Materials and Methods

Horses—Horses with osteoarthritis (n = 10) were purchased for research procedures approved by the institutional animal care and use committees. Horses were selected by use of clinical inclusion criteria, including forelimb lameness localized to the MCP joint ≥ 2/5 on the American Association of Equine Practitioners grading scale³⁰ and radiographic evidence of osteoarthritis in that joint, specifically osteophytes, entheseophytes, subchondral sclerosis, or lysis and joint space narrowing.³¹ Control horses (n = 11) were euthanatized for a nonconflicting study^a and were selected as similar-age horses without lameness and with visibly, palpably, and radiographically normal MCP joints.

Specimen collection—Immediately after euthanasia via overdose of pentobarbital (IV), the joints were opened and imaged with high-resolution digital photography (focal distance, 150 mm) to quantify and document the gross distribution of the lesions. Gross characteristics of osteoarthritis were evaluated for lesion extent, without knowledge of group assignment, by use of software,^b and assigned numeric scores for severity, including visual assessment for the presence of score lines, surface fibrillations, cartilage erosions or ulcerations, and osteophytes, on the basis of the following scoring system: 0 = normal, 1 = scattered surface fibrillations, 2 = mild partial-thickness erosions, 3 = moderate partial-thickness erosions with full-thickness fissures, and 4 = severe fullthickness cartilage ulcerations extending to subchondral bone (Figure 1).

View Full Size

Digital photographs of the metacarpal condyles of a horse with osteoarthritis (A) and an age-matched control horse (B).

Citation: American Journal of Veterinary Research 67, 8; 10.2460/ajvr.67.8.1299

• Download Figure
• Download figure as PowerPoint slide

Histologic grading—For histologic grading of severity of osteoarthritis, full-thickness cartilage specimens were cut from representative dorsal and palmar regions, processed, sectioned, and stained with H&E and toluidine blue.¹⁹ Slides were assessed by investigators (ALB, KJS, and SEW) without knowledge of group or site. Mean scores and indices were determined for histomorphologic structure, hypocellularity, and histochemical staining intensity according to descriptions adapted from the Mankin^13,32 scoring system (range, 0 [normal] to 12 [severely abnormal]; maximal sum index, 36), as follows: structure (0 = normal, 1 = surface irregularities, 2 = irregularities down to the tangential layer, 3 = cleft to 50% depth of radial zone, 4 = cleft to 100% depth of radial zone); hypocellularity (percentage of empty lacunae; 0 = 0% cell reduction [normal], 1 = 1% to 25% cell reduction, 2 = 26% to 50% cell reduction, 3 = 51% to 75% cell reduction, 4 = 76% to 100% cell reduction [severe hypocellularity]); and matrix-staining intensity (0 = normal, 1 = ≤ 25% reduction, 2 = 26% to 50% reduction, 3 =51% to 75% reduction, 4 = 76% τo 100% reduction in staining intensity).

Extraction of RNA—Immediately after specimens were harvested for histologic examination, the surface of MC3 was split frontally into dorsal and palmar halves. The entire dorsal and palmar surfaces were harvested via sharp curettage and snap-frozen separately in liquid nitrogen prior to storage at −80°C until required for RNA isolation. The RNA was extracted separately from each region with specimens individually processed to prevent inadvertent mixing of nuclear material from different regions or from different joints. Cartilage was ground under liquid nitrogen with a mortar and pestle,¹⁶ avoiding sample thawing. Each milligram of milled cartilage powder was mixed with 10 mL of reagent^c and homogenized with a rotor-stator tissue homogenizer for 1 minute followed by incubation for 5 minutes at 4°C.³³ Samples were centrifuged at 12,000 X g for 5 minutes at 4°C, and the clear supernatant was transferred to a new tube. Chloroform (0.2 mL) was added to each 1-mL aliquot of sample, and samples were shaken vigorously for 15 seconds. After samples were held for 3 minutes at 4°C, samples were centrifuged for 15 minutes at 4°C for phase separation. The colorless supernatant was transferred to a new tube, and isopropanol (0.5 mL) was added prior to incubation for 10 minutes at room temperature (approx 20°C). Samples were centrifuged for 10 minutes, and the RNA pellet was washed with 75% ethanol, dried, and resuspended in 50 μL of RNase-free water. The amount of nucleotide extracted was calculated by measuring UV absorbance at 260 to 280 nm. The RNA was assessed for quantity and integrity by use of capillary electrophoresis^d to measure fluorescence of polynucleotides. The degree of fluorescence provided information on DNA or salt contamination sustained during extraction, as indicated by signal intensities of 28S and 18S rRNA. After extraction and preparation, 14 RNA specimens from the osteoarthritis group (dorsal, n = 6; palmar, 8) and 18 RNA specimens from the control group (dorsal, 10; palmar, 8) were suitable for placement on individual equine microarrays (32 total equine microarrays).

Preparation of RNA—The method used for RNA preparation has been detailed in the literature.^27,33 The RNA extracted from each specimen was processed individually, with 1 specimen hybridized with 1 gene chip. Total RNA (5 μg) was reverse transcribed into double-stranded cDNA by use of a polymerase.^e Biotinylated cRNA was synthesized by use of a polymerase labeling kit^e and fragmented prior to overnight hybridization at 45°C for 16 hours with the equine microarray gene chip.^f Microarrays were washed and stained with streptavidin-phycoerythrin in accordance with an established protocol.^g The equine microarray gene chip^f has been validated for performance and gene expression by use of reverse transcriptase-PCR for select genes.^34,35 Light was emitted from the fluorescent reporter group only when bound to the probe. Light emitted from the perfect match oligoprobe, compared with the single base pair mismatched oligoprobe, was detected and analyzed by use of bioinformatics software.^h

Statistical analysis—Quantitative data were analyzed via 2-factor ANOVA for disease (osteoarthritis and control) and region (dorsal and palmar). Semiquantitative data were analyzed by use of a Mann-Whitney U rank test with Bonferroni correction. Microarray probe-level data were analyzed with software.^36,i Array normalization was performed with the invariant set procedure, and model-based expression indices were computed with the perfect-matchonly model. Probe-set level data that were classified as array outliers were omitted and considered to be missing data in subsequent analyses. After model-based expression indices computation and log transformation of the values, data were imported into commercially available software^j for statistical comparisons. Only probe sets that had a notable amount of variation in expression among specimens were considered for further analysis, and probe sets classified as absent for > 75% of the specimens were omitted. Specimens were clustered by use of hierarchal clustering, with mean linkage and 1 – Pearson correlation as the distance measurement. Results of statistical tests were considered significant at values of P ≤ 0.005 to compensate for multiple comparisons. Genes were annotated for name and function by computer programs.^k,l

Results

Macroscopic lesions—The median gross degeneration score for articular cartilage was significantly greater in the palmar region of osteoarthritic joints, compared with the dorsal region of osteoarthritic joints and also compared with the corresponding regions of control joints. Osteoarthritic specimens were all of grades 2 to 4, and control specimens were of grade 0 or 1. Clinically normal joints all had grossly normal cartilage (grade 0) in the dorsal region. The articular cartilage obtained from the palmar region of clinically normal joints and the dorsal region of osteoarthritis-affected joints had mild evidence of gross degeneration with surface fraying and score lines. Cartilage harvested from the palmar aspects of the diseased joints was severely degenerated with deep fissures, erosions, and ulcerations down to subchondral bone. Median gross score for articular cartilage was different (P = 0.013) between region (dorsal vs palmar) and between group (osteoarthritis vs normal).

Histologic lesions—Histologic lesions were significantly greater in both regions of osteoarthritic joints, compared with corresponding regions in control joints. Cartilage from the palmar aspect of diseased joints consistently had the highest histologic scores (median index score, 28; range 11 to 31) with significantly (P = 0.013) greater indices than dorsal osteoarthritic sites or either region in control joints. In contrast, specimens harvested from the dorsal region of affected joints were characterized by mild to moderate degeneration, with obvious changes evident but of lower magnitude than the palmar region (median index score, 9; range, 7 to 16). The histologic index scores of cartilage from control joints were low (median index score, 0; range, 0 to 2), and there was mild degeneration in palmar regions (median index score, 3; range, 0 to 18).

Microarray analysis—Microarray quality characteristics, including signal intensity and percentage positive signal calls, were not different between regions or groups, indicating consistency of gene chip performance and tissue expression. Each array met internal standard requirements analyzed by the bioinformatic core before being included in further analyses.

Gene expression profiles clustered significantly by disease (osteoarthritis or control) and severity (mild or severe). One of the 2 main hierarchal clusters consisted almost entirely of control specimens (Figure 2). A different set of genes for dorsal osteoarthritis (mild) and palmar osteoarthritis (more severe) was identified in association with greater and lesser gene expression (Tables 1 and 2).

View Full Size

Dendrogram for microarray clustering analysis of RNA from metacarpal condyle articular cartilage of 14 specimens from horses with osteoarthritis and 18 specimens from control horses. N = Control. RP = Right palmar. OA = Osteoarthritis. LD = Left dorsal. LP = Left palmar. RD = Right dorsal.

Citation: American Journal of Veterinary Research 67, 8; 10.2460/ajvr.67.8.1299

• Download Figure
• Download figure as PowerPoint slide

Analysis revealed more evidence of anabolic pathways and anti-apoptotic mechanisms in dorsal specimens affected with osteoarthritis. Articular cartilage obtained from the dorsal region had > 2 times the expression of genes central to cartilage maintenance, compared with cartilage obtained from corresponding regions of control joints. These included genes encoding the macromolecules of the extracellular matrix (collagen types I and II), as well as significantly greater SOD-2 and genes coding for degradative enzymes, catabolic proteins, and transcriptional regulation (Table 1). Articular cartilage from specimens with palmar osteoarthritis predominantly yielded lesser expression, including that of genes involved in protein biosynthesis, anti-apoptotic function, and extracellular proteoglycan matrix formation. Expression of genes encoding collagen I and II was ≥ 2 times that in cartilage harvested from dorsal and palmar regions of osteoarthritic joints, compared with the equivalent regions in control specimens (Table 2).

Discussion

To the authors' knowledge, the present study was the first to quantitatively characterize metacarpal osteoarthritis by region via comparative, gross, histologic, and gene expression profiling. By use of macroscopic and histologic grading systems, it was established that specimens from clinically normal adult middle-aged horses had unaffected cartilage in the dorsal aspect of the joint and mild degenerative changes in certain palmar regions. In contrast, a marked difference in regional distribution of cartilage lesions was evident in condyles with osteoarthritis. Degenerative changes were most marked in the palmar region of affected condyles, consistent with severe osteoarthritis. Cartilage lesions in the dorsal region were evident but of a markedly lesser magnitude and were characteristic of mild osteoarthritis.

Radiographic assessment, in conjunction with clinical and surgical findings, has suggested involvement of MC3, specifically, the palmar condyle region, but this has remained minimally studied. In contrast, histologic grading systems^13,32 have been used extensively in humans and veterinary species to classify the severity of disease in affected cartilage,³⁷ but are limited to the microscopic areas retrieved. This disadvantage is particularly important in diseased specimens that are highly heterogeneous or have regional variability in lesions, as in the MCP joint.³⁸ This focal nature of osteoarthritis is thought to suggest a role for biomechanical loading, superimposed on age-related changes, in the distribution of lesions throughout the joint. The injury progressively results in morphologic changes and matrix degradation, not only in the directly injured area but also in the surrounding cartilage.^7,14 Furthermore, the biomechanical and biochemical properties of articular cartilage adjacent to the lesion are substantially weaker than that of grossly normal cartilage from remote, unaffected parts of the joint.^29,38,39 Previous studies^38,40,41 of affected MCP joints have focused on the biochemical degeneration of the articular surface of the first phalanx. A distribution pattern of site-related changes in the biochemical composition of the cartilage of the first phalanx has been identified,^40–42 with the medial side more frequently affected than the lateral side.⁴³ It has been postulated that the dissimilarity in response between dorsal and palmar regions may be related to differences in the transmission and dissipation of force experienced with sudden loading during exercise.⁴³ The medial to lateral distribution pattern is thought to be caused by asymmetrical loading of the articular surface caused by the eccentric position of the center of gravity of the horse in relation to the axis of the limb.

The effect of the location of diseased cartilage has been evaluated in experimental animal models other than those involving horses by use of traditional methods to investigate location-specific changes in the expression of a limited number of genes and proteins pertaining to cartilage metabolism.^11,14 In the present study, specimens from the mildly affected dorsal region and severely affected palmar region had significantly different gene expression profiles, compared with cartilage from equivalent regions in normal joints. Articular cartilage from the dorsal region was predominantly characterized by greater expression of genes for extracellular matrix macromolecules. In addition, there was significant expression of the gene encoding SOD-2, which has an important role in reduction of apoptosis caused by several classes of oxidative stress inducers.⁴⁴ Expression of the SOD-2 gene indicates the chondrocyte's attempts to maximize structural viability by reducing programmed cell death and may imply a survival response by the chondrocyte, early in the stage of disease, as described in other species.¹⁴ Other genes of greater expression encoded for proteolytic enzymes (cathepsin L and L2, cathepsin S, and MMP 3) and proteins involved in anabolic processes (squalene synthase, integral membrane protein 2C, and cartilage link protein 1). Our results were consistent with the findings of other investigators and indicated high activity of anabolic pathways in mild or moderate disease simultaneous with high activity of pathways for matrix degradation. In a developmental osteoarthritis cartilage explant model,¹⁸ chondrocytes have high expression of genes for type II collagen; aggrecan; biglycan MMP-1, -3, and -13; and TIMP-1. The present study of specimens with mild osteoarthritis supported findings of other investigators that early degeneration is accompanied by greater expression of collagenase activity¹³ and expression of genes associated with matrix macromolecules.²⁵ Furthermore, it was possible to verify that in early osteoarthritis, among all the proteases, MMP 3 had the strongest expression, which is in accordance with previous work.¹⁷ Matrix metalloproteinease 13 is detectable only in late-stage specimens, suggesting that early stages are characterized more by degradation of other matrix components than by degradation of type II collagen fibers.²⁴ The results presented here suggest that mild osteoarthritis, as is typical of the dorsal region of diseased joints, involves proteolysis in conjunction with enhanced transcriptional activity of chondrocytes as an attempt to maintain the integrity of the matrix.

In contrast to the profile of cartilage affected with mild osteoarthritis, the expression profile of severely affected cartilage had less transcriptional activity, with most differences in gene expression representing lesser gene expression. Results indicated significantly less expression of genes encoding transcriptional proteins, including H3 histone and translation initiation factor 3. Several ribosomal proteins also had lower expression. Lower expression of a biosynthetic matrix enzyme (heparin sulfate 3-O-sulfotransferase) was consistent with the overall lower extracellular proteoglycan matrix formation detected histochemically. Lower expression of genes with anti-apoptotic function was detected (B-cell chronic lymphocytic leukemia) in addition to immediate early response protein 3, which have an inhibitory effect on programmed cell death by directly interacting with various proteins involved in control of apoptotic pathways,⁴⁴ and DEAD box, involved in the positive regulation of cell proliferation. Greater collagen II expression was still present, as in mild osteoarthritis of the dorsal region, which may suggest that progression into later stage osteoarthritis is initially a failure of chondrocyte anabolic activity, as well as sustained resistance to senescence, and not loss of ability to produce type II collagen. This profile represented overall lower transcription, less ability to generate extracellular glycosaminoglycan matrix protein, and less chondrocyte ability to prevent cell death. It has been proposed that in the later, more severe stages of disease, the biosynthetic machinery of the chondrocyte is unable to keep up with anabolic demands and a net depletion of extracellular matrix subsequently occurs. Our results confirmed lower expression of genes for proteins with anti-apoptotic properties, adding further evidence that cell death is a feature of the pathogenesis of osteoarthritis late in the stage of disease.

The present findings on gene expression suggest potential therapeutic targets for molecular treatments of osteoarthritis that could be beneficial. Research of molecular mechanisms involved in the progression of osteoarthritis has identified a small number of known molecular mediators associated with osteoarthritis and has contributed to the subsequent development of new protein and gene-based approaches.⁴⁵ Initial experimental studies of the effect of protein and gene therapy strategies in the treatment of osteoarthritis have yielded encouraging results.² Protein therapies such as interleukin-1 receptor antagonist^m can ameliorate disease but requires continuous injections. Gene transfer offers the capability to achieve sustained, localized presentation of bioactive proteins, or gene products, to sites of tissue damage,⁴⁶ with the emphasis predominantly on the genes whose products enhance synthesis of the cartilage matrix or inhibit its breakdown. Results of the present study suggest that anti-apotoptic genes and genes trophic to cell health may have the most impact for long-term joint health, particularly in chronic or severe osteoarthritis. Genes that enhance collagen II production may be less effective.

The contribution of the chondrocyte as a source of inflammatory signals in osteoarthritis has been unclear. Our data suggest that the chondrocyte may not be the primary source for inflammatory cytokines in the progression of osteoarthritis. The classical inflammatory proteins that have been implicated in osteoarthritis include inducible nitric oxide synthetase, tumor necrosis factor-α, cyclooxygenase-2, and interleukin-β and were not differentially expressed in the present study. Inflammatory pathways are important in naturally occurring disease.²⁸ Two of the few studies^24,25 that used microarrays to evaluate chondrocyte gene expression in naturally occurring osteoarthritis did not detect substantial changes in inflammatory mediators, supporting our findings in equine osteoarthritis. In consideration of the experimental conditions in which inflammatory mediators from chondrocytes have been identified, our results may be more indicative of the native state.

This study revealed location-specific morphologic and gene expression patterns of the MC3 condyles of horses. Results of profile analysis indicated that several biological processes, including inflammatory, anabolic, catabolic, apoptotic, and degradative pathways, contribute to the pathogenesis of osteoarthritis at different stages of disease. Mild osteoarthritis was typified by an increase in genes encoding collagen, chondrocyte anabolism, and anti-apoptotic proteins. More severe osteoarthritis was characterized by a decrease in transcription, maintenance of proteoglycan matrix macromolecules, and the ability of the chondrocyte to regulate apoptosis, but persistence of increased collagen expression. In naturally occurring osteoarthritis cartilage, chondrocyte-derived inflammatory mediators were not a featured phenotype. Further characterization of these genes and pathways will advance our understanding of the basic mechanisms responsible for the initiation and progression of osteoarthritis and may aid the development of novel therapies.