TNF-α+ CD4+ T cells dominate the SARS-CoV-2 specific T cell response in COVID-19 outpatients and are associated with durable antibodies

Abstract

Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2)-specific CD4⁺ T cells are likely important in immunity against coronavirus 2019 (COVID-19), but our understanding of CD4⁺ longitudinal dynamics following infection and of specific features that correlate with the maintenance of neutralizing antibodies remains limited. Here, we characterize SARS-CoV-2-specific CD4⁺ T cells in a longitudinal cohort of 109 COVID-19 outpatients enrolled during acute infection. The quality of the SARS-CoV-2-specific CD4⁺ response shifts from cells producing interferon gamma (IFNγ) to tumor necrosis factor alpha (TNF-α) from 5 days to 4 months post-enrollment, with IFNγ^-IL-21^-TNF-α⁺ CD4⁺ T cells the predominant population detected at later time points. Greater percentages of IFNγ^-IL-21^-TNF-α⁺ CD4⁺ T cells on day 28 correlate with SARS-CoV-2-neutralizing antibodies measured 7 months post-infection (⍴ = 0.4, p = 0.01). mRNA vaccination following SARS-CoV-2 infection boosts both IFNγ- and TNF-α-producing, spike-protein-specific CD4⁺ T cells. These data suggest that SARS-CoV-2-specific, TNF-α-producing CD4⁺ T cells may play an important role in antibody maintenance following COVID-19.

Trial registration: ClinicalTrials.gov NCT04331899.

Keywords: CD4; COVID-19; SARS-CoV-2; T cells; TNF-α; Tfh; antibodies; cytokines; longitudinal; neutralizing antibodies.

Graphical abstract

Figure 1

Identification of four distinct SARS-CoV-2-specific CD4⁺ T cell subsets that differ between S- and MN-protein stimulation PBMCs of COVID-19 outpatients on day 28 post-enrollment were stimulated with spike (S) or a combination of membrane (M) and nucleocapsid (N) proteins in vitro and were stained and analyzed by flow cytometry. (A and C) Representative flow plots of (A) cytokine-producing (TNF-ɑ, IFNγ, and IL-21) or (C) AIM-expressing (CD137⁺OX40⁺) CD45RA^- CD4⁺ T cells that are stimulated with “media only” or MN or S proteins. (B) The absolute percentage (scatterplot; black line, median) of each individual combination of TNF-ɑ-, IFNγ-, and IL-21-producing non-naïve CD4⁺ T cells stratified by antigenic stimuli (red: MN, blue: S; n = 99). The pie charts show the relative proportion (mean) of each individual combination of cytokine-producing non-naïve CD4⁺ T cells (e.g., pie slice 4 represents TNF-ɑ⁺IFNγ^-−IL-21^−-) among the total population of antigen-specific cells, stratified by antigenic stimuli (MN: n = 102, S: n = 100). The p value shown under the pie charts is calculated using a partial permutation test. (D) Shown are absolute percentages of OX40⁺CD137⁺ CD45RA^- CD4⁺ T cells stratified by antigenic stimuli with the black line indicating the median (n = 104). (B and D) p values shown above the scatterplots are calculated using Wilcoxon matched-pairs signed rank test. Values shown are background (media-only condition) subtracted. See also Figures S1–S3 and S5.

Figure 2

Kinetics of four distinct cytokine-producing SARS-CoV-2-specific CD4⁺ T cell populations over time and the density of CCR7 receptor expression on these populations (A and B) PBMCs from 24 COVID-19 outpatients sampled at days 5, 14, and 28 and month 4 (day 120) (n = 10 also sampled on day of enrollment [day 0]) were stimulated with MN (left) or S (right) proteins in vitro and were stained and analyzed by flow cytometry. (A) The mean and SEM of the absolute percentage of background-subtracted IFNγ^-IL-21^-TNF-ɑ⁺- (blue), IFNγ⁺IL-21^-TNF-ɑ^-- (yellow), IFNγ^-IL-21⁺TNF-ɑ^-- (red), or IFNγ⁺IL-21^-TNF-ɑ⁺-producing non-naïve CD4⁺ T cells (green) of sequential timepoints post-enrollment are shown. These four populations were the dominant populations identified in Figure 1. (B) The relative proportion of each individual combination of background-subtracted, TNF-ɑ-, IFN-γ, and IL-21-producing non-naïve CD4⁺ T cells stratified by sequential days are depicted in the pie charts (mean). The p values indicated on top of the pie charts are calculated using the partial permutation test, which tests the association between MN- and S-protein stimulation of each indicated day post-enrollment. The p values in the tables indicate the significance of the associations between the different days post-enrollment calculated using the partial permutation test (left: MN-protein-stimulated T cells; right: S-protein-stimulated T cells). (C) Representative flow cytometry plots of CCR7- and CD45RA-expressing CD4⁺ T cells (left) with an overlay of MN-protein-specific IFNγ^-IL-21^-TNF-ɑ⁺- (blue), IFNγ⁺IL-21^-TNF-ɑ^-- (yellow), IFNγ^-IL-21⁺TNF-ɑ^--(red), or IFNγ⁺IL-21^-TNF-ɑ⁺-producing non-naïve CD4⁺ T cells (green, right). (D) Mean fluorescence intensity (MFI) of CCR7 on MN- (left, n = 72) or S-protein-specific (right, n = 74) single positive TNF-ɑ- (blue), IFNγ- (yellow), IL-21- (red), or TNF-ɑ- and IFNγ-producing non-naïve CD4⁺ T cells (green) from day 28 post-enrollment. The p values were calculated using the Friedman test with Dunn’s multiple comparisons test (black line, median). See also Figures S1, S2, and S4–S6.

Figure 3

The kinetics of SARS-CoV-2-specific AIM⁺ CD4⁺, activated cTfh, and SARS-CoV-2-specific cTfh cell percentages and their correlation with cytokine-producing CD4⁺ T cells PBMCs of COVID-19 outpatients on days 5, 14, 28, and 120 post-enrollment were stimulated with MN or S proteins in vitro and were stained and analyzed by flow cytometry. (A) The kinetics of the absolute percentage of paired MN- (left, n = 22) or S- (right, n = 19) protein-stimulated AIM⁺ CD45RA^- CD4⁺ T cells over time. (B) The gating strategy of PD-1⁺CXCR5⁺CD4⁺ (cTfh), PD-1⁺CXCR5⁺ICOS⁺CD4⁺ (ICOS⁺ cTfh), and PD-1⁺CXCR5⁺OX40⁺CD137⁺CD4⁺ (AIM⁺ cTfh) cells. (C and D) The kinetics of the absolute percentage of paired (C) cTfh (n = 21) and (D) ICOS⁺ cTfh (n = 21) cells of unstimulated cells (media only) are depicted. (E) The kinetics of the absolute percentage paired AIM⁺ cTfh cells stimulated with MN (left, n = 22) or S (right, n = 19) protein over time are shown. (A and C–E) The p values were calculated using the Friedman test with Dunn’s multiple comparisons test. (F) The heatmap shown depicts Spearman’s correlations (Benjamini-Hochberg corrected) between the percentages of MN- (left, n = 101) or S-protein-stimulated (right, n = 100) cTfh cell populations (cTfh, ICOS⁺ cTfh, and AIM⁺ cTfh) or AIM⁺ CD45RA^- CD4⁺ T cells measured in AIM experiments and cytokine-producing non-naïve CD4⁺ T cells measured in ICS experiments. The measurements that are significantly associated are indicated by asterisks (∗p < 0.05, ∗∗∗p < 0.001). The scale bar indicates the Spearman's correlation rho (ρ). (G) Scatterplots comparing antigen-stimulated (MN: left/red, n = 101; S: right/blue, n = 100) IFNγ^-IL-21^-TNF-ɑ⁺-producing T cells with ICOS⁺ cTfh (top), AIM⁺ cTfh (middle), or AIM⁺ CD45RA^- CD4⁺ T (bottom) cells are shown (n = 101). The rho (ρ) and p values were calculated using Spearman’s correlation (Benjamini-Hochberg corrected). The lines represent the fitted linear relationship between the indicated cell populations. See also Figure S1.

Figure 4

Late T cell responses are positively correlated with durability of antibody titers, while early T cell responses are not associated with the peak magnitude of antibody titers (A and C) In the heatmaps, Spearman’s correlations (Benjamini-Hochberg corrected) between indicated cTfh cell populations, AIM⁺ CD45RA^- CD4⁺, and cytokine⁺ non-naïve CD4⁺ T cell data and antibody titers (S protein IgG binding [area under the curve (AUC) IgG] or neutralizing antibody (Neut) are shown. The measurements that are significantly associated are indicated by asterisks (∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001). The scale bar indicates the Spearman's correlation rho (ρ). (A) Correlations are performed between indicated T cell data collected at day 5 post-enrollment and indicated antibody titers at day 28 post-enrollment. (B) Scatterplots comparing MN- (left/red) and S-protein-stimulated (right/blue) IFNγ^-IL-21^-TNF-ɑ⁺-producing non-naïve CD4⁺ T cells collected on day 5 (MN: n = 55; S: n = 51) post-enrollment with neutralizing antibody titers collected on day 28 post-enrollment are depicted. (C) Correlations are performed between indicated T cell data collected on day 28 post-enrollment and indicated antibody titers at day 28 or 210 post-enrollment. (D) Scatterplots comparing antigen-stimulated IFNγ^-IL-21^-TNF-ɑ⁺-producing non-naïve CD4⁺ T cells collected on day 28 (MN: n = 81; S: n = 78) post-enrollment with neutralizing antibody titers collected on day 28 post-enrollment are shown. (E) Scatterplots comparing antigen-stimulated IFNγ^-IL-21^-TNF-ɑ⁺-producing non-naïve CD4⁺ T cells (left, MN: n = 48, S: n = 47) or ICOS⁺ cTfh cells (right, MN: n = 52, S: n = 50) collected on day 28 with neutralizing antibody titers collected on day 210 post-enrollment are shown. (B, D, and E) The neutralizing antibody titers are presented in natural logarithm, and we added +1 to allow for inclusion of participants with no neutralizing activity. The rho (ρ) and p values were calculated using Spearman’s correlation (Benjamini-Hochberg corrected). The lines represent the fitted linear relationship between the indicated data. See also Figures S5 and S7.

Figure 5

Associations between early immune response and T cell responses on day 28 post-enrollment (A) Gene-Ontology-based immune pathways (left) and plasma proteins (right) associated with the absolute percentage of SARS-CoV-2-specific IFNγ^-IL-21^-TNF-ɑ⁺ non-naïve CD4⁺ T cells on day 28. (B) Scatterplots showing associations between the MCP-3 on day 0 or 5 and the percentage of SARS-CoV-2-specific IFNγ^-IL-21-TNF-ɑ⁺ non-naïve CD4⁺ T cells on day 28. (C) Gene-Ontology-based immune pathways (left) and plasma proteins (right) associated with the absolute percentage of ICOS⁺ cTfh cells on day 28. (D) Scatterplots showing associations between the MCP-3 on day 0 or 5 and ICOS⁺ cTfh cells on day 28. (A and C) We used regression models to test the association while controlling for the sampling time of immune pathways or proteins (day 0 or 5) and the stimulation type (MN or S) of the T cell response. (B and D) The color of the points represents the stimulating antigens. The spearman correlations (rho [ρ]) between MCP-3 and the T cell percentages and the corresponding p values are reported. See also Tables S2 and S3.

Figure 6

mRNA vaccination boosts IFNγ⁺IL-21^-TNF-ɑ^-- and IFNγ⁺IL-21^-TNF-ɑ⁺-producing CD4⁺ T cells and AIM⁺ cTfh cells PBMCs of COVID-19 outpatients on month 10 post-enrollment were stimulated with S or MN proteins in vitro and were stained and analyzed by flow cytometry. (A) The absolute percentage (scatterplot; black line, median) of each individual combination of TNF-ɑ-, IFNγ-, and IL-21-producing non-naïve CD4⁺ T cells stratified by vaccination status (closed triangle: unvaccinated, n = 20; open triangle: vaccinated, n = 20). Top panel depicts MN-protein-stimulated CD4⁺ T cells (red), and bottom panel depicts S-protein-stimulated CD4⁺ T cells (blue). p values shown above the scatterplots are calculated using the Mann Whitney test. The pie charts show the relative proportion (mean) of each individual combination of cytokine-producing non-naïve CD4⁺ T cells (e.g., pie slice 4 represents TNFa⁺IFNγ⁻IL-21⁻) among the total populaton of antigen-specifc cells, stratified by antigenic stimuli. The p value shown under the pie charts is calculated using a partial permutation test. (B) The ratio between month 10 and day 20 of the absolute percentage of IFNγ⁺IL-21^-TNF-ɑ⁺- (IFN + TNF, left), IFNγ⁺IL-21^-TNF-ɑ^-- (IFNsp, middle), or IFNγ^-IL-21^-TNF-ɑ⁺-producing (TNFsp, right) non-naïve CD4⁺ T cells stimulated with the S protein is depicted (scatterplot; black line, median). (C) The absolute percentage (left) and the ratio between month 10 and day 28 of AIM⁺ (CD137⁺/OX40⁺) cTfh cells (right) stimulated with the MN (red) or S (blue) proteins and stratified by vaccination status are shown. (B and C) The p values shown are calculated using the Mann Whitney test. See also Figure S8.