Single-cell multiplexed cytokine profiling of CD19 CAR-T cells reveals a diverse landscape of polyfunctional antigen-specific response

Abstract

Background: It remains challenging to characterize the functional attributes of chimeric antigen receptor (CAR)-engineered T cell product targeting CD19 related to potency and immunotoxicity ex vivo, despite promising in vivo efficacy in patients with B cell malignancies.

Methods: We employed a single-cell, 16-plex cytokine microfluidics device and new analysis techniques to evaluate the functional profile of CD19 CAR-T cells upon antigen-specific stimulation. CAR-T cells were manufactured from human PBMCs transfected with the lentivirus encoding the CD19-BB-z transgene and expanded with anti-CD3/anti-CD28 coated beads. The enriched CAR-T cells were stimulated with anti-CAR or control IgG beads, stained with anti-CD4 RPE and anti-CD8 Alexa Fluor 647 antibodies, and incubated for 16 h in a single-cell barcode chip (SCBC). Each SCBC contains ~12,000 microchambers, covered with a glass slide that was pre-patterned with a complete copy of a 16-plex antibody array. Protein secretions from single CAR-T cells were captured and subsequently analyzed using proprietary software and new visualization methods.

Results: We demonstrate a new method for single-cell profiling of CD19 CAR-T pre-infusion products prepared from 4 healthy donors. CAR-T single cells exhibited a marked heterogeneity of cytokine secretions and polyfunctional (2+ cytokine) subsets specific to anti-CAR bead stimulation. The breadth of responses includes anti-tumor effector (Granzyme B, IFN-γ, MIP-1α, TNF-α), stimulatory (GM-CSF, IL-2, IL-8), regulatory (IL-4, IL-13, IL-22), and inflammatory (IL-6, IL-17A) functions. Furthermore, we developed two new bioinformatics tools for more effective polyfunctional subset visualization and comparison between donors.

Conclusions: Single-cell, multiplexed, proteomic profiling of CD19 CAR-T product reveals a diverse landscape of immune effector response of CD19 CAR-T cells to antigen-specific challenge, providing a new platform for capturing CAR-T product data for correlative analysis. Additionally, such high dimensional data requires new visualization methods to further define precise polyfunctional response differences in these products. The presented biomarker capture and analysis system provides a more sensitive and comprehensive functional assessment of CAR-T pre-infusion products and may provide insights into the safety and efficacy of CAR-T cell therapy.

Keywords: CD19 CAR-T cell product; Microfluidic microdevice; Polyfunctionality; Precision profiling; Single-cell proteomics.

Fig. 1

Multiplexing single cell measurement of CAR-T cells in microchambers. a Schematic outline of CD19TL19 CAR-T cell generation, sorting and stimulation. The CD19-BB-z transgene lentiviral vector was used for CAR-T cell generation. CAR-T cells were sorted and then stimulated by anti-CAR beads prior to the SCBC assay. b The validated 16-plex panel including 4 groups of cytokines: effector, stimulatory, regulator and inflammatory. c Major work flow for single-cell functional proteomic analysis. (i) Schematic depiction of a microchamber fabricated in PDMS used for isolating single cells and assaying a panel of proteins/cytokines secreted from the entrapped cell with an antibody barcode array patterned on the glass slide. Each device has 12,000 microchambers. (ii) Motorized miscopy allows for automated imaging of the entire microchamber PDMS device for locating and counting T cells in microchambers. Protein secretion profile is obtained by quantifying the fluorescence signals corresponding to single-cell secretions in each microchamber. Overlay of these two data sets allows for the identification of single-cell protein secretion profiles. (iii) Quantitative analysis, statistics and advanced informatics (e.g., CytoSpeak package) were applied in this project to investigate the effector function (cytokine) landscape of single CAR-T cells. d A representative measurement of IFN-γ in supernatants from CAR-T cells stimulated by either IgG beads or anti-CAR beads by ELISA

Fig. 2

CAR-T cells show high polyfunctionality in anti-tumor effector and stimulatory functions. a Polyfunctional breakdown of CD3+, CD4+ and CD8+ T cells at the single-cell level across 4 donors. The T cells of donors 1, 3 and 4 show a 6 to 47-fold increase in overall polyfunctionality when stimulated with anti-CAR beads, compared to IgG stimulation. By comparison, donor 2 only shows a 2-fold increase. b Polyfunctional strength index (PSI) computed for CD3+, CD4+, and CD8+ T cells at the single-cell level across 4 donors. The profiles of donors 1, 3 and 4 is dominated by effector and stimulatory cytokine subsets. c Heat map and dendrogram visualization applied to the CD3 T cell secretion data. There is one heat map per donor, and each column corresponds to a single cytokine, while rows correspond to individual cells. Non-secreting cells are excluded from the heat maps. The colors indicate log transformed secretion intensities (red = low, green = high). At a high level, this visualization illustrates some differences across donors and which cytokines are commonly secreted in tandem. However, the clustering is done individually per donor, and it is difficult to map clusters to functional subsets

Fig. 3

Higher dimensional data is difficult to visualize concisely. a With highly-multiplexed single-cell data, displaying the breakdown of different functional groups being secreted by a sample can increase by up to a factor of 2× with the addition of x cytokines. When this analysis is performed across a set of donors or stimulation conditions, effectively highlighting the key secretion differences is challenging. In this standard bar graph visualization of functional groups secreted by CD4+ CAR-T cells of four donors, it is cumbersome to see which are the major functional groups being secreted by each donor, and what are the biggest fold differences across donors. b-c Reducing the dimensionality of the dataset is a different approach to more effective and understandable visualizations. PCA (principal component analysis) uses an orthogonal transformation to convert the original dataset into a set of linearly uncorrelated principal components, where the number of components is smaller than the number of original variables. The transformation is defined in such a way that the first principal component has the largest possible variance (accounting for as much variability as possible within the dataset), followed by the second component, and so on. While reducing the dimensionality to two principal components may still result in some loss of information, the benefit is that the transformed data points can then be visualized on a two-dimensional scatterplot. In this figure, PCA is applied to the 4-donor CD4+ CAR-T secretion dataset. Each cell’s secretions (signal intensity of each cytokine) are log transformed prior to dimensionality reduction. b is color-coded by donor, while c is color-coded by some of the individual cytokines. The combination of these graphs reveals some information, such as the low overall polyfunctionality of donor 2, and the high Granzyme B + MIP-1a + polyfunctionality of Donor 4. However, more detailed information about upregulated and/or distinct polyfunctional subsets is less clear

Fig. 4

Polyfunctional heat map and PAT PCA reveal distinct CD4+ CAR-T cell profiles across donors. a Polyfunctional heat map displaying major functional subsets secreted across the 4 donors’ CD4+ CAR-T samples. Hierarchical clustering is applied to attain a condensed set of functional groups that still faithfully represent the overall profile of the donors. The color-coding indicates how commonly each donor secrets the corresponding functional group/cluster. Donor 1, closely followed by donor 4, has the highest frequencies of most expressed functional groups. Donor 3 is less polyfunctional, while donor 2 has virtually no secreted polyfunctional groups. The group GM-CSF, Granzyme B, IL-13 and TNF-α is expressed exclusively by the CD4+ CARs of donors 1 and 4, but not by the CARs of donor 2 or donor 3. Similarly, the 7-plex group containing GM-CSF, Granzyme B, IFN-γ, IL-8, IL-13, MIP-1α, and TNF-α is unique to these two donors. Functional groups not containing GM-CSF or IL-13 are expressed at similar frequencies by donor 3 as they are by donors 1 and 4. b PAT PCA visualization of the same dataset. Data points are color-coded based on donor. Those representing the same functional group are randomly offset, but remain within a radius proportional to the secretion frequency of the corresponding group (i.e., large groups = large circles, small groups = small circles). The principal components are labeled according to their correlation with specific cytokines. The lack of donor 2 (orange) subsets indicates the lower polyfunctionality of this sample, while the presence of numerous donor 1 (blue) and 4 (green) groups in the right area of the graph indicates the highly-polyfunctional makeup of these two samples. Donor 3 has generally less polyfunctional subsets, typically including combinations of Granzyme B, MIP-1α, IL-8, and TNF-α but lacking IFN-γ, IL-13, and GM-CSF. Donor 4 largely spans the polyfunctional profiles of both donors 1 and 3

Fig. 5

Polyfunctional heat map and PAT PCA reveal distinct CD8+ CAR-T cell profiles across donors. a Polyfunctional heat map displaying major functional subsets secreted across the 4 donors’ CD8+ CAR-T samples. Hierarchical clustering is applied to attain a condensed set of functional groups that still faithfully represent the overall profile of the donors. The color-coding indicates how commonly each donor secrets the corresponding functional group/cluster. Donors 1 and 4 are significantly more polyfunctional than donors 2 and 3, and also uniquely secrete the 7-plex group containing GM-CSF, Granzyme B, IFN-γ, IL-8, IL-13, MIP-1α, and TNF-α. Donor 1 has a higher number of unique polyfunctional groups than donor 4, particularly groups not containing IL-8. The only polyfunctional groups secreted by all four donors contain Granzyme B, MIP-1 α with smaller amounts of IFN- γ. b PAT PCA visualization of the same dataset. Data points are color-coded based on donor. Those representing the same functional group are randomly offset, but remain within a radius proportional to the secretion frequency of the corresponding group (i.e., large groups = large circles, small groups = small circles). The principal components are labeled according to their correlation with specific cytokines. Like the CD4+ CAR-T samples, donor 2 (orange) has low polyfunctionality, donors 1 (blue) and 4 (green) are highly polyfunctional. Donor 3 has generally less polyfunctional subsets often comprised of combinations of Granzyme B, MIP-1α, IL-8, and TNF-α but lacks IFN-γ, and IL-13. Donor 4 largely spans the polyfunctional profiles of both donors 1 and 3, which can also be seen in the heat map