DEVELOPMENT AND CHARACTERIZATION OF A GRADED, IN VIVO, COMPRESSIVE, MURINE MODEL OF SPINAL CORD INJURY

Mital Joshi

A thesis submitted in conformity with the requirements for the degree of Master of Science at the lnstitute of Medical Science,

University of Toronto

O Copyright by Mital Joshi 2000

National Library Bibliothèque nationale l*l of Canada du Canada

Acquisitions and Acquisitions et Bibliographie Services seivices bibliographiques

395 Wellington Street 395, rue Wellington Ottawa ON K1A ON4 Ottawa ON K1 A ON4 Canada Canada

The author has granted a non- L'auteur a accordé une licence non exclusive licence allowing the exclusive permettant à la National Library of Canada to Bibliothèque nationale du Canada de reproduce, loan, distribute or sel1 reproduire, prêter, distribuer ou copies of this thesis in microfonn, vendre des copies de cette thèse sous paper or electronic formats. la forme de microfiche/film, de

reproduction sur papier ou sur format électronique.

The author retains ownership of the L'auteur conserve la propriété du copyright in this thesis. Neither the droit d'auteur qui protège cette thèse. thesis nor substantial extracts fiom it Ni la thèse ni des extraits substantiels may be printed or otherwise de celle-ci ne doivent être imprimés reproduced without the author's ou autrement reproduits sans son permission. autorisation.

DEVELOPMENT AND CHARACTERlZATlON OF A GRADED, IN VIVO,

COMPRESSIVE, MURINE MODEL OF SPINAL CORD INJURY

By Mital Joshi, November 2000, Master of Science, lnstitute of Medical Science,

University of Toronto

ABSTRACT

In order to take advantage of various genetically manipulated mice available to study

the pathophysiology of spinal cord injury (SCI) we adapted an extradural clip

compression injury model to the mouse. The dimensions of the modified aneurysrn clip

blades were customized for application to the rnouse spinal cord. Three clips with

different springs were made to produce differing magnitudes of closing force (39, 89 and

249). The clips were calibrated regularly to ensure that the closing force rernained

constant. The surgical procedure involved a laminectomy at T3 and T4, followed by

extradural application of the clip at this level for one minute to produce SCI. Three injury

severities (39, 89 and 24g), sham (passage of dissector extradurally at T3-4) and

transection control groups were examined (n=l Zgroup). Quantitative behavioural

assessments using the Basso, Beattie and Bresnahan (BBB) and lnclined Plane (IP)

tests showed a significant graded increase in neurological deficits with increasing

severity of injury (pc0.0001, two-way repeated measures ANOVA). By Day 14, the motor

recovery of the mice plateaued. Morphomettic analyses of H&E/Luxol Fast Blue stained

sections at every 50pm from the injury epicenter indicated that with greater injury

severity there was a progressive decrease in residual tissue (p<0.0001, two-way

ANOVA). Counts of retrogradely labeled neurons in the brainstern, midbrain and cortex

also indicated decreased integrity of descending axons through the injury epicenter with

increasing injury seventy (p<0.0001, one-way ANOVA). Correlations between the

functional deficits and anatomical outcome measure were identified. Specifically, the

BBB openfield locomotor test was found to preferentially assess the integrity of

raphespinal and corticospinal tracts, whereas the inclined plane preferentially assessed

the integrity of vestibulospinal and rubrospinal tracts in mice. This novel, graded

compressive model of SC1 will facilitate future studies of the pathological mechanisms of

SC1 using transgenic and knockout murine systems.

ACKNSWLEDGEMENTS

I am very grateful to my supervisor, Dr. Michael Fehlings, for his excellent guidance,

support and for providing me with an opportunity to expand my horizons into spinal cord injury

research. He has been and will be both a mentor and a role mode1 as I continue my training in

research and medicine.

I am also very thankful to the members of my prograrn advisory committee, Dr. John

Roder and Dr. Charles Tator for their invaluabte support and advice.

1 would like to thank Dr. Raad Nashrni, Dr. Qi Wan and Dr. Andrei Krassioukov for their

help in teaching me the numerous intricacies of spinal cord injury research. Also, I am very

appreciative of the technical assistance of Karen Scales. In the laboratory, I was very fortunate

to work with a dedicated team: Dr. Jennifer Brunton, Dr. Steven Casha, Lori Edwards, Edward

Eng, Dr. Chunbo Hu, Dr. Yang Liu, Nav Persaud, Dr. Nicolas Phan, Gwen Schwartz and Dr.

Wenru Yu, ail of whom taught me important, unique lesson(s) about research and helped me in

countless ways.

The studentship funding provided by the Ontario Neurotrauma Foundation was also

greatly appreciated.

Finally, I thank God and I will always cherish the love and support provided by my farnily

and close friends.

iii

TABLE OF CONTENTS

INTRODUCTION 1

RATION ALE STATEMENT OF THE HYPOTHESES

Overall Hypothesis: Specific Hypotheses:

OBJECTIVES PATHOLOGY OF SPINAL CORD INJURY PATHOPHYSIOLOGY OF SPINAL CORD INJURY

Ischemia Altered lonic Homeostasis Lipid Hydrolysis, Lipid Peroxidation and Free Radical Release Glutamate Excitotoxicity Inflammation Apoptosis

IN VIVO EXPERIMENTAL SPINAL CORD INJURY MODELS Kinetic Impact Models

Weight-Drop Models Electromechanical Contusion Model Clip Compression Model Crush Models l nf latable Balloon/Cuff Other Kinetic Models

Static Load Models Transection Models Other SC1 Models

OUTCOME MEASURES TO ASSESS SEVERITY OF CHRONIC SPINAL CORD INJURY IN VIVO Neurological Neurop hysiological HistopathologicaVAnatomical

IN VIVO MURINE MODELS OF SPINAL CORD INJURY

CLIP DESIGN AND CALIBRATION 26 MOUSE SPINAL CORD INJURIES 27 QUANTITATIVE ASSESSMENTS OF MOTOR FUNCTION 28 RETROGRADE LABELING OF NEURONS 29 TISSUE EXTRACTION AND PROCESSING 29 MORPHOMETRIC ASSESSMENTS OF SPINAL CORD TISSUE 30 COUNTS OF RETROGRADELY ~ B E L E D NEURONS IN KEY MOTOR AREAS OF THE CEREBRUM AND BRAINSTEM 31 STA~ST~CAL ANALYSE 31

EFFECTS OF COMPRESSIVE SPINAL CORD INJURY SE VER^ ON MORTALITY AND NEUROLOGICAL FUNCTION 38 QUALITATIVE EXAMINATION OF INJURY SITE PATHOLOGY FOLLOWING COMPRESSIVE SPINAL CORD INJURY 39

Acute 39 Chronic 40

QUANTITATIVE MORPHOMETRIC EXAMINATION OF THE COMPRESSIVE SPINAL CORD INJURY SITE 40 EFFECTS OF COMPRESSIVE SPINAL CORD INJURY SEVERITY ON THE ~NTEGRIW OF DESCENDING MOTOR TRACTS 42 RELATIONSHIPS BETWEEN NEUROLOGICAL FUNCTION, HISTOPATHOLOGY AND INJURY SEVERITY 43

ADVANTAGES AND DISADVANTAGES OF THE CLIP COMPRESSIVE MODEL OF SC1 BEHAVIOURAL RESPONSES OF MICE TO SPINAL CORD INJUFIY PATHOLOGICAL RESPONSES OF THE MOUSE SPINAL CORD TO INJURY RELATIONSHIP OF TRACTS TO BBB AND INCLINED PLANE SCORES LIMlTAT10NS OF TRACT TRACING TECHNIQUES METHODOLOGICAL CONSIDERATIONS APPLICATION OF MURINE MODELS TO SC1 FUTURE DIRECTIONS

CONCLUSIONS 72

REFERENCES 73

LIST OF FIGURES Page

FIGURE 1: Mechanisms of injury involved in spinal cord trauma

FIGURE 2: Schematic diagram of the mouse clip dimensions

FIGURE 3: in vivo application of the mouse clip

FIGURE 4: Neurological recovery aWer spinal cord injury

FIGURE 5: Lesion site pathology afier spinal cord injury: Acute Time points

FIGURE 6: Lesion site pathology after spinal cord injury: Chronic Time point

FIGURE 7: Distribution of residual tissue following spinal cord injury

FIGURE 8: Cross-sectional area of cavitation and residual tissue at the injury epicentef

FIGURE 9: Fluoro-Goid labeled neurons in the red nucleus after spinal cord injury

FIGURE 10: Mean counts of Fluoro-Gold labeled neurons in the mouse brain

LIST OF TABLES

TABLE 1 : Murine models of spinal cord injury

TABLE 2: BBB hindlirnb locomotor rating scale

TABLE 3: Surnmary of SNK post-hoc comparisions of BBB scores

Page

25

TABLE 4: Summary of SNK post-hoc cornparisions of IP scores 47

TABLE 5: Counts of neurons retrogradely labeled by Fluoro-Gold 60

TABLE 6: lndependent correlations between histopathology and behaviour 61

TABLE 7: Multiple linear regression correlations between cell counts and behaviour 61

vii

AMPA ANOVA ATP BBB C O C

CC

CBS DNA 9 GFAP GluR H&E IP kg L mg min mm MEP mGluR MRI msec N NF200 nm NMDA NO NYU PMN R R~ SC1 S E M . SNK SSEP T TNF

al p ha-amino-3-hydroxy-5-met h yl-4-isoxazoe pro pionic acid analysis of variance adenosine triphosphate Basso, Beattie and Bresnahan cervical degrees Celsius cubic centimeter combined behavioural score deoxyribonucleic acid gram glial fibrilliary acid protein non-NMDA ionotropic glutamate receptor hematoxylin and eosin lnclined plane kilograrn lumbar milligram minute millimeter motor evoked potential metabotropic glutamate receptor magnetic resonance imaging millisecond Newton neurofilament protein (200kDa) nanometer N-methyl-D-aspartate nitric oxide New York University polymorphonuclear neutrophils Pearson correlation coefficient stepwise multiple linear correlation coefficient spinal cord injury standard error of the mean Student-Newman-Keuls test somatosensory evoked potential t ho racic tumour necrosis factor micrometer wildtype allele pair mutant allele pair

viii

INTRODUCTION

Rationak

Spinal cord injury is an important cause of morbidity and mortality, particularly in young

adults and children. In North America more than 10 000 peoplelyear sustain a spinal cord injury

(SC!). Over 200,000 people live with serious and permanent disabilities as a result of these

injuries, resulting in profound socioeconomic costs. Current treatments for SC1 include

rnethylprednisolone and GM-1 ganglioside but these result in only minimal neurological

improvements. lmproved neuroprotective strategies should be directed at the rnechanisms

involved in the loss of tissue and the neurological deficits sustained. In order to ascertain which

molecules should be targeted, the role of specific genes in the pathophysiology of spinal cord

injury must first be examined. However, one of the limitations to rapid progress is the tardy

development of pharmaceutical agents specific for individual gene products. Furthemore, the

pharmaceutical agents which have been developed have been relatively non-specific in their

targets, thus not facilitating studies to identify key genes in the pathophysiology of SCI. An ideal

solution would be to apply recent breakthroughs in mouse genetics to target a given gene in

mice and create mice that lack a detectable gene product. These knockout mice for specific

gene product(s) could then be tested for their neurologicai, physiological, cellular and molecular

responses to SCI. In addition to applying targeted knockout mice to study the

pathophysiological mechanisms of SCI, mice with naturally occurring/spontaneouç mutations

could also be studied. In order to facilitate SC1 studies with genetically altered mouse systems,

a reliable, reproducible, graded model of SC1 must first be established in the mouse.

Overall Hypofhesis:

A graded, murine model of spinal cord injury can be established that is reliable and

reproducible which will ultimately facilitate studies of spinal cord injury in mutant mice.

Specific Hypo theses:

1. lncreasing levels of compressive spinal cord injury severity will result in increasing

neurological deficits following injury.

2. Increasing levels of compressive spinal cord injury severity will result in increased tissue

loss and descending supraspinal neuronal projections involved in motor function.

. . iect ives

In order to develop and charactenze a reliable, reproducible, graded model of spinal cord injury

in the murine spinal cord the following specific aims were exarnined:

To design a custom modified aneurysm clip for application to the murine spinal cord to

produce three severities of SC1 (39, 89 and 249).

To per fon neurological analyses (Basso, Beattie and Breshnahan Hindlimb Locomotor

Test and lnclined Plane Test) in order to assess the degree of injury seventy following three

levels of SC1 (39, 89 and 24g), sham and transection controls.

To perfom hstopathological examinations by the retrograde tracing of descending motor

tracts using the fluorescent tracer, Fluoro-Gold, in order to assess the integrity of

descending motor fiben through the injury epicenter in three injury severities (39, 89 and

24g), sham and transection controls.

To perform morphornetric injury site tissue analyses in order to quantify lesion size and

neuronal tissue sparing at the injury epicenter, rostral and caudal to the injury epicenter in

the three injury severities (39, 89 and 249) and sham controls.

Pathologv of Spinal Cord 1-

Spinal cord injury is initiated by a primary mechanical insult to the spinal cord. The

primary injury mechanisms include: compression, contusion, stretch and laceration, though

most often, SC1 has a component of impact plus sustained compression. The primary injury is

mediated by vertebral burst fractures, vertebral fracture dislocation, missile injuries and

ruptured discs leading to tissue distortion (Tator, 1995). Following the pnmary injury, a

sequelae of pathophysiological events occur at and adjacent to the injury site.

lnitially following SCI, edema occurs and many petechial hemorrhages are observed,

especially in the gray matter (Tator, 1995). At the injury epicenter, where maximal tissue

damage occurs, there is an infiltration and activation of macrophageslmicroglia and

extravasation of neutrophils as the blood-central nervous system bamer is compromised

(Bunge et al., 1997; Kakulas, 1984). These invading immune cells then proceed to

phagocytose debris from necrotic tissue and after they leave, a cystic cavity remains. Around

the central necrotic zone, a translational zone of damage and a peripheral subpial rim of

residual white matter are found. There exists a gradient of necrosis extending both rostrally and

caudally from the focal epicenter (Balentine, 1978; Basso et al.. 1996; Bresnahan et al., 1976;

Bunge et al., 1997; Griffiths and McCulloch, 1983; Kakulas, 1984; Tator, 1995). Astroglial scar

tissue, often termed gliosis, is found to serve as a boundary between the central cavity and

adjacent tissue (Kakulas, 1984). Gliosis is marked by the hypertrophy of astrocytes, as shown

by the immunohistochemical labeling of GFAP (Bunge et al., 1997; lizuka et al., 1987).

Furthermore, it has been shown that with injuries of increasing magnitude, the amount of tissue

damage also increases (Basso et al.. 1996; Bresnahan et al., 1976; Fehlings and Tator, 1995;

Grifiths and McCulloch, 1983). In penetrating laceration injuries. or injuries of massive

compression, where the glia limitans (pial tissue on the cord surface) is breached as dense

mass of connective scar tissue remains in the daniaged region (Bunge et al., 1997).

A progressive axonal pathology has been carefully examined following SC1 using

electron microscopy (Anthes et al., 1995; Balentine, 1978; Bresnahan, 1978; Griffiths and

McCulloch, 1983). Shortly after SCI, investigators have described penaxonal space formation

berneen the axon and its myelin sheath (GriffÏths and McCulloch, 1983) (Balentine, 1978)

(Anthes et al.. 1995; Bresnahan, 1978). Distal fibres undergo Wallerian degeneration and. at

the proximal stump, there is an accumulation of organelles, presumably due to the interruption

of anterograde transport (Balentine, 1978). Notably, axons from injured tissue have a

characteristic appearance as swollen or giant and have thin or absent myelin sheaths (Anthes

et al., 1995; Balentine, 1978; Bresnahan, 1978; Griffiths and McCulloch, 1983). A time course

study by Bresnahan (Bresnahan, 1978) indicated that small diameter fibers degenerate first,

followed by large diameter fibres, but a chronic study by Fehlings (Fehlings and Tator, 1995)

indicated a preferential loss of large diameter fibres. lnvestigators have described a vesicular

degeneration, rupture and phagocytosis of myelin sheaths (Anthes et al., 1995; Balentine,

1978). Griffiths and McCullouch (Griffiths and McCulloch, 1983) described remyelination of

surviving axons by oligodendrocytes or Schwann cells. Thus. partially myelinated axons in

addition to completely demyelinated axons are found chronically following SCI.

Pathophvsiology of S ~ h a l Cord Injury

Ensuing the initial mechanical insult to the spinal cord, are host of secondary injury

processes which exacerbate the primary injury. This cascade of pathophysiological

mechanisms includes: ischemia, altered ionic homeostasis, free-radical release/lipid

peroxidation. glutamate-mediated excitoxicity, inflammation and apoptosis. It is also important

to note that a cornplex interplay exists between the primary and secondary injury processes

which ultirnately leads to paralysis. Figure 1 (pg.24) summarizes the major processes of

primary and secondary injury implicated in SCI.

lschemia

Ischemia is thought to be a key mechanism of secondary injury propagating further

cellular damage following SC1 by rendering tissue anoxic and thus depleting cellular energy

stores. The reduction of cellular ATP by ischemia results in an imbalance of energy demand

and need. Acidosis ensues as the cell undergoes anaerobic respiration leading to an

accumulation of lactate and hydrogen ions within the cell (Du et al., 1999; Hovda et al., 1992).

Microangiographic and electron microscopic studies have shown a reduction in the

microcirculation and spinal cord blood flow after SC1 (Koyanagi et al., 1993; Koyanagi et al.,

4

1993; Koyanagi et al., 1993; Tator and Fehlings, 1991 ; Tator and Koyanagi, 1997). This

ischemia seems to persist and does not involve the larger blood vessels, but rather entails a

loss of microcirculation involving capillaries, arterioles and venules at, and adjacent to, the

injury site (Koyanagi et al., 1993; Koyanagi et al., 1993; Koyanagi et al., 1993; Tator, 1995;

Tator and Fehlings, 1991; Tator and Koyanagi, 1997). Vasospasm by the release of vasoactive

amines, cord compression and edema are thought to be among the mechanisms that rnediate

ischemia in SC1 (Amar and Levy, 1999; Anthes et al., 1996; Tator and Fehlings, 1991 ). Also

hemorrhage, thrornbosis and platelet aggregation might contribute to ischernia following SCI.

Functionally, post-traumatic ischemia has been linked to axonal dysfunction after spinal cord

trauma and is proportional to the amount of ischemia sustained by the spinal cord (Fehlings et

al., 1989). Systemically, neurogenic shock results; there is a loss of autoregulation, an initial

increase in heart rate followed by a subsequent decline in the arterial pressure and cardiac

output of rats after SC1 (Tator, 1995; Tator and Fehlings, 1991 ). Post-traumatic ischemia has

been of major research interest because it precedes many of the other secondary injury

mechanisms and is postulated to mediate many of them (i.e. excitotoxicity, altered ionic

homeostasis) (Tator and Fehlings, 1 991 ).

AItered lonic Homeostasis

The ischemia resulting from SC1 causes many perturbations in the cell and one of the

eariiest is the depletion of cellular ATP which disrupts energy-rjependent mechanisms, such as

cellular ionic homeostasis. With this loss of control, sodium, potassium, chloride and calcium

pass across the cell membrane according to their respective concentration gradients (Amar

and Levy, 1999). There is a net effiux of potassium and magnesium, an influx of sodium,

calcium and chloride accompanied by a movernent of water into the cell, thus resulting in

cellular edema (Amar and Levy, 1999; Du et al., 1999; LoPachin et al., 1999; Stys and

Lopachin, 1 998).

The altered ionic homeostasis also produces conduction abnormalities in axons, such

as conduction block, rnediated by the altered expression and distribution of potassium channels

(Fehlings and Nashmi, 1996; Hansebout et al., 1993; Nashmi et al., 2000). Studies have shown

that sodium influx from the extracellular space may occur through a vanety of mechanisrns

including: voltage-gated sodium channels, sodium-potassium pump dysfunction, glutamate

receptors and the sodium-hydrogen exchanger (Agrawal and Fehlings, 1997: Agrawal and

Fehlings, 1996; Fehlings and Agrawal, 1995; LoPachin et al.. 1999; Stys and Lopachin, 1998).

lncreased intracellular sodium concentrations may then result in acidosis, membrane

depolarization and the release of excitotoxic arnino acid neurotransmitters, such as glutamate,

thereby precipitating further secondary injury cascades (Amar and Levy, 1999). Also, the rise in

intracellular sodium levels elicits a rise in intracellular calcium by the reverse activation of the

sodium-calcium exchanger (Amar and Levy, 1999; LoPachin and Lehning, 1997; Stys and

Lopachin. 1998).

Reverse action of the sodium-calcium exchanger is not the sole mechanism for an

increased cellular load of calcium following SCI. Voltage-gated calcium channels, dysfunction

of calcium pumps, glutamate receptors, release from intracellular calcium stores and leak

through voltage-sensitive sodium channels al1 mediate rises in intracellular calcium also

(Agrawal et al., 2000; Du et al., 1999; Stys and Lopachin, 1998; Tymianski and Tator, 1996;

Young, 1992). This rise in intracellular calcium concentrations produces devastating effects for

the cell which are stereotypical regardless of mechanism of central nervous system trauma.

Elevated calcium potentiates fuNer increases in calcium, by the release of neurotransmitters,

thus increasing cell excitability (Young, 1992). The electron transport chah is interrupted by

calcium binding to mitochondrial membrane, further depleting cellular energy stores (Young,

1992). Furthemore, calcium activates a host of enzymes such as: calpains, other proteases,

phospholipases, protein kinases, and endonucleases (Agrawal et al., 2000; Du et al., 1999;

LoPachin and Lehning, 1997; Schumacher et al., 1999; Schumacher et ai., 2000; Tymianski

and Tator, 1996; Young. 1992). Activation of these enzymes will resuit in an increase in free

radicals, cytoskeletal breakdown, apoptosis and necrosis (Tymianski and Tator, 1996; Young,

1 992).

Lipid Hydrolysis, Lipid Peroxida tion and Free Ra dical Release

Lipid hydrolysis leads to the release of fatty acids by the activation of membrane

phospholipases and lipases. Early after SCI, there is a marked increase in the release of

arachadonic acid which is a precursor to the cyclooxygenase pathway which produces

eicosanoids (Anderson and Hall, 1993). High levels of eicosanoids (prostaglandins and

thromboxane) and, in sorne species leukotrienes, are detected in spinal cord tissue following

trauma (Anderson and Hall, 1993). Eicosanoids and leukotrienes can then in turn mediate the

inflarnmatory response, exacerbating tissue damage (Hsu et al., 1996).

Following SCI, there is a nse in the oxygen free radical-mediated fatty acid peroxidation

products, such as malonyldialdehyde, as well as a decline in the levels of antioxidants, and an

inhibition of lipid peroxidation-sensitive proteins, such as the Na+/K* ATPase (Hall, 1996).

Methylprednisolone, 21-arninosteroids and vitamin E have been shown to improve neurological

function in spinal cord injured cats through their antioxidant actions (Hall, 1996). In particular,

methylprednisolone and U-74006F (a 21 -aminosteroid) have also decreased tissue damage

after SC1 (Hall. 1996). Furthemore, it has been hypothesized that lipid peroxidation rnight play

a rote in ischemia as studies of methylprenisolone have associated decreased lipid

peroxidation with decreased tissue ischemia following spinal cord trauma (Hall, 1996).

Glufamafe Excifofoxic~ty

Glutamate is a u biquitous neurotransmitter that rnediates its excitotoxic actions through

a variety of ionotropic (NMDA/AMPA/Kainate) and metabotropic (group 1, II and III) receptors.

Both ionotropic and rnetabotropic receptors have been implicated in the secondary injury after

SCI. It has been shown that the levels of glutamate released during SC1 contribute to the

pathophysiology of SC1 (Liu et al., 1999). The excitotoxic effects of glutamate have been

demonstrated by an increase in the influx of ca2' prier to secondary damage (Mills et al., 1995;

Regan and Choi, 1991). Furthenore, glutamate has also been linked in mediating other

secondary injury processes such as inflammation and apoptosis (Matute. 1998; Matute et al.,

1997; Nakai et al., 2000; Sanchez-Gomez and Matute, 1999; Tenneti and Lipton, 2000; Wada

et al., 1999).

Studies have shown that the application of ionotropic glutamate (NMDAIAMPAlkainate)

receptor antagonists. both in vitro and in vivo, have reduced tissue loss and functional deficits

associated with SC1 (Agrawal and Fehlings. 1997; Arnar and Levy, 1999; Faden et al., 1988;

Faden and Simon, 1988; Gaviria et al., 2000; Li and Stys, 2000; Rosenberg et al.. 1999; Tator,

1995; Wrathall et al., 1994; Wrathall et al., 1996; Wrathall et al., 1997). It has also been shown

that ionotropic glutamate receptor-rnediated excitoxicity was ca2' dependent in the white

rnatter of the rat spinal cord (Li and Stys, 2000). Furthermore, the antagonism of group I

mGluR0s and of phospholipase C has been shown to improve the recovery of compound action

potentials. In contrast, the selective activation of group I mGluR receptors has led to

impairments in the compound action potential recovery following a clip compression injury to

the in vitro rat dorsal column (Agrawal et al.. 1998).

lnfïammation

lmmediately after SCI, an inflammatory reaction is elicited with two waves of invading

leukocytes: neutrophils followed by macrophages. Polymorphonuclear neutrophils (PMN's)

infiltrate damaged tissue through the recognition of surface adhesion molecules on ?MN'S and

endothelial cells, thus mediating further damage (Hsu et al., 1996; Taoka et al., 1997). Studies

with double knockout mice for the adhesion molecules, ICAM-1 and P-selectin, showed

improved neurological recovery following spinal cord injury further implicating PMN's in

secondary injury processes (Farooque et al.. 1999). PMN's generate free radicals, release

proteases, cytokines, nitric oxide. eicosanoids and kinins thus potentiating other injury

rnechanisms such as free-radical mediated lipid peroxidation and cellular protein degradation

(Hsu et al., 1996). Platelet deposition has also been observed following SC1 (Hsu et al., 1996).

Platelet degranulation results in the release of inflammatory mediators and proteolytic

enzymes, activating PMN's (Hsu et al., 1996). Blight (Blight, 1985) described a secondary

pathology of invading macrophages which phagocytosed cellular debris at the injury site.

Furtherrnore, the reduction of macrophages has been shown to improve neurological function

and neuronal survival following SC1 (Hsu et al.. 1996). Studies indicate that mice which have a

delayed Wallerian degeneration response, also have a delayed activation of

macrophages/microglia after spinal cord crush injury (Fujiki et al.. 1996).

Apoptosis

Apoptosis is a highly organized f o n of programmed cell death that is characterized by

the activation of specific genes leading to DNA fragmentation. chromatin condensation and

cytosolic condensation. Recent reports have indicated that apoptotic cascades contribute to the

secondary injury rnechanisms following SC1 and occur in macrophageslmicroglia, neurons and

glia (Casha et al., 2000; Crowe et al., 1997: Li et al., 1999; Liu et al., 1997; Shuman et al.,

1997; Springer et al., 1999; Yong et al., 1998). Oligodendroglial apoptosis has been

demonstrated in apposition to degenerating axons (Casha et al., 2000; Emery et al., 1998; Li et

al., 1999; Liu et al., 1997). Furthermore, specific cell death signalling molecules have been

identified are associated with the apoptosis of oligodendrocytes, through the expression of FAS

and p75 on the cell surface (Casha et al.. 2000). In addition, caspases, which are cysteine

proteases that cleave at aspartate residues, are key downstream effectors of apoptosis. These

enzymes have been implicated in apoptosis following SCI, including caspase, 3.8 and 9

(Casha et al., 2000; Emery et al.. 1998; Springer et al., 1999). Another cell death pathway

which has been implicated in apoptosis after spinal cord trauma is the p53 pathway (Saito et

al., 2000).

In Vivo Experirnental Spinal Cord Iniury Models

The goals of rnodeling SC1 in vivo are: to produce transient andior permanent

paraplegia; to screen potential pharmacological therapies for SC1 and; to study the

mechanisms which underlie and augment neurological deficits following SCI. Important criteria

for modeling SC1 are that a potential model must be relevant. reliable and reproducible in order

to enable researchers to draw conclusions from studies using a particular rnodel.

Kinetic Impact Models

All kinetic impact models have a component of contusion where a force is imparted to

the cord rapidly, usually in less than 1 second, producing an impulse (energy) (Fehlings and

Tator, 1988). Models that utilize this strategy include: the weight-drop models, the clip

compression model, the forceps crush model, screw compression model, balloon compression

rnodels and vertebral displacement models. All of these models are applied to study

pathophysiological mechanisms of SC1 as well as to screen potential therapeutic agents

because kinetic models mimic the acute mechanical trauma that is sustained by the spinal

cord.

Weight-Drop Models

In modeling SCI. the trend has been to take the cornplex situation of SC1 and simplify it

by using a quantifiable method. The first modem experimental rnodel of SC1 was the Allen

weight-drop model in dogs where, after a laminectomy. a given weight could be dropped from a

known height through a hollow tube, producing a quantifiable and standardized contusive injury

(Allen. 191 1). Since Allen's original rnodel, many researchers have adapted the original weight-

drop model to many different species, including non-human primates. sheep, pigs, cats, ferrets

and rats, making it the most widely used mode1 of SC1 (Anderson, 1982; Balentine, 1978;

Balentine, 1978; Basso et al., 1996; Beattie, 1992; Behrmann et al.. 1992; Black et al., 1988;

Bresnahan, 1978; Bresnahan et al,, 1976; Dohrrnann et al., 1976; Ford, 1983; Griffiths and

McCulloch. 1983; Gruner, 1992; Kearney et al., 1988; Metz et al., Neurotrauma 2000 Jan;

Noyes, 1987; Southem et al., 1990; Stokes, 1992: Stokes et al.. 1992; Wrathall et al., 1985;

Yeo et al., 1975). A number of groups have modified it further to obtain gradations of injury that

are reliable and reproducible in the rat (Basso et al.. 1996; Behrmann et al., 1992; Gniner,

1992; Stokes, 1992; Stokes et al., 1992; Wrathall et al., 1985).

A common variation of the weight-drop model has been to drop a known weight from a

given height along a metal rod. The weight falls ont0 a gauge which measures the force of

impact which subsequently hits an impounder resting on the dorsal surface of the spinal cord

(Black et al., 1988; Dohrmann et al., 1976: Wrathall et al.. 1985). In order to ensure

reproducibility, this method requires that the animal be placed in a stereotaxic frame and that

the spinal column is stabilized by clamps or forceps (8lack et al., 1988; Dohrmann et al., 1976;

Wrathall et al., 1985). In order to achieve different grades of injury, the height andlor the weight

can be varied. The units of injury quantification were described in gram-centimeters, which was

the product of the weight and height. however this was later shown to be an inappropriate

measure as different weights and heights that produced the same gram-centimeter product did

not inflict the same grade of injury severity (Dohrmann et al., 1976).

In addition to the device described above, other variations of the weight-drop method

have been developed. For example. to ensure that ventral displacement of the spinal column

does not absorb the energy of the impact. a model which places an anvil under the ventral

aspect of the cord has been presented (Ford. 1983). Furthemore, a model less invasive than

the version of the weight-drop model above has been developed. A weight is dropped through

a plexiglass guide tube which impacts an impounder that rests on the dorsal surface of the

vertebrae (no laminectomy) and c m be a placed to produce three different types of trauma:

compression. flexioncompression and extension-compression (Southem et al., 1990).

With the demonstration that inherent sources of variability existed in the biomechanical

parameters of weight-drop techniques. especially in small anirnals (multiple impacts,

differences in force of impact, velocity of impact, rib and spinal column movements,

reproducibility) investigators sought to engineer devices to control for these problems (Das,

1989; Khan and Griebel, 1983; Khan et al., 1985; Koozekanani et al., 1976). Pneumatic

techniques have been applied to weight-drop models of spinal cord injury which attempt to

control certain biomechanical parameters such as the amount and velocity of compression

sustained by the spinal cord during trauma (Anderson, 1982; Kearney et al., 1988). Another

refined variation of the weight-drop technique is the NYU contusion model in rats (Gruner,

1992). This model monitors biomechanical parameters such as rate of tissue compression,

vertebral movernent and impact velocity through digital optical potentiometen sensors when a

l og rod falls along a guiding rod from a set height above the surface of the spinal cord (Basso

et al., 1996; Gmner, 1992). Monitoring of biomechanical parameters allows the experirnenter to

discard an animal whose impact does not fall within a set range of parameters for a particular

grade of injury (Gruner, 1992). This model has produced behaviourally and anatornically

graded, reproducible injuries (Basso et al., 1 996).

Electromechanical Contusion Mode1

Recently, a very sophisticated electromechanical spinal injury device has been created

that attempts to minimite many of the sources of error that exist in the weight-drop technique

(Behnann et al., 1992; Stokes, 1992; Stokes et al.. 1992). In this device (Ohio State University

SC1 device), an electromagnetic driver and impact pattern generator cornpress the spinal cord,

producing trauma in less than 25 msec to the rat spinal cord (Stokes et al., 1992). The key

elements of this model are: 1) the elimination of multiple impacts; 2) the use of sensitive

transducers to measure biomechanical parameters; 3) elirnination of errors in the calculation of

the force due to the mass of the injury probe and 4) the measurement of a starting point by

measuring the touch sensitivity (Stokes, 1992). During injury, either the force of impact or

spinal cord displacernent is the controllad variable and a rat is discarded, should either of the

variables exceed the acceptable range. This injury paradigm has produced a consistent,

graded injury in rats (Behrmann et al., 1992).

Clip Compression Mode/

Other kinetic expenmental SC1 models involve not only contusion, but also a persisting

compression which is similar to what frequently occun in clinical SC1 (Tator, 1995) thus,

modeling both the primary rnechanical trauma and the subsequent ischernia. The Rivlin and

Tator rat clip compression model is an example of such a technique. A rnodified Kerr-Lougheed

aneurysm clip with cuwed blades is used to impact and cornpress the spinal cord extradurally.

After laminectorny, the lower blade is passed anteriorly between the spinal cord and the

vertebral bone and the upper blade is allowed to close rapidly on the spinal cord, producing

both dorsal and ventral compression (Dolan and Tator, 1979; Dolan et al., 1980; Rivlin and

Tator, 1978). The grade of injury in this method can also be varied when the spring of the clip is

changed, resulting in different forces of blade closure. Regular calibration of clip closing forces

must be performed to ensure that a reproducible injury can be obtained with this method (Dolan

and Tator, 1979). This technique has been shown to produce a graded severities of SC1 based

on anatomical, behavioural and electrophysiological outcomes (Fehlings and Tator. 1995;

Fehlings et al., 1989; Fehlings et al.. 1987; Khan and Griebel, 1983; Khan et al., 1985;

Krassioukov and Fehlings, 1999; Maiorov et al., 1998; Midha et al., 1987; Nashmi et al., 1997).

A variation of the Rivlin and Tator rat clip compression injury model has been recently reported

(von Euler et al., 1997).

Crush Models

Another SC1 model that produces kinetic impact injury is the modified forceps crush

model developed by Blight in guinea pigs (Blight, 1991). In this technique, the degree of injury

is controlled by compressing the spinal cord to a set thickness with a pair of modified forceps.

After laminectomy, the fiat tips of the forceps are inserted so that they are on either side of the

spinal cord and compressed the spinal cord laterally towards the midline. Compression occurs

over 1 second and persists for 15 seconds (Blight, 1991; Gniner et al., 1996). This model has

been applied to the rat spinal cord producing a graded injury depending on the thickness of the

compressed spinal cord when behavioural and histological outcome measures were examined

(Gruner et al., 1996). Hashimoto and Fukuda (Hashimoto and Fukuda, 1990) have developed a

model whereby screws are inserted into a small hole made in the vertebrae of rats. The

principle of injury is similar to the modified forceps crush model in that the controlled variable is

the extent of spinal cord compression. Screws of different lengths are used to impart

compression from 5 min to 1 hour thus producing a graded injury. The model takes advantage

of an intact vertebrae exerting pressure onto the spinal cord during injury.

ln fla table Balloon/Cuff

Rapid compression by an inflatable balloonlcuff has also been described (Khan and

Griebel, 1983; Khan et al., 1985; Martin et al., 1992; Martin and Bloedel, 1973; Tarlov, 1957;

Tator, 1973). Circumferential compression by an infiatable cuff was produced in monkeys by

passing a Silastic cuff around the exposed spinal cord in the extradural space and rapidly

infiating the cuff to a set pressure by a sphygmomanometer. The duration of compression could

also be varied from 2 to 5 minutes, resulting in an injury that resembles the fracture-dislocation

injury observed in man (Tator, 1973). However, application of this model to smaller animals,

such as the rat, using a Fogarty balloon catheter resulted in a steep, dose response curve,

making graded injuries difficult (Khan and Griebel, 1983). This is compounded by the fact that

the pressure exerted on the spinal cord is extremely difficult to measure with accuracy

(Fehlings and Tator. 1988). Recently, it has been shown that inflatable balloons inserted into

the subarachnoid space can produce graded injuries in rats however a disadvantage of this

technique is the uncertainty of the spinal injury level when the balloon is inserted (Martin et al.,

1992).

Ofher Kinetic Models

A vertebral displacement model has been described which offers the advantage of not

requin'ng a laminectorny to produce injury (Fialho et al., 1982). This technique produces spinal

cord trauma by luxation between the L1 and 12 vertebrae or fractures of L I and L2 in dogs.

14

Unfortunately, this method varies in the force that is applied to the spinal column (Le. 1848N to

2666N), thus adding additional variability to the system.

Static Load Models

Static load models apply compression slowly to the spinal cord (Wada et al., 1999). In

contrast to the kinetic impact models, static load models do not replicate the primary

mechanical insult to the spinal cord. However, the static load rnodels do contribute information

about the secondary injury processes which follow prirnary mechanical insults to the spinal cord

(Wada et al., 1999: Yu et al., 1999). Eidelberg (Eidelberg et al., 1976) (Eidelberg et al.. 1977)

produced static load SC1 in ferrets by adding weights to a pressor device which resulted in

submaximal SCI. From static balloon inflation techniques in monkeys, it was concluded that

blood fiow changes do not appear to be significant pathological mechanisms in this type of

injury (Kobrine et al., 1978; Kobrine et al., 1979). Furthermore, other investigators concluded

that duration of load compression did not affect outcome and that load force did not correlate

with the inclined plane test for neurological outcome (Black et al., 1986; Kushner et al., 1987).

Unfortunately, the static load models have the limitation of not mimicking the pathology of

human spinal cord injury (Black et al.. 1986; Fehlings and Tator. 1988). Specifically, static load

injury models do not produce the same vascular damage as is obsenred in kinetic models

(Fehlings and Tator, 1988); moreover Black and colleagues (Black et al., 1986) concluded that

the damage is primarily located in the posterior half of the spinal cord.

Transection Models

Transection and hemisection models al1 involve cutting the spinal cord with a knife or

blade. Complete transection models are useful in studies of regeneration, provided that the

experimenter is certain that they have achieved a complete transection. lncomplete lesions

can lead to incorrect conclusions of regenerating axons (Das, 1989). Complete transection

studies can also permit studies of segmental refiexes, serve as controls for studies of

retrograde tracing and locomotor recovery (Basso et al., 1996; Edgerton et al., 1992; Ritz et al.,

1992). For studieç of descending pathways, hemisections are also important, when done with

precision, because they may provide information about the role of specific pathways in

locomotion (Fernandes et al., 1999; Jenkins et al., 1993; Little et al., 1988; Ritz et al.. 1992;

Windle et al., 1951).

Other SC/ Models

A nurnber of other mechanical and non-mechanical approaches have been used to

create experimental SCI. All of the mechanical approaches outlined above utilize posterior

approaches to producing injury, however. one study has detailed an anterior approach (Benzel

et al., 1990). In this rnethod, an aneurysm clamp is inserted into the upper lumbar,

retroperitoneal space, ensuring that the clamp was immediately anterior to the vertebral column

using an operating microscrope. The flat end of a knife handle is placed between the clamp

and the dorsal part of the spinal colurnn before the clamp is closed. This model minirnizes focal

injuries to the dorsal spinal cord so that a focal injury is from the ventral side (Benzel et al.,

1990). This type of injury, although clinically appealing, is not ideal because the ventrally

applied force is not graded. Another closed model of acceleration-deceleration producing a C1-

2 subluxation has been reviewed, although the variability in injury severity and location

preclude its use in controlled experimental studies (Fehlings and Tator, 1988).

Among the non-mechanical, non-graded approaches. freezing injury has been

described as a method for studying spinal cord tissue injury and repair in the rat (Collins et al..

1986). An interesting finding was that the experirnental conditions demonstrated axonal

outgrowths and may provide an altemate model for studying regeneration (Collins et al., 1986).

Photochemical induced vascular stasis is another non-mechanical, non-graded injury

procedure that has been developed (Gaviria et al., 2000; Watson et al., 1986). A

photosensitizing dye (rose bengal) is injected into a rat before the expoçed spinal cord is

subjected to irradiation at a wavelength of 560nm for 40 min. The dye generates an oxygen

16

free-radical thus peroxidizing endothalial cells, stimulating platelet adherence and producing

vascular occlusion. The pathology described by the investigators included hernorrhagic

necrosis of the gray matter, edema and vascular congestion (Watson et al., 1986). The element

of induced vascular stasis may make this mode1 useful in studying vascular dysfunction

following SCI. Microwave hyperthemia has also been applied to the rabbit spinal cord to

produce non-invasive graded injuries (Sutton, 1988). Somatosensory evoked potentials were

measured following microwave irradiation at 91 5MHz which raised intraspinal temperatures

from 40 to 43°C for time periods ranging from 15 to 60 minutes and complete paraplegia was

obtained after irradiation for 30 minutes at 42% (Sutton, 1988). Finally, lesioning by

radiofrequency in the cat spinal cord produces a non-graded, consistent injury that does not

spread like the mechanical approaches discussed previously (Haghighi et al., 1996).

utcome Measures to Assess Se veritv of Chronic SQ inal Cord Iniury ln Vivo

Neurological

Openfield locomotor function has been widely used to assess recovery from in vivo SCI.

Modifications to the Tarlov locomotor scale were the most commonly used indicators of

hindlimb rnotor recovery. The modified Tarlov scale ranges from no spontanous movement to

normal walking (Blight, 1996; Fehlings et al., 1989; Kerasidis et al., 1987; Stokes and Homer,

1996). Recently, a more sensitive and reliable method for assessing openfield walking in rats

has been reported, termed the BBB (Basso. Beattie and Bresnahan) Locomotor Rating Scale

(Basso et al., 1995). This scale discrirninates between slight and extensive joint movements

and; between occasional, frequent and consistent coordination and stepping in rats and has

been applied to rnice (Basso et al., 1995; Jakeman et al., 2000; Ma et al., 1999).

In addition to testing openfield walking, postural function can also be reliably tested

using the lnclined Plane Test developed by Rivlin and Tator (Rivlin and Tator, 1977) which has

been shown to correlate with anatornical outcome measures in rats (Fehlings and Tator, 1995).

17

The lnclined Plane Test requires that the rat is placed horizontally on a rubber mat and the

maximal angle which a rat could maintain its balance for 5 seconds is recorded (Rivlin and

Tator, 1977).

lt has been proposed that a combination of locomotor and reflex tests can be applied to

assess behavioural function in rats following SC1 (Kunkel-Bagden et al., 1993). The Combined

Behavioural Score (CBS), introduced by Wrathall, has attempted to use this principle to assess

functional deficits in both rats and mice (Kerasidis et al., 1987; Kuhn and Wrathall, i998;

Wrathall et al., 1985). This testing paradigm consists of a battery of behavioural tests including:

motor score, toe spread reflex, righting reflex, withdrawal reflexes. placing reflex, swim test,

lnclined Plane test, platform hang, wire mesh descent, rope walk and tail flick response. The

scores from each test are combined, resulting in the combined behavioural score for an animal.

Footprint analyses, rotarod tests and grid walking tests are also often used by investigators to

assess motor function in rats and mice (Behrmann et al., 1992; Crawley and Paylor, 1997).

However, these tend to detect deficits in coordination among rodents with slight to moderate

neurological deficits and often do not detect differences between rodents with cornplete

injuries. In general. when behavioural tests are applied, two independent, blinded observers

are required in order to ensure that objectivity is maintained and that bias is eliminated.

Neuroph ysiological

The measurernent of evoked potentials is also useful to study functional tissue

responses following SCI. Specifically, somatosensory and motor evoked potentials are the

most predorninantly used tests (Fehlings et al., 1989; Fehlings et al.. 1987; Nashmi et al.,

1997). The attraction of using electrophysiological recordings measures is their objectivity in

quantifying the integrity of sensory and motor tracts in the spinal cord. Despite these attributes,

SSEP's and MEP's have important limitations. For example, in a recent report myoelectric

motor evoked potentials (MEP's) were concluded to not differentiate well between different

severities of SC1 (Nashmi et al.. 1997). Also, evoked potentials are technically challenging,

selective for specific tracts and the responses are non-linearly related to the tissue damage in

those tracts (Blight, 1992).

Histopa thological/A na tomical

There are many histopathological outcome measures that are suitable for quantifying

the degree of injury severity and present the opportunity to infer structure/function relationships

following SC1 when combined with neurological andlor neurophysiological techniques. Digital

morphometry provides an accurate method of measuring residual tissue at and adjacent to the

injury site (Agrawal and Fehlings, 1997; Basso et al., 1996; Beattie, 1992; Behnann et al.,

1992; Noble and Wrathall, 1985; Schumacher et al., 2000). This requires staining the fixed

spinal cord tissue with Luxol Fast Blue for myelin and counterstaining with Nissl stain, Cresyl

Violet or Hematoxylin and Eosin. Usually every fifth or tenth transverse section is analyzed

under brightfield microscopy for the cross-sectional area of cavity, neuron sparing, white and

gray matter and lesion length. Furthermore, three-dimensional reconstructions can be done of

the injury site to describe the shape of the lesion.

Light rnicroscopic studies can also be performed to measure the surviving axons and

the degree of myelination through the injury site (Blight, 1991; Blight and Decrescito, 1986;

Eidelberg et al., 1977; Fehlings and Tator, 1995; Gruner et al., 1996). One method involves

staining transverse sections with toluidine blue and observing the sections under oil immersion

at 1000x magnification. Visualized axons can then be counted by a line-sampling technique as

well, rnyelination indices can be calculated (Blight, 1991 ; Blight and Decrescito, 1986; Fehlings

and Tator, 1995).

Another histopathological outcome measure which assesses the integrity of axons

through the injury site is retrograde tracing by molecules such as horse-radish peroxidase

(HRP) or Fluoro-Gold (Eideiberg et al., 1981 ; Fehlings and Tator, 1995; Hogan, 1992; Midha et

al., 1987; Naso et al., 1993). This can be accomplished by irnplanting the retrograde tracer

several segments caudal to the injury site and allowing uptake and transport of the tracer to the

19

bnin. Following fixation, the brain can be extracted, sectioned, processed and examined for

labeled neurons in regions of the brain stem that are involved in motor function (i.e. raphe

nuclei, reticular nuclei, vestibular nuclei, red nuclei. rostro-ventrolateral medulla and motor

cortex). ldeally histopathological analyses should be performed in a blinded fashion in order to

minimize observer bias in experirnental studies, especially in those of a pre-clinical nature.

In Vivo Murine Models of Spinal Cord Injury

In recent years, rnice have been used to study spinal cord injury in vivo. Table 1 (pg.25)

outlines the different models and outcome measures used to produce experimental SC1 in

mice. lnherent physiological differences of mice compared to other animals used to study SCI,

along with their small size, present technical challenges as it may not be feasible to apply

corn plex or large apparatuses reliably to produce graded injuries in mice. Thus, transection

studies have been cornmonplace to study spinal cord pathophysiology and repair in mice

(Bartholdi and Schwab, 1997; Bjugn et al.. 1997; Huang et al., 1999; Pekny et al., 1999; Wang

et al., 1997; Zhang et al., 1998). Wang et al. (Wang et al., 1997) studied GFAP knockout mice

following dorsal hemisections at T8 using immunofluorescence for extracellular matrix

molecules and found that the absence of GFAP in reactive astrocytes does not have an effect

on axonal sprouting or regeneration. Pekny et ai. (Pekny et al.. 1999) studied GFAP and

vimentin double knockout mice following an upper thoracic hemisection and a 5mm longitudinal

section of the spinal cord through the donal funiculus for nestin immunofluorescence and in

situ hybridization. as well as, morphological assessment of blood vesse1 diameter and glial scar

formation. This group found that both GFAP and vimentin are necessary for glial scar formation

following CNS trauma (Pekny et al., 1999). Lateral transections were performed at T8 in

C57B116 mice and mutants for a delayed onset of Wallerian degeneration (Zhang et al., 1998).

It was found that mice with a delayed Walleflan degeneration response had a deficit in tissue

repair responses as rneasured by the quantitative morphometry of cavity and motor recovery

(Zhang et al., 1 998). Bartholdi and Schwab (Bartholdi and Schwab. 1997) hemisectioned the

spinal cord of C57Bll6 mice at T9 in order to study the expression of pro-indammatory

cytokines and chernokines following SC1 using immunofluorescence and in situ hybridization.

This study indicated that TNFa, II-1, MIP-la and MIP-1 P are upregulated within 1 hour

following SC1 and that resident microglia are the likely source of the mRNA for these molecules

(Bartholdi and Schwab, 1997). In another expenment, the number of distal ventral hom

neurons were not found to change after spinal cord transection at Tg-Tl 2 (Bjugn et al., 1997).

Most recently, extensive spinal cord regeneration was reported by investigators, using the

techniques of retrograde tracing, anterograde tracing and motor function, following complete

spinal cord transection in BALBlc mice immunized against myelin and inhibitory proteoglycans

(Huang et al., 1999).

An ischemia model of SC1 was reported in C57BlI6NCr18R mice following aortic arch,

left subclavian artery and intemal mammary artery cross-clamping for 9 or 11 minutes (Lang-

Lazdunski et al.. 2000). In this model, motor deficit and spinal cord tissue damage were the

outcornes chosen to evaluate the injury severity. Radiation has also been used to generate SC1

in C3HfJSedlKam mice (Lavey et al., 1994; Lo et al., 1993; Lo et al., 1992). In these studies it

was determined that onset and recovery from paralysis depended on the extent, timing and

fraction size of radiation delivered to the lower thoracichpper lumbar spinal cord.

Non-graded, crush injuries have also been applied to the mouse spinal cord using

forceps. The degree of cavitation, GFAP immunohistochemistry and the activation of

rnicroglia/macrophages was compared in mice with a mutation for delayed Wallerian

degeneration were against C57BV6 mice following crush injury with forceps at TB in a time

course study (Fujiki et al., 1996; Zhang et al., 1996). The mice with delayed Wallerian

degeneration exhibited a delayed activation of microglia/macrophages and astrocytes, and

deficient tissue repair (Fujiki et al., 1996; Zhang et al.. 1996). The transport system for TNFa

across the blood-brain barrier was upregulated and functional deficits were observed after

forceps cmsh injury at LI-L2 in ICR mice (Pan et al., 1997; Pan et al., 1999). lmproved open-

field walking ability was obsewed after CD1 mice were treated with an antiangiogenic

polysaccharide (CM1 01) following spinal cord crush injury with forceps at T l 0-TI 1 (Wamil et

al., 1998).

Graded, weight-drop models of SC1 have also been recently applied to murine systems

(Jakeman et al., 2000; Kuhn and Wrathall, 1998; Ma et al., 1999; Ma et al., 1998). The classic

weight drop model has been developed and characterized using neurological tests (the

combined behavioural score) and anatomical assessments of the injury site following two

graded injuries (and sham injury) at T8 in C5761i6 mice (Kuhn and Wraüiall, 1998). in addition,

correlations between behavioural indices (inclined plane, mesh descent, rope walk, bar grab

and motor score with anatomical indices (residual white matter and lesion length) were done to

further quantify this model (Kuhn and Wrathall, 1998). The Ohio State University SC1 device

was also adapted in order to produce a controlled contusion injury in C57B116, BalbIC and

BIOIPL mice (Jakernan et al., 2000; Ma et al., 1999; Ma et al., 1998). Data from these studies

indicate that graded injury can be produced using this device and that injury severity and BBB

scores are correlated with the percentage of spared white rnatter and biornechanical

parameters of injury (Le. impulse) (Jakernan et al., 2000; Ma et al., 1999).

A graded, static load model has also been adapted to examine double knockout mice

for P-selectin and CAM-1 (Farooque et al., 1999). Following two severities of injury at T8,

these mice were cornpared to wild-type mice for behavioural differences in response to static

load SC1 (Farooque et al., 1999). This group concluded that knockout of these genes resulted

in improved neurological recovery following SCI, however uninjured shams were not examined

in this study, thus evidence was not provided to ensure that the operative procedures did not

induce injury. A follow-up study with ICAM-1 knockout rnice indicated that, this molecule does

not mediate secondary tissue damage alone following traumatic SC1 (Isaksson et al., 2000).

There are important limitations of the present techniques used in murine systems.

Firstly, the transection, ischemia, irradiation and crush mouse models of SC1 are not graded.

Transection models have the additional drawback of not mimicking the elernents of clinical SCI,

typically involving an injury mediated by bunt fracture of the vertebral body followed by a

persisting circumferential compression (Fehlings and Tator, 1988; Tator, 1973; Tator, 1983).

Secondly, the graded contusive models lack a rigourous quantitation of anatomical and

neurological outcornes and the relationships between them in the mouse. Moreover, animal

stability, low variability and ease of injury production are still not fully assessed in the mouse

contusion rnodels,

The selection of strains is an important issue in the development of a murine model of

SCI. There are two general categories of mice: inbred and outbred strains. An inbred strain is

one where brother-sister matings have been performed for at least twenty generations, and the

line can be traced back to a single breeding pair (Picciotto and Wickman, 1998). As a result of

this breeding strategy, inbred mice are homozygous at every gene allele (Picciotto and

Wickman, 1998). In contrast, outbred mice have not undergone such mating schemes and thus

have more genetic variability and resilience among its memben. Selection of a strain is

important factor because different inbred strains have different physiological characteristics

and, in some cases, direct comparisons cannot be made between different inbred strains

(Cutler. 1993; Davis et al., 1995; Gao and Cutler, 1992; Gerlai, 1998; Gerlai, 1998; Homanics

et al., 1999; Picciotto and Wickrnan, 1998; Tsao et al., 1999).

In the present study, the adaptation of an existing, well-developed rat model of SC1 to a

resilient, outbred strain (CDI) has been used to develop and characterize the general response

of the murine spinal cord to injury at T3-4.

FIGURE 1: Mechanisms of lnjury lnvolved in Spinal Cord Trauma

PRIMARY INSULTS

1 Contusion I

1 *Persisting Compression 1 A

SECONDARY INJURY PROCESSES

Glutamate Excitotoxicity *? cellular sodium, calcium

4 energy supply

*acidosis

bfree radical release lonic Disruptions *T cellular sodium, calcium and chloride

Lipid Hydrolysis

*? arachadonic acid, eicosanoids, leukotrienes

*& cellular potassium and magnesium

medema, membrane depolarization, release of EAA's

*activation of proteases, phospholipases and endonucleases

Inflammation *activation of macrophages and

velease of proteases, free radical, Apoptosis

cytokines, NO, eicosanoids *oligodendroglial, neuronal death

TABLE 1: Murine Spinal Cord lnjury Models

lschemia

Radiation

Forceps Crush

Weight Drop Contusion

Static Load Compression

References

Bartholdi and Schwab, 1997

Bjugn et al., 1997

Wang et al., 1997

Zhang et al., 9998

Huang et al., 1999

Peknyetal., 1999

Lang-Lazdunski et al., 2000

Loetal., 1992

Lo et al., 1993

Lavey et al., 1 994

Fujiki et al., 1996

Zhang et al., 1996

Pan et al., 1997

Pan et al., 1999

Warnil et al., 1998 - -. . -- - - - - - --

O Kuhn and Wrathall, 1998

Ma et al., 1998

Ma et al., 1999

Jakeman et al., 2000

Farooque et al., 1999

Farooque. 2000

Isaksson, 2000

Level of lnjury

T9

Tg-Tl 2

T8

T8

Lower thoracic

Upper-mid thoracic

Lower thoracid

Upper lurnbar

Lower thoracicl

Upper lumbar

--

Outcome Measures -. .

lmmunohistochemistry, Ni situ hybridization

Cell counts, cell morphometry

immunohistochemistry

Motor function, lesion morphometry

Motor function, retrograde & anterograde tracing

Immunohistochemistry, in situ hybridization - Motor function, lesion size

Motor function, lesion size

Lesion morphornetry, immunohistochernistry

Motor function, serunl protein radiolabeling

Motor function, MRI scan

Motor function, lesiori size

Motor function, immunohistochemistry, lesion

size, lesion site recoristruction

Motor function, lesior1 histopathology

METHODS

Clip Desian and Calibratioii

The clip used to produce spinal cord injury was a modified Kerr-Lougheed aneurysm

clip, which has already been applied to produce SC1 in rats (Dolan et al.. 1980). However. in

order to adapt the clip to the mouse spinal cord, several design modifications were made which

are illustrated in Figures 2A and 28 (pg.33). The clip consists of two curved, flattened blades

that are 1 mm in width. The curved ends of the blades are tapered so that thickness at the tips

is 0.3 mm to a maximal thickness of 0.6 mm. The length of the blades was also shortened to 13

mm in order to allow application to the murine spinal cord. A C-shaped spring closes the blades

of the clip, which rotate about a ball. serving as a fulcnim. The closing force of the blades can

be calculated by the following relationship of Hooke's Law: F = kx, where F is the closing force.

k is the spring constant and x is the displacement of the blades. Thus, at maximal blade

displacement, the closing force is maximal. A modified pair of forceps, with tips to fit the flanges

of the blades. was used as the applicator. Clips of three magnitudes of closing force were used

to produce traumatic SC1 in mice: 39, 89 and 249. These closing forces were selected by

testing intermediate forces in pilot experiments on 24 mice until three distinct severities of injury

were achieved based on neurological assessments (BBB and lnclined Plane tests) .

The mouse clips were calibrated regularly dunng the course of the experiments in order

to ensure that the closing force of the clip remained constant. Clip calibration was performed by

hooking wires onto the grooves of the blades. and one of the wires was attached to a double

a m balance. lncreasing weights were applied to the balance and resulted in increasing blade

displacements, thereby measuring the opening forces of the clip. Applying linear regression to

the measured blade displacements at increasing weights the relationship of Hooke's Law was

derived for the clip and the force applied perpendicular to the cord was deterrnined. See Dolan

and Tator (Dolan and Tator. 1979) and Fehlings and Nashmi (Fehlings and Nashrni. 1997) for

further details.

ouse Soinal Cord lniuries

All experirnental protocols of this study were approved by the animal care committee of

the Toronto Western Hospital Research lnstitute in accordance with the policies established in

the Guide to the Care and Use of Experimental Animals prepared by the Canadian Council of

Animal Care.

Fernale, adult CD1 mice (20-249) (Charles River, Canada) were deeply anesthetized

with halothane (2%), nitrous oxide and oxygen (1 :l lmin), as evidenced by lack of response to a

nociceptive stimulus. A 0.5cc injection of saline was given subcutaneously and during surgery,

the mice were placed on a heating pad at 37°C. Under aseptic conditions, a longitudinal

incision was made on the midline of the back exposing the superficial muscle layers. These

muscle layers were biuntly dissected away and the muscle attached to the vertebrae was

removed using a No. 15 scalpel blade, exposing the T2-T5 vertebrae. Using the large spinous

process of the T2 vertebrae as a landrnark, a laminectomy was perforrned on the T3 and T4

vertebrae with a pair of microscissors. In addition, part of the pedicles of the T3 and T4

vertebrae were also removed. A dissecting hook was then applied to clear an extradural path

between the spinal cord and the vertebral body that remained on the ventral side. Gelfoam,

saturated with saline, was used to control excessive bleeding during the laminectomy

procedure.

The spinal cord was compressed anteriorly and posteriorly at T3-T4 by extradurally

applying a modified Kerr-Lougheed aneurysm clip for 1 minute (Walsh Manufacturing Ltd.,

Mississauga, Canada). One minute was chosen as the length of clip compressive injury based

on pilot studies with differing times of clip compression. The one minute duration of clip

compression produced the most reliable neurological injurylrecovery profile. The lower blade

was passed undemeath the cord, in the path cleared by the dissecting hook, and the upper

blade was released and allowed to close rapidly by release of the applicator designed for the

mouse clips (Walsh Manufacturing Ltd., Mississauga, Canada), thus producing the

compressive injury (Figure 3, pg.35). The clip was allowed to compress the spinal cord for 1

27

minute before rernoval. Five groups of mice (n=12 per group) were injured for the chronic

behavioural and anatomical data in this study. Three groups were subjected to varying degrees

of clip compression: 39, 89 and 249 closing forces. In order to produce three levels of injury

severities, three identical clips were manufactured with different springs. thus producing three

levels of closing force. The remaining two groups studied were subjected to sham and

transection control injuries. The sham control mice were subjected to a laminectomy and hook

dissection, but the clip was not applied to the spinal cord. The transection control animals were

also subjected to a laminectomy, the dura mater was slit at T3 and T4 and a No. 11 scalpel

blade was used to completely transect the spinal cord. Complete transection was confirmed by

direct visual inspection. When the injury procedure was completed, the superficial muscle

layers were sutured using a 6.0 continuous suture and the skin incisions were closed using

small Michel clips. An additional eight animals were injured (four at 89 clip compression and

four at 249 clip compression) for 1 day and 7 day lesion site analysis after SCI.

Following surgery, 1 .Occ of saline was administered subcutaneously in order to replace

the blood volume lost during the surgery. During recovery from anesthesia, the mice placed on

a wam heating pad and covered with a warrn towel. The mice were singly housed in a

temperature-controlled roorn at 27OC for a survival period of 28 days. Food and water were

provided to the rnice ad libitum. During this time period, the animals' bladders were manually

voided twice a day until the mice were able to regain normal bladder function. The incidence of

hematuria was low (6 cases) and gentamycin was administered once daiiy in the event of

hematuria (20mglkg) subcutaneously once a day for 5 days. Concornitantly 0.5cc of saline was

administered everyday that gentamycin was given.

ssessments of Motor Functioq

Prior to injury, al1 of the mice were preconditioned to the openfield envimnment used to

test locomotor function. All neurological assessments were performed on Days 1,4,7, 10, 14,

21 and 28 post injury during the active phase of the rodent sleeplwake cycle. The five groups of

28

mice were blindly assessed for functional recovery by adapting the Basso, Beattie and

Bresnahan (BBB) hindlimb locomotor test using two independent observers (Basso et al.,

1995). The BBB scale used is shown in Table 2 (pg.37). Neurological function was also

assessed using the lnclined Plane (IP) test developed by Rivlin and Tator (Rivlin and Tator,

1977). This test involved placing the mice horizontally on a rubber mat and recording the

maximal angle of the board at which the animal could maintain balance for 5 seconds. The

rubber mat used for the IP test was modified for the mice such that the grooves in the mat were

thinner, shallower and closer together.

etroarade Labelina of Neurons

On day 28, the mice were deeply anesthesized again with halothane, nitrous oxide and

oxygen. Another incision was made on the back exposing the superficial muscle layers. These

muscle layers were bluntly dissected away and the muscle attached to the vertebrae was

removed using a No. 15 scalpel blade. exposing the dorsal vertebral surface at T9

(approximately 5-6 mm caudal to the clip compression injury site). The mice undenvent a T9

laminectomy and subsequently the dura was cut open using a No. 11 scalpel blade and the

spinal cord was transected with a single slice using the No. 11 scalpel blade. Once the

bleeding had ceased, a Gelfoam pledget (3mm x Imm) saturated with 4% Fluoro-Gold (FG,

Fluorochormes Inc., Colorado, USA) in sterile saline was introduced at the transection site and

placed at the rostral cut end of the spinal cord. The transection site was covered with petroleum

jelly (Vaseline) in order to prevent diffusion of Fluoro-Gold. The superficial muscle layers were

sutured using a 6.0 continuous suture and the skin incisions were closed using small Michel

clips.

Tissue Extraction and Frocessing

Five days following Fluoro-Gold introduction, ail of the mice from the chronic subset

were sacrificed. From the eight acutely injured rnice, two 89 and two 249 mice were sacrificed

29

at 1 day post injury. and the other two 89 and two 249 mice were sacrificed at 7 days post

injury. All of the mice were perfused after deep anesthesia with 0.2cc of sodium pentobarbital

(Somnotol). as evidenced by lack of response to a nociceptive stimulus. After ventricular

injection of heparin, the mice were perfused transcardially with I Z m l of 4% paraformaldehyde.

The injury site and adjacent spinal cord tissue was extracted and post-fixed in 4%

paraformaldehyde. The spinal cords underwent manual processing through increasing

alcohols, chlorofom and paraffin wax for paraffin embedding of spinal cord blocks. The brains

were also removed and post-fixed in 4% paraforrnaldehyde and then placed in 30% sucrose for

cryopreservation.

c Assessments of S ~ i n a l Cord Tissue

Paraffin embedded blocks of spinal cord tissue from twenty four randomly selected

anirnals (n=6lgroup) 5 weeks after injury and the eight spinal cords at 1 day and 7 days post

injury were cut into 10pm transverse sections and stained with Luxol Fast Blue (LFB), to

identify residual white matter, and counter-stained with Hematoxylin and Eosin (H&E), to

identify non white matter. Spinal cords taken from the sham, 39, 89 and 249 injury seventy

groups 5 weeks after injury underwent quantitative morphometric analysis in a blinded fashion.

The injury epicenter and sections at every 50pm until 1500pm rostral and caudal to the injury

epicenter were examined under 40x (4x objective, 10x secondary) magnification using

brightfïeld microscopy (Nikon EBOO Microscope) with an Imagepro Plus image analysis system

for quantification of spared white matter. spared gray matter and cavity size after clip

compressive SCI. A low power microscopic image of each transverse section to be analyzed

was digitalized using the image analysis software and calibrated in mm2. The image of the

section was outlined and the total cross-sectional area was measured with the image analysis

software. Areas of cavity within the section were next identified by the technique of colour

selection. using the background colour of the slide as the criterion for areas of cavity formation.

The total cavity area was expressed and also recorded. The areas of spared gray and white

30

matter were identified by their cytoarchitectural appearance, as well as, by their histochernical

staining properties. The regions of residual gray and white rnatter were then outlined, and the

enclosed area was also rneasured.

Counts of Retroaradelv Labeled Neurons in Kev Motor Areas of the Cerebrum and

rainstem

Serial 35pm frozen coronal sections were cut from the brains of thirty randomly selected

animals (n=6lgroup) in a cryostat. All five experirnental groups (sham, 39, 89, 249 and

transection control) were analysed for retrograde labeling of descending spinal rnotor tracts.

Altemate sections were mounted and stained with cresyl violet for the identification of the nuclei

of interest (raphe nuclei. reticular nuclei. vestibular nuclei, red nuclei and rnotor cortex). The

remaining unstained sections were mounted and covenlipped with Mowiol for the detection of

retrogradely labeled neurons with Fluoro-Gold after graded. compressive SCI. Counts were

taken, by a blinded observer, of Fluoro-Gold labeled neurons under 100x (1 0% objective, 10x

secondary) magnification fluorescence microscopy (Nikon E800 Microcscope) with a UV 2A

filter (330-380 nm excitation wavelength).

Anaiysis

Differences in motor performance following SC1 were deterrnined by ho-way repeated

measures analysis of variance (ANOVA). The subsamples chosen for histopathological

analysis from the experimental groups were tested to ensure that they were representative of

the anirnals tested for behavioural function. Differences in the rnorphometric assessrnents

between groups were detenined by two-way analysis of variance (ANOVA). Differences in cell

counts in brain nuclei between groups were detemined by one-way analysis of variance

(ANOVA). For behavioural and histopathological data, al1 pairwise cornparisons were made

between groups using a Student Newman-Keuls multiple range post-hoc test. Correlation

coefficients were calculated as Pearson Correlation Coefficients (R) and as multiple linear

regression correlation coefficients (R~). A variance stabiliu'ng square root transformation was

performed on the counts of neurons retrogradely labeled with Fluoro-Gold before analysis of

variance was applied. Data are presented in al1 graphs as mean _+ S E M . and differences were

reported as statistically significant at pc0.05.

FIGURE 2: (A) Side view of modified mouse aneurysm clip depicting the shortening of blade

length to 13mm and tapering of blade thickness from 0.6rnm to 0.3mm at the tips. (8) Front

view of modified aneurysm clip depicting a blade width of 1 .Omm.

FIGURE 3: A view of the mouse spinal cord (A) before and (B) during 89 in vivo compressive

injury following a laminectomy at T3 and T4.

TABLE 2: BBB Hind Limb Locomotor Rating Scale

Score

No hind limb rnovement

~ l i g h f movement 1-2 joints

Function

~xtensive'~ rnovement 1 joint

Score

Extensive rnovement 2 joints

Slight movement 3 joints

Slight movement 2 joints, extensive movement 1 joint

Extensive movement 2 joints, slight movernent 1 joint

Extensive movement 3 joints

Sweeping or plantar placement, no weight support

Plantar placement or dorsal placement, weight support

Occasional' weig ht-supported plantar steps, no coordinations

Function

Weig ht-supported plantar steps, no coordination

Weig ht-supported plantar steps, occasional coordination

Weig ht-supported plantar steps, frequent" coordination

Weig ht-supported plantar steps, consistenp coordination

Consistent plantar stepping and coordination, no/occasional toe clearance

Consistent plantar stepping and coordination, frequent toe clearance

Consistent plantar stepping and coordination, frequent toe clearance, paw parallel throughout gait

Consistent plantar stepping, coordination and toe clearance

Consistent plantar stepping, coordination and toe clearance, tail down

Consistent plantar stepping , coordination and toe clearance, tail up, trunk unstable

Consistent plantar stepping. coordination and toe clearance. tail up, tmnk stable

.. 'c 50% range of motion > 50% range of

motion tt 50-94% of the time 95-1 00% of the

time

040% of the time

ftontlhind limb coordination

RESULTS

Effects of compressive Spinal Cord Iniuw Severity on Mortalie and Neuroloaical

Function

Mortality was less than 10% during the injury procedure and. usually occurred during

the larninectomy due to excessive blood loss andlor respiratory failure due to the anaesthetic

sensitivities of the mice. The incidence of mortality during the post-injury neurological

assessment period was zero. Following Fluoro-Gold introduction, two anirnals from the 249

injury group and two animals from the transection control group died. Following traumatic spinal

cord injury al1 mice exhibited varying levels of paraplegia and a progressive recovery of function

was observed until Day 14 post injury, where BBB and IP scores reached a plateau.

BBB Openfield Locomotor test scores are shown in Figure 4A (pg.45) as the mean

combined BBB score for each injury severity 2 S.E.M. Sham control mice displayed no

neurological deficits, thus indicating that any obsewed deficits were attributable to the force of

clip compression on the spinal cord. A graded decrease in locomotor performance was

observed with increasing severities of injury immediately post injury and continued for the

duration of the 28 day neurological assessrnent period. At the end of the 28 day post injury

assessment period. the sham control mice perfomed to a mean score of 21 on the BE8 scale,

presenting no deficits. The 39 injured mice perforrned to a mean score of 17 on the BBB scale,

presenting mild deficits. The 89 injured mice performed to a mean score of 13 on the BBB,

presenting moderate deficits. The 249 injured mice performed to a mean score of 7 on the

BBB, presenting severe deficits and; the transection control mice perfomed to a mean score of

6 on the BBB scale (See Table 2, pg.37, for a description of neurological presentation

associated with a given BBB score). A two-way repeated measures ANOVA indicated

significant differences with respect to injury seventy (p<0.0001), time (p<0.0001) and the injury

severity x time interaction (p<0.0001). Posthoc painvise comparisons indicated significant

differences between al1 groups at al1 time points (pe0.05, Student-Newman-Keuls multiple

range test) except betwaen the 39 injury and the 89 injury at Day 7 and no significant

differences are found between the 249 injury and transection controls (Table 3, pg.47).

lnclined plane test scores are shown in Figure 48 (pg.45) as the mean inclined plane

angle for each injury severity + S.E.M. There was also a graded decrease in inclined plane

performance with increasing severities of injury immediately following injury and continued for

the duration of the IP assessment period. Maximal IP performance was observed in the sham

control group at an angle of 60' at 28 days. The 39 injury group presented with a mean angle

of 5 6 O (rnild deficits). The 89 injury group presented with a rnean angle of 5 3 O (mild-moderate

deficits). The 249 injury group presented with a mean angle of 33O (severe deficits) and; the

transection control group presented with a rnean angle of 32O on the IP 28 days post injury. A

two-way repeated measures ANOVA indicated significant differences with respect to injury

severity (p<0.0001), tirne (p<0.0001) and the injury severity x tirne interaction (pc0.0001). Post-

hoc pairwise comparisons (Table 4. pg.47) indicated significant differences between ail groups

at al1 tirne points (p<0.05, Student-Newman-Keuls multiple range test) except between the 39

injury and the 89 injury at Day 4, 10. 14. 21, 28 and between the 39 injury and sham controls at

Day 7, 10, 14, 21, 28. A significant difference was only found behnreen the 249 injury and

transection controls at Day 10.

lniunr

Acute

The injury epicenten were examined 1 day and 7 days after 89 and 249 clip

compression injury (Figure 5, pg.48). One day following 8g injury, sparing of dorsal, lateral and

to a lesser extent, ventral, white matter, central hemonhagic necrosis in the gray matter and

medial white matter was evident. At 7 days following 89 injury, necrotic regions were

diminished, white matter vacuolization, sparing of dorsal and lateral white matter and an

infl ammatory infiltrate (mononuclear leukocytes) appeared to be present. In the 249 injury, at 1

day and 7 days, there was more extensive hemorrhagic necrosis and white matter sparing was

only seen in the subpial rim. By 7 days following 249 injury, a central cyst had formed. In al1 of

the injuries, cells that appeared to be rnononuclear leukocytes, were observed at the injury site

and there appeared to be preferential damage in the ventral regions of the spinal cord.

Chronic

The injury epicenters in al1 three compressive injury severities (39, 89 and 24g), at 5

weeks, were most notably charactenzed by an extensive network of scar tissue that may be of

glial (astrocytes, oligodendrocytes, Schwann cells) or fibrous origin (Figures 6A, B, C and F,

pg .5O) which extended both rostrally and caudally. Diffuse microcystic cavitation was also

evident in the three compressive injuries and occurred to a greater degree in the ventral

regions of the spinal cord (Figure 6D, pg.50). The appearance of a robust infiltration of

inflammatory cells (mononuclear leukocytes) was frequently observed (Figure 6E. pg.50).

Large, central zones of cavitation which are commonly seen in rats, monkeys, and humans

were not found (Bresnahan et al., 1976; Bunge et al., 1997; Fehlings and Tator, 1995; Tator,

1995). However, two animals (one 89 and one 249 injury) displayed a smaller, but not

microcystic, central cavity (Figure 6C. pg.50). The extent of cavitation and scar tissue

progressively diminished with increasing distance frorn the injury epicenter. Examination of

tissue adjacent to the injury epicenter indicated a preferential loss of ventral homs and ventral

columns. All but one sham control appeared normal with no damage to the spinal cord.

* . c Examination of the Comp - inal Cord lniurv Site

Digital morphometric spinal cord analyses of the three compressive injury severities (39,

89 and 249) and the sham control involved quantification of the cross-sectional area of residual

tissue (normal appearing gray matter and white matter) and cavitation of sections sampled at

50pm increments from the injury epicenter to a distance of 1500pm rostral and 1500pm caudal

40

from the injury epicenter. The injury epicenter was defined as the section with the least amount

of normalized residual tissue. Measurements of residual tissue cross-sectional area and cavity

cross-sectional area were expressed as a percent of the total cross-sectional area of the

section according to the following equations in order to account for biological differences in

spinal cord dimensions between animals.

Percent residual tissue = (Cross-sectional residual tissue area/Total cross-sectional area) x 100

Percent cavity = (Cross-sectional cavity areafrota! cross-sectional area) x 100

The following equation could be applied to calculate the cross-sectional area of residual

tissue, cavity and scar tissuelinflammatory infiltrate:

Total cross-sectional area = Cross-sectional area of residual tissue + Cross-sectional area of

cavity + Cross-sectional area of scarlinflammatory tissue

Figures 6A, 6 and C (pg.50) show representative sections of the injury epicenter from

each injury severity. There was a significant difference in the preservation of percent residual

tissue between the three injury severity and the sham control groups (pe0.0001, two-way

ANOVA), at discrete 50pm distances from the epicenter (p<0.0001, two-way ANOVA) and the

group x distance interaction (p<0.0001, two-way ANOVA). Figure 7 (pg.52) depicts the rnean

percent residual tissue at every sampled interval. The sham control displayed a very minor loss

of intact tissue at the epicenter which will be further explored in the Discussion section. The

injury resulting from the 39 clip compression extended for a total of 1000pm (mild injury). The

injury from the 89 clip compression extended for a total of 1500pn (moderate injury) and; the

injury from the 249 clip compression extended for more than 2000pm (severe injury). The

severe 249 injury most clearly displayed a non-symmetrical spread of injury, with a greater

extent of injury expansion in the caudal direction. A similar non-symmetrical pattern of injury

distribution was not observed in the 39 and 89 injury groups. In the 89 and 249 injuries, the

injury spread for a distance longitudinally that was greater than the actual portion which was in

contact with the clip (1 000p) . However, the spread of 39 injury did not exceed the width of the

clip. At the injury epicenter the percent cross-sectional area of residual tissue was found to be

significantly different between al1 groups (39. 89, 249 and sham) by post hoc pairwise testing

(ps0.05, Student-Newman-Keuls multiple range test) as illustrated in Figure 8 (pg.54).

Further analyses of the percent cross-sectional a n a of cavitation also indicated

significant differences between the injury severity and sham control groups (p<0.0001, two-way

ANOVA). at distances from the epicenter (p~0.0001, two-way ANOVA) and the group x

distance interaction (peO.01, two-way ANOVA). Post hoc pairwise testing dernonstrated that at

the injury epicenter, the percent cross-sectional area of cavitation was significantly different

between al1 groups except the 39 and 89. and the 39 and sham controls (p4.05, Student-

Newman-Keuls multiple range test), as shown in Figure 8 (pg.54).

Effects of Com~ress ive Spinal Cord lnjurv Severitv on the Intear itv of Descendina Motor

Tracts

Retrograde labeling of brainstem, midbrain and cortical neurons with Fluoro-Gold was

applied to assess the integrity of axons that descended through the injury epicenter. The origin

of spared axons through the injury epicenter were labeled with Fluoro-Gold and neuron counts

were taken of the raphe nuclei. reticular formation, vestibular nuclei. red nucleus and the motor

cortex in the 39, 89, 249, sham and transection control groups. Examples of Fluoro-Gold

labeled neurons from the red nucleus are shown in Figure 9 (pg.56). The mean raw counts are

shown with the corresponding S.E.M. in Figure 10 (pg.58) and range of raw counts in Table 5

(pg.60). Due to unequal variances, one-way ANOVA was performed on the square root

transformation of the raw count data and indicated that there were statistically significant

differences between the injury severity, sham and transection control groups in al1 five areas

that were examined (pc0.0001). Results detailing al1 paiwise Student-Newman-Keuls multiple

range test post hoc tests between al1 the injury groups in each of motor regions examined are

shown in Table 5 (pg.60) by postscnpts for the sake of simplicity. Mean neuron counts

repeated the trends seen with the behavioural and morphometric assessments; 39 clip

compression produced mild injury, 89 clip compression produced rnoderate injury and 249 clip

compression produced severe injury.

Relationships Between Neuroloaical Function. Histopatholoav and Injurv Severi&

The counts of Fluoro-Gold labeled neurons and residual tissue were significantly

correlated with neurological function when analyzed independently (Table 6, pg.61). The

arnount of norrnalized area of residual tissue at the injury epicenter was very strongly

correlated with the BBB scores at Day 28, using the Pearson Correlation Coefficient (R)

(R=0.946. p<0.0001). The counts of neurons retrogradely labeled with Fluoro-Gold were also

strongly correlated with the BBB scores at Day 28. The extrapyramindal tracts: raphespinal

(R=0.814, p<0.0001), reticulospinal (R=O.812, pç0.0001) and vestibulospinal (R=O.813,

p<0.0001) tracts were correlated with BBB scores at Day 28 post SCI. The extrapyramidal

rubrospinal (R=0.747, p<0.0001) tracts, as well as the pyramidal corticospinal (R=0.776,

pc0.0001) tract were conelated with the BBB scores at Day 28.

Strong, positive correlations were identifed between the IP scores (28 days post SCI)

and number of retrogradely labeled cells in the extrapyramidal tracts: rubrospinal tract

(R=0.836. pc0.0001), raphespinal tract (R=0.801. p<0.000 1 ), reticulospinal tract (R=0.782,

pq0.0001) and vestibulospinal tract (R=0.790, p<0.0001). A weaker, positive correlation was

found between the IP scores at Day 28 following SC1 and the number of Fluoro-Gold labeled

cells in the pyramidal corticospinal tract (R=0.583, pc0.0001). In addition, the percent area of

cross-sectional residual tissue at the injury epicenter was significantly correlated with IP

function at Day 28 post SCI (R=0.781, pc0.0001).

In order to ascertain which motor tracts were contributed most significantly to the

neurological outcornes. the neuron counts from al1 the brain regions were analyzed together for

correlation with the BBB scores and IP scores at day 28 post-injury, using stepwise multiple

linear regression (Table 7, pg.61). The neuron counts from the raphe nuclei (p<0.0001) and the

motor cortex (pc0.0001) were found to be signif'icantly related to the BBB scores. The R~

(multiple correlation coefficient) was detenined to be 0.805 (F=55.89, p<0.0001) when the

counts from the raphe nuclei and motor cortex were correlated with BBB scores. Similarly, the

counts from al1 the brain regions were analyzed together, using stepwise multiple linear

regression, for determining which regions correlated significantly with the IP scores at day 28

post-injury. The counts from the vestibular nuclei (pc0.05) and the red nucleus (pc0.001) were

found to correlate significantly with the IP scores at day 28 post-injury. The R~ (multiple

correlation coefficient) was calculated to be 0.754 (F=41.34. pc0.0001).

Functional Recovery Assessed By BBB Locomotor Performance Following Spinal Cord lnjury

t S ham Con trol (n= 1 2) + 3g (n='I2)

89 (n=12) + 24g (n=12) + Transection ( ~ 1 2 ) Functional Recovery Assessed By lnclined Plane

Performance Following Spinal Cord lnjury

FIGURE 4: (A) BBB Openfield Locomotor test scores are shown as mean combined BBB score

for each injury group t S.E.M. (8) lnclined Plane test scores are shown as mean inclined plane

angle for each injury group I S.E.M.

TABLE 3: Summary of Student-Newman-Keuls Post-Hoc Comparisons of BBB Tests

' means are significantly different by one-way ANOVA (p<0.0001) a rneans are significantly different from the 3 g injury mean (pc0.05)

Day 1 Day 4 Day 7 Day 10 Day 14 Day 21 Day 28

p rneans are significantly different from the 8 g injury mean (pc0.05) y means are significantly different from the 24 g injury mean ( ~ ~ 0 . 0 5 ) S means are significantly different from the sham control mean ( ~ ~ 0 . 0 5 ) E means are significantly different from the transection control mean (p<0.05)

TABLE 4: Summary of Student-Newman-Keuls Post-Hoc Comparisons of IP Tests

Sham* 1 3g* 1 8g* 24g* 1 Transection* ay& aySs

y SE U ~ S E a y S ~ uy6s W~SE

aB6 apS ap6 aps ap6 af3S

aBys apys apya aPy& upys a&& a P y ~

Day 1 Day 4

* means are significantly different by one-way ANOVA (pc0.0001) a rneans are significantly different from the 3 g injury mean ( ~ ~ 0 . 0 5 ) p means are significantly different from the 8 g injury mean (pc0.05) 7 rneans are significantly different from the 24 g injury mean (pc0.05)

aPS apS apS aps apS apS

PI~E P y 8 ~ YSE P y 6 ~ ( 3 ~ 8 ~ p y 6 ~ py6s

~ a y 7

6 means are significantly different from the sham control mean (pe0.05) E means are significantly different from the transection control mean ( ~ ~ 0 . 0 5 )

a86 1 aPS

Sham* aDys aBrs

apys

3g* P{~E YSE 'JSE

Day 10 ' a P y ~

8g* a y S ~ ?SE

76s y& y 9 ~

Day 14 Day 21

*/6&

apys a Pys

24g* afls ai36

y€)& 76s $ 5 ~

Transection* a06 aB8

aPS a06 ~ P S E as6 aPS

aPy8 a06 a@

Dor

t Ventral

FIGURE 5: Cross-sections of mouse spinal cords. stained with Luxol Fast Blue and H&E, at the

injury epicenter following clip compression (A) 1 day after 89 injury, (B) 7 days after 89 injury.

(C) 1 day after 249 injury and (D) 7 days after 249 injury. Closed arrows indicate areas of white

matter sparing, open arrows indicate regions of hemorrhage and double arrows indicate zones

of cavity. Scale bar is equal to 100 microns.

FIGURE 6: Cross-sections of rnouse spinal cords, stained with Luxol Fast Blue and H&E, 5

weeks following (A) 39, (B) 89, (C) 249 clip compression spinal cord injury viewed under low

power magnification. Examples of (D) ventral microcystic cavity (E) clusters of inflammatory

cells (mononuclear leukocytes) and (F) swirls of scar tissue (astrocytes, oligodendrocytes,

Schwann cells or fibroblasts) 5 weeks after SC1 viewed under high power rnagnification. Scale

bars are equal to 100 microns.

Distribution of Residual Tissue Following SC1

--

- i ~ - Sharn Control t 39 lnjury -t 89 lnjury

249 lnjury

-2000 -1500 -1000 -500 O 500 1000 1500 2000

Distance from Epicenter (pm)

FIGURE 7: Extent of tissue damage at 50pm rostral-caudal increments from injury epicenter

shown as by the measurement of percent cross-sectional area of residual tissue. lnjury

epicenter is denoted by O with positive (+) values indicating caudal to the injury epicenter and

negative (-) values indicating rostral to the epicenter.

CrossSectional Area of Cavitation and Residual Tissue at the lnjury Epicenter

Residual Tissue Cavitation

I I

Sham 39 89 249

lnjury Group

FIGURE 8: Percent cross-sectional area of residual tissue and cavitation + S.E.M. at the injury

epicenter. An overall significant difference was found in the area of residual tissue and

cavitation (p<0.0001, two-way ANOVA). The percent cross-sectional area of residual tissue

was significantly different between al1 groups (pe0.05, Student-Newman-Keuls test). The

percent cross-sectional area of cavitation was significantly different between al1 groups except

between the 39 and 89, and the 39 and sham controls (~4.05, Student-Newman-Keuls test).

FIGURE 9: Neurons in the red nucleus retrogradely labeled by fluorogold (introduced into the

cord at Tg) 5 weeks following (A) 39, (8) 89 and (C) 249 compressive spinal cord injury at T3-4.

Scale bar is equal to 100 microns.

Mean Counts of Fluoro-Gold Labeled Neurons

Raphe Nuclei l=l Reticular Formation

Vestibular Nuclei Red Nucleus

i i Motor Cortex

39 89 Transection

lnjury Group

FIGURE 10: Mean number of neurons retrogradely labeled with Fluoro-Gold in the raphe

nuclei, reticular formation, vestibular nuclei, red nucleus and the motor cortex in each injury

group k S.E.M. An overall significant difference was found between the groups (p*0.0001, one-

way ANOVA).

TABLE 5: Counts of Neurons Retrogradely Labeled by Fluoro-Gold

Raphe Nuclei

Range: 545-1 559

Mean: 921

Range: 41 5-978

Mean: 557'"

24g*

Sham*

Transection*

Reticular Formation

Range: 1 191 -6573

Mean: 424IY"

Range: 1-331

Mean: 1 14"~*~

Range: 347-1400

Mean: 7 6 F

Range: 0-0

Mean: o"~'~'

Range: 7 624-3837

Mean: 2320""

Range: 7-674

Mean: 270"~'"

* means are significantly different by one-way ANOVA (p<0.0001) a means are significantly different from the 3 g injury mean by Student-Newman-Keuls post hoc test set et pc0.05. p means are significantly different from the 8 g injury mean by Student-Newman-Keuls post hoc test set at p4.05. y means are significantly different from the 24 g injury mean by Student-Newman-Keuls post hoc test set at pq0.05. 6 means are significantly different from the sham control mean by Student-Newman-Keuls post hoc test set at pc0.05. E means are significantly different from the transection control mean by Student-Newman-Keuls post test set at pc0.05.

Vestibular Nuclei

Range: 578-1 473

Mean: 1 062"~'

Range: 2387-6445

Mean: 3847w

Range: 0-0

Mean: odh"

Red Nucleus Motor Cortex

Mean: 6 0 V Mean: 940['~*

Range: 315-820

Mean: 56 1

Range: 1-1 77

Mean: 56"'Wi

Range: 159-2360

Mean: 13431'w

Range: 0-0

Mean: o"~*

Range: 479-1 034

Mean: 660'"

Range: 7-325

Mean: 124"

Range: 543-1 025 Range: 1 488-3001

Mean: 746Y" Mean: 204laflYc

Range: 0-0 Range: 0-0

Mean: 6 Mean: O""

Range: 0-26

Mean: 13""

Range: 0-3

Mean: lu"

TABLE 6: lndependent Correlations Between Histopathological Outcome Measures and Neurological Outcome Measures

Number of Cells Labeled in Raphe Nuclei

BBB Score at Day 28

1 Nurnber of Cells Labeled in / Reticular Nuclei

IF Score at Day 28

Number of Celis LabeIed in Vestibular Nuclei

Number of Cells Labeled in Red Nucleus

Normalized Cross-Sectional Area of Residual Tissue at

Epicenter

Number of Cells Labeled in Motor Cortex

TABLE 7: Summary of Correlations of Cell Counts from Brain Regions using Multiple Linear Regression

R=0.747*

R=0.776*

R=0.836*

868 scores at Day 28

Raphe Nuclei Motor Cortex

Regression Conelaïion Coefficient (R~)

- Neurological Outcomes

IP scores at Day 28

Brain Regions which Correlated with Neurological Outcornes

l Vestibular Nudei l Red Nucleus

* indicates p<0.0001

DISCUSSION

The results of this study demonstrate that SC1 can be modeled in the murine spinal cord

using rapid compression induced by a modified aneurysrn clip. Quantitative behavioural and

histopathological analyses characterized three distinct grades of injury severity (39, 89 and

249) in a dose-response style. Moreover, the behavioral and histopathological paramaters were

found to correlate with each other. In particuiar, the counts of Fluoro-Gold labeled neurons in

the raphe nuclei and the motor cortex correlated with BBB scores and counts in the vestibular

nuclei and red nucleus correlated with IP scores.

dvantaaw and Disadvantaaes of the Clip Compressive Model of SC1

This technique of producing injury is an ideal method due to its simplicity in application

and clinical relevance. The modified aneursym clip has the following advantages compared to

other kinetic models of injury:

The injury consists of an initial contusion, plus a persisting compression, which is a

common feature of clinical SCI.

The spinal cord sustains both anterior and posterior mechanical damage, which is also a

feature of clinical SCI.

The concem of spinal column stabilization during injury is abolished due to the presence of

the lower blade of the clip.

Low variability in the injury event due to regular maintenance and calibration of the clip.

The relatively low cost and ease of use of the clip.

Some disadvantages of the clip compressive model of SC1 in mice exist. These

limitations include:

The surgical technique required to produce injury requires a laminectomy procedure. thus

reducing the pressure exerted by the bone following traumatic SCI. After human SCI, the

spinal cord sustains added pressure from the vertebrae.

2. The laminectomy and hook dissection prior to clip application requires practice to ensure

that the surgical procedure does not produce additional injury to the spinal cord.

3. The absence of large central cavities in the mouse spinal cord following injury, compared to

SC1 in other rnammals (including humans), may indicate that there are important

pathophysiological differences between mice and other species which need to be

considered when results from mouse SC1 studies are extrapolated to other species, such as

humans.

Behavioural Responses of Mice to Spinal Cord lniurv

The BBB scores from the three clip compression injury groups were well distributed

along the scale, demonçtrating light. moderate and severe injury severities. However, the IP

scores did not exhibit a moderate injury severity after Day 28, rather the scores appeared to

cluster either above an angle of 50° or below 35". This finding may indicate that mice recover

postural stability more rapidly than locomotor function since moderate IP scores were attained

at earlier time points following injury. Also, this bimodal distribution in IP scores may highlight

the need for another injury severity between 89 and 249 to be tested for IP performance.

Neurological recovery is non-linear in rnice and plateaus in half the time that rats recover (by

Day 14 post-injury) as illustrated by BBB and IP (Agrawal and Fehlings, 1997; Basso et al.,

1995; Nashmi et al., 1997: Schumacher et al., 2000). This decreased length of recovery to

plateau could be due to a number of factors including: a decreased length of time for the

resolution of primary and secondary injury events, different cellular repair responses, synaptic

plasticity, sprouting andlor differences in compensatory pathways (Blight, 1993; Tator. 1998).

An expeditious reorganization of spared pathways after incomplete injury could explain the

eariier neurological recovery plateau seen in rnice compared to rats. Another recovery

mechanism that could account for the earlier plateau of motor function is an earlier

denouement of conduction block of descending fibres by a reduction in edema, hematomyelia

and remyelination of axons following spinal cord trauma (Little et al., 1999). Moreover,

63

differences in the central pattern generators behnreen rats and mice might contribute to the

differences seen in the rate of recovery between these rodent species.

Transection control mice also underwent recovery from early timepoints following injury

until the end of the neurological assessrnent period. This recovery exhibited by the transection

control mice could also be attributable to a number of events, including the resolution of

pnmary and secondary injury events. resolution of spinal shock, transition from a hyporeflexic

state to one of hyperreflexia and spacticity (Wight, 1993; Little et al., 1999; Tator, 1998).

Indeed. the lack of supraspinal input has been thouçh! !O be an important mechanism of

producing hyperreflexia because the local rnotoneuron pools have an increased excitability due

to a lack of descending inhibition.

Patholo~ical Responses of the Mouse Spinal Cord to Injury

60th methods (rnorphometric analyses and counts of Fluoro-Gold labeled neurons)

used to assess histopathological differences between the clip compression injury severities

established three distinct injury severities: light, moderate and severe. At acute time points

following injury, central cavities appeared in the mouse spinal cord following clip compression

injury. The cavities were not well-defined and were found in the 89 group 1 day after injury, as

well as in the 249 group 1 day and 7 days after injury.

The response of the murine spinal cord to compressive injury at chronic time points

verified previous accounts of injury site morphology (Jakeman et al., 2000; Kuhn and Wrathall,

1998; Ma et al., 1999; Wang et al., 1997; Zhang et al., 1996; Zhang et al., 1998). A lack of

central cavitation in the chronically injured mouse spinal cord and microcystic cavitation

appean to charactenze the response of the murine spinal cord to injury. In place of this

cavitation, a robust response of reactive non-neuronal cells and glial cells seems to occur

(Fujiki et al., 1996; Kuhn and Wrathall, 1998; Ma et al., 1999; Wang et al.. 1997; Zhang et al.,

1996; Zhang et al., 1998). The reasons for these differences have been postulated to be

attributable to differences in secondary injury processes and cellular responses of the tissue

64

(Kuhn and Wrathall, 1998; Wang et al., 1997). This rnay be mediated by differences in the

immune system or vascular structure of the mouse spinal cord cornpared to other species that

undergo central cavitation following SCI. Indeed, studies of vascular changes in the rat and

human spinal cord have been shown to mediate ischemia and result in hemorrhagic necrosis

following SC1 (Koyanagi et al., 1993; Koyanagi et al., 1993; Koyanagi et al., 1993; Tator and

Koyanagi. 1997). There may also be other genetic differences between rats and mice, and

behnleen strains of mice (Steward et al., 1999). For example, unpublished results have shown

that the GIuR2 receptor is localized to the white matter of the CD1 mouse spinal cord, contrary

to what has been found in the rat (Joshi and Fehlings, unpublished data). This difference in

glutamate receptor distribution may contribute to mechanisms of excitotoxic cell death and lead

to the differing pathologies seen in the rat and the mouse following spinal cord trauma.

Counts of neurons retrogradely labeled with Fluoro-Gold were shown to be a reliable

technique to assess injury severity. In particular, neuron countç in the sensory-motor cortex

were most susceptible to injury, as labeling was obliterated in this area with moderate SCI. In

contrast, robust labeling was observed in the reticular formation in ail injury severitieç.

elationshi~ of Tracts to BBB and Inçl ned Plane Scores

Strong independent positive correlations were found between the Fluoro-Gold labeled

cells in key motor nuclei, residual tissue at injury epicenter and neurological outcornes (BBB

and IP tests). The results of the independent correlational analyses implicated that the

extrapyramidal tracts, not the pyramidal tracts, were strongly related to IP function, confirming

the findings of Fehlings and Tator (1995) in the rat. Furthemore, independent Pearson

Correlation Coefficients described correlations of both the extrapyramidal (raphespinal,

reticulospinal, vestibulospinal and nibrospinal) and the pyramidal (corticospinal) tracts and. the

BBB openfield locomotor test.

The results of stepwise multiple Iinear regression indicated that the BBB scores were

related to counts of retrogradely labeled cells in the raphe nuclei, and the motor cortex. Since

65

these counts represent the integrity of specific tracts through the injury epicenter, the BBB

openfield locomotor test rnay be an important outcome to assess the survival of the raphespinal

and corticospinal tracts in rnice.

The finding that the raphespinal pathways correlated with BBB function was an

unexpected result given that this pathway has been implicated in the descending control of pain

(Paxinos, 1995). Indeed, the raphespinal tracts may also play an important role in the openfield

movement by maintaining control of the trunk during locomotion. Also, the mathematical

modeling using stepwise multiple linear regression might have caused an artefactual result due

to a bystander effect. The raphespinal tract projects to the dorsolateral quadrant of the spinal

cord, very close to the rubrospinal projections. This reg ion of the spinal cord may be

preferentially spared following SCI, thus involved in mediating locomotor recovery. It is then

possible that during the stepwise multiple regression procedure that rubrospinal counts could

have been the actual region that correlated with the BBB test. However, because of a higher

variance, the neuron counts from the red nucleus were eliminated from the correlation model,

leaving the raphe nuclei counts in the model due to a lower variance. It is also conceivable that

during the counting procedure, some of the recticular formation neurons labeled with Fluoro-

Goid were mistaken and counted as labeled raphe nuclei neurons, as these nuclei are

juxtaposed to each other in the medulla.

The corticospinal tracts have been implicated in the control of skilled movements,

especially those involving precise foot placement and locomotion (Armstrong, 1988).

Locomotor recovery has not been previously reported to wholly rely on the integrity of the

corticospinal tract in rats (Eidelberg and Yu, 1981 ; Little et al.. 1988; Muir and Whishaw, 1999).

Rather, it was shown that lesioning of the corticospinal tract resulted in temporary locomotor

deficits (Eidelberg and Yu, 1981 ; Muir and Whishaw, 1999). However, it is important to note

that these findings do not diminish the importance of the corticospinal tract in locomotion.

Altematively, the initial transient deficits reported by othen (Eidelberg and Yu, 1981; Muir and

Whishaw, 1999), combined with our results that openfield hindlimb locomotion correlated

significantly with the integrity of corticospinal tracts, may imply that the pyramidal systems are

an important source of locomotor control, but that they may be compensated for by other

descending motor systems.

Using the same techniques of stepwise multiple linear regression, IP scores were

related to counts of retrogradely labeled cells in the vestibular nuclei and the red nucleus.

These findings are in agreement with a previous study in our laboratory that implicated the

vestibulospinal and rubrospinal tracts in IP performance following clip compression SC1 in rats

(Fehlings and Tator, 1 995). The vestibulospinal tracts have been considered an integral

descending motor pathway mediating locomotor recovery, especially in their role in maintaining

postural control during locomotion (Eidelberg et al., 1981; Yu and Eidelberg, 1981). It is,

therefore, possible that the IP task requires the vestibulospinal pathway for its control of

equilibrium as the plane is raised. A recent study indicated that the rubrospinal pathway was

important in producing non-skilled movements during locomotion in rats (Muir and Whishaw,

2000). Indeed, it was demostrated that ablation of the red nucleus produced an asymmetrical

gait, even during unperturbed locomotion over a flat surface (Muir and Whishaw, 2000). The

present study results suggest that IP performance can predicted by the survival of the

vestibulospinal and rubrospinal tracts in mice, and imply that that these tracts rnay be important

systems for mediating gross postural recovery .

na Technigues

When studies of tract tracing are undertaken, important controls must be established to

ensure the validity of the results. Positive (sham) and negative (transection) control animals

were examined in the present study. The sharn controls exhibited the greatest amount of

labeling, whereas, the transection control animals exhibited no labeling. These controls

provided benchmarks that allowed the comparisons of the graded clip compressive injury

severities. Also, the introduction site of the retrograde tracer, Fluoro-Gold, was examined to

detemine the passive spread of the tracer during implantation. In this study, the diffusion of

67

Fluoro-Gold, on the proximal stump to the injury was not found to be in excess of two segment.

Thus, the F luoro-Gold introduction site (at T9-10) was sufficiently caudal to the injury site (T3-

4), ensuring that none of the tracer diffused to the site of injury and did not result in non-specific

labeling of brainstem, midbrain and cortical neurons. Furtherrnore, the duration of incubation

with Fluoro-Gold before tissue extraction was sufficiently short (5 days) to eliminate the

possibility of transport of the tracer by gap-junctional coupling of cells.

One caveat that must be considered is that Our findings of iraci integriry by the upiake

of the retrograde tracer, Fluoro-Gold, may not fully represent the functional integnty of these

tracts. To fully ascertain which tracts are involved in locomotion, in vivo conduction studies of

rnotor evoked potentials and studies that stimulate specific motor nuclei must be perforrned.

along with corresponding anatomical studies. The execution of such experiments was beyond

the scope of the present investigation.

Another limitation of the current study is that the correlations identified behrveen specific

motor pathways and behavioural function do not f o m a causal relationship. Rather the R~

values (multiple correlation coefficient) are a measure of the proportion of the variance in the

neurological performance explained by the counts of retrogradely labeled neurons in key motor

regions of the brainstern, midbrain and cerebral cortex. In order to confirrn the findings that

have been presented in the present study. studies which specifically lesion motor nuclei or

pathways, combined with detailed behavioural analyses, would address the causal

relationships between specific descending motor pathways and neurological function.

ethodologcal Cons

In the present study, one sham control mouse was found to have minor damage in the

spinal cord. This could be ascribed to two possible events. Firstly, in the sham control group,

after a two-level laminectomy. the ventral aspect of the spinal cord was dissected with a hook

to clear an extradural path; thus the sham control was not simply a laminectomy control. The

spinal cord manipulation by the dissecting hook might have caused some grossly unobserved

damage. Secondly, the possibility of a subdural hematoma, causing damage, following Fluoro-

Gold introduction can not be excluded. Although, the contribution of this observation to the

results was minimal since no statistically significant functional or histopathological deficits were

seen in the sham controls, similar controls must be performed when SC1 experiments have

manipulation of the spinal cordispinal column other than the intended injury.

. . ication of Murine Models to SC1

The motivation behind developing a graded murine rnodel of SC1 has been the desire to

apply genetically altered rnurine systems in order to study the pathophysiology of SCI. This

study characterizes a clip compressive injury at T3-4, which is anatomically higher in the spinal

cord than in the mouse SC1 models that have been presented to date in the literature. This

level of SC1 was chosen because the hook dissection was more easily accomplished at T3-4

compared to the lowar thoracic level. Also. an injury at T3-4 will affect the regulation of the

mid-thoracic intermediolateral neurons which are responsible for autonornic (cardiovascular)

control. Thus. an injury at T3-4, also models the potenüal for cardiovascular disturbances

following SC1 including: neurogenic shock, bradycardia and autonornic dysreflexia. In order to

apply the clip compressive injury model to another spinal level, sharn controls and injured

(untreated) controls must be examined to enable valid comparisons to be made between

genetically altered animals and their wildtype counterparts.

There are also other important issues that must be considered when applying injury

models to knockout mouse mutants. Physiological differences have been noted between

different inbred and knockout strains of rnice. For example, differences in motor function, size

of mouse, response to excitatory amino acid administration, inflammatory responses and

neurogenesis arnongst inbred strains have been reviewed (Crawley et al., 1997; Crawley and

Paylor, 1997; Schauwecker and Steward, 1997; Steward et al., 1999).

With these differences in mind, SC1 research in knockout mice requires additional

attention when designing studies with knockout mice. Methods to control for such differences

can be incorporated into the project design. Fundamental to any SC1 study with mutant mice,

relevant and appropriate controls must be tested. This is important because knockout mutant

mice are usually a hybrid between two background inbred strains of mice, thus any responses

in the background strain to a given stimulus could be propagated into the knockout breed. In

order to minimize artefactual findings, control mice (these are also called wildtype (+/+)

littermate mice in the literature) that are the same sex, from the same generation and have the

same parents as the knockout mutants, (identified as -1- in the literaturej shouid undergo the

same experimental procedures as the knockout mutant mice.

Genetic deletion can also present limitations in which outcome measures can be

applied to study SC1 in mutant mice. Certain knockout mutants have lethal mutations that

prohibit their use; or have syndromes which complicate their use in SC1 studies. For example,

mGluRl knockout mice display decreased coordination and decreased synaptic plasticity

(Conquet et al., 1994) and GIuR2 knockout mice also display ataxia (Jia et al., 1996). Such

phenotypic abnomalities do not preclude the use of such animals in experimental SCI, rather

these traits preclude the use of some behavioural outcome measures (i.e. openfield locomotor

testing) to study the functional recovery in the injured mice.

Future Directions

The development and characterization of an in vivo murine model of SC1 has pemitted

studies to progress with mutant mice as well as, raised many questions which merit further

study. To further characterize the model, time course light microscropic and electron

microscopie studies should be done of the onset of edema, extravasation of cellular

components, necrosis, cavitation, demyelination, rernyelination and fibro-gliotic scar formation.

Experiments that characterize the molecular events following injury are also important to

validate and further characterize this model of spinal cord injury. For example, a detailed

account of the apoptotic cascades, and their time course, that are activated in the rnouse spinal

cord would be useful to comparelcontrast with other animals, including humans. Comparative

studies of the differences in infiammatory reactions between species rnay also be useful in

ascertaining the role of certain immune responses following SCI. In addition. the integrity of

cytoskeletal protein complexes, such as NF200 and spectrin, might provide data on the

secondary injury mechanisms that ensue in the murine spinal cord. Time course studies of the

secondary injury events would be very useful in understanding which factors contribute to the

earlier recovery plateau in mice.

Differences in the pathology between SC1 in mice and in other mammais are especially

interesting. A study of the vascular structure of the mouse spinal cord would address the issue

that the lack of cavity formation observed in the chronically injured mouse spinal cord might be

due to differences in the vascular supply of the murine spinal cord. Microangiographic and

electron rnicroscopic techniques, similar to those applied by Koyanagi and Tator (Koyanagi et

ai.. 1993; Koyanagi et al.. 1993: Koyanagi et al., 1993) could be applied to the injured and non-

injured mouse spinal cord. Furthemore, such studies could also provide insights into murine

recovery mechanisms that could then be compared with those in humans. Further questions

into the mechanisms of early cavity formation and dissappearance at late tirnepoints need to be

posed and the answers might provide neurotrauma research with valuable information about

wound-healing and recovery responses.

Finally, application of the compressive, munne mode1 of SCI is already underway.

These studies are important in terms of validating the model in addition to testing hypotheses

about the role of specific genes in the pathophysiology of SCI. For example, the responses of

knockout mutant mice for the GIuR2 gene to SC1 could result in important information about the

role of calcium permeable AMPA receptors in SCI. Thus. the development of a compressive, N,

vivo, murine model of spinal cord injury has not only provided investigaton with a tool to further

study the pathophysiological mechanisms of SCI, but it has also evoked novel examinations

which can be pursued.

CONCLUSIONS

A reliable, graded, in vivo model of compressive SC1 has been established in the mouse

by customizing an existing rnethod of producing injury in the rat. Neurological and

histopathological outcome measures were found to accurately discriminate between three

grades of injury severity: 39 clip compression (light injury), 89 clip compression (moderate

injury) and 249 clip compression (severe injury). In addition, the histopathological outcome

measures correlated highly with the behavioural outcorne measures. lndependent correlational

analyses implicated that the extrapyramidal tracts, not the pyramidal tracts, were strongly

related to IP function. Furthemore, independent correlations were identified behnreen both the

extrapyramidal (raphespinal, reticulospinal, vestibulospinal and rubrospinal) and the pyramidal

(corticospinal) tracts and, the BBB openfield locomotor test. This model of in vivo SC1

establishes, for the first time, the use of the clip compression method to produce a reliable SC1

technique in mice. This model will enable future projects to be undertaken to investigate the

role of specific genes and molecules in spinal cord injury.

REFERENCES

Agrawal, S. K., and Fehlings, M. G. (1997). The effect of the sodium channel blocker QX-314

on recovery after acute spinal cord injury. J Neurotrauma 14, 81-8.

Agrawal, S. K., and 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 ine Na(+)-Ca2+

exchanger. J Neurosci 16,545-52.

Agrawal, S. K., and Fehlings, M. G. (1997). Role of NMDA and non-NMDA ionotropic glutamate

receptors in traumatic spinal cord axonal injury. J Neurosci 77, 1055-63.

Agrawal. S. K., Nashmi, R., and Fehlings, M. G. (2000). Role of L and N-Type Calcium

Channels in the Pathophysiology of Traumatic Spinal Cord White Matter Injury. Neuroscience

In press.

Agrawal, S. K., Theriault, E., and Fehlings, M. G. (1998). Role of group 1 rnetabotropic

glutamate receptors in traumatic spinal cord white matter injury. J Neurotrauma 75, 929-41.

Allen. A. (191 1). Surgery of experimental experimental lesion of spinal cord equivalent to crush

injury of fracture dislocation. J. Am. Med. Ass. 57, 878-880.

Amar, A. P., and Levy, M. L. (1999). Pathogenesis and phamacological strategies for

mitigating secondary damage in acute spinal cord injury. Neurosurgery 44, 1027-39; discussion

1039-40.

Anderson, D. K., and Hall. E. D. (1993). Pathophysiology of spinal cord trauma. Ann Emerg

Med 22, 987-92.

Anderson, T. E. (1982). A controlled pneumatic technique for experimental spinal cord

contusion. J Neurosci Methods 6, 327-33.

Anthes, O. L., Thenault, E., and Tator, C. H. j1995j. Characteriration of axonal uitras~ructurai

pathology following exparimental spinal cord compression injury. Brain Res 702, 1-1 6.

Anthes, O. L., Thenault, E., and Tator. C. H. (1996). Ultrastructural evidence for arteriolar

vasospasm after spinal cord trauma. Neurosurgery 39,804-14.

Armstrong, D. M. (1988). The supraspinal control of mammalian locomotion. J Physiol (Lond)

405, 1-37.

Balentine, J. D. (1978). Pathology of experimental spinal cord trauma. 1. The necrotic lesion as

a function of vascular injury. Lab lnvest 39, 236-53.

Balentine, J. D. (1978). Pathology of experimental spinal cord trauma. II. Ultrastructure of

axons and rnyelin. Lab lnvest 39, 254-66.

Bartholdi, D., and Schwab, M. E. (1 997). Expression of pro-inflammatory cytokine and

chemokine rnRNA upon expefimental spinal cord injury in mouse: an in situ hybndization study.

Eut J Neurosci 9, 1422-38.

Basso, D. M., Beattie, M. S., and Bresnahan, J. C. (1 996). Graded histological and locomotor

outcornes after spinal cord contusion using the NYU weight-drop device versus transection.

Exp Neurol 739, 244-56.

Basso, 0. M., Beattie, M. S., and Bresnahan, J. C. (1995). A sensitive and reliable locomotor

rating scale for open field testing in rats. J Neurotrauma 12, 1-21.

Beattie, M. S. (1992). Anatomic and behavioral outcome after spinal cord injury produced by a

displacement controlled impact device. J Neurotrauma 9, 157-9; discussion 159-60.

Behrmann, D. L., Bresnahan, J. C., Beattie, M. S., and Shah, B. R. (1992). Spinal cord injury

produced by consistent mechanical displacement of the cord in rats: behavioral and histologie

analysis. J Neurotrauma 9, 197-21 7.

Benzel, E. C., Lancon, J. A., Thomas, M. M., Beal, J. A., Hoffpauir, G. M., and Kesterson, L.

(1990). A new rat spinal cord injury model: a ventral compression technique. J Spinal Disord 3,

334-8.

Bjugn, R., Nyengaard. J. R.. and Rosland, J. H. (1997). Spinal cord transection-no loss of

distal ventral horn neurons. Modem stereological techniques reveal no transneuronal changes

in the ventral homç of the mouse lumbar spinal cord after thoracic cord transection. Exp Neurol

748, 179-86.

Black, P.. Markowitz, R. S., Cooper, V.. Mechanic, A., Kushner, H., Damjanov, 1.. Finkelstein,

S. D.. and Wachs, K. C. (1986). Models of spinal cord injury: Part 1. Static load techniqüe.

Neurosurgery 19,752-62.

Black, P., Markowitz, R. S., Damjanov, 1.. Finkelstein, S. D., Kushner, H., Gillespie, J., and

Feldman, M. (1988). Models of spinal cord injury: Part 3. Dynamic load technique.

Neurosurgery 22, 51-60.

Blight, A. R. (1 985). Delayed demyelination and macrophage invasion: a candidate for

secondary cell damage in spinal cord injury. Cent New Syst Trauma 2, 299-315.

Blight, A. R. (1991). Morphometric analysis of a mode1 of spinal cord injury in guinea pigs, with

behavioral evidence of delayed secondary pathoiogy. J Neurosurg Sci 35, 131-8.

Blight, A. R. (1996). An overview of spinal cord injury rnodels. In Neurotrauma, R. K. Narayan,

J. E. Wilberger and J. T. Povlishock, eds. (New York: McGraw Hill), pp. 1367-1379.

Blight, A. R. (1993). Remyelination, revascularization, and recovery of function in experimental

spinal cord injury. Adv Neurol 59, 91 -1 04.

Blight, A. R. (1992). Spinal cord injury models: neurophysiology. J Neurotrauma 9. 147-9;

discussion 149-50.

Blight, A. R., and Decrescito, V. (1986). Morphornetric analysis of expenmental spinal cord

injury in the cat: the relation of injury intensity to survival of myelinated axons. Neurosurgery 79,

752-62.

Bresnahan, J. C. (1 978). An electron-rnicroscopic analysis of axonal alterations following blunt

contusion of the spinal cord of the rhesus monkey (Macaca mulatta). J Neurol Sci 37,59-82.

Bresnahan, J. C.. King, J. S., Martin, G. F., and Yashon, D. (1 976). A neuroanatomical analysis

of spinal cord injury in the rhesus monkey (Macaca mulatta). J Neurol Sci 28, 521-42.

Bunge, R. P.. Puckett, W. R., and Hiester, E. D. (1997). Observations on the pathology of

several types of human spinal cord injury, with emphasis on the astrocyte response to

penetrating injuries. Adv Neurol 72, 305-1 5.

Casha, S., Yu, W. R., and Fehlings, M. G. (2000). Oligodendroglial apoptosis occurs along

degenerating axons and is associated with FAS and P75 expression following spinal cord

injury. Neuroscience (accepted with revisions).

Collins. G. H., West, N. R.. and Parmely, J. D. (1986). The histopathology of freezing injury to

the rat spinal cord. A light and electron microscope study. J. Neuropathol. Exp. Neurol. 45, 742-

757.

Conquet, F., Bashir, Z. I., Davies, C. H., Daniel, H., Ferraguti, F., Bordi. F.1 Franz-Bacon. K.,

Reggiani, A., Matarese, V.. Conde, F., and et al. (1994). Motor deficit and impairrnent of

synaptic plasticity in rnice lacking mGluR1 [see cornrnents]. Nature 372, 237-43.

Crawley, J. N.. Belknap, J. K., Collins, A., Crabbe, J. C., Frankel, W., Henderson, N.,

Hitzemann, R. J., Maxson, S. C., Miner, L. L., Silva, A. J., Wehner, J. M., Wynshaw-Boris, A.,

and Paylor, R. (1 997). Behavioral phenotypes of inbred mouse strains: implications and

recornrnendations for molecular studies. Psychopharmacology (Berl) 732, 107-24.

Crawley, J. N., and Paylor, R. (1997). A proposed test battery and constellations of specific

behavioral paradigms to investigate the behavioral phenotypes of transgenic and knockout

mice. Hom Behav 31, l97-2I 1.

Crowe, M. J., Bresnahan, J. C., Shuman, S. L., Masters, J. Na, and Beattie, M. S. (1997).

Apoptosis and delayed degeneration after spinal cord injury in rats and monkeys [published

erratum appears in Nat Med 1997 Feb;3(2):240]. Nat Med 3, 73-6.

Cutler, M. G. (1993). Comparison of the effects of yohimbine and clonidine on the behaviour of

female mice during social encounters in an "approach-avoidance" situation.

Neuropharmacology 32,411-7.

Das. G. D. (1989). Perspectives in anatomy and pathology of paraplegia in experimental

animais. Brain Res Bull 22, 7-32.

Davis, S., Heal, D. J., and Stanford, S. C. (1995). Long-lasting effects of an acute stress on the

neurochemistry and function of 5-hydroxytryptarninergic neurones in the mouse brain.

Psychopharmacology (Berl) 178,267-72.

Dohrmann, G. J.. Panjabi, M. M., and Wagner, F. C., Jr. (1 976). An apparatus for quantitating

experimental spinal cord trauma. Surg Neurol 5,315-8.

Dolan, E. J., and Tator, C. H. (1979). A new rnethod for testing the force of clips for aneurysrns

or experimental spinal cord compression. J Neurosurg 57, 229-33.

Dolan, E. J., Tator. C. H.. and Endrenyi, L. (1980). The value of decompression for acute

experimental spinal cord compression injury. J Neurosurg 53, 749-55.

Du, S., Rubin, A., Klepper, S., Barrett, C., Kim, Y. C., Rhim, H. W., Lee, E. B.. Park, C. W.,

Markelonis, G. J., and Oh, T. H. (1999). Calcium influx and activation of calpain I mediate acute

reactive gliosis in injured spinal cord. Exp Neurol 757, 96-105.

Edgerton, V. R., Roy. R. R., Hodgson, J. A., Prober, R. J., de Guzman, C. P., and de Leon, R.

(1 992). Potential of adult marnmalian lumbosacral spinal cord to execute and acquire improved

locomotion in the absence of supraspinal input. J Neurotraurna 9, S119-28.

Eidelberg, E., Staten, E., Watkins, J. C., McGraw, D., and McFadden, C. (1976). A model of

spinal cord injury. Surg Neurol 6, 35-8.

Eidelberg, E., Story, J. L., Walden, J. G., and Meyer, B. L. (1981). Anatomical correlates of

retum of locomotor function after partial spinal cord lesions in cats. Exp Brain Res 42,81-8.

Eidelberg, E., Straehley, D., Erspamer, R., and Watkins, C. J. (1977). Relationship between

residual hindlimb-assisted locomotion and surviving axons after incornplete spinal cord injuries.

Exp Neurol 56, 31 2-22.

Eidelberg, E., and Yu. J. (1981). Effects of corticospinal lesions upon treadrnill locomotion by

cats. Exp Brain Res 43, 101-3.

Emery. E., Aldana, P., Bunge. M. B., Puckett. W.. Srinivasan, A., Keane, R. W., Bethea, J., and

Levi, A. D. (1998). Apoptosis after traurnatic human spinal cord injury. J Neurosurg 89, 91 1-20.

Faden, A. 1.. Lemke, M., Simon, R. P., and Noble, L. J. (1988). N-methyl-D-aspartate

antagonist MK801 irnproves outcorne following traurnatic spinal cord injury in rats: behavioral,

anatornic, and neurochemical studies. J Neurotrauma 5,3345.

Faden, A. I., and Simon, R. P. (1 988). A potential role for excitotoxins in the pathophysiology of

spinal cord injury. Ann Neurol 23, 623-6.

Farooque, M., Isaksson, J., and Olsson, Y. (1 999). lmproved recovery after spinal cord trauma

in CAM-1 and P-selectin knockout mice. Neuroreport 10, 131 -4.

Fehlings, M. G., and Agrawal, S. (1995). Role of sodium in the pathophysiology of secondary

spinal cord injury. Spine 20, 2187-91.

Fehlings, M. G., and Nashmi, R. (1996). Changes in pharmacological sensitivity of the spinal

cord to potassium channel blockers following acute spinal cord injury. Brain Res 736, 135-45.

Fehlings, M. G., and Nashmi, R. (1997). A new mode1 of acute compressive spinal cord injury

in vitro. J Neurosci Methods 77. 21 5-24.

Fehlings. M. G., and Tator, C. H. (1995). The relationships among the severity of spinal cord

injury, residual neurological function, axon counts, and counts of retrogradely labeled neurons

after experimental spinal cord injury. Exp Neurol 132,220-8.

Fehlings, M. G., and Tator, C. H. (1988). A review of models of acute experimental spinal cord

injury. In Spinal Cord Dysfunction: Assessment, L. S. Illis, ed. (Oxford, UK: Oxford University

Press), pp. 3-33.

Fehlings, M. G., Tator, C. H., and Linden, R. D. (1989). The relationships among the severity of

spinal cord injury, motor and somatosensory evoked potentials and spinal cord blood flow.

Electroencephalogr Clin Neurophysiol 74, 244 -59.

Fehlings, M. G., Tator, C. H., Linden, R. D., and Piper, 1. R. (1987). Motor evoked potentials

recorded from normal and spinal cord-injured rats. Exp Neurol 95, 548-70.

Fernandes, K. J., Fan, D. P., Tsui, B. J., Cassar, S. L., and Tetzlaff, W. (1999). Influence of the

axotomy to cell body distance in rat rubrospinal and spinal motoneurons: differential regulation

of GAP-43, tubulins, and neurofilament-M. J Comp Neurol 41 4,495-51 0.

Fialho, S. A., Lumb, W. V., and Scott. R. J. (1 982). Pneumatically powered vertebral

displacement device for dogs. Am J Vet Res 43, 1254-7.

Ford, R. W. (1983). A reproducible spinal cord injury model in the cat. J Neurosurg 59, 268-75.

F ujiki, M., Zhang, Z., Guth, L.. and Steward, 0. (1996). Genetic influences on cellular reactions

to spinal cord injury: activation of macrophageslmicroglia and astrocytes is delayed in mice

canying a mutation (WldS) that causes delayed Wallerian degeneration. J Comp Neurol 371,

469-84.

Gao, B., and Cutler, M. G. (1992). Effects of acute and subchronic administration of propranolol

on the social behaviour of mice; an ethophannacological study. Neuropharmacology 37,749-

56.

Gaviria, M., Privat, A., diArbigny, P., Kamenka, J. M., Haton, H., and Ohanna, F. (2000).

Neuroprotective effects of gacyclidine after experimental photochernical spinal cord lesion in

adult rats: dose-window and tirne-window effects. J Neurotrâuma 17, 19-30.

Geriai, R. (1 998). Contextual leaming and cue association in fear conditioning in mice: a strain

comparison and a lesion study. Behav Brain Res 95,191-203.

Gerlai, R. (1 998). A new continuous alternation task in T-maze detects hippocampal

dysfunction in mice. A strain comparison and lesion study. Behav Brain Res 95, 91-1 01.

Griffiths, 1. R., and McCulloch, M. C. (1983). Nerve fibres in spinal cord impact injuries. Part 1.

Changes in the myelin sheath during the initial 5 weeks. J Neurol Sci 58, 335-49.

Gruner, J. A. (1992). A monitored contusion model of spinal cord injury in the rat. J

Neurotrauma 9, 123-6; discussion 1 26-8.

Gruner, J. A., Yee, A. K., and Blight, A. R. (1996). Histological and functional evaluation of

experimental spinal cord injury: evidence of a stepwise response to graded compression. Brain

Res 729,90-101.

Haghighi, S. S., Perez-Espejo, M. A., Rodriguez, F., and Clapper, A. (1 996). Radiofrequency

as a lesioning model in experirnental spinal cord injury. Spinal Cord 34, 214-9.

Hall, E. 0. (1996). Free radicals and lipid peroxidation. In Neurotrauma, R. K. Narayan. J. E.

Wilberger and J. T. Povlishock, eds. (New York: McGraw Hill), pp. 1405-141 9.

Hansebout, R. R., Blight, A. R., Fawcett, S., and Reddy, K. (1993). 4Aminopyridine in chronic

spinal cord injury: a controlled, double-blind, crossover study in eight patients [see cornments].

J Neurotrauma 10, 1-1 8.

Hashimoto. T., and Fukuda, N. (1990). New spinal cord injury mode1 produced by spinal cord

compression in the rat. J Pharmacol Methods 23, 203-12.

Hogan, E. L. (1 992). Problematic issues in spinal cord injury. J Neurotrauma 9, 161 -2;

discussion 1 62-3.

Homanics, G. E., Quinlan, J. J., and Firestone, L. L. (1999). Pharrnacologic and behavioral

responses of inbred C57BU6J and strain 129iSvJ mouse lines. Pharmacol Biochem Behav 63,

21 -6.

Hovda, D. A., Becker, D. P., and Katayama. Y. (1 992). Secondary injury and acidosis. J

Neurotrauma 9, S47-60.

Hsu, C. Y.. Hu, Z., and Doster, S. K. (1996). Cell-mediated injury. In Neurotrauma, R. K.

Narayan, J. E. Wilberger and J. T. Povlishock. eds. (New York: McGraw Hill). pp. 1433-1444.

Huang, D. W., McKerracher, L., Braun, P. E., and David, S. (1999). A therapeutic vaccine

approach to stimulate axon regeneration in the aduit mammalian spinal cord. Neuron 24, 639-

47.

lizuka, H., Yamamoto. H., Iwasaki, Y.. Yamamoto, T.. and Konno, H. (1 987). Evolution of tissue

damage in compressive spinal cord injury in rats. J Neurosurg 66, 595-603.

Isaksson, J., Farooque, M., and Olsson. Y. (2000). Spinal cord injury in CAM-l-deficient mice:

assessrnent of functional and histopathological outcorne. J Neurotrauma 17, 333-44.

Jakeman, L. B., Guan, Z., Wei, P., Ponnappan, R., Dzwonczyk, R., Popovich, P. G., and

Stokes, B. T. (2000). Traumatic spinal cord injury produced by controlled contusion in mouse

[In Process Citation]. J Neurotrauma 17, 299-31 9.

Jenkins, R., Tetzlaff, W., and Hunt, S. P. (1993). Differential expression of immediate early

genes in rubrospinal neurons following axotorny in rat. Eur J Neurosci 5, 203-9.

Jia, 2.. Agopyan, N., Miu, P., Xiong, Z., Hendenon, J., Gerlai, R., Taverna, F. A., Velumian, A.,

MacDonald, J., Carlen, P., Abramow-Newerly, W., and Roder, J. (1 996). Enhanced LTP in

mice deficient in the AMPA receptor GIuR2. Neuron 17, 945-56.

Kakulas, B. A. (1 984). Pathology of spinal injuries. Cent New Syst Trauma 1, 1 1 7-29.

Kearney, P. A., Ridella, S. A.. Viano, D. C., and Anderson, T. E. (1988). Interaction of contact

velocity and cord compression in detennining the severity of spinal cord injury. J Neurotrauma

5, 187-208.

Kerasidis, H., Wrathall, J. R., and Gale, K. (1 987). Behavioral assessment of functional deficit

in rats with contusive spinal cord injury. Paraplegia 25, 217-20.

Khan, M., and Griebel, R. (1983). Acute spinal cord injury in the rat: cornparison of three

expenmental techniques. Can J Neurol Sci 10, 16 1-5.

Khan, M., Griebel, R., Rozdilsky, B., and Politis, M. (1985). Hemonhagic changes in

experimental spinal cord injury rnodels. Can J Neurol Sci 12, 259-62.

Kobrine, A. I., Doyle, T. F., and Martins, A. N. (1978). Correlation of spinal cord blood flow and

function in experimental compression. Surg. Neurol. 70, 54-59.

Kobrine, A. I., Evans, D. E., and Rizzoli, H. V. (1979). Experirnental acute balloon compression

of the spinal cord. Factors affecting disappearance and retum of the spinal evoked response. J

Neurosurg 57, 841 -5.

Koozekanani, S. H., Vise, W. M., Hashemi, R. M., and McGhee, R. B. (1976). Possible

mechanisms for observed pathophysiological variability in experimental spinal cord injury by

the method of Allen. J Neurosurg 44,429-34.

Koyanagi, I., Tator, C. H., and Lea, P. J. (1993). Three-dimensional analysis of the vascular

system in the rat spinal cord with scanning electron microscopy of vascular corrosion casts.

Part 1 : Normal spinal cord. Neurosurgery 33, 277-83; discussion 283-4.

Koyanagi, I., Tator, C. H., and Lea, P. J. (1993). Three-dimensional analysis of the vascular

system in the rat spinal cord with scanning electron microscopy of vascular corrosion casts.

Part 2: Acute spinal cord injury. Neurosurgery 33, 285-91 : discussion 292.

Koyanagi, I., Tator, C. H., and Theriault, E. (1993). Silicone rubber microangiography of acute

spinal cord injury in the rat. Neurosurgery 32,260-8; discussion 268.

Krassioukov, A. V., and Fehlings, M. G. (1999). Effect of graded spinal cord compression on

cardiovascuIar neurons in the rostro-ventro-lateral medulla. Neuroscience 88,959-73.

Kuhn, P. L., and Wrathall, J. R. (1998). A mouse rnodel of graded contusive spinal cord injury.

J Neurotrauma 15, 125-40.

85

Kunkel-Bagden, E., Dai, H. N., and Bregman, B. S. (1 993). Methods to assess the

developrnent and recovery of locomotor function after spinal cord injury in rats. Exp Neurol 179,

1 53-64.

Kushner, H., Markowih. R. S., Mechanic, A., and Black, P. (1987). Models of spinal cord injury:

Part 2. A mathernatical model. Cent New Syst Trauma 4, 149-59.

Lang-Lazdunski, L., Matsushita, K., Hirt, L.. Waeber, C., Vonsattel, J. P., Moskowitz, M. A.. and

Dietrich. W. D. (2000). Spinal cord ischernia. Development of a model in the mouse. Stroke 31,

208-1 3.

Lavey, R. S., Taylor, J. M., Tward, J. D., Li. L. T., Nguyen, A. A.. Chon, Y., and McBride, W. H.

(1994). The extent. tirne course, and fraction size dependence of mouse spinal cord recovery

from radiation injury. Int J Radiat Oncol Biol Phys 30, 609-1 7.

Li, G. L., Farooque, M., Holtz, A., and Olsson. Y. (1999). Apoptosis of oligodendrocytes occun

for long distances away from the primary injury after compression trauma to rat spinal cord.

Acta Neuropathol (Berl) 98,473-80.

Li. S., and Stys, P. K. (2000). Mechanisms of ionotropic glutamate receptor-mediated

excitotoxicity in isolated spinal cord white matter (In Process Citation]. J Neurosci 20, 1190-8.

Little, J. W., Ditunno, J. F., Jr., Stiens, S. A., and Hams, R. M. (1999). lncornplete spinal cord

injury: neuronal mechanisrns of motor recovery and hyperreflexia. Arch Phys Med Rehabil 80,

587-99.

Little, J. W., Harris, R. M., and Sohlberg, R. C. (1 988). Locomotor recovery following subtotal

spinal cord lesions in a rat model. Neurosci Lett 87, 189-94.

Liu, D., Xu, G. Y., Pan, E., and McAdoo, D. J. (1999). Neurotoxicity of glutamate at the

concentration released upon spinal cord injury. Neuroscience 93, 1383-9.

Liu, X. Z.. Xu. X. M.. Hu, R., Du. C.. Zhang, S. X., McDonald. J. W., Dong, H. X., Wu, Y. J..

Fan. G. S., Jacquin, M. F., Hsu. C. Y.. and Choi, D. W. (1997). Neuronal and glial apoptosis

after traumatic spinal cord injury. J Neurosci 7 7, 5395-406.

Lo, Y., Taylor. J. M.. McBride, W. H.. and Withers. H. R. (1993). The effect of fractionated

doses of radiation on mouse spinal cord. Int J Radiat Oncol Biol Phys 27, 309-17.

Lo, Y. C., McBride, W. H.. and Withen, H. R. (1992). The effect of single doses of radiation on

mouse spinal cord. Int J Radiat Oncol Biol Phys 22, 57-63.

LoPachin, R. M., Gaughan. C. L.. Lehning, E. J., Kaneko, Y., Kelly, T. M.. and Blight. A. (1999).

Experimental spinal cord injury: spatiotemporal characterization of elemental concentrations

and water contents in axons and neuroglia. J Neurophysiol 82, 2143-53.

LoPachin, R. M., and Lehning, E. J. (1 997). Mechanism of calcium entry during axon injury and

degeneration. Toxicol Appl Pharmacol 743, 233-44.

Ma, M., Walters. D. M., Basso, D. M.. Stokes. B. T., and Jakeman, L. B. (1999). Recovery of

function and histological outcome following controlled contusion injury in rnice. Soc. Neurosci.

1999, Abst. 126.8.

Ma, M., Walters, P., Basso. D. M., Stokes, B. T., and Jakeman, 1. B. (1 998). Time course of

behavioral recovery after contusion injury in mouse. J. Neurotrauma 15,882.

Maiorov, D. N., Fehlings, M. G., and Krassioukov, A. V. (1998). Relationship between severity

of spinal cord injury and abnormalities in neurogenic cardiovascular control in conscious rats. J

Neurotrauma 15, 365-74.

Martin, O., Schoenen, J., Delree, P.. Gilson. V., Rogister, B., Leprince, P.. Stevenaert, A.. and

Moonen, G. (1992). Experimental acute traumatic injury of the adult rat spinal cord by a

subdural infiatable balloon: methodology, behavioral analysis, and histopathology. J Neurosci

Res 32,539-50.

Martin, S. H., and Bloedel, J. R. (1973). Evaluation of experimental spinal cord injury using

cortical evoked potentials. J. Neurosurgery 39, 75-81.

Matute, C. (1998). Characteristics of acute and chronic kainate excitotoxic damage to the optic

nerve. Proc Natl Acad Sci U S A 95, 10229-34.

Matute, C., Sanchez-Gomez. M. V., Martinet-Millan. L., and Miledi, R. (1 997). Glutamate

receptor-mediated toxicity in optic nerve oligodendrocytes. Proc Natl Acad Sci U S A 94, 8830-

5.

Metz. G. A.. Curt, A., van de Meent, H., Klusrnan, I., Schwab, M. E., and Dietz, V. (2000).

Validation of the weight-drop contusion model in rats: a comparative study of human spinal

cord injury. J Neurotrauma 17, 1-1 7.

Midha, R., Fehlings, M. G., Tator, C. H., Saint-Cyr, J. A., and Guha. A. (1987). Assessment of

spinal cord injury by counting corticospinal and rubrospinal neurons. Brain Res 410, 299-308.

Mills, L. R., Fehlings, M. G., Theriault, E., and Agrawal, S. (1995). Calcium signalling in axons

and glial cells of spinal cord white matter after trauma. J. Neurotrauma 12,432.

Muir, G. D., and Whishaw. 1. Q. (1999). Complete locomotor recovery following corticospinai

tract lesions: measurement of ground reaction forces during overground locomotion in rats.

Behav Brain Res 703,4553.

Muir, G. D., and Whishaw, 1. Q. (2000). Red nucleus lesions impair overground locomotion in

rats: a kinetic analysis. Eur J Neurosci 12, 1 1 1 3-22.

Nakai, M., Qin, Z. H., Chen, J. F., Wang, Y., and Chase. T. N. (2000). Kainic acid-induced

apoptosis in rat striaturn is associated with nuclear factor-kappa8 activation. J Neurochem 74,

647-58.

Nashmi. R., Imamura, H., Tator. C. H.. and Fehlings, M. G. (1997). Serial recording of

sornatosensory and myoelectric motor evoked potentials: role in assessing functional recovery

after graded spinal cord injury in the rat. J Neurotrauma 14, 151 -9.

Nashmi, R., Jones, O. T.. and Fehlings, M. G. (2000). Abnomal axonal physiology is

associated with altered expression and distribution of kvl .l and kv1.2 K+ channels after

chronic spinal cord injury [In Process Citation]. Eur J Neurosci 12, 491-506.

Naso, W. B., Cox, R. D., McBryde, J. P., and Perot, P. L., Jr. (1993). Rubrospinal neurons and

retrograde transport of fluoro-gold in acute spinal cord injury-a dose-response curve. Neurosci

Lett 755, 1 25-7.

Noble, L. J., and Wrathall, J. R. (1985). Spinal cord contusion in the rat: rnorphometric analyses

of alterations in the spinal cord. Exp Neurol 88, 135-49.

Noyes, D. H. (1987). Correlation between parameters of spinal cord impact and resultant injury.

Exp Neurol 95, 535-47.

Pan, W., Banks, W. A., and Kastin, A. J. (1 997). Blood-brain barrier permeability to ebiratide

and TNF in acute spinal cord injury. Exp Neurol 746, 367-73.

Pan, W., Kastin, A. J., Bell, R. L., and Olson, R. D. (1999). Upregulation of tumor necrosis

factor alpha transport across the blood-brain barrier after acute compressive spinal cord injury.

J Neurosci 79, 3649-55.

Paxinos, G. (1995). The Rat Nervous System, Second Edition (Sydney: Academic Press), pp.

1136.

Pekny, M., Johansson, C. B., Eliasson, C., Stakeberg, J., Wallen, A., Perlmann, T., Lendahl,

U., Betsholtz, C., Berthold, C. H., and Frisen, J. (1999). Abnomal reaction to central nervous

system injury in mice lacking glial fibrillary acidic protein and vimentin. J Cell Biol 145, 503-14.

Picciotto, M. R., and Wickman, K. (1998). Using knockout and transgenic mice to study

neurophysiology and behavior. Physiol Rev 78, 1 131 -63.

Regan, R. F., and Choi, D. W. (1991). Glutamate neurotoxicity in spinal cord cell culture.

Neuroscience 43. 585-91.

Ritz, L. A., Friedman, R. M., Rhoton, E. L., Sparkes. M. L., and Vierck, C. J., Jr. (1992). Lesions

of cat sacrocaudal spinal cord: a minimally disruptive model of injury. J Neurotrauma 9, 219-30.

Rivlin, A. S.. and Tator, C. H. (1978). Effect of duration of acute spinal cord compression in a

new acute cord injury model in the rat. Surg Neurol 70, 38-43.

Rivlin, A. S., and Tator, C. H. (1977). Objective clinical assessrnent of motor function after

experimental spinal cord injury in the rat. J Neurosurg 47, 577-81.

Rosenberg, L. J., Teng, Y. D.. and Wrathall, J. R. (1999). 2,3-Dihydroxy-6-nitro-7-sulfamoyl-

benzo(f)quinoxaline reduces glial ~JSS and acute white matter pathology after experimental

spinal cord contusion. J Neurosci 79,464-75.

Saito, N.. Yamamoto. T., Watanabe, T., Abe, Y., and Kumagai, T. (2000). Implications of p53

protein expression in experimental spinal cord injury. J Neurotrauma 7 7, 173-82.

Sanchez-Gomez, M. V., and Matute, C. (1999). AMPA and kainate receptors each mediate

excitotoxicity in oligodendroglial cultures. Neurobiol Dis 6. 475-85.

Schauwecker, P. E., and Steward, 0. (1997). Genetic deterrninants of susceptibility to

excitotoxic cell death: implications for gene targeting approaches. Proc Natl Acad Sci U S A 94,

41 03-8.

Schumacher, P. A., Eubanks, J. H., and Fehlings, M. G. (1999). lncreased calpain 1-mediated

proteolysis, and preferential loss of dephosphorylated NF200, following traumatic spinal cord

injury. Neuroscience 9 1, 733-44.

Schumacher, P. A., Siman. R. G., and Fehlings, M. G. (2000). Pretreatment with calpain

inhibitor CEP-4143 inhibits calpain I activation and cytoçkeletal degradation, improves

neurological function, and enhances axonal survival after traumatic spinal cord injury. J

Neurochem 74, 1646-55.

Shuman, S. L., Bresnahan, J. C., and Beattie, M. S. (1997). Apoptosis of microglia and

oligodendrocytes after spinal cord contusion in rats. J Neurosci Res 50, 798-808.

Southern, E. P., Oxland, T. R., Panjabi, M. M., and Duranceau, J. S. (1990). Cervical spine

injury patterns in three modes of high-speed trauma: a biomechanical porcine model. J Spinal

Disord 3,316-28.

Springer, J. E.. Azbill, R. D., and Knapp, P. E. (1999). Activation of the caspase-3 apoptotic

cascade in traumatic spinal cord injury. Nat Med 5, 943-6.

Steward, O., Schauwecker, P. E., Guth, L., Zhang, Z., Fujiki, M., Inman, D., Wrathall, J.,

Kempermann, G., Gage, F. H., Saatman, K. E., Raghupathi, R., and Mclntosh, T. (1999).

Genetic approaches to neurotrauma research: opportunities and potential pitfalls of murine

modeis. Exp Neurol 157, 19-42.

Stokes, B. T. (1 992). Experimental spinal cord injury: a dynarnic and verifiable injury device. J

Neurotrauma 9, 129-31; discussion 131 -4,

Stokes, B. T., and Homer, P. J. (1996). Spinal cord injury modelling and outcome assessment.

In Neurotrauma, R. K. Narayan, J. E. Wilberger and J. T. Povlishock, eds. (New York: McGraw

Hill), pp. 1395-1403.

Stokes, B. T., Noyes, D. H., and Behrmann, D. L. (1992). An electromechanical spinal injury

technique with dynamic sensitivity. J Neurotraurna 9, 187-95.

Stys, P. K., and Lopachin, R. M. (1998). Mechanisms of calcium and sodium fluxes in anoxic

rnyelinated central nervous system axons. Neuroscience 82, 21 -32.

Sutton, C. H. (1988). Graded spinal cord injuries produced in rabbits with non-invasive

microwave hyperthermia. J Am Paraplegia Soc 11.26-34.

Taoka, Y., Okajima, K., Uchiba, M., Murakami, K., Kushimoto, S., Johno, M., Naruo, M.,

Okabe, H., and Takatsuki, K. (1 997). Role of neutrophils in spinal cord injury in the rat.

Neuroscience 79, 1 177-82.

Tarlov, 1. M. (1957). Spinal cord compression: mechanism of paralysis and treatrnent

(Springfield, Illinois: Charles C. Thomas).

Tator, C. H. (1973). Acute spinal cord injury in primates produced by an indatable extradural

cuff. Can J Surg 16.222-31.

Tator, C. H. (1 998). Biology of neurological recovery and functional restoration after spinal cord

injury. Neurosurgery 42, 696-707; discussion 707-8.

Tator, C. H. (1983). Spine-spinal cord relationships in spinal cord trauma. Clin. Neurosurg. 30,

479-94.

Tator, C. H. (1995). Update on the pathophysiology and pathology of acute spinal cord injury.

Brain Pathol 5, 407-13.

Tator, C. H., and Fehiings, M. G. (1 991 j. Review of ine secondary injury lheory of acute spinal

cord trauma with ernphasis on vascular mechanisms [see comments]. J Neurosurg 75, 15-26.

Tator, C. H., and Koyanagi, 1. (1997). Vascular mechanisms in the pathophysiology of human

spinal cord injury [see comments]. J Neurosurg 86,483-92.

Tenneti. L.. and Lipton, S. A. (2000). Involvement of activated caspase-3-like proteases in N-

methyl-D-aspartate-induced apoptosis in cerebrocortical neurons. J Neurochem 74, 13442.

Tsao, J. W., Paramananthan, N.. Parkes, H. G., and Dunn, J. F. (1999). Altered brain

metabolism in the C57BUWld mouse strain detected by magnetic resonance spectroscopy:

association with delayed Wallerian degeneration? J Neurol Sci 168, 1-1 2.

Tymianski, M., and Tator, C. H. (1996). Normal and abnormal calcium homeostasis in neurons:

a basis for the pathophysiology of traumatic and ischemic central nervous system injury.

Neurosurgery 38, 1 176-95.

von Euler, M., Seiger, A.. and Sundstrom, E. (1997). Clip compression injury in the spinal cord:

a correlative study of neurological and morphological alterations. Exp Neurol 145, 502-10.

Wada, S., Yone, K., Ishidou, Y ., Nagamine, T., Nakahara, S., Niiyama, T., and Sakou, T.

(1 999). Apoptosis following spinal cord injury in rats and preventative effect of N-methyl-D-

aspartate receptor antagonist. J Neurosurg 91, 98-104.

Warnil, A. W., Wamil, B. D., and Hellerqvist, C. G. (1998). CM101-mediated recovery of walking

ability in adult mice paralyzed by spinal cord injury. Proc Nat1 Acad Sci U S A 95, 13188-93.

Wang, X., Messing, A., and David, S. (1997). Axonal and nonneuronal cell responses to spinal

cord injury in mice lacking glial fibrillary acidic protein. Exp Neurol 748, 568-76.

Watson. 0. D., Prado, R., Dietrich. W. D., Ginsberg. M. D., and Green, B. A. (1986).

Photochemically induced spinal cord injury in the rat. Brain Res 367, 296-300.

Windle, W. F., Smart, J. O., and Beers, J. J. (1951). Residual function after subtotal spinal cord

transection in adult cats. Neuroogy 8, 51 8-521.

Wrathall, J. R., Choiniere, D., and Teng. Y. 0. (1994). Dose-dependent reduction of tissue loss

and functional impairment after spinal cord trauma with the AMPNkainate antagonist NBQX. J

Neurosci 74, 6598-607.

Wrathall, J. R., Pettegrew, R. K., and Harvey, F. (1985). Spinal cord contusion in the rat:

production of graded, reproducible, injury groups. Exp Neurol 88, 108-22.

Wrathall, J. R., Teng, Y. D., and Choiniere, D. (1 996). Amelioration of functional deficits from

spinal cord trauma with systemically administered NBQX, an antagonist of non-N-methyl-D-

aspartate receptors. Exp Neurol 737, 1 1 9-26.

Wrathall, J. R., Teng, Y. D., and Mamott, R. (1 997). Delayed antagonism of AMPNkainate

receptors reduces long-term functional deficits resulting from spinal cord trauma. Exp Neurol

145, 565-73.

Yeo, J. D., Payne, W., Hinwood, B., and Kidman, A. D. (1 975). The experimental contusion

injury of the spinal cord in sheep. Paraplegia 12. 279-98.

Yong, C., Arnold, P. M., Zoubine, M. N., Citron. B. A., Watanabe, 1.. Berman, N. E., and Festoff,

B. W. (1998). Apoptosis in cellular compartments of rat spinal cord after severe contusion

injury. J Neurotrauma 15, 459-72.

Young, W. (1992). Role of calcium in central nervous systern injuries. J Neurotrauma 9, S9-25.

Yu, J., and Eidelberg, E. (1981). Effects of vestibulospinal lesions upon locomotor function in

cats. Brain Res 220, 179-83.

Yu, W. R., Westergren, H., Farooque. M., Holtz, A., and Olsson, Y. (1999). Systemic

hypothermia following compression injury of rat spinal cord: reduction of plasma protein

extravasation demonstrated by immunohistochemistry. Acta Neuropathol (Berl) 98, 15-21,

Zhang. Z., Fujiki, M., Guth, L., and Steward, 0. (1996). Genetic influences on cellular reactions

to spinal cord injury: a wound-healing response present in normal mice is impaired in mice

canying a mutation (WldS) that causes delayed Wallerian degeneration. J Comp Neurol 371,

485-95.

Zhang, Z., Guth. L., and Steward, 0. (1998). Mechanisms of motor recovery after subtotal

spinal cord injury: insights from the study of mice canying a mutation (WIdS) that delays

cellular responses to injury. Exp Neurol 149, 221-9.