Structural changes in the collagen network of joint tissues in late stages of murine OA

Osteoarthritis (OA) is the most prevalent degenerative joint disease, resulting in joint pain, impaired movement, and structural changes. As the ability of joint tissue to resist stress is mainly imparted by fibrillar collagens in the extracellular matrix, changes in the composition and structure of collagen fibers contribute to the pathological remodeling observed in OA joints that includes cartilage degeneration, subchondral bone (SCB) sclerosis, and meniscal damage. Using the established OA model of destabilization of the medial meniscus (DMM) in C57BL/6J mice, we performed a comprehensive analysis of the content and structure of collagen fibers in the articular cartilage, subchondral bone, and menisci using complementary techniques, which included second harmonic generation microscopy and immunofluorescence staining. We found that regions exposed to increased mechanical stress in OA mice, typically closest to the site of injury, had increased collagen fiber thickness, dysregulated fiber formation, and tissue specific changes in collagen I and II (Col I and Col II) expression. In cartilage, OA was associated with decreased Col II expression in all regions, and increased Col I expression in the anterior and posterior regions. Col I fiber thickness was increased in all regions with disorganization in the center region. In the superficial SCB, all regions exhibited increased Col I expression and fiber thickness in OA mice; no changes were detected in the deeper regions of the subchondral bone except for increased Col I fiber thickness. In the menisci, OA led to increased Col I and Col II expression in the vascular and avascular regions of the anterior meniscus with increased Col I fiber thickness in these regions. Similar changes were observed only in the vascular region of the posterior meniscus. Our findings provide, for the first time, comprehensive insights into the microarchitectural changes of extracellular matrix in OA and serve as guidelines for studies investigating therapies that target collagenous changes as means to impede the progression of osteoarthritis.


Results
Microarchitectural changes are evident in tissues of the medial compartment of late-stage OA knee joints. 12 weeks following DMM, mice knees showed loss of articular cartilage, decreased bone marrow space (arrows, Fig. 1), and increased sclerosis in the subchondral bone ( Fig. 1A-C), as we have previously showed 1 . For a preliminary evaluation of the collagen microarchitecture, we stained knee joints of both sham and DMM-operated mice with Picrosirius red (PSR) 2,3 , in which samples were imaged with a polarized light microscope to selectively image collagen fibers (Fig. 1D). Red birefringence indicates thicker fibers, while green birefringence indicates thinner fibers 3,25 . In the cartilage, DMM mice had reduced uncalcified cartilage area (Fig. 1C) with a trend towards reduced area and intensity of the green birefringence (Fig. S1A). This green birefringence originates from the radial collagen network, which is composed of thin fibers mainly formed by Col II, cross-linked with minor fibrillar collagens like Col III, V, and XI 26,27 . In the subchondral bone, DMM mice showed decreased marrow space (blue arrows, Fig. 1D), with increase in both red and green birefringence area and intensity (Fig. 1E). In both anterior and posterior menisci, DMM mice had a significant increase in red birefringence area and intensity with only a trend towards increase in green birefringence area and intensity (Fig. 1F,G). These results indicated reduction in uncalcified tibial cartilage and increased collagen fiber deposition in the subchondral bone and both the anterior and posterior menisci (Fig. S1A, Fig. 1E-G). We used these results as a preliminary guide to confirm that OA progression affects the collagenous structure of not only cartilage, but also SCB and menisci. Although PSR is useful for quick analysis of tissue remodeling, PSR cannot be used to distinguish different collagen types 25 . Thus, we performed the regional analysis of the medial knee-joint compartment with IF staining and SHG imaging for more comprehensive specific analysis of fibrillar collagen changes.
OA results in substantial changes in the microstructure of collagen fibers in the articular cartilage. First, to analyze how the expression of Col I and Col II changes in OA tibial cartilage, we performed immunofluorescence (IF) staining of Col I and Col II in sham and DMM mice and analyzed the area % and average intensity of the expression of both proteins in the anterior, center, and posterior regions of the cartilage ( Fig. 2A-D). Because the superficial and transitional zones are degraded in late-stage OA, we imaged the remaining radial zone in the uncalcified cartilage. Results indicated that Col I expression area was very low in the sham cartilage, but significantly increased in the anterior and posterior regions of the cartilage of DMM mice (Fig. 2C), with increased average intensity in the anterior region (Fig. S1B). This increase in Col I expression in DMM mice was accompanied by a significant reduction in Col II expression area in all analyzed regions (Fig. 2D) and unchanged average intensity (Fig. S1B). This reduction in Col II expression is consistent with cartilage degeneration (Fig. 1A-C). Supporting these results, SHG analysis of FSHG signal, which is strongly produced by Col I fibers, and BSHG signal, which is produced by both Col I and Col II fibers, demonstrated a significant increase in FSHG/BSHG volume fraction ratio in the anterior and posterior regions of the tibial cartilage (Fig. 2E). We then analyzed the microstructure of collagen fibers. To avoid confounding the data by analyzing the BSHG signal that is generated by both Col I and Col II fibers which generally colocalize in OA cartilage, we focused our analysis on the FSHG signal (i.e. Col I fibers). Results indicated significant thickening of FSHG-producing fibers in all regions of DMM cartilage (Fig. 2F). Further, to characterize how OA pathology affects the arrangement Col I fibers, we computed the orientation index in all regions of the articular cartilage. The central region exhibited less organized FSHG-producing fibers in DMM samples as compared to sham samples (Fig. S1C). Interestingly, in sham mice, collagen fiber arrangement was significantly more oriented in the central region compared to the anterior region (Fig. S1D), whereas no significant difference in fiber orientation was noted among different regions of DMM cartilage (Fig. S1E).  (Fig. S2A,B). FSHG-producing fibers (Col I) were also consistently thicker than normal in all regions of both the superficial and deep zones (Figs. 3F, 4F). No change in the orientation of FSHG-producing fibers was observed in any of the analyzed regions of the superficial or deep SCB (Fig. S2C,D). Interestingly, Col II expression area increased in the central region of the superficial SCB (Fig. 3C,E), with no change in any of the deep SCB regions (Fig. 4C,E). Col II average intensity remained unchanged throughout all regions in both superficial and deep SCB (Fig. S2A,B). FSHG/BSHG VF ratio was not calculated for a measure of expression of Col I relative to Col II in the SCB, as IF staining showed that their expression increased simultaneously.
OA is accompanied by changes in collagen composition and fiber thickness in the menisci. Increased vascularization of the menisci is known to occur after knee injury, resulting in calcification and chondrocyte hypertrophy 28,29 . Therefore, we assessed changes in collagen composition in the vascular (outer 10-30% of the meniscus) and avascular (inner 70-90% of the meniscus) regions 30

Discussion
We present a comprehensive analysis of the microstructural changes that occur in collagen fibers in late OA stages, which leads to the structural remodeling of the knee joint. Previous studies of the collagenous network in human knee joints have found that OA progression results in collagenous disorganization of the posterior meniscal horn 21 and misaligned, thickened fibers in the superficial and deep zones of the articular cartilage 31 , which contribute to structural changes leading to tissue degradation. Our study extends the scope of previous studies that focused on examining the initial changes that occur in the cartilage during early OA 4,16,32 , or singular joint tissues 21 . Here, we define regional differences in the medial compartment of the knee in late OA, which provides indication of how load is being re-distributed in later stages of the disease. Furthermore, as OA is known to impact not only cartilage, but also multiple joint tissues including subchondral bone and menisci, we provide a comprehensive characterization of OA-associated collagenous changes in different regions and tissues composing the knee joint. Late-stage OA is characterized by extensive cartilage degeneration and increased Col I expression that mediates mineralization of the articular cartilage 16,33 . Our comprehensive analyses of fibrillar collagen using complementary PSR staining, IF staining, and SHG techniques confirm that the uncalcified region of DMM mice exhibits reduced expression of Col II protein, loss of Col II perpendicular collagen fibers in the radial zone, as well as pathologically elevated expression of Col I ( Fig. 2A-E). Regionally, Col I expression increases in the anterior and posterior, but not the central, regions of the DMM articular cartilage ( Fig. 2A,C,E). This non-uniform pattern of increased expression can be attributed to changes in the joint structure that accompany OA and altered force distribution to create more loading toward the periphery, as opposed to even force distribution across the tibial plateau in normal knees.
Previous SHG analysis of early stage OA showed that collagen I and II fibers in the radial zone of the cartilage become thinner and more organized 16 . Here, we show that as OA proceeds to late stage, Col I fibers undergo thickening in all regions of the cartilage (Fig. 2F) and become markedly disorganized in the central region (Fig. S1C). In general, the central region of intact cartilage experiences more direct contact with the femur in comparison to other regions of the tibial plateau due to the absence of meniscal fibrocartilage protection 34 , resulting in an increasingly displaced collagenous network 34 that undergoes re-orientation when exposed to stress 35,36 .   www.nature.com/scientificreports/ Therefore, the observed fiber re-orientation in this region of OA cartilage (Fig. S1C) might act, together with fiber thickening, as a compensatory mechanism for the lost support and the resultant increased mechanical loads 35,36 .
As crosstalk between the SCB and the cartilage impacts chondrocytes homeostasis and OA progression 5,37 , understanding the SCB structural changes in late-stage OA is crucial for uncovering the underlying mechanisms contributing to cartilage loss with OA progression. The SCB is separated from the calcified cartilage by the chondro-osseous junction and is composed of the SCB plate and trabecular portion 3,5,38,39 . The SCB plate resists stress and undergoes dynamic re-orientation in response to environmental stimuli and external force 3,30,31 . Accordingly, the SCB thickens in response to OA-associated cartilage degeneration as a result of increased loading, which decreases the shock-absorbing ability of the bone, noted most acutely in regions peripheral to the area closest to articulation 40,41 , which may lead to an increase in microcracks and SCB lesions 38,42 . Notably, SCB thickening co-localizes with cartilage lesions, with the least thickening detected in places where the overlying cartilage is maintained 43 , which provides an indication of the regions exposed to the highest amounts of stress after injury 43,44 . Consistent with this, our results in DMM mice illustrate increased Col I expression in all regions of the superficial (Fig. 3B,D), but not the deep (Fig. 4B,D), SCB, indicating that the superficial zone experiences higher compressive stress, resulting in more drastic changes in the collagenous content. However, both the superficial and deep zones exhibit thickening of Col I fibers (Figs. 3F, 4F) in late OA, indicating overall subchondral bone stiffening.
The meniscus absorbs shock and limits force transfer to the underlying cartilage 8 , allowing for mechanical stability and supporting up to 70% of the compressive load of the knee 45 . Therefore, meniscal injuries predispose individuals to OA due to increased loading on the cartilage 46 . The menisci are wedge-shaped in cross-section, with a thicker, outer vascular region, rich in circumferentially arranged collagen I fibers, and a thin, inner avascular region, with a high abundance of collagen II fibers arranged parallel to the surface of the meniscus 9,45,47 . Thus, the outer, vascular region, which constitutes the peripheral 10-30% of the meniscus 30,48,49 , has higher capacity for healing, in comparison to the inner, avascular portion. Pathological changes in an injured meniscus include meniscal hypertrophy and mineralization, as well as neovascularization in the vascular region 48,49 . Fibrosis of the avascular portion of the menisci is seen to correlate with tissue stiffness 50,51 , which results in decreased mechanical stability and transmission of load through the joint 45,48 . Though several studies have shown increased Col I and II expression in OA menisci, architectural changes of the collagen network which occur in the meniscal extracellular matrix in response to injury are yet to be characterized 2,18,35 . Our results in DMM mice provide insight on the meniscal regions most susceptible to stress in the medial compartment. We found contribute to meniscal mineralization and hypertrophy that typify late-stage OA 28,29,52 . This study provides a comprehensive examination of collagenous changes in the medial compartment of the knee, which undergoes the most significant degenerative changes in OA, and highlights regional differences in each joint tissue. Future studies examining the central and lateral compartments could provide further elucidation of the changes throughout the entire knee during disease progression.

Methods
Study design. This study follows the ARRIVE guidelines 2.0. Sample size was determined as previously described 1 using the G*Power 3.1 software. Based on our previous experience and studies analyzing histologic parameters affected by DMM mice in comparison to sham mice, we calculated the sample size required for each group to reach 5% significance and 0.80 power. Power analysis showed that at least 4 mice in each group are required. Thus, we increased the number of mice to 5 in the sham and DMM groups to account for any individual variances. No animals were excluded from the research group. Animals were randomized for either sham or DMM surgery. Samples collected were coded and analyzed in a blinded manner until data was obtained and quantification completed. Data were then decoded to be graphed under corresponding treatment groups, with results charted and statistical analysis performed as described. Outliers in data sets were removed using Grubb's test for exclusion. Results were confirmed by repetition 3 times.
Animals. Twelve-week-old C57BL/6J male mice were purchased from The Jackson Laboratories. Groups of 3-5 mice per cage were housed in standard conditions (micro-isolator cage, 12-h light/dark schedule, dry pellet food and tap water available throughout the day). All animal procedures were performed according to the National Institute of Health (NIH) Guide for the care and use of laboratory animals and approved by the Animal Care and Use Committee of Pennsylvania State University. www.nature.com/scientificreports/ meniscotibial ligament (MMTL) as previously described 53 . Briefly, mice were prepped for surgery with intraperitoneal injection of ketamine (60 mg/kg) and xylazine (4 mg/kg), and the skin of the medial knee was cleaned with betadine. A 5 mm incision was made through the skin to reveal joint structures and the patellar tendon was identified. The MMTL was transected with a #11 scalpel allowing for increased movement of the medial meniscus within the joint. To close the incision site, 4-0 silk sutures in an interrupted pattern were placed and the site cleaned with betadine. Sham surgeries were performed on the left knees, in which the same procedure was performed without disruption of the MMTL. Analgesia was provided by intraperitoneal injection of buprenorphine (0.5 mg/kg) every 12 h for 3 days. Sutures were removed after 7 days. After 12 weeks post-DMM, mice were sacrificed by whole animal perfusion (with 10% NBF). Mouse knee joints were harvested, fixed with 10% neutral buffered formalin (NBF) for one week, then decalcified in two changes of decalcification buffer (14% EDTA Tetrasodium, 1.5% glacial acetic acid in dH 2 O) over one week. Samples were then embedded in paraffin wax and serial sections of DMM and sham knee joint samples were obtained at 5 µm thickness with a rotational microtome. Sections were then used for Safranin-O/Fast Green, Picrosirius red (PSR), and immunofluorescence staining and histological analysis. Safranin-O/Fast Green images were acquired with OsteoMeasure Software (https:// www. osteo metri cs. com/).

Post-traumatic induction of
Picrosirius red staining, imaging, and quantification. 5 µm sections of the medial knee joint were deparaffinized and stained with picrosirius red solution for 1 h at RT. Slides were then rinsed 2 × in 0.5% acetic acid solution, then dehydrated, cleared in xylene, and mounted with resinous medium 2,3 . Slides were imaged at 10 × magnification with polarized light microscopy and were quantified using ImageJ. Images were split into red and green channels (threshold: 75-184 for green channel, 147-225 for red channel), then areas were selected and area percentage and average intensity were quantified.
Second harmonic generation image acquisition. We have previously described the general SHG microscopy setup 13,15 . Briefly, we utilized a InSight DS + mode-locked single-box laser system, set to acoustooptic tunable filters (AOTF) to protect samples from damage. Care was taken to ensure that power setting was consistent between sample imaging. Samples were imaged with a Nikon A1 MP + Multi-Photon upright Microscope system (Nikon Instruments) with water immersion objective lens (CFI75 Apo Water 25X/1.1 LWD 2.0mm WD) for FSHG collection, and 440/20 nm band pass filters for BSHG detection. Safranin-O/Fast Green stained knee joint sections were used for SHG microscopy. Tibial cartilage, subchondral bone, and anterior and posterior horns of the medial meniscus from DMM and sham knee joints were imaged using SHG microscopy, in order to compare and quantify collagenous changes within specific regions of interest. Images of the tibia were acquired at anterior, middle, and posterior regions in reference to the anterior and posterior meniscal horns, for an even representation of the cartilage (radial zone of uncalcified cartilage), superficial subchondral bone (inferior to the osteochondral junction), and deep subchondral bone, immediately superior to the growth plate. Because the superficial and transitional zones in our DMM samples were too degraded to measure, we imaged the remaining uncalcified cartilage in the radial zone. The anterior and posterior horns of the medial menisci were analyzed by comparing vascular (external) and avascular (internal) regions for collagenous changes due to differences in blood supply. Three ROIs were captured from each area for a representative measure of the entire sample.
Each ROI was captured at a lower laser power to measure collagen content and at a higher laser power to measure fiber orientation and width. Using lower laser power prevented oversaturation of pixels between ROIs of any samples, while the higher laser power images visualized smaller fibers with lower expression during later analysis. Notably, care was taken to avoid oversaturation in each imaging session by utilizing the saturation map during image capture, as the width of discrete fibers are impossible to define if excess signal merges the fibers together.

SHG image analysis.
Utilizing the 3D reconstructions created from Z-stacked SHG images, analyses were performed as previously described 15 using Volocity software (PerkinElmer, United Kingdom), to analyze collagen content, fiber width, and orientation index.
Collagen content. To determine the collagen content in the ROIs, volume fraction (VF) for both BSHG and FSHG signal was calculated and confirmed with immunofluorescence staining. For each ROI, volume fraction was calculated by dividing the volume in the ROI producing BSHG or FSHG signal by the total volume of the image in order to quantify the amount of signal within the ROI. Thicker fibers allow SHG signal to traverse the structure to the FSHG detector, while thinner fibers reflect SHG signals to the BSHG detector. Thus, the amount of FSHG and BSHG signal within a ROI is indicative of the amount of thick (Col I) and thin (Col II) fibers respectively. FSHG/BSHG VF ratio was then calculated as a measure of expression changes in Col I relative to Col II. Ratios from each ROI were averaged by sample, then plotted with GraphPad Prism. The FSHG/BSHG VF ratio was only calculated for cartilage because other joint tissues experienced a simultaneous increase in both Col I and II expression.
Fiber width. Fiber width was determined by measuring four discrete fibers per ROI by using the line tool in Volocity to draw a line perpendicular to a fiber, measuring FSHG and BSHG signal intensity against distance across the line. FSHG signal intensity v. distance was plotted using Origin software, yielding peaks of width equal to that of the fiber. For each ROI, 3-5 fibers were measured, and the resulting widths plotted in Graph Pad by region. Fiber widths were averaged by sample. Although FSHG/BSHG ratio has been commonly used as a parameter to evaluate the thickness of collagen fibers 16  www.nature.com/scientificreports/ fibers that produce both FSHG and BSHG signal, and can only be used as a measure in locations with only one type of collagen if statements are to be made about the thickness of Col I v. Col II fibers. It therefore does not provide a concrete measurement of the fibers themselves, but rather provides a relative measure of the amount of thick and thin fibers present in a ROI.
Orientation index. Fiber orientation was determined using a MatLab Fast Fourier Transform (FFT) script, in order to quantify the organization of collagen fibers in the ECM. This calculation determines how aligned the fibers are within the sample, providing a numerical value for the angle at which the fibers are placed in relationship to each other (0 = perpendicular orientation, 90 = parallel orientation).
Immunofluorescence. Immunofluorescence staining was utilized to determine localization of collagen types I and II. 5 µm sections were deparaffinized. Antigen retrieval was achieved by incubation in 0.5% Pepsin/0.5 N HCl solution at 37 °C for 30 min, followed by incubation in hyaluronidase solution (1.5 mg/1 mL 1xTBS) at 37 °C for 5 min. The slides were then permeabilized by incubation in 0.03% Triton X in 1xTBS for 30 min. Samples were blocked in 10% normal goat serum at RT for two hours, then incubated overnight at 4 °C with rabbit anti-mouse Col II antibody (ab34712, 1:100). Slides were then washed 3 times with 1X TBS, incubated with goat-anti rabbit AF568 antibody (1:100) for 1 h at RT. Then, samples were incubated in goat biotinylated Col I antibody (ab24821, 1:50) overnight at 4C, washed in 1X TBS, and incubated with Streptavidin AF647 Conjugate (S32357, 1:100) for 1 h at RT. Negative control slides for Col I and II were stained as described, but incubated O/N at 4 °C with IgGs instead of primary antibodies. All samples were mounted with DAPI Prolong Gold, cover-slipped, and imaged with a Zeiss Axio Observer Inverted Microscope at 40 × magnification. Images were acquired with Zeiss Zen Blue 2.6 edition (https:// www. zeiss. com/ micro scopy/ us/ produ cts/ micro scope-softw are/ zen. html).
Immunofluorescence staining analysis. IF stains for Col I and Col II were quantified utilizing Zen Blue imaging analysis software. In each 40 × image, 3 ROIs (235 × 235 um to replicate the magnification of the SHG images) were selected for measurement of average fluorescence intensity (pixel intensity) and area percentage (number of pixels within an image containing fluorescence) of each collagen of interest. Threshold was set specifically for each collagen in each tissue type (cartilage, SCB, and menisci fibrocartilage) to quantify fluorescence in the matrix, and was consistent throughout analysis of each sample.
Statistical analyses. Data for each SHG and IF parameter studied was averaged per sample for n = 5 for both sham and DMM mice. Data from sham and DMM samplers were compared using unpaired Welch's t-test. Comparison between the three regions of the cartilage (anterior, center, and posterior) was performed using one-way ANOVA. All calculations were performed using the GraphPad Prism 9.1.0 program. Graphed values are expressed as mean ± SEM.

Data availability
The datasets used and/or analyzed during the current study available from the corresponding author on reasonable request.