A Pre-Clinical Model of Synovitis Using Ex Vivo Human Synovial Tissue with Preserved Function and Architecture

Abstract

Arthritis is an inflammatory state within joints resulting in cartilage damage, pain, and loss of mobility. Recent advances in arthritis research specifically demonstrate that the joint capsule (e.g., synovium) is an important source of this inflammation, but there are no human models that replicate essential synovial architecture. To address this, the Joint Space Analysis System, or JSAS, was created. Anterior synovium was obtained intra-operatively from patients undergoing Total Knee Arthroplasty (TKA). Synovium was dissected and sectioned in 3 mm biopsy cores. Cores were placed in the upper well of a 5 μm or 0.4 μm transwell with 300 μl of DMEM with10% FBS. In the bottom well, 600 μL of media was added, and exchanged every 2-3 days. Viability was assessed out to 7 days in hyperoxic (50%), atmospheric/standard (~21%) and physiologic (5%) incubation conditions. Stimuli in the bottom well included monocyte chemoattractant protein 1 (MCP-1/CCL2), lipopolysaccharide (LPS), N-acetyl cysteine, S. aureus, and B. burgdorferi. Media was stored for ELISA, and tissue was stored for formalin fixed paraffin embedded (FFPE) analysis. In standard conditions, synovium remained fully viable for 3 days. Stimulus modified the structure and function of intimal lining and sublining synovial cells including loss of the resident macrophage border, sublining expansion, upregulation of pathogenic fibroblasts, and production of cytokines IL-1β and TNFα. Immune cells and fibroblasts migrated to the bottom chamber (5 μm pores) per flow cytometry analysis. Mobile B. burgdorferi migrated into tissue at the 0.4 μm pore size while non-motile S. aureus did not. Relevant cytokines were expressed in sufficient quantity for ELISA. JSAS is a modular system capable of studying acute alterations to human synovium, allowing for the complexity of 3D structures in a pre-clinical model while maintaining biologically relevant structure and function.

SUMMARY:

We developed a human ex vivo transwell system capable of evaluating acute inflammatory, infectious, and structural changes in synovium.

INTRODUCTION:

Arthritis afflicts over 20% of all U.S. adults and 50% of those over age 65 with chronic conditions¹. It is also a leading cause of disability worldwide². Arthritis includes not only cartilage damage, but also inflammation of the joint capsule or synovium (e.g., synovitis). Chronic synovitis is likely a major source of pro-inflammatory cytokines that potentiate bone and cartilage damage in osteoarthritis and rheumatoid arthritis^3,4. Therefore, understanding the immune-mediated changes within synovium is likely essential for developing new therapeutic and possibly preventative options for patients with arthritis.

Synovium is an organized structure composed of two major layers: the cellular intimal lining (IntL) and the sublining (SubL). The IntL interfaces with synovial fluid (SF), and the SubL makes SF components and contains vascular structures⁵. Within the IntL, there are both macrophages and fibroblast-like synoviocytes (FLS). The macrophages of the IntL are resident cells, locally renewed by interstitial macrophages within the SubL⁶. These resident synovial macrophages (RSMs) are M2-skewed by expression of CD206 and TREM2, epithelialized, and express tight junction markers. This sub-structure of the IntL may be key in functionally and physically maintaining joint space homeostasis^5,6. This has been observed in rheumatoid arthritis (RA). For example, when the IntL breaks down, patients with rheumatoid arthritis experience flare symptoms, which resolve when the lining rebuilds^6,7. Spontaneous RA remission is specifically associated with MerTK ⁺CD206⁺ synovial macrophages⁷. RSMs were also identified in SF of patients with acute joint pain, finding that the severity of infectious or inflammatory disease correlated with the quantity of RSMs and inflammatory cytokines⁸. Unfortunately, the relationship between the IntL and SubL structures, functional alterations of RSMs and FLS, and the production of inflammatory mediators that may go on to damage cartilage and/or bone are difficult to query, especially in humans.

To query underlying mechanisms of synovitis in multiple clinically relevant settings, a human ex vivo model called the Joint Space Analysis System (JSAS) was developed. This system produces reproducible results with modular elements to allow broad evaluation of inflammatory and infectious arthritis. This manuscript describes the processes of tissue acquisition from common open-joint surgeries such as total joint replacement, JSAS design, the histomorphometry necessary to quantify the acute structural alterations in synovium, and the functional responses of tissue including by immunofluorescence (IF), RNAscope, and enzyme linked immunosorbent assay (ELISA). JSAS differs from common human in vitro models by maintaining the IntL and SubL structures, allowing for near in vivo insights into acute or acute on chronic synovitis that leads to cartilage damage. By maintaining the native structure of synovium, it is possible to dissect the RSM and FLS intrinsic capabilities to dampen or perpetuate inflammation. As an ex vivo model, it neither requires differentiation of peripheral immune cells into synovial macrophages nor oversimplification of synovium to 1-2 discrete cell types only. It differs from animal models such as Collagen Induced Arthritis (CIA) by using naturally diseased human tissue, allowing pathologic-specific evaluation that is human-relevant. Finally, multiple infections can be simulated, allowing for a model of septic arthritis in addition to osteoarthritis and inflammatory arthritis. The main limitation is culture to 3 days in standard conditions prior to the onset of apoptosis.

PROTOCOL:

Patient consent and tissue processing protocol was approved by local Institutional Review Board (IRB).

1. Tissue acquisition and biopsy:

1.1 Identify patients pre-operatively based on surgeon schedules. Recruit patients aged 18-65 who are receiving open joint surgery such as total knee arthroplasty (TKA) or Tibial Tubercle Osteotomy (TTO, also known as Fulkerson’s) who are not on systemic immunomodulatory medication.

NOTE: Ideally exclude patients who have more than one arthroscopy or more than one glucocorticoid injection, as these patients had a higher intra-operative failure of tissue acquisition than other populations due to fibrosis. Given risk of infection transfer, defer recruitment of patients with history of Hepatitis C or Human Immunodeficiency Virus (HIV) unless pertinent to outcomes. Practice safe handling practices at all times with fresh human tissue regardless of infectious disease status.

1.2 Orthopedic surgeons perform arthrotomy and dissection as part of standard operating procedure. Identify the anterior synovium immediately proximal to the trochlear groove and use electrocautery to excise the synovium en bloc while avoiding dissection of the articularis genu muscle or near the medial and lateral epicondyles. Synovectomy is relatively neutral for TKA outcomes, and in inflammatory conditions, may be beneficial^9,10. Pass synovium sterilely to the research team member and place in a sterile specimen container with cold Dulbecco’s phosphate buffered saline (DPBS) on ice.

NOTE: Terminate tissue acquisition if there are abnormalities noted by surgical team, or if synovectomy would impair closure or surgical outcome. The latter in particular would be highly uncommon in TKA, as synovium excision may be utilized as an extensile procedure to loosen tight joints. Quantity of excision depends upon the size of the experiment. It is possible to obtain approximately four 3 mm biopsy cores per 1 cm² of tissue. In many cases, it is possible to obtain 24 cm² or more of tissue. Tissue weight is highly dependent on surgical dissection thus is not as reliable as surface area for determine adequacy for the experiment.

1.3 Return synovium to the lab and wash twice in cold DPBS for 10 minutes on a plate rocker.

1.4 Transfer synovium to a 10 cm or larger sterile petri dish under sterile conditions in a biosafety level 2 (BSL2) certified biosafety cabinet. Add cold, sterile, low glucose Dulbecco’s Modified Eagle Medium (DMEM) media containing sodium pyruvate and L-glutamine, enough to fill the dish halfway. Identify the synovial lining by the pearlescent IntL layer (Figure 2A), as opposed to the cauterized side (Figure 2B).

Figure 2:

Knee anterior synovium with the intimal lining held by hemostats to visualize the delicate nature of this layer, which has peeled away (A). The posterior side of the excised tissue demonstrating electrocautery (black arrows) and coagulation (white arrows) (B).

NOTE: The IntL is fragile. Careful not to peel it away or it will dissociate from the sublining (Figure 2A, gripped by hemostat). Pearlescence may be altered by prior history of hemarthrosis (appearing yellow-orange due to hemosiderin), high BMI (covered in adipose), or inflammation (red, friable, and villous). It is absolutely essential to correctly identify the IntL prior to proceeding. Examine carefully for signs of surgical excision, cautery, bleeding, and coagulation to determine following steps are not occurring on the IntL.

1.5 Dissect away extraneous adipose and stromal tissue on the cauterized side to isolate synovium, maintaining approximately 4-5 mm of tissue depth in all areas. Iris scissors or scalpels may be used per personal preference. Change the petri dish media as necessary to maintain field of view.

NOTE: Posterior stromal or adipose tissue is easily excised. However, difficulty with dissection or feeling great resistance is indicative of retained ligament, tissue metaplasia such as inappropriate cartilage deposition, or cutting through the IntL.

1.6 Once dissected to appropriate depth, obtain 3 mm biopsy cores as needed for a minimum of 2 technical replicates per tested condition. Perform with the synovial lining facing the coring tool as opposed to against the surface of the petri dish. A quality biopsy punch may be used approximately 10-15 times before needing replacement due to dulling, as a dull biopsy punch will lead to damage of the intimal lining structure. Perform coring with forceful perpendicular load and gentle twisting. If this twists the tissue, the biopsy tool has become dull and needs replacement. Stabilize with non-dominant hand using forceps or hemostats.

NOTE: 3 mm cores were used to maximize tissue perfusion as well as the number of cores per patient and guarantee enough technical replicates for these experiments. Alternative size cores may be used, but only 3 mm was tested for this protocol.

NOTE: excess retained adipose from insufficient dissection may cause the core to float in JSAS, which is suboptimal. If the cores float to the surface rather than remaining submerged, return to dissection or choose a submerged core.

1.7 Open a second petri dish and fill with 5-10 mL of warmed (37°C) media with 10% FBS. With flat or non-toothed forceps, transfer cores to this dish to wash away debris with 1-2 seconds of gentle mechanical agitation. Do not pinch the cores tightly or this may damage the intimal lining.

NOTE: May pause here for ≤ 24 hours by placing tissue cores in incubator at standard setting of 37°C and 5% CO₂.

2. Set up of Joint Space Analysis System (JSAS).

2.1 Prepare the bottom well of JSAS with 600 μL of warmed (37°C) low glucose DMEM with 10% FBS. Though 1% penicillin-streptomycin has been tested without noticeable negative effect in this system, antibiotics are not typically added to media if sterility has been adequately maintained.

NOTE: If co-culturing with bacteria or a cell line that requires a different media, the bottom well may be a different media. Avoid antibiotic media if co-culturing with bacteria.

2.2 To the bottom well, add the treatment of choice. For example, lipopolysaccharide (LPS). To avoid osmolarity differences, it is recommended volume of treatment not exceed 2.5% of the total volume (e.g., ≤15 μL for 600 μL). Buffer any pH changes from treatments to maintain pH 7.0-7.4. Otherwise, utilize vehicle controls.

2.3 Choose transwell pore size depending on desired experimental outcome (Figure 3).

Figure 3:

Joint Space Analysis System and utility of different pore sizes, stimulus treatments, and outcome measures. Biopsy cores are placed in the upper well (A,B). Depending on desired outcome measures, need to co-culture with autogenic, allogenic, or xenogenic tissue, need to co-culture with bacteria, different media and/or pore sizes should be utilized (C,D).

2.4 Add 300 μL of low glucose DMEM with 10% FBS to the transwell.

2.5 Gently transfer individual cores with sterile forceps into the media of the transwell. The orientation of the tissue cannot be controlled; however the adiposity and natural buoyancy of the stromal tissue naturally makes the intimal lining face down for the majority of tissue cores. Carefully visualize the cores as the process of coring may disrupt the intimal lining, and avoid using cores that appear frayed at the start of the experiment. If the core is floating, improve dissection and chose an alternate core.

NOTE: consider collagen matrix if tissue orientation is important, however this will impact the time to develop a chemical gradient with the bottom well.

2.6 Place in incubator at 37°C and 5% CO₂ for desired length of time.

NOTE: Earliest time point recommended for biologic changes is 8 hours. Latest validated time point for biologic activity is 3 days.

3. Experimental Takedown and Analysis

3.1 For histology, use sterile, non-toothed forceps to transfer tissue to 3-5 mL of 10% neutral buffered formalin (NBF) for formalin-fixed, paraffin-embedded (FFPE) analysis. Store in NBF for 3-7 days prior to embedding and sectioning. Do not let tissue sit in NBF for extended periods as this worsens autofluorescence. Perform histologic analysis including hematoxylin and eosin (H&E) ¹¹ staining for histomorphometry, immunofluorescence (IF) ¹², and/or RNAscope with IF ¹³ according to standard methods.

NOTE: careful not to puncture the transwell with forceps when removing the tissue. Be extremely gentle in transferring the core or the intimal lining may dissociate. Gently place the core in fixative and do not agitate.

NOTE: NBF should be used inside a certified chemical fume hood, must be disposed of according to local environmental health and safety (EHS) guidelines, and may not be poured down sinks. Unused human tissue from the experiment is placed in excess 10% bleach for 30 minutes, then placed in secondary containment per local EHS requirements, then disposed of in labeled biohazard bins for incineration. The bleach solution may be poured down the sink if approved by local EHS. Follow local guidelines, which may differ based on institution.

3.2 Embed technical replicates in the same FFPE block. Use screened cassettes to avoid loss of sample. Section in the longitudinal axis at 5 μm thickness. Place three serial sections on one slide (Supplemental Figure 1).

3.3 Image H&E slides at 10X and 40X using brightfield microscopy. Identify the IntL as the organized cellular layer at the interface between tissue and empty space. Acquire 3 representative 10X sections and 10 representative 40X sections across a minimum of 2 technical replicates per treatment condition, maintaining the IntL in field of view. Capture images as high-resolution TIFFs. Perform histomorphometry analysis:

• 3.3.1Perform measurements in ImageJ/FIJI (version 1.54p). First, Set Scale using the Analyze menu → Set Scale. Set Scale requires use of a calibration slide imaged on each respective microscope at each objective. First, use the straight line tool to create a line of known length based on the calibration slide image. Assign the scale globally. At 10X using AmScope T390B-3M with AmScope MU1000, the scale is 3 px/μm. At 40X, the scale is 12 px/μm.
• 3.3.2Measure SubL thickness using the 10X images with the straight line tool. Measure thickness as a plumb depth from the IntL surface to the bottom of the SubL (μm). After drawing each line, use keyboard shortcut “m” to measure the length of the line. Average a minimum of 5 measurements per each 10X image (Figure 4A, B). The transition between SubL and stromal tissue is demarcated by transition to less organized, less cellular areas containing collagen and adipose (Figure 4B). Bisect the transition point between tissue compartments.
• 3.3.3For 40X measurements of IntL thickness, IntL cellularity, IntL integrity, and SubL cellularity, begin by opening the 40X TIFF in Adobe Photoshop (version 27.1.0, or any similar image editor). The IntL is defined as a layer ~1-5 cells thick organized into a distinct cellular layer at the interface between tissue and empty space (e.g., synovial fluid) (Figure 4C).
• 3.3.4Create a new layer. Select the Polygonal Lasso tool. Using a mouse click to initiate the lasso as well as anchor inflections in the line, bisect the transition between the IntL and the SubL across the entirety of the image. Then, continue the lasso just outside the IntL in the empty space. Close the lasso with a double click. Invert the selected area (select → inverse,), then fill the space with black using the paint bucket tool. This will cover all areas that are not the IntL. Save this image as a high resolution JPG representing the IntL, then hide the layer.
• 3.3.5Create a new layer. Under polygonal lasso settings, select “intersection” feature which will only select where the new lasso area and old lasso areas intersect. Lasso around the sublining layer by clicking through the middle of the IntL, around the left and right edges of the image into the background canvas, and then bisect the transition between SubL and acellular tissue behind it. If the bottom of the SubL exceeds the dimensions of the 40X image, simply select the background canvas around the remaining edge. Again select the inverse of this area as above, then use paint bucket tool to fill with black. Save this image as a high resolution JPG representing the SubL. Save this layered image as a PSD for future reference.

  NOTE: Erythrocytes are common in these slides and may interfere with future cell count measurements. If there are large areas of vasculature, use the polygonal lasso tool, the paintbrush tool, or other tools of choice to create masks on a separate layer covering these areas

• 3.3.6In FIJI, open the IntL mask image. Measure IntL thickness (μm) as an average of a minimum of 5 measurements over the 40X image using the straight line tool as in 3.3.2-3.
• 3.3.7Measure the total length (μm) of the IntL using the segmented line tool. Then measure the length of only the intact areas, also with the segmented line tool (Figure 4D, E). IntL sections that are not intact are demarcated by gaps between cells, moth-eaten appearance, and breaking or flaking away of IntL from tissue. Calculate the % intact IntL border in as (intact IntL/total IntL)*100. This must be compared to same time point controls to compensate for any sectioning artifact.
• 3.3.8Measure cellularity of the IntL using the IntL mask image. First, transform to 8-bit (Image→Type→ 8-bit). Then threshold (Image→Threshold) to minimize noise and maximize cell identification. In the threshold menu, deselect “dark background” and use the bottom toggle for adjustments. Manually determine threshold per image aided by disappearance of noise (e.g., background artifact) while retaining original number of nuclear bodies. Compare frequently between original image to maintain fidelity. When noise is minimized (false positive) but hematoxylin signal is maximized (true nuclear positive), it is appropriate to set threshold (Supplemental Figure 2).

  NOTE: Threshold will be consistent across a single slide scan, but will differ between slide scans and also between any individual 10X or 40X images. It will also differ between batches of H&E, as the depth of color is unlikely to be consistent. Thus, simple macros are inappropriate if there are differences in any of the following categories between images: patient, H&E batch, section thickness, lighting conditions (room or microscope light brightness/color temperature), software used to acquire the image, microscope used to acquire image, and use of white balance or color balancing algorithms. This list is not exhaustive. Using AmScope T390B-3M microscope with AmScope MU1000 camera and AmScope camera software (v.10.11.2024), with microscope light at maximum in a dimly lit room, utilizing automatic white balance settings, 10 random 40X images encompassing 3 patients, the threshold ranged from 96-182, with an average of 137. This broad range reflects the differences in the underlying 8-bit histograms, not user inconsistency. Thus, assuring visual fidelity between original image and threshold image is essential, not the exact number of threshold. Automating this process is out of the scope of the typical bench scientist and would require programming from an image analysis professional.

• 3.3.9Use Analyze Particles (Analyze→analyze particles) with cell size 8-150 μm² to identify and count nuclei as a surrogate for cell number. Select “overlay masks” in the settings to visualize the identified cells, again comparing to the original image. As the IntL is primarily a linear, not volumetric, structure, normalize this 100 μm of IntL length (e.g., #cells/100 μm). This averaged approximately 13.19 (range 5.4-21.9, SD 3.9) cells per 100 μm in freshly retrieved synovial tissue.

  NOTE: The range 8-150 μm² was determined based on Set Scale and iterative visual comparison of thresholded image to original image. The limits determined empirically through the dataset correspond to shape assumptions based on macrophages having a cell diameter that can range from approximately 8-20 μm^14,15. Larger sizes were excluded as this often represented multiple discrete nuclei in close proximity being read as one cell.

  NOTE: Similar workflows may be established in, for example, QuPath.

• 3.3.10Open the SubL mask with erythrocytes excluded (Figure 4G). Perform 8-bit transformation, threshold and Analyze Particles as 3.3.7-3.3.8. As the SubL is a volumetric layer, normalize the number cells counted to 100 μm2 of SubL area (e.g., measurement of non-masked areas) (Figure 4H).

Figure 4:

Histomorphometry of human synovium detailing methodology of quantitative analysis in FIJI (“FIJI is Just ImageJ”). Scale bar = 100 μm (A, B at 10X), otherwise 20 μm (40X). SubL depth is performed on 10X images, while all other measurements are performed on 40X. The IntL is identified (C) and masks are created to isolate it for measuring IntL depth and length (D). Loss of cells at the IntL border indicates an incomplete structure (E). The length of the dashed black line divided by the length of the dashed red and black lines indicates the percent of the IntL border intact. On the masked image, convert to 8 bit and threshold while comparing the original image to maximize true nuclear signal while minimizing noise or background dots. Color overlay masks are used to identify nuclei identified as cells after Analyze Particles is utilized (F). Black represents signal not identified as cells. Masks are also created on the SubL (G), with blacked out areas representing endothelium. This minimizes quantification of erythrocytes. After converting to 8-bit, setting threshold, and Analyze Particles, color mask overlay again represents identified cells, while black represents uncounted signal.

3.4 For flow cytometry analysis of migratory cells in the bottom well, centrifuge at 300 x g (1200 rpm) at 4°C for 10 minutes and keep the cell pellet on ice. Expect ~20,000 live, migratory cells per 3 mm biopsy core at 24 hours. Flow cytometry methods are published previously ⁸.

3.5 For ELISA, centrifuge media at max speed (> 5000 x g) for 5 minutes at 4°C to pellet any cells or cell debris. Store at desired temperature. ELISA methods are published previously ⁸.

RESULTS:

Optimizing oxygen culture conditions

Physiologic tissue perfusion is approximately 6-8% oxygen while standard incubation conditions for cell culture are at atmospheric oxygen levels (~21%)¹⁶. Comparatively, hyperoxygenation may improve cell culture viability or be toxic¹⁷. Adequate utilization of JSAS necessitated early determination of synovial tissue viability, therefore apoptosis was assessed by caspase-3 immunohistochemistry or immunofluorescence on day 0, 1, 2, 3, 5, and 7 at 3 oxygen conditions: physiologic (5% O₂), atmospheric/standard (21% O₂), and hyperoxic (50% O₂). Apoptosis began at day 3 in standard incubation conditions, and peaked at day 5. In both hyperoxic and hypoxic conditions, apoptosis was visualized at days 1-2 before peaking on day 5. For this reason, standard incubation conditions were chosen for further experiments (Figure 5). However, depending on study of choice, alternative oxygen conditions may be used with apoptosis remaining comparable on days 1 and 2 to day 0.

Figure 5:

Apoptosis by caspase-3 signal in hypoxic (5% O₂), normoxic (21% O₂), and hyperoxic (50% O₂) conditions. Statistics not performed as this was exploratory with n=1 patient per condition, with 2 technical replicates, and ten 40X images analyzed per bar. Error bar represents standard deviation.

Simulation of clinically relevant pathology

Lyme Arthritis

Lyme Arthritis (LA) models are currently performed in in vivo models, particularly rodent models. This may not be preferred, as the white footed mouse Peromyscus is the natural host species for Borrelia burgdorferi, the causative agent of LA¹⁸. Further, immune responses in mice may not capture what happens in humans¹⁹. Lastly, LA in particular results in a relapsing-remitting synovitis, and therefore a model to evaluate synovitis specifically is essential to understanding the pathogenesis of LA²⁰. To test JSAS as a model for LA, the ability of B. burgdorferi (B31, clone 5A1; highly arthritogenic OpsC Type A) to 1) survive in the bottom well and 2) infect tissue in the upper well using the 0.4 μm pore transwell was evaluated. Using standard sterile culture technique in a BSL2 laboratory, B. burgdorferi was thawed from glycerol stock and spiked into 10 mL of warmed Barbour-Stoenner-Kelly (BSK)-H media in a 15 mL conical. This was cultured at 34°C for 7-10 days and counted using a Petroff-Hausser counting chamber under Ph2 phase light to visualize the spirochetes. The bottom well was then inoculated with 1x10⁶, 5x10⁵, and 2.5x10⁵ live B. burgdorferi spirochetes from the late exponential phase into 50% DMEM + 10% FBS and 50% BSK-H media in the bottom well. Inoculate volume was generally <10 μL but varied depending on bacterial growth on the day of counting. Tissue was cultured for 1-3 days at standard incubation conditions. It was found that B. burgdorferi migrated through the 0.4 μm transwell and infected tissue as demonstrated by RNAscope targeting 23srRNA with paired immunofluorescence (Figure 6)²¹. This was consistent at all inoculate doses and all days of culture at normoxic conditions. These inoculate doses and the amount of media in the bottom well were determined with preliminary experiments examining initial inoculate dose (100,000 vs. 1,000,000) and starting media volume using a 50:50 mixture of low glucose DMEM with 10% FBS and BSK-H (Figure 6B,C). At 3 days of culture, the end of maximal viability in synovial tissue, there was no discernable difference in B. burgdorferi viability at any of the starting media volumes, but there were differences by oxygen conditions by SYBR Green: Propidium Iodide ratio²². Therefore 600 μL starting volume in standard incubation conditions for 3 days was used for Lyme Arthritis simulations.

Figure 6:

Dual immunofluorescence (CD68, ZO-1) and RNAscope (23srRNA of B. burgdorferi) visualizing B. burgdorferi overlaid with synovial cells (inset, right) at 24 hours of co-culture (A). B. burgdorferi viability by ratio of SYBR Green: Propidium Iodide over 7 days with starting inoculation of 100,000 spirochetes (B) and 1,000,000 spirochetes (C). A Two-way ANOVA with Tukey post-hoc correction was used for statistical significance where * = p<0.05, ** = p<0.01, *** = p<0.001, and **** = p<0.0001. a = comparison between 21% and 50%, b = comparison between 21% and 5%, and c = comparison between 50% and 5%. Differences in media volume was not assessed.

NOTE: As the cores are 3 dimensional structures, exact cell counts cannot be established, thus MOI cannot be accurately reported. Hypothetically, a biopsy core of 3 mm diameter and 3 mm length is a volume of 28.27 mm³, and averaged over 10 random patients in the dataset, the average number of cells per 100 μm² was 2.6. Using a first order volumetric extrapolation from 2D histologic counts, this equates to approximately 1.5 x10⁸ cells possible in each biopsy core. Thus, MOI (spirochete:cell) is roughly 1:600 (2.5x10⁵ spirochetes), 1:300 (5x10⁵ spirochetes), 1:150 (1x10⁶ spirochetes). Additionally, as B. burgdorferi does not form colonies on a plate, CFU cannot be reported.

Septic Arthritis

Lyme Arthritis is rare and atypical form of septic arthritis (SA) in comparison to Staphylococcus aureus SA, a common cause of SA in human native and prosthetic joints^23,24. Evaluation of septic arthritis is may be performed using in vivo models with direct injection into the joint space²⁵, or retrospective evaluation of retrieved tissue from humans. To determine if B. burgdorferi is unique in its ability to infect tissue in this system, it was compared to the clinically relevant USA 300 MRSA strain and an isogenic agr deletion mutant^26,27. Both strains were transformed with the pCM29 plasmid to mediate GFP expression with chloramphenicol resistance^28,29. As the agr operon drives the expression of multiple virulence factors, deletion results in diminished toxin production, yet provides survival benefits that include increased expression of adhesion molecules, superior biofilms, and possibly intracellular persistence^30-32. Thus, Δagr mutants are important in chronic and/or nosocomial infections.

As S. aureus is not considered motile, the transwell and bottom chambers were inoculated in separate experiments using the 0.4 μm pore transwell. For infection of the transwell chamber, it was expected S. aureus would directly adhere to tissue. In the bottom well, an indirect, toxin-mediated effect was expected, but no S. aureus infecting the tissue. Inoculation was approximately 1x10⁶ CFU, corresponding to 5 μL. Twenty-four hours after infection, tissue was fixed in 10% NBF and then cryoembedded to preserve the GFP signal. S. aureus was observed on tissue only with direct inoculation of the transwell. Unlike B. burgdorferi, S. aureus remained on the surface of tissue at 24 hours (Figure 7). The Δagr strain was observed to have a population in excess of the WT strain. No S. aureus was observed co-located with tissue when the bottom well was inoculated.

Figure 7:

Fluorescence microscopy of GFP-tagged S. aureus, WT (A), and Δagr (B). In B, green signal overlying cells is collagen autofluorescence. Scale bar = 40 μm.

Sterile infectious and inflammatory arthritis

To evaluate the response of synovial tissue to simulated infectious (LPS 2.5, 25, 250 μg/mL final concentrations) and inflammatory (MCP-1 10, 100, 200 ng/mL final concentrations) arthritis, synovium was cultured with increasing doses of each stimulant and assessed effect on tissue at 24 hours using methods described above. This resulted in distinct IntL morphology changes and especially alterations in synovial cytokine response (Figure 8). Specific morphologies observed included friable IntL (Figure 8A, arrows), and thickened IntL with decreased cellular density (compare brackets). MCP-1 treatment is consistent with OA and possibly RA environments, and induced pathogenic fibroblast marker Podoplanin (PDPN) in the sublining^33,34 (Figure 8B). Comparatively, LPS, a bacterial cell wall component consistent with a septic arthritis environment, increased PDPN in both the IntL and SubL. LPS also increased IL-1β signal in the IntL. In contrast, IL-4 appeared primarily at the endothelial border with both treatments. This demonstrates treatments with JSAS induce structural and functional changes in synovium.

Figure 8:

H&E 40X micrographs of synovial tissue exposed to MCP-1 and LPS, focusing on the changes to the intimal lining structure (A). IF images demonstrating IL-4, IL-1β, and TNFα induction with treatment (B). Arrows denote PDPN positivity in sublining cells.

Evaluation of migratory immune cells present in synovial tissue

Though the peripheral immune system is not present in JSAS, immune cells already present in the tissue at the time of acquisition have the capacity to respond to stimulus. To assess the presence of migratory immune cells, flow cytometry was performed on the cells in the bottom well with the 5 μm pore size, using previously described methods⁸. There were neutrophils (CD66b+), NK cells (CD56+), NKT cells (CD3+CD56+), T cells (CD3+CD56−), B cells or dendritic cells (CD20 or CD11c, “dump” channel), monocytes/macrophages (CD11b+), and M2 macrophages (CD11b+TREM2+). Of these, T cells were most abundant (Figure 9A). Statistically, there was no difference in cell migration based on bottom well stimulus. Cell counts ranged from approximately 2,000-20,000 living cells per well at 24 hours. As the only immune cells able to migrate were those already present in the tissue, this represents baseline migratory leukocytes in OA tissue. It would be expected that in, for example, septic arthritis patients, higher neutrophil counts may be present. Comparatively, in RA tissue, higher lymphocytes may be present. Regardless, cells in the bottom chamber represent what cells would be likely to migrate into synovial fluid in vivo, demonstrating a synovium and synovial fluid simulated chamber with JSAS.

Figure 9:

Analysis of live cells and secretome within JSAS. Percent of live, migratory immune cells (Live/Dead Blue negative, CD45+) in the bottom well at 24 hours (A). Data from 3-4 patients per bar. There was no significance by two-way ANOVA with Tukey post-hoc. MCP-1 200 ng/mL not performed. ELISA of OPG and sRANKL, n= 2-3 patients per bar with 2 technical replicates per time point, e.g. 4-6 data points (B). Multiple Mann-Whitney U tests.

Secretion of arthritis-relevant proteins into supernatant

To analyze if proteins were being secreted by synovium, ELISA was performed after 24, 48, and 72 hours untreated in JSAS for osteoprotegerin (OPG/TNFRSF11B) and soluble receptor/activator of NF-kB (sRANKL). Tissue constitutively expressed OPG with a daily increase that was significant by 72 hours (Figure 9B). There was low expression of sRANKL compared to previously published values for SF. For reference, healthy SF sRANKL ranged from 0-123 ng/mL, while inflammatory SF sRANKL ranged from 14-192 ng/mL, and septic SF from 74-277 ng/mL⁸. Likewise, OPG ranged from 0-1 ng/mL in healthy SF, 0.4-2 ng/mL in inflammatory SF, 0.18-3.5 ng/mL in septic SF. Conversely, all measurements in JSAS were in picogram range, though at 72 hours the amount of OPG in media approximated 1 ng/mL. The ratio of OPG:RANKL may be an important measure of osteoclast activation, and hence, bone resorption, both in arthritis and in metastatic settings^35,36. As this system does not contain osteoblasts unless co-cultured with bone, only the soluble form of RANKL is present for this demonstration, which is likely derived from lymphocytes³⁷.

The utility of JSAS as a pre-clinical therapeutic model

The bridge from bench to bedside is often broken in the translation of results from animal models to human trials. Assessing the functional changes of human synovium to novel therapeutics may therefore be advantageous to quickly distinguish what may be beneficial or harmful to human disease. For example, oxidative stress has been implicated in the development of OA and RA, and therefore antioxidants have been evaluated as a potential therapy^38-41. One such agent, N-acetyl cysteine (NAC), currently lacks standardized preparation and delivery, and also has resulted in contradictory effects in humans compared to rats^42-44. NAC likely supplements intracellular antioxidants to help prevent or combat oxidative stress utilizing thiol pathways, such as glutathione⁴⁵.

As an example of the pre-clinical use for JSAS in development and/or validation of arthritis therapeutics, synovium was treated with 10 mg/mL (final concentration) of freshly prepared NAC buffered to a pH of 7.4 with NaHCO₃. This was added to the bottom well. There was high expression of inducible nitric oxide synthase (iNOS) in the IntL at Day 0. Without NAC treatment, iNOS expression in the IntL was nearly absent at 24 hours, but iNOS was maintained with NAC (Figure 10).

Figure 10:

IF of CD68 (macrophages, green) and iNOS or MMP9 (red) in serial sections from human arthritic synovium cultured for 24 hours in JSAS with and without NAC. Scale bar = 50 μm.

The role of iNOS in the IntL has not been established, however iNOS is also required for nitric oxide (NO) signaling to neighboring macrophages, which may be pro-regulatory and anti-antigen presentation in chronic inflammatory (e.g., cancer, or potentially, OA) settings⁴⁶. However, reactive nitrogen species (RNS) generated by iNOS are also important in anti-microbial settings, and excessive generation of RNS is implicated in inflammatory joint disease^47,48. Therefore, iNOS expression may be anti-inflammatory, inflammatory, or both. Conversely, NAC did not alter MMP9 expression, a matrix metalloproteinase and marker of both synovitis and potential cartilage damage⁴⁹. Maintained generation of NO likely indicates diminished reactive oxygen species (ROS) generation by mitochondria, as ·NO and O₂·⁻ exist in opposition^50,51. This indicates the protective effect of NAC in arthritis may be due to decreased oxidative stress but not decreased macrophage migration through MMP secretion. It may be necessary to target both ROS and MMPs for better arthritis treatment, which illustrates how JSAS may be useful in the development of arthritis therapeutics.

DISCUSSION:

This work demonstrated JSAS capabilities, specifically maintenance of living synovial tissue with intact, functionally responsive synovial architecture. There was minimal cell death out to 72 hours of culture in most incubation conditions. To evaluate the microscopic changes of synovial structure, a unique histologic measuring system was described, quantifying structural alterations of the IntL and SubL. Such changes are indicative of acute or acute-on-chronic synovitis. This quantification may be particularly beneficial when paired with cellular, mechanistic analyses as traditional classification systems rely on lymphocytic infiltrates and longer time courses⁵². There was consistency in histomorphometry and ELISA results between tissue, without need to normalize results to compare patients in the nearly fifty patients tested. In addition to reproducible histomorphometry, JSAS is capable of immunohistochemistry, flow cytometry, and ELISA. Not tested here, but certainly of utility, would be bulk qPCR, flow cytometry of synovium enzymatically digested into a single cell suspension, single cell RNA-sequencing, or spatial transcriptomics. JSAS is also extremely cost effective if the lab is trained in recruitment and sterile tissue acquisition from the operating room. Thus, majority of costs are for the reagents, standard lab equipment, and dissection tools.

JSAS does not exist in isolation. There are many existing models of arthritis and synovitis, particularly in vivo. For RA, this primarily includes the collagen-induced arthritis (CIA) model in mice and rats, which utilizes immunization against Collagen II, the main collagen component of articulating joints, to initiate disease⁵³. Transgenic mice and serum or auto-antibody transfer may also be used for induction of arthritic disease⁵³. While the CIA model recapitulates essential biologic features such as rheumatoid factor positivity and Th17 involvement, there may be significant variability in lesion location, severity, and incidence⁵³. For OA, there are large and small animal models available for both primary OA and secondary OA, the latter of which may include post-traumatic OA though surgically or injury-induced models⁵⁴. In the case of primary OA, the stifle joint is typically studied as quadrupeds do not have a knee joint. Primary OA also requires a large portion of an animal’s lifespan to develop. However, large animals such as horse and sheep may present particularly good models that mirror human disease, cartilage biology, tissue volume, and biomechanics⁵⁴. The biggest challenges remain 1) the cost-prohibitive nature of animal work to develop experiments with appropriate power and rigor, particularly for large animal models, and 2) the biologic relevance of non-human responses that require further validation, possibly in multiple animal models prior to use in humans⁵⁴. These issues are not unique to arthritis research. While variability of animal-to-animal outcomes may also be considered a limitation, this is merely consistent with human work and thus is not unique a limitation.

In comparison to animal models of arthritis, evaluation of human arthritis pathogenesis is non-invasive or minimally invasive, typically in the form of epidemiologic evaluation, imaging studies, or core needle biopsies (for example, ^55-58). Models, animal or otherwise, focusing on synovium-mediated arthritis are particularly limited⁴. Mechanistic studies in humans are further limited as this would require invasive and/or serial sampling of joint space compartments including synovium, cartilage, and synovial fluid. Thus, there is a particular need for improved models of human arthritis that JSAS can fully or partially address.

Other groups have also attempted to address the challenges of in vivo models through in vitro models such as joint-on-a-chip⁵⁹, synovium-on-a-chip^60,61, or co-culture of FLS and macrophages⁶². In addressing the challenges through simplifying the system to maximize affordability and reproducibility, one introduced weakness of these models is the loss of complexity of the functional IntL and SubL layers. No in vitro or organ-on-a-chip model yet describes formation of an IntL, which clinically appears to be essential in understanding the response of synovium to inflammation or inflammation resolution^6-8. Additionally, the macrophages utilized for in vitro models may include cells lines, bone marrow derived macrophages (BMDMs), or monocyte derived macrophages from blood sampling, but RSM cell lines do not exist, and the differences in baseline functional responses are not established between peripherally derived macrophages and RSMs. Therefore, there may be unclear clinical relevance of simplified models.

Demonstration of the clinical relevance of JSAS was accomplished with simulated (LPS) and actual infections (B. burgdorferi, S. aureus), simulated inflammatory arthritis with MCP-1^33,34, and therapeutic testing with NAC. While OA tissue was used, other tissue is also possible to obtain, including RA tissue or healthy synovium from patients undergoing tibial tubercle osteotomy (TTO). Volume of synovium from TTO patients is significantly less, but they represent an important “healthy control,” as these patients typically undergo a surgical procedure to correct patellofemoral misalignment, ideally prior to the development of OA⁶³. To determine if a tissue response is pathology-specific or is generalizable it will be necessary to obtain healthy tissue comparisons. It is possible to obtain 20-30 3 mm cores from TTO patients, compared to 50-70 from TKA patients due to limited synovial overgrowth seen in healthy joints. Regardless of the underlying pathology leading to the open joint surgery, the primary benefit of JSAS is the focus on relevant human biology, which makes it ideal for testing therapeutics in addition to mechanistic studies in infectious and inflammatory arthritis and/or synovitis. The primary difficulty of the system is the inherent complexity of obtaining human tissue, necessitating close affiliation with collaborative orthopedic surgeons.

Troubleshooting with JSAS is primarily limited to patient selection, dissection technique, and preparation of cores (1.1-1.7). Younger patients with a minimal medication list had more cellular tissue than older patients with more comorbidities. Cellular tissue capable of responding to stimuli is essential to study alterations to synovial structure. Poor sterile technique will lead to undesired contamination, thus it is essential that from the moment the tissue leaves the body to its takedown that strict sterile conditions are maintained. Contaminated culture will be visible in 24-48 hours. Yeast was the most common contaminant in initial experiments until appropriate dissection guidelines were determined. Having sterile 70% ethanol solution and sterile PBS inside the BSC to first sanitize and then rinse tools, and doing so frequently, is essential for sterility. Allowing the tissue to sit for more than 2 hours on ice prior to dissection and plating also led to poor viability outcomes. Other troubleshooting involved selection of appropriate biopsy punches. While discounted punches may be available from online vendors, blunted, inexpensive tools will rip the IntL, causing suboptimal embedding conditions. Being gentle with the tissue was also essential; squeezing too hard with forceps or vigorously mixing cores with NBF led to dissociation of IntL from SubL. As the tissue is fragile, technical replicates per condition are therefore essential, as is a time 0 for reference.

Limitations of JSAS include simplification of in vivo processes through limited experimental time courses and lack of peripheral immune system involvement, as immune infiltrates from systemic circulation are certainly important for pathogenesis of, for example, RA. In this manuscript, co-culture with osteochondral cores is not addressed, as osteochondral culture has already been discussed elsewhere^64,65. However, osteochondral cores may be co-cultured in the bottom well to evaluate synovium-bone-cartilage communication. Depending on the pore size and the concern for immunogenic responses between tissue of different human or species origin, it would be possible to use the 0.4 μm pore size for allogenic or xenogenic tissue, specifically, porcine or bovine cartilage. This would allow for cytokine mediated communication without cell-to-cell contact, especially considering the T cells and NK cells identified migrating from synovial tissue. In autogenic tissue, for example, osteochondral cores from the tibial plateau, the 5 μm pore size could be utilized to evaluate cell migration. Osteoblast and/or osteoclast co-culture, supernatant transfer experiments, and synovial fluid transfer experiments can also be considered in JSAS.

The critical actions to use JSAS are 1) maintaining tissue hydration and cold temperature from the time of tissue acquisition until prompt dissection, coring, and plating can be accomplished, 2) appropriate dissection and washing of extraneous cellular components that could be inflammatory (e.g., necrosed cells or broken cellular components), and 3) maintaining sterility throughout. While it may be tempting to obtain synovium from other joints such as the hip, this group found hip synovium (e.g., pulvinar) is more difficult to identify surgically, smaller in volume, and has different morphology than knee synovium. For reproducibility, it is therefore recommended to compare only within-joint replicates (e.g., hip-hip, knee-knee, but not hip-knee). Other joints such as shoulder were not assessed.

In conclusion, JSAS adequately addresses the challenges of biologic relevance, cost, safe tissue acquisition, and reproducibility in a human model that has high utility as a pre-clinical and mechanistic test ground for infectious and inflammatory arthritis. It specifically preserves IntL and SubL responses such as iNOS, MMP9, IL-1β, TNFα, IL-4, PDPN, sRANKL, and OPG production. Histomorphometry may be used to compare structural changes to biologic responses, tying together IntL breakdown (%IntL intact), synovitis (depth and cellularity), cell migration, and immune function. This provides clarity as to biologic responses within the joint space that are relevant to multiple disease processes.

Supplementary Material

supplementary figure 2

Supplementary Figure 2: Visualization of thresholding ranges in FIJI. In full color, there are 4 nuclear bodies seen, which are also visualized on transition to 8-bit image. If under thresholded (too much background), cells merge together (cells #3 and 4) and background is identified as cells. If over thresholded, cells are lost (cell #2). Appropriate threshold setting should reflect the original image. Size cut offs (8-150 μm2) restrict any small noise artifacts from being counted as cells

supplementary figure 1

Supplementary Figure 1: Orientation of tissue in FFPE blocks on cassette. To evaluate the histomorphometry, it is essential to embed the tissue in the proper orientation so that sections are cut through the long or longitudinal axis of the core (top). This places the IntL and SubL at opposing ends of the section. The authors prefer to embed all technical replicates into the same block, and cut three serial sections per slide to facilitate immunofluorescence co-localization (bottom).

Figure 1: Illustration of protocol steps 1.2-1.7.

All steps are performed in a sterile environment. Tissue is obtained intra-operatively and is typically not marked for orientation. It is kept on ice for no longer than 2 hours (1.2). Tissue is brought back to lab, washed on a rocker (1.3), and then examined in a BSC to identify the IntL (1.4). Necessary dissection equipment includes tools such as hemostats, iris scissors, and forceps, which may differ based on preference. Tools are sterilized prior to use in autoclave, and frequently re-sanitized frequently in 70% ethanol followed by rinsing in PBS during dissection (1.4). Multiple sterile petri dishes are used for dissection, and tissue removed during dissection is placed in a separate waste petri dish (1.4). Dissection is performed with the stromal side facing up, gently stabilizing the tissue with hemostats while small pieces are removed with iris scissors (1.5). This is continued until the tissue is trimmed to approximately 4-5 mm thick. Tissue is then ready for biopsy coring (1.6-1.7). Cores are transferred to a separate dish containing DMEM with 10% FBS for washing of debris and while awaiting transfer to the top well of JSAS.

DATA AVAILABILITY:

There is no database to share regarding this study.