Early detection and metabolic pathway identification of T cell activation by in-process intracellular mass spectrometry

Abstract

Background aims: In-process monitoring and control of biomanufacturing workflows remains a significant challenge in the development, production, and application of cell therapies. New process analytical technologies must be developed to identify and control the critical process parameters that govern ex vivo cell growth and differentiation to ensure consistent and predictable safety, efficacy, and potency of clinical products.

Methods: This study demonstrates a new platform for at-line intracellular analysis of T-cells. Untargeted mass spectrometry analyses via the platform are correlated to conventional methods of T-cell assessment.

Results: Spectral markers and metabolic pathways correlated with T-cell activation and differentiation are detected at early time points via rapid, label-free metabolic measurements from a minimal number of cells as enabled by the platform. This is achieved while reducing the analytical time and resources as compared to conventional methods of T-cell assessment.

Conclusions: In addition to opportunities for fundamental insight into the dynamics of T-cell processes, this work highlights the potential of in-process monitoring and dynamic feedback control strategies via metabolic modulation to drive T-cell activation, proliferation, and differentiation throughout biomanufacturing.

Keywords: cell therapy; lab-on-a-chip; metabolic pathway; process analytical technology.

Figure 1.

The sample-to-analysis platform for at-line analysis of cell culture systems. (A) The cell processing device enables rapid pre-concentration, rinsing, and extraction of intracellular metabolites for direct-infusion ESI-MS analysis of. (B) MS chromatogram showing the total ion count, base peak, and extracted ion traces of selected amino acids. Each trace is normalized by the maximum signal intensity of the trace in the period shown. Amino acid traces are of the protonated monoisotopic mass [M+H]⁺ and are extracted with mass accuracy of 5 ppm. (C) Microfabricated cell processing device showing integrated electrodes (gold) adjacent to microfluidic channel (thin horizontal line). The transparent Borofloat cover allows for visual access to the channel during processing. (D) Zoomed-in view of the cell immobilization features after loading ca. 200 intact cells.

Figure 2.

Multivariate analyses of full scan mass spectrometry spectra reveal overall trends between sample classes and across time. (A) Comparison of media and cell lysate spectra shows distinct clustering between media and lysate signals regardless of time or activation condition; 95% confidence intervals shown. (B) Comparison of media for the activated and unactivated conditions displays scattered distribution at early time points with slight divergence along PC2 at later time points. (C) Comparison of cell lysate for the activated and unactivated conditions displays clustering regardless of activation condition at early time points. Distinct divergence of the activated condition is observed after 12 hr along the PC2 axis. (D) PLS-DA analysis of early (0–6 hr), intermediate (12–24 hr), and late (24–48) time points for the activated condition reveals progressive cluster rotation as activation proceeds (12 hr data point is common between early and intermediate-time point classifications; 24 hr data point is common between intermediate and late time point classifications); 95% confidence intervals shown. (E) The activated and unactivated conditions distinctly cluster in the heatmap with sub-clustering of the activated condition, demonstrating progressive clustering from early through intermediate to late cell states. (A-E) Label numbers correspond to the time in hours following exposure to the activating agent.

Figure 3.

PLS-DA score plots, associated VIP scores of the top twenty-five features, and volcano plots comparing the activated vs unactivated conditions at various time points. (A-C) Early time points (6–12 hr) reveal 27 upregulated and 19 downregulated features in the volcano plot. (D-F) Intermediate time points (12–24 hr) reveal 25 upregulated and 43 downregulated features in the volcano plot. (G-I) Late time points (24–48 hr) reveal 17 upregulated and 82 downregulated features in the volcano plot. Volcano plots generated with p-value threshold of 0.1 and fold change (FC) threshold of 1.5; positive values indicate upregulation for the activated condition. Features common between the VIP and those receiving putative annotation in the metabolic pathway analysis are denoted with a black box.

Figure 4.

GSEA pathway analysis of the activated vs unactivated conditions from 12–48 hr. (A) Results of pathway analysis showing metabolic pathways that are enriched for the activated (red) or unactivated (green) condition. Pathways with p-value ≤ 0.1 are shown with p-values given. (B) and (C) Example extracted ion traces of metabolites identified in multiple enriched pathways for the (B) activated and (C) unactivated conditions with pathways given in parentheses. Each trace is normalized by the maximum signal intensity of the trace in the period shown. Succinic acid and glutamic acid traces are sodiated monoisotopic masses [M+Na]⁺; all other traces are protonated monoisotopic masses [M+H]⁺. All analyte traces are extracted with mass accuracy of 5 ppm.

Figure 5.

Culture dynamics as captured by standard characterization methods. (A) Initial drop in viability is observed regardless of the activation condition, with activated cells exhibiting greater overall reduction, after which viability began to increase. Lactate levels exhibit negligible change for the initial 24 hours following antigen exposure after which the levels in the activated system steadily increase. (B) Secreted factors in the media were measured and display increased concentrations following activation. Values above/below calibration range are saturated at the calibration limit (upper limit denoted by dashed line). IL-1 beta is not shown as it was below the detectable range at all time points. (C) Surface marker analysis of the activated condition at 0, 24, 48, and 72 hours as obtained by flow cytometry for both the activated (top) and unactivated (bottom) conditions; error bars represent one standard deviation from the mean as calculated from technical replicates (n=3). *Indicates variation of surface maker expression relative to the initial condition with p-value < 0.05 as calculated using two-tailed student t-test; ranges include 0 hr for simplicity.