High spatial resolution analysis using indentation mapping differentiates biomechanical properties of normal vs. degenerated mouse articular cartilage

Characterizing the biomechanical properties of articular cartilage is crucial to understanding processes of tissue homeostasis vs. degeneration. In mouse models, however, limitations are imposed by their small joint size and thin cartilage surfaces. Here we present a 3D automated surface mapping system and methodology that allows for mechanical characterization of mouse cartilage with high spatial resolution. We performed repeated indentation mappings, followed by cartilage thickness measurement via needle probing, at 31 predefined positions distributed over the medial and lateral femoral condyles of healthy mice. High-resolution 3D x-ray microscopy (XRM) imaging was used to validate tissue thickness measurements. The automated indentation mapping was reproducible, and needle probing yielded cartilage thicknesses comparable to XRM imaging. When comparing healthy vs. degenerated cartilage, topographical variations in biomechanics were identified, with altered thickness and stiffness (instantaneous modulus) across condyles and within anteroposterior sub-regions. This quantitative technique comprehensively characterized cartilage function in mice femoral condyle cartilage. Hence, it has the potential to improve our understanding of tissue structure-function interplay in mouse models of repair and disease.


Introduction 39
The articular cartilage in synovial joints has a specialized three-dimensional structure and biochemical 40 composition that provides low-friction, wear-resistance, and load-bearing properties to the tissue. 41 Mechanical loading is essential to tissue homeostasis and influences gene expression, chondrocyte 42 metabolism, extracellular matrix (ECM) maintenance, and associated interstitial fluid permeability 1,2 . spontaneous and induced tissue degeneration and regeneration [8][9][10][11] . In this regard, mouse models serve 54 as powerful tools for the targeted assessment of cellular and molecular processes and the discovery of 55 novel therapeutics related to cartilage regeneration and OA. Yet, while the structural integrity and 56 biochemical composition of murine cartilage are routinely assessed through histological and 57 molecular approaches, the evaluation of how these features translate into mechanical function is 58 limited. The main challenge in mechanical function assessment stems from their small joint size and 59 thin cartilage found in mice relative to other species 12 . Prior efforts to overcome these challenges 60 include finite element modelling and optimization of small-scale indentation techniques 3,13-15 . 61 62 Mechanical indentation (including AFM techniques), performed using either creep or stress-relaxation 63 protocols, is widely employed for assessing the biomechanical behavior of cartilage and OA-related 64 changes in many species [16][17][18] , and considered the gold standard for small animal joints 19 . Unlike 65 confined and unconfined compression tests, no sectioning or subsampling of tissue (i.e., cylindrical

Automated indentation mapping reliability 108
While previous studies have employed indentation mapping on mouse cartilage, to our knowledge 109 none have measured its accuracy and precision. Therefore, we performed three repeated mappings per 110 position for all specimens to undertake a reliability analysis of the automated surface indentation 111 technique. Each of the 31 predefined positions distributed over the femoral condyles of ten C57Bl6 112 mice was included in this analysis ( Figure 1A). The setup was developed and optimized ( Figure S1) to 113 assess the load-bearing regions of the femoral condyles and achieve non-destructive retrieval of 114 specimen post-testing for subsequent 3D XRM imaging analysis. The imposed step deformation (i.e., 115 indentation depth) on the femoral cartilage yielded typical stress-relaxation behavior, characterized by 116 a sharp increase in force followed by gradual relaxation over time until equilibrium ( Figure 1B). 117 Assessment of stress-relaxation and corresponding force-displacement curves ( Figure 1B) 118 demonstrated consistency among repeated measurements for single positions and visible differences 119 in peak reaction forces between condyles. These observations were further evidenced by the spatial 120 distribution of peak force values across condylar testing sites ( Figure 1C, Table S1). A total of 930 121 indentation measurements were retrieved, out of which only 21 produced invalid curves (seven testing 122 sites at specimen's periphery with higher angles yielded noisy signals, Figure 1C), representing a 2.26 123 % error rate during data acquisition. High reliability and absolute agreement between repeated 124 measures for individual testing sites were observed, with 4.7% intra-assay average coefficient of 125 variation (CV) (Table S1) and intraclass correlation coefficients -ICC (lower 95%, upper 95%) -126 ranging from 0.974 (0.966, 0.981) for the lateral condyle to 0.971 (0.963, 0.978) for the medial 127 condyle. Mean peak force values illustrate site-specific variations within and between condyles 128 ( Figure 1D). The lateral condyle values varied significantly per position (p < 0.0001), ranging from 129 0.07 to 0.15 N and showed a trend for higher values at outermost positions, with a slight decrease in 130 force posteriorly. The latter was also seen for the medial condyle, wherein heterogeneities in peak 131 force were also apparent (p < 0.0001) and had a wider rangefrom 0.15 to 0.294 N. Since the 132 analysis per testing site also reflects inherent deviations due to anatomical positioning across 133 specimens, data was pooled for regional (between condyles) and sub-regional (between and within 134 anteroposterior locations) comparisons. increase with indentation depth (bottom) (C) Normal peak force recorded for all three repeated 142 measures (a-c) considering each of the 31 testing sites, L1-L14 at lateral condyle and M1-M17 at 143 medial condyle, for each specimen (n = 10, S01-S10), demonstrates general agreement for intra-144 specimen measurements on both condyles (D) Mean peak force values varied within and between 145 condyle locations and higher within medial condyle testing sites. Data is presented as mean ± SD. As Table 1 shows, the average peak force was significantly higher on the medial condyle and on both  152 its anterior and posterior sub-regions when compared to lateral counterparts (Lat/Ant vs. Med/Ant and 153 Lat/Post vs. Med/Post). Interestingly, no significant differences were observed between sub-regions of 154 the lateral condyle (Lat/Ant vs. Lat/Post). In contrast, the mean peak force yielded at the Med/Post 155 sub-region was 20% lower than on the Med/Ant (p < 0.01). As cartilage thickness variations between 156 and within condyle locations could affect peak forces measured at same indentation depth 34  Needle probing (NP) thickness mapping was performed on all ten femoral condyles, which were 169 subsequently scanned using contrast enhanced XRM imaging. As shown in Figure 2A, reconstructed 170 3D datasets of murine distal femurs allowed us to validate the spatial distribution of NP testing sites, 171 thereby the corresponding ROI coordinates for imaging processing could be determined. Additionally, 172 the 2D slices confirmed that the needle probe pierced the full length of the cartilage, reaching the 173 subchondral bone (Figure 2A). The cartilage surface and cartilage-subchondral bone interface 174 positions were identified using the load-displacement curves from NP ( Figure 2B) and the cartilage 175 thickness for each position was then calculated considering the surface angle (Methods). There was a 176 2.58% rate of needle probe failure (8 out of 310 measurements, testing sites near the specimen's edge) 177 during data acquisition. NP and contrast enhanced XRM imaging yielded similar cartilage thickness 178 distributions on both condyles ( Figure 2C), demonstrating a highly significant correlation between 179 paired values (R = 0.842, n = 302, p < 0.0001) ( Figure 2D). Further method-comparison using a 180 Bland-Altman plot ( Figure 2E) illustrated XRM measurements were approximately 6.8 µm thicker on 181 average than needle probing, representing only a 1.55 voxels difference (4.39 µm resolution -XRM). 182 Moreover, no significant differences were found between pairwise mean thickness values (NP vs. showing overall agreement between methods, with average difference of 6.8 µm in thickness. Dotted 199 black lines show upper and lower 95% limit of agreement F) Pairwise assessment of mean cartilage 200 thickness NP vs. XRM per position for the lateral and medial condyles (* p < 0.05, ** p < 0.01, two-201 way ANOVA). Symbols represent the means and error bars the standard deviation.

203
Unlike peak force distributions, spatial heterogeneities in cartilage thickness were apparent among 204 individual testing sites only on the lateral condyle (p < 0.0001), with averaged values ranging from 46 205 to 76 µm; whereas thickness distributions within the medial condyle was more uniform (p = 0.06), 206 ranging from 36 to 45 µm. Nevertheless, no significant differences were observed within condyles 207 when comparing their anteroposterior sub-regions; whereas the medial condyle was significantly 208 thinner than its lateral counterpart, both in its anterior and posterior sub-regions (Tables 2 and S2). 209 Together, cartilage thickness appears as a contributing factor but not the sole explanation to 210 mechanical variations, which is also affected by differences in composition and morphology. 211 212 As expected, pairwise comparisons for individual testing sites demonstrated a significant negative 220 correlation between peak force and thickness measurements within lateral (R = −0.554, n = 135, p < 221 0.001, Figure 3A) and medial (R = −0.463, n = 167, p < 0.001, Figure 3B Altered biomechanical properties in degenerated murine articular cartilage 241 To assess the changes in mechanical response within the context of cartilage degeneration, we 242 employed the same testing protocol on age-matched Proteoglycan 4 (PRG4) knockout mice (n = 6) 243 and compared the outcomes to the C57Bl/6 controls. PRG4 is a mucin-like glycoprotein highly 244 conserved across species 36,37 and functionally relevant in joint homeostasis and lubrication 33,38,39 . 245 PRG4 loss of function, as seen in knockout mice (Prg4 -/-), leads to degenerative joint changes 246 recapitulating the phenotype of human camptodactyly-arthropathy-coxavara-pericarditis (CACP) 247 syndrome 33,40 . Histological alterations of articular cartilage have been comprehensively described, 248 and include surface roughness, tissue thickening, and loss of collagen parallel orientation at the 249 superficial layer, progressing to irreversible tissue damage with age 33,41 . Yet, the microscale 250 assessment of site-specific mechanical variations, aside from friction, has not been described so far 251 for Prg4 -/knee cartilage surfaces. Mapping of biomechanical parameters allowed site-specific 252 differences in Prg4 -/cartilage to be visualized, and outcomes were largely reproducible across 253 femoral specimens ( Figure 4A-B). These findings were supported by the combined quantitative 254 assessment between genotypes for the different condyle regions and subregions ( Figure 4C-F). Prg4 -/-255 mean cartilage thickness on both lateral (67.9 ± 3.5 µm, p = 0.032) and medial (56.3 ± 3.7 µm, p < 256 0.001) sides of the knee was higher compared to controls ( Figure 4C). Differences in cartilage 257 thickness between genotypes on anteroposterior sub-regions, however, were only detected on the 258 medial condyle ( Figure 4E). The site-specific thickness measurements enabled us to determine the 259 instantaneous modulus at each testing site for the same mechanical strain. The utilized Hayes et al.  Pairwise comparison between genotypes for mean values on the lateral and medial condyles and 275 anteroposterior sub-regions (Mann-Whitney U with Bonferroni-Dunn correction, * p < 0.05, ** p < 0.01, 276 *** p < 0.001). 277

Discussion 278
Using a novel microscale instrumented apparatus, we were able to reliably detect and quantify spatial was significantly thicker than in control littermates, even though mice were younger (10-weeks-old) 329 than in our study. 330 Reduction in the instantaneous compressive modulus in our study was confined to the medial 332 compartment. We attribute that to differences in thickening in Prg4 -/femoral cartilage compared to 333 controls, 12.6% versus 41% on lateral and medial condyles respectively, as well as Prg4 -/-334 compromised structural integrity 33,38,51 having a bigger functional impact on the medial compartment, Male C57Bl6 mice were purchased from Jackson Laboratories (Bar Harbor, ME). We based on 370 sample on a recent paper that used indentation (but not needle probing) in mice 26 , however, because 371 of the addition of needle probing we decided to increase the sample size. Animals were housed under 372 a standard light cycle and had free access to feed (standard diet) and water. Ten mice were euthanized 373 at 16-weeks of age, and hind limbs (n = 10 right) were harvested for biomechanical testing and 3D X-374 ray microscopy (XRM) imaging. Age-matched PRG4 knockout mice (Prg4 -/-, n = 6) were generated 375 and maintained on a C57BL/6 genetic background, as previously described 57 . Limbs were 376 disarticulated at the hip, followed by transection of the ligaments and careful isolation of distal femurs 377 from tibiae and menisci with the help of a dissection microscope (Leica). Femurs were preserved 378 gently wrapped in Kimwipe soaked in phosphate-buffered saline (PBS, pH 7.4) until the time of 379 assessment. All samples were mechanically tested no longer than 3h after dissection to prevent tissue 380 degradation. 381

Automated Indentation Mapping 382
The shafts of isolated femurs were glued into a 0.1-10 µL pipette tip (VWR) using cyanoacrylate 383 adhesive, fixed into a stainless-steel hex nut (Paulin, Model 848-216) and secured to the sample 384 holder ( Figure S1). This customized setup allowed for simple and proper positioning of the sample 385 exposing the load-bearing region of the condyles 27 for data acquisition, as well as non-destructive 386 retrieval of samples after testing, such that subsequent XRM imaging could be carried out. A  Figure 1A). 407