neurosurgery,

80(2), 301-313.

[79] De Stefano, N., Matthews, P.M., Fu, L., Narayanan, S., Stanley, J., Francis, G.S.,

& Arnold, D.L. (1998) Axonal damage correlates with disability in patients with

relapsing-remitting multiple sclerosis. Results of a longitudinal magnetic

resonance spectroscopy study. Brain, 121(8), 1469-1477.

[80] Wolf, J. A., Meaney, D. F., Lusardi, T. A., & Smith, D. H. (1999). Calcium influx

and membrane permeability in axons after dynamic stretch injury in vitro. J.

Neurotrauma, 16, 966.

BIBLIOGRAPHY

179

[81] Li, S., Mealing, G. A., Morley, P., & Stys, P. K. (1999). Novel injury mechanism

in anoxia and trauma of spinal cord white matter: glutamate release via reverse

Na+-dependent glutamate transport. The Journal of neuroscience, 19(14), RC16.

[82] Waxman, S. G., Black, J. A., Ransom, B. R., & Stys, P. K. (1994). Anoxic injury

of rat optic nerve: ultrastructural evidence for coupling between Na+ influx and

Ca2+-mediated injury in myelinated CNS axons. Brain research, 644(2), 197-

204.

[83] Waxman, S. G., Black, J. A., Stys, P. K., & Ransom, B. R. (1992). Ultrastructural

concomitants of anoxic injury and early post-anoxic recovery in rat optic

nerve. Brain research, 574(1), 105-119.

[84] Stys, P. K., Waxman, S. G., & Ransom, B. R. (1992). Ionic mechanisms of anoxic

injury in mammalian CNS white matter: role of Na+ channels and Na+-Ca2+

exchanger. The Journal of neuroscience, 12(2), 430-439.

[85] Agrawal, S. K., & Fehlings, M. G. (1996). Mechanisms of secondary injury to

spinal cord axons in vitro: role of Na+, Na+-K+-ATPase, the Na+-H+ exchanger,

and the Na+-Ca 2+ exchanger. J Neurosci, 16, 545-552.

[86] Maxwell, W. L., Kosanlavit, R., McCreath, B. J., Reid, O., & Graham, D. I.

(1999). Freeze-fracture and cytochemical evidence for structural and functional

alteration in the axolemma and myelin sheath of adult guinea pig optic nerve

fibers after stretch injury. Journal of neurotrauma, 16(4), 273-284.

[87] Maxwell, W. L., McCreath, B. J., Graham, D. I., & Gennarelli, T. A. (1995).

Cytochemical evidence for redistribution of membrane pump calcium-ATPase

BIBLIOGRAPHY

180

and ecto-Ca-ATPase activity, and calcium influx in myelinated nerve fibres of the

optic nerve after stretch injury. Journal of neurocytology, 24(12), 925-942.

[88] McCracken, E., Hunter, A. J., Patel, S., Graham, D. I., & Dewar, D. (1999).

Calpain activation and cytoskeletal protein breakdown in the corpus callosum of

head-injured patients. Journal of neurotrauma, 16(9), 749-761.

[89] Büki, A., Siman, R., Trojanowski, J. Q., & Povlishock, J. T. (1999). The Role of

Calpail-Mediated Spectrin Proteolysis in Traumatically Induced Axonal Injury.

Journal of Neuropathology & Experimental Neurology, 58(4), 365-375.

[90] Posmantur, R., Hayes, R. L., Dixon, C. E., & Taft, W. C. (1994). Neurofilament

68 and neurofilament 200 protein levels decrease after traumatic brain


[91] Pike, B. R., Zhao, X., Newcomb, J. K., Posmantur, R. M., Wang, K. K., & Hayes,

R. L. (1998). Regional calpain and caspase‐3 proteolysis of α‐spectrin after

traumatic brain injury. Neuroreport, 9(11), 2437-2442.

[92] Pierce, J. E., Trojanowski, J. Q., Graham, D. I., Smith, D. H., & McIntosh, T. K.

(1996). Immunohistochemical characterization of alterations in the distribution of

amyloid precursor proteins and beta-amyloid peptide after experimental brain

injury in the rat. The Journal of neuroscience, 16(3), 1083-1090.

[93] Sherriff, F. E., Bridges, L. R., & Sivaloganathan, S. (1994). Early detection of

axonal injury after human head trauma using immunocytochemistry for β-amyloid

precursor protein. Acta neuropathologica, 87(1), 55-62.

BIBLIOGRAPHY

181

[94] Gentleman, S. M., Nash, M. J., Sweeting, C. J., Graham, D. I., & Roberts, G. W.

(1993). β-Amyloid precursor protein (βAPP) as a marker for axonal injury after

head injury. Neuroscience letters, 160(2), 139-144.

[95] Blumbergs, P. C., Scott, G., Vis, J. M., Wainwright, H., Simpson, D. A., &

McLean, A. J. (1995). Topography of axonal injury as defined by amyloid

precursor protein and the sector scoring method in mild and severe closed head


[96] Yaghmai, A., & Povlishock, J. (1992). Traumatically induced reactive change as

visualized through the use of monoclonal antibodies targeted to neurofilament

subunits. Journal of Neuropathology & Experimental Neurology, 51(2), 158-176.

[97] Grady, M. S., Mclaughlin, M. R., Christman, C. W., Valadka, A. B., Fligner, C.

L., & Povlishock, J. T. (1993). The use of antibodies targeted against the

neurofilament subunits for the detection of diffuse axonal injury in humans.

Journal of Neuropathology & Experimental Neurology, 52(2), 143-152.

[98] Hall, G. F., & Lee, V. M. Y. (1995). Neurofilament sidearm proteolysis is a

prominent early effect of axotomy in lamprey giant central neurons. Journal of

Comparative Neurology, 353(1), 38-49.

[99] Povlishock, J. T., Marmarou, A., McIntosh, T., Trojanowski, J. Q., & Moroi, J.

(1997). Impact acceleration injury in the rat: evidence for focal axolemmal change

and related neurofilament sidearm alteration. Journal of Neuropathology &

Experimental Neurology, 56(4), 347-359.

BIBLIOGRAPHY

182

[100] Chen, X. H., Meaney, D. F., Bai-Nan, X. U., Nonaka, M., Mcintosh, T. K., Wolf,

J. A., ... & Smith, D. H. (1999). Evolution of neurofilament subtype accumulation

in axons following diffuse brain injury in the pig. Journal of Neuropathology &

Experimental Neurology, 58(6), 588-596.

[101] Okonkwo, D. O., Pettus, E. H., Moroi, J., & Povlishock, J. T. (1998). Alteration

of the neurofilament sidearm and its relation to neurofilament compaction

occurring with traumatic axonal injury. Brain research, 784(1), 1-6.

[102] Povlishock, J. T., Giebel, M. L., & Pettus, E. (1995). Traumatically induced

alterations in axolemma permeability are associated with a distinct subset of

intraaxonal cytoskeletal changes. Journal of Neurotrauma, 12, 417.

[103] Pettus, E. H., Christman, C. W., Giebel, M. L., & Povlishock, J. T. (1994).

Traumatically induced altered membrane permeability: its relationship to

traumatically induced reactive axonal change. Journal of neurotrauma, 11(5),

507-522.

[104] Sia, S. K., & Whitesides, G. M. (2003). Microfluidic devices fabricated in poly

(dimethylsiloxane) for biological studies. Electrophoresis, 24(21), 3563-3576.

[105] Hosmane, S. (2012). Innovative Technologies to Study Mechanisms of CNS

Axonal Degeneration. (Unpublished doctoral dissertation). Johns Hopkins

University, Baltimore, MD.

[106] Banker, G., & Goslin, K. Culturing Nerve Cells, 2nd edn, The MIT Press,

Cambridge, MA, 1998.

BIBLIOGRAPHY

183

[107] Hosmane, S., Yang, I. H., Ruffin, A., Thakor, N., & Venkatesan, A. (2010).

Circular compartmentalized microfluidic platform: study of axon–glia

interactions. Lab on a Chip, 10(6), 741-747.

[108] Bernal, R., Pullarkat, P. A., & Melo, F. (2007). Mechanical properties of axons.

Physical review letters, 99(1), 018301.

[109] Anstis, G. R., Chantikul, P., Lawn, B. R., & Marshall, D. B. (1981). A critical

evaluation of indentation techniques for measuring fracture toughness: I, direct

crack measurements. Journal of the American Ceramic Society, 64(9), 533-538.

[110] Lee, S. J., Chan, J. C., Maung, K. J., Rezler, E., & Sundararajan, N. (2007).

Characterization of laterally deformable elastomer membranes for microfluidics.

Journal of Micromechanics and Microengineering, 17(5), 843.

[111] Coleman, M. (2005). Axon degeneration mechanisms: commonality amid

diversity. Nature Reviews Neuroscience, 6(11), 889-898.

[112] Verma, P., Chierzi, S., Codd, A. M., Campbell, D. S., Meyer, R. L., Holt, C. E., &

Fawcett, J. W. (2005). Axonal protein synthesis and degradation are necessary for

efficient growth cone regeneration. The Journal of neuroscience,25(2), 331-342.

[113] Fournier, A., Rajbhandari, L., Shrestha, S., Venkatesan, A., & Ramesh, K.T.

(2014). In-Vitro and In Situ Visualization of Cytoskeletal Deformation under

Load: Traumatic Axonal Injury. In review.

[114] Gurtin, M. E., Fried, E., & Anand, L. (2010). The mechanics and thermodynamics

of continua. Cambridge University Press.

BIBLIOGRAPHY

184

[115] Ludin, B., & Matus, A. (1998). GFP illuminates the cytoskeleton. Trends in cell

biology, 8(2), 72-77

[116] Banik, N.L., Matzelle, D.C., Gantt-Wilford, G., Osborne, A., & Hogan, E.L.

(1997) Increased calpain content and progressive degradation of neurofilament

protein in spinal cord injury. Brain research, 752(1), 301-306

[117] Ouyang, H., Nauman, E., & Shi, R. (2010, June) Contribution of cytoskeletal

elements to the mechanical property of axons. In Micro/Nano Symposium

(UGIM), 2010 18th Biennial University/Government/Industry (pp. 1-5). IEEE

[118] Fournier, A., Hogan, J.D., Rajbhandari, L., Shrestha, S., Venkatesan, A., &

Ramesh, K.T. (2014). Neurofilament and Microtubule Evolution under Focal

Axon Compression. In review.

[119] De Mey, J., Moeremans, M., Geuens, G., Nuydens, R., & De Brabander, M.

(1981). High resolution light and electron microscopic localization of tubulin with

the IGS (immuno gold staining) method. Cell biology international reports, 5(9),

889-899.

[120] De Mey, J., Lambert, A. M., Bajer, A. S., Moeremans, M. D., & De Brabander,

M. (1982). Visualization of microtubules in interphase and mitotic plant cells of

Haemanthus endosperm with the immuno-gold staining method. Proceedings of

the National Academy of Sciences, 79(6), 1898-1902.

[121] De Mey, J., & Moeremans, M. (1986). The preparation of colloidal gold probes

and their use as marker in electron microscopy. In Advanced techniques in

biological electron microscopy III (pp. 229-271). Springer Berlin Heidelberg.

BIBLIOGRAPHY

185

[122] Hermann R, Walther P, Müller M (1996) Immunogold labeling in scanning

electron microscopy. Histochemistry and Cell Biology, 106(1), 31-39

[123] Aberts B, Johnson A, Lewis J, Raff M, Roberts K, and Walter P. Molecular

biology of the cell (5th ed.). New York (NY): Garland Science; 2008

[124] Maxwell, W. L. (1996). Histopathological changes at central nodes of Ranvier

after stretch‐injury. Microscopy research and technique, 34(6), 522-535.

[125] Ahmadzadeh, H., Smith, D. H., & Shenoy, V. B. (2014). Viscoelasticity of Tau

Proteins Leads to Strain Rate-Dependent Breaking of Microtubules during Axonal

Stretch Injury: Predictions from a Mathematical Model. Biophysical

journal, 106(5), 1123-1133.

[126] Nixon, R. A., Paskevich, P. A., Sihag, R. K., & Thayer, C. Y. (1994).

Phosphorylation on carboxyl terminus domains of neurofilament proteins in

retinal ganglion cell neurons in vivo: influences on regional neurofilament

accumulation, interneurofilament spacing, and axon caliber. The Journal of cell

biology, 126(4), 1031-1046.

[127] Meaney, D.F., Smith, D.H., Shreiber, D.I., Bain, A.C., Miller, R.T., Ross, D.T., &

Gennarelli, T.A. (1995). Biomechanical analysis of experimental diffuse axonal


[128] Elkin, B. & Morrison, B III. (2007) Region-specific tolerance criteria for the

living brain. Stapp Car Crash J, 51, 127-138.

[129] Smith, D. H., Hicks, R., & Povlishock, J. T. (2013). Therapy development for

diffuse axonal injury. Journal of neurotrauma, 30(5), 307-323.

BIBLIOGRAPHY

186

[130] Saatman, K. E. (2012). The complexity of traumatic axonal injury. Traumatic

Brain and Spinal Cord Injury: Challenges and Developments, 138.

[131] Cai, J., Jones, D.P. (1998) Superoxide in apoptosis. Mitochondrial generation

triggered by cytochrome c loss. J Biol Chem 273(19), 11401–11404

[132] Reed, J.C. (1998) Bcl-2 family proteins. Oncogene 17(25), 3225–3236

[133] Saikumar, P., Dong, Z., Weinberg, J.M., Venkatachalam, M.A. (1998)

Mechanisms of cell death in hypoxia reoxygenation injury. Oncogene 17(25):

3341–3349

[134] Bartus, R.T., Hayward, N.J., Elliott, P. J., Sawyer, S. D., Baker, K.L., Dean, R.L.,

Akiyama, A., Straub, J.A., Harbeson, S.L., & Li, Z. (1994). Calpain inhibitor

AK295 protects neurons from focal brain ischemia. Effects of postocclusion intra-

arterial administration. Stroke, 25(11), 2265-2270.

[135] Hong, S.C., Goto, Y., Lanzino, G., Soleau, S., Kassell, N.F., & Lee, K.S. (1994).

Neuroprotection with a calpain inhibitor in a model of focal cerebral

ischemia. Stroke, 25(3), 663-669.

[136] Lee, K.S., Yanamoto, H., Fergus, A., Hong, S.C., Kang, S.D., Cappelletto, B.,

Toyoda, T., Kassell, N.F., Bavbek, M., & Kwan, A. L. (1997). Calcium‐Activated

Proteolysis as a Therapeutic Target in Cerebrovascular Diseasea. Annals of the

New York Academy of Sciences, 825(1), 95-103.

[137] Markgraf, C.G., Velayo, N.L., Johnson, M.P., McCarty, D.R., Medhi, S., Koehl,

J.R., Chmielewski, P.A. & Linnik, M.D. (1998). Six-hour window of opportunity

for calpain inhibition in focal cerebral ischemia in rats. Stroke, 29(1), 152-158.

BIBLIOGRAPHY

187

[138] Banik, N.L., Shields, D.C., Ray, S., Davis, B., Matzelle, D., Wilford, G., &

Hogan, E.L. (1998). Role of Calpain in Spinal Cord Injury: Effects of Calpain and

Free Radical Inhibitorsa. Annals of the New York Academy of Sciences, 844(1),

131-137.

[139] James, T., Matzelle, D., Bartus, R., Hogan, E. L., & Banik, N.L. (1998). New

inhibitors of calpain prevent degradation of cytoskeletal and myelin proteins in

spinal cord in vitro. Journal of neuroscience research, 51(2), 218-222.

[140] Kruman, I.I., & Mattson, M. P. (1999). Pivotal role of mitochondrial calcium

uptake in neural cell apoptosis and necrosis. Journal of neurochemistry, 72(2),

529-540.

[141] Posmantur, R., Kampfl, A., Siman, R., Liu, S.J., Zhao, X., Clifton, G.L., & Hayes,

R.L. (1997). A calpain inhibitor attenuates cortical cytoskeletal protein loss after

experimental traumatic brain injury in the rat. Neuroscience, 77(3), 875-888.

[142] Saatman, K.E., Murai, H., Bartus, R.T., Smith, D.H., Hayward, N.J., Perri, B.R.,

& McIntosh, T.K. (1996). Calpain inhibitor AK295 attenuates motor and

cognitive deficits following experimental brain injury in the rat. Proceedings of

the National Academy of Sciences, 93(8), 3428-3433.

[143] Saatman, K.E., Zhang, C., Bartus, R.T., & McIntosh, T.K. (2000). Behavioral

efficacy of posttraumatic calpain inhibition is not accompanied by reduced

spectrin proteolysis, cortical lesion, or apoptosis. Journal of Cerebral Blood Flow

& Metabolism, 20(1), 66-73.

188

Vita

Adam John Fournier was born at Camp Darby, Italy in 1980. He earned his B.S.

degree in Biomedical Engineering from Wright State University in 2002 and his M.S.

degree in Biomedical Engineering from Purdue University in 2004. In the following

years, Adam worked for private industry and the U.S. Army Aberdeen Test Center (ATC)

evaluating personal protection equipment and assessing the performance of vehicular

platforms for survivability and lethality. His position within ATC provided an

opportunity to return to graduate school by enrolling in the Mechanical Engineering

Ph.D. program at Johns Hopkins University in 2010 where he received his M.S.E. degree

at Johns Hopkins University in 2012. His research interests are injury biomechanics and

multiscale experimental approaches for understanding biological tissues and cells. On

top of his research, Adam has served as a board member for the Mechanical Engineering

Graduate Association at Johns Hopkins at Johns Hopkins University from 2011-2013.