Enhanced detection of antigen-specific T cells by a multiplexed AIM assay

Abstract

Broadly applicable methods to identify and characterize antigen-specific CD4⁺ and CD8⁺ T cells are key to immunology research, including studies of vaccine responses and immunity to infectious diseases. We developed a multiplexed activation-induced marker (AIM) assay that presents several advantages compared to single pairs of AIMs. The simultaneous measurement of four AIMs (CD69, 4-1BB, OX40, and CD40L) creates six AIM pairs that define CD4⁺ T cell populations with partial and variable overlap. When combined in an AND/OR Boolean gating strategy for analysis, this approach enhances CD4⁺ T cell detection compared to any single AIM pair, while CD8⁺ T cells are dominated by CD69/4-1BB co-expression. Supervised and unsupervised clustering analyses show differential expression of the AIMs in defined T helper lineages and that multiplexing mitigates phenotypic biases. Paired and unpaired comparisons of responses to infections (HIV and cytomegalovirus [CMV]) and vaccination (severe acute respiratory syndrome coronavirus 2 [SARS-CoV-2]) validate the robustness and versatility of the method.

Keywords: 4-1BB (CD137); Antigen-specific CD4(+) T cells; CD40L (CD154); CD69; CP: Immunology; OX40 (CD134); activation-induced marker (AIM) assay; antigen-specific CD8(+) T cells; flow cytometry; infection-induced T cell responses; vaccine-induced T cell responses.

Graphical abstract

Figure 1

Quantification of Ag-specific CD4⁺ and CD8⁺ T cell responses using Boolean OR gating (A) Representative FACS plots depicting the multiplexed (6xAIM) strategy to identify Ag-specific CD4⁺ (orange gate) and CD8⁺ (blue gate) T cells. For simplicity, the example focuses on the HIV Gag stimulation. (B) FACS plots illustrating the AND/OR Boolean gating strategy. For simplicity, the example focuses on the AIM pairs CD69⁺OX40⁺ (blue population) and CD69⁺CD40L⁺ (red population). (C and D) Raw frequencies of (C) AIM⁺ CD4⁺ and CD8⁺ T cell responses following an ex vivo 15 h stimulation of PBMCs with a pool of HIV Gag, Pol, Envelope (Env), or Nef peptides, or (D) following stimulation with HCMVA pp65 peptides. PBMCs were left unstimulated as a control (gray bars). Numbers of responders reaching >2× over the unstimulated condition are written below the histograms for each stimulation. The median 6xAIM values and Wilcoxon tests are shown. (C) n = 16 ART participants; (D) n = 15 CMV⁺ ART participants.

Figure 2

The multiplexed 6xAIM assay improves detection of HIV- and CMV-specific CD4⁺ T cells (A) Net HIV-specific CD4⁺ and CD8⁺ T cell responses. The net HIV Gag, Pol, Env, and Nef responses were summed to assess the “total” HIV-specific responses. (B) Net CMV-specific CD4⁺ and CD8⁺ T cell responses. (C) Comparisons between the net responses of the 6xAIM strategy and benchmark AIM pairs to detect Ag-specific CD4⁺ and CD8⁺ T cell responses. Medians, median fold increases (FIs), and Wilcoxon tests are shown. (D) Proportion of background signal in HIV Ag-stimulated conditions. The Ag-stimulated conditions are normalized to 100% to better represent the autologous unstimulated signal. Light colors represent the total of stimulated cells; dark colors represent the background. (E) Donut charts depicting the proportion of participants for whom each AIM pair yielded the highest magnitude of HIV-specific CD4⁺ and CD8⁺ T cell responses. (F) Individual contributions of each AIM marker for the detection of HIV-specific CD4⁺ and CD8⁺ T cell responses. The detection levels were normalized to the value yielded by the 6xAIM. Light colors represent the signal lost when the indicated AIM marker was removed from the analysis; dark colors represent the residual signal still detected. (G) Spider charts depicting the comparisons between the normalized amplitudes of the net HIV- (teal) and CMV-specific (amber) CD4⁺ and CD8⁺ T cell responses. Within each cohort, responses were normalized to their 6xAIM values. Medians and Wilcoxon tests are shown for each pair (see also Figure S2H). ∗p < 0.05, ∗∗p < 0.01, and ∗∗∗p < 0.001. Matched donors who had at least one HIV stimulation and the CMV stimulation reaching >2× over the unstimulated conditions were kept. For CD4⁺ and CD8⁺ T cells, n = 13 CMV⁺ ART-matched participants. (A and B) The 6xAIM and individual AIM pairs are ranked in decreasing order of the median, based on their respective CD4⁺ T cells. The tables above the graphs represent the pairwise multiple comparisons performed using the Friedman test with Dunn’s post hoc test. (D and F) Pairwise multiple comparisons were performed using the Friedman test with Dunn’s post-hoc test. (A, B, D, and F) The bars represent medians ± interquartile ranges. Thicker borders represent donors who had at least one HIV stimulation, or had the CMV stimulation, reaching >2× over the unstimulated conditions. (A, C, and D–F) n = 16 ART participants; (B) n = 15 CMV⁺ ART participants.

Figure 3

CD69, 4-1BB, OX40, and CD40L are differentially expressed on HIV-specific CD4⁺ T cell subsets Phenotypic analysis of HIV-specific CD4⁺ T cell responses. (A) Proportion of 6xAIM⁺ HIV-specific CD4⁺ T cells expressing CD69, 4-1BB, OX40, and CD40L in univariate analyses (see also Figure S3A for gating strategy). (B) Multiparametric global UMAP representation of 6xAIM⁺ HIV-specific CD4⁺ T cells. The colors identify 12 populations clustered by unsupervised analyses and labeled on the UMAP. These populations could be further grouped in “superclusters” driven by their chemokine receptor expression. Clusters that were not polarized toward CCR6⁺, CXCR3⁺, or CXCR5⁺ T cells were classified as “no polarization.” (C) Heatmap summarizing the mean fluorescence intensity (MFI) of each loaded parameter. (D) Relative frequency of each identified cluster within 6xAIM⁺ HIV-specific CD4⁺ marker⁺ T cell subpopulations (colored dots) compared to within the reference parental 6xAIM⁺ cells (white dots). Arrows pointing to the right indicate that a population is overrepresented in the HIV-specific population compared to the reference CD4⁺ T cells, while arrows pointing to the left indicate an underrepresentation. Medians and only significant results with the Wilcoxon tests (see also Figure S3C) are shown. ∗p < 0.05, ∗∗p < 0.01, and ∗∗∗p < 0.001. (E) Example of univariate CCR6⁺, CXCR3⁺, and CXCR5⁺ gatings on total and HIV Gag-specific CD4⁺ T cell populations. (F) Enrichment scores of the AIMs in the Gag-specific CD4⁺ CCR6⁺ (green), CXCR3⁺ (yellow), and CXCR5⁺ (purple) T cell populations. The scores were calculated by dividing, for each individual AIM, the Gag-specific CD4⁺ phenotype⁺ population by the Gag-specific CD4⁺ population, irrespective of the polarization (see also Figure S3D). Wilcoxon tests (see also Figure S3E) are shown. ∗p < 0.05, ∗∗p < 0.01, and ∗∗∗p < 0.001. (G) Proportion of HIV-specific CCR6⁺ and CXCR5⁺ CD4⁺ T cells still detected after ad hoc removal of the indicated AIM molecules. The values were normalized to those of the 6xAIM. Wilcoxon tests are shown. (A, F, and G) Medians ± interquartile ranges are shown. (A and F) Pairwise multiple comparisons were performed using the Friedman test with Dunn’s post-hoc test. (D and F) The graphs are on a log₂ scale. To avoid contaminating phenotype profiling with excessive background, donors who had at least one HIV stimulation (A–D) or the HIV Gag stimulation (F and G) reaching >2× over the unstimulated condition were kept. (A–D) n = 14 ART participants; (F and G) n = 13 participants.

Figure 4

Infection-specific (HIV) and vaccine-induced (SARS-CoV-2) responses are detected by the same AIM pairs (A) Net SARS-CoV-2 Spike-specific CD4⁺ (orange) and CD8⁺ (blue) T cell responses. The 6xAIM and individual AIM pairs are ranked in decreasing order of the median, based on the CD4⁺ T cells. The tables above the graphs represent the pairwise multiple comparisons performed using the Friedman test with Dunn’s post hoc test. (B) Individual contributions of each AIM marker for the detection of Spike-specific CD4⁺ (left) and CD8⁺ (right) T cell responses. The detection levels were normalized to the value yielded by the 6xAIM. Light colors represent the signal lost when the indicated AIM marker is removed from the analysis; dark colors represent the residual signal still detected. Pairwise multiple comparisons were performed using the Friedman test with Dunn’s post-hoc test. (C) Spider charts depicting the comparisons between the normalized amplitudes of net HIV- (teal) and Spike-specific (lavender) CD4⁺ and CD8⁺ T cell responses. Within each cohort, responses were normalized to their 6xAIM values. Medians and Mann-Whitney tests are shown for each pair (see also Figure S4D). ∗p < 0.05, ∗∗p < 0.01, and ∗∗∗p < 0.001. Donors who had at least one HIV stimulation reaching >2× over the unstimulated conditions were kept, as well as those whose Spike stimulation reached the same threshold. For CD4⁺ T cells, n = 14 ART participants and n = 23 SARS-CoV-2-vaccinated participants; for CD8⁺ T cells, n = 15 ART participants and n = 16 SARS-CoV-2-vaccinated participants. (A and B) The bars represent medians ± interquartile ranges. Thicker borders represent donors whose stimulation reached >2× over the unstimulated conditions. n = 23 SARS-CoV-2-vaccinated participants.