In recent years, changes in the metabolic landscape of OC have been recognized as important drivers of the two major obstacles to effective treatment of OC patients: metastasis and chemoresistance. Thus, a deeper understanding of the cellular mechanisms that underlie these phenomena is warranted. Organoid models have presented a major advance to understanding OC pathobiology, providing an in vitro tool that effectively recapitulates OC cell and tumor characteristics. However, studying metabolism in these organoid models remains technically challenging and therefore largely unexplored. While cold reagents are applied to quench enzymatic reactions when isolating lysate for metabolomics and enzymatic cycling assays in 2D experiments, lysate isolation from 3D cultures requires matrix digestion, during which metabolic changes may occur. Assays that measure tertiary metabolic readouts like oxygen consumption rate (OCR) or extracellular acidification rate (ECAR) are similarly challenging to apply to organoid cultures, as the diffusion of perturbative reagents through the extracellular matrix required for organoid growth remains highly variable.
Genetically encoded, fluorescent metabolic biosensors present a potential opportunity to study OC metabolism in a robust manner. In this study, we investigated metabolic changes in 15 OC cell lines expressing four metabolic biosensors, iNap, Peredox, Perceval, and HyPer, to measure NADPH/NADP+, NADH/NAD+, ATP/ADP, and intracellular H2O2 respectively. These sensors, in addition to providing valuable, dynamic measurements of bioenergetic states, included internal controls which allowed us to account for differences in sensor expression. HyPer, Perceval, and iNap are ratiometric sensors, while Peredox includes a conjugated mCherry for normalization. While other genetically encoded, metabolic biosensors have been described and are being developed [17, 28], this study was a preliminary investigation into the feasibility of using these sensors to explore the metabolic dynamics in OC cells.
Application of these sensors in OC cells revealed changes in cellular metabolism when cells were exposed to OCM, which serves as a surrogate for the omental microenvironment. The OCM-induced increases in intracellular oxidative stress measured using HyPer are particularly interesting in this context, as the precise mechanisms that underlie aggressive metastatic growth in the omentum are not well understood. Integrating this observation with heterogenous NADPH/NADP + and NADH/NAD + responses of cell lines may suggest that cellular antioxidant mechanisms are differentially activated to allow cells to alter their metabolism in response to the omental microenvironment. Indeed, further investigation into this speculative model of omental metastasis may open avenues to target metastatic OC tumors more effectively.
Another common clinical challenge in the treatment of patients with OC remains the development of chemoresistance, particularly in response to first-line platinum-based therapeutics. Understanding the role of metabolic machinery in promoting carboplatin resistance may similarly yield insights to guide therapeutic approaches. Unsurprisingly, treatment with carboplatin induced alterations in OC metabolism, including increased oxidative stress and generally, increased NADPH/NADP + ratios. Interestingly, as a general trend, the degree to which oxidative stress increased appeared to be associated with the carboplatin resistance of the OC cell line.
In our endeavor to understand if these observed changes could be recapitulated in 3D organoids, we established 3D organoids from SKOV3-HyPer expressing cells. We also injected SKOV3-HyPer expressing cells intraperitoneally and imaged omental seeding of these cells through an intravital window. Performing live imaging in these three contexts (2D, 3D, and in vivo) allowed us to determine which most closely modeled in vivo metabolism. Organoids appeared to be a robust model for in vivo metabolism, an observation in line with a growing body of literature that has demonstrated similar results using organoids to model other important cellular characteristics.
Perhaps more importantly, the strength of the organoid approach for modeling OC cell metabolism in vivo allowed us to investigate the relationship between OC carboplatin resistance and oxidative stress in patient-derived samples. Indeed, we found using this approach that in PDOs established from OC patients, the oxidative stress measured in organoids derived from carboplatin-resistant patients was significantly higher than that measured in organoids from carboplatin-sensitive patients following carboplatin treatment.
The observation of greater oxidative stress in OC cell lines and in OC PDOs could be important for our understanding of the mechanisms of OC, however further studies are necessary to fully elucidate the mechanistic underpinnings of this finding. It is interesting to speculate that activation of oxidative stress pathways could activate cellular response mechanisms that may select for resistant cells. Indeed, a recent study that used longitudinal single-cell RNA-seq to analyze metastatic OC patient tumors revealed that therapy-induced stress clonally selected for resistant phenotypes . Taken together, the observed relationship between oxidative stress and carboplatin-resistant OC may provide an opportunity for synergistic therapy with oxidative stress-reducing agents that could sensitize resistant cells to carboplatin.
Our work represents early pre-clinical work to establish the validity of using biosensors as tools to interrogate OC metabolism. Further investigations are necessary to expand the scope of the conclusions of this study. For example, it remains unclear how faithfully organoids recapitulate in vivo tumor dynamics. Although the sensors used here provided valuable information, an important limitation is that they compete with endogenous enzymes for intracellular metabolites, thereby potentially altering the cellular processes they are intended to measure. Although previous studies have suggested that this effect is minimal, it requires ongoing consideration [19, 21]. Moreover, appropriate controls using exogenous stimuli to establish biosensors’ responses are necessary.
Finally, while this initial analysis of OC organoids derived from carboplatin-resistant and carboplatin-sensitive patients presents a promising proof-of-concept, the sample size was small (n = 4 per group). The high rates of establishment for ascites-derived OC organoids suggests ascites may be a robust source material. In clinical settings, ascites is often drained for symptomatic management and discarded. Further studies using larger sample sizes from more heterogenous patient populations may expand our understanding of the predictive capability of biosensor signals. Despite these limitations, this study illustrates that genetically encoded, fluorescent biosensors can be used to study OC organoids, presenting an unprecedented opportunity to study OC metabolism and pathobiology.