A High Throughput Model of Post-Traumatic Osteoarthritis using Engineered Cartilage Tissue Analogs

Abstract

(1) Objective

A number of in vitro models of post-traumatic osteoarthritis (PTOA) have been developed to study the effect of mechanical overload on the processes that regulate cartilage degeneration. While such frameworks are critical for the identification therapeutic targets, existing technologies are limited in their throughput capacity. Here, we validate a test platform for high-throughput mechanical injury incorporating engineered cartilage.

(2) Method

We utilized a high throughput mechanical testing platform to apply injurious compression to engineered cartilage and determined their strain and strain rate dependent responses to injury. Next, we validated this response by applying the same injury conditions to cartilage explants. Finally, we conducted a pilot screen of putative PTOA therapeutic compounds.

(3) Results

Engineered cartilage response to injury was strain dependent, with a 2-fold increase in GAG loss at 75% compared to 50% strain. Extensive cell death was observed adjacent to fissures, with membrane rupture corroborated by marked increases in LDH release. Testing of established PTOA therapeutics showed that pan-caspase inhibitor (ZVF) was effective at reducing cell death, while the amphiphilic polymer (P188) and the free-radical scavenger (NAC) reduced GAG loss as compared to injury alone.

(4) Conclusions

The injury response in this engineered cartilage model replicated key features of the response from cartilage explants, validating this system for application of physiologically relevant injurious compression. This study establishes a novel tool for the discovery of mechanisms governing cartilage injury, as well as a screening platform for the identification of new molecules for the treatment of PTOA.

Introduction

The primary function of articular cartilage is as a load-bearing structure that supports and distributes the high stresses generated during normal physiological activities¹. While cartilage generally functions well over a lifetime of use, acute instances of supra-physiologic loading (e.g. accident or other traumatic event), often result in tissue damage that initiates degenerative processes within the joint. Indeed, a subset of osteoarthritis (OA), termed post-traumatic osteoarthritis (PTOA) represents the significant fraction of patients who develop OA secondary to such joint trauma. Based on the incidence of knee, hip, and ankle OA for patients with a history of joint injuries², it is estimated that up to ~6 million individuals are burdened with PTOA in the US alone. Cartilage pathology and PTOA incidence generally correlate with the intensity of the original injury; patients with ligamentous or meniscal injuries are 10-fold more likely, and those with articular fractures are 20-fold more likely to develop knee OA compared to individuals without previous joint injuries^3,4. Despite a growing understanding of the mechanical thresholds that instigate PTOA, the molecular pathogenesis and mechanisms of disease progression are not yet well understood.

To that end, a number of in vitro, ex vivo, and in vivo models of cartilage injury have been developed to explore the temporal patterns of anabolic and catabolic events that culminate in cartilage degeneration. These models serve as useful platforms in which to explore variables that regulate the extent of damage, including the impact energy, peak stress/strain, and stress/strain rate. Common markers of load induced injury include tissue swelling and fibrillation⁵, cell death at or near the injury site^6,7, and increased expression of proteases and inflammatory cytokines^7,8. Biologic mediators of PTOA act collectively to decrease chondrocyte matrix biosynthesis⁵ and instigate a loss of proteoglycans and other matrix elements^5,6,9,10. Together, these molecular and compositional changes culminate in a loss of tissue mechanical integrity^5,7.

The timeline of activation of these degenerative processes (and the controlling signaling mechanisms) is particularly important, as the different stages of response post-injury may represent opportunities for therapeutic intervention¹¹. Indeed, previous studies have focused on small molecules targeting the early events, including mechanisms that lead to cell death, release of inflammatory mediators, and proteoglycan loss. Examples of such compounds include pan-caspase inhibitors^12,13 to decrease cell death, amphiphilic surfactants^14–16 to repair disrupted cell membranes, oxidative free radical scavengers¹³ to limit early inflammatory processes, as well as growth factors^17,18 and glucocorticoids¹⁹ to increase anabolic response post-injury. In the above studies, these factors have shown varied success in reducing cell death and matrix degradation in in vitro and in vivo models of PTOA, indicating that such early pathologic changes are appropriate targets for therapeutic intervention.

To date, the selection of agents that might abrogate PTOA initiation has been based on their roles in canonical pathways involved in cell physiology and/or OA progression. Since the mechanisms of PTOA have not yet been fully elucidated, there may be other agents not previously known to play a role in PTOA that could have chondro-protective effects. In recent work, Sampson et al. showed that parathyroid hormone (clinically used to improve bone mass) administered to mice after meniscus destabilization surgery was chondro-protective (or regenerative) in that it limited hypertrophic changes after onset of instability²⁰. Wang et al. also showed that the inflammatory complement system regulated cartilage degradation in mouse models of joint instability²¹. These studies illustrate the significant role that such non-canonical pathways may play in mediating the degenerative response in situations of chronic overload; the acute injury response may similarly initiate heretofore unexplored signaling pathways.

High throughput (HT) screening enables the rapid evaluation of small molecule libraries for the discovery of novel compounds relevant to tissue development and healing without prior knowledge of the mechanism of action. Recently, Johnson et al. developed an image-based high throughput screening system to identify molecules that promoted chondrogenic differentiation of MSCs²². From the 1000s of molecules screened in that study, several “hits” were identified, with the small molecule kartogenin emerging as the most promising. Follow-up secondary in vitro assays (e.g. RT-PCR) and tertiary in vivo investigations (rodent joint instability models) illustrated that kartogenin also had a chondro-protective effect and acted by disrupting the binding of a specific transcription factor subunit to an actin associated protein. Given the non-intuitive mechanism of action, this study highlights the need for unbiased screening tools to guide molecular discovery specific to a particular disease process.

To enable such screens in the context of PTOA, we developed a high throughput in vitro mechanical injury platform that is compatible with drug screening. While in vitro models of injury using explants have been valuable in elucidating regional changes in cell viability and matrix loss, explants are not ideal for high throughput screening due to the large number of samples required and variation in the cellular and molecular stratifications found throughout the joint. Cartilage tissue engineering, which aims to mimic the biochemical and mechanical properties of native cartilage for joint repair, can generate cartilage-like analogs with which to study the pathogenesis of PTOA. Engineered cartilage can also be fabricated in a uniform manner and in large quantity, and as such are ideal for high throughput screening applications. In particular, we have studied a scaffold-less method to generate cartilage tissue analogs (CTAs) that closely mimic native cartilage both in terms of extracellular matrix composition and biomechanical properties^23–25.

Here, we adapted our high throughput mechanical testing system²⁶ to apply compressive injury to CTAs in a rapid and reproducible manner. The primary goals of this study were to determine the strain and rate dependent response of engineered cartilage to compressive injury, to evaluate the progression of degeneration, and to validate this response with respect to native articular cartilage explants treated similarly. Our findings validate the use of engineered cartilage as a surrogate for studying mechanisms of PTOA pathogenesis and introduce a new screening tool with which to identify novel compounds that can attenuate degeneration following cartilage injury.

Methods

Fabrication of Cartilage Tissue Analogs

Engineered cartilage tissue analogs (CTAs) were produced as described previously^24,25. Briefly, articular cartilage was harvested from juvenile bovine knees (2–6 months old, Research 87, MA), finely minced, and digested overnight (12–16 hours) in DMEM containing collagenase Type II (298U/mL Worthington, NJ). Tissue digests were filtered (70µm pore mesh), washed with PBS containing 200U/mL penicillin, 200µg/mL streptomycin, 5µg/mL Fungizone (PSF, Life Technologies, NY), and centrifuged at 1750 rpm for 15 minutes at 12°C (3X) until collected into a single suspension. Chondrocytes were resuspended at 5×10⁶ cells/mL in complete medium (high glucose DMEM containing 10% FBS, 100U/mL penicillin, 100µg/mL streptomycin, 2.5µg/mL Fungizone, 1% MEM Vitamin Solution (Gibco), 25mM HEPES buffer, 50µg/mL ascorbic acid²⁵). This cell suspension was plated into ultra-low adhesion (polyHEMA coated) 96 well plates (Corning, NY) at 1×10⁶ cells/well, where chondrocytes coalesced within 24 hours to form a CTA²⁵. CTAs were cultured for a minimum of 14–16 weeks in complete medium prior to injury.

Injurious Compression of CTAs and Native Tissue Explants

To determine the level of injury necessary to induce pathological changes in CTAs that mimic changes in cartilage explants, four different injurious compression protocols were applied in a single-sample manner based on previously established injury parameters^5,6,8. CTAs were subjected to either 50 or 75% strain at one of two strain rates, 10% strain/sec or 50% strain/sec, followed by a hold period for a total ramp-hold compression time of 10 seconds. Constructs were then cultured for 5 days after injury, and both CTAs and media were harvested at 12, 24, and 120 hours post-injury for evaluation of biochemical content and presence of soluble catabolic markers as described below. Based on the outcomes of single-sample injury, high throughput injury was applied to constructs at 75% strain at 50% strain/sec. For this, a custom high throughput mechanical screening device²⁶ was used. The device consisted of an aluminum housing with linear bearings to guide the vertical displacement of a loading platen which included a force-sensitive resistor (FSR) array for real-time monitoring of compressive forces during injury. In the current version of the device, 48 samples are housed in a standard 48-well plate and are compressed via PTFE indenters, with load recorded continuously during injury using a NI-DAQ board (National Instruments, USB-2665) and a custom Labview program (National Instruments, V8.6) with post-processing in MATLAB (Mathworks, R2012a). Injurious strain and strain rate were calculated based on the average height of constructs. After injury, constructs were cultured for 5 days with sample harvest at 24, 48, and 120 hours post-injury. As a positive control, constructs were treated with IL-1β (10ng/mL) for 5 days^27–29. To validate the injury response in CTAs, cartilage explants were injured in a similar manner. Full-thickness articular cartilage (chondral only) explants (4mm diameter) were harvested from the trochlear groove of juvenile bovine knees and trimmed to 3–4mm thickness, keeping the superficial layer intact. Cartilage cylinders were subjected to 75% strain at 50% strain/s using a single-sample injury protocol matching the high throughput injury of the CTAs. Cartilage cylinders were cultured for 5 days and evaluated as above in order to make comparisons between native and engineered cartilage response to injury.

Treatment with Putative Therapeutic PTOA Compounds

In a subset of studies, and immediately following injury at 75% strain applied at 50%/sec with the high throughput device, engineered cartilage was treated with one of the three following agents: (1) N-Acetyl-Cysteine (NAC, 2mm, Sigma, MO) a reactive oxygen species scavenger, (2) Z-VAD-FMK (ZVF, 100µM, Promega, WI) a pan-caspase inhibitor, or (3) Polaxamer 188 (P188, 8mg/mL, Corning, NY), an amphiphilic polymer capable of inserting into the cell membrane. Each compound was included in the culture medium for the initial 48 hours post-injury at levels previously reported to have beneficial effects in the context of cartilage injury^12–15. Harvest time points and outcome measures were the same as described above.

Biochemical and Molecular Evaluation of Injury Response

Following injury, construct wet weight and dry weight (following lyophilization) were determined. Samples were then papain digested and glycosaminoglycan (GAG) content determined using the dimethylmethylene blue assay, with chondroitin-6-sulfate as a standard, as previously described³⁰. DNA per construct was measured using the PicoGreen assay (Life Technologies, NY). Matrix content was measured per construct or normalized to DNA, and swelling ratio was calculated as the ratio of wet to dry weight at the time of harvest. Medium was assayed at all harvest time points for GAG release and lactate dehydrogenase activity (LDH; CytoTox-ONE Homogeneous Membrane Integrity Assay, Promega, WI), which is released upon disruption of the cell membrane and is a measure of cell injury.

Histological Analysis of Injury Response

For viability analysis, constructs were stained using the Live/Dead staining kit (Live/Dead Viability/Cytotoxicity Kit, Life Technologies. NY) and imaged on a Nikon Eclipse TE2000-U (excitation wavelengths: Live-420–495nm and Dead- 532–587nm) using 2× or 10× objectives. Additional samples were fixed in 4% paraformaldehyde (Affymetrix, CA), dehydrated, embedded in paraffin, and sectioned to 8µm thickness. Sections were stained for proteoglycan distribution with Alcian Blue (Rowley Biochemical Institute, MA) as previously described³¹.

Statistical Analysis

Effect of strain and strain-rate for single sample injury, differential effects of injury or IL-1 β as a function of time, and comparisons to chondral explants were assessed by two-way ANOVA with Bonferroni’s post-hoc test (p<0.05). PTOA compound effects were compared to injury alone using a single sample t-test (p<0.05). All statistical analysis was conducted using the SYSTAT13 software (v.13.00.05, San Jose, CA).

Results

Effect of Strain Magnitude and Strain Rate on CTA Injury Response

CTAs were injured in a single sample manner, with compression to 50 or 75% strain at 10 or 50% strain/sec. Representative stress versus time profiles of injured CTAs showed that both strain and strain-rate significantly increased peak stress (with values reaching 1–2 MPa) (Figure 1). Interestingly, compression of CTAs up to 75% strain resulted in multiple peaks, likely indicative of construct fracture and re-compression during loading. Construct failure at high strains was confirmed by histological analysis. At 50% strain, there was some internal fissuring of the construct and focal areas of GAG loss (Figure 2G). At the higher 75% strain level, there was obvious surface fibrillation, loss of construct shape, and widespread GAG depletion (Figure 2H). In comparison, control constructs maintained uniform GAG distribution throughout the construct (Figure 2F). Rapid application of 50% strain resulted in cell death throughout the intact construct thickness, while 75% strain primarily resulted in cell death in areas adjacent to fissures (Figure 2, C–E). GAG within the construct and released to the medium was measured 0–12, 0–24, and 24–120 hours post-injury. 75% strain applied at either strain rate reduced GAG/DNA in constructs for all time points post-injury (Figure 2A). Furthermore, injury at both 50 and 75% strain significantly increased GAG released to the medium compared to un-injured controls. Application of 75% strain resulted in an ~2-fold greater increase in GAG release compared to 50% strain applied at the same rate (Figure 2B).

Figure 1.

Injurious compression protocol. (A,B) Engineering strain and stress profiles of CTAs for the four injury protocols tested: 50% or 75% strain applied at either 10 or 50% strain/sec for a total compression time (ramp and hold) of 10 sec. Profiles are representative of constructs injured in that group (N=24/group, all CTAs fabricated with pooled chondrocytes of a single animal). (C) Strain and strain-rate both significantly influence peak stress, with 75% strain applied at 50% strain/s yielding the highest peak stress.

Figure 2.

Effect of strain and strain rate on matrix retention and loss following injury. (A) GAG content normalized to DNA content within CTAs (N=3/group, all CTAs fabricated with pooled chondrocytes of a single animal) showed consistent loss of matrix following injury applied to 75% strain at 10 and 50% strain/sec. (B) GAG released to the medium (N=3/group, all CTAs fabricated with pooled chondrocytes of a single animal) mirrored that of the construct, but with significant loss observed in a strain-dependent manner at both 50 and 75% strain, regardless of the strain-rate during injury. (C–E) Live/Dead staining (green: viable; red: non-viable) 24 hours post-injury for the highest strain rate illustrated that 50% strain resulted in focal regions of cell death with internal fissuring while control constructs contained viable cells throughout. 75% strain caused more extensive cell death superficially and adjacent to large and full-depth fissures (N=2/group). (D–F) Alcian blue staining for proteoglycans showed control constructs with well distributed matrix, that 50% strain caused internal fissuring with local matrix loss (black arrows), and that 75% strain caused wide-spread matrix damage with fainter staining for proteoglycans throughout the construct (N=2/group). *p vs. control and ^#p vs. 50% strain for a respective time point; p<0.001 for * and # symbols alone.

High Throughput Injury of Engineered Cartilage

Based upon the outcomes of single sample injury of CTAs, 75% strain applied at 50% strain/sec was chosen for the high throughput application of compressive injury. Injury was carried out using the custom high throughput mechanical injury device, shown in Figure 3A. While CTAs had slight variations in height due to their free form assembly and growth during pre-culture, an applied target strain of 75% resulted in applied strains ranging from 59% to 99% in individual CTAs, with a mean strain of 78+/−10% for a full 48-well plate of constructs (Figure 3B). Simultaneous compression resulted in peak loads in CTAs comparable to single-sample compression; a 3D graphical representation of the real time peak voltages (peak load) is provided in Figure 3C, showing similarity across constructs. GAG released from high throughput constructs was ~2 to 3-fold higher than un-injured controls for all time points (Figure 4A), a response similar to that evoked by the single sample injury. Interestingly, while IL-1 treatment initially caused GAG release at a similar magnitude as injury, by 120 hours of continuous treatment, GAG release was ~9-fold higher than un-injured controls and 4 to 5-fold greater than injured samples. Evaluation of LDH in the medium indicated that injury resulted in dramatic loss in viability over the short term (Figure 4B). Conversely, IL-1 treatment did not disrupt membrane integrity and so resulted in little LDH release relative to baseline levels. Tissue swelling was calculated as a measure of degeneration, as swelling is observed clinically in early stages of osteoarthritis. While injured constructs swelled significantly within 24 hours, control constructs did not change (Figure 4C). In contrast, treatment with IL-1 only affected swelling after 5 days due to continued matrix degradation.

Figure 3.

HTMS device for applying compressive injuries to CTAs. (A) The HTMS injury device consisted of an aluminum frame with linear bearings to guide the vertical motion of the sensor loading platen. The sensor adhered to the underside of the loading platen comes into contact with a well plate assembly consisting of PTFE indenters aligned with a standard 48 well plate containing engineered constructs. Average sample height was measured prior to testing, with a target injurious compression of 75% strain at 50% strain/sec applied to constructs. (B) Example distribution of sample heights and applied strains for each construct (N=48/plate). Average strain was 78+/−10% strain for the population. (C) Peak voltage recordings in each well showed the uniformity of peak load responses in engineered cartilage during compressive injury.

Figure 4.

Release of ECM and cellular enzymes and alterations in CTA properties following high throughput mechanical injury. (A) Injury of engineered cartilage significantly increased GAG release to the medium in a manner similar to that of IL-1 treatment alone for the first 48 hours; after 48 hours, IL-1 causes a 4 to 5-fold higher level of GAG release from the construct compared to injury alone (N=8 media collections per sample combined into 4 aliquots, Sol-GAG is an average value per aliquot/group). (B) LDH release (a measure of cell viability) indicated that injury caused a large increase in chondrocyte membrane disruption in the first 24 hours after injury, while IL-1 resulted in little membrane damage with continuous exposure for 5 days (N=8 media collections per sample combined into 4 aliquots/group, LDH is an average value per aliquot). (C) The swelling ratio (calculated as the ratio of wet to dry weight) indicated that that injury caused gross tissue damage and swelling within 24 hours post-injury. In contrast, IL-1 treatment resulted in increased swelling only after 120 hours of treatment (N=4 samples/group; all CTAs fabricated with pooled chondrocytes of a single animal). *p vs. control and ^#p vs. IL-1 for a respective time point; p<0.001 for * and # symbols alone.

Validation of Engineered Cartilage as an Analog for Injury of Native Tissue

To validate the injury response of CTAs, explants were subjected to the same injury protocol and outcomes. A representative peak stress profile for a cartilage explant shows multiple peaks, similar to that observed with the engineered cartilage, with the first peak concurrent with gross tissue failure, followed by a second or third peak due to the further compression of the fragments (Figure 5A, inset). The first peak in stress occurred at ~16 MPa for cartilage explants, similar to previously reported values for explants subject to injurious compression at high strains or strain-rates^5,32. GAG release from explants and CTAs 24 hours post-injury showed similar loss in matrix content per construct (Figure 5A). Similarly, LDH release showed comparable findings, with injured CTAs and explants both showing evidence of cell membrane damage compared to un-injured controls, though LDH release from explants was ~2-fold greater than from CTAs (Figure 5B).

Figure 5.

Comparison of native and engineered cartilage injury response. Bovine cartilage explants were compressed to 75% strain at 50%/s. (A-inset) Representative peak stress profile of explants during injurious compression showing a first peak concurrent with gross tissue failure, followed by a second peak resulting from further compression of the fragments. Average first peak stress was 16.8+/−4.3 MPa (N=12, all explants harvested from a single animal). Soluble factors released to the medium in the first 24 hours showed that (A) GAG release was comparable between injured explants and CTAs and was significantly greater than un-injured samples (N=4 for explants, and N=8 media collections per sample combined into 4 aliquots/group). (B) In contrast, LDH release was approximately 2-fold greater in explants compared to CTAs following injury. (N=4 for explants, and N=8 media collections per sample combined into 4 aliquots/group). p<0.001 for * vs control within tissue type and # vs CTAs.

Response of Engineered Cartilage to Putative PTOA Therapeutic Compounds

As a secondary validation of this high throughput injury system, we subsequently screened several putative therapeutic PTOA compounds. In the initial 24 hours, ZVF treatment significantly reduced LDH release by 20% compared to injury alone; however NAC and P188 did not alter membrane disruption (Figure 6A). By 120 hours post-injury, however, both NAC and P188 decreased GAG loss (by 18 and 20% respectively) compared to injury alone, although neither agent restored GAG content to control levels (Figure 6B). Despite increasing initial cell viability after injury, ZVF treatment did not alter GAG content in injured samples.

Figure 6.

Effect of putative PTOA therapeutics on matrix content and cell death after injury. (A) LDH release 24 hours post-injury was reduced by ~30% with ZVF treatment while NAC and P188 had no effect compared to injury alone (N=8 media collections per sample combined into 4 aliquots/group). (B) GAG content in the construct 120 hours post-injury showed that NAC and P188 treatment resulted in retention of ~20% more GAG compared to injury alone (N=4/group). Dashed red line demarcates results from two separate experiments (CTAs for each experiment fabricated with pooled chondrocytes of a single animal) evaluating the effects of NAC, ZVF, and P188.

Discussion

Traumatic joint injury initiates a cascade of catabolic and anabolic processes, the imbalance of which often results in further cartilage degeneration. However, the pathways that underlie these irreversible changes remain poorly understood. As such, in order to conduct an efficient and unbiased evaluation of molecules that may modulate PTOA biologic processes, high throughput screening would be a valuable tool to evaluate compound libraries after injury. In this study, we utilized engineered cartilage analogs (CTAs) in conjunction with a high throughput mechanical injury device to develop a platform for studying PTOA pathology and to enable the discovery of potential therapeutics.

Primary markers of cartilage damage following traumatic injury include matrix disruptions, GAG loss from the matrix, and cell death. Using CTAs as an in vitro cartilage surrogate for studying PTOA, our objective was to define the thresholds for inducing such a response, to determine the uniformity of response using a high throughput device, and to benchmark the response against that of native cartilage. To define thresholds for injury, single sample compression was applied to constructs at strains (50% and 75%) and strain rates (10% and 50% strain/sec) previously explored^5,6,8. In explants, while strains larger than 50% cause permanent deformation and surface fibrillation (stresses >15MPa), strains greater than 80% (stresses >20MPa) cause deep fibrillation and complete destruction of matrix integrity³². In CTAs cultured for up to 16 weeks, whose properties approach that of native tissue²⁵, application of 75% strain resulted in widespread fissuring, with a 2-fold increase in release of GAG to the medium compared to 50% strain. The greater release of GAG observed at this higher strain may be due to the increased surface area for diffusion. DiMicco et al. observed with injury of osteochondral explants an initially high rate of GAG release not blocked by MMP or biosynthesis inhibitors, indicating that this early release (≤4 days post-injury) likely consisted of diffusion of larger proteoglycan molecules out of the tissue rather than enzymatically-cleaved fragments, as is observed at later time points⁹.

In our study, for both levels of injury, extensive cell death was observed adjacent to surface fissures and internal cavities. In cartilage explant studies, the extent and depth of cell death has been reported to be both strain and strain-rate dependent^6,7,33,34. In addition, similar to our observations, extensive loss of cell viability occurs along fissure lines/regions both in whole joint³⁵ and osteochondral explant^6,36 models with cell viability increasing with distance from the fissure line. However, with time, cell death expands to these non-fissure regions, which may suggest two mechanisms by which cell viability decreases with injury. Under high loading rates, cells may not be able to “recruit” sufficient membrane components in order to deform under compression, and in areas of cartilage with fissuring cells experience high strains and these cells rupture³⁷. The expansion of the region of cell death to non-fissure regions with time may be the result of diffusible, soluble factors that induce apoptosis and contribute to the propagation of the injury response^35,36,38.

Upon determining that 75% strain at 50% strain/sec induced a degenerative response in CTAs, we next used our high throughput device to apply consistent compressive injury to up to 48 samples simultaneously. While there were small variations in applied strain due to differences in construct height, the response was sufficiently uniform so as to provoke a ~2–3 fold increase in GAG release compared to un-injured controls, comparable to the single sample response. Monitoring LDH release as a quantitative measure of cell death similarly confirmed that cell membrane damage is a repeatable effect of compressive injury in this engineered model of PTOA.

We subsequently validated the CTA response against that of native cartilage. Peak stresses during compressive injury of native tissue were ~20-fold higher than CTAs. This difference could explain the 2-fold greater increase in LDH release as compared to CTAs in the first 24 hours. Indeed, strain rate and peak stress dependent increases in LDH release have been observed in cartilage explants^39,40. Additional factors which may also influence the extent of cell death include matrix composition and organization⁴¹. In our study, GAG release was comparable between explants and CTAs, with a 4-fold increase in GAG release to the medium compared to un-injured controls within the first 24 hours. Previous work has similarly reported an ~2-fold increase in GAG release in the first 24 hours from explants subjected to injury at fracture levels (e.g. 50% strain at 100% strain/sec^9,10 or high strain rates of 50% or 70% strain/sec⁶). Given that the CTA can be injured at any point in its maturation, from the cell-rich, matrix-poor more fetal-like state, through to the mature, matrix-rich, cell poor adult-like state²⁵, a range of studies may be performed to determine how injury and tissue maturation state interact.

Using our CTAs, we observed a consistent and marked effect of load-induced injury; however, in PTOA, degeneration is also potentiated by the presence of inflammatory cytokines, such as IL-1β, which may differentially govern the chondrocyte response in our in vitro system. While mechanical injury caused extensive cell membrane damage as measured by LDH release, IL-1 had a minimal effect. In contrast, while IL-1 and injury initially increased GAG release to similar levels, by 120 hours of exposure to IL-1, GAG release was 3-fold higher than injury alone. IL-1 is known to cause matrix degradation via increased NF-κB activation, expression of proteases (e.g. MMP and ADAMTS), and other pro-inflammatory molecules (e.g. NO and COX-2)^42,43. Such temporal patterns of matrix disruption may also explain why injury caused rapid construct swelling (mechanical disruption of the nascent collagen network), while this swelling response took longer with IL-1 mediated degradation (sustained enzymatic cleavage). These observations are consistent with reports of injury-induced increases in water content of explants in a strain dependent manner^5,32, as well as of chondrocyte-seeded agarose hydrogels following a crush injury⁴¹.

Finally to determine whether this CTA model is useful for screening new PTOA therapeutics in a high throughput manner, bioactive molecules previously reported to reduce cell death and proteoglycan loss post-injury were evaluated. Application of these compounds to CTAs resulted in early protection against loss of viability (ZVF) and late protection against matrix loss (P188 and NAC) after injury. ZVF has been observed to increase cell viability by 15–20% with 48 hours of treatment following compression of explants (30% strain at 0.6s^{−1 12} or impact at 7J/cm^{2 13}). Although NAC has likewise been shown to increase viability by ~30% following injury¹³, here we found no effect on reducing membrane damage. However, NAC was effective at reducing GAG loss, consistent with observations of a ~20% reduction in GAG loss from chondral explants following injury (7J/cm^{2 13}). While our findings also showed that P188 improved GAG retention, literature findings have been variable, with some studies noting a ~20% increase in cell viability¹⁴ with minimal changes in GAG loss¹⁵. One possible explanation is that P188 insertion into the membrane may prevent the expulsion of intracellular contents which in turn may preserve local ion concentration gradients⁴⁴ to allow living cells to function normally. It is important to note that although these compounds did have acute effects on matrix retention and cell viability, none were able to return constructs to control levels, highlighting the importance of continuing therapeutic discovery.

Taken together, these studies illustrate that injurious compression of CTAs replicates key markers of the injury response in native cartilage explants, validating this approach as a model system for studying the processes that govern cartilage degeneration in PTOA. While our current model focuses on mimicking articular fracture, this system can be adapted for applying insults that do not produce structural damage to mimic subtle injury scenarios and also for cyclic overloading injuries. In addition, our testing platform, along with the ability to form large numbers of these cartilage analogs in a micro-scale format, sets the stage for high throughput screening of large chemical libraries to more rapidly identify therapeutics to attenuate progressive degenerative joint changes after injury.