A human meniscus explant model for studying early events in osteoarthritis development by proteomics

Degenerative meniscus lesions have been associated with both osteoarthritis etiology and its progression. We, therefore, sought to establish a human meniscus ex vivo model to study the meniscal response to cytokine treatment using a proteomics approach. Lateral menisci were obtained from five knee‐healthy donors. The meniscal body was cut into vertical slices and further divided into an inner (avascular) and outer region. Explants were either left untreated (controls) or stimulated with cytokines. Medium changes were conducted every 3 days up to Day 21 and liquid chromatography–mass spectrometry was performed at all the time points for the identification and quantification of proteins. Mixed‐effect linear regression models were used for statistical analysis to estimate the effect of treatments versus control on protein abundance. Treatment by IL1ß increased release of cytokines such as interleukins, chemokines, and matrix metalloproteinases but a limited catabolic effect in healthy human menisci explants. Further, we observed an increased release of matrix proteins (collagens, integrins, prolargin, tenascin) in response to oncostatin M (OSM) + tumor necrosis factor (TNF) and TNF+interleukin‐6 (IL6) + sIL6R treatments, and analysis of semitryptic peptides provided additional evidence of increased catabolic effects in response to these treatments. The induced activation of catabolic processes may play a role in osteoarthritis development.

been attributed to increased proteolytic activity of matrix-degrading enzymes, including matrix metalloproteinases (MMPs), collagenases, aggrecanases, and proteinases belonging to serine and cysteine families. 2While previous ex vivo models of OA using cartilage [3][4][5][6][7] have provided insights about the disease, studying the degenerative changes in the extracellular matrix of human menisci may contribute with valuable new knowledge about the degenerative role of cytokines in a complex disease affecting the whole joint.The network of cytokines as key mediators of inflammation and their involvement in catabolic processes receive increasing attention.It has been reported that synthesis and mechanisms of cytokines may vary during the disease process of OA. 8 The disruption of homeostasis that cytokines cause is, particularly in tissues often subjected to high mechanical load, of key interest when studying OA disease progression.The progressive degeneration involves processes of inflammation, degradation, and synthesis, which, together often result in the loss of joint function and pain. 9,10 vitro and ex vivo models have been extensively used to study pathological changes, molecular pathways, and the effect and role of cytokines in certain conditions, and proinflammatory stimulation of human meniscus has previously been studied using cell cultures. 11However, no inflammatory disease model has been established for human meniscus using an explant tissue model. 12man OA tissue samples are often collected at the end-stages of disease, making it difficult to study early changes and factors that are involved in the disease progression. 13In vitro and ex vivo models overcome these limitations by offering a controlled environment in which initial disease mechanisms may be simulated and studied.
In the current proof-of-concept study, we establish a human meniscus ex vivo model to explore the meniscal response of healthy human meniscus to cytokine treatment using a proteomics approach to study and follow the release of matrix proteins.

| Explant harvest and treatment
Lateral donor menisci (N = 5, non-OA, four male and one female, age range 69-80) were harvested within 40 h postmortem without known chronic joint disease.The procedure was approved by the Lund University ethics committee.The meniscus was visually inspected to be macroscopically intact.
Vertical slices (1 mm wide) were cut radially from the meniscal body and were divided into an inner (inner 1/3rd) and outer region (outer 2/3rd).The inner region represents the white/white avascular region while the outer region is vascularized.Each slice is weighed ("wet weight," later used for weight correction in the quantitative analysis) and placed in medium in a 24-well plate.Incubation at 37°C, 5% CO 2 for 24 h to let the explant slices equilibrate.Medium for culture with treatment: 500 mL Dulbecco's Modified Eagle Medium (DMEM)/F12 HEPES (SH30023.01),PEST 5 mL (1:100), L-Proline 0.4 mM, insulin, transferrin supplemented media (ITS) 5 mL, vitamin C 50 µg/mL + Fungizone (first two time points).
Explants were untreated (controls) or treated with cytokines (Figure 1, Supporting Information: Table S7) and cultured in serum-free DMEM media (supplemented with ITS) for 21 days.
Medium changes were carried out every 3 days and used medium was collected (96-deepwell plate) and stored at −20°C until analysis.Media was then added to the explants (1 mL for outer, 0.5 mL for inner explant wells).We first evaluated treatment with IL1 in four samples in total (MEX0-3).Then, we evaluated seven treatments in sample MEX3.Sample MEX4 was included as a validation of results in MEX3, without separating inner/outer zones.

| Metabolic activity and glycosaminoglycan (GAG) release
GAG-release was evaluated in all samples using a 1,9dimethylmethylene blue (DMMB) assay.DMMB solution was prepared according to Farndale et al. 14  Fold-changes (treatment vs. control) were estimated using mixed linear regression models in R. Base-2 log-transformed GAG-release was used as response.Treatment, time point and zone with interactions between all terms were used as fixed effects, and subject and zone were used as nested random effects terms for MEX0-3, which had multiple biological replicates with repeated measurements.In MEX3, only zone was used as random effect.
The metabolic activity was evaluated at for time points d0, d6, d12, and d21, where 400 µL Alamar mix (10% AlamarBlue [BioRad] in medium without Fungizone and vitamin C) was added to explants after harvesting the media and incubated for 3 h at 37°C.Samples were transferred to a 96-well plate and the absorbance (570/600 nm) was measured to determine the metabolic activity of the explants.

| Mass spectrometry (MS) preparation and identification
Culture medium (50 µL) was prepared for MS analysis as previously described 15 : explant culture media (50 µL) was reduced by 4 mM dithiothreitol for 30 min at 56°C, alkylated by 16 mM iodoacetamide for 60 min in the dark at room temperature, ethanol precipitated (9:1) and then digested by 0.25 µg trypsin gold (Promega) in 0.1 M ammonium bicarbonate (AMBIC) pH 7.8 for 16 h on a shaker at 37°C.After drying, samples were resuspended in 100 µL AMBIC with 0.5 M NaCl, run through 30 kDa filter (PALL Life Sciences) and desalted with reversed-phase C18 cartridges (AssayMAP, Agilent Technologies) using a Bravo robot.Discovery MS was performed using a quadrupole Orbitrap benchtop mass spectrometer (Q-Exactive HFX, Thermo Scientific) with prior separation of peptides using a liquid chromatography system (EASY-nLC 1000, Thermo Scientific) on an analytical column (PepMap RSLC C18, 75 µm × 25 cm, Thermo Scientific) coupled on-line using a nano-electrospray ion source with a column temperature at +45°C (EASY-Spray, Thermo Scientific) using a flow rate of 300 nL/min and a 1 h binary gradient.Protein This was performed using the same combined searches as above but in series.The protein false discovery rate (FDR) was 0.01 for both searches.

| Statistical analysis
Proteins with maximum five missing values per treatment group in MEX0-3, and maximum three missing values per treatment group in MEX3-4, were considered to have sufficient data points for statistical analysis.Based on this criterium, MEX0-3 included 804 proteins, MEX3 included 822 proteins, and MEX4 included 508 proteins for statistical analysis.
Statistical analysis was performed using mixed linear regression models in R, using the lme4 package. 16Three linear regression models were used for the different data sets.In each model, base-2 log-transformed intensity was used as response.Treatment, time point, and zone with interactions between all terms were used as fixed effects, and subject and zone were used as nested random effects terms for MEX0-3, which had multiple biological replicates with repeated measurements.Treatment, time point, and zone (inner/outer) with interactions between all terms were used as fixed effects, and zone as random effect term were used for MEX3.
Treatment and time point with interactions were used as fixed effect terms for the MEX4 data set.Contrasts between treatment and control were specified using the emmeans package 17 and are reported with 95% confidence intervals based on restricted maximum likelihood estimates using Kenward-Rogers method for estimation of degrees of freedom.Estimates (base 2 log fold changes) were extracted as means across time points.Model diagnostics was conducted to validate model fit.Proteins that had a 95% confidence interval not spanning zero were considered differentially abundant.Given the exploratory nature of the study and use of mixed models that minimize the multiplicity problem, 18 we did not apply any further corrections for multiplicity, but rather we report all derived estimates to inform future studies and metaanalyses.

| Principal component analysis (PCA)
PCA was conducted to examine whether treatment effect and time effect contributed to clustering of samples, and whether similarity in release profiles between treatments could be observed.We used the PCA function in the mixOmics R package, 19 which uses multilevel decomposition for repeated measurements on data which excluded proteins with any missing values, the outcome was log transformed (base 2), then scaled and centered using the IMIFA R package. 20PCA was visualized using the plotly R package. 21

| Analysis of semitryptic peptides
As a possible indicator of degradation and catabolic effect, we studied the number of semitryptic peptides identified by MS/MS in the different treatment groups.Tryptic peptides were filtered out and only peptides identified at FDR < 0.01 in at least two files using Sequest HT search engine were counted.

| Time series clustering
Unsupervised clustering was performed to identify groups of proteins with similar treatment response profiles over time.We excluded proteins with any missing values.Protein abundances were standardized to have mean = 0 and SD = 1.Then, for each treatment and time point, we took the mean value of each protein's replicates.We calculated pairwise slope distances, and clusters were obtained using the getClusters function in the R package tscR. 22The number of clusters for each treatment was decided by visually interpreting cluster dendrogram of the slope distance.The trajectories were plotted using the R package ggplot2. 23

| Protein classification and interaction analysis
Differentially upregulated proteins with a log2 fold change >1.5 was illustrated as interaction networks using the R-package igraph. 24teraction data was collected from STRINGdb. 25Community detection was conducted using the cluster_louvain algorithm to emphasize dense subgraphs.Nodes were labeled in accordance with the protein's classification in pantherdb. 26Communities were colored according to the most common protein class in each community.

| Complementary analysis (Western blot, proximity extension assays [PEAs], and quantitative polymerase chain reaction [qPCR])
Explant media from outer menisci Days 12 and 21 (MEX3) were thawed and triplicates from the different days and treatments were pooled (15 µL each).Samples were run nonreduced on NuPage Bis/ tris gels 4%-12% in MOPS buffer.Western blots (against a COMP neoepitope fragment 27 ) were run in Tris-Glycine buffer +10% methanol and proteins transferred to polyvinylidene difluoride membranes.Filters were blocked in 3% bovine serum albumin (BSA) T-TBS and all antibody incubations were done in 3% BSA T-TBS.As substrate Super signal West Dura from Pierce were used.We used the Olink ® Explore 384 inflammation panel (Olink proteomics AB) to obtain complementary data.Explant culture media from all time points were pooled for each replicate (n = 3) and region in the MEX3 experiment using treatments: control, IL1, (oncostatin M [OSM] + tumor necrosis factor [TNF]) and (TNF + IL6 + sIL6R).Samples were randomized in order and 40 µL was used for the analysis and relative quantification performed at Olink.Proteins with linear normalized protein expression (NPX) greater than or equal to two times limit of detection was selected for differential abundance analysis, including 102 proteins.Fold-changes (treatment vs. control) were estimated using mixed linear regression models in R. Base-2 logtransformed NPX was used as response.Treatment and zone with interactions were used as fixed effects, and zone were used as random effect.
As an additional complementary analysis, we used qPCR to study expression changes between control and IL1, (OSM + TNF), or (TNF + IL6 + sIL6R) treatment groups in explant tissue from Day 21 of MEX4.Briefly, the explant tissue samples were pulverized in liquid N 2 and total RNA extracted and purified using the RNAqueous kit (Invitrogen, #AM1912).RNA concentration and purity were determined using a NanoDrop spectrophotometer (Thermo Fisher Scientific) and cDNA was synthesized from 60 ng of total RNA using the Maxima First Strand cDNA Synthesis kit (Thermo Fisher Scientific, #K1641).qPCR was performed with 3 ng cDNA per reaction, and TaqMan Fast Advanced Master Mix (Applied Biosystems, #4444556) on an Applied Biosystems StepOnePlus Real-Time PCR System.

| Metabolic activity and GAG release
The reduction of AlamarBlue was consistently higher in explant plugs compared to negative control (Figure 2), suggesting explants were healthy and viable throughout the experiment.GAG-analysis by DMMB assay revealed IL1 had a slightly increased effect on the release of GAGs than control.Log2 fold-changes was greater for IL1 Days 6-12 in the inner zone (Supporting Information: Figure S1).The GAG release for the multigroup comparison is shown in Supporting Information: Figure S2 showing small differences with various treatments.

| Identification and quantification
Across all data sets, a total of 2248 proteins were identified and quantified in the explant media (1528 in MEX0-3 inner zone, 1090 in MEX0-3 outer zone, 1303 in MEX3 inner zone, 1434 in MEX3 outer zone, and 1307 in MEX4).

| PCA
PCA revealed cytokine treatment, and to a lesser extent, release over time, as the main contributing factors to the observed clustering effect exhibited in Figure 3.

| ZONAL DIFFERENCES
We selected a set of 33 extensively researched proteins based on their functional classifications as ECM proteins, MMPs, proteases, and protease inhibitors.We performed hierarchical clustering on the mean values of standardized abundances (Z-score) of these proteins, which revealed a similar clustering pattern between the inner and outer zones (Figure 5).In both zones, we found the clusters consisting of proteins with the highest Z-score contained the proteins COL3A1, ACAN, MMP1, and MMP3.
In Supporting Information: Figure S6, a Venn diagram of the inner and outer meniscal zones illustrates the number of unique and shared differentially abundant proteins that were identified for OSM + TNF and TNF + IL6 + sIL6R.The greatest number of unique differentially abundant proteins were identified in the OSM + TNF treatment in the outer zone, with 66 unique proteins.We identified 63 proteins which were differentially abundant in both zones and treatments.
Comparing protein-protein interactions for IL1 versus Control of the inner and outer zones (Figure 6), the main similarities observed between the zones were that IL1 induces the release of inflammatory mediators such as IL6, CXCL1, CXCL6, and CXCL8.Specifically for the outer zone, our analysis revealed the interaction between IL6 and superoxide dismutase (SOD2) as well as C-C motif chemokine 2 (CCL2), along with the interactions between cystatin B (CSTB) and tumor necrosis factor-inducible gene 6 protein (TNFAIP6) and CXCL1.

| Immune system response
The cytokine treatment increased the release of proteins involved in inflammation, such as interleukins, chemokines, and MMPs.For example, some of the most upregulated proteins in both inner and outer zones of the MEX0-3 replicates were growth-regulated alpha protein (CXCL1), interleukin-8 (CXCL8), IL6, neutrophil gelatinaseassociated lipocalin (LCN2) and CXCL6 (Supporting Information: Table ST1).

| Catabolic effect
While the immune system response was most prominent in the contrast IL1 versus control (Figure 4), we observed upregulation of multiple ECM proteins in the contrast OSM + TNF (MEX3 inner zone) (Figure 6).In both the contrasts OSM + TNF versus control and TNF + IL6 + sIL6R versus control we found upregulated ECM proteins von Willebrand factor and versican interacting with upregulated cytokines and protease inhibitors.In both the inner and outer zone, we identified ECM "communities" of primarily collagens with interactions to proteases such as MMPs and cathepsins.Upregulation of ECM proteins were also prominent in IL1 + OSM versus control (MEX3 inner zone).However, besides interleukins, few cytokines were estimated to be upregulated in this treatment group (Supporting Information: Table ST2, Supporting Information: Figure S3).

| Time-dependent clustering
The time-dependent effect of treatments was studied by clustering analysis, performed on standardized release at each time point (Figure 7).Each set of proteins annotated to a cluster for a specific treatment was also plotted for control group as reference.For most treatments, we observed three clusters: one with initial increase from Day 3, one which displayed peak in release on Day 9, and one where no clear increase occurred.The patterns were not seen to the same extent in the control clusters.

| Release of COMP fragment
Western blot analysis of the COMP neoepitope (QQS 77 ) 27 (Supporting Information: Figure S5) supported the findings of increased catabolic effect, compared to the control that showed no fragment release, particularly by treatments OSM + TNF and TNF + IL6 + sIL6R at Day 12 while the release was almost absent at Day 21.

| Analysis of semitryptic peptides
Semitryptic were studied as a possible indicator of induced catabolism.In Supporting Information: Table ST6, the number of semitryptic peptides identified in each experiment and treatment group, as well as the total number of identified peptides in each experiment.In MEX3 (inner and outer zones), the highest number of  Eight proteins upregulated in OSM + TNF were not upregulated in TNF + IL6 + sILR.These were interleukin-1 beta, placenta growth factor, matrilin-2, growth-regulated alpha protein (CXCL1), SPARCrelated modular calcium-binding protein, endothelial cell-specific molecule, serpin B8, and toll-like receptor 3.
Comparing differentially abundant proteins in Olink (Supporting Information: Table ST4) and MS data (MEX3, Supporting Information: Table ST2), we found that in the inner zone, the number of upregulated proteins overlapping between Olink and MS methods were less than or equal to seven.Eight proteins in the contrast OSM + TNF versus control (outer zone) were not identified in any MS experiment, and included growth factors, cytokines, and signal receptors (Supporting Information: Table ST1-ST4).

| qPCR
The gene expression results revealed upregulation for MMP1 and MMP13 in meniscus explant tissue in all three treatments, while the ACAN, COL1A1, COL3A1, CHI3L1, and TIMP2 genes were downregulated in all three treatments (Supporting Information: Figure S7).
The results found by qPCR were consistent with the expression patterns and differential abundance results of the MS-proteomics data for MMPs in the culture medium, while the reverse was found for most of the ECM proteins.These findings support that the cytokine treatment have a catabolic effect on the meniscus, resulting in decreased expression of ECM components and increased expression of proteolytic enzymes.

| DISCUSSION
We have developed a human meniscus degeneration ex vivo model and studied protein release over time.While treatment with IL1 leads to increased release of GAG and upregulation of several ECM proteins, treatment with OSM + TNF or TNF + IL6 + sIL6R appear to induce the strongest catabolic effect.Dysregulation of protein homeostasis mediated by mechanical injury, oxidative stress, and inflammation contribute to ECM catabolism in OA. 30 Damage to the ECM and its association with joint destruction has been well studied in cartilage, [31][32][33] and while multiple studies have confirmed the clinical importance of the meniscus in OA disease, [34][35][36][37] few have investigated the meniscal response to cytokine treatment in human tissue.In this study we used a proteomics approach to study early events of OA in an ex vivo model, and provide evidence that a catabolic response can be induced in such explant model.Type I collagen is the most abundant in meniscus and constitutes 98% of the total collagen content. 38The protein is susceptible to degradation by multiple collagenases, such as metalloproteinases.
Our data showed upregulation of both MMP1 and MMP2 and several ECM proteins, particularly in OSM + TNF and TNF + IL6 + sIL6R treatments (Supporting Information: Table ST2).In MEX4, we found strong upregulation (log2 fold-change >3) of MMP1, MMP3, MMP10, and MMP14.We found the treatments OSM + TNF and TNF + IL6 + sIL6R to release matrix proteins and proteases involved in ECM breakdown (Supporting Information: Table ST2-ST3), suggesting these treatments can induce catabolic processes in meniscus.The most highly upregulated proteins released from tissue stimulated with IL1ß were primarily cytokines, chemotactic factors, and inflammatory regulators (Supporting Information: Table ST1).The low catabolic response to IL1 in this study compared to observations in previous publications [39][40][41] may be attributed to multiple factors, particularly donor age, or due to dissimilarities between the studies species.
We studied the time-dependent effect of treatments using timeseries cluster analysis and found a reoccurring pattern of peak release on Day 9 in multiple treatments, but not to the same extent in controls (Figure 7).Clusters with this pattern was particularly noticeable for treatments IL1 (MEX0-3 inner and outer zone), IL1 + IL17 (MEX3 outer zone), and OSM + TNF (MEX3 inner and outer zone).While further research would be needed to confirm these results, these initial findings suggest these treatments may have a time-dependent effect on the release of proteins as previously shown for articular cartilage. 42Clusters of proteins with high release at the earliest time point both in treatment group and control may be an artifact of the meniscus being cut upon harvest or cut into slices.
The inner zone of the meniscus is more cartilage-like, while the outer zone, located further to the edge of the joint, is more fibrous in structure. 39,43The difference in composition of the tissue is reflected in observed differences in differentially abundant proteins between the zones.In both zones, IL1 induces the release of inflammatory mediators such as IL6, CXCL1, CXCL6, and CXCL8, which may play a role in recruiting immune cells to the site of inflammation. 44ecifically for the outer zone, our analysis revealed interactions between IL6 and SOD2, a protein involved in oxidative stress response, and the chemokine CCL2.Additionally, the interactions between CSTB and TNFAIP6 and CXCL1 were identified for the RYDÉN ET AL.
| 2775 outer zone.CSTB is involved in protease inhibition, which could play a role in regulating ECM degradation. 45,46While overall there was many overlapping upregulated proteins between the zones, some treatments, such as TNF + IL6 + sILR, is seemingly preferably for inducing release of ECM proteins and proteases particularly in the inner zone (Figure 6, Supporting Information: Table ST5).However, the distinction between inner and outer zone is not clear cut and could contribute to unwanted sample variability.Given the overall similar response in both zones, studying the whole meniscus (and thus studying the average response across meniscal zones) may be the preferable approach.
Semitryptic peptides have been digested by trypsin at one end and the other terminal by an unknown protease.These peptides can be used as markers of catabolic processes, or the breakdown of cellular components. 47The highest ratio of semitryptic peptides to all identified peptides were found in treatments TNF + IL6 + sILR (MEX3 inner zone) and OSM + TNF (MEX3 outer zone and MEX4).
Olink PEA was used both as a way to expand the MS results using a highly sensitive and targeted proteomics technique, and to confirm the presence and relative abundance of proteins identified by MS, in particularly to assess the catabolic effect of treatments IL1, OSM + TNF, and TNFIL6sILR compared to controls.One notable limitation of this method is that only proteins specifically targeted to proteins in the Olink inflammation panel can be identified and will thus not provide knowledge about other important proteins.The Olink technology is optimized for plasma and serum samples while the performance in explant media is not much explored.In agreements with the MS results, the highest number of upregulated proteins were estimated in OSM + TNF outer zone (Supporting Information: Table ST4).Seven of these proteins were upregulated in MEX3 outer zone by MS and eight proteins were only found using the Olink biomarker panel.These proteins included growth factors, cytokines, and signal receptor which may contribute to a complex interplay of proteins contributing to dysregulated proteostasis in a catabolic state.
In comparison to the proteomics results, the qPCR analysis displayed similar expression patterns for MMP proteins, while the reverse pattern was found for most of the ECM proteins.While the proteomics data showed upregulation of ACAN, COL1A1, COL3A1, and MMP1 (MMP13 was excluded due to extensive missing values in control), and downregulation of CHI3L1 in two out of the three treatments, the qPCR analysis showed downregulation of five genes including ACAN, CHI3L1, COL1A1, COL3A1, and TIMP2.Additionally, while TIMP2 was not differentially abundant in the proteomics data, the qPCR analysis showed downregulation of TIMP2 in all three treatments.The downregulation of ECM genes observed in the qPCR analysis suggests that cytokine treatment may disrupt the synthesis of ECM components as previously shown in articular cartilage at protein level 15 and at mRNA level for articular cartilage as well as in menisci. 39Upregulation of the corresponding proteins in the proteomics analysis is attributed to the cytokine treatment leading to breakdown of ECM and release into explant culture media.
The experimental design of ex vivo meniscus explant models is influenced by existing models using articular cartilage.Cytokine treatment on nonhuman cartilage explants have been demonstrated to have an effect of increased release of ECM-proteins and proteins involved in catabolic response of cartilage degradation into the explant culture media when compared to controls. 48mbined mechanical loading and cytokine treatment on porcine and bovine meniscus was recently reported by McNulty et al. in   which two mechanical treatments were used; cell stretching and dynamic loading, using transcriptomics. 49The inflammatory response was highly modulated by mechanical loading.In our study of menisci, we find release of ECM-and catabolic proteins, although the specific proteins released are to a large degree different than previous studies using a cartilage model.For example, in the proteomic study on bovine knee articular cartilage, MMP13 was selected as a "representative protein" of the response group "cytokines versus control." 48By contrast, in the current study MMP13 had a seemingly negligible effect, upregulated only in IL1 versus control (MEX0-3 inner zone) and with a log2 estimate of 0.69 (lower CL 0.03, upper CL 1.35) (Supporting Information: Table ST1).
GAG-analysis may also be less informative, as GAG content in the meniscus is much lower than in cartilage. 39Thus, it is not surprising that GAG release was much more modest in our meniscus model than previously reported for cartilage. 48 important strength of the current report is use of human tissues.Many previous studies on tissue breakdown were conducted in bovine cartilage 4,5,7,48,50,51 and thus generalizability of results to humans could be limited, especially given typical ages used but also different loading patterns in animal versus human joints. 52,53We examine the effect of multiple cytokine treatments, two of which we propose as possible triggering agents of catabolism in human meniscus.
While our model allowed us to examine the effects of cytokines on the meniscus in a controlled environment, there are important limitations to consider.Mechanical loading plays a critical role in the regulatory mechanisms of the function of the meniscus and ECM remodeling.The absence of mechanical stimulation in our model limits our ability to fully capture the complex interplay between cytokine signaling and mechanical loading.Our sample size is small, but given the novel nature of the model, we think it still provides important insights into feasibility of the approach.The meniscus is a complex and delicate structure, and it can be difficult to cut slices that are consistently the same size and thickness.This potential inconsistency may lead to variability in the samples being studied.
To minimize this variability, we randomized the cut-out slices into treatments.By doing so, any differences between slices were assumed to be random rather than systematic.Moreover, the meniscus is composed of different zones that gradually change A 20 µL of sample was mixed with 200 µL DMMB-solution and absorbance was measured at F I G U R E 1 Schematic overview of the study.*Mass spectrometry analysis was not conducted for time points 6, 12, or 18 for MEX3 and MEX4.standard.Wells were washed three times.
identification was performed in Proteome Discoverer 2.5 (Thermo Scientific) using two search engines in parallel: a tryptic search against the UniProt human (UP000005640 from January 2021) sequence database combined with an MSPep spectral search against the NIST_human_Orbitrap_HCD_20160923 library (mass tolerance: 10 and 20 ppm in MS1, MS2 respectively.Other Sequest search settings were modifications: carbamidomethylation (fixed: C), oxidation (variable: M, P) missed cleavages (max 2), mass tolerance (MS1-10ppm, MS2-0.02Da).Label-free protein abundance quantification was obtained by averaging peak area intensities from the top three unique peptides for each protein.To determine individual peptide abundances, we performed a semitryptic database search to enable identification of nontryptic cleavages within the data set.
strong as expected.Thus, in the follow-up experiment, we expanded the number of treatments.In this multigroup treatment comparison, the most notable were comparisons of OSM + TNF versus control (MEX3), where 215 proteins were upregulated in the inner zone and 261 proteins were upregulated in the outer zone (Supporting Information: Table ST5).Corresponding comparison of TNF + IL6s + ILR versus control (MEX3) had 191 upregulated proteins in the inner zone and 95 upregulated proteins in the outer zone (Supporting F I G U R E 2 Mean reduction of AlamarBlue (metabolic activity) of explant culture in MEX0-3; treatments interleukin-1 (IL1) and control.Blue line: metabolic activity for meniscus plugs.Red line: negative control.RYDÉN ET AL. | 2769 Information: Table ST5).A majority of highly upregulated (log2 fold change >3, Figure 4) proteins in OSM + TNF versus control and TNF + IL6s + ILR (MEX3 inner and outer zones) had the pantherdb classification "extracellular matrix protein."Contrarily, none of the highly upregulated proteins in IL1 versus control (MEX0-3 inner and outer zones) were annotated as ECM proteins according to pantherdb classification.

F
I G U R E 3 Principal component analysis of media proteome (MEX3) in treatments (oncostatin M [OSM] + tumor necrosis factor [TNF] and TNF + interleukin-6 [IL6] + sIL6R) for inner and outer zones.There is a clear treatment effect for cytokines versus control particularly in the inner zone.

F I G U R E 4
Differentially abundant proteins in MEX3 inner and outer zones.Proteins with a log 2 fold change >3 were included.

F I G U R E 5
Standardized (Z-score) proteomic abundances of selected proteins in inner and outer meniscal zones (MEX3) across treatments, color annotated by functional classification.Heatmap shows the standardized proteomic abundances of selected proteins across treatments and the two meniscal zones (inner and outer).The color scale represents the relative Z-score (red: high, blue: low).Proteins are annotated by functional classification.F I G U R E 6 Protein-protein interaction network for treatment versus control in inner (left) and outer (right) zones.Proteins which had a log2 fold-change >1.5, annotated by classification.Red lines denote between-community interactions and black lines denote within-community interactions.See additional network figures in Supporting Information: Figure S4.

F I G U R E 7
Release profiles for upregulated proteins by cluster for treatments interleukin-1 (IL1) and oncostatin M (OSM) + tumor necrosis factor (TNF) and the respective release patterns for control group in MEX3.The blue line indicates the mean standardized abundance for proteins within a cluster.Additional cluster figures are found in Supporting Information: FigureS4.identified semitryptic peptides was for OSM + TNF.The highest number of identified semitryptic peptides in MEX3 inner zone was identified in the group treated with TNF + IL6 + sIL6R, while in both MEX3 outer zone and MEX4 the largest fractions were identified in the OSM + TNF treatment.5.4.3 | Olink PEADifferential abundance analysis of Olink was conducted as a complement to and comparison with the MS results.Differential abundance of Olink protein data revealed that most upregulated proteins were in the outer zone.In both the inner and outer zone, OSM + TNF versus control and TNF + IL6 + sILR versus control, we estimated more upregulated proteins than in IL1 versus control.In the inner zone, the same nine proteins were upregulated in OSM + TNF and TNF + IL6 + sILR.Twenty of the proteins in the outer zone overlapped between OSM + TNF and TNF + IL6 + sILR.