Our results showed that plating material influence the hypoxia-driven retention of [18F]FAZA profoundly in all cell lines, whereas a suppressive effect of polystyrene on [18F]FDG uptake only was observed in SiHa cells. Of note, [18F]FAZA retention data suggests that the average effective cellular oxygen pressure experienced during the tracer loading period is above 0.5% O2 in polystyrene attached cells even after a 2h pre-equilibration period under shaking conditions. This equals a pO2 of ~3.5 mmHg assuming a normal pressure of 760 mmHg and a water vapor pressure of 47 mmHg when using a gas humidifier. Besides posing a problem for the development of hypoxia-selective tracers, such differences may influence any experiment where exact control of oxygen levels are critical, including studies on hypoxia-induced gene expression and hypoxia activated prodrugs. Our results do not decisively rule out other explanations for reduced anoxia-driven tracer uptake in polystyrene grown cells, but our observations corroborate with seminal studies showing that radiosensitivity during an anoxic gas challenge is elevated when using polystyrene and that it relates to the presence of substantial and releasable quantities of O2 in plastics [6,7]. Other types of plastic such as permanox may reduce, but not fully eliminate, the problem [10]. To avoid unwanted influx of oxygen from the substrate, pre-equilibrated plastic-ware has been applied in some studies on bioreductive cytotoxins and the kinetics of hypoxia-induced genes [11]. However, such an approach requires a hypoxia work-station and more importantly, prevents seeding and growing of cells under identical standard culturing conditions prior to experimentation and gas challenge, which in turn may induce unknown biological differences and experimental bias that cannot be resolved unequivocally. The confounding influence of plastic-ware in hypoxia research have generally gained footing in radiotherapy research, but is less appreciated outside this field. In most cell lines, [18F]FDG uptake was uninfluenced by plating material, suggesting that glycolysis is fully activated at pO2 levels higher than those required for significant reduction of [18F]FAZA. This may appear somewhat surprising since the partial pressure at which respiration is reduced to 50% of the maximal rate at saturating oxygen is reported to be around or below 1 mmHg in intact cells [3]. However, during more prolonged hypoxia (hours) adaptive changes typically cause respiration to drop, and glycolysis to increase, at much higher values (so-called oxygen conformance) with a P50 in the range of 5-15 mmHg. In accordance, a 6h exposure to relatively mild hypoxia of 1.5% O2 (~10 mmHg) stimulated FDG retention in 2 tumor cell lines [14]. Interestingly, in SiHa cells, which unlike typical tumor cells have a largely non-glycolytic phenotype under aerobic conditions, and thus experience larger adaptive glycolytic flux changes when mitochondrial ATP synthesis ceases, a significant difference was observed between polystyrene and glass. Whether this relates to true cell-to-cell line differences in pO2 threshold values that affects respiration due to difference in oxygen conformance behavior (see above) or rather reflects increased assay sensitivity in the highly responsive SiHa cell line is unclear, but highlights that also in studies on hypoxia-induced changes in energy metabolism the choice of culturing material may influence results. Indeed, in such studies the use of well-plates is common and this may further exaggerate the problem of prolonged oxygen release and also result in differences between centrally and peripherally located wells, so-called edge effects. Of note, in a hypothetical study conducted in polystyrene dishes, one would erroneously conclude that [18F]FDG and [18F]FAZA performs equally well as markers of severe hypoxia in SiHa cells (compare figure 1 and 2) and that [18F]FAZA displays no hypoxia-selectivity in MDAMB-231 cells (figure 1).
Another, often underappreciated, problem is that cells alter their local and global environment by consumption of tracers, O2, metabolites and possibly even drugs, which may lead to hypoxia and insufficient diffusive delivery of tracers and drugs. During long-term experiments in dense cultures alterations in medium composition (e.g., glucose) may also be a concern (see also next paragraph). For example, Pettersen et al. [12] showed that pericellular oxygen tension in confluent stagnant cultures may deviate substantially from pO2 in equilibration gasses, and concluded that this may affect any study that correlates cell biology to oxygen levels. One approach to optimize control of the pericellular microenvironment, is the use of freshly trypsinized cells maintained in shaken suspension in glass vials. This procedure has been valuable in several settings including the testing of hypoxia PET tracers [13,14], but may be less useful for delicate metabolic and gene expression studies, since several studies have shown that detached cells are stressed, and may behave differently than attached cells [15,16]. Therefore, when possible the use of attached cells are ideal. Not all cells attach firmly to glass, but coating may effectively overcome these problems. A simple, but not commonly used, way to overcome or limit diffusion barriers is the use of gentle orbital shaking, since convection will ensure fast and effective gas equilibration of medium and reduce the effective thickness of still layers. This may be particularly relevant when studying low, but non-zero, levels of oxygenation, where the relative influence by cell self-consumption may lead to large relative changes in oxygenation. Surprisingly, our results showed that [18F]FDG retention was unaffected by shaking conditions even at an O2 level of 0.5% in dense cultures. A similar result was obtained for [18F]FAZA retention in SiHa cells grown on glass (figure 1). These results suggests that even in experiments where proper control of pO2 is required non-shaking conditions may be appropriate at least when using the experimental conditions in our study. Still orbital shaking may be necessary in other settings including very dense cultures or when testing tracers/drugs that diffuses more slowly (e.g., higher MW).
A key advantage of using tracers such as [18F]FDG, 11C-acetate and 11C-methionine to assess cellular metabolism, is that measurements can be performed rapidly under reasonably stable culturing conditions. In contrast, traditional enzymatic assays determines uptake rates based on concentration changes in metabolites in medium samples, which by necessity prevents measurements of fluxes at stable medium and cell number conditions. A second problem is that detectable changes in medium composition may require long incubation periods, which leads to substantial evaporation even when using humidified gasses. In our acute experiments, the total cellular tracer retention was typically below 2% of total added dose (quantified in medium samples prior to cell harvest), suggesting that medium-volume-to-cell content was sufficient to ensure near stable conditions. Nonetheless, in many settings, such as studies on cell metabolism under chronic tumor-microenvironment-mimicking conditions with nutritional deprivation, further reducing the cell number may be advantageous or even required to dampen inappropriate changes in medium metabolites that affects tracer retention by competition (e.g., glucose, acetate) or indirectly by changing cellular metabolic state (e.g., lactic acidosis). However, very low cell numbers may compromise accuracy due to random variations in the radioactive disintegration rate and background. This may partly be circumvented by prolonging the post-harvest radioactivity measurement time or by increasing the added dose, but the latter strategy may not be acceptable due to radioactivity concerns or costs (for long-lived tracers). In accordance, we also assessed the reliability of downscaling the number of cells, without adjusting measurement time and added dose, while maintaining a high degree of confluence, by growing cells in discrete areas in polystyrene dishes or on glass cover slips. Downscaling typically resulted in an acceptable total radioactivity of 2000-3000 CPM under oxic conditions. In addition, the hypoxia induced stimulation of 18F]FDG was in agreement with results obtained when growing cells as dense cultures in whole dishes. Whether this is true in other settings and for other tracers with lower uptake (such as [18F]FAZA under oxic conditions) must be determined before a given application. An added advantage when using cover slips is that cell washing and radioactivity assessment is very easy to perform since meticulous cell scraping and cell suspension collection is not required, and that cells do not, or only slowly, spread to the Petri dish bottom. The coverslip approach is thus particular useful when doing prolonged incubations where strictly defined environmental conditions are required at a high degree of confluence. In cells on cover slips, SiHa typically retained around 0.25% of added [18F]FDG under anoxia in 1h in MEM (≈5 mM glucose). Assuming that the relative affinity for [18F]FDG retention and glucose use (the so-called “lumped constant”) is one [17] and that glucose metabolic rate is maintained under hypoglycemic conditions, then a 24h anoxic pre-incubation of SiHa cells at 1 mM glucose would lower medium glucose by approximately 30%. This quantity can be further reduced to ~7.5%, when only using a single cover slip, suggesting that by appropriate downscaling such experiments are feasible.