Development of a unified mass cytometry assay integrating activation-induced markers (AIMs) and cytokine profiling for high-throughput assessment of antigen-specific T cell responses

ABSTRACT

Accurate profiling of antigen-specific T cells by measuring T cell activation markers and cytokine production has been limited by inconsistent marker usage across studies and substantial methodological variability. To overcome these challenges, we established a unified mass cytometry (CyTOF) platform that integrates optimized activation-induced marker (AIM) panels, intracellular cytokine staining (ICS), and a cadmium-based barcoding system for high-throughput, multiplexed analysis of T cell responses. We optimized stimulation conditions (18–24 h) and validated highly sensitive dual AIM combinations, including CD25⁺CD134⁺ and CD25⁺CD69⁺ in CD4⁺ T cells, as well as CD25⁺CD137⁺ and CD25⁺CD69⁺ in CD8⁺ T cells. These combinations were stable under protein transport inhibition and closely paralleled cytokine-producing populations. In addition, we developed a cadmium-tagged β2 M/CD298 barcoding strategy that supports 21-plex sample processing without signal distortion. Validation in two independent cohorts confirmed the platform’s high sensitivity, reproducibility, and its ability to simultaneously detect AIM⁺ T cells and cytokine-producing subsets. Together, this integrated workflow provides a robust and scalable framework for comprehensive immune monitoring in vaccine development and infectious disease research.

GRAPHICAL ABSTRACT

Introduction

Precise identification and functional characterization of antigen-specific T cells are fundamental for advancing vaccine development, immune monitoring, and therapeutic interventions against infectious diseases and cancer. Established assays such as Enzyme-Linked Immunospot (ELISPOT), intracellular cytokine staining (ICS), and activation-induced marker (AIM) profiling have significantly advanced our understanding of T cell specificity and function. However, each method has inherent limitations. ELISPOT is highly sensitive and widely used in clinical trials for quantifying cytokine-secreting T cells, but it provides no phenotypic information.^1,2 ICS enables per-cell cytokine quantification and phenotype characterization, but typically targets a restricted cytokine panel, potentially underestimating polyfunctional responses.^3,4 AIM assays identify surface activation markers that are upregulated upon antigen engagement^5,6, enabling sensitive cytokine-independent detection of antigen-specific T cells across diverse subsets. However, they do not directly assess effector functions.^7,8 These constraints are further compounded by inconsistent AIM selection across studies^9–17 – ranging from CD40L⁺CD200⁺ to OX40⁺CD25⁺ or CD69⁺CD40L⁺ – hindering cross-study comparability and the establishment of universal correlates of protective immunity.

Given these limitations, recent studies have emphasized the need to combine activation markers with functional readouts to more completely capture the breadth of antigen-specific T cell responses.^18,19 For example, Binayke’s meta-analysis of antigen-specific T cell responses¹⁸ revealed that integrating AIM and ICS assays yields a more comprehensive understanding of T cell activation and functionality than either assay alone. In a similar vein, Bowyer et al.¹⁹ highlighted substantial variability in AIM selection – such as CD25, CD134(OX40), PD-L1(CD274), CD107a, and CD40L – that undermines comparability across studies and suggests that no single assay can fully capture the complexity of polyfunctional T cells elicited by vaccines or infections. Nevertheless, no standardized and sensitive protocol has yet been established for the simultaneous detection of AIM and ICS, particularly in clinical cohort samples. These findings highlight the urgent need for high-dimensional, standardized platforms capable of concurrently profiling activation status, effector function, and phenotypic features of antigen-specific T cells.

Although conventional and spectral flow cytometry are capable of measuring 30 or more parameters and are certainly applicable for AIM/ICS assays. These workflows typically require extensive panel optimization and remain susceptible to spectral spillover and fluorescence spread, especially in large clinical cohorts. However, mass cytometry (CyTOF) presents a transformative solution to the specific limitations of spectral overlap and compensation inherent in fluorescence-based techniques.²⁰ By overcoming these issues, CyTOF enables the simultaneous measurement of over 40 parameters with minimal background, encompassing surface activation markers, intracellular cytokines, and signaling intermediates at a higher resolution for low-abundance targets. In our study, we directly address the historical inconsistency in AIM selection by systematically evaluating a comprehensive panel of commonly used markers – CD69, CD137, OX40, CD25, CD40L, PDL1, CD200, and CD107a – to select an optimized, universal AIM panel with maximal sensitivity and broad applicability. Critically, this optimized AIM panel is combined with ICS within a unified CyTOF framework, establishing a single multiplexed assay capable of concurrently resolving both the phenotypic and functional dimensions of antigen-specific T cells.

Importantly, our protocol includes metal-isotope cell barcoding (MCB) to enable multiplexed sample processing. MCB allows multiple samples to be barcoded prior to staining and acquistion, significantly increasing throughput, reducing antibody consumption by 10- to 100-fold, and minimizing technical batch effects.^21,22 This enhancement is indispensable for large-scale vaccine trials, and longitudinal immune monitoring. Moreover, CyTOF offers heightened sensitivity for quantifying low-abundance AIMs and cytokines by eliminating spectral overlap. Its use of rare-metal isotopes provides a near-zero background signal, improving detection resolution, especially across heterogeneous cell population.²³ By integrating MCB, AIM assays, and ICS within a unified CyTOF workflow, we establish a scalable, high-dimensional assay that concurrently resolves both phenotypic activation markers and functional cytokine responses. This combined approach addresses key reproducibility challenges and enhances cross-study comparability, setting a new benchmark for antigen-specific T cell analysis in immunology.

Materials and methods

Blood samples collection and ethical considerations

Twenty-four healthy adult volunteers (18–59  years of age) with no prior SARS-CoV-2 infection were recruited from the Affiliated Hospital of Yunnan University and received two doses of COVID-19 inactivated vaccine between September 2022 and February 2023. Among them, 12 individuals received a third dose of COVID-19 mRNA vaccine in June 2023, and these samples were used for methodological optimization, while the remaining 12 individuals received only two doses of inactivated vaccine and were used for validation of the optimized workflow.

In addition, 46 healthy adult volunteers were recruited from the Jiangsu Provincial Center for Disease Control and Prevention and received two doses of COVID-19 inactivated vaccine between January 2022 and March 2023. A subset of these individuals experienced SARS-CoV-2 infection between June and December 2023, and all subsequently received a third dose of the COVID-19 mRNA vaccine in March 2024.

Whole blood samples were collected from study participants at the Affiliated Hospital of Yunnan University and the Jiangsu Provincial Center for Disease Prevention and Control – after obtaining written informed consent. The study was approved by the Ethics Committee of the Affiliated Hospital of Yunnan University (2022026) as well as the Jiangsu Provincial Center for Disease Prevention and Control (JSJK2023-B004-02). All procedures involving human blood complied with the Declaration of Helsinki. PBMCs were isolated from heparinized whole blood using a density gradient separation medium (Stemcell) and cryopreserved in fetal bovine serum (FBS, Gibco) containing 10% dimethyl sulfoxide (Sigma-Aldrich) until analysis.

Antibody conjugation and titrations

Surface and intracellular metal–antibody conjugations were performed using Maxpar X8 Antibody Labeling Kits (Standard Biotools). Briefly, 100 μg of antibody was loaded into a 50 kDa filter (Millipore), adjusted to 400 μL with R-Buffer, and centrifuged at 12,000 × g for 10 min. Samples exceeding 400 μL were pre-concentrated before adjustment. A 4 mM TCEP solution (8 μL of 0.5 M stock in 992 μL R-Buffer) was added (100 μL per 100 μg antibody) for partial reduction, followed by incubation at 37°C for 30 min and washing with C-Buffer (300 μL, then 400 μL) with centrifugation. Partially reduced antibodies were incubated with Ln-loaded X8 polymer (resuspended in 60 μL C-Buffer) at 37°C for 90 min. Conjugates were washed four times with W-Buffer (200 μL once, 400 μL three times), diluted in 80 μL W-Buffer, quantified by NanoDrop, and centrifuged to remove the residual buffer. Antibody Stabilizer PBS (Boca Scientific) was added to reach 0.5 mg/mL. Conjugated antibodies were stored in Protein LoBind tubes (Eppendorf) at 4°C until titration, diluted to the optimal working concentration, and kept at 4°C. Final titers were determined by the highest S/N or D/W ratio.

For the anti-human β2-microglobulin (β₂M) antibody (Biolegend) and anti-human CD298 antibody (Biolegend) used for cell barcoding, another commercial metal-conjugated antibody kits Maxpar MCP9 Antibody Labeling Kits, Standard Biotools were used to bridge the metal (Cd) and antibodies. Partial antibody reduction was performed similarly to the Maxpar X8 Antibody Labeling Kits protocol. For washing after metal conjugation, the labeled antibodies were washed using a 100 kDa filtration unit (Millipore) centrifuged at 5000 × g for four times. After the final wash, the conjugate is diluted with 75 μL W-Buffer, mixed, and quantified using a NanoDrop spectrophotometer. The filtration unit is centrifuged again to remove the W-Buffer. Finally, the volume of HRP-Protector peroxidase stabilizer (Boca Scientific) needed to achieve a final concentration of 0.5 mg/mL is calculated and added to the filtration unit. The conjugated antibody is transferred to a Protein LoBind tube, sealed, and stored at 4°C until titration. After titration, the antibody is diluted to the optimal working concentration with the HRP-Protector peroxidase stabilizer and stored at 4°C. The final titer for anti-human β₂M antibody staining was 1:1600. The final titer for anti-human CD298 antibody staining was 1:100.

Titration experiments were performed to determine the optimal concentration of antibody staining following recommended CyTOF antibody optimization procedures. Antibody performance was evaluated quantitatively by calculating both the staining index (SI) and signal-to-noise ratio (SNR). The SI was calculated as (GeoMean_positive − GeoMean_negative)/(2 × SD_negative), where GeoMean and SD refer to the geometric mean and standard deviation of the metal intensity of the gated positive and negative cell populations, respectively. The signal-to-noise ratio (SNR) is defined as the ratio of the geometric mean intensity of the positive population to that of the negative population (SNR = GeoMean_positive/GeoMean_negative).

Freezing and thawing of PBMCs

For thawing, cells were placed in a 37°C water bath until nearly thawed, then transferred to a 6 mL RPMI 1640 medium (Gibco) supplemented with 10% fetal bovine serum and 1% penicillin–streptomycin (Gibco). Cells were centrifuged at 500 × g for 5 min at room temperature, the supernatant was discarded, and the pellet was resuspended in 6 mL of the same medium in six-well plates. The cells were then incubated at 37°C with 5% CO₂ for 1 h to allow recovery.

T cell experiments

15-mer peptide pools spanning the length of the SARS-CoV-2 spike protein (ancestral strain) with 11 amino acid overlaps (GenScript, RP30020) were used at a concentration of 2 µg/ml to stimulate spike-specific T cells. PBMCs were cultured in 96-well U-bottom plates at 2 million cells/200 µl per well, as previously described.^{14, 24–26} The cells were co-stimulated with 1 µg/ml of anti-CD28/CD49d (BD Biosciences, USA; Cat. No. 347690). Stimulation with an equimolar amount of dimethyl sulfoxide (DMSO) was performed with negative control. 5 µg/ml Phytohemagglutinin (PHA, Roche) stimulated cells were used as positive controls. After incubation for 18 h at 37°C in 5% CO₂, an additional incubation of 6 h was carried out by adding brefeldin A (Sigma-Aldrich) and Golgi-Stop containing monensin (BD Biosciences). For AIM assays, PBMC was incubated for 15 min at 37°C, with CD40 blocking antibody (BD Biosciences) at a final concentration of 0.5 µg/mL prior to stimulation. Antigen-specific CD4⁺ and CD8⁺ T cells were measured using background (DMSO)-subtracted data, with a minimal DMSO level set to 0.005%. Positive responses were defined here as greater than twofold over background in the unstimulated condition (defined as “net responses” and calculated as the signal in the antigen stimulated condition subtracted the signal in the unstimulated condition).

Cell barcoding

Using a 7-choose-2 combinatorial scheme, we pooled 12 barcoded samples, as illustrated in Figure 4a. To thoroughly remove unbound antibodies from the samples and ensure data reliability, we used the Curiox Laminar Wash HT2000 system for cell barcoding. Cells were transferred to a 96-well laminar wash plate (Curiox), and different combinations of anti-human β₂M and anti-human CD298 antibodies were added to the respective samples. The plate was then incubated at 4°C for 40–50 minutes until the cells settled on the plate. The Curiox Laminar Wash HT2000 was configured with the following settings: initial input volume, 80 µL; 25 washes; and a flow rate of 10 µL/s. These parameters were used for all subsequent washing steps. Cells were washed with PBS containing 0.5% bovine serum albumin (Sigma-Aldrich) using a laminar flow device.

Figure 4.

Barcoding of 12 samples using β2 M and CD298-based live-cell barcoding and a 7-choose-2 barcoding scheme. (a) Barcoding strategy. Individual samples were barcoded with a combination of two different Cd-tagged β₂M and CD298 antibodies, pooled, stained and acquired. (b) Histograms showing the signal intensities of the 12 barcoded samples of 7 Cd isotopes labeled with β₂M and CD298 antibodies. (c) Pooled samples (n = 12) were subjected to dimension reduction and clustering using UMAP algorithms. Each color code corresponds to a single sample. (d) The debarcoding efficiency of β₂M and CD298-based barcoded unique samples post-boolean gating was compared for a representative 7-choose-2 experiment. (e, g) Frequency of CD4⁺ and CD8⁺ T cell subsets between individual processed samples and barcoded samples. (f, h) Frequency of CD4⁺ and CD8⁺ AIM⁺ cells and cytokine-producing T cells after background subtraction between groups.

Cell staining

The experimental workflow for mass cytometry largely parallels that of conventional fluorescence flow cytometry, with specific adjustments to accommodate the system’s unique requirements. After cell isolation, samples were processed and stained according to a detailed protocol for multi-marker analysis. The cell pellet was initially resuspended in 990 µL of pre-warmed PBS without Ca^{2 +}/Mg^{2 +}. In this suspension, 10 µL of Cisplatin Reagent (Polaris Biology) was added, gently vortexed, and incubated at room temperature for 5 min. The reaction was quenched by adding 2 mL of LunaStain Cell Staining Buffer (Polaris Biology), followed by centrifugation at 500 × g for 5 min, after which the supernatant was removed.

For surface marker staining, an extracellular antibody cocktail was prepared using LunaStain Cell Staining Buffer. Each sample was resuspended in 100 µL of this cocktail, gently mixed to avoid bubble formation, and incubated at room temperature for 30 min. After incubation, 5 mL of LunaStain Cell Staining Buffer was added, followed by centrifugation at 500 × g for 5 min, after which the supernatant was discarded. This washing step was repeated twice.

For fixation, LunaFix Cell Fix Buffer (Polaris Biology) was diluted 1:1 with PBS. A total of 100 µL of this fixative mixture was added to the surface-stained cells, gently mixed, and incubated at room temperature for 5 min. Subsequently, 2 mL of LunaStain Cell Staining Buffer was added, followed by centrifugation at 800 × g for 5 min, after which the supernatant was removed. For permeabilization, the cell pellet was resuspended in 2 mL of LunaPerm Cell Perm Buffer (Polaris Biology), vortexed to mix, and centrifuged at 800 × g for 5 min at room temperature, after which the supernatant was discarded. This step was repeated, and during the second resuspension, the cells were incubated at room temperature for 30 min before centrifugation and supernatant removal.

For intracellular staining (ICS), an intracellular antibody cocktail was prepared using LunaPerm Cell Perm Buffer. Each sample was incubated with 100 µL of this antibody mixture, gently vortexed, and left at room temperature for 45 min. After staining, 5 mL of LunaPerm Cell Perm Buffer was added, mixed, and centrifuged at 800 × g for 5 min, after which the supernatant was discarded. This washing step was repeated twice. Finally, for Ir-DNA intercalator staining, 100 µL of LunaFix Cell Fix Buffer was combined with 2 µL of Ir-DNA Intercalator Reagent (Polaris Biology) to prepare the staining solution. Each sample was treated with 100 µL of this solution, mixed carefully, and incubated at room temperature for 10 min. Samples were then either processed immediately or stored at 4°C until the Ir-DNA staining solution was removed prior to data acquisition.

Data acquisition and analysis of FCS files

Cell acquisition was performed at a rate of 400–500 events per second on a mass cytometer (Polaris Biology, StarionX1). The acquired mass cytometry data were normalized and exported as standard FCS 3.0 files. Barcoded FCS files were manually debarcoded based on the assigned metal-isotope barcoding scheme using hierarchical gating on barcode channels in FlowJo (BD Biosciences) to reassign events to their original sample identities. To confirm accurate deconvolution, we applied UMAP and clustering analyses, which showed sample-specific clustering patterns, validating the debarcoding results. To visualize the overall structure of the immune compartment, t-distributed stochastic neighbor embedding (t-SNE) was applied with the following opt-SNE parameters: 1000 iterations, a perplexity of 30, and a learning rate of 0.5. Uniform manifold approximation and projection (UMAP) was further employed to illustrate the clear separation of the 12 barcoded samples into discrete clusters.

Statistical analysis

Data visualization and statistical analyses were carried out using R and FlowJo_v10.10.0. Correlations were assessed with Spearman’s correlation test, while paired non-parametric data were analyzed using the two-sided Wilcoxon signed-rank test. Unless otherwise specified, bar graphs display the full range from minimum to maximum values. Statistical significance is indicated as follows: ****p < .0001, ***p < .001, **p < .01, *p < .05.

Results

Multiparametric antibody panel optimization for comprehensive T cell subset profiling

T cells are highly heterogeneous, comprising numerous phenotypically and functionally distinct subsets. The first critical step in developing a comprehensive T cell assay is selecting markers that robustly distinguish major lineages and capture key phenotypic and functional attributes of antigen-specific responses. We used CD3, CD4, and CD8 to differentiate helper (CD4⁺) and cytotoxic (CD8⁺) T cells. For T cell subset classification, we included CD45RA, CCR7, CXCR5, PD-1, ICOS, CXCR3, and CCR6, enabling identification of memory subsets (naive, central memory [TCM], effector memory [TEM], terminally differentiated effector memory [TEMRA], and follicular helper T cells [Tfh].

To capture functional status of antigen-specific T cells, we included cytokines IFN-γ, IL-2, IL-4, TNF-α, granzyme B (GZMB), IL-17, and IL-21. In parallel, we measured TCR-triggered AIMs: CD25, CD69, CD134 (OX40), CD137 (4-1BB), CD200, CD40L, PD-L1, and CD107a to identify activated T cells. Additional activation markers CD38 and HLA-DR, together with the long-lived memory marker CD127, provide enhanced resolution of T cell functional states. A comprehensive list of all metal-isotope-tagged antibodies is provided in Table 1.

Table 1.

Antibody used in this study.

	MARKER	CLONE	METAL	PURPOSE
1	CD3	UCHT1	148Nd	T cells
2	CD4	RPA-T4	142Nd	T helper cells
3	CD8a	RPA-T8	144Nd	CD8 T cells
4	CD45RA	HI100	150Nd	T cell memory
5	CCR7(CD197)	G043H7	166Er	T cell memory
6	CCR6	G034E3	171Yb	Tfh subsets
7	CXCR5	J252D4	153Eu	Tfh subsets
8	CXCR3	G025H7	163Dy	Tfh subsets
9	CD279 (PD-1)	EH12.2H7	155Gd	Activation/Exhaustion
10	CD278(ICOS)	C398.4A	159Tb	Tfh subsets
11	CD28	CD28.2	158Gd	T cell memory
12	CD95	DX2	164Dy	T cell memory
13	IFN-γ	B27	170Er	Th1 Cytokines
14	IL-2	MQ1-17H12	145Nd	Th1 Cytokines
15	TNF-α	MAb11	143Nd	Th1 Cytokines
16	GZMB	QA16A02	149Sm	Th1 Cytokines
17	IL-4	MP4-25D2	172Yb	Th2 Cytokines
18	IL-17A	BL168	147Sm	Th17 Cytokines
19	IL-21	3A3-N2	173Yb	Tfh Cytokines
20	CD40L(CD154)	24–31	175Lu	AIM
21	CD200	OX-104	156Gd	AIM
22	CD69	FN50	160Gd	AIM
23	CD134(OX40)	Ber-ACT35	162Dy	AIM
24	CD137(4-1BB)	4B4-1	167Er	AIM
25	CD107a	H4A3	161Dy	AIM
26	PD-L1	29E.2A3	168Er	AIM
27	CD127(IL-7Ra)	A019D5	165Ho	Longevity memory
28	CD25 (IL-2R)	TÜ69	169Tm	Activation,AIM marker
29	HLA-DR	L243	141Pr	Activation
30	CD38	HIT2	174Yb	Activation

To ensure accurate and reproducible detection of both surface and intracellular proteins, we optimized antibody concentrations via serial titration. For surface antibody (Supplementary Figure 1), we began with twice the manufacturer’s recommended concentration and applied six successive 1:2 dilutions. For AIM and intracellular markers (Supplementary Figure 3), peripheral blood mononuclear cells (PBMCs) were stimulated with phytohemagglutinin (PHA) prior to staining to ensure expression of activation-induced targets. At each dilution, we calculated both the staining index (SI, D/W) and the signal-to-noise ratio (S/N), selecting titers that provided the highest SI or S/N with minimal background signal. The final antibody concentrations were determined based on the dilution that maximized resolution – bright positive signal with low background (Supplementary Figure 2, 4).

Time-resolved AIM dynamics reveal different combinatorial efficacy of AIM for vaccine immunogenicity assessment

Selecting appropriate activation-induced markers (AIMs) is essential for accurately detecting antigen-specific T cells and evaluating vaccine immunogenicity. A critical variable in AIM/ICS assays is stimulation duration, which significantly influences marker expression and varies depending on both the antigen and the specific marker. However, the precise expression kinetics of different AIMs on SARS-CoV-2-specific T cells remains underexplored.

To address this gap, we first profiled the temporal expression of eight AIMs on SARS-CoV-2 peptide stimulated PBMCs from 12 healthy vaccine recipients. We measured CD69, CD25, PD-L1, CD134, CD137, CD40L, and CD200 on CD4⁺ T cells, as well as CD69, CD25, PD-L1, CD134, CD137, CD40L, and CD107a on CD8⁺ T cells at five time points (6, 12, 18, 24, and 30 h). After subtracting the background signal from unstimulated controls (Supplementary Figure 5–7), most CD4⁺ and CD8⁺ T cell responses peaked within the 18–24 h (Figure 1a,b and Supplementary Data 1–2), confirming this as the optimal stimulation period. Given the potential contribution of Tregs to constitutive CD25 expression (Supplementary Figure 8), all analyses of CD25-containing AIM combinations were performed after excluding CD25⁺CD127⁻ Tregs. For the peaking magnitudes in CD4⁺ T cells, CD69 showed the strongest response (~3%), followed by PD-L1, CD25, and CD134 (~1.8). Markers like CD40L, and CD200 showed minimal activation (<0.5%) (Figure 1a). Similarly, in CD8⁺ T cells, CD69 was the most robust (>3%), while CD25, PD-L1, and CD137 produced moderate responses, and the remaining markers exhibited low-level expression (Figure 1b). These results align with prior studies⁵ showing that AIMs such as CD69 and CD137 reliably indicate antigen-specific activation around 18-24 h post-stimulation, whereas markers like CD40L or CD200 may exhibit delayed or low-level expression in short stimulation protocols.

Figure 1.

Distinct kinetics of AIM among activated T cells. (a,b) Comparison of the net SARS-CoV-2 specific CD4⁺ and CD8⁺ T cell responses for each individual AIM across different stimulation durations. Data points represent the mean of 12 individual samples. The single-AIM gating strategy is shown in Supplementary Figure 9. (c,d) Comparison of the net SARS-CoV-2 specific CD4⁺ and CD8⁺ T cell responses across 21 AIM pairs at different stimulation durations. Data points represent the mean of 12 individual samples. The dual-AIM gating strategy is shown in Supplementary Figure 10. (e,f) Heatmap depicting the correlation between results from different AIM combinations performed on PBMCs isolated from the same subjects, based upon spearman’s rank-order correlation. The spearman’s correlation coefficient are summarized in Supplementary data 1–2. (g, h) barplot illustrate overlap between results from different AIM combinations performed on PBMCs isolated from the same subjects.

While dual-marker AIM assays enhance specificity relative to single-marker approaches, variability across different marker pairings remains a notable challenge. To optimize assay performance, we systematically evaluated all 21 unique dual-marker combinations derived from our selected AIM panel over five timepoints. In CD4⁺ T cells, the CD134⁺PD-L1⁺ pair yielded the highest peak responses (~0.42%), with CD25⁺PD-L1⁺, CD69⁺CD134⁺, and CD69⁺PD-1⁺ also performing robustly (Figure 1c, Supplementary Data 3). Most combinations involving CD40L⁺ or CD200⁺ were markedly less sensitive (<0.2%). CD8⁺ T cells followed a similar trend: CD25⁺PD-L1⁺ peaked at ~0.28%, and other effective combinations included CD25⁺CD69⁺, CD25⁺CD134⁺, and CD69⁺PD-L1⁺, CD25⁺CD40L⁺, CD25⁺CD137⁺. Other AIM pairs elicited minimal activation (<0.1%), including those featuring CD107a⁺ or CD40L⁺ (Figure 1d, Supplementary Data 4). These observations highlight the varying sensitivity of different marker pairings.

Given the variability in antigen-specific T cell magnitudes detected by the 21 AIM combinations over time, we next investigated whether these AIM pairs could be used interchangeably. We focused on the 24 h post-stimulation time point and conducted correlation analyses across all combinations to evaluate their degree of concordance and potential interchangeability. We observed strong positive correlations among most AIM pairs (Figure 1e,f, Supplementary Data 5–6), implying that they largely identify overlapping antigen-specific T cell populations. However, pairs involving PD-L1 on CD4⁺ T cells, as well as CD40L⁺CD69⁺ and CD40L⁺CD134⁺ on CD8⁺ T cells exhibited weaker correlations with other AIM combinations. This suggests that these particular marker combinations may delineate distinct T cell subsets that are not captured by the majority of AIM pairings.

To evaluate the overlap among antigen-specific T cells identified by different AIM combinations, we used the widely adopted CD134⁺ CD137⁺ set as the reference for CD4⁺ T cells and the CD69⁺ CD137⁺ for CD8⁺ T cells. For each tested combination, we calculated the percentage of dual-positive cells that overlapped with the corresponding reference population. In CD4⁺ T cells, combinations such as CD69⁺ CD134⁺, CD134⁺ PD-L1⁺, and CD69⁺ CD137⁺ showed the highest co-identification with the reference CD134⁺ CD137⁺ population (mean overlap >75%) (Figure 1g, Supplementary Data 7), indicating these combinations capture largely overlapping antigen-responsive T cell subsets. In contrast, combinations like CD69⁺ CD200⁺ or PD-L1⁺ CD200⁺ had much lower overlap rates (<50%), suggesting they label distinct or only partially overlapping subsets. Similarly, in CD8⁺ T cells, most AIM combinations demonstrated strong co-detection (>70% overlap, Figure 1h, Supplementary Data 8), while combinations such as CD137⁺ PD-L1⁺, CD40L⁺ CD134⁺, and CD107a⁺ CD137⁺ showed noticeably lower overlap. These findings suggest that, even though many dual-marker AIM combinations have not been previously described, they can nonetheless reliably identify antigen-specific T cell populations.

Assessing the impact of protein transport inhibitors on AIM assay performance

An essential consideration when combining AIM assays with cytokine detection is whether protein secretion inhibitors, such as brefeldin A (BFA) and monensin (MN), affect AIM surface expression. These inhibitors block endosomal trafficking and may therefore interfere with both AIM surface presentation and cytokine secretion.^27,28 To investigate this, we measured both surface and intracellular AIM expression in the presence of BFA and monensin, using surface expression levels without inhibitors as the reference standard. In our protocol, inhibitors were added during the final 6 h of a 24-h antigen stimulation, ensuring initial surface accumulation of AIM before trafficking disruption. Correlation analyses were then conducted to assess how the inhibitors impacted marker detection. This approach allowed us to determine which AIMs remain reliable in combined AIM-ICS assays and which are susceptible to underestimation in the presence of secretion blockers.

For most CD4⁺ T cell AIMs, surface expression under protein transport inhibition remained highly correlated with the standard condition (without inhibitors). In particular, surface staining (SS) of CD25 (R² = 0.94, Figure 2a), CD69 (R² = 0.93, Figure 2b) and CD134 (R² = 0.95, Figure 2c) demonstrated the strongest concordance, while CD137 (R² = 0.87, Figure 2d) also maintained good agreement. Intracellular expression of these markers exhibited slightly lower correlations with surface expression under inhibitor-free conditions, indicating that they are minimally affected by protein transport disruption. In contrast, markers such as CD40L, PD-L1, and CD200 – whether detected extracellularly or intracellularly – were significantly impacted by protein transport inhibitors (Supplementary Figure 11a–c), indicating greater susceptibility to protein transport inhibition.

Figure 2.

Correlation between surface and intracellular staining of AIM in the presence of protein transport inhibitors and surface staining in the absence of protein transport inhibitors. (a–d) Correlation between the expression levels of CD4⁺ AIMs stained with surface and intracellular staining in the presence of protein transport inhibitors and surface stained CD4⁺ AIMs in the absence of protein transport inhibitors. (e–h) Correlation between the expression levels of CD8⁺ AIMs stained with surface and intracellular staining in the presence of protein transport inhibitors and surface stained CD8⁺ AIMs in the absence of protein transport inhibitors. (i) Frequency of the expression levels of CD4⁺ AIM combination based on surface staining after background subtraction between groups. (j) Frequency of the expression levels of CD8⁺ AIM combination based on surface staining after background subtraction between groups. ****p <0 .0001, ***p < 0.001, **p <0 .01, *p <0 .05.

For CD8⁺ T cell AIMs, intracellular expression of CD25 (R² = 0.97, Figure 2e), CD69 (R² = 0.92, Figure 2f) and CD137 (R² = 0.96, Figure 2h) showed the strongest correlations, and the extracellular staining of these markers was also stable. In contrast, CD134 displayed higher correlation when stained extracellularly (R² = 0.91, Figure 2g) than intracellularly (R² = 0.88, Figure 2g). Markers such as CD40L, PD-L1, and CD107a were consistently sensitive to protein transport inhibitors (Supplementary Figure 11d–f).

Given the superior performance of surface staining under protein transport inhibition, we adopted surface-based AIM detection for subsequent analyses. Markers sensitive to protein transport inhibitors – CD40L, PD-L1, CD200, and CD107a – were excluded from further evaluation. We then assessed six AIM combinations composed of the remaining markers, comparing conditions with and without protein transport inhibitors, to evaluate their consistency in detecting antigen-specific T cells. In CD4⁺ T cells (Figure 2i), most AIM combinations remained stable under both conditions, except CD69⁺ CD137⁺, which showed a significant reduction with inhibitor treatment. In CD8⁺ T cells (Figure 2j), combinations including CD134⁺ CD137⁺, CD25⁺ CD134⁺, and CD69⁺ CD134⁺, were significantly affected by inhibitors. In comparison, CD25⁺ CD137⁺, CD25⁺ CD69⁺, and CD69⁺ CD137⁺ demonstrated consistent detection regardless of inhibitor presence.

AIM assays capture a broader spectrum of antigen-responsive cells than ICS alone

To better understand the relationship between T cells identified by AIM assay and those detected by intracellular cytokine staining (ICS), we analyzed T cells from the same individuals using both assays. Subsequently, we performed spearman correlation analyses between six representative AIM combinations and ICS results, focusing on IFN-γ, Granzyme B (GZMB), IL-2 and TNF-α expression in both CD4⁺ and CD8⁺ T cell subsets.

Correlation analysis revealed that the frequency of the AIM pair CD25⁺ CD134⁺ showed the strongest and most consistent positive correlation with cytokine-producing CD4+ T cells, particularly with IFN-γ (p <0 .01, Figure 3a). Another combination, CD25⁺ CD69⁺ also exhibited significant correlations, whereas other four combinations were negatively associated with cytokine responses. Similar trends were observed in CD8⁺ T cells, where the CD25⁺ CD137⁺, CD25⁺ CD134⁺, and CD25⁺ CD69⁺ combinations remained correlated with cytokine responses, particularly with IFN-γ (p <0 .05, Figure 3b). These findings indicate a strong correlation between CD25⁺-based AIM pairs and functional T cell responses, such as IFN-γ release.

Figure 3.

Experimental correlation between AIM assay and ICS assay. (a,b) Heatmap depicting the level of correlation between results from AIM and ICS assays performed on PBMCs isolated from the same subjects (n = 10), based upon spearman’s rank-order correlation. The ICS gating strategy is shown in Supplementary Figure 12. Stars represent p-values, and the intensity of color represents R-values (spearman’s rank correlation coefficient); ****p <0 .0001, ***p <0 .001, **p <0 .01, *p <0 .05. (c,d) Barplot showing the proportion of AIM⁺ cells among IFN-γ⁺ T cells in PBMCs isolated from the same individuals. (e,f) Barplot showing the proportion of cytokine-positive T cells among AIM⁺ cells in PBMCs isolated from the same individuals.

We next quantified the frequency of AIM⁺ cells within the pool of IFN-γ⁺ T cells. In CD4⁺ T cells, a substantial proportion of IFN-γ⁺ cells co-expressed CD25⁺ CD69⁺ (median ~75%), followed by CD25⁺ CD134⁺ and CD134⁺ CD137⁺ (Figure 3c). In CD8⁺ T cells, the CD25⁺CD69⁺ combination again showed the highest overlap with IFN-γ⁺ cells (Figure 3d), suggesting that cytokine-producing cells preferentially upregulate AIMs, particularly the CD25⁺ CD69⁺ pair.

To further characterize the functional capacity of AIM⁺ T cells, we quantified the proportion of cytokine-producing cells within AIM⁺ CD4⁺ and CD8⁺ populations (Figure 3e,f). Among CD4⁺ AIM⁺ T cells (Figure 3e), less than 20% produced any individual cytokine (GZMB, IFN-γ, IL-2, or TNF-α), with the highest frequencies observed for IFN-γ and TNF-α. Similarly, in CD8⁺ T cells (Figure 3f), fewer than 20% produced any single cytokine (GZMB, IFN-γ, IL-2, or TNF-α). This finding aligns with previous reports²⁹ showing that the AIM assay detects higher frequencies of antigen-specific T cells compared to the ICS assay. Therefore, AIMs not only capture the antigen-responsive T cells identified by conventional ICS but also reveal additional subsets of antigen-reactive cells.

Impact of barcoding on the detection of antigen-specific T cell responses

Single-cell barcoding enables the simultaneous processing of multiple samples, thereby increasing throughput while reducing technical variability during sample preparation. Early mass cytometry barcoding strategies primarily utilized amine- or sulfhydryl-reactive heavy metals, which required fixation and permeabilization due to the intracellular abundance of these reactive groups.^30,31 This requirement limited their compatibility with epitopes that are sensitive to fixation or permeabilization. Although CD45-based palladium barcoding later addressed this limitation for PBMCs and other CD45-expressing cells^32,33, its broader application is restricted by the tissue-specific expression cells. To overcome these challenges, alternative surface markers such as beta-2-microglobulin (β₂M) and Na⁺/K⁺ -ATPase (CD298) conjugated to platinum have been proposed for live-cell barcoding. However, the limited availability of palladium and platinum isotopes constrains the number of samples that can be multiplexed.

Here, we introduce a cadmium (Cd)-based barcoding strategy employing seven Cd-conjugated β₂M and CD298 antibodies. This system, following a 7-choose-2 combinatorial scheme, enables multiplexing of up to 21 samples while preserving doublet discrimination. In this study, we pooled 12 barcoded samples (Figure 4a), followed by surface staining, fixation, and permeabilization. Barcode-positive and barcode-negative populations were clearly distinguishable for both β₂M and CD298 (Figure 4b). Dimensionality reduction via t-SNE revealed distinct clustering of the 12 barcoded samples (Figure 4c), with high debarcoding efficiency confirmed by comparison to individually stained reference samples (Figure 4d).

To evaluate the potential impact of barcoding on antigen detection and biological interpretation, we divided the 12 samples into two groups: one processed without barcoding and the other barcoded using the β₂M/CD298 system prior to staining with the same antibody panel. A strong overall correlation (correlation coefficient >0.99, Figure 4e,g) was observed in the relative distribution of major CD4⁺ and CD8⁺ T cell populations between barcoded and individually processed samples. Notably, good correlations (correlation coefficient >0.95, Figure 4f,h) were also detected for rare subsets, including AIM⁺ cells and cytokine-producing cells (IFN-γ, IL-2, TNF-α, and GZMB) within both CD4⁺ and CD8⁺ T cell compartments. These findings demonstrate that our cadmium-based barcoding strategy preserves signal integrity and enables reliable immune profiling without introducing technical bias.

Application of the optimized panel

To validate the performance of the optimized AIM and cytokine detection panel, we first applied the workflow to 46 healthy adult volunteers (age >18) who predominantly had hybrid immunity to SARS-CoV-2. PBMCs from these donors were stimulated with SARS-CoV-2 peptide pools for 24 h, followed by barcoding, pooling, and staining with a 30-parameter immunophenotyping master mix. After acquisition and debarcoding, t-SNE analysis identified 12 major PBMC subpopulations (Figure 5a), with the corresponding marker expression profiles shown in Figure 5b.

Figure 5.

Quantification of AIM and cytokine production in stimulated and unstimulated samples to SARS-CoV-2 peptide. (a) Phenotype clusters identified by t-SNE clustering. (b) Dotplot showing expression of markers in the different cell clusters identified in the CyTOF dataset. (c) Frequency of CD4⁺ AIM⁺ cells and cytokine-producing T cells among two groups. (d) Frequency of CD8⁺ AIM⁺ cells and cytokine-producing T cells among two groups. (e) Frequency of CD4⁺ AIM⁺ cells and cytokine-producing T cells among two groups before and after vaccination. (f) Frequency of CD8⁺ AIM⁺ cells and cytokine-producing T cells among two groups before and after vaccination. Mann-whitney analyses between stimulation conditions within each population and between the same stimulation conditions in different populations. Medians and inter-quartile range (IQR) shown. ****p <0 .0001, ***p <0 .001, **p <0 .01, *p <0 .05, ns: not significant (p > 0.05).

In this cohort with hybrid immunity, SARS-CoV-2 peptide stimulation elicited significant increases in both AIM⁺ and cytokine-producing CD4⁺ T cells (IFN-γ, IL-2, and TNF-α), whereas GZMB⁺ CD4⁺ T cells remained unchanged (Figure 5c). CD8⁺ T cells similarly showed significant elevations in both AIM⁺ populations and all measured cytokine-producing subsets (Figure 5d).

To further validate the reliability of this assay, we analyzed an additional cohort of 12 donors with paired PBMC samples collected both prior to vaccination and 14 d after receiving two doses of the inactivated SARS-CoV-2 vaccine. As illustrated in Figure 5e,f, pre-vaccination samples exhibited low frequencies of AIM⁺ and cytokine-producing T cells across both CD4⁺ and CD8⁺ subsets. Following vaccination, however, the optimized AIM combinations and cytokine readouts increased robustly and significantly. Notably, less optimal AIM pairs, such as CD40L⁺ CD137⁺, CD40L⁺ CD69⁺, CD200⁺ CD137⁺ in CD4⁺ T cells, and several CD107a-based CD8⁺ combinations (Supplementary Figure 13–14), failed to discriminate pre- from post-vaccination samples, further confirming the superior sensitivity of the optimized AIM panel.

Together, analysis from both hybrid-immunity and antigen-naive cohorts confirms the robustness, specificity, and scalability of our cadmium-barcoded CyTOF workflow. This platform supports high-dimensional, multiplexed immune profiling and enables reliable discrimination of SARS-CoV-2-specific T-cell responses from nonspecific activation.

Discussion

In this study, we established a high-throughput mass cytometry (CyTOF) platform capable of simultaneously detecting activation-induced markers (AIMs) and intracellular cytokines in SARS-CoV-2-specific memory T cells, providing a robust and sensitive tool for immune profiling following vaccination. Our findings emphasize the importance of carefully selecting marker combinations and optimizing assay conditions to ensure the reliable detection of antigen-specific T cells.

Throughout the COVID19 pandemic, AIM and ICS assays have been widely employed to characterize T cell responses. However, substantial methodological variation has hindered direct comparisons across studies – including differences in stimulation duration, stimulant concentrations, use of co-stimulants, and cell input numbers. Despite this, inconsistent marker selection across assays likely accounts for the wide variability in reported T cell response ranges, even among studies examining comparable cohorts.^{9, 10, 14, 34–37} To address this issue, we established a standardized, universal AIM/ICS protocol and then apply this protocol to longitudinally assess cellular immunity in a cohort of healthy individuals with hybrid immunity. This approach would enable precise determination of protective T cell response thresholds and facilitate their use in next-generation vaccine trials.

Using SARS-CoV-2 vaccines as a model, our systematic profiling of eight AIMs across five time points revealed a conserved peak activation window at 18–24 h post-stimulation for both CD4⁺ and CD8⁺ T cells – consistent with prior kinetic studies of antiviral T cell responses, now extended to SARS-CoV2 antigens.^7,19 Our analysis found strong concordance across the 21 dual-marker AIM combinations derived from these eight markers in their capacity to quantify antigen-specific CD4⁺ and CD8⁺ T cell responses, despite several combinations not previously reported. Although each combination defined largely overlapping AIM⁺ populations, none were identical, which underscores the robustness and specificity of AIM assays in capturing antigen-responsive T cells.

CD69 emerged as the most sensitive single AIM in both CD4⁺ and CD8⁺ T cell subsets, confirming its value as an early activation indicator. In contrast, markers like CD40L (CD4⁺) and CD107a (CD8⁺) performed poorly when used alone – yielding a lower antigen-specific T cell response – emphasizing the limitations of single-marker assays. Notably, our dual-marker strategy identified CD134⁺ PD-L1⁺ and CD25⁺ PD-L1⁺ as the most sensitive combination across CD4⁺ and CD8⁺ subsets. This suggests that simultaneous expression of activation and checkpoint molecules reflects robust T cell activation – a pattern previously observed in chronic viral infection settings.^7,24,38 However, markers such as PD-L1, CD200, CD40L, and CD107a showed significantly reduced correlation under protein secretion inhibitors (brefeldin A and monensin) compared to untreated conditions. These inhibitors disrupt Golgi- and endosomal-mediated trafficking, which is essential for the surface expression of these markers.^27,28 For instance, CD40L requires active Golgi-to-cell-surface transport and is rapidly internalized, while CD107a depends on granule fusion during degranulation – both processes are blocked by secretion inhibitors.³⁹ Similarly, PD-L1 and CD200 rely on vesicle-based trafficking for their regulated surface expression. Although brefeldin A completely blocks cell-surface expression of CD69 following stimulation, it does not impair its intracellular expression – whereas monensin does not inhibit surface CD69 expression at all.²⁷ Thus, co-administration of monensin can effectively counteract the surface-blocking effect of BFA. Other markers such as CD25, CD137, and CD134 are rapidly upregulated directly at the plasma membrane, independent of complex intracellular transport, and thus remain reliable even in the presence of secretion inhibitors.

The relationship between T cell activation markers and effector function in vaccine-induced immunity remains poorly defined. By systematically correlating AIM co-expression with intracellular cytokine production, our study demonstrates that AIM⁺ cells co-expression CD25 with CD134, and CD69 on CD4⁺ and CD8⁺ T cells significantly correlate with the IFN-γ⁺ functional responses, reinforcing prior findings that CD25-based AIM pairs are superior indicators of SARS-CoV-2-specific helper T cell responses in vaccinated or convalescent individuals.^18,40 Notably, CD25-based AIM identification in CD4+ T cells can be influenced by constitutive CD25 expression on regulatory T cells; therefore, inclusion of markers (CD127 or FOXP3) in the panel is recommended to exclude Tregs prior to AIM gating and downstream analysis. Interestingly, AIM pairs such as CD134⁺ CD137⁺ (CD4⁺) and CD69⁺ CD137⁺ (CD8⁺) – despite being commonly used to identify antigen-specific T cells – showed poor correlation with cytokine production. This suggests that while they effectively mark antigen recognition, they may reflect signaling pathways geared more toward survival or memory formation rather than immediate effector output, consistent with CD134 (OX40) and CD137 (4-1BB) biology.

CD25⁺ CD69⁺ demonstrated strong overlap with cytokine-producing T cells, indicating that early activation (CD69) together with IL-2 receptor upregulation (CD25) identifies transitional cells that are acquiring effector function. A key practical advantage is that the CD25⁺ CD69⁺ pair enables simultaneous assessment of both CD4⁺ and CD8⁺ T cell responses within a single panel, unlike earlier AIM approaches that relied on separate marker pairs for each subset. This compatibility simplifies direct comparisons across studies. Moreover, AIM assays consistently detect higher frequencies of antigen-specific T cells than ICS by capturing cytokine-independent activation events. Given the heterogeneity of antigen-specific T cells and the limitations of ICS in detecting noncytokine-producing cells, relying solely on ICS underestimates the response. In contrast, AIM assays anchored in TCR-dependent activation marker upregulation provide a broader, cytokine-independent view of the antigen-responsive repertoire. This enriched detection capability is especially valuable in clinical trials and studies involving rare cell subsets.

Importantly, our live-cell barcoding strategy – using antibodies against β₂M and CD298 conjugated with seven cadmium (Cd) isotopes (Cd-106, 110, 111, 112, 113, 114, 116) – provides robust multiplexing without compromising AIM⁺ or cytokine-producing T cell detection. Unlike CD45, whose expression varies across peripheral immune subtypes and is often low or absent in granulocytes and nonimmune cells⁴¹, β₂M and CD298 are stably expressed on virtually all live human cells, including non-hematopoietic and tissue-derived populations.^42,43 While most existing barcoding workflows rely on palladium isotopes (Pd-102, Pd-104, Pd-105, Pd-106, Pd-108, and Pd-110) applied to after permeabilization.^31,32,44 Our cadmium-based approach enables pre-staining of live PBMC before surface antibody staining and fixation. Cadmium isotopes lie outside the lanthanide mass detection window (139–176 Da), minimizing interference with conventional antibody channels, and offer stronger signal resolution than Pd-based barcodes due to superior chelation chemistry (MCP9 vs mDOTA).⁴⁵ Our data confirm that this Cd-based barcoding does not perturb t-SNE dimensionality reduction or the representation of 12 manually gated PBMC and T cell populations. The protocol’s design – including antigen stimulation and fixation – supports robust detection of intracellular/intranuclear markers, rare subpopulations, and phosphorylation events. This Cdconjugated β₂M/CD298 barcoding strategy combines live-cell compatibility, universal expression coverage, prestaining flexibility, and enhanced signal clarity, delivering a highly scalable, accurate, and versatile platform for multiplexed CyTOF immune profiling.

In this study, we demonstrate that integrating AIM and ICS assays represents an effective approach to comprehensively assess both the breadth and functionality of antigen-specific T cells. However, our analyses were exclusively performed on cryopreserved peripheral blood samples. Future protocol refinements should incorporate fresh peripheral blood and tissue-derived immune cells to more comprehensively reflect antigen-specific T cell responses. Although our barcoding strategy supports the simultaneous analysis of up to 21 PBMC samples, we barcoded only 12 samples in practice, since the extended acquisition time required for a 21sample run could jeopardize instrument stability and lead to signal drift over prolonged measurement periods. When the number of cells collected from each sample is small, pooling up to 21 barcoded samples remains a robust and dependable strategy. Moreover, our current panel was restricted to SARS-CoV-2 antigens. Expanding this methodology to include other vaccine- or infection-induced antigen-specific T cells would further enhance its applicability. In summary, we present a standardized and scalable CyTOF-based workflow for high-dimensional profiling of SARS-CoV-2-specific T cell responses. The integration of optimized AIM combinations, co-stimulation, and sample barcoding establishes a powerful platform for immune monitoring in vaccine development and infectious disease research.

Biographies

Wei Yang is an Assistant Professor at the Bio-X Interdisciplinary Innovation Center, Yunnan University, China. Her research focuses on inflammation, cancer immunology, and vaccine immunity. She applied integrated immunology and molecular biology approaches to elucidate the roles and mechanisms of inflammatory factors in infection immunity and tumor development, with the goal of identifying novel cancer targets and screening small-molecule drugs. Her work employs knockout mouse models, tumor cell lines, and clinical samples to validate key findings and therapeutic strategies. In addition, she combines medical immunology with single-cell multi-omics to investigate the interplay between immune imprinting and the antigenic distance of viral variants, providing strategic insights for vaccine updates against highly mutated pathogens.

Jingxin Li is a Professor at the School of Public Health, Nanjing University, and the Jiangsu Provincial Center for Disease Control and Prevention, China. Her research focuses on systems immunology of vaccines and clinical evaluation of immunization strategies. She pioneered a homologous boosting strategy for recombinant Ad5-vectored Ebola vaccines, challenging the prevailing belief that homologous vectors have poor boosting efficacy. She also led the first clinical trial of an mRNA COVID-19 vaccine in the Chinese population, providing critical data for global mRNA vaccine development and its application in Asian cohorts. In addition, she directed groundbreaking studies on heterologous sequential boosting following inactivated COVID-19 vaccination, demonstrating substantial enhancement of immune responses and contributing to the development of China’s sequential immunization policy.

Yanli Chen is a Postdoctoral Researcher at the Bio-X Interdisciplinary Innovation Center, Yunnan University, China. Her research focuses on T cell responses, vaccine immunology, and single-cell analysis. She employs non-human primate models to investigate the infection and pathogenic mechanisms of clinically significant human respiratory viruses, including EV71, CA16, influenza virus, and SARS-CoV-2. Additionally, she integrates single-cell transcriptomics with immunological approaches to elucidate the molecular mechanisms underlying SARS-CoV-2 infection and vaccine-induced immunity, providing key insights for the development of next-generation vaccines and therapeutic strategies.

Funding Statement

This work was supported by a grant [2023YFC2307600, to Z. S. Z.] from the National Key Research and Development Program of China; a grant [C615300504062, to Y. C.] from Caiyun Postdoctoral Program in Yunnan Province of China; a grant [202401BF070001-020, to W. Y.] from Yunnan Fundamental Research Project.

Abbreviations

CyTOF

mass cytometry

AIM

activation-induced marker

ICS

intracellular cytokine staining

ELISPOT

enzyme-Linked Immunospot

MCB

Metal-isotope cell barcoding

PHA

phytohemagglutinin

PBMC

peripheral blood mononuclear cell

SI

staining index

S/N

signal-to-noise ratio

SARS-CoV-2

severe acute respiratory syndrome coronavirus 2

BFA

brefeldin A

MN

monensin

SS

surface staining

IS

intracellular staining

β₂M

beta-2-microglobulin

CD298

Na⁺/K⁺ -ATPase

COVID19

coronavirus disease 2019

t-SNE

t-distributed stochastic neighbor embedding

UMAP

Uniform Manifold Approximation and Projection

FBS

fetal bovine serum

DMSO

dimethyl sulfoxide

Disclosure statement

ZJZ, YLC, and NW filed a patent application on the integration of AIM and ICS for SARS-CoV-2-specific T cell detection using mass cytometry. Other authors declare no competing interests.

Data availability statement

All data relevant to the findings of this study are included in the article and its supplementary materials. Raw datasets are available from the corresponding author upon reasonable request.

Supplementary material

Supplemental data for this article can be accessed online at https://doi.org/10.1080/21645515.2025.2610068