Discussion
This study demonstrates that cartilaginous tissue formation can be significantly upregulated by controlling nutrient metabolism and that this response appears to be mediated through resultant changes in HIF-1α signalling. Chondrocytes, although characteristically described as anaerobic cells (Khan et al. , 2009), appear to switch their metabolism in response to nutrient availability. Under low-to-intermediate volumes of media (< 2 mL/106 cells), chondrocytes display an anaerobic phenotype. At higher media volumes (> 2 mL/106 cells), glucose uptake increases and metabolism switches to a mixed aerobic-anaerobic phenotype with the aerobic pathway appearing to be more utilized with further increases in media volume. This response is similar to the Crabtree effect — the phenomenon where aerobic cells also excrete fermentation products (e.g. ethanol or lactate) in the presence of oxygen and high external glucose (De Deken, 1966). While several potential mechanisms have been postulated to explain the Crabtree effect (e.g. catabolite repression, catabolite inactivation, limited respiratory capacity), recently it has been explained by the overflow of glucose metabolites caused by the saturation of TCA cycle capacity (Pronk et al. , 1996). Here, we postulate an inverse response (“inverse Crabtree effect” ) where under high glucose availability, increased glucose uptake leads to the saturation of the fermentation pathway causing metabolite overflow to the TCA cycle (Figure 5). However, it should be noted that additional studies to quantify metabolic pathway flux (e.g. 13C flux analyses) are required to confirm this notion.
Previous studies have also shown that increased glucose uptake occurs under elevated media volumes (Heywood et al. , 2006; Suitset al. , 2008) and that metabolic switching can occur in response to glucose availability (Otte, 1991; Lee & Urban, 1997; Heywoodet al. , 2010; Heywood et al. , 2014). Under low glucose, increases in oxidative phosphorylation have been observed, providing evidence for another Crabtree-like effect under such conditions (Otte, 1991; Lee & Urban, 1997; Heywood et al. , 2010; Heywood et al. , 2014). Taken together, this suggests that chondrocytes may have several different metabolic phenotypes depending on the availability of glucose (in the presence of oxygen): predominantly anaerobic under intermediate glucose levels (Warburg-like effect) and increased aerobic activity under either low or high glucose levels (Crabtree-like effects).
One important consequence of changes in metabolic phenotype was the increased biosynthetic response (2.2- to 3.5-fold) observed at the transition between metabolic states (~ 2 mL/106 cells). Previous studies have demonstrated that increased nutrient availability (by media volume) can improve cartilaginous tissue growth (Mauck et al. , 2003; Heywood et al. , 2006; Khan et al. , 2009; Oze et al. , 2012) as well as that chondrocytes are more synthetically active under mixed aerobic-anaerobic metabolism (Lane et al. , 1977; Obradovicet al. , 1999; Khan et al. , 2009). When cultured at, or near, the point of transition between anaerobic to mixed aerobic-anaerobic metabolism, a state of pseudo-hypoxia occurs resulting in the initiation of HIF-1 signaling leading to increased cartilaginous tissue formation. This response is believed to manifest as a result of the pooling of intracellular metabolites during the transition to different metabolic states, which in turn, can stabilize HIF-1α by interfering with PHD2 (Lu et al. , 2002; Kim et al. , 2010; Ren et al. , 2011; De Saedeleer et al. , 2012; Bailey & Nathan, 2018). Of the 14 metabolites investigated, only intracellular lactate and succinate were correlated with PHD2 activity. Both lactate (De Saedeleer et al. , 2012) and succinate (Bailey & Nathan, 2018) can affect PHD2 enzymatic activity; however, by different means. Lactate inhibits PHD2 by competing with its substrate α-ketoglutarate (De Saedeleer et al. , 2012) whereas succinate affects PHD2 activity through product inhibition at high concentrations (Bailey & Nathan, 2018). While additional work is needed to determine the relative contributions of intracellular lactate and succinate pools on PHD2 activity, most likely this affect can be attributed to lactate due to observation that succinate was only present in trace amounts.
HIF-1α stabilization occurred primarily at intermediate volumes, leading to the regulation of hypoxia-induced gene expression (Lu et al. , 2002; Kim et al. , 2010; Ren et al. , 2011; De Saedeleeret al. , 2012; Bailey & Nathan, 2018). Although HIF-1 has many target genes, several support chondrogenesis and regulate cartilage homeostasis (Kim et al. , 2010), including glucose uptake (GLUT1), TCA cycle suppression (PDK1), and chondrogenic differentiation (SOX9); each of which was upregulated (by 2.0- to 2.7-fold) under these conditions. Lastly, loss of function experiments using YC-1 (to degrade HIF-1α) confirmed the involvement of the HIF-1α pathway in these studies. While HIF mediated gene transcription can be induced by other factors, they most likely do not play as prominent a role in the current study. HIF-1α can also be regulated by FIH-1 (factor inhibiting HIF) (Masoud & Li, 2015). Similarly, in the presence of oxygen, FIH-1 hydroxylates HIF-1α to prevent interaction with p300 and blocks transcriptional activation (Masoud & Li, 2015); however, intracellular pyruvate does not affect FIH-1 expression (Dalgard et al. , 2004) or its activity (Hewitson et al. , 2007). Additionally, the other known HIF transcription factors (HIF‑2α, HIF‑3α) are also probably not involved as intracellular pyruvate does not affect HIF-2α (Ren et al. , 2011) and HIF‑3α has multiple variants with different and opposing functions (Duan, 2016). Lastly, while direct hypoxia has been investigated as an anabolic stimulus for chondrocytes (Coyle et al. , 2009; Yodmuang et al. , 2013), conflicting results have been observed, most likely due to the fact that chondrocytes require oxygen to a certain degree and prolonged hypoxia has been definitively shown to inhibit ECM synthesis (Gibson et al. , 2008).
It is also recognized that other factors could potentially contribute to the observed response. To account for changes in hydrostatic pressure between cultures of different media volumes, these studies were conducted using different sized culture plates (i.e. 24-, 12- and 6-well plates). Estimates of the maximum hydrostatic pressure difference between conditions were relatively small (≤ 50 Pa) due to the minimal changes in media height above the cultures (≤ 5 mm). Previous studies have shown that hydrostatic pressures of several orders of magnitude higher (kPa to MPa range) are required to elicit changes in ECM synthesis (anabolic or catabolic) (Elder & Athanasiou, 2009) indicating that the potential influences were negligible. Limitations in oxygen delivery to the cultures can also be a concern with varying culture volumes (Place et al. , 2017). However, the distance from the cells to the media surface was also accounted for by using different sized culture plates and held relatively constant across groups (within 5 mm). In addition, as media buffering capacity changes proportionally with volume, there may have been an influence of extracellular pH. Chondrocytes are sensitive to pH with relatively small changes influencing ECM synthesis (Wilkins & Hall, 1995). However, observed differences in extracellular pH did not correlate with the changes in ECM deposition, which was maximal at intermediate media volumes. Lastly, as the cultures were only evaluated after a 4-week culture period (or the last 48-hour media exchange cycle), additional studies are required to determine whether these effects manifest throughout the culture period.