Differentiated Activities of Decorin and Biglycan in the Progression of Post-Traumatic Osteoarthritis

Abstract

Objective:

To delineate the activities of decorin and biglycan in the progression of post-traumatic osteoarthritis (PTOA).

Design:

Three-month-old inducible biglycan (Bgn^iKO) and decorin/biglycan compound (Dcn/Bgn^iKO) knockout mice were subjected to the destabilization of the medial meniscus (DMM) surgery to induce PTOA. The OA phenotype was evaluated by assessing joint structure and sulfated glycosaminoglycan (sGAG) staining via histology, surface collagen fibril nanostructure and calcium content via scanning electron microscopy, tissue modulus via atomic force microscopy-nanoindentation, as well as subchondral bone structure and meniscus ossification via micro-computed tomography. Outcomes were compared with previous findings in the inducible decorin (Dcn^iKO) knockout mice.

Results:

In the DMM model, Bgn^iKO mice developed similar degree of OA as the control (0.44 [−0.18 1.05] difference in modified Mankin score), different from the more severe OA phenotype observed in Dcn^iKO mice (1.38 [0.91 1.85] difference). Dcn/Bgn^iKO mice exhibited similar histological OA phenotype as Dcn^iKO mice (1.51 [0.97 2.04] difference versus control), including aggravated loss of sGAGs, salient surface fibrillation and formation of osteophyte. Meanwhile, Dcn/Bgn^iKO mice showed further cartilage thinning than Dcn^iKO mice, resulting in the exposure of underlying calcified tissues and aberrantly high surface modulus. Bgn^iKO and Dcn/Bgn^iKO mice developed altered subchondral trabecular bone structure in both Sham and DMM groups, while Dcn^iKO and control mice did not.

Conclusion:

In PTOA, decorin plays a more crucial role than biglycan in regulating cartilage degeneration, while biglycan is more important in regulating subchondral bone structure. The two have distinct activities and modest synergy in the pathogenesis of PTOA.

INTRODUCTION

Decorin and biglycan are the two most abundant small proteoglycans present in the extracellular matrix (ECM) of articular cartilage¹. Classified as class I small leucine-rich proteoglycan (SLRP), decorin and biglycan have highly similar structure, with ≈ 57% homology along their leucine-rich, horseshoe-shaped protein cores². Decorin has one chondroitin sulfate/dermatan sulfate glycosaminoglycan (CS/DS-GAG) chain attached to its N-terminus, while biglycan has two². Given their similarities, the two SLRPs share many activities in their interactions with other matrix molecules, growth factors and cell surface receptors³. In fibrous tissues such as tendon and cornea, both decorin and biglycan regulate the matrix collagen fibril assembly⁴ and cell signaling⁵, where their activities can be compensatory and synergistic^6,7. In cartilage, decorin is present in both the pericellular matrix (PCM) and further-removed territorial/interterritorial extracellular matrix (T/IT-ECM)⁸, while biglycan is localized in the PCM⁹. Such differentiated distributions indicate that they may have distinct activities in cartilage. To this day, although the importance of the two SLRPs in cartilage function and pathology has been well recognized¹⁰, it is unclear how decorin and biglycan individually or synergistically regulate the initiation and progression of osteoarthritis (OA)¹¹, the most prevalent musculoskeletal disease characterized by the irreversible breakdown of cartilage.

In human cartilage, both decorin and biglycan are significantly up-regulated in OA^12-14. This up-regulation is hypothesized to be chondrocytes’ compensatory attempt to attenuate cartilage degeneration¹¹. This hypothesis is supported by our recent finding that decorin functions as a “physical linker” to increase the molecular adhesion of aggrecan, thereby increasing the integrity of aggrecan networks in normal cartilage ECM⁸, and slowing the loss of fragmented aggrecan from degenerative cartilage¹⁵. This role is supported by the more severe OA phenotype developed in decorin-null (Dcn^−/−) mice¹⁶ relative to the wild-type (WT) when subjected to the destabilization of the medial meniscus (DMM) surgery¹⁵. It also explains why, despite its up-regulation in OA cartilage, decorin is not released to the synovial fluid at an increasing level, contrary to the case of biglycan¹⁷. To this day, it is unclear if biglycan also plays an active role in the progression of OA, and if the two SLRPs have synergistic activities.

This study sought to delineate the individual roles of decorin and biglycan in the progression of post-traumatic osteoarthritis (PTOA), and to elucidate if they have synergistic and/or additive activities. We focus on their activities in PTOA, because it is the most prevalent form of OA in young adults, and leads to long term detrimental influence on the quality of life¹⁸. Following our recent study on the inducible decorin knockout mice^8,15, we applied the DMM surgery to the inducible biglycan knockout and decorin/biglycan compound knockout murine models¹⁹. In these mice, we allowed for normal joint growth to maturity, and induced the knockout of each or both SLRPs at the time of DMM. The resulted phenotype thus represented the impact of loss of SLRPs during DMM-induced cartilage remodeling, with pre-deposited developmental defects of cartilage being minimized. In these models, we studied the progression of OA by assessing the morphology, sulfated GAG (sGAG) staining, modulus, fibrillar structure and surface calcium content of cartilage, as well as the structure of subchondral bone and meniscal ossicles.

METHODS

Animal models.

Decorin, biglycan and decorin/biglycan-compound inducible knockout mice (Dcn^flox/flox/Rosa26Cre^ER, Bgn^flox/0/Rosa26Cre^ER and Dcn^flox/flox/Bgn^flox/0/Rosa26Cre^ER, or Dcn^iKO, Bgn^iKO and Dcn/Bgn^iKO) in the C57BL/6 strain were generated as previously described¹⁹, and were housed in the Calhoun animal facility at Drexel University. To induce the knockout of SLRP genes, three consecutive daily intraperitoneal (i.p.) injections of tamoxifen were applied at a dosage of 3 mg/40 g body weight in the form of 20 mg/ml suspended in sesame oil (S3547, Sigma) with 1% volume/volume benzyl alcohol (305197, Sigma) at 3 months of age (first injection started at one week before DMM). Quantitative polymerase chain reaction (qPCR) was performed on femoral head cartilage at day 5 to confirm the tamoxifen-induced gene excision of Dcn and/or Bgn, and to detect any compensatory up-regulation between the two genes (Primers: Dcn, forward: 5’-TGAGCTTCAACAGCATCACC-3’, reverse: 5’-AAGTCATTTTGCCCAACTGC-3’; Bgn, forward: 5’-CTACGCCCTGGTCTTGGTAA-3’, reverse: 5’-ACTTTGCGGATACGGTTGTC-3’). Control mice include those with normal expressions of decorin and biglycan, including the inducible knockout mice of decorin, or biglycan, or compound injected with vehicle (the same amount of sesame oil and benzyl alcohol but without tamoxifen), and WT mice injected with tamoxifen at the same dose and frequency. All the mice used here were genotyped, following standard procedures¹⁹. Animal work was approved by the Institutional Animal Care and Use Committee at Drexel University.

The DMM surgery was performed on the right hind knees of skeletally mature, 3-month-old male mice, following the established procedure²⁰. Briefly, after anesthesia, the joint capsule was opened and medial meniscotibial ligament was cut to destabilize the medial meniscus. Sham surgery was performed on the contralateral left knee by opening the joint capsule in the same fashion to expose the ligament, but without further damage. Mice were euthanized at 8 weeks after surgery for further analyses (n = 11, except n = 19 for the control). In our recent study, Dcn^iKO and control mice were subjected to the same DMM model, and analyzed following the same paradigm by the same researchers (BH, QL, CW) as the current study¹⁵. These results were thus analyzed together with Bgn^iKO, Dcn/Bgn^iKO and additional control mice.

Histology and immunofluorescence imaging.

Whole hind knee joints (n = 6, except n = 10 for the control) were harvested and fixed in 4% paraformaldehyde, first used for μCT analysis, and then, decalcified in 10% EDTA for 4 weeks, and embedded in paraffin. Serial 6-μm-thick sagittal sections were prepared, and two sections with every consecutive six sections of the medial side of the Sham and DMM knees were stained with Safranin-O/Fast Green, following the established procedure¹⁵. For each joint, approximately 15 sections were obtained and scored by two blinded observers (QL and CW) using the modified Mankin method²¹. Each section was assigned a score based on the sum of cartilage structure (0-5), chondrocytes (0-3), Safranin-O staining (0-5) and tidemark (0-1). The score of each knee was taken as the maximum of all scored sections²². Thicknesses of uncalcified and total cartilage were determined by averaging six values evenly distributed across the entire cartilage in the load bearing medial region.

To validate the reduction of decorin and biglycan by tamoxifen injection at the protein level, immunofluorescence (IF) imaging was performed. Additional paraffin sections were treated with 0.1% pepsin (P7000, Sigma) for antigen retrieval, and blocked with 5% BSA in PBS for 1 hour at room temperature. Sections were first incubated with primary antibody (decorin, LF-114, biglycan, LF-159, Kerafast, 1:100 dilution) overnight at 4°C, and then, with secondary antibody (AlexaFluor 594, ThermoFisher, 1:500) for 2 hours at room temperature (n = 4). Sections were washed with PBS, counter-stained, mounted with DAPI (Fluoromount-G, 0100-20, SouthernBiotech), and imaged with a Carl Zeiss Axio Observer Microscope.

AFM-based nanoindentation

AFM-based nanoindentation was applied to freshly dissected femoral condyle cartilage (n = 5, except n = 9 for the control), using custom-made polystyrene microspherical colloidal tips (R ≈ 5 μm, nominal k ≈ 8.9 N/m, HQ:NSC35/Tipless/Cr-Au, cantilever A, NanoAndMore) and a Dimension Icon AFM (BrukerNano) at 10 μm/s indentation rate up to a maximum load of ≈ 1 μN in 1 × PBS with protease inhibitors (Pierce 88266, ThermoFisher). Following the established procedure²³, at least 10-15 locations were tested on each joint to account for spatial heterogeneity. The indentation modulus, E_ind, of each location was calculated by fitting the entire loading portion of each force-indentation depth (F-D) curve with the Hertz model, and the average value from each joint was taken as one biological repeat.

Scanning electron microscopy

Scanning electron microscopy (SEM) was applied to quantify the fibril diameter and alignment on the load-bearing region of medial condyle cartilage surfaces (n = 4), following the established procedure²⁴. Immediately after AFM-nanoindentation, condyle cartilage joints were processed for proteoglycan removal, fixed, dehydrated, air dried overnight, coated with ≈ 6 nm thick platinum-palladium mixture, and then, imaged via a Supra 50 VP SEM (Carl Zeiss). Collagen fibril diameter d_col, and alignment angle, θ, were measured using ImageJ. Values of θ were fitted with von Mises probability density function to calculate the von Mises concentration parameter, κ, a quantitative measure of fibril alignment²⁵, following the established procedure²⁶. In addition, since one major distinction between the calcified versus uncalcified cartilage layers is the presence of higher calcium content, we applied SEM with the Energy Dispersive X-ray Spectroscopy (SEM/EDS) to assess the weight percentage of calcium on cartilage surface, which is an indicator of the exposure of underlying calcified layer.

Micro-computed tomography

Micro-computed tomography (μCT) scanning was performed to assess concurrent changes of subchondral bone and meniscus ossification after DMM. For mice purposed for histology, prior to demineralization, knee joints (n = 5) were scanned ex vivo using MicroCT 35 (Scanco Medical AG) at 6 μm isotropic voxel size and smoothed by a Gaussian filter (sigma = 1.2, support = 2.0). Following the established procedure^27,28, we estimated the thickness of subchondral bone plate (SBP.Th), the bone volume fraction (BV/TV), trabecular number (Tb.N), and trabecular thickness (Tb.Th) of subchondral trabecular bone (STB) on the central load-bearing region of medial tibial plateau, as well as meniscal ossicle volume and osteophyte formation on the medial side.

Statistical analysis

Linear mixed effect model was applied to test quantitative outcomes including Mankin score, cartilage thickness, E_ind, d_col, calcium content and μCT outcomes using the R package lme4 (version 1.1-19)²⁹. In all the tests, genotype, surgery type, and position (anterior versus posterior for meniscal ossicle volume) were treated as fixed effect factors when applicable, with interaction terms between genotype and surgery, or between genotype and position. Individual animal effect was treated as a randomized factor, and surgery type and position were considered as within-subject factors. Prior to the test, Shapiro-Wilk test was applied to residuals to confirm that these outcomes did not deviate significantly from normal distribution, and likelihood ratio test was applied to determine the choice of two covariance structures, unstructured versus compound symmetry. For all comparisons, p-values were adjusted for family-wise type I errors via Tukey-Kramer test between genotypes for each surgery type, and via Holm-Bonferroni correction for multi-contrasts between surgeries or between positions across multiple genotypes. For fibril orientation data, Mardia and Jupp test of concentration equality²⁵ was applied to compare the von Mises concentration κ between genotypes and surgery types, followed by the Holm-Bonferroni correction. All quantitative and statistical outcomes were summarized in Tables S1-S5.

RESULTS

Reduction of decorin and biglycan expressions in the induced knockout models.

In all inducible knockout models, daily injection of tamoxifen for 3 days substantially reduced the expressions of each or both SLRPs by day 5. In single knockout mice, we did not notice the up-regulation of biglycan in response to the loss of decorin, and vice versa, illustrating limited compensatory effects (Fig. 1a). In alignment with gene expression results, IF imaging also showed marked reduction in the staining of decorin in Dcn^iKO and Dcn/Bgn^iKO cartilage (Fig. 1b), and biglycan in Bgn^iKO and Dcn/Bgn^iKO groups (Fig. 1c), in both Sham and DMM knees following the tamoxifen injection. In control mice, we detected increased staining of decorin, but not biglycan, after DMM (Fig. 1b,c), consistent with our previous observations on WT mice^15,30. In comparison to the control, Dcn^iKO and Bgn^iKO mice showed similar staining patterns of biglycan and decorin, respectively. Also, in Dcn^iKO mice, biglycan remained to be localized in the PCM after DMM. These results supported limited compensatory effects, corroborating the qPCR results. Thus, we validated these murine models for studying the impact of the loss of individual or both SLRPs on DMM-induced PTOA.

Figure 1.

a) Confirmation of the induced knockout of Dcn and/or Bgn gene in Dcn^iKO , Bgn^iKO and Dcn/Bgn^iKO mice. In 3-month-old mice, intraperitoneal (i.p.) injection of 3 mg tamoxifen (TM)/40 g body weight for 3 consecutive days reduces the expression of decorin (Dcn) and/or biglycan (Bgn) to the baseline level by day 5, as measured from femoral head cartilage (mean ± 95% CI, n = 6 biological repeats, different letters indicate significant differences between genotypes, p < 0.001). Injection of vehicle does not alter the level of Dcn or Bgn expression relative to the wild-type (WT). In single inducible knockout mice, loss of Dcn does not alter the expression of Bgn, and vice versa. b,c) Immunofluorescence (IF) images show the reduction of b) decorin and c) biglycan protein content following their respective induced knockout at 8 weeks after DMM and Sham surgeries (inset: negative control, blue: DAPI, n = 4).

Impacts of decorin and biglycan on the degradation of cartilage after DMM.

By 8 weeks after DMM, all four genotypes exhibited salient histological signs of OA, including reduced sGAG staining, surface fissures, increased chondrocyte hypertrophy and cartilage thinning (reduced t_uncalcified and t_total), contributing to increased modified Mankin scores (Fig. 2). Amongst the genotypes, Bgn^iKO (6.22 [5.68 6.76] Mankin score, mean [95% CI]) did not show appreciable differences relative to the control post-DMM (5.78 [5.35 6.21], 0.44 [−0.18 1.05] difference, p = 0.273) (Fig. 2a,b). This was distinct from the more severe OA observed in Dcn^iKO mice (7.17 [6.89 7.44], 1.38 [0.91 1.85] difference versus control). Thus, loss of biglycan had a lesser impact on DMM-induced cartilage degeneration than the loss of decorin.

Figure 2.

a) Representative histological images of Safranin O-Fast Green-stained cartilage specimens from control, Dcn^iKO, Bgn^iKO and Dcn/Bgn^iKO mice at 8 weeks after Sham and DMM surgeries, with more severe cartilage damage observed in the Dcn^iKO and Dcn/Bgn^iKO mice. b) Modified Mankin score and, c) thickness of uncalcified cartilage, t_uncalcified, and d) thickness of total cartilage, t_total, in the medial femur at 8 weeks after Sham and DMM surgeries (mean ± 95% CI, n = 6, except n = 10 for the control). Each data point represents the value from one animal, different letters indicate significant differences between genotypes for each surgery type, ^#: p < 0.05 between Sham and DMM surgeries for each genotype. Data for Dcn^iKO mice are adapted from Ref. ¹⁵ with permission.

In Dcn/Bgn^iKO mice, the Mankin score (7.29 [6.87 7.71]) was higher than that of control (1.51 [0.97 2.04] difference) and Bgn^iKO (1.07 [0.47 1.67] difference) mice, but similar to that of Dcn^iKO mice (0.13 [−0.32 0.57] difference, Fig. 2b). On the other hand, Dcn/Bgn^iKO mice developed lower uncalcified cartilage thickness, t_uncalcified (16.4 [14.0 18.9] μm), compared to Dcn^iKO mice (24.4 [21. 6 27.2] μm, 8.0 [4.7 11.2] μm difference, Fig. 2c). This further reduction, however, did not lead to higher Mankin scores, as cartilage erosion has not extended into the calcified cartilage layer in both genotypes, yielding similar scores on cartilage structure. Thus, upon the loss of decorin, the concomitant loss of biglycan did not markedly aggravate the progression of OA based on histological analysis. In the Sham group, unlike the single knockout mice, Dcn/Bgn^iKO mice showed mild Mankin scores (2.17 [1.62 2.71]) and reduced cartilage thickness (Fig. 2b-d), indicating a modest impact of concomitant loss of decorin and biglycan on cartilage post-natal maintenance.

Impacts of decorin and biglycan on cartilage surface fibrillar structure and modulus after DMM.

In healthy joints, cartilage surface is characterized by transversely random mesh of collagen fibrils⁸, as present in all Sham groups and the DMM groups of control and Bgn^iKO mice (Fig. 3a). In contrast, both Dcn^iKO and Dcn/Bgn^iKO cartilage surfaces developed highly aligned collagen fibrils along the mediolateral orientation, that is, the direction subjected to extensive shearing by the destabilized medial meniscus. These changes were signified by the higher von Mises concentration, κ, indicating salient surface fibrillation (Fig. 4a). Meanwhile, we did not notice significant changes in average surface fibril diameter, d_col, amongst the four genotypes, or between Sham and DMM groups (Fig. 4b).

Figure 3.

a) Representative scanning electron microscopy (SEM) images showing the nanostructure of collagen fibrils on condyle cartilage surfaces at 8 weeks after Sham and DMM surgeries. Red arrows denote the mediolateral orientation. b) Comparison of the fibril orientation distributions for all four genotypes after Sham and DMM surgeries. Results are from ≥ 300 fibrils pooled from n = 4 animals per group following setting the average angle to 0° for each joint. Data for control and Dcn^iKO mice in panel a) are adapted from Ref.15 with permission.

Figure 4.

a) Degree of fibril alignment, as denoted by the von Mises concentration parameter κ, and b) Distribution of collagen fibril diameter, d_col, for all four genotypes at 8 weeks after Sham and DMM surgeries. Results are from ≥ 300 fibrils pooled from n = 4 animals per group. Different letters indicate significant differences between genotypes for each surgery type, ^#: p < 0.05 between Sham and DMM surgeries for each genotype. Data for control and Dcn^iKO mice in panel a) are adapted from Ref. ¹⁵ with permission.

The EDS analysis did not detect appreciable calcium content in all Sham groups (data not shown). In DMM groups, low calcium content was detected on control, Dcn^iKO and Bgn^iKO cartilage surfaces (≤ 0.5% wt.), as expected for uncalcified cartilage. The surface of Dcn/Bgn^iKO cartilage, however, showed significantly higher calcium amount (1.08 [0.59 1.56]% wt., n = 6, Fig. 5a). This concentration was much lower than the calcium content in bone (≈ 26.6% wt. based on 67% dry weight of hydroxyapatite in bone³¹), affirming the histological finding that cartilage has not undergone full erosion by 8 weeks. This observation suggested that the underlying calcified cartilage has started to be partially exposed, which was not yet detectable by histology, but apparent at the nanoscale.

Figure 5.

a) Weight percentage of calcium on the surfaces of condyle cartilage subjected to the DMM surgery, as measured by SEM Energy Dispersive X-ray Spectroscopy (EDS) analysis (n = 6, except n = 5 for Bgn^iKO mice). Each data point represents the averaged value measured from one animal. b) Effective indentation modulus, E_ind, of medial condyle cartilage surface, as measured by AFM-nanoindentation (n = 5, except n = 9 for the control). Each data point represents the average value of ≥10 locations measured from one joint. Panels b-d: mean ± 95% CI, different letters indicate significant differences between genotypes for each surgery type, ^#: p < 0.05 between Sham and DMM surgeries for each genotype. Data for Dcn^iKO mice in panel b) are adapted from Ref. ¹⁵ with permission.

In Dcn^iKO , Bgn^iKO and control mice, the DMM group showed significantly lower modulus, E_ind, than the Sham group (Fig. 5b), illustrating impaired cartilage load-bearing function in OA. In Dcn/Bgn^iKO mice, however, we observed much higher modulus than other genotypes after DMM. This finding can be explained by the partial exposure of the stiffer calcified layer, as evidenced by the higher calcium content (Fig. 5a). This aberrantly high modulus does not represent the restoration of cartilage biomechanical function, but rather, is an indicator of salient cartilage erosion, and thus, represents more severe cartilage damage²³. Taken together, despite not having higher Mankin scores than Dcn^iKO mice after DMM (Fig. 2b), Dcn/Bgn^iKO mice exhibited more severe cartilage damage, as signified by lower t_uncalcified (Fig. 2c), higher calcium content (Fig. 5a) and aberrantly high surface modulus (Fig. 5b).

Altered subchondral bone structure with the loss of biglycan.

For the control mice, by 8 weeks after DMM, we did not detect significant changes in subchondral bone plate (SBP) or subchondral trabecular bone (STB) structure, except for a mild increase in STB Tb.N (Fig. 6a-e). This is in agreement with literature showing that in the DMM model, subchondral bone changes only occur after full erosion of cartilage in late OA³². Comparing the four genotypes, we did not notice significant differences between the control and Dcn^iKO mice for both surgery groups (Fig. 6a-e). Bgn^iKO and Dcn/Bgn mice, however, developed significantly altered STB structure relative to the control and Dcn^iKO mice, marked by decreased BV/TV, Tb.N and Tb.Th, for both surgery groups. In contrast, we detected the formation of osteophytes, another sign of more advanced OA³³, in Dcn^iKO and Dcn/Bgn^iKO joints, but not in control or Bgn^iKO joints (Fig. 7a). At both the anterior and posterior horns of the meniscus, all four genotypes showed increased ossification after DMM, and this increased ossification was similar amongst all genotypes (Fig. 7b,c). Collectively, findings from μCT suggested that biglycan has a more direct impact on subchondral bone remodeling, while decorin has a more important role in OA.

Figure 6.

a) Representative 2D μCT frontal plane images of the knee joint at 8 weeks after Sham and DMM surgeries (L: lateral, M: medial). b) Subchondral bone plate thickness (SBP.Th) and c-e) Subchondral trabecular bone structural parameters, including c) BV/TV: bone volume fraction, d) Tb.N: trabecular number, and e) Tb.Th: trabecular thickness, as measured from the μCT images of medial tibia. Panels b-e: mean ± 95% CI, n = 5, each data point represents the averaged value measured from one animal. Different letters indicate significant differences between genotypes for each surgery type, ^#: p < 0.05 between Sham and DMM surgeries for each genotype. Data for control and Dcn^iKO mice are adapted from Ref. ¹⁵ with permission.

Figure 7.

a) Representative reconstructed 3D μCT images showing the presence of osteophyte in both Dcn^iKO and Dcn/Bgn^iKO joints at 8 weeks after DMM (white arrowheads), but not in other groups. b) Representative reconstructed 3D μCT images (top view) of meniscal ossicles showing increased ossification after DMM for all genotypes. c) Meniscal ossicle volume at both anterior and posterior ends at 8 weeks after Sham and DMM surgeries. Panel c: mean ± 95% CI, n = 5, each data point represents the averaged value measured from one animal. Different letters indicate significant differences between genotypes for each surgery type, ^#: p < 0.05 between Sham and DMM surgeries for each genotype. Data for control and Dcn^iKO mice are adapted from Ref. ¹⁵ with permission.

DISCUSSION

Differentiated activities of decorin and biglycan in DMM-induced PTOA.

This study highlights the differentiated activities of decorin and biglycan in the progression of DMM-induced PTOA. In degenerative cartilage, decorin functions as a “physical linker” to increase the retention of fragmented aggrecan, thereby slowing aggrecan loss¹⁵. In the DMM model, loss of decorin leads to accelerated sGAG depletion, cartilage fibrillation, and thus, more severe OA¹⁵, while loss of biglycan does not have a marked impact on cartilage degradation or OA progression (Fig. 2). Such contrast can be attributed to differences in their distribution, binding activities, as well as response to DMM and inflammation. In healthy cartilage, biglycan shows much higher binding affinity than decorin to PCM-specific molecules, such as collagen VI and matrilins^34,35, which may contribute to its localization in the PCM. Thus, unlike decorin, biglycan possibly primarily regulates the integrity of PCM, not the ECM as a whole. Also, different from decorin, biglycan does not undergo salient changes in its concentration or distribution after DMM (Fig. 1b)³⁰. This corroborates observations in bovine³⁶ and murine¹⁵ cartilage explant models, in which, stimulation by inflammatory cytokine IL-1β increases the expression of decorin, but not biglycan. Thus, although biglycan may be crucial to the integrity of cartilage PCM, and may mediate the canonical Wnt signaling of chondrocytes³⁷, its activities are not markedly altered or stimulated by DMM, indicating that biglycan may not be an essential player in OA pathogenesis in the DMM PTOA model.

In both Sham and DMM groups, loss of biglycan alters the structure of subchondral trabecular bone, while loss of decorin does not (Fig. 6). This is consistent with literature showing that biglycan plays a more crucial role than decorin in regulating bone homeostasis and remodeling^38-41. Despite these changes in subchondral bone, Bgn^iKO mice do not show altered cartilage degradation (Figs. 2, 3), suggesting limited cross-talk between subchondral bone remodeling and cartilage degradation in DMM-induced OA. Furthermore, the observation that the Sham and DMM knees of Bgn^iKO mice have similar subchondral bone structure (Fig. 6) suggests that the regulation of subchondral bone by biglycan does not directly influence the pathogenesis of DMM-induced OA, and vice versa.

Our results do not rule out a potential critical role of biglycan in more advanced PTOA or other forms of OA. In late stage human OA, biglycan is up-regulated, undergoes substantial fragmentation and has an increased presence in the T/IT-ECM^13,14. Also, in late OA, unlike the case of decorin, an increasing amount of soluble, fragmented biglycan is released to the synovial fluid, which may accelerate sGAG loss through elevating NF-κB activities¹⁷. This potential adverse effect of biglycan is not observed here, as by 8 weeks after DMM, the PCM has not yet lost its distinction to the bulk ECM³⁰, and biglycan remains to be sequestered within the PCM (Fig. 1b). It is possible that, at a more advanced stage, when biglycan fragments are released from the damaged PCM, biglycan could become an important player in OA pathogenesis. In addition, both decorin and biglycan could have different activities in other forms of OA. For example, Dcn^−/− mice demonstrated higher resistance to forced running-induced OA⁴², which can be attributed to different OA etiology in the over-exercised OA model^8,15. Also, Bgn^−/− mice develop accelerated OA during aging⁴³, illustrating a potential role of biglycan in overall cartilage health and spontaneous OA.

Modest synergy between decorin and biglycan in DMM-induced PTOA.

In the DMM model, Dcn/Bgn^iKO mice develop similar Mankin score (Fig. 2b), surface fibrillation (Figs. 3b and 4a) and osteophyte formation (Fig. 7a) as Dcn^iKO mice, but show accelerated cartilage thinning (Fig. 2c, d), higher surface calcium content (Fig. 5a) and aberrantly higher surface modulus (Fig. 5b). These results illustrate a moderately higher degree of cartilage damage in Dcn/Bgn^iKO mice. This modest synergy between the two SLRPs could be due to their coordinated impacts on the PCM integrity. In cartilage, the PCM plays a crucial role in modulating cell-ECM interactions and chondrocyte mechanotransduction⁴⁴. Degradation of the PCM, and associated alteration of chondrocyte mechanotransduction are among the earliest events that precede the initiation of PTOA³⁰. Loss of decorin is expected to aggravate the degradation of PCM by accelerating the depletion of fragmented aggrecan¹⁵, which may accelerate the disruption of chondrocyte mechanotransduction⁴⁵. Concomitant loss of biglycan, a crucial PCM constituent, could further accelerate the PCM disruption, and thus, impair chondrocyte mechanotransduction and exacerbate cartilage degradation. Meanwhile, Dcn/Bgn^iKO mice develop similar subchondral bone changes as Bgn^iKO mice (Fig. 6b-e), illustrating limited synergy of the two SLRPs in subchondral bone remodeling. Our future studies will thus focus on the activities of the two SLRPs in advanced OA, in which, they may show stronger synergy in regulating chondrocyte mechanobiology and cartilage degradation.

On the technical front, our study highlights the importance of assessing cartilage degradation beyond the scope of standard histological analysis. By 8 weeks after DMM, mice develop moderate OA, and the four genotypes only exhibit modest differences in Mankin scores (Fig. 2b). However, results from SEM, AFM and μCT analyses show clear signs of more advanced OA in Dcn^iKO and Dcn/Bgn^iKO mice, including surface fibrillation, exposure of calcified cartilage and formation of osteophytes. Therefore, integrating histology with more focused structural and biomechanical tools can provide a more sensitive and in-depth assessment of OA etiology and cartilage damage.

Comparison of decorin and biglycan activities in other connective tissues.

The limited compensation and synergy of decorin and biglycan in cartilage is in stark contrast to their activities in fibrous tissues such as tendon and cornea, in which, they share similar roles of regulating the assembly of collagen fibrils⁴. In cornea, biglycan is up-regulated in the deficiency of decorin, but not vice versa, whereas their compound loss leads to more severe defects in collagen fibril nanostructure in Dcn^−/−/Bgn^−/− mice⁶. A similar compensatory pattern is observed in the flexor digitorum longus (FDL) tendon of juvenile Dcn^−/− mice at 1 month of age⁷, but not in the patellar tendon of old Dcn^iKO mice at 16 months of age⁴⁶. In aged tendon, the compound induced knockout of decorin and biglycan also does not result in more severe biomechanical changes than single knockouts.⁴⁶

In the knee joint, biglycan has shown synergistic activities with other SLRPs, including fibromodulin, a class II SLRP, and epiphycan, a class III SLRP. Both Bgn^−/−/Fmod^−/−47 and Bgn^−/−/Epn^−/−43 mice develop more severe spontaneous OA than the respective single knockouts. Given that biglycan has the highest structural homology with decorin², our ongoing work aims to query the potential synergy of the two in more advanced PTOA and aging-associated spontaneous OA. In particular, aging of cartilage is associated with reduced chondrocyte autophagy⁴⁸. Decorin and biglycan can both evoke autophagy in other cell types^49,50, we will study their roles in regulating the autophagy of chondrocytes.

CONCLUSIONS

In summary, decorin and biglycan have differentiated activities in the progression of early-to-intermediate PTOA. Unlike decorin, biglycan does not play a major role in regulating cartilage degradation in DMM-induced PTOA, and the compound loss of both SLRPs shows modest synergistic impacts. While decorin is more crucial in regulating cartilage integrity, biglycan has a stronger impact on subchondral bone structure. These observations are distinct from the highly coordinated activities of the two SLRPs in fibrous tissues. Therefore, decorin, rather than biglycan, could serve as a potential target for developing effective intervention strategies for attenuating the progression of PTOA.