Early-onset degeneration of the intervertebral disc and vertebral end plate in mice deficient in type IX collagen




Type IX collagen is an important component of the intervertebral disc extracellular matrix. Mutations in type IX collagen are associated with premature disc degeneration in mice and a predisposition to disc disorders in humans. The aim of this study was to assess the prevalence and timeline of intervertebral disc degeneration in mice homozygous for an inactivated Col9a1 gene.


Intact spine segments were harvested from wild-type (WT) and type IX collagen–knockout (Col9a1−/−) mice at 3, 6, and 12 months of age. Sagittal spine sections were evaluated for evidence of histologic changes, by 2 blinded graders, using a semiquantitative grading method.


There was evidence of more degeneration of the disc and end plate in the spines of Col9a1−/− mice compared with those of WT controls, at most time points. These findings were significant for the disc region at 3 and 6 months (P < 0.01) and at 12 months (P < 0.10) and for the end plate region only at 6 months (P < 0.10). Degenerative changes in the disc consisted of cellular changes and mucous degeneration. Degeneration in the end plates was associated with more cell proliferation, cartilage disorganization, and new bone formation.


A deletion mutation for type IX collagen is associated with connective tissue changes characteristic of musculoskeletal degeneration in bony and cartilaginous tissue regions. Some of the observed changes were similar to cartilage changes in osteoarthritis, while others were more similar to disc degenerative changes in humans. The finding of premature onset of intervertebral disc degeneration in this mouse model may be useful in studies of the pathology and treatment of human disc degeneration.

Arthritic degenerative disorders of the spinal intervertebral disc are the most common of the musculoskeletal conditions and have a major impact on society because of the frequency of occurrence and the economic consequences (1). Environmental factors such as physical activity and mechanical loading may explain only a subset of intervertebral disc disorders when compared with inherited genetic factors (2–5). Intervertebral disc disorders and degeneration have been associated with mutations or polymorphisms in genes encoding matrix proteins, including type IX collagen (6) and aggrecan (7), as well as with genes encoding interleukin-1 (IL-1) (8), IL-6 (9), cartilage intermediate-layer protein (10), and vitamin D receptor (11).

Mice with genetic mutations in select extracellular matrix proteins, including type II collagen (12) and type I collagen (13), have been shown to acquire structural and functional matrix alterations in the intervertebral disc. In one study, mice carrying an inactivated allele of the Col2a1 gene showed early vertebral end plate ossification and decreased glycosaminoglycan concentration in the vertebrae, end plate, and annulus fibrosus beginning at 1 month, with differences between mutant mice and wild-type (WT) controls no longer evident by the age of 9 months (12). In another study, mice heterozygous for a type I collagen mutation (Mov13 strain) were reported to show decreased compressive and tensile stiffness of the intervertebral discs, thus providing evidence of functional changes associated with collagen deficiency (13). Taken together, the results of these studies demonstrate that collagens are an important contributor to the organization and function of the intervertebral disc extracellular matrix and may represent a precipitating factor, or key partner, in the cascade of degeneration events.

Genetic analyses of human populations have identified substitution mutations in the chains of the type IX collagen molecules as being linked to a predisposition to intervertebral disc disorders (6, 14–16). Type IX collagen is a member of the fibril-associated collagen with interrupted triple helix group of collagens. These collagens act as molecular bridges between fibrillar collagens and other extracellular matrix components (17, 18). The type IX collagen molecule, in combination with type XI collagen, is a key player in the type II/IX/XI heterofibril that is an important stabilizing element in cartilaginous tissue composed of type II collagen (19). Mice expressing a transgene for Col9a1 associated with a shortened collagen α1(IX) chain exhibit degenerative disc changes that include shortening of vertebrae, reduced matrix staining for mucous material, matrix disorganization, and end plate irregularities (20). Mice homozygous for an inactivated Col9a1 gene have also been generated and present with degenerative changes in articular cartilage, beginning at an early time point (6 months of age) (21, 22). These mice also exhibit changes in the expression of proteases and mechanical function that are consistent with human osteoarthritis (OA) (23, 24).

Although type IX collagen is a major component of intervertebral disc extracellular matrix collagen (18, 25, 26), no corresponding information is available regarding the impact of a type IX collagen deletion on intervertebral disc structure and function. In this study, the lumbar spines of mice homozygous for the inactivated Col9a1 gene (Col9a1−/−) were evaluated to identify age-related and genotype-related changes in key spinal structures. The intervertebral disc and end plate regions of spines from WT and type IX collagen–deficient mice were evaluated using histologic and immunohistochemical methods. A semiquantitative grading scheme was used to quantify degeneration of the intervertebral disc and vertebral end plates. Differences noted between WT and knockout mouse spines suggest that the type IX collagen deletion leads to premature intervertebral disc degeneration.


Mouse genotyping.

Genomic DNA was isolated from the tail tips of mice (∼0.5 cm) for genotyping, as described previously (24). Primers were used that flank the neocassette for the genetic mutation (for Col9a1, forward 5′-CCAGTGAACTCCCCTTCCATT-3′ and reverse 5′-CTGGCCCTTTCTAACAACTCA-3′), with separate primers for the neomycin gene used to select for the clones (21) (for neo, forward 5′-GATCTCCTGTCATCTCACCT-3′ and reverse 5′-AGAACTCGTCAAGAAGGCGA-3′). In WT mice, the Col9a1 gene does not contain the neocassette or the neomycin gene, and only the primers Col9a1 forward and Col9a1 reverse produce a polymerase chain reaction (PCR) product of 377 bp. In the homozygous mutant mouse, the Col9a1 forward and Col9a1 reverse primers produce no byproduct, while the neo forward and neo reverse primers produce a 490-bp PCR product.

Mice were bred from a colony bred at Harvard University (24) on a strain background of C57BL/6 and were housed at Duke University. Wild-type (Col9a1+/+) and homozygous (Col9a1−/−) mice were bred and housed until they reached the desired ages (3, 6, and 12 months). A separate colony of mice were bred similarly at Harvard University and killed at 9 months of age, for use in a related study (24). Spines from these mice were harvested, wrapped in moist gauze and parafilm, and sent to Duke University via overnight delivery.

Histologic evaluation and semiquantitative grading.

For histologic evaluation, intact spine segments were harvested from WT mice (n = 4 at each time point) and Col9a1−/− mice (n = 4 at each time point), dissected of all ligamentous and muscular tissue, transected at the mid-thoracic level, and fixed for histologic evaluation and semiquantitative grading. Contiguous thoracolumbar spinal segments were explanted and fixed to wooden rods with suture material to maintain alignment during fixation and processing. Segments were formalin-fixed, decalcified in formic acid, and embedded in paraffin, using routine methods. Noncontiguous sagittal spine sections were evaluated to provide representative samples of each intervertebral disc, consistent with current practices for examining pathology in animal models of knee OA (27, 28). Nonconsecutive sections (7 μm) were obtained across each spine segment from 6 zones at a spacing of 140 μm, then stained with hematoxylin and eosin and Safranin O–fast green. In order to assess degeneration in each spine, the 2 most distal lumbar disc levels were evaluated in 3 nonadjacent zones per spine (280-μm separation).

Grading was performed on 2 motion segments from each of 4 mice at each time point and for each genotype. Sections were independently evaluated for histologic changes in the intervertebral disc and end plate regions by 2 graders (LMB, KDA) blinded to animal genotype, using a semiquantitative grading scheme (29). For each time point, images (10× magnification) of hematoxylin and eosin– and Safranin O–fast green–stained sections were evaluated by the blinded graders, using the 11 criteria specified by the grading system (Table 1). The scores assigned by the reviewers were averaged for each section for each criterion then summed to give the total intervertebral disc and end plate degeneration scores for each histologic section. This grading scheme allows for separate evaluation of the intervertebral disc (scale 0–22 points, where 22 = maximum degeneration) and the end plate (scale 0–18 points, where 18 = maximum degeneration), and allows for broad evaluation of these criteria across the entire intervertebral disc, making no provision for separate evaluation of the annulus fibrosus or nucleus pulposus regions.

Table 1. Grading for intervertebral disc and end plate regions in Col9a1−/− and WT mouse spines*
Criteria (range)Col9a1−/−WT
3 months6 months12 months3 months6 months12 months
  • *

    The 11 criteria were derived from those described by Boos et al (29). Values are the mean ± SD; n = 8 intervertebral discs per genotype and time point. WT = wild type.

Intervertebral disc region      
 Chondrocyte proliferation/ density (0–6)1.38 ± 0.451.29 ± 0.382.21 ± 0.530.88 ± 0.350.88 ± 0.501.79 ± 0.40
 Mucous degeneration (0–4)0.38 ± 0.330.96 ± 0.381.17 ± 0.500.17 ± 0.360.38 ± 0.210.96 ± 0.42
 Cell death (0–4)0.23 ± 0.180.63 ± 0.381.04 ± 0.380.15 ± 0.110.29 ± 0.281.13 ± 0.40
 Tear/cleft formation (0–4)0.58 ± 0.240.63 ± 0.211.21 ± 0.620.58 ± 0.430.54 ± 0.251.21 ± 0.25
 Granular changes (0–4)0.17 ± 0.310.02 ± 0.060.27 ± 0.470.0 ± 0.00.02 ± 0.060.0 ± 0.0
Vertebral end plate region      
 Cell proliferation (0–4)1.33 ± 0.311.42 ± 0.431.33 ± 0.311.25 ± 0.460.96 ± 0.421.25 ± 0.46
 Cartilage disorganization (0–4)0.54 ± 0.351.38 ± 0.382.47 ± 0.390.88 ± 0.591.04 ± 0.522.21 ± 0.25
 Cartilage cracks (0–4)0.19 ± 0.250.79 ± 0.310.83 ± 0.620.29 ± 0.280.63 ± 0.521.54 ± 0.82
 Microfracture (0–2)0.0 ± 0.00.02 ± 0.060.38 ± 0.280.13 ± 0.150.04 ± 0.080.08 ± 0.15
 New bone formation (0–2)0.98 ± 0.111.0 ± 0.01.17 ± 0.150.56 ± 0.320.96 ± 0.080.98 ± 0.11
 Bony sclerosis (0–2)0.0 ± 0.00.08 ± 0.130.42 ± 0.350.06 ± 0.180.0 ± 0.00.15 ± 0.19

For analysis of histologic scoring, a 2-factor repeated-measures analysis of variance (ANOVA) with Bonferroni post hoc adjustment was used to detect differences in intervertebral disc and end plate scores. The intervertebral disc and end plate grades for nonadjacent sections (240-μm separation) of each motion segment were treated as 3 repeated measures. Differences between intervertebral disc and end plate histologic grades were considered significant at a 95% confidence level.

Observer reliability was determined by calculating the percentage of interrater agreement and the kappa coefficient, using JMP 6 statistical software (SAS Institute, Cary, NC). The kappa coefficient is a measure of correlation between categorical variables and is often used as a measure of interrater variability for ordinal ratings (30). According to a scheme proposed by Landis and Koch, the strength of observer agreement from kappa was defined as follows: 0.0–0.20 = slight, 0.21–0.40 = fair, 0.41–0.60 = moderate, 0.61–0.80 = substantial, and 0.81–1.00 = almost perfect (31). When the percentage of agreement between reviewers on a grading criterion was <70%, consensus was reached by both graders in conference while reviewing images in a blinded manner. Using this approach, average kappa coefficients were achieved that represented “almost perfect” strength of agreement between observers (for 3 months, κ = 0.88; for 6 months, κ = 0.96; for 12 months, κ = 0.94). Additional grading data from a prior study of 9-month-old mice by our group are also presented (32). In that study, 3 investigators blinded to animal genotype performed grading of single midline histology sections (no repeated measures) from 3 Col9a1−/− mice (n = 9 motion segments) and 2 Col9a1+/+ mice (n = 6 motion segments).

Immunohistochemical analysis.

The matrix composition and fibrillar architecture in knockout and WT mice were qualitatively assessed using immunohistochemical staining for type IX collagen and type II collagen. Spine motion segments were formalin-fixed, decalcified in formic acid, paraffin-embedded, and stained with monoclonal antibodies to type II collagen (II-II6B3; Developmental Studies Hybridoma Bank [DSHB], University of Iowa, Iowa City) and type IX collagen (D1-9; DSHB, University of Iowa). A special staining kit (HistoMouse-Plus; Zymed, South San Francisco, CA) was used to block background rodent antibodies on murine tissue sections. Prior to staining, pepsin digestion (Digest-All 3; Zymed) was used to aid in antigen retrieval. Negative (no primary antibody) controls were processed with the spine sections.


For each time point, 8 motion segments from mice of each genotype were evaluated, and these were treated as 8 independent measures, due to minimal intersubject variability. It should be noted that there was no significant intersubject variability for the intervertebral disc or end plate grading (P > 0.05 by repeated-measures ANOVA), with the exception of intervertebral disc scores at 3 months for Col9a1+/+ mice only (P = 0.04) and end plate scores at 12 months for Col9a1−/− mice only (P = 0.03).

Intervertebral disc.

The histologic evidence of intervertebral disc degeneration was greater in the spines of Col9a1−/− (knockout) mice compared with those of WT controls, as shown by higher average histologic grades (Figure 1). There was also a progressive increase in the cumulative degeneration score with increasing age, for both knockout and WT mice. The higher grade of degeneration for the Col9a1−/− mice was statistically significant for the intervertebral disc at 3 months (P < 0.01 by ANOVA with Bonferroni post hoc adjustment), 6 months (P < 0.01 by ANOVA), and at the 12-month time point (P = 0.054 by ANOVA). These findings are consistent with our previous work grading the spines of 9-month-old mice, which also demonstrated a significantly higher degeneration score for the Col9a1−/− mice compared with WT mice (P < 0.01 by ANOVA) (32). Features of intervertebral disc degenerative changes in the Col9a1−/− model, as noted by individual grading criteria, included progressive fibrochondrocyte degeneration with chondrocyte clustering and proliferation, mucous degeneration, and cell death (Table 1). Few tears or disruptions in the annulus fibrosus were noted, and these did not appear to differ between knockout mice and WT mice (Table 1). Figures 2 and 3 show representative histologic sections from Col9a1−/− mice and WT mice, respectively.

Figure 1.

Degeneration scores for the intervertebral discs (22-point scale) and end plates (18-point scale) of type IX collagen–knockout (KO) and wild-type (WT) mice. Bars show the mean and SD scores for 8 motion segments (3 nonconsecutive sections), from 2 blinded reviewers. † = mean and SD values from a prior study (32); not included in the statistical analysis. ∗ = P < 0.01 versus WT mice; ∧ = P < 0.10 versus WT mice, by analysis of variance with post hoc adjustment.

Figure 2.

Safranin O–fast green–stained motion segments from Col9a1−/− mice at 3 months (A), 6 months (B), and 12 months (C). Progressive changes in individual features with aging were apparent and included areas of chondrocyte proliferation and mucous degeneration at 3 months and 6 months, with evidence of cartilage disorganization, end-plate disruption, and sclerosis/calcification on the inferior end plate at 12 months. Bars = 0.5 mm. (Original magnification × 10.)

Figure 3.

Safranin O–fast green–stained motion segments from wild-type mice at 3 months (A), 6 months (B), and 12 months (C). Minimal degenerative changes in the intervertebral disc or end plate were observed at 3 months, while degenerative changes in the intervertebral disc and new bone formation on the end plate were more apparent at 6 months and 12 months. Bars = 0.5 mm. (Original magnification × 10.)

Vertebral end plate.

The histologic evidence of vertebral end plate degeneration was equivalent or greater in the Col9a1−/− mouse spines compared with the WT controls, as shown by higher average cumulative scoring grades (Figure 1). The higher grade of degeneration for the Col9a1−/− mice was statistically significant for the end plate, however, only at the 6-month time point (P = 0.064 by ANOVA with Bonferroni post hoc adjustment). This is consistent with related results for the 9-month-old mice: there was no evidence of a significant difference in the degeneration score in type IX collagen–deficient mice compared with WT mice (P > 0.5 by ANOVA) (32). There was also a progressive increase in scores with increasing age, for both knockout and WT mice. Features of end plate degenerative changes in the Col9a1−/− and WT mice included abnormal cell proliferation, cartilage disorganization, and new bone formation (Table 1). At 12 months of age, there was evidence of increased microfracture and bony sclerosis in the Col9a1−/− mice compared with WT mice, but no evidence of a difference in the cumulative scores (see Figures 2 and 3).

Immunohistochemical staining was performed on representative histologic segments at 3 months (Figure 4) and 12 months (Figure 5). At 3 months, the WT mice showed intense staining for type II collagen in the central nucleus pulposus and peripheral annulus fibrosus regions of the intervertebral disc and end plates (Figure 4C). The Col9a1−/− mice showed reduced staining for type II collagen in both regions of the intervertebral disc (Figure 4A). Staining for type IX collagen appeared more intense in the end plate region of the WT mice compared with a pattern of less intense, more diffuse staining of the nucleus pulposus and annulus fibrosus regions (Figure 4D), suggesting that the collagen type IX deficiency manifests more substantially in the cartilaginous regions of the end plate than in the substructures of the intervertebral disc. Aging-related differences were also observed in the staining for type II collagen and were similar in both Col9a1−/− (Figure 5A) and WT (Figure 5C) spines, including significant type II collagen disorganization at 12 months of age, with diffuse regions of bony nodule formation in the end plates that did not react with the antibody. The level of staining for type IX collagen in the spines of 12-month-old WT mice appeared reduced from that at 3 months, with evidence of increased staining in the vertebral end plate (Figure 5D). There was no evidence of type IX collagen staining in the spines of 3-month-old or 12-month-old Col9a1−/− mice, consistent with what was observed in the negative controls (Figures 4B and 5B).

Figure 4.

Aminoethylcarbazole (AEC) chromagen staining for anti–type II collagen (A and C) and anti–type IX collagen (B and D) in spine segments from 3-month-old Col9a1−/− mice (A and B) and wild-type mice (C and D). Bar = 0.5 mm. (Original magnification × 10.)

Figure 5.

Aminoethylcarbazole (AEC) chromagen staining for anti–type II collagen (A and C) and anti–type IX collagen (B and D) in spine segments from 12-month-old Col9a1−/− mice (A and B) and wild-type mice (C and D). Bar = 0.5 mm. (Original magnification × 10.)


Early signs of degeneration were observed in both the intervertebral disc and end plate regions of the type IX collagen–deficient mice beginning at 3 months of age and 6 months of age, respectively, using the grading scheme described by Boos and coworkers (29). According to this system, features of morphologic change are evaluated in the intervertebral disc and end plate regions that are associated with aging and atypical changes in humans. A degenerative or pathologic tissue is thus classified as one that is graded as older or younger than would be expected for its chronologic age. Many of the observed changes in the intervertebral disc region were common to human intervertebral disc degenerative disease, including mucous degeneration, granular changes, and cell death. Evidence of chondrocyte proliferation involving multiple chondrocytes in small groups, and regions of staining for Safranin O, were also observed but are more consistent with OA cartilage changes and disc herniation changes than the senescent changes often reported for intervertebral disc aging and degeneration (33, 34).

Cell death appeared to be a factor of interest only at the 6-month time point and was noted as areas in which we could not identify any cell nuclei. There was minimal evidence of tear or cleft formation in the intervertebral disc in the annulus fibrosus or nucleus pulposus regions and no differences in these categories between mutant and WT mice. The accelerated degeneration of the intervertebral disc regions in Col9a1−/− mouse spines was apparent at the earliest time point (3 months) and certainly preceded that in the end plate, supporting the hypothesis that disc degeneration initiates in the central intervertebral disc (35). The extent of intervertebral disc degeneration in the mice, even at 12 months, was relatively mild when compared with the full range of the grading scheme (22 points). This observation is consistent with the ages of the mice (3–12 months), as compared with a maximum age for C57BL mice of 30 months (36, 37). In comparison, severe degeneration of the intervertebral disc in humans is seen primarily in those beyond the fifth decade of life, as demonstrated via macroscopic (35, 38) and microscopic (29) analysis of the spine motion segments.

More than 30 different grading schemes have been used to assess the degenerative state of the lumbar intervertebral disc and adjacent cartilaginous end plate regions (39). For the assessment of macroscopic anatomy or histologic sections, one of the most widely used systems is the grading scale proposed by Thompson and coworkers (38), which assigns a degeneration score from grade I to grade V to the intervertebral disc, end plate, and vertebral body regions and is associated with good to excellent interobserver agreement. However, it is not well-suited to the grading of histologic sections and does not assess cellular changes apparent by histology.

Multiple grading schemes developed for evaluation of histologic sections are available (40, 41), although information on observer reliability measures via either kappa statistics or intraclass correlation coefficient measures is not always available with these schemes. The comprehensive and well-validated grading system developed by Boos and coworkers (29) was chosen for the current evaluation of mouse spines. Use of that approach has demonstrated good correlation with the scale of macroscopic degeneration grading described by Thompson et al and found utility in predicting age-related degeneration in human samples. Although that system was not specifically developed for evaluation of animal histology, the histologic features appear to be equally relevant to the murine spine with age and degeneration, as noted in this study. With our use of the grading scale proposed by Boos et al, interobserver variability values in the grader evaluations were in the “almost perfect” range and were consistent with those reported for human tissue evaluation (29). For these reasons, it was concluded that the intervertebral disc and end plate grading schemes described by Boos and coworkers were appropriate for use in the current study.

An observation of note was that summed grades across individual criteria could not be expected to correlate with progression of degeneration in all cases, and that the full range of the scheme (e.g., 0–22 for the intervertebral disc region) would not be expected for even the most severely degenerative changes in humans. Thus, the finding of relatively small grade differences, as observed for the type IX collagen mutation model in this study, should not be interpreted as “mild” degenerative changes but rather identification of specific changes in the intervertebral disc or end plate. Use of a statistical approach designed to test for differences in specific features of this grading scheme, rather than the cumulative score, may be useful in future studies.

Intervertebral disc and end plate degeneration related to a type IX collagen mutation may have important consequences in the clinical setting with respect to understanding and treating intervertebral disc degeneration and low back pain. The type IX collagen–deficient mice may be a useful model of premature and spontaneous intervertebral disc and end plate degeneration, with the benefit of age-matched and littermate controls. Furthermore, models of spontaneous degenerative disease require no surgical, mechanical, or chemical injury for initiation of intervertebral disc changes (42). This mouse model may also provide additional insight into mechanisms of intervertebral disc and end plate degeneration in humans. There are morphologic features reported in humans having single-point mutations in genes encoding α2(IX) or α3(IX) chains that coincide with those observed in the Col9a1−/− mice studied. These common features (assessed using magnetic resonance imaging in humans) include annular tears (16, 43) and end plate fracture or irregularity (16, 44).

Current findings suggest that end plate changes may be only a minor contributor to the pathology, with metabolic changes in the cell population being a major factor. In a related study of knee OA changes in this same animal model, elevated staining and expression of collagenases relevant to collagen and proteoglycan degradation were expressed in cartilage of the Col9a1−/− mice at a young age, suggesting that biochemical changes and subsequent structural changes could be precipitating factors for the progression of OA (24). Related to these changes was a later loss of compressive stiffness and an elevated level of permeability, which are consistent with functional changes that occur in OA cartilage. Whether these same mechanisms contribute to changes in the intervertebral disc matrix or end plate regions in the Col9a1−/− mouse spine are not known, although the implications for proteolytic involvement could be meaningful in designing treatment strategies for prevention of progressive intervertebral disc degeneration. Ongoing studies will explore the expression of proteolytic enzymes in the spines of mice carrying this type IX collagen gene mutation, as well as studies directed at understanding a role for the collagen deficiency in modulating nutrient and metabolism of the intervertebral disc cells.


Dr. Setton had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study design. Boyd, Jing, Li, Chen, Setton.

Acquisition of data. Boyd, Richardson, Allen, Flahiff, Jing, Chen.

Analysis and interpretation of data. Boyd, Allen, Flahiff, Li, Chen, Setton.

Manuscript preparation. Boyd, Richardson, Chen, Setton.

Statistical analysis. Boyd, Setton.


We thank Steve Johnson for his help with the colony maintenance and animal care.