The knee menisci are fibrocartilaginous structures that serve to protect the articular cartilage from excessive stress. Their functional extracellular matrix (ECM) is mainly composed of type I collagen (Fox et al., 2012; Makris et al., 2011). The inner meniscus regions contain lesser amounts of hyaline cartilage-type matrix components such as type II collagen and aggrecan (Fox et al., 2012; Makris et al., 2011). Meniscus ECM is synthesized by cells known as meniscus fibrochondrocytes (MFCs) (Sanchez-Adams and Athanasiou, 2009). In adults, the inner regions are avascular and unable to heal upon injury (Arnoczky and Warren, 1982). These injuries are associated with early osteoarthritis (OA) development (Lohmander et al., 2007). The objective of meniscus tissue engineering is to generate cellular replacements that prevent OA development upon implantation.

Hypoxia (HYP) and mechanical loading are characteristic features of the native MFC environment that influence development of the avascular fibrocartilage phenotype in the inner meniscus regions (Fox et al., 2012; Makris et al., 2011; Lund-Olesen, 1970). For example, the meniscus exists in synovial fluid with oxygen saturation ranging from 1–7% and experiences axial compressive strains of ∼5–10% under body weight (Lund-Olesen, 1970; Freutel et al., 2014). Individual HYP and mechanical loading [e.g., dynamic compression (DC), cyclic hydrostatic pressure (CHP)] treatments and their combination (“mechano-hypoxia conditioning”) can be useful tools for engineering cartilaginous tissues in cells including MFCs and mesenchymal stem cells (MSCs) through the molecular actions of factors including hypoxia-inducible factor-1 (HIF-1), SRY-Box transcription factor 9 (SOX9), and transforming growth factor-beta (TGF-β) (Mauck et al., 2000; AufderHeide and Athanasiou, 2004; Adesida et al., 2006; Suzuki et al., 2006; Baker et al., 2010; Egli et al., 2011; Huey and Athanasiou, 2011; Tan et al., 2011; Furumatsu et al., 2012; Kanazawa et al., 2012; Guilak et al., 2014; McNulty and Guilak, 2015; Zellner et al., 2015; Anderson and Johnstone, 2017; Salinas et al., 2018; Szojka et al., 2019). Hypoxia and mechanical loading do not always have beneficial effects with regards to modulation of ECM components in engineered meniscus (e.g., types I and II collagen, aggrecan), however, suggesting the need to identify appropriate culture parameters (Szojka et al., 2021a). In human engineered meniscus, a short-term (1-day) mechano-hypoxia treatment (30–40% dynamic compressive strain, 3% O₂) promoted fibrocartilage gene expression (Szojka et al., 2021b) after a static pre-culture period, but the effects of long-term treatments and different loading types and parameters have not been explored (Szojka et al., 2021b). Our overarching hypotheses were that longer mechano-hypoxia treatments would also have pro-fibrocartilage effects and that they would translate into beneficial outcomes at mechanical and protein levels.

Thus, we first compared mechano-hypoxia treatments (DC and CHP) and characterized the transcriptome response to a longer (5 days) course of treatment than before (Szojka et al., 2021b). We then manipulated loading incident duration, applied strain, and pre-culture duration to better understand how they affect gene expression. Using this information, we finally investigated the long-term (3 weeks) outcomes of mechano-hypoxia conditioning after a pre-culture period for initial matrix formation.

Most culture methods and assays were performed identically to those described in previous work (Adesida et al., 2012; Liang et al., 2017). The experiments are outlined in Figure 1. Experimental parameters in each experiment were selected based upon previous experiences and new data as it became available, including from pilot studies of scaffold seeding density, orbital shaking, and number of loading days (Supplementary Material) (Bornes et al., 2016; Szojka et al., 2021a; Szojka et al., 2021b; Szojka et al., 2021c).

This article investigated “mechano-hypoxia conditioning” as a strategy for tissue engineering the inner meniscus. Hypoxia (HYP) and dynamic compression (DC) 30–40% affected gene expression in human meniscus fibrochondrocyte (MFC)-based engineered tissues in an overwhelmingly additive manner after 5 days. It is likely that some more synergistic interactions existed than were detected but that detection power was limiting for the RNA-seq analysis. HYP and DC 30–40% also drove an increase in expression of hyaline cartilage-related precursors, indicating that the mechano-hypoxia treatment was promoting fibrocartilage formation. Thus, our hypothesis that the 5-days mechano-hypoxia treatment would have pro-fibrocartilage effects at the gene expression level was supported. Previously, Pauwels proposed that hydrostatic pressure, developing during compressive loading of tissues, is the specific stimulus to induce formation of cartilaginous matrix in the body (Pauwels, 1980). Krompecher later proposed that the hydrostatic pressure generated during mechanical loading forces vessels closed by overcoming blood pressure; this reduces blood supply to related tissues and results in a hypoxic environment that is instead the specific chondrogenic stimulus (Krompecher and Tóth, 1964). The outcomes of experiment I indicate that mechanical loading can be a direct promoter of cartilage-related gene expression independently of low oxygen. These effects were primarily driven by hypoxia rather than DC, however, which is generally consistent with Krompecher’s view. The previous studies that directly evaluate combined hypoxia and mechanical loading usually showed pro-chondrogenic effects of hypoxia but mixed outcomes from cyclic and intermittent hydrostatic pressure and dynamic compression (Domm et al., 2000; Hansen et al., 2001; Szojka et al., 2021a; Szojka et al., 2021b; Wernike et al., 2008; Meyer et al., 2010; Parker et al., 2013). One of these studies, applying porcine mesenchymal stem cells (MSCs), noted that, “low oxygen is a more potent promoter of chondrogenesis than dynamic compression.” The outcomes of experiment I (Figures 2, 4) affirm this statement for human MFCs at the gene expression level (Meyer et al., 2010).

Outcomes in experiments I/II suggest that the magnitude of the gene expression response to mechanical loading is closely related to the aggression of the loading regime (Figures 2, 6). To be useful for meniscus tissue engineering, Cyclic hydrostatic pressure (CHP) loading and hybrid displacement-load control regimes such as the DC 1%/2 kPa group introduced here will require more aggressive parameters than those used here, such as higher hydrostatic pressures (Gunja et al., 2009; Gunja and Athanasiou, 2010) and applied strains per cycle (e.g., 1.5%/2 kPa). Platen liftoff has long been recognized as a problem in DC loading studies and is typically addressed by use of static strain offsets or reduced loading frequencies (Mauck et al., 2000). The hybrid DC regime introduced here provides a unique solution to the problem of platen liftoff that could be useful in other tissue engineering applications of DC.

Experiment II showed that to induce gene expression changes: i. the early loading period is most important, and ii. there is a dose effect with increasing strain (Figures 5, 6). Previous researchers identified that DC systems reached a mechanical steady state within minutes, which we confirmed at the gene expression level (Puetzer and Bonassar, 2016). The (5c + 5r) × 6 group showed partial recovery of the stress response for each loading incident but without differential gene regulation to the other 60-min groups; longer cyclic rest periods may thus be more promising (Figure 5). The results complement what has been described as a fundamental rule in bone: the adaptive loading response saturates quickly, causing diminishing returns for extended loading incident durations (Turner, 1998). Many DC parameters need to be appropriately selected in tissue engineering to achieve beneficial outcomes (Anderson and Johnstone, 2017; Salinas et al., 2018). Loading events as short as 5 min make it much more feasible to optimize these parameters, which is a useful outcome for the tissue engineering field. The observed dose effect with strain (an increased static strain offset from 10 to 30% with a fixed dynamic strain amplitude of 10%) indicates that DC can directly induce SOX9 expression rather than only doing so indirectly through enhanced medium mixing and platen-induced hypoxia (Figure 6). More investigation is required regarding the effects of pre-culture duration on gene regulation in response to DC, as the two available donors showed different trends.

Mechano-hypoxia conditioning (3 weeks) supported increases in mechanical properties with mixed matrix outcomes that included suppression of hypertrophy at the mRNA level (Figures 7, 8). Normal growth with culture time surely contributed to the mechanical gains as previously reported, but it is doubtful that it could have led to the sigmoid-shaped mechanical growth curves (Figure 7) (Szojka et al., 2021c). These could instead have been the product of factors such as inflammation early on and depletion of growth factors in the medium in the late stages. Each donor had a different growth profile, suggesting that efforts to tailor loading regimes on a donor-to-donor basis could be fruitful (Figure 7). The increase in final dynamic modulus in experiment II (for the available donors), conducted in parallel from 3 to 6 weeks in free-swelling culture compared to the mechano-hypoxia treatment in experiment III (after adjustment for higher strain levels in the second 2 weeks) were 2.2 vs. 5.1-fold (Donor D) and 1.3 vs. 1.8-fold (Donor F) (Figures 6, 7). This suggests that the mechano-hypoxia treatment indeed accelerated mechanical property development. Mechanical loading has previously been reported to have beneficial outcomes on mechanical properties (Baker et al., 2010; Huey and Athanasiou, 2011; Puetzer and Bonassar, 2016). DC led to superior outcomes at 2 but not 6 weeks compared to static controls with bovine MFCs (Ballyns and Bonassar, 2011), which may be consistent with the tendency for mechanical properties to plateau here. Matrix staining for sulphated proteoglycans appeared qualitatively reduced in HYP/DC compared to static controls, which could be due to GAG loss to the medium (Figure 7B) (Puetzer et al., 2012). There was no significant upregulation of c-FOS in HYP/DC, which is accounted for by the 6-h delay in harvest after the final loading event (Figure 8). Interestingly, matrix precursor expression in the mechano-hypoxia treatment was inferior to free-swelling controls despite HYP/DC having increased SOX9. Thus, altogether our hypothesis that extended mechano-hypoxia conditioning would lead to superior mechanical properties and extracellular matrix (ECM) protein expression was only partially supported. Future investigations are needed to understand the processes underlying this outcome in order to achieve gains in ECM properties with long-term loading (Figure 8) (Huang et al., 2010).

This study had several limitations. Inclusion of more donors in all experiments would have increased detection power; at best three donors were included in each experiment albeit with multiple tissue replicates and experimental designs favourable to detection of main effects, e.g., the factorial investigation in experiment I. A second limitation is that RNA-seq was only performed after the relatively short 5-days mechano-hypoxia treatment in experiment I. In longer-term mechano-hypoxia conditioning, tissue properties would diverge due to the unique gene expression profiles in each group (as identified in experiment I), likely affecting the reception of and response to mechano-hypoxia signals. Thus, many more gene interactions could be identified with longer-term treatments due to differing tissue properties; nevertheless, the results here indicate that the early molecular response to the treatments has few interactions (Figure 4A). The finding in experiment II that 5 min of loading was adequate to saturate much of the gene expression response was dependent on the state of maturity of the tissues (Figure 5). A longer course of loading than 5 min could be required for a more mature tissue that had undergone less stress relaxation by this time point; thus, to determine the appropriate loading incident duration for a different tissue the stress relaxation profile needs to first be assessed. Finally, experiment III had two main limitations. First, the medium was not changed during the whole 3-weeks culture period (though it was initially supplied in 5× excess and there was no colour change to indicate an accumulation of cell metabolites). It is possible that the growth factors degraded during this time, resulting in a tendency for mechanical properties to plateau (Figure 7). Second, the static control groups were free swelling due to the limited number of bioreactors available; it would have been ideal to have 0, 30, and 40% static compression controls within replicate bioreactors. These would have accounted for the reduced exposed surface area in HYP/DC, which likely contributed to the outcome of reduced matrix staining in HYP/DC compared to the free-swelling controls.

A priority in future work would be to address the matrix-poor core regions by increasing the local density of viable ECM-forming cells. Possibilities include using a gentle DC regime to promote media supply to the inner regions, forced perfusion during the pre-culture phase, or constructs incorporating medium channels (Timmins et al., 2007; Schulz et al., 2008; Daly et al., 2018). As well, the suppression of hypertrophic markers in combined HYP/DC suggests mechano-hypoxia conditioning may have utility in engineering inner meniscus and articular cartilage with MSCs. These have some superior functional matrix-forming capabilities to MFCs but with higher native expression of constituents such as type X collagen that are undesirable for engineering cartilaginous tissues (Elkhenany et al., 2020). The bioreactor in experiment III required little user input during the culture period, which is practical for long-term investigations of mechanical loading that are usually labour intensive. Mechano-hypoxia conditioning with a similar bioreactor could also be employed towards building a functional meniscus. The timing of the treatment could be delayed until an organized and appropriately mature matrix of collagen fibres was formed within a biomimetic meniscus shape, potentially using boundary conditions and cell contraction or polymer fibres with a compression or tension-loading component to guide fibre formation (Ballyns and Bonassar, 2011; Puetzer and Bonassar, 2016; Puetzer et al., 2021). Then, mechano-hypoxia conditioning could be initiated to promote mechanical property development and fibrocartilage maturation in the inner regions, mimicking an aspect of human meniscus development.

In conclusion, mechano-hypoxia conditioning is promising for meniscus tissue engineering as it may support fibrocartilage formation and compressive property development. However, further research is required to achieve continual development of mechanical properties towards native levels and to incorporating a native meniscus-like collagen architecture.