Metabolic pathways analysis gains considerable interest in the field of immunology because of its impact on immune cell effector function. Seahorse technology makes these analysis easy to implement in every lab, as soon as it is provided with the equipment. However, many critical environmental parameters can affect data reliability and including appropriate experimental controls is critical to be confident in scientific results. In this article, we described a simple procedure to standardize metabolic analysis using XFe96 Seahorse analyzer.
We sought to validate a method involving JURKAT cell line as an internal control for metabolic potential assessment according to ICHQ2 (R1) guidelines. In this setting, we confirmed that the optimal concentration of JURKAT, in our setting, is 100 000 cells/well, corresponding to an about 90 % cell confluence, as recommended by Agilent Technologies and Plitzko et al. [26]. The basal OCR and ECAR values obtained for this concentration are comprised between 100 and 300 pmol/minutes and greater than 20 mpH/minutes respectively. These values are considered as acceptable by Divakaruni et al. [24] and TeSlaa et al. [15]. The g-intercept we obtained for basal ECAR during method linearity studies was around 18, strenghtening TeSlaa’s statement. For this concentration, basal OCR/ECAR ratio is less than 4, allowing us to conclude that ECAR parameter is mainly glycolysis related. This conclusion is supported by the limited effect of FCCP on ECAR measurement. Nevertheless, it should be kept in mind that the relative contribution of respiration and glycolysis to ECAR is likely to depend on cellular type, evaluated cell lines respiratory proton production rate ranging from about 9 to 76 % [27].
OCR and ECAR profiles obtained for JURKAT cell line at basal state and after oligomycin and FCCP exposure are in accordance with expected profiles, given the mechanism of action described for these two compounds. These profiles are also in accordance with those described in other models [26]. It is well known that phenotypically less differentiated and less activated T cells showed a more OXPHOS-dependent basal metabolism and a minority resort to glycolysis than more differentiated and activated T cells [28]. Concomitantly, our team realized a study about cultured T cells differentiation status (Marton et al., unpublished data). Thus, authors demonstrated that metabolic potential of T cells is correlated with their differentiation profile evaluated by flow cytometry. Taken together, these data confirmed the appropriate capacity of the evaluated method to measure relevant analytes. Moreover, our experimental observations enabled us to invalidate the hypothesis according to which presence of impurities in assay medium was likely to interfere with analytes measurement. Furthermore, T cell culture conditions with a confluence equal to about 20 % resulted in an absence of detected signal in term of OCR or ECAR (data not shown). Altogether, capacity to identify analytes in our experimental model and adequate impurities management permitted to conclude that specificity of the method is conform to expectations.
Excellent accuracy of the Agilent’s provided method was confirmed by calculating the systematic bias of the method associated to Seahorse calibration data. Thus, the bias was largely less than 5 % for both O2 and pH emission.
Concerning method precision assessment, we first showed that CV are inferior to 15 %, the cut-off value of CV associated with a good repeatability. Then, CV values calculated for intermediate precision are all inferior to 30 % and are considered as acceptable. In this last setting, despite a thick 95 % confidence interval, we can conclude that precision of the method is consistent with expected performance. Yépez et al. [21] studied the OCR of adherent primary fibroblasts DHNF using Seahorse technology and demonstrated the precision of the method. The CV for repeatability and intermediate precision evaluation are respectively similar and higher than those we obtained in this study. It indicates that the precision of our method is similar to those of this already published method, although we seeded suspension cells instead of adherent cells. It particularly highlights the fact that the step of cells attachment, in our study, is appropriate and well controlled.
According to the acceptance criteria chosen for the determination coefficient r2, i.e. 0.92, we demonstrated that the method is linear for a JURKAT concentration comprised between 25 000- 150 000 cells/well. This acceptance criteria is particularly stringent, since, as a rule of thumb, a strong positive correlation has an r-value more than 0.7 [29], demonstrating the analytic relevance of our method. Moreover, the range within which precision is acceptable is comprised between 50 000-150 000 JURKAT/well. Plitzko et al. [26] studied metabolic potential of different cell lines using a Seahorse XF24 analyzer. Authors determined a seeding density likely to induce OCR and ECAR values being within the linear response. They described, at basal state, acceptable linearity of their method and an optimal density seeding of 20 000 and 35 000 cells/well respectively for melanoma cell lines and colon-derived cell line. These values are largely inferior to those we determined for JURKAT cell line for an about 90 % confluence. It could be explained by the cell size and the larger place held by adherent cells on the plastic support. It is notable that JURKAT cell line occupy the plastic area with a confluence of about 90 % at a concentration of 100 000 cells/well. We could expect that a cell concentration of 150 000 cells/well results in an over-confluence of cells. It is finally not the case, the achieved confluence is quite similar to those previously obtained, because of an obvious cytosolic retraction of the attached JURKAT cells. Similar observations have already been reported by Luciani et al. concerning HeLa cell line [30]. Nevertheless, it seems that, at a dose of 200 000 cells/well, JURKAT are not able to contract their cytosol enough to maintain their plastic attachment, inducing a turbidity in wells during the transient micro-chamber formation, likely to interfere with analytes detection. We could assume that the determined range of the method is closely related to confluence: this one could be insufficient under 50 000 JURKAT/well and too important over 150 000 JURKAT/well.
Indeed, we demonstrated here the specificity, the precision, the linearity of the method involving JURKAT cell line as an internal control in the setting of metabolic potential assessment and according to ICHQ2 (R1) [22]. Furthermore, we determined the range with acceptable precision and linearity and confirmed the accuracy of the standard provided method.
The use of JURKAT cell line as experimental internal control and the validated method proposed here could be further extended to even better match with potential user scientific inquiries.
Analysis of inter-plate variability could be further controlled, as described by Yépez et al. [21], by using a mathematical modeling of the inter-assay variability of the internal control, then integrated in experimentally determined data values. In our assay, we took into account for analysis all three sequential measured values obtained for each basal and post compounds injections. According to Divakaruni et al. [24], it could be appropriate to analyze only the minimum or last measurement in the presence of oligomycin because its bioavailability can often result in a time-dependent effect. Similarly, only the maximum measurement after FCCP exposure, due to potentially occurring cell consumption of the molecule, should be considered. This kind of analysis could improve the method precision even more. In our study, we performed all assays using a standard concentration of FCCP, i.e. 1.5 µM. A further characterization of respiratory reserve and capacity could be realized by assaying a dose escalation of FCCP, as advised by Van der Windt et al. [31].
Cell resort to OXPHOS and glycolysis could be further investigated using the Seahorse technology by studying the metabolic impact of other compounds. Indeed, the injection of rotenone plus antimycin A, targeting the electron transport chain and by this way reducing OCR to a minimal value, would allow the evaluation of proton leak-linked respiration and a more refined analysis of basal mitochondrial respiration. Rotenone and antimycin A addition would likewise enable to assess qualitative contribution of OXPHOS-related CO2 production to ECAR, as described by Divakaruni et al. [24]. More specific glycolysis assessment could be implemented by basal ECAR evaluation performed using a glucose non-containing assay medium, then by sequential adding of glucose, oligomycin and 2-deoxy-glucose (2DG) in ports A, B and C of the Seahorse cartridge. Thereby, measuring ECAR just after glucose exposure is likely to refine basal glycolysis analysis and 2DG exposure, by blocking glucose cell uptake and utilization, would contribute to evaluate more slightly non-glycolysis related medium acidification [25]. Taking these considerations into account, further investigations could be performed to extend the validation of the use of JURKAT cell line as an internal quality control for metabolic potential determination.
It is important to note that a relevant control of fitness and proliferative potential of JURKAT cells should be applied. Here, we demonstrated that an appropriate control of passage number and mycoplasma contamination [32], as well as viability, seeding and attachment of JURKAT cells are likely to provide acceptable method precision and linearity. Methods and strategies to further normalize XF metabolic data to cellular parameters are available and could be used to improve these key validation parameters and ensure even more accurate results [33]. Total cellular protein assay is a quick and inexpensive method to normalize data but it is also not applicable if there are significant variations in the amount of extracellular matrix protein present among different experimental groups or if plates are coated with protein derived compounds, as poly-D-lysine. Nuclear DNA quantification represents an alternative to protein assessment and is commonly admitted to be correlated linearly with cell number. Cells counting remains the most robust normalization method. It involves cells counting in each well of the microplate via direct imaging or staining cell nuclei.
To conclude, we validated a method using JURKAT cell line as internal quality control to manage the inter-assay variability of a Seahorse technology based-method. This would allow researchers to compare independent experiments and to improve the robustness of the method. The metabolism analysis methodology developed here presents an adaptation potential to a myriad of scientific inquiries in the setting of T cell metabolism studies, either in term of pathways evaluation or in term of analysis refinement. Our study could represent the first founding element to the use of Seahorse technology to evaluate metabolic potential as a monitoring parameter of ATMP such CAR T cells, in a GMP-compliant environment. Well characterized samples of JURKAT cell line could constitute a reference material to standardize metabolic potential analysis, confirm the accuracy of the method in routine analysis and evaluate the reproducibility of the method between different ATMP quality control laboratories.