We developed an in vitro platform allowing to recapitulate loading-induced self-heating in articular cartilage and study corresponding cell response. Our system consisted of two essential components: (i) a poro-viscoelastic scaffold, with comparable dissipative capacity and equilibrium compressive stiffness to cartilage, seeded with human chondro-progenitor cells (ECPs); (ii) a modular bioreactor to independently control applied mechanical loading, temperature increase, as well as gas concentration and humidity levels during stimulation. We have used ECPs because they provide an off-the-shelf cell source and have potential chondro-differentiation capability and reportedly they can produce cartilage matrix in vivo (Studer et al., 2017). By using this cell source, we were able to have a standardized platform for all of our experimentation with the high quality of cells and full safety features for experimental models (Laurent et al., 2020; Laurent et al., 2021).

As a 3D support for cells, a fatigue-resistant hydrogel (Figure 2a, b) was developed by combination of flow-dependent and flow-independent dissipation sources to recapitulate cartilage viscoelastic properties. This was accomplished by a hybridly crosslinked (combination of covalent and physical bonds) pHEMA network and a poorly permeable macrostructure (Figure 2—figure supplement 1). Rearrangement of polymeric chains, dissociation of reversible bonds, and fluid frictional drag mechanisms contributed to viscoelastic behavior of the developed hydrogel as described in detail elsewhere (Nasrollahzadeh et al., 2019). The stress-strain curve under cyclic loading was minimally varied showing that the developed hydrogel can preserve applied mechanical cues during multiple physiological stimulations. The poro-viscoelastic scaffold was also functionalized by RGD peptides (Figure 2—figure supplement 2) to enhance cell attachment and cell-matrix interactions during loading. The X-ray photoelectron spectroscopy (XPS) survey scan recorded on pure (non-modified) and RGD functionalized samples (Figure 2c), confirmed the effectiveness of the RGD grafting process. As the pure pHEMA hydrogel does not contain any nitrogen atom in its molecular structure, no signal was observed in respective range for pure samples contrary to RGD functionalized samples with a peak for nitrogen at 402 eV.

Figure 2 with 7 supplements see all

Download asset Open asset

In addition, cell proliferation and DNA content (Figure 2—figure supplement 3) of the functionalized constructs were significantly higher than the pure samples without RGD motifs. We could minimize the destructive effect of the bio-conjugation process on the crosslinked network and could maintain the dissipative capacity and equilibrium stiffness of the functionalized hydrogels in the range of cartilage tissue (Mak, 1986; Mow et al., 1992; Abdel-Sayed et al., 2014a; Abdel-Sayed et al., 2014b) after modification (Figure 2d). While we could not observe a noticeable change on the surface morphology of the polymeric scaffolds after functionalization process, the SEM images showed that cells could spread and attach to multiple binding points on the pores of hydrogels with RGD motifs (Figure 2—figure supplement 3). We also observed more detached cells in non-modified hydrogels without endogenous binding sites for cells when cyclic compression was combined with temperature rise (Figure 2—figure supplement 4).

Apart from tailored biomaterial’s properties, reliable control over applied stimuli are required to understand the effects of loading-induced self-heating on chondrogenesis. Specifically, we need to simulate the cells thermal and/or mechanical environment during joint loading. We custom designed our bioreactor system as shown in Figure 2f, employed an adjustable loading mechanism, and applied a convection dominant heat transfer system inside the wells to ensure consistent stimulation of samples. The performance of the developed bioreactor in different modes of action relevant to our study was then validated (Figure 2—figure supplement 5). Briefly, we confirmed that samples in different wells receive similar external stimuli according to the defined operation protocol. The controlled culture variables were verified by external calibrators. The temperature increase was modeled by Equation 1 and its parameters were estimated based on reported in vivo data during jogging (Becher et al., 2008) using the least square method.

The optimally fitted temperature evolution function (A = 31.62°C, B = 7.99°C, and c = 0.023 s^–1) predicted around 7°C temperature rise after 2 hr (Figure 2e) with a good correlation coefficient (R = 0.95). We controlled the evolution of the culture temperature inside the wells based on the model prediction by an external heat source to simulate intra-articular knee temperature during joint loading. Given the volume of the culture medium in each well, the size of the sample and the system boundary conditions, the accumulated lost energy from the dissipative hydrogel negligibly changed the culture medium temperature (<0.5°C). Our heat transfer simulation confirmed this observation when the well interface temperature was kept constant at 32°C (Figure 2—figure supplement 6). However, in an unrealistic adiabatic condition, our numerical simulation showed a temperature rise (Figure 2—figure supplement 7) due to hydrogel’s dissipative capacity. The isolated temperature and loading control strategy in our bioreactor design allowed us to decompose direct and indirect effects of mechanical loading on cell responses during self-heating of cartilage.

We first evaluated the viability of human chondro-progenitor cells (origin and characterization described elsewhere; Darwiche et al., 2012) within the functionalized hydrogels at two different culture temperatures. The cell-seeded hydrogels were cultured for 12 days at knee intra-articular temperature at rest (32.5°C) and the conventional in vitro condition (37°C) corresponding to the core body temperature. We confirmed stable attachment, high viability, and normal proliferation of distributed cells within the RGD functionalized hydrogels (Figure 3a-c). Our results indicated that cells metabolic activity, the amount of extracted DNA and RNA (Figure 3—figure supplement 1a) were slightly higher at 32.5°C compared to 37°C. These data suggest that adjusting the culture temperature at 32–33°C suits human chondro-progenitor cells metabolism which is in agreement with the reported results for human and porcine chondrocytes in monolayer and pellet cultures (Ito et al., 2015; Ito et al., 2014).

Figure 3 with 2 supplements see all

Download asset Open asset

Figure 3—source data 1

Fold change of samples in different studied groups.

https://cdn.elifesciences.org/articles/72068/elife-72068-fig3-data1-v1.xlsx

Download elife-72068-fig3-data1-v1.xlsx

To understand the influence of temperature and mechanical load on biological response, gene expression data was compared by applying coupled or isolated signals (Figure 3—figure supplement 2). Our findings demonstrated that the loading-induced self-heating (simultaneous temperature evolution and mechanical loading) positively enhances chondrogenic expression of cells and outperforms mechanical or thermal stimulus alone (Figure 3d-i). Importantly, we obtained a boosted effect on the expression of Sox9, as shown in Figure 3d, following biomimetic thermo-mechanical stimulation designated as Thermo-Mech (32.5–39°C). Sox9 is a transcription factor and its role as one of the key regulators of chondrogenic differentiation processes is well established in the literature (Akiyama et al., 2002). Upregulation of Sox9 gene under thermo-mechanical stimulation was significantly higher than the control, mechanical, and thermal groups. In parallel, Twist-related protein 1 (Twist1), which is known as an inhibitor of chondrogenesis (Gu et al., 2012), exhibited a significantly downregulated expression following the thermo-mechanical stimulation (Figure 3f). Highest expression was also observed for functional chondrogenic markers, Col2a1, Agc1, and COMP, when cells received the combined physical cues. Col1a1 expression (Figure 3g) was not significantly different between stimulated and control groups. The obtained relative fold-change ratio for Col2a1/Col1a1 (Figure 3—figure supplement 1b) confirms enhanced chondrogenic differentiation capacity regardless of the absolute level of Col2 and Col1 genes. Collectively, the applied thermo-mechanical stimulus provided a chondro-inductive signal by simulating the cartilage native environment under loading.

While still a promoter of chondrogenesis, isolated mechanical or thermal stimuli were not as effective as thermo-mechanical stimulus combined. Notably, application of intermittent temperature increases (from 32.5°C to 39°C during 1.5 hr) on cells significantly upregulated expression of Sox9, Col2a, COMP and downregulated Twist1. Unlike traditional mechanobiological in vitro models with unrealistic culture temperature of 37°C, we evaluated the effect of steady-state incubation temperature on chondrogenic cellular responses. Compared to the control group, a significant fold increase was detected on expression of COMP by loading at 32.5°C. An enhanced expression of Col2a was also observed following mechanical stimulation without temperature variation. A statistically significant difference in fold increase of Sox9 was obtained between stimulated samples at 37°C and 32.5°C. This implies that chondrogenic cells respond better to compressive loading at higher temperatures which supports the common mode of action in most bioreactors for cartilage tissue engineering. Yet, simulation of loading-induced self-heating condition better promoted chondrogenic markers compared to mechanical stimulus at constant temperatures (32.5°C or 37°C). This indicates that cells could sense the transient temperature rise. The transient effect should therefore be considered as a potential regulatory cue besides external stress.

Knowing that oscillation of intracellular calcium is an important regulator of chondrogenesis, the role of Ca²⁺ signaling in thermo-mechanotransduction process was evaluated. We also examined the contribution of multimodal TRPV4 ion channels to this process as a potent Ca²⁺ modulator (Figure 4a). We have measured time-dependent gene expression of TRPV4 channels in the scaffolds and confirmed their functional expression on human chondro-progenitor cells over our study period (Figure 4—figure supplement 1). When cells cultured in scaffolds without any stimulation at 32.5°C, we observed downregulation of the TRPV4 genes at days 6 and 10 compared to day 2. However, a significant upregulation in expression of TRPV4 gene was detected by application of biomimetic temperature evolution, mechanical loading, and combination thereof. Immunostaining of cells using a specific antibody confirmed the presence of TRPV4 channels on cells (Figure 4b) seeded in functionalized hydrogels. The functioning of TRPV4 was confirmed by live florescence imaging of cells following administration of 10 nM TRPV4 agonist, GSK1016790 (GSK101). We could detect multiple dynamic peaks after GSK101 delivery (following 50-70 s baseline recording) when analyzing intracellular calcium signals as shown in Figure 4c for three representative cells (see Video 1 for the complete response of a group of cells). We then verified significant inhibition of TRPV4-mediated calcium signaling following 10 nM GSK101 delivery (Video 2) by using 10 µM TRPV4 antagonist (GSK205) in culture medium 1 hr before imaging. Moreover, full chelation of extracellular calcium sources was obtained by employing 2 mM EGTA in the culture medium which completely diminished cytosolic calcium variation. We also evaluated the cell viability in presence of effective quantities of employed chemicals (DMSO as carrier, EGTA, GSK101 and GSK205 as agonist/antagonist) in our pathway study. Results of the assay (Figure 4d) for all tested conditions showed more than 85% viability at day 8 after using chemicals in culture medium for 4 hr during 2 different days.

Figure 4 with 2 supplements see all

Download asset Open asset

Figure 4—source data 1

Fold change of samples in different groups of thermo-mechanotransduction pathways study.

https://cdn.elifesciences.org/articles/72068/elife-72068-fig4-data1-v1.xlsx

Download elife-72068-fig4-data1-v1.xlsx

Video 1

Download asset

Video 2

Download asset

The inhibition of TRPV4 channels with 10 µm GSK205 diminished the elevated expression of the main chondrogenic markers when cells received the combined physical cues. As shown in Figure 4e-h, a significant downregulation for expression of Col2a and Agc1 genes was obtained in stimulated samples with TRPV4 antagonist. We also observed a slight decrease in Sox9 and Col1a expression with blocked TRPV4 channels. Inhibiting TRPV4 activity in free-swelling samples did not affect relative expression of the Col2a, Sox9, Col1a, and COMP (Figure 4i). In contrast, we could detect a significant downregulation of Twist1 in control samples with GSK205 (Figure 4j). Consequently, further downregulation was obtained for Twist1 gene by thermo-mechanical stimulation when TRPV4 activity was inhibited. Our results also indicated that TRPV4 acts as signal integrator for thermal and mechanical stimuli (Figure 4—figure supplement 2). By comparing the Col2a and Sox9 gene expression for either cue alone, with and without GSK205, one can observe reduced levels of these markers when TRPV4 is inhibited. The downregulation in blocked TRPV4 channels condition is statistically significant for Sox9 only in case of mechanical loading and for Col2a only in case of thermal stimulus.

Without application of external physical stimulus, we could not observe a significant change on relative gene expression of cells in presence of 2 mM EGTA. Similarly, the promoted expression of chondrogenic markers following thermo-mechanical stimulation was almost abrogated without external Ca²⁺ sources. In particular, we measured a significant reduction in expression of Col2a, COMP, and Agc1 genes and a significant increase in expression of Twist 1 (Figure 4e, f, i, j). The average Sox9 expression (Figure 4g) was also decreased; however, the downregulation was not statistically significant in the absence of external Ca²⁺. Taken together, our results suggest that TRPV4 mediates thermo-mechanotransduction processes and the Ca2⁺ signaling participates more in modulation of functional ECM markers (collagen type 2, cartilage oligomeric matrix protein, and aggrecan genes) than Sox9 transcription factor.

As already well established, the biophysical cues significantly influence the cellular responses in musculoskeletal tissues (Ingber, 2003; Guilak et al., 2014; Humphrey et al., 2014; Jaalouk and Lammerding, 2009). In a mechanistic view point, structural and material properties of ECM microenvironment are acting in association with external stresses to generate physical cues for cells (Humphrey et al., 2014; Panadero et al., 2016; Goetzke et al., 2018). The cartilage dissipative capacity originates from intrinsic matrix viscoelasticity and solid-fluid interactions under an applied loading. These dissipative sources are associated with the matrix deformation, hydrostatic pressure, and interstitial fluid flow around cells which were previously considered as mechanobiological regulators (Mow et al., 1999; O’Conor et al., 2013; Salinas et al., 2018). Cartilage dissipation, therefore, could be a functional property for chondrogenesis overarching all solid- and fluid-related mechanical signals (Abdel-Sayed et al., 2014a; Nasrollahzadeh et al., 2019). In parallel, energy dissipation in avascular cartilage during joint loading generates heat and causes the temperature evolution over time (Abdel-Sayed et al., 2014b). The absolute temperature rise in cartilage microenvironment also depends on the joint boundary condition (e.g. the ambient temperature and heat flux by surroundings tissues). Regardless of the heat source(s) and of the complex heat transfer mechanisms inside articular cavity, the varied temperature during joint loading could affect chondrocyte biological responses. To simulate such conditions and study cartilage thermo-mechanobiology, we developed a fatigue-resistant dissipative hydrogel for 3D culture of cells and a modular bioreactor to apply respective biophysical cues in vitro as there was not an available platform for this purpose elsewhere.

The developed modular bioreactor is necessary for evaluating the role of the temperature rise during mechanical loading in a laboratory design. Yet, the use of human articular cartilage explants instead of cell-seeded constructs in our customized bioreactor could better reproduce the in vivo thermo-mechanical environment. However, the difficulty in obtaining healthy human articular cartilage with similar size and physiological state was the main reason for which we developed a reproducible model and used cell-laden hydrogels. Indeed, this system has its own limitation and the spread morphology of the cells in our dissipative porous hydrogels is one of them. The ideal construct would be a mechanically capable scaffold that provides an environment for moderate cells adhesion and fairly maintains round morphology of cells. This might be realized by cells encapsulation into an adhesive soft hydrogel (Karami et al., 2021) with endogenous binding sites for cells which is firmly integrated in a stiff and porous 3D support with load-bearing capacity (Liao et al., 2013; Boere et al., 2014). The limitation of our porous dissipative hydrogel for this method of construct preparation is its intentional low permeability (order of E^–14 m²) to induce frictional drag dissipation. The permeability can be increased by enlarging the size of the pores and providing full interconnectivity within the porous structure. Nonetheless, recapitulating cartilage dissipation might not be achievable in this condition. The focus of current research was studying the effect of loading-induced temperature evolution on cells behavior by a reasonable and reproducible in vitro model. Given the fact that we employed identically prepared cell-scaffolds constructs for all of our experimentation, we assumed that the isolated role of externally applied stimuli on cells chondrogenic response could be evaluated irrespective of cells morphology.

The reciprocal relationship between cells and ECM governs cartilage hemostasis as chondrocytes receive physical cues and regulate tissue turnover to maintain its functional characteristics (e.g. dissipation). An altered gene expression is generally considered as the first indicator for cells responsiveness to conveyed cues which could ultimately lead to long-term matrix deposition depending on other contributing factors. Collagen type II (Col2a), aggrecan (Agc1), and SRY-related HMG-box gene 9 (Sox9) genes are the most utilized chondrogenic markers at the molecular level for mechanobiological studies (O’Conor et al., 2013; Anderson and Johnstone, 2017). Not only does Sox9 contribute to early chondrogenesis and matrix synthesis (Akiyama et al., 2002), but it also inhibits chondrocyte terminal differentiation (Leung et al., 2011). Col2a and Agc1 are anabolic markers of main ECM components in cartilage tissue and COMP gene controls synthesis of the linkage protein with multiple binding possibilities to integrate and stabilize the ECM (Koelling, 2006). Twist1 is known to hinder chondrogenesis by direct inhibition of Sox9 as the master regulator of the process (Gu et al., 2012). Our customized model system enabled us to show that chondrogenic cells immediately responded to biomimetic thermo-mechanical cues by altering their gene expression and that cells are able to sense the transient temperature rise. We evaluated cellular reactions after three rounds of stimulation to consider their probable desensitization and estimate a robust transcriptional response. However, given the transient nature of the chondrogenic markers to an applied stimulation (Scholtes et al., 2018), analysis of results at different time points could help to better understand the gene expression profile. Additionally, it still remains to be determined to what extent the promoted chondrogenic gene expression could enhance downstream protein expression and functional properties.

The expression of chondrogenic markers was enhanced by the applied physical stimulus in all groups except for Agc1 and COMP in one group (Figure 3). The obtained gene expression data for Col2a1, Agc1, Col1a, Sox9, Twist1, and COMP clearly show that thermo-mechanical stimulus generates more potent signaling to elicit cell responses compared to either cue alone. Moreover, Col2a1/Col1a relative fold-change ratio confirmed that combined biomimetic stimuli enhanced progenitor cell chondrogenic capacity. Yet, prolonged culture of ECP cells with growth factor supplements (at least 21 days) was needed to reach mature chondrogenic differentiation state where the absolute expression of Col2 could exceed the absolute expression of Col1 as reported by other studies with this cell line (Studer et al., 2017; Laurent et al., 2021). One compelling aspect of this finding is that employing bioreactors which are capable of controlling temperature evolution during dynamic loading should be preferred for tissue engineering applications. Indeed, chondro-inductivity can be enhanced by the condition which simulates the in vivo cartilage more similarly as shown by other multi-functional bioreactors (O’Conor et al., 2013; Schätti et al., 2011). For instance, simulating complex physiological loading conditions in articulating joints by combined compression, sliding, and shear was shown to be more effective than a single loading mode (Schätti et al., 2011). In addition, combining hypoxia culture with multi-directional loading performed better than normaxia (Wernike et al., 2008) as low oxygen level could direct chondrogenesis. Accordingly, investigating controlled oxygen tension combined with thermo-mechanical stimulation might shed more light on the in vivo chondrogenic process.

Dynamic temperature evolution also had promising impact on transcription of chondrogenic markers in the current study where applied thermal stimulus outperforms pure mechanical loading. Significant upregulation in Col2a1, COMP, and Sox9 expression as well as downregulation of Twist1 gene clearly indicate enhanced cellular responses to isolated self-heating cue (temperature rise without loading). In line with our findings, available evidence in the literature suggests that proper heat stimulation can enhance and accelerate chondrogenic differentiation processes (Tonomura et al., 2008). Over stimulation by high temperature, however, reduces the cell viability and inhibits cartilage matrix synthesis (Ito et al., 2015). Therefore, it is important to simulate the dynamics of thermal environment in vivo and optimize the influence of thermal stimuli by defining a biomimetic intermittent temperature rise. Considering clinical translation, it is encouraging to have chondro-inductive cues only with application of careful heating to simulate joint thermal environment during physical activity. Such stimulus can be simply delivered to patients in need by development of a customized heat therapy device for joint disease. It remains to elucidate if this argument is true for osteoarthritic cartilage as damaged chondrocytes may respond differently to biophysical cues as reported recently (Agarwal et al., 2021).

Understanding the translation mechanisms of external stimuli and their interaction with molecular signaling pathways is essential for development of therapeutic products to modulate the target regulators. It is well established that TRPV4 is a key mediator in transduction of mechanical signals in chondrocytes which is necessary for maintenance of cartilage tissue (O’Conor et al., 2014; White et al., 2016; Agarwal et al., 2021). In parallel, the reported activation temperature around 30°C for TRPV4 (White et al., 2016; Güler et al., 2002) makes this channel a potential mediator of loading-induced self-heating in cartilage. Our study revealed that TRPV4 is a potent modulator for expression of Col2a and Agc1 genes in response to thermo-mechanical cues. These results are in agreement with the data reported for post-transcriptional markers determined by protein expression and functional properties of cell-encapsulated hydrogels under mechanical stimulation (O’Conor et al., 2014; Gan et al., 2018). The obtained gene expression profile of Sox9 in this study suggested a partial contribution of TRPV4 to its transcription in response to thermo-mechanical cues. This means that there are parallel mechanisms for sensing the applied stimuli and eliciting Sox9 enhancement which could be other transmembrane channels, primary cilium or integrin (O’Conor et al., 2013; Panadero et al., 2016; Lv et al., 2018; Krupkova et al., 2017). Integrin-mediated pathways arising from cell-matrix interaction are mechanisms by which cells may sense the mechanical cues in the surrounding microenvironment (Humphrey et al., 2014; Loeser, 2014). Since RGD motifs were conjugated onto the pores of the developed hydrogels, a stable and effective cell-matrix interaction during loading can be expected. According to our results, significant upregulation of Sox9 in the mechanically loaded samples with and without GSK205 (Figure 4—figure supplement 2) could imply the contribution of an integrin pathway. Of note, previous studies also showed that integrin could regulate activation of ion channels in chondrocytes and cell-matrix interactions could modulate calcium signaling dynamics in response to physical cues (Degala et al., 2011; Han et al., 2012; Jablonski et al., 2014). We could not find any direct thermo and/or mechano-biological study in the literature reporting even partial effect of TRPV4 blocking on Sox9 expression in chondrogenic cells. However, in agreement with our findings, in vivo and in vitro studies have shown that pharmacological activation/inhibition of TRPV4 channels could mediate Sox9 expression (Atobe et al., 2019; Muramatsu et al., 2007).

TRPV4 activity can modulate intracellular calcium transients and literature findings indicate that a well-adjusted TRPV4-mediated Ca²⁺ signaling is necessary for correct biological functioning of chondrocytes (O’Conor et al., 2014; Krupkova et al., 2017; O’Conor et al., 2016; Atobe et al., 2019; Clark et al., 2010). We found that full removal of external calcium sources could further diminish the enhanced chondrogenic response obtained by thermo-mechanical cues when compared with TRPV4 inhibition alone. This behavior was consistent in gene expression profiles of the investigated markers and induced transcription returned toward control levels in the presence of EGTA. These findings indicate that calcium is a major signaling mechanism for thermo-mechanotransduction. Intracellular calcium oscillation is a key signaling mechanism in chondrocytes and can be triggered by different types of mechanical stimuli such as strain (Lv et al., 2018; Han et al., 2012) and fluid flow (Degala et al., 2011). Additionally, the dynamic characteristics of calcium signaling (e.g. amplitude and duration of Ca²⁺ transients) were shown to be temperature dependent (Han et al., 2012). Our results suggest that the signaling cascade of thermo-mechanotransduction process is almost incomplete without external calcium sources. Previous studies also showed that extracellular Ca²⁺ source is necessary for calcium signaling processes and increase in cytosolic Ca²⁺ is not possible without extracellular Ca²⁺, even when intracellular stores are full of calcium (Pingguan-Murphy et al., 2006; Lv et al., 2018).

When thermo-mechanical cues are applied, downregulation of Agc1 and COMP and upregulation of Twist1 were significantly different in the presence of EGTA compared to the condition in which TRPV4 antagonist was used (see Figure 4). This different response could arise from two plausible reasons. First, TRPV4 channels were not fully blocked by GSK 205 and the conveyed signal was strong enough to partially activate some of the channels. This is consistent with other studies (Zhou et al., 2019; Lv et al., 2018) where quantitative results indicated that 10 µm GSK205 significantly inhibited but did not fully abolish TRPV4 channel activity. A second possibility would be that, parallel thermo-mechanically responsive pathways besides TRPV4 were involved in Ca²⁺ mediation in order to make this process more effective and robust. This is not surprising as accumulating evidence in the literature suggests that distinct calcium pathways, including stretch-activated channels (e.g. PIEZO1), voltage-gated calcium channels (e.g. T-type VGCC), purinergic receptors (e.g. P2Y or P2X), PLC-IP3-induced endoplasmic reticulum, and TRP family channels (e.g. TRPV3), are directly or indirectly influenced by an externally applied stimulus (Pingguan-Murphy et al., 2006; Lv et al., 2018; Servin-Vences et al., 2017; Krupkova et al., 2017). Determining other probable contributing mediators in thermo-mechanotransduction processes are thus important next steps and need further investigation.

In summary, different aspects involved in cartilage thermo-mechanobiology following joint loading were analyzed by developing a novel customized in vitro model. Our findings have demonstrated the superior effect of thermo-mechanical cues on chondrogenesis compared to either cue alone. Indeed, the applied synergetic stimulus provided a chondro-inductive signal by simulating the cartilage native environment under loading. This study also has shown that biomimetic temperature evolution as by-product of mechanical loading could itself induce biological responses of cells. Moreover, TRPV4 ion channels were identified as a key mediator of thermo-mechanotransduction process. Thermo-mechano-responsive nature of TRPV4 channels makes them a key target for future investigation on thermo-mechanobiology. We found calcium signaling as a contributing mechanism of action translating the effect of the externally applied cues to intracellular biochemical signals for cells. However, other pathways are also playing parallel roles, especially for Sox9 transcription, which therefore would necessitate further understanding of thermo-mechano-transduction processes overall.

ECPs were isolated from the proximal ulnar epiphysis of a 14-week gestation male donor (Ethics Committee Protocol # 62/07). It has been demonstrated that this progenitor cell source is very stable over many passages and gives reproducible behavior for chondrogenic differentiation (Darwiche et al., 2012). Cell growth medium consisted of DMEM containing L-glutamine, 4.5 g/l D-glucose and sodium pyruvate (Life Technologies) which was supplemented with 10% FBS (Sigma) and 1% L-glutamine and penicillin (Life Technologies). Cells were grown to 90% confluence, trypsinized and redistributed again in 2D culture up to passage 5 before seeding into scaffolds (~1.2 mio/scaffold). Stimulation medium was prepared with cell growth medium without FBS but having 10% ITS IV (Life Technologies) and 1% vitamin C. MTT staining Kit I (Roche) was used to control the cell distribution and Viability Assay Kit (Biotium) was utilized for live/dead evaluation according to the manufacturer’s protocols. By using the MTT reagent, the purple formazan is produced due to the cells metabolic activity and therefore reveals cell distribution within the hydrogels. Viability assay was performed via a cell-permeable dye (calcine) for staining live cells and a cell-impermeable dye (ethidium homodimer) for staining dead cells having a damaged cell membrane. The Hoechst 33258 dye (Thermo Fisher Scientific) was employed for DNA quantification to evaluate the impact of hydrogel functionalization on cell attachment. Briefly, the pure and RGD-modified hydrogels (with and without cells) were cut in small pieces and incubated overnight inside 500 ml papain digestion buffer at 65°C. Then, by dilution of 10 µl of digested solution in 140 µl of the dye (0.2 µg/ml), the emission signal of samples was measured at 460 nm by a Wallac microplate reader after excitation at 355 nm. The DNA content of the samples was finally determined by using a standard curve extracted from sequential dilutions of Calf Thymus DNA (Thermo Fisher Scientific) as calibrators. PrestoBlue viability kit (Thermo Fisher Scientific) was also used to monitor the cell proliferation inside the hydrogels during culture period (Sonnaert et al., 2015) based on manufacturer’s protocol. Briefly, the PrestoBlue reagent was 10 times diluted in culture medium and samples were incubated inside. The fluorescent signal was then measured with Wallac 1420 VICTOR2 microplate reader for excitation/emission wavelengths of 544/590 nm, respectively.

The bioreactor chamber was custom designed and supplemented with different modules to reliably alter dynamic culture conditions. The mechanical design of the bioreactor chamber was optimized to ensure sterility and application of reproducible spatiotemporal stimuli on samples. The chamber, as shown in Figure 2a, contains 12 circularly arranged cylindrical wells for culture of cell-laden hydrogels. The set-up is compatible with the Instron uniaxial loading machine (Instron-E3000, Norwood, MA) to apply the mechanical loading (Figure 2—figure supplement 5) according to reported values for knee cartilage deformation during walking (Liu et al., 2010). The bioreactor was fixed to the Instron apparatus at its base by a support, and was stimulated from above by a set of pistons attached to the actuator. To control culture temperature, a thermo-foil heater (Minco-HM6975) was embedded below an aluminum conductive disc supporting culture wells. The temperature was measured by a miniature RTD sensor (Minco-S308) placed inside one of the wells. The evolution of culture temperature was feedback controlled by a PID microprocessor (Minco-CT16A, Minneapolis, MN) based on the error signal between the desired and monitored temperatures. The parameters of the PID controller were tuned experimentally based on the standard Ziegler-Nichols method. The heat controller was tuned to regulate the culture temperature defined by sequential ramps with different slopes according to the model prediction. This strategy was employed to simulate the temperature increase (occurring in knee joint with its irregular boundary conditions due to cartilage self-heating) in our small-size scaffolds (Ø:8, t: 2.2 mm) surrounded by the relatively large volume of culture medium (~2 ml) inside the bioreactor wells. To minimize the temperature gradient due to conduction within the sample from bottom to top, a medical graded thermoplastic part (PPSU) was incorporated within the stainless-steel wells. The geometry of PPSU part was carefully designed to make convection the dominant heat transfer mechanism to samples. A flexible silicon support was also employed to keep the position of samples on the center during dynamic loading. To maintain sterility, the wells were isolated from the outside environment with perforated solid caps covered with gas-permeable filters (OpSite Flexigrid, Smith & Nephew, Hull, UK). The CO₂/O₂ concentration and humidity inside the chamber were regulated by an external gas mixer (ibidi-Gas Incubation System, Martinsried, Germany) injecting humidified gas into the bioreactor. The humidity and gas concentration was directly measured inside the chamber with external calibrator probes to ensure correct functioning of the gas regulator, humidifier, and embedded sensors of the device.

Salt leaching method was employed for porous hydrogel fabrication by polymerization of HEMA-EGDMA precursor (4.8% molar ratio) inside molds containing sieved salt particles (150–250 µm) as described previously (Nasrollahzadeh et al., 2019; Nasrollahzadeh and Pioletti, 2016). Energy dissipation was determined by calculating the hysteresis area of stress-strain curve during loading and unloading of cycle 100. Equilibrium Young’s modulus (Eeq) was measured by application of sequential stress relaxations and finding the slope of the corresponding relaxed stress in range of 10–20% strain values.

The functionalization of pHEMA porous hydrogels with RGD peptide was performed following a published protocol with slight change (Tugulu et al., 2007; Figure 2—figure supplement 2). The pendant hydroxyl groups of pHEMA hydrogels were firstly activated by 4-nitrophenyl chloroformate (NPC) and then RGD peptides were conjugated. For this purpose, lyophilized hydrogels were vacuum-swelled in the freshly prepared activation mixture of 1.41 g NPC in 15 ml acetone and 1 ml distilled triethylamine. The activation process was continued by agitation of samples in the solution for 1 hr at room temperature. The samples were then extensively rinsed with acetone and methanol several times to remove excess NPC. Afterward, the NPC-activated samples were transferred to the methanol solution containing 1 mM RGD peptides and 2 mM 4-dimethylamino-pyridine. To complete the functionalization process, the samples were incubated overnight at room temperature under gentle shaking in the reactor covered with aluminum foil. The RGD functionalized samples were then rinsed three times with pure methanol for 2 min. To ensure removal of any non-reacted NPC group, the samples were incubated in 0.5 M solution of ethanolamine in methanol for 20 min under gentle shaking. The samples were finally rinsed several times with methanol and bi-distillated water to wash non-specifically adsorbed peptides and entrapped NPC molecules inside the porous hydrogels.

To ensure independent control over applied loading and temperature evolution in bioreactor culture wells, we developed a finite element (FE) heat transfer model using COMSOL Multiphysics software (Burlington, MA). The hydrogel dissipation was considered as a heat source generating 9000 W/m³ (determined experimentally based on area of hysteresis loop at 1 Hz) in each step of cyclic compression. The solid and the fluid phases of the porous scaffold were coupled based on theory of poro-elasticity and a linear viscoelastic behavior for the solid part was assumed as described in our previous work (Nasrollahzadeh et al., 2019). Moreover, the heat transfer module was coupled to the resultant pressure/velocity field in porous media to consider convective heat transfer mechanism. In the model, the 2D axisymmetric cross section of cylindrical samples (2.2 mm thickness and 4 mm radius) and culture medium were divided into FEs and boundary conditions were applied as shown in Figure 2—figure supplement 6 and Figure 2—figure supplement 7. The cyclic displacement (10% strain @ 1 Hz) was applied on the interface of the piston with sample-medium domains and stress, strain, velocity, pressure, and temperature fields were solved in a fully coupled manner. The model was examined for different mesh sizes to ensure that results are not changed with elements size. We assumed isotropic mechanical behavior for the poro-viscoleastic scaffold with material properties according to Table 1.

Human epiphyseal chondro-progenitor cells were expanded in 2D culture according to standard cell culture protocols (Darwiche et al., 2012) before seeding them into porous hydrogels. Hydrogels were disinfected by graded ethanol. Cells (~1.25 million per sample) were infused into them by an optimized compression released induced suction method (Nasrollahzadeh et al., 2017). The cell-seeded construct was prepared as a batch and then randomly distributed in different study groups. The effect of the different biophysical stimuli was then assessed by application of respective signals on cell-laden hydrogels. The reported in vivo data in the literature for deformation of tibio-femoral cartilage during walking (Liu et al., 2010) and temperature rise during jogging (Becher et al., 2008) activities were used to apply respective stimulus in vitro. After the seeding step, all samples were pre-cultured for 4 days in cell growth medium inside standard incubators (32°C or 37°C and 5% CO₂). After this step, a cell stimulation medium was used and the thermal (32–39°C), mechanical (10% pre-strain, 10% amplitude at 1 Hz), or thermo-mechanical (their combination) stimulation were applied on samples starting from day 6. Three intermittent stimulations in alternate days were applied on the treated samples for 90 min while the control group was cultured in equivalent conditions but without receiving any stimulus (Figure 3—figure supplement 2).

Total RNA was extracted using the NucleoSpin RNA (Macherey-Nagel) after preparation steps. Briefly, samples were put in a 2 ml Eppendorf tube containing 300 µl Trizol. Hydrogels were disrupted with a polytron (Kinematica AG, Switzerland), while keeping them cold on dry ice. Then, 100 µl chloroform was added and samples were centrifuged for 5 min at 12,000 rpm at 4°C. The aqueous phase was transferred to 2 ml phase lock tubes and centrifuged for an additional 5 min at 12,000 rpm at 4°C. The aqueous phase was carefully transferred to 1.5 ml Eppendorf tubes and the extraction was completed by adding a RNA carrier and following the XS kit protocol. The RNA was quantified using the Nanodrop Lite Spectrophotometer (Thermo Scientific) and reverse transcription of 1 µg RNA was carried out using Taqman Reverse Transcription Reagents (Applied Biosystems, Bedford, MA). Fast SYBR Green Master Mix (Applied Biosystems) was used for PCR amplification in a final volume of 20 μl containing 10 ng of synthesized cDNA. The PCR amplification was performed for each sample by StepOnePlus Real-Time PCR device (Applied Biosystems) using specific human primers for different genes (Table 2). The cycling steps were defined as an initial 95°C for 2 min followed by 40 cycles of amplification. Gene expression data were analyzed using the comparative ΔΔCt (Livak and Schmittgen, 2001) method with RLP13a as the reference gene (Studer et al., 2017).

Characterization of TRPV4 protein on human chondro-progenitor cells was conducted by employing an anti-TRPV4 antibody (ab191580, Abcam) via standard immunostaining technique. Briefly, samples were fixed by PAF (4%) and cells were permeabilized by triton (0.25%). After rinsing by PBST, samples were incubated for 1 hr in blocking solution (1% BSA) at room temperature and then overnight in TRPV4-specific antibody (5 µg/ml) at 4°C. Finally, the secondary antibody (goat-anti rabbit) conjugated with Alexa Fluor 488 was used to visualize TRPV4 channels by a confocal fluorescent microscope (LSM 700-Zeiss).

Calcium imaging was performed by staining samples with 5 µM Flou 4-AM reagent (F14201, Thermofisher) according to manufacturer’s protocol. Activation of TRPV4 channels was assessed by using 10 nM GSK101 in HBSS solution. The samples were imaged by spinning disc confocal microscopy (Visitron) up to 10 min and the effect of TRPV4 agonist was analyzed. To diminish total extracellular calcium and inhibit TRPV4-mediated calcium influx, free calcium ions were chelated with 2 mM EGTA and blocked the channel with 10 µM TRPV4 antagonist, GSK205, respectively.

Gene expression of cell-laden hydrogels in different groups was tested by using medium containing vehicular control, TRPV4 antagonist, and calcium chelator. The cell-seeded samples were prepared as a batch and then randomly distributed in different study groups. The samples were stimulated according to our standard thermo-mechanobiological experiment protocol (Figure 3—figure supplement 2) while the medium was supplemented with 0.2% DMSO as control vehicle, 10 µM GSk205, or 2 mM EGTA. These chemicals were added to the culture wells 1 hr before start of stimulation and removed half an hour after stimulation to limit exposure of cells to pathway antagonists. Afterward, the samples were washed twice by fresh medium and stimulation medium was added without chemicals.

All biological and mechanical experiments were analyzed with at least three replicates per test condition. The statistical significance between different study groups was determined by Student’s t-test (Welch assumption). Statistical comparisons were performed with GraphPad Prism software (GraphPad). The bar charts show mean and SD values. The box and whiskers charts show mean, median, min, and max values. Calcium imaging evaluations were performed at different time points each covering more than 50 cells per test condition during TRPV4 activation, inhibition, and Ca²⁺ chelation.

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

We thank Sandra Jaccoud, Theofanis Stampoultzis, Dr Nathalie Burri for their technical support. We acknowledge the contribution of the Polymers Laboratory of EPFL and in particular Prof Harm Anton Klok for his advice on functionalization of pHEMA hydrogels. We appreciate the support of EPFL mechanics workshop and in particular Marc Jeanneret in design of the bioreactor chamber. The service given by CIME facility at EPFL to prepare SEM images is also acknowledged. This work was supported by the Swiss National Science Foundation (#310030_149969/1 and #CR23I3_159301).

Human subjects: The work involves human subjects as the employed cell source for this study was established in 2009 after standardized processing of a fetal cartilage sample obtained from the proximal ulnar epiphysis of a 14-week gestation male donor (Ethics Committee Protocol # 62/07). The process was registered under the Swiss Federal transplantation program and its dedicated biobank, complying with the laws and regulations within both framework programs and stipulates that the donor is anonymous. The mother donor provided written informed consent that the tissue could be used for development of biobanks for clinical research purposes. The resulting biobank is the property of one of the authors (LAL) and cells have been provided for academic purposes and has been mutually agreed to publishing through a Material Transfer Agreement.

© 2022, Nasrollahzadeh et al.

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.