TSpace Research Repository tspace.library.utoronto.ca

Osteolytic and mixed cancer metastasis modulates collagen and mineral parameters

within rat vertebral bone matrix

Burke M., Atkins A., Akens M., Willett T., and Whyne C.

Version Post-Print/Accepted Manuscript

Citation (published version)

Burke M, Atkins A, Akens M, Willett T, Whyne C. Osteolytic and mixed cancer metastasis modulates collagen and mineral parameters within rat vertebral bone matrix. J Orthop Res. 2016 Mar 30. doi: 10.1002/jor.23248. [Epub ahead of print]. PMID: 27027407

Publisher’s Statement This is the peer reviewed version of the following article: Burke M, Atkins A, Akens M, Willett T, Whyne C. Osteolytic and mixed cancer metastasis modulates collagen and mineral parameters within rat vertebral bone matrix. J Orthop Res. 2016 Mar 30, which has been published in final form at https://dx.doi.org/10.1002/jor.23248. This article may be used for non-commercial purposes in accordance with Wiley Terms and Conditions for Self-Archiving.

How to cite TSpace items

Always cite the published version, so the author(s) will receive recognition through services that track citation counts, e.g. Scopus. If you need to cite the page number of the TSpace version (original manuscript or accepted manuscript) because you cannot access the published version, then cite the TSpace version in addition to the published version using the permanent URI (handle) found on the record page.

Osteolytic and mixed cancer metastasis modulates collagen and

mineral parameters within rat vertebral bone matrix

Authors

Mikhail Burke1,2, Ayelet Atkins1, Margarete Akens3,4,Tom Willett2,3, Cari Whyne1,2,3

1Orthopaedics Biomechanics Laboratory, Sunnybrook Research Institute, Toronto, ON; 2Institute

of Biomaterials and Biomedical Engineering and 3Department of Surgery, University of Toronto,

Toronto, ON; 4Techna, University Health Network, Toronto, ON

Corresponding Author: Dr. Cari Whyne, Orthopaedics Biomechanics Laboratory, Sunnybrook

Research Institute, 2075 Bayview Ave., Room S620, Toronto, ON M4N 3M5. Phone: 416-480-

6100, ext. 5056. Email: cari.whyne@sunnybrook.ca.

Author Contribution Statements:

Mikhail Burke – Primary Author. Responsible for co-ordinating resources, study design, sample

preparation prior to analysis, data and statistical analysis and drafting the paper.

Ayelet Atkins – Responsible for Raman spectrographic study design and data acquisition and

review of paper.

Margarete Akens – Contributed to study design. Responsible for tumour cell line maintenance

and inoculation of rats with tumour cells to optimize vertebral metastases growth and review of

paper.

Tom Willett – Contributed to study design. Facilitated HPLC data acquisition and helped to

critically revise paper draft.

Cari Whyne – Senior Author. Developed study design. Provided direct links to study resources,

guidance on data analysis, critically revised paper draft.

All authors have read and approved the final submitted manuscript.

Abstract

Metastatic involvement in vertebral bone diminishes the mechanical integrity of the spine,

however, minimal data exists on the potential impact of metastases on the intrinsic material

characteristics of the bone matrix. Thirty-four (34) female athymic rats were inoculated with

HeLa (N=17) or Ace-1 (N=17) cancer cells lines producing osteolytic or mixed (osteolytic &

osteoblastic) metastases respectively. A maximum of 21 days was allowed between inoculation

and rat sacrifice for vertebrae extraction. High performance liquid chromatography (HPLC) was

utilized to determine modifications in collagen-I parameters such as proline hydroxylation and

the formation of specific enzymatic and non-enzymatic (pentosidine) cross-links. Raman

spectroscopy was used to determine relative changes in mineral crystallinity, mineral

carbonation, mineral:collagen matrix ratio, collagen quality ratio and proline hydroxylation.

HPLC results showed significant increase in the formation of pentosidine and decrease in the

formation of the enzymatic cross-link deoxy-pryridinoline within osteolytic bone compared to

mixed bone. Raman results showed decreased crystallinity, increased carbonation and collagen

quality(aka 1660/1690 sub-band) ratio with osteolytic bone compared to mixed bone and healthy

controls along with an observed increase in proline hydroxylation with metastatic involvement.

The mineral:matrix ratio decreased in both osteolytic and mixed bone compared to healthy

controls. Quantifying modifications within the intrinsic characteristics of bone tissue will provide

a foundation to assess the impact of current therapies on the material behaviour of bone tissue in

the metastatic spine and highlight targets for the development of new therapeutics and

approaches for treatment.

Introduction

Metastatic involvement in vertebral bone occurs in up to 1/3 of cancer patients1 and can manifest

as bone-resorbing (osteolytic), bone forming (osteoblastic) or a mixture of the two. Vertebral

metastasis has been linked with diminished mechanical integrity of the spine with bone

metastases leading to a high incidence (~66%) of skeletal related events (SREs), which can

significantly impact quality of life and physical function2.

Previous work has focused on structural, mechanical and mineral density changes in

metastatically involved bone with respect to the risk and pattern of fracture3. Significant co-

relations between specific CT-based stereological parameters and mechanical behaviour have

also been established4.However, these detected modifications in total bone structure or mass may

not fully account for observed SREs due to modified mechanical behaviour. The intrinsic

material characteristics of bone has been shown to be impacted by pathological conditions such

as type-I diabetes5 and should be accounted for in metastatic involvement.

Bone is a complex two-phase composite material whose material properties are dependent on

both its organic (collagen) and mineral (hydroxyapatite) components and their combined

ultrastructure. In its organic phase, the rectilinear array of collagen type-I fibrils provides bone

with enhanced ductility increasing the tissue's toughness6. This observed fibrillar structure, into

which the helical collagen-I molecules aggregate, is maintained and stabilized in part by the

formation of covalent cross-links between these molecules. These cross-links can be formed

through enzymatic pathways connecting the non-helical telopeptides of adjacent collagen

molecules7 to form immature divalent cross-links or mature trivalent cross-links or through non-

enzymatic pathways connecting the helical peptides8. A variety of enzymatic and non-enzymatic

cross-links exist within bone but studies tend to focus on the analysis of the trivalent mature

enzymatic pyridinium cross-links: pyridinoline (pyr) and deoxy-pyridinoline (dpyr); and the non-

enzymatic crosslink: pentosidine (pen)5, 9, 10. This focus is due to the natural fluorescence of these

cross-links making them easily identifiable markers in high performance liquid chromatography

(HPLC)5, 11. Decreases in pyridinium cross-links has been associated with a reduction in bending

strength and the elastic modulus of bone12. Modifications in the ratio between immature cross-

links and mature pyridinium cross-link levels within bone have been associated with changes in

observed bone strength or toughness9. Increase in non-enzymatic crosslinks have been shown to

negatively impact bone ductility, toughness and post-yield properties13-15. While no data exists

for metastatically involved bone, pathologic bone secondary to diabetes and osteoporosis has

demonstrated increases in pen levels suggesting pen level’s potential as a biomarker to

pathologic bone tissue.5, 9, 16. Note, these studies utilized bone samples for direct and absolute

measurement of cross-link concentration. While this allows for assessment of cross-link changes

due to both bone turnover and pathologically formed bone, pen levels can be inferred by

concentration detected in urine or serum due to turnover8.

Formation of these cross-links could be impacted by numerous factors. An imbalance between

produced reactive oxygen species (ROS) and the local system's ability to counteract with

sufficient anti-oxidising mechanisms and repair potential damage (known as oxidative stress)

could provide an environment conducive to oxidative reactions associated with advanced

glycation endproduct (AGE) formation9, 17. Additionally, hydroxylation of residues such as lysine

in the collagen molecule are critical in dictating the pathways of enzymatic cross-link

formation18. Modifications in collagen cross-links and hydroxylation of residues are detectible

via techniques such as high powered liquid chromatography.

Bone's mineral component primarily exists as hydroxyapatite crystals, which nucleate and grow

within the spacing between collagen fibrils. This mineral presence within bone is critical in

providing bone its strength and stiffness. While it has been theorized that mineral dimensions

would impact bone stiffness, studies have shown conflicting data between modified mineral size

and mechanical behaviour19. Modified chemistries of the calcium mineral can occur due to ionic

substitution of phosphate ion within the environment; the most common of those being with the

carbonate ion, which make up roughly 7wt% of the apatite lattice20. This type-B carbonate

substitution is suspected to be driven by the presence of non-collageneous proteins in bone and

the presence of carbonate ions has been linked to modifications of metabolic activity within

bone's localized biosystem such as buffering acidity potentially connected with bone resorption21

and stimulating bone growth22. Carbonate within the apatite structure has also been associated

with a modification in crystal structure with observed decreases in crystal dimensions or

increased lattice strain23. Some work has started to study the impact of metastasis on the mineral

crystal structure through Raman spectroscopy showing patterns of decreased crystallinity, crystal

size and increased carbonation24. This impact could also be cyclic in nature as factors such as

crystallinity and crystal size have been shown to impact metastatic cell behaviour25, 26. Raman

spectroscopy can also be utilized to assess the mineral/collagen ratio within bone tissue

indicative of tissue mineral density instead of observing total bone mass.

To understand the material behaviour of metastatically involved bone requires evaluation of

changes which occur in both organic and mineral phases. As such, this study aims to evaluate

collagen cross-linkage and mineral crystal structure in osteolytic and mixed metastatically

involved vertebral bone using well established in vivo rat models27, 28. It is hypothesized that due

to the associated increase of reactive oxygen species (ROS) production with cancer cells 29, the

increase in the oxidative stress within the bone microenvironment due to metastatic involvement

would lead to increase oxidation reactions and hence an increase in the formation of the non-

enzymatic pentosidene. Furthermore, an increase in the hydroxylation of specific collagen amino

residues is hypothesized to occur within metastatically involved bone as cancer cells have been

associated with an increase in the expression of hydroxylation enzymes such as and prolyl

hydroxylase (PH)30 and lysyl hydroxylase (LH)31. Hydroxylation impacts enzymatic cross-

linkage pathways as the cross-linkage formed differs by whether substrates/intermediates were

hydroxylated8. Additionally, previous work utilizing a murine metastatic lung-cancer model has

linked metastatic involvement to increased expression of LH2 and subsequent modification in

enzymatic collagen cross-link concentrations31. Factors such as hypoxia-inducible factor–1(HIF-

1) and signal transducer and activator of transcription 3 (STAT3), which were linked to the

observed increase in LH2, are also associated with the development or presence of bone

metastases32, 33. This, combined with the fact that the coexpression of LH and lysyl oxidase (LO),

an oxidizing enzyme key in the formation enzymatic crosslink intermediates, can occur because

both genes are targets for HIF-1 binding and are upregulated by extracellular signals associated

with metastatic behaviour31, and increases the likelihood of observed changes in the quantity of

various enzymatic cross-links formed. Finally, we predict decreased crystallinity, collagen

mineralization, crystal size and increased carbonation with metastatic involvement based on

trends observed previously24.

Material/Methods

Animal model and metastatic inoculation: Well established rat models for studying spinal

metastases were utilized27, 28, 34. These models use systemic inoculation of tumour cells to

simulate the physiologic development of vertebral metastases. The animal use protocol was

approved by the Ontario Cancer Institute prior to performing the study. Randomly assigned 5–6

week old athymic female rnu/rnu rats were inoculated with human HeLa cervical cancer cells

(previously misidentified as MT-1 cells) to create osteolytic metastases or canine Ace-1 prostate

cancer cells to produce mixed (osteolytic/osteoblastic) metastases (N=17 per group). An

additional 12 rats were used as healthy controls. The HeLa and Ace-1 cell lines were cultured in

RPMI and DMEM/F-12 media respectively at 37°C with 5% CO2 and stably transfected with the

luciferase gene to enable bioluminescent image monitoring of tumor growth within the rat via

subcutaneous injection of luciferin. Cultured cell viability was determined utilizing a trypan blue

exclusion. Rats were anesthetised using nose-cone inhalation of a 2% halothane/air mixture.

Intracardiac injections containing ~1.5×106 cells (in 0.2mL of media) were injected into the left

heart ventricle. After 14 and 21 days, bioluminescence imaging (Xenogen) was performed for

detection and semi-qualitative assessment of the degree of metastatic growth within the rat, and

particularly in the spine. The rats were euthanized 21 days after initial inoculation via CO2

asphyxiation, their vertebrae harvested and wrapped in saline dampened gauze and stored in -

20˚C freezer until testing. In cases of premature rat death the vertebrae were removed

immediately and noted accordingly.

High Performance Liquid Chromatography: HPLC was performed to quantify crosslinks using

a previously published protocol11, 35, 36. Briefly, soft tissue was removed from extracted T12-T13

vertebrae via papain digestion. Samples were defatted and then hydrolyzed in 11 M HCl at

110°C for 24 h. Samples were diluted and added to a sample buffer containing 10% acetonitrile,

1% HFBA and water plus an internal standard (pyridoxine). Pyridinoline, deoxy-pyridinoline

and pentosidine were quantified against standards of pentosidine (PolyPeptide Group,

Strasbourg, France) and pyridinoline (Qiagen, Hilden, Germany). Agilent Zorbax Eclipse XDB-

C18 Reversed-Phase C18 HPLC columns were used (150 × 4.6 mm, 5 μm particle size, 80 Å

pore size, end-capped; Agilent Technologies, Mississauga, Canada). The mobile phase for

crosslink quantification incorporated A: 26% methanol+0.1% heptafluorobutyric acid (HFBA),

B: 85% acetonitrile+0.1% HFBA and C: 38% methanol +0.08% (HFBA) at different elution

times (A: 0-18 min, 40-50 min, B: 18-30min, C: 30-40 min). A separate HPLC method utilizing

the same type of column was used to measure collagen content via hydroxyproline quantification

of the same samples using both hydroxyproline and amino acid standards (Sigma-Aldrich).

Samples were diluted with borate buffer and homoarginine (internal standard) and underwent

derivatization via the cycled addition of fluorenylmethyloxycarbonyl chloride (FMOC-Cl) +

acetone, pentane and extraction of waste top layer. .For each sample run, an elution profile

measuring fluorescence vs. elution time was obtained. The areas under the peaks were measured

and compared to a standard curve in order to calculate the concentration of each crosslink,

hydroxyproline and proline. Cross-link concentration was then normalized to collagen content.

The degree of proline hydroxylation was also determined by hydroxyproline/proline ratio.

Raman Spectroscopy: Extracted T8 vertebrae were cut in the sagittal plane with a diamond blade

(Buehler, Illinois, USA) then polished with grit papers (600, 1200, 2400, 4000 and diamond

paste 3um and 1um). The bone specimens were thawed in a 4°C refrigerator 24 hours prior to

Raman spectra acquisition. Acquisition was done with an inVia Confocal Raman Microscope

(Renishaw, Gloucestershire, UK) which includes a Renishaw spectrometer coupled into a

DMI6000 epifluorescence microscope (Leica, Wetzar, Germany). A 785 nm laser light was used

to focus on the vertebral body through a Leica HC PLAN APO 20x/0.70NA (laser power 3 mW,

spot size 4um in the y direction and 0.5um in the x direction). The bone sample was kept wet

with PBS, each sample had 3-5 locations tested and 4-6 spectra were acquired from each location

(Figure 1a), consisting of two accumulations with an exposure time of 15s. The number of

testing points chosen reflected that which was utilized in previous studies24 and point locations

were chosen on trabecular bone adjacent to abnormal voids in trabecular structure attributed to

metastatic involvement. Custom script written in Matlab was used to correct the baseline of each

spectrum. Peak intensities of phosphate ν1 (959cm−1), carbonate (1070cm−1), proline (855cm−1),

hydroxyproline (876cm−1) and amide-1(1665-1670cm−1), were determined for the mineral and

collagen properties of the bone (Figure 1b)24. For peak analysis, background signal noise was

removed by subtracting a baseline polynomial curve that best fit the data. To verify the results, a

second derivative of the spectrum was taken to attenuate the low-frequency components. Three

sub-bands (1650-1660cm−1, 1665-1670cm−1, 1680-1690cm−1) in the amide-1 region (1650-

1690cm-1) were identified and were Gaussian in nature.

From these intensities, a number of ratios were calculated for analysis. The mineral/matrix ratio

(phosphate ν1/proline+hydroxyproline) and carbonate/matrix ratio (carbonate/

proline+hydroxyproline) illuminate the degree of mineralization of the bone tissue.

Proline+hydroxyproline were chosen as the matrix factors to utilize in the mineral/matrix and

carbonate/matrix ratios due to those factors having a distinct tie to collagen-I37. Mineral

crystallinity (1/FWHM of phosphate ν1 peak), carbonate/phosphate ν1 ratio (indicative of

mineral carbonation), hydroxyproline/proline ratio (indicative of proline hydroxylation), and the

collagen quality parameter (mature/immature enzymatic cross-link ratio) were also calculated

and compared between sample groups.

Statistical data analysis: For each data set, the normality of the data was evaluated utilizing the

Shapiro-Wilk test. If the normality assumption held, then a test for the homogeneity of variance

between groups via Levene's test was performed. In data sets without equal variance, Welsh

followed by Tamhane post hoc tests for multiple comparisons were performed to test for

differences between sample groups. When equal variance and normality could be assumed,

analysis of variance followed by the Tukey post hoc test for multiple comparisons between

samples groups were used. In cases where the normality assumption could not be held, the

Kruskal-Wallis H test followed by the Dunn-Bonferroni post-hoc approach for multiple

comparisons between sample groups were used. All data are presented as mean±standard

deviation, with a significance level of p<0.05 (SPSS v23, SPSS, Chicago, IL, USA).

Results

Metastatic Involvement: Bioluminescence analysis found that 14/17 rats injected with HeLa

cells (~82% success rate) and 12/17 rats injected with Ace-1 cells (~71% success rate) displayed

metastatic tumour growth in the spine (including the approximate location of the T8-T13

vertebrae as displayed in Figure 2), as well as in the heart, lungs and femurs. The degree of

metastatic growth in each rat was random. A total of 25 rats (HeLa: N=9; Ace-1: N=7; Healthy:

N=9) were utilized for HPLC analysis. For Raman, 25 samples were used (HeLa: N=10; Ace-1:

N=9; Healthy: N=6).

Cross-link and Proline Hydroxylation Quantification via HPLC: The non-enzymatic cross-link

pentosidine showed a large statistically significant increase (2.4x, p<0.05) in observed pen levels

in osteolytic vertebrae compared to healthy controls (Figure 3a). Although total pyridinium

cross-links content showed no significant difference between healthy and metastatic involved

bone, a trend (p = 0.087) was observed between osteolytic and mixed bone samples (Figure 3b).

Deoxy-pyridinoline levels were significantly lower (p<0.05) in the osteolytic compared to the

mixed vertebrae (Figure 3c). There were no significant differences in pyridinoline levels between

the 3 groups (Figure 3d). There was a significant increase in hydroxyproline formation (~16%,

p<0.05) in osteolytic bone compared to healthy controls (Figure 4a) with no statistically

significant changes in proline concentration (Figure 4b). Osteolytic bone samples demonstrated a

trend toward an increase (~6%) in the hydroxyproline/proline ratio (Figure 4c) compared to

mixed bone and healthy controls (p = 0.053, p = 0.077, respectively).

Modifications in Mineralization, Collagen Quality and Proline Hydroxylation Ratios assessed

via Raman: Modifications in both crystallinity and crystal structure were found due to metastatic

involvement. The 1/FWHM value of ν1Phosphate peak showed a small but statistically

significant decrease (~3%) in the osteolytic group compared to both the mixed model and

healthy controls, indicative of decreased crystallinity (Figure 5a). The carbonate/phosphate ratio

in the osteolytic samples was elevated (~6%) compared to the other groups, implying increased

carbonate ion substitution in the crystal lattice (Figure 5b).

The mineral/matrix ratio showed a statistically significant decrease in the presence of the

phosphate component of hydroxyapatite relative to the proline/hydroxyproline residues specific

to collagen-I for both osteolytic (~ 15%) and mixed groups (~11%) versus healthy controls

(Figure 5c). There was also a decrease in the carbonate/matrix ratio in bone with metastatic

involvement (Figure 5d).

The collagen quality parameter, thought to be indicative of the level of the ratio between mature

and immature cross-links, was moderately reduced (5%) in osteolytic samples compared to both

mixed and healthy controls (Figure 5e). Additionally, there were observed differences in the

hydroxyproline/proline ratio between all three samples groups with osteolytic samples exhibiting

the highest ratio (~ 10% increase) (Figure 5f).

Discussion

Both the organic and mineral phases of metastatically involved vertebrae demonstrated changes

as compared to healthy controls, suggesting that tumour induced changes impact the remaining

bone at a material level. The most striking differences were seen in pentosidine (pen) levels. Pen

is an AGE crosslink between lysine and arginine found in the helical domains of adjacent

collagen molecules. Pathological conditions such as Type 1 and Type 2 diabetes considered to

negatively impact bone quality have been shown to modify the quantity of collagen cross-link

formed within bone tissue5, 9. In each case, pen levels were shown to increase. This, combined

with increased pen levels being associated with diminished bone mechanical properties13, 14,

implies an importance of pen as a potential biomarker for poor bone quality and a risk factor for

fracture. In our study, osteolytic metastatic involvement within the vertebrae (HeLa) was shown

to greatly increase the formation of pen in the bone. In general, an increase in ROS species

concentration associated with the presence of cancer cells29could facilitate the oxidation of open

chain aldehyde metabolic intermediates leading to the production of AGEs such as pen.

However, Ace-1 metastatic involvement did not impact pen levels in a statistically significant

manner. This contrast to the HeLa involvement could be due to variability in tumour burden

among samples. An increase in pen has been linked to diminished bone quality and may act as a

marker for oxidative stress and AGE production which can impact bone remodelling13, 14.

Amplified concentrations of ROS can damage the extracellular matrix (collagen) and impact

cellular metabolic reactions (osteoblasts and osteoclasts remediated bone remodelling)38.

Collagen degradation, both in terms of fragmentation and susceptibility to enzymatic

degradation, has been shown to be facilitated by ROS presence39, 40. All these factors combined

with other AGE presence may have an impact on bone's mechanical behaviour.

Pyridinoline (pyr) and deoxypyridinoline (dpyr) are formed via a hydroxyallysine dominant

(ahyl) pathway. Since ahyl versus allysine (alys) formation is dependent upon the availability of

hydroxylysine (hyl), both the action of lysyl hydroxylases (LH) and lysyl oxidases (LO) on

lysine (lys) are important in the formation of these cross-links. The level of Lys hydroxylation

has been shown to play an important role in the tissue-specific pattern of enzymatic cross-links

and defining the functionality of type I collagen in tissue18. Metastatic involvement in other

systems has been linked to increased LH and LO levels within the microenvironment31.

Additionally, ROS presence could also induce non-enzymatic hydroxylation of amino residues41.

Therefore, one might anticipate an increase in the formation of ahyl dominant enzymatic cross-

links with metastatic involvement. However, this was not the case for the observed mature

enzymatic cross-links with no significant difference between osteolytic or mixed bone compared

to healthy bone. The significant relative decrease in dpyr formation in osteolytic compared to

mixed bone however, does suggest that collagen cross-link formation pathways are impacted

differently by the type of cancer involved. Although such modifications in enzymatic cross-

linking (both mature and immature) have been shown to impact bone mechanical behaviour 12, 42,

the presence or directionality of the change in pyridinium cross-link levels in bone can differ

under different pathological conditions5, 9, 10, 43. Note, post-hoc power analysis showed that the

comparison of dpyr differences was underpowered (power=0.48). This may have impacted the

ability to detect a statistically significant difference between osteolytic and healthy bone

(p=0.114). This general decrease in pyridinium cross-link formation is likely due to competitive

utilization of lys and hlys residues in AGE formation creating a deficiency in the formation of

enzymatic cross-links. The emphasis on dpyr decrease compared to pyr might be due to the

potential increase in the hydroxylation of lysine, limiting the available of helical lys which is

required in dpyr formation but not for pyr formation. Unfortunately, we did not measure lysine

hydroxylation in this study.

Hydroxylation of proline in collagen is mediated by the enzyme prolyl hydroxylase. The

presence of hydroxyproline in collagen is critical for the stability of the collagen triple helix via

hydrogen bonding interactions. Additionally, the polar interactions between protruding hydroxyl

groups of collagen molecules within collagen fibrils could impact both the chemical and

observed physiological properties of the fibril. For example, it has been shown that

hydroxylation is favourable for HA nucleation and growth44. The expression of prolyl

hydroxylase has been linked to cancer metastases in primary tumours often associated with bone

metastasis, likely to facilitate increased extracellular matrix production to providing structure for

tumour cell adhesion, growth and taxis30, 45. Additionally, ROS species associated with metastatic

involvement could potentially lead to non-enzymatic hydroxylation of amino residues41.

Therefore metastases could facilitate an increase in non-enzymatic hydroxylation and the prolyl

hydroxylase levels in the bone micro-environment promoting increased hydroxyproline

expression and the observed increase in the hydroxyproline/proline ratio in metastatic bone

groups compared to healthy controls. Note that in this study, the modifications in

hydroxyproline/proline ratio observed among sample groups was different between HPLC and

Raman. HPLC provides absolute values on the quantity of proline and hydroxyproline residue

within all bone in the vertebrae whereas Raman analysis does not provide absolute values (just

the ratio), is focal in nature, and locations of analysis were chosen by their likely proximity to

tumour. This difference in data collection methodology could account for Raman’s perceived

increased sensitivity in detecting differences in the hydroproline/proline ratio in metastatically

involved bone and suggests that tumour impact on bone collagen and bone is proximal in nature.

Note, although hydroxyproline levels may be modified with metastatic involvement, its

expression as the gold standard in measuring collagen content was utilized.

Proline and hydroxyproline were chosen as the matrix factors to utilize in the mineral/matrix and

carbonate/matrix ratios in Raman due to those factors having a distinct tie to collagen-I37. The

observed decrease in the mineral/matrix ratio for both the osteolytic and mixed bone samples

agrees with decreases in tissue mineral density associated with metastatic involvement observed

via other modalities (i.e. µCT scanning + segmentation , pycnometry + mass measurements)4, 27,

34.

Metastatic involvement has been previously shown to both influence and be influenced by

hydroxyapatite (HA) crystallinty and crystal size25, 26. Our results revealed a decrease in

crystallinity in osteolytic samples compared to healthy controls (p<0.05) with a similar pattern

(p<0.05) in the mixed metastasis groups. This could be attributed to the observed increase in

carbonate/phosphate ratio, implying increased carbonate substitution crystal defects or may be

associated with changes in HA crystal dimensions. The observed increase in the relative

carbonation of the HA mineral could be indicative of the increased acidity of the local bone

biosystem due to both increased overall osteolytic activity associated with tumour involvement

in both metastatic models and the acidic extra-cellular fluid of the tumour tissue46. It has been

shown that acidic microenvironments favor the dissolution of phosphate compared to carbonate

in the cross-sectional areas of bone47.

In Raman spectroscopic analysis, the 1660cm-1 and the 1690 cm-1 peaks within the amide-1

region (1650-1690cm-1) are associated with the presence of mature pyr and immature deH-

DHLNL respectively. Although there is contention on the direct link of the 1660/1690 to the

ratio of mature/immature crosslinks in the bone48, it is still commonly used in Raman to provide

information on bone age and the quality of the collagen (as immature cross-links mature over

time). HPLC analysis showed no significant changes in pyr formation between the groups.

However, a decreasing 1660/1690 ratio in the osteolytic samples implies that there is a change in

the amount of immature cross-links formed in the collagen. This potential modification in the

ratio between mature and immature cross-links could impact bone’s material behaviour9.

While the results of this study show clear changes in the formation of collagen cross-links and

modifications in mineralization of the metastatically involved bone matrix, some limitations

must be noted with respect to the approach utilized. Although the presence of tumour within the

spinal area of inoculated rats was established via bioluminescence imaging, tumour burden was

not volumetrically quantified within each tested vertebrae. Therefore the potential impact of the

quantity of tumour on collagen-cross-link formation and mineralization was not evaluated. As

well, the tumour cell lines utilized were from different animal and tissue sources, however, the

immune compromised nature of the rats should minimize any impact of cell species differences

in the study. Additionally, potential sub differences between existing bone/new matrix and

osteolytic /osteoblastic bone tissue within the Ace-1 mixed metastatic model were not evaluated.

Raman provided qualitative, comparative assessment of bone quality between samples groups,

yet it does not provide absolute value assessment for specific phase characteristics (such as the

dimensional changes in the HA crystals). Finally, while pen is an AGE product and a

relationship between pen fluorescence and bulk fluorescence of all AGEs has been shown in

other studies, no clear relationship between pen formation and the formation of other non-

enzymatic cross-links has been proven

Conclusion

This study demonstrates that metastatic involvement has a clear impact on the formation of

specific non-enzymatic and mature enzymatic cross-links in vertebral bone. Future work will

need to link these cross-link modifications to potential changes in collagen microstructure, bone

mineralization and the overall material behaviour of the pathologic bone matrix. Additionally,

specific elucidation of potential differences between osteolytic and osteoblastic tissue will be

important in characterising focal impact and distinguishing sub-differences within mixed

metastatic involvement. Also, studying the relative impact of these tissue level changes to the

mechanical behaviour of vertebral bone is important for future analysis. Ultimately, by

quantifying the influence of cancer on bone tissue, a fundamental basis for understanding the

mechanics of structural changes can be established; which will be key in developing new

integrative strategies to mechanically model metastatically involved vertebral bone and highlight

areas of focus in the evaluation of new and existing treatment options.

Acknowledgements

Grant funding for this study was provided by the Canadian Institutes of Health Research (#68911

and #125886). The authors declare that there are no conflicts of interest.

Figure Legend

Figure 1 - a) Photograph of cross section of representative rnu/rnu rat T8 vertebrae. White

arrows point to Raman spectra collection points. Points were chosen on trabeculae adjacent to

seemingly larger abnormal voids in trabecular structure. b) Example of Raman spectrograph

from each sample group (osteolytic, mixed and healthy bone). Selective bands were marked with

biochemical assignments.

Figure 2 – Representative bioluminescent images of rats inoculated with HeLa (a) and Ace-1 (b)

21 days post-inoculation. Metastatic growth was observed in various regions of the rat anatomy

with the region of interest being the spinal column area (indicated by oval overlay). The dotted

square overlay indicates the approximate location of T8-T13 vertebrae.

Figure 3 - Quantitative assessment of a) pentosidine b) total pyridinium c) deoxypyridinoline

and d) pyridinoline collagen crosslinks concentrations from HPLC analysis of rat vertebral bone

within each sample group. Osteolytic bone showed increases in expression of the non-enzymatic

cross-link pentosidene and decreases in the mature enzymatic cross-link deoxypyridionline

compared to healthy and mixed bone respectively. A total of N=25 rats (HeLa: N=9; Ace-1:

N=7; Healthy: N=9) were assessed.

Figure 4 - Quantitative assessment of a) hydroxyproline and b) proline collagen residue

concentrations from HPLC analysis and the calculated c) hydroxyproline/proline ratio of rat

vertebral bone within each sample group. Osteolytic bone showed a statistical trend towards

higher hydroproline/proline ratios compared to healthy and mixed bone. A total of N=25 rats

(HeLa: N=9; Ace-1: N=7; Healthy: N=9) were assessed.

Figure 5 - Raman spectrograph-derived parameters of bone composition between sample groups.

Characteristics analyzed included a) mineral crystallinity, b) carbonate/phosphate, c)

mineral/matrix, d) carbonate/matrix ratios, e) the collagen quality parameter and f)

hydroxyproline/proline ratio. All parameters showed statistical difference between healthy and

metastatic groups. A total of N=25 rats (HeLa: N=10; Ace-1: N=9; Healthy: N=6) were assessed.

References

1. Wong DA, Fornasier VL, MacNab I. 1990. Spinal metastases: The obvious, the occult, and the

impostors. Spine (Phila Pa 1976) 15: 1-4.

2. Toma S, Venturino A, Sogno G, et al. 1993. Metastatic bone tumors. nonsurgical treatment.

outcome and survival. Clin Orthop Relat Res (295): 246-251.

3. Kaneko TS, Bell JS, Pejcic MR, Tehranzadeh J, Keyak JH. 2004. Mechanical properties,

density and quantitative CT scan data of trabecular bone with and without metastases. J Biomech

37: 523-530.

4. Nazarian A, von Stechow D, Zurakowski D, Muller R, Snyder BD. 2008. Bone volume

fraction explains the variation in strength and stiffness of cancellous bone affected by metastatic

cancer and osteoporosis. Calcif Tissue Int 83: 368-379.

5. Silva MJ, Brodt MD, Lynch MA, et al. 2009. Type 1 diabetes in young rats leads to

progressive trabecular bone loss, cessation of cortical bone growth, and diminished whole bone

strength and fatigue life. J Bone Miner Res 24: 1618-1627.

6. Wang X, Shen X, Li X, Agrawal CM. 2002. Age-related changes in the collagen network and

toughness of bone. Bone 31: 1-7.

7. Robins SP, Duncan A. 1983. Cross-linking of collagen. location of pyridinoline in bovine

articular cartilage at two sites of the molecule. Biochem J 215: 175-182.

8. Saito M, Marumo K. 2010. Collagen cross-links as a determinant of bone quality: A possible

explanation for bone fragility in aging, osteoporosis, and diabetes mellitus. Osteoporos Int 21:

195-214.

9. Saito M, Fujii K, Mori Y, Marumo K. 2006. Role of collagen enzymatic and glycation induced

cross-links as a determinant of bone quality in spontaneously diabetic WBN/Kob rats.

Osteoporosis Int 17: 1514-1523.

10. Oxlund H, Mosekilde L, Ortoft G. 1996. Reduced concentration of collagen reducible cross

links in human trabecular bone with respect to age and osteoporosis. Bone 19: 479-484.

11. Burton B, Gaspar A, Josey D, et al. 2014. Bone embrittlement and collagen modifications

due to high-dose gamma-irradiation sterilization. Bone 61: 71-81.

12. Oxlund H, Barckman M, Ortoft G, Andreassen TT. 1995. Reduced concentrations of

collagen cross-links are associated with reduced strength of bone. Bone 17: 365S-371S.

13. Hernandez CJ, Tang SY, Baumbach BM, et al. 2005. Trabecular microfracture and the

influence of pyridinium and non-enzymatic glycation-mediated collagen cross-links. Bone 37:

825-832.

14. Viguet-Carrin S, Roux JP, Arlot ME, et al. 2006. Contribution of the advanced glycation end

product pentosidine and of maturation of type I collagen to compressive biomechanical

properties of human lumbar vertebrae. Bone 39: 1073-1079.

15. Vashishth D, Gibson GJ, Khoury JI, et al. 2001. Influence of nonenzymatic glycation on

biomechanical properties of cortical bone. Bone 28: 195-201.

16. Odetti P, Rossi S, Monacelli F, et al. 2005. Advanced glycation end products and bone loss

during aging. Ann N Y Acad Sci 1043: 710-717.

17. Willett TL, Pasquale J, Grynpas MD. 2014. Collagen modifications in postmenopausal

osteoporosis: Advanced glycation endproducts may affect bone volume, structure and quality.

Curr Osteoporos Rep 12: 329-337.

18. Yamauchi M, Shiiba M. 2008. Lysine hydroxylation and cross-linking of collagen. Methods

Mol Biol 446: 95-108.

19. Adele Boskey. 2003. Bone mineral crystal size. Osteoporos Int 14 Suppl 5: S16-20;

discussion S20-1.

20. Wopenka B, Pasteris JD. 2005. A mineralogical perspective on the apatite in bone. Materials

Science and Engineering: C; NATO Advanced Study Institute (ASI on Learning from Nature

How to design New Implantable Biomaterials: From Biomineralization Fundamentals to

Biomimetic Materials and Processing Routes) NATO Advanced Study I(TRUNCATED) 25:

131-143.

21. Bushinsky DA, Lam BC, Nespeca R, Sessler NE, Grynpas MD. 1993. Decreased bone

carbonate content in response to metabolic, but not respiratory, acidosis. Am J Physiol 265:

F530-6.

22. Gibson IR, Bonfield W. 2002. Novel synthesis and characterization of an AB-type carbonate-

substituted hydroxyapatite. J Biomed Mater Res 59: 697-708.

23. LeGeros RZ, Kijkowska R, Bautista C, LeGeros JP. 1995. Synergistic effects of magnesium

and carbonate on properties of biological and synthetic apatites. Connect Tissue Res 33: 203-

209.

24. Bi X, Sterling JA, Merkel AR, et al. 2013. Prostate cancer metastases alter bone mineral and

matrix composition independent of effects on bone architecture in mice--a quantitative study

using microCT and raman spectroscopy. Bone 56: 454-460.

25. Pathi SP, Lin DDW, Dorvee JR, Estroff LA, Fischbach C. 2011. Hydroxyapatite

nanoparticle-containing scaffolds for the study of breast cancer bone metastasis. Biomaterials 32:

5112-5122.

26. Zhu W, Wang M, Fu Y, et al. 2015. Engineering a biomimetic three-dimensional

nanostructured bone model for breast cancer bone metastasis study. Acta Biomater 14: 164-174.

27. Wise-Milestone L, Akens MK, Rosol TJ, et al. 2012. Evaluating the effects of mixed

osteolytic/osteoblastic metastasis on vertebral bone quality in a new rat model. J Orthop Res 30:

817-823.

28. Engebraaten O, Fodstad Ø. 1999. Site-specific experimental metastasis patterns of two

human breast cancer cell lines in nude rats. Int J Cancer 82: 219-225.

29. Storz P. 2005. Reactive oxygen species in tumor progression. Front Biosci 10: 1881-1896.

30. Gilkes DM, Chaturvedi P, Bajpai S, et al. 2013. Collagen prolyl hydroxylases are essential

for breast cancer metastasis. Cancer Res 73: 3285-3296.

31. Chen Y, Terajima M, Yang Y, et al. 2015. Lysyl hydroxylase 2 induces a collagen cross-link

switch in tumor stroma. J Clin Invest 125: 1147-1162.

32. Hiraga T, Kizaka-Kondoh S, Hirota K, Hiraoka M, Yoneda T. 2007. Hypoxia and hypoxia-

inducible factor-1 expression enhance osteolytic bone metastases of breast cancer. Cancer Res

67: 4157-4163.

33. Abdulghani J, Gu L, Dagvadorj A, et al. 2008. Stat3 promotes metastatic progression of

prostate cancer. Am J Pathol 172: 1717-1728.

34. Won E, Wise-Milestone L, Akens MK, et al. 2010. Beyond bisphosphonates: Photodynamic

therapy structurally augments metastatically involved vertebrae and destroys tumor tissue. Breast

Cancer Res Treat 124: 111-119.

35. Bank RA, Beekman B, Verzijl N, et al. 1997. Sensitive fluorimetric quantitation of

pyridinium and pentosidine crosslinks in biological samples in a single high-performance liquid

chromatographic run. J Chromatogr B Biomed Sci Appl 703: 37-44.

36. Bank RA, Jansen EJ, Beekman B, te Koppele JM. 1996. Amino acid analysis by reverse-

phase high-performance liquid chromatography: Improved derivatization and detection

conditions with 9-fluorenylmethyl chloroformate. Anal Biochem 240: 167-176.

37. Mandair GS, Morris MD. 2015. Contributions of raman spectroscopy to the understanding of

bone strength. Bonekey Rep 4: 620.

38. Banfi G, Iorio EL, Corsi MM. 2008. Oxidative stress, free radicals and bone remodeling.

Clin Chem Lab Med 46: 1550-1555.

39. Mukhopadhyay CK, Chatterjee IB. 1994. Free metal ion-independent oxidative damage of

collagen. protection by ascorbic acid. J Biol Chem 269: 30200-30205.

40. Hawkins CL, Davies MJ. 1997. Oxidative damage to collagen and related substrates by metal

ion/hydrogen peroxide systems: Random attack or site-specific damage? Biochimica et

Biophysica Acta (BBA) - Molecular Basis of Disease 1360: 84-96.

41. Stadtman ER, Levine RL. 2003. Free radical-mediated oxidation of free amino acids and

amino acid residues in proteins. Amino Acids 25: 207-218.

42. Knott L, Whitehead CC, Fleming RH, Bailey AJ. 1995. Biochemical changes in the

collagenous matrix of osteoporotic avian bone. Biochem J 310 ( Pt 3): 1045-1051.

43. Bank RA, Robins SP, Wijmenga C, et al. 1999. Defective collagen crosslinking in bone, but

not in ligament or cartilage, in bruck syndrome: Indications for a bone-specific telopeptide lysyl

hydroxylase on chromosome 17. Proc Natl Acad Sci U S A 96: 1054-1058.

44. Toworfe GK, Composto RJ, Shapiro IM, Ducheyne P. 2006. Nucleation and growth of

calcium phosphate on amine-, carboxyl- and hydroxyl-silane self-assembled monolayers.

Biomaterials 27: 631-642.

45. Chakravarthi BV, Pathi SS, Goswami MT, et al. 2014. The miR-124-prolyl hydroxylase

P4HA1-MMP1 axis plays a critical role in prostate cancer progression. Oncotarget 5: 6654-6669.

46. Yoneda T, Hata K, Nakanishi M, et al. 2011. Involvement of acidic microenvironment in the

pathophysiology of cancer-associated bone pain. Bone 48: 100-105.

47. Bushinsky DA, Smith SB, Gavrilov KL, et al. 2002. Acute acidosis-induced alteration in

bone bicarbonate and phosphate. Am J Physiol Renal Physiol 283: F1091-7.

48. Farlay D, Duclos ME, Gineyts E, et al. 2011. The ratio 1660/1690 cm(-1) measured by

infrared microspectroscopy is not specific of enzymatic collagen cross-links in bone tissue. PLoS

One 6: e28736.

Phosphate ν1

Hyd

roxy

pro

line

Pro

line C

arb

on

ate

Am

ide-

1

Figure 1

a)

b)

a) b)

Figure 2

Figure 3

a)

c) d)

b)

p<0.05

p<0.05

P=0.087

Figure 4

a) b)

c)

p<0.05

P=0.077 P=0.053

Figure 5

b) a)

c) d)

e) f)

p<0.01 p<0.01 p<0.05 p<0.05

p<0.01

p<0.01

p<0.05 p<0.05

p<0.05

p<0.05

p<0.05 p<0.05

p<0.05