In this study, we observed that chondrocytes, when cultured in a medium depleted of all glucose and replaced with galactose, underwent a metabolic reprogramming. Control chondrocytes in glucose-rich medium exhibited a high dependence on the glycolysis pathway for ATP production, a dependence that became more prominent when the cells were pro-catabolically activated with IL-1β treatment. This dependence on glycolysis was inversely echoed by a deficit in ATP produced through mitochondrial respiration. As such, these conditions in glucose-rich medium mimic the mitochondrial dysfunction associated with OA 2,14−18. In a galactose-replaced medium, the use of glycolysis by chondrocytes was substantially reduced; mitochondrial respiration increased concomitant with an increase in overall mitochondrial function. More importantly, the development of pro-catabolic features following IL-1β treatment, including MMP13 production and proteoglycan loss from cartilage explants, were blocked within vitro culture in galactose-replaced media.
Previously we reported that the overexpression of hyaluronan synthase-2 (HAS2-OE) blocked MMP13 and MMP3 synthesis in activated chondrocytes but by way of intracellular changes and not feedback due to the accumulation of extracellular hyaluronan (HA) 1. Treatment of activated chondrocytes with the HA inhibitor 4MU also blocked the pro-catabolic response 4. Given that both seemingly opposing treatments had in common, the depleted UDP-glucuronic acid pools (via hyaluronan biosynthesis or substrate sequestration) led us to examine potential changes in overall chondrocyte metabolism. Enhanced HA production, as well as HA inhibition by 4MU, both reduced the use of glycolysis in activated chondrocytes and rescued IL1β-induced deficits in mitochondrial respiration 2. These metabolic changes were coincident with a reduction in enzymatic degradation of the extracellular matrix. To further document that the metabolic shifts of HAS2-OE and 4MU were due, in part, to depletion of intracellular precursor pools of glucose derivatives, we examined the low-dose use of inhibitors of the glycolysis pathway, namely 2DG and dichloroacetate (DCA) 2. As with HAS2-OE and 4MU, both glycolysis inhibitors 2DG and DCA rescued mitochondrial activity in activated chondrocytes, resulting in a downstream inhibition of the OA-like pro-catabolic phenotype. However, all these experiments were performed in a standard glucose-rich culture medium.
It is known that many cells in glucose-rich culture adapt to the prominent use of the glycolysis pathway, even under aerobic conditions 19. This has made it difficult to study the role of mitochondria and oxidative phosphorylation. In some cells, this is termed “Crabtree Effect,” where there is enhanced use of glycolysis even when mitochondria are still present and functional 20. Data in Fig. 3 of this paper suggests that functional mitochondria are present in control chondrocytes even though ~ 75% of ATP was being generated by glycolysis (Fig. 2). To overcome these adaptive changes, investigators have used galactose-replacement. Galactose can be metabolized by the Leloir Pathway to glucose-1-phosphate and then enter the glycolysis pathway after modification to glucose-6-phosphate. However, the multi-step conversion of galactose to glucose-6-phosphate is slower than starting with glucose and results in conditions that promote oxidative phosphorylation to generate ATP 5. Galactose can also be metabolized to glyceraldehyde-3-phosphate and pyruvate through the phosphate pentose shunt but, that process generates no net ATP and thus again forces cells to rely on mitochondrial oxidative phosphorylation for ATP 21. Fortunately, cultured cells generally grow well under galactose-replacement conditions—conditions that allow for a more accurate evaluation of the role of mitochondria in a host of cellular activities 6,8.
Using galactose-replacement conditions in our study, we observed that lactate production (measured as the rate of proton acidification of the medium or PER) was highly reduced; OCR (oxygen consumption related to oxidative phosphorylation of the tricarboxylic acid or Kreb’s cycle) was very highly enhanced as compared to glucose-rich conditions (Fig. 1). These results provide evidence that control chondrocytes could switch from ~ 75% (ATP produced from glycolysis) / 25% (ATP generated from mitochondrial respiration) to 12% / 88% in galactose-replaced medium (Fig. 2). Even though the control chondrocytes were highly glycolytic, they exhibited mitochondria with an intact, functional mitochondrial membrane potential (Fig. 3). After treatment with IL-1β in glucose-rich media, the otherwise normal, healthy chondrocytes developed a pro-catabolic phenotype reminiscent of events associated with human OA. Mitochondrial function diminished (but not the number of mitochondria, Fig. 3), nitric oxide (NO), inducible Nitric Oxide synthase (iNOS), and reactive oxygen species (ROS) levels became elevated (Fig. 4). But critically, markers indicative of OA-like destruction of the extracellular matrix, namely MMP13 mRNA and loss of proteoglycan from cartilage explants (indicative of ADAMTS4 activity), were prominently enhanced (Fig. 6). All of these IL1β-induced features were blocked when the experiments were performed in parallel in galactose-replaced medium. Mitochondria of IL1β-treated chondrocytes in galactose-replaced medium regained membrane potential, pAMPK levels spiked coordinate with decreases in NO, iNOS mRNA, ROS, and NMOC (Figs. 3–5). These data suggest that the induced mitochondrial dysfunction and associated pro-catabolic effect in chondrocytes are reversible similar to the Crabtree Effect observed in several other cell types 20.
We used IL-1β to activate chondrocytes in this study. IL1β is commonly used by many investigators to induce and mimic OA-like features, including changes in metabolism associated with OA, such as an increase in NO, leading to mitochondrial damage and release of ROS and the activation of extracellular enzymatic cartilage damage—the major issues of OA 22. However, the genesis of human OA is more complex and multifaceted. In the natural state, the onset of OA is due to trauma, aging, inflammation, and the effects of feedback by the degradation products of tissue damage (termed DAMPs) 15,23−25. IL1β may not induce all of the features of human disease, but it is highly reproducible, and since it is used by many investigators, it allows cross-comparisons between studies. In our previous studies, we made comparisons between IL-1β other commonly employed inducers, including; the inflammatory cytokine TNFα, three relevant examples of DAMPS, namely fibronectin fragments, HA oligosaccharides, and LPS 1,2. Each of these inducers was capable of activating molecular markers and features of OA (such as MMP13 and proteoglycan release from cartilage) in bovine or human OA chondrocytes as well as cartilage explants. More importantly, all of these pro-catabolic features were blocked by co-incubation of each inducer and HAS2-OE 1, 4MU 2,4, 2DG or DCA 2. In addition, non-induced human chondrocytes derived from human OA patients 1,2,4 were also examined. Like in this study (Fig. 7), albeit with variability from patient-to-patient, baseline MMP13 was elevated without the need of an inducer such as IL-1β. These elevated MMP13 levels could also be reversed by treatment with HAS2-OE, 4MU, 2DG, DCA or, in this case, by galactose-replacement 1–3. After establishing these results, only IL1β was employed as an inducer in this study.
It may very well be that OA chondrocytes have undergone long-term metabolic reprogramming and mitochondrial dysfunction, and even depletion of mitochondria. This long-term metabolic reprogramming is the distinguishing feature between the Warburg Effect (long term) and the Crabtree Effect (short term, reversible) in tumor cells 20. It should be noted that control human OA chondrocytes have enhanced MMP13 protein levels and non-detectable pAMPK at baseline even before treatment with IL1β (Fig. 7). Nonetheless, galactose-replacement substantially reduced both baseline and IL1β-activated MMP13 protein levels. Moreover, baseline pAMPK and PGC1α (proteins associated with functional mitochondria) were also substantially enhanced. Thus, although chondrocytes derived from human OA patient tissues may exhibit levels of permanent mitochondrial dysfunction, our galactose-replacement studies suggest that both mitochondrial activity and pro-catabolic activities remain reversible. It is possible that we saw fewer green mitochondria in OA chondrocytes because of autophagy—an attempt by the cells to remove damaged mitochondria to prevent further cellular damage.
There are several limitations to this study. As mentioned above, one limitation of this study was that chondrocytes must be grown in culture to do this kind of study, wherein they readily adapt to a high dependence on glycolysis for ATP. This is typical of many studies on metabolism that require the use of cultured cells, including many tumor cell lines 7,8,26 and primary cultures such as myotubes 6,27. This study, like others 7,8,26 used the galactose-replacement approach to overcome this issue. A second limitation was limited access to human cartilage tissue. As such, studies on human OA cells was limited. It could be concluded that the data presented in Fig. 7 at least matches the same trends in results shown in bovine articular chondrocytes. More work will be done in the future as samples become available. It is recognized that all of this work because it is mechanistic in nature, was performed in vitro. Thus, many questions arise, such as do these chondroprotective results (of glycolysis inhibition/ TCA enhancement) change due to oxygen levels, serum levels, glucose concentrations, cell density used, the timing of experiments, etc. We recently addressed this by examining the effects of 4MU in an animal (mouse) model of induced OA 28. In this study, feeding animals 4MU provided a protective effect on the development of OA (in vivo) following medial meniscal ligament transection-induced OA. These OA protective effects were observed in vivo in the natural settings of oxygen, nutrients, growth factors, and loading that occur within a knee joint. This suggests that our in vitro observations of chondroprotection (by metabolic shifting) can be replicated in vivo.
It is unlikely that we could replicate galactose-replacement in vivo. However, that was not the goal of this study. Rather, this study demonstrated that reducing the pro-catabolic phenotype of OA may be obtained by targeting the root cause; namely fixing mitochondrial dysfunction, and that this may be a better approach than blocking cytokines or inhibiting MMPs. Even with control human OA chondrocytes, undetectable levels of pAMPK could be recovered by galactose-replacement, leading to inhibition of MMP13 protein production downstream. Thus, finding more promising pharmaceutical paths to generate mitochondrial reactivation in articular joint chondrocytes may lead to more effective therapeutic strategies to treat OA (or at least reduce OA progression) if used early enough in the disease. And it may not take full recovery of all mitochondria. Galactose-replacement rescued mitochondrial function better than 2DG (and 2DG matches HAS2-OE, 4MU, and DCA). However, galactose-replacement blocked increased MMP13 and cartilage breakdown to a similar extent as 2DG. Thus, forcing the activation of mitochondria, even to a small extent (as in 2DG), is sufficient for this rescue of the pro-catabolic phenotype.