The Critical Role of Hypoxia in the Re-Differentiation of Human Articular Chondrocytes

The preservation of the chondrogenic phenotype and hypoxia-related physiological microenvironment are major challenges in the 2D culture of primary human chondrocytes. To address this problem, we develop a 3D culture system generating scaffold-free spheroids from human chondrocytes. Our results highlight the chondrogenic potential of cultured human articular chondrocytes in a 3D system combined with hypoxia independently of the cartilage source. After 14 days of culture, we developed spheroids with homogenous diameter and shape from hyaline cartilage donors. Spheroids generated in hypoxia showed a significantly increased glycosaminoglycans synthesis and up-regulated the expression of SOX9, ACAN, COL2A1, COMP, and SNAI1 compared to those obtained under normoxic conditions. Therefore, we conclude that spheroids developed under hypoxic conditions modulate the expression of chondrogenesis-related genes and native tissue features better than 2D cultures. Thus, this scaffold-free 3D culture system represents a novel in vitro model that can be used for cartilage biology research.


Introduction
Osteoarthritis (OA) is a degenerative joint disease associated with cartilage damage, synovitis, and altered bone remodeling [1]. Globally, OA is the most prevalent chronic joint disease, causing motor disability in the elderly where the quality of life is detrimentally affected [2,3]. Moreover, its adverse impact on health systems and economies will persist with the worldwide aging population. For decades, there has been a relentless struggle to find bioactive molecules for cartilage tissue engineering and to treat the early stages of OA that could decrease its progression in the long term. However, one of the challenges we face is designing in vitro models that mimic intra-articular conditions, making the translation of new drugs to humans highly complex.
Three-dimensional (3D) culture of chondrocytes and hypoxic conditions has been proposed as novel approaches to assay cartilage repair treatments [4][5][6][7]. These insightful culture methods provide an attractive option to mimic the tissue microenvironment since they can establish the cell-cell and cell-matrix interactions required for maintaining chondrocyte function and phenotype frequently lost in the monolayer culture [8][9][10]. Nevertheless, establishing an accurate 3D culture system using human de-differentiated articular chondrocytes as starting material represents a challenge due to cartilage source limitation and the marked de-differentiation of human chondrocytes as a result of in vitro expansion required to reach confluency [11,12].
Furthermore, the hypoxic culture condition has been described as a critical factor in emulating the cartilage microenvironment retaining chondrocyte phenotype and reducing cell de-differentiation [4]. Moreover, several studies have also evaluated hypoxia as an innovative culture platform to induce chondrogenic differentiation of mesenchymal stem cells [13][14][15]. In this sense, human chondrocyte culture in 3D systems combined with hypoxia could encourage a sustainable expression of cartilage-specific genes, supporting the re-differentiation of de-differentiated chondrocytes to assemble the native 3D structure of cartilage.
This work represents an innovative method to develop a hypoxic 3D culture system of human osteoarthritic chondrocytes that modulates the expression of chondrogenic genes. The cell culture under hypoxia allows us to use articular chondrocytes derived from osteoarthritic donors as starting material for generating scaffold-free spheroids with native tissue features required for cartilage biology research.

Human Articular Chondrocytes
Hyaline cartilage biopsies (macroscopically healthy) were obtained from a non-weight bearing area of the knees from patients who underwent either knee arthroscopy or total knee replacement. This study was conducted following the Declaration of Helsinki and approved by the Ethics and Research Committee of the Instituto Nacional de Rehabilitación Luis Guillermo Ibarra Ibarra (39/20 SP-1). All participants were appropriately informed about the study's aim and signed the informed consent form. Cartilage samples from 15 donors were included in this study: four anterior cruciate ligament (ACL) reconstructions, four partial meniscectomies, one lateral meniscus fixation, and six total knee replacements. All patients underwent surgery in the usual manner and received the same type of anesthesia (spinal with sedation).

Primary Isolation and Culture of Human Chondrocytes
Immediately after the harvesting process, cartilage samples were subjected to mechanoenzymatic breakdown. Cartilage samples were cut into small pieces. Chondrocytes were isolated by digestion with 2% type II collagenase (Gibco, Thermo-Fischer Scientific, Waltham, MA, USA) for 6 h in an incubator at 37 • C under constant stirring (Enviro-Genie, Scientific Industries Inc., Bohemia, NY, USA) [16]. Isolated cells were cultured in DMEM/F12 medium supplemented with 10% fetal bovine serum (FBS) and 1% antibiotics (Gibco, Thermo-Fischer Scientific, Waltham, MA, USA). Cell cultures were maintained at 37 • C with 5% O 2 and 5% CO 2 at saturation humidity (hypoxia) or 37 • C with 21% O 2 and 5% CO 2 at saturation humidity (normoxia). Chondrocytes from passage 2 (P2) were used to generate the spheroids.

Estimating the De-Differentiation Rate of Human Articular Chondrocytes in Monolayer Culture
The percentage of cells with a chondrocyte-like morphology was analyzed by considering the shape and diameter previously described [17,18]. Briefly, images from cells cultured under hypoxia or normoxia were taken with light-microscopy using an inverted microscope (EVOS FL Auto, Thermo-Fischer Scientific, Waltham, MA, USA). A total of 100 cells were counted in three independent events.

Manufacturing of Non-Adhesive Microwells
Ultrapure agarose was dissolved in PBS (4% w/v), heated, and poured on the bottom of a 12-well plate to stamp negative polydimethylsiloxane (PDMS) customized mold (Research Micro Stamps, Clemson, SC, USA) containing circular pillar structures. Molds with pores of 250 µm diameter and a depth of 250 µm were generated. Once the agarose polymerized at room temperature, the agarose microwell was separated from the PDMS mold. Subsequently, each 12-well plate was collocated under UV radiation for 60 min and then stored at 4 • C until use.

Live/dead Viability Assay
Spheroids in agarose microwell were washed with PBS and incubated with calcein-AM (2 µg mL −1 ) and propidium iodide (2 µg mL −1 ) (Thermo-Fischer Scientific, Waltham, MA, USA). After 30 min of incubation, viability was evaluated using an inverted fluorescence microscope and Pearl Scope 64 image software (EVOS FL Auto, Thermo-Fischer Scientific, Waltham, MA, USA). ImageJ software (NIH, Bethesda, MD, USA) was used to quantify the fluorescence signal. The mean fluorescence signal for calcein and propidium iodide was obtained.

Morphometric Evaluation
The morphology of spheroids was analyzed using Pearl Scope 64 image software (EVOS FL Auto, Thermo-Fischer Scientific, Waltham, MA, USA). The diameter, area (A), and perimeter (p) were measured throughout 14 days of hypoxia and normoxia culturing. Circularity was calculated using the formula f circularity = (4πA)/p 2 .

Glycosaminoglycans Staining
Spheroids were flushed out of agarose microwell and centrifuged in a 12-well plate at 1500 rpm for 1 min. The medium was removed, and spheroids were washed with PBS and fixed in 4% formaldehyde. Staining for glycosaminoglycans (GAG) of the extracellular matrix (ECM) was performed using Alcian blue in 3% acetic acid (pH 2.5) for 40 min. The quantification of staining was performed using the ImageJ software. Color thresholding was used to select and measure blue stained areas (Alcian blue). The percentage of Alcian blue staining was calculated by dividing the blue-stained area by the total spheroid area.

Gene Expression Analysis
We used qRT-PCR to analyze the expression of chondrogenesis-related genes. Primary human chondrocytes and spheroids were dissolved in 0.6 mL Lysis Buffer (Pure Link RNA, Thermo-Fischer Scientific, Waltham, MA, USA), and total RNA was isolated according to the manufacturer's protocol. A total RNA (500 ng) was reverse transcribed in a 20 µL volume reaction using High-Capacity cDNA Reverse Transcription Kit (Applied Biosystems, Thermo-Fischer Scientific, Waltham, MA, USA). cDNA products were stored at −20 • C until used. The qRT-PCR was carried out using 10 µL of RT2 SYBR Green FAST Mastermix (Qiagen, Germany) in a final volume of 20 µL under the following condition:  Table 1). The qRT-PCR amplification was performed by the Rotor-Gene Q thermocycler (Qiagen, Germany).

Statistical Analysis
The normality of the input data was tested with the Shapiro-Wilk normality test. For spheroid diameter and Alcian blue staining of hypoxic spheroids, one-way analysis of variance (ANOVA) followed by the Tukey post hoc test was performed to compare multiple groups. For the analysis of the percentage of cells with a chondrocyte-like morphology in normoxia and hypoxia throughout the culture time, we performed a two-way ANOVA test.
The student t-test was performed to analyze the difference between normoxia and hypoxia-related gene expression in spheroids, GAG staining, and viability assay. The Mann-Whitney U test was carried out for values that were not normally distributed. Among multiple groups was performed with the Kruskal-Wallis test followed by Dunn's pairwise comparison as a post hoc test for gene expression in primary cell culture. Significance was assigned at p < 0.05.
If not otherwise stated, data are presented as median with interquartile range (IQR) or mean with 95% confidence interval (CI). The statistical analysis was carried out using the STATA v.13 statistical package (StataCorp, College Station TX, USA). Graphics were generated using GraphPad Prism 9.0.1 (GraphPad Software, San Diego CA, USA).

Anthropometric Features of Cartilage Donors
We analyze the impact of age, body mass index, gender, and procedure stratified by normoxic or hypoxic culture. There was no significant difference in age (p = 0.16), body mass index (p = 0.33), gender (p = 0.59), and procedure (p = 0.71) ( Table 2). Macroscopically, cartilage samples showed a smooth and intact surface.

Morphology of Passaged Articular Chondrocytes
Human articular chondrocytes from cartilage biopsy were cultivated in hypoxic or normoxic conditions until confluence and designated as passage 0 (P0). The articular chondrocytes showed a rounded, polygonal morphology at P0 (Figure 1a,b). However, there was a progressive change to fibroblast-like morphology through the passages, mainly observed in chondrocytes from normoxic culturing conditions (Figure 1b). Chondrocytes at P2 still exhibited a native chondrocyte morphology when cultured in a low oxygen tension (5% O 2 ) incubator ( Figure 1c). Significant differences in the number of cells with chondrocyte-like morphology were mainly observed at P0 and P1 between normoxia and hypoxia. Moreover, articular chondrocytes cultured in hypoxia presented a higher growth rate by reaching confluency on average after 25 days versus 31 days required in normoxia.

Spheroid Morphology
Cells aggregated spontaneously in agarose microwells, and regular-shaped spheroids were formed, maintaining stable circularity throughout culturing. After one day in hypoxia, compact spheroids with smooth surfaces were observed; nevertheless, spheroids generated in normoxia showed irregular borders making circularity less stable ( Figure 2). Moreover, a steady size throughout the culture was kept until day 14 with hypoxic conditions (143 µm (95% CI 139-145)). In contrast, spheroids became rounded and compact with increasing culture time in normoxia, showing a final diameter of 124 µm (95% CI 120-129) on day 14 ( Figure 2). roids were formed, maintaining stable circularity throughout culturing. After one day in hypoxia, compact spheroids with smooth surfaces were observed; nevertheless, spheroids generated in normoxia showed irregular borders making circularity less stable ( Figure 2). Moreover, a steady size throughout the culture was kept until day 14 with hypoxic conditions (143 μm (95% CI 139-145)). In contrast, spheroids became rounded and compact with increasing culture time in normoxia, showing a final diameter of 124 μm (95% CI 120-129) on day 14 ( Figure 2).

Viability
We estimated the viability of cell-forming spheroids by live/dead assay. After 14 days of hypoxic culturing conditions, the spheroids exhibited minimal area of dead cells conserving their metabolic activity (Figure 3a). Nevertheless, a significant reduction of cell viability was detected predominantly in spheroids from normoxia, showing a low calcein signal (78.85% (95% CI 75.27, 82.44)), but not in spheroids cultured in hypoxia (97.49% (95% CI 96.52, 98.46)) (p < 0.001) (Figure 3b). Calcein and propidium iodide intensities were normalized to the spheroid area and expressed as a percentage of the relative intensity of fluorescence.

Viability
We estimated the viability of cell-forming spheroids by live/dead assay. After 14 days of hypoxic culturing conditions, the spheroids exhibited minimal area of dead cells conserving their metabolic activity (Figure 3a). Nevertheless, a significant reduction of cell viability was detected predominantly in spheroids from normoxia, showing a low calcein signal (78.85% (95% CI 75.27, 82.44)), but not in spheroids cultured in hypoxia (97.49% (95% CI 96.52, 98.46)) (p < 0.001) (Figure 3b). Calcein and propidium iodide intensities were normalized to the spheroid area and expressed as a percentage of the relative intensity of fluorescence.

Extracellular Matrix Production
The production and distribution of GAGs into spheroids were analyzed. On day 14, Alcian blue staining showed broadly distributed GAGs throughout the area in spheroids cultured in a hypoxic environment (Figure 4a). To quantify differences in GAG distribution among samples, we determined the percentage of stained area concerning total area using ImageJ software. Significant differences were found in the rate of GAGs in normoxia (70.35 ± 5.97%) compared to hypoxia (91.93 ± 2.08%) (p < 0.001). In this context, GAGs distribution was also conserved independently of the sample source when spheroids were generated under hypoxic conditions (Figure 4b). Interestingly, geometric features of spheroids cultured in hypoxia were preserved independently of the sample source (Figure 4c,d). A total of 10 human cartilage samples from different donors were analyzed, showing a homogeneity in size and circularity after 14 days of cell culture with a chondrogenic medium under hypoxic conditions.

Extracellular Matrix Production
The production and distribution of GAGs into spheroids were analyzed. On day 14, Alcian blue staining showed broadly distributed GAGs throughout the area in spheroids cultured in a hypoxic environment (Figure 4a). To quantify differences in GAG distribution among samples, we determined the percentage of stained area concerning total area using ImageJ software. Significant differences were found in the rate of GAGs in normoxia (70.35  5.97 %) compared to hypoxia (91.93  2.08 %) (p < 0.001). In this context, GAGs distribution was also conserved independently of the sample source when spheroids were generated under hypoxic conditions (Figure 4b). Interestingly, geometric features of spheroids cultured in hypoxia were preserved independently of the sample source ( Figure  4c

Gene Expression Profile
Re-differentiation of chondrocytes forming spheroids in normoxia and hypoxia was assessed to identify the impact of these culturing conditions on the regulation of chondrogenic gene expression. Regarding spheroids generated in normoxia, there was a signifi-

Gene Expression Profile
Re-differentiation of chondrocytes forming spheroids in normoxia and hypoxia was assessed to identify the impact of these culturing conditions on the regulation of chondrogenic gene expression. Regarding spheroids generated in normoxia, there was a significant difference in the expression ratio of SOX9 (2.2-fold increase, p < 0.001) and COL2A1 (1.6-fold increase, p < 0.019), taking P2 as the reference group; meanwhile, ACAN expression did not change drastically. Moreover, it was observed a higher COL1A1 and COL10A1 expression of 5.2-fold increase (p < 0.001) and 9484-fold increase (p < 0.001), respectively.

Discussion
This study demonstrates the influence of 3D culture conditions combined with hypoxia maintaining the expression of chondrogenic genes and native tissue features in chondrocytes derived from human samples with different inflammatory compromises. Previous cartilage engineering and tissue repair studies have reported similar findings related to the chondrogenic potential by combining a 3D culture system and hypoxia in rat and pig chondrocytes [6,19]. Nevertheless, to our knowledge, this work represents the first report highlighting the use of articular chondrocytes derived from osteoarthritic donors as starting material for generating scaffold-free spheroids under hypoxic conditions to promote the gene expression characteristic of native cartilage.
The in vitro expansion of human chondrocytes resembling the physiological environment of cartilage is essential to test clinically used cartilage treatments and to explore potential new therapies. However, 3D cultures also represent an opportunity to evaluate the development and progression of articular pathologies as OA at early stages [20]. Nevertheless, there is no consensus on the ideal in vitro model representing particular features of OA due to different mechanisms to induce catabolic metabolism in chondrocytes. Our data support the generation of human cartilage spheroids with homogeneous GAGs distribution resembling the original native tissue using a low-cost 3D culture method.
For the generation of spheroids in our study, we used primary cells with a certain degree of de-differentiation and phenotype alteration due to the inflammatory environment from which biopsies came. Morphology changes observed in primary chondrocytes at early passages have been previously reported. The de-differentiation phenomenon occurs due to forcing cells to grow in a 2D environment where nutrient and oxygen gradients do not exist [7,21]. In this sense, chondrocytes are prone to acquire a fibroblast-like phenotype associated with an altered gene expression of SOX9, COL2A1, and ACAN [11,22]. This research used chondrocytes at P2 to avoid a marked cell de-differentiation in monolayer culture under normoxia and hypoxia. Our 2D culture showed only altered expression of the chondrogenic marker, COL2A1, being higher affected in chondrocytes cultured in normoxia ( Figure 1d); this represents a major inconvenience for in vitro culture of chondrocytes as, in turn, decreased COL2A1/COL1A1 ratio indicates de-differentiation. Our study observed a significant decrease in COL2A1/COL1A1 ratio with sequential passaging under normoxic conditions even when COL1A1 expression was not altered (Figure 1d). Not surprisingly, COL2A1/COL1A1 ratio was stable in chondrocytes cultured under hypoxia as COL1A1 showed lower expression and COL2A1 was less affected (Figure 1e).
It is worth mentioning that allowing chondrocytes to proliferate until reaching confluence may explain the steady RNA-level expression of SOX9 and ACAN observed in cells at P0 vs. passaged chondrocytes from hypoxia and normoxia in our study. Nev-ertheless, further studies have reported that chondrocyte marker genes decreased some orders of magnitude compared to freshly isolated chondrocytes from cells at first passage [19,23]. However, it has been noticed a changeless expression of chondrogenic genes (e.g., SOX9 and ACAN) in chondrocytes from human donors and animal specimens at early passages [19,21,23]. Marked changes in SOX9 expression were observed for up to the sixth passage, while COL2A1 declined progressively at the first cell passages of human articular chondrocytes leading to a progressive loss of the chondrogenic phenotype [21,23].
On the other hand, prolonged monolayer culture triggers the development of fibrous cartilage, which is highly undesirable because it exerts articular homeostasis imbalance [24][25][26]. Chondrocytes cultured in hypoxic conditions showed a significant decreasing trend in the expression of COL10A1, a marker of chondrocyte hypertrophy; nevertheless, normoxia promoted a higher expression resulting in an altered COL2A1/COL10A1 ratio. These findings are consistent with previous studies reflecting de-differentiation of passaged chondrocytes under normoxia [24,27,28].
Culturing chondrocytes in a 3D environment has been widely proposed as a promising approach as it can mimic native articular signaling by re-establishing a specific phenotype and cellular functions of the cartilage compared with other previously reported methods not related to the spheroid generation [29][30][31]. Significantly, the 3D model triggers a tissuespecific phenotype due to cell-cell and cell-ECM interactions. Re-differentiation of passage cells has been demonstrated in many studies where an increased expression of COL2A1 and ACAN was achieved in scaffold-based 3D culture versus 2D culture [32][33][34][35]. In our study, gene expression analysis of scaffold-free spheroids comparing monolayer culture at P2 confirmed the re-differentiation process exerted mainly under hypoxic culturing conditions. As shown in Figure 5a, SOX9, ACAN, and COL2A1 were up-regulated in hypoxia. Nevertheless, no significant differences were observed for ACAN expression when spheroids were generated under normoxia. This expression profile is similar to those reported in previous studies using pellet culture and scaffold-free spheroids under hypoxia and normoxia [6,[36][37][38].
It should be noted that the spheroids had an increased expression of COL10A1, a hypertrophic cartilage-related marker under hypoxia. High expression of hypertrophic genes has been identified during re-differentiation of chondrocytes regardless of their origin and in vitro culture method [36,39,40]. Although an elevated expression ratio for COL10A1 was expected for normoxia, hypoxia could not deplete it (Figure 5a). However, COL10A1 expression was significantly lowered in spheroids under hypoxic conditions than in normoxia after 14 days of culture with a chondrogenic medium (Figure 5b). It has been demonstrated in recent studies that TGF-β, found in the chondrogenic medium, can drive hypertrophic differentiation in chondrocytes while also inducing its leading role as a transforming growth factor in cartilage differentiation [41][42][43][44].
Furthermore, the differences in COL2A1/COL1A1 and COL2A1/COL10A1 ratio in spheroids indicate that hypoxia represents a better culturing system to control the chondrocyte features (Figure 5c). These results represent a success in our study as spheroid-forming chondrocytes maintained a genotype and function despite their osteoarthritic origin. Moreover, source limitation and time for in vitro expansion represent crucial factors that should always be considered for their implication for triggering de-differentiation in primary human chondrocytes.
Using human donor cartilage for 3D culture implies the diversification of chondrocyte phenotype obtained in diverse samples due to different OA grades observed in each patient. Regardless of the anthropometric features of donors recruited in this study, we developed size-controlled spheroids from ten hyaline cartilage using a scaffold-free system and resembling the articular environment with hypoxic culturing conditions ( Figure 2). Since cartilage is a tissue with a low oxygen tension environment, it has been proposed to use an oxygen concentration of 5% for the chondrocyte in vitro culturing [45].
As the oxygen gradient has been described as an essential factor regulating cell viability dependent on spheroid size, we evaluated viability by Live/Dead assay on day 14 of human chondrocyte spheroids generated in hypoxia. In this context, our results revealed that hypoxia did not induce toxicity in the core of our in vitro 3D cartilage model (Figure 3a). Supporting our data, different studies mention that hypoxia plays a vital role in the growth and survival of mesenchymal stem cells embedded in scaffolds [46,47], being able to regulate anabolism under a hypoxic microenvironment. Although hypoxia represents the best scenario for cartilage 3D culture, adequate development of chondrocyte spheroids after 14 days of culture in normoxia combined with a chondrogenic medium was also demonstrated in this study ( Figure 2). Nevertheless, spheroids from normoxic conditions showed increased dead cell areas (Figure 3a). These findings agree with a recent study where spheroids from the ATDC5 cell line were developed using a polyethylene glycolcoated chip and chondrogenic differentiation medium. The spheroids showed spotted areas of dead cells after 14 days under normoxia culturing conditions [37]. According to Omelyanenko et al., the nonviable cell population has increased significantly from days 7 to 21. Moreover, the accumulation of specific extracellular matrix components such as type II collagen and aggrecan was closely correlated with time [48], which agrees with the low GAGs synthesis observed in our spheroids under normoxia (Figure 4a).
Our results show that spheroids cultured in hypoxia showed better geometrical features and increased GAGs levels resembling the native articular cartilage than normoxia (Figures 2 and 4). In this context, under hypoxic conditions, the 3D culture system produces an extensive ECM characteristic of native cartilage, suggesting that our in vitro system can regulate chondrogenic phenotype. Therefore, our in vitro 3D culture system represents a novel application for developing disease-modifying OA drugs. Our high throughput system to create human chondrocyte spheroids is based on a PDMS mold that can be easily adopted as specialized equipment is not required. Moreover, this in vitro novel tool represents an alternative for 2D cultures, facilitating the screening of active biological molecules for cartilage engineering and tissue repair, making our model easy to reproduce.
It is essential to state that further characterization of our cartilage in vitro model is required to elucidate the molecular mechanism underlying chondrogenesis in hypoxia.

Conclusions
In conclusion, we have developed scaffold-free spheroids in hypoxia from human chondrocytes derived from osteoarthritic donors, mimicking the in vivo-like cartilage microenvironment that makes possible a cartilage-specific phenotype and functions. This work details how hypoxia represents a critical issue in human chondrocyte culture and the combination with a 3D system to modulate the expression of chondrogenesisrelated genes and native tissue features. This 3D model potentially provides a novel approach to elucidate the mechanism underlying chondrogenesis and clarify the molecular mechanism exerted by new molecules in the cartilage.  Informed Consent Statement: Informed consent was obtained from all subjects involved in this study.