Spatiotemporal regulation of type I interferon expression determines the antiviral polarization of CD4⁺ T cells

Abstract

Differentiation of CD4⁺ T cells into either follicular helper T (T_FH) or type 1 helper T (T_H1) cells influences the balance between humoral and cellular adaptive immunity, but the mechanisms whereby pathogens elicit distinct effector cells are incompletely understood. Here, we analyzed the spatiotemporal dynamics of CD4⁺ T cells during infection with recombinant vesicular stomatitis virus (VSV), which induces early, potent neutralizing antibodies or recombinant lymphocytic choriomeningitis virus (LCMV), which induces a vigorous cellular response, but inefficient neutralizing antibodies, expressing the same T cell epitope. Early exposure of dendritic cells to type I interferon (IFN), which occurred during infection with VSV, induced the production of the cytokine IL-6 and drove T_FH cell polarization, while late exposure to type I IFN, which occurred during infection with LCMV, did not induce IL-6 and allowed differentiation into T_H1 cells. Thus, tight spatiotemporal regulation of type I IFN shapes antiviral CD4⁺ T cell differentiation, and might instruct vaccine design strategies.

CD4⁺ T cells are key players in adaptive immune responses against pathogens. After priming in secondary lymphoid organs (SLOs), antigen-specific CD4⁺ T cells undergo clonal expansion and differentiation into specialized effector T cell subsets. Viral infection may result in the generation of follicular helper T cells (T_FH cells) and type 1 helper T cells (T_H1 cells) ^1,2. T_FH cells, which express the transcription factor Bcl-6, migrate into the B cell follicles where they promote the formation of high-affinity neutralizing antibodies^3,4. T_H1 cells, which express the transcription factor T-bet⁵, promote the activation of macrophages and CD8⁺ T cell responses. As such, T_FH and T_H1 cell subsets contribute to adaptive immune responses by specifically supporting humoral or cellular immunity, respectively.

Although humoral or cellular immunity can be found in a state of “competitive coexistence”⁶, one response usually emerges as dominant after viral infection and is responsible for most of the antiviral activity. Viruses that induce direct cell damage (cytopathic viruses), such as the vesicular stomatitis virus (VSV), typically induce early, potent neutralizing antibody responses⁷, while non-cytopathic viruses, such as the lymphocytic choriomeningitis virus (LCMV), usually elicit robust cellular responses, but weak and inefficient neutralizing antibody responses⁷. The relative inefficiency of non-cytopathic viruses to induce neutralizing antibodies has been attributed, among other factors, to the properties of viral surface proteins, the frequency of germline-encoded immunoglobulin variable regions, the CD8⁺ T cell-induced immunopathological changes in secondary lymphoid organs and the inflammatory monocyte-mediated suppression of B cells^7,8. Here, we investigated whether CD4⁺ T cell polarization had a role in the dichotomous response to VSV and LCMV and characterized the spatiotemporal dynamics of the differentiation of antiviral CD4⁺ T cells. We found that CD4⁺ T cells differentiated mostly to T_FH cells upon infection with VSV and mostly to T_H1 cells upon infection with LCMV. Regardless of the differentiation outcome, priming of CD4⁺ T cells occurred in the outer paracortex and in the interfollicular areas of the lymph nodes in both infections. The dichotomous T cell differentiation could not be explained by distinct cellular composition of the priming niches. Instead, the spatiotemporal regulation of type I interferon (IFN) expression determined whether dendritic cells (DC) in the lymph node produced the cytokine IL-6 and, consequently, shaped antiviral CD4⁺ T cell polarization.

Results

VSV and LCMV induce distinct antiviral CD4⁺ T cell polarization

First, we compared CD4⁺ T cell polarization upon VSV or LCMV infection in C57BL/6 mice. We adoptively transferred naïve VSV-specific (Tg7)⁹ or LCMV-specific (SMARTA)¹⁰ transgenic CD4⁺ T cells into C57BL/6 mice 24 h before subcutaneous intra-footpad infection with VSV Indiana (VSV Ind) or LCMV WE, respectively. CD4⁺ T cell polarization into Bcl-6⁺ CXCR5⁺ T_FH cells and T-bet⁺ CXCR5^- T_H1 cells was analysed at 3, 5, 7 and 14 days after infection in the footpad-draining popliteal lymph nodes (dLNs) (Extended Data 1a). On day 5 after VSV infection, >40% of Tg7 T cells differentiated into T_FH cells, with little or no differentiation into T_H1 cells (Fig. 1a); at the same time point, >80% of SMARTA T cells were T_H1 cells, and <15% became T_FH cells (Fig. 1a). This pattern of CD4⁺ T cell polarization during VSV and LCMV infections was also observed when we analysed the endogenous CD44^hiCD4⁺ T cell response in dLNs of mice subcutaneously infected with different viral strains (VSV Ind, VSV New Jersey, LCMV Armstrong, LCMV WE and LCMV clone 13) (Extended Data 1b, c), and it occurred independently of the route of infection (intrafootpad and intravenous infections) (Extended Data 1d) and the size of viral inoculum (10²-10⁷ pfu or ffu/mouse) (Extended Data 1e).

Fig. 1. VSV and LCMV infections result in distinct antiviral CD4⁺ T cell polarization and in vivo dynamics.

(a) Representative flow cytometry plots of transferred CD45.1⁺ Tg7 CD4⁺ T cells or CD45.1⁺ SMARTA CD4⁺ T cells (1 x 10⁶ each) in the footpad-draining popliteal LNs (dLNs) of CD45.2⁺ wild-type recipient mice 5 days post intrafootpad infection with VSV Ind (left) or LCMV WE (right), respectively (Extended Data 1a for schematic representation of experimental set up). Numbers indicate percentages within the indicated gates. Plots are representative of at least 5 independent experiments. (b) Representative flow cytometry plots of transferred CD45.1⁺ SMARTA CD4⁺ T cells (1 x 10⁶) in the footpad-draining popliteal LNs (dLNs) of CD45.2⁺ wild-type recipient mice 5 days post intrafootpad infection with rVSV (left) or rLCMV (right). Numbers indicate percentages within the indicated gates. Plots are representative of at least 5 independent experiments. (c) Quantification of T_FH cells (left) or T_H1 cells (right) as percentages of the transferred Tg7 (in case of VSV Ind infection) or SMARTA (in case of infection with rVSV or rLCMV) CD45.1⁺ CD4⁺ T cells in dLNs of CD45.2⁺ wild-type recipients at 0, 3, 5, 7 and 14 days post-infection. Mean+/- SEM is shown. Day 0 n = 3 (VSV), 4 (rVSV and rLCMV); day 3 n = 5 (VSV), 7 (rVSV and rLCMV); day 5 n = 5 (VSV), 9 (rVSV), 10 (rLCMV); day 7 n = 3; day 14 n = 3 (VSV and rVSV), 6 (rLCMV). Black and blue stars indicate significance of VSV and rVSV samples, respectively, compared to rLCMV samples. A two-way Anova with LSD post-test was applied. * p < 0.05; ** p < 0.01; *** p < 0.001; **** p < 0.0001 (d) Confocal micrographs of dLNs in wild-type (WT) mice injected with 1 x 10⁶ purified GFP⁺ Tg7 (VSV Ind) or SMARTA (all other groups) CD4⁺ T cells (green) and 3 x 10⁷ purified deep red-labelled polyclonal B cells (grey), 1 day prior to injection with PBS, VSV Ind, rVSV and rLCMV. The images on the bottom are the same as the images on top except that the channel pertaining to polyclonal B cells was removed to improve clarity. Images were collected 3 days after infection and are representative of at least 3 independent experiments. Scale bars represent 200 μm. (e) Quantification of follicular Ag-specific CD4⁺ T cells in dLNs of mice as in d. Mean+/- SEM is shown. n=2 (PBS), 4 (VSV Ind, rVSV, rLCMV). A one-way Anova with Bonferroni’s post-test was applied. *** p < 0.001 (f) Snapshots from multiphoton intravital imaging of dLNs in WT mice injected with 1 x 10⁶ purified GFP⁺ Tg7 (VSV Ind) or SMARTA (all other groups) CD4⁺ T cells (green) and 3 x 10⁷ purified deep red-labelled polyclonal B cells (grey), injected 1 day prior to injection with PBS, VSV Ind, rVSV and rLCMV (see Supplementary Movie 1). Cell tracks are coloured based on track speed mean values (blue = 0 μm/sec; red =1 μm/sec; see Supplementary Movie 1). Dashed lines define B cell follicles and were depicted based on polyclonal B cell positioning. Data are representative of at least 2 independent experiments. (g) Intravital cytometry plots of the movies in g showing either Tg7 (VSV Ind) or SMARTA (all other groups) CD4⁺ T cell XY cell positions over time (top), or meandering index versus track speed mean (bottom, see Supplementary Movie 1). Gates represent percentages of follicular CD4⁺ T cell tracks (top) and meandering index^hitrack speed mean^hi CD4⁺ T cell tracks (bottom). Numbers indicate the percentage of tracks within the indicated gates. Data are representative of at least 2 independent experiments.

TCR signal strength influences CD4⁺ T cell fate, with high-affinity antigen recognition associated to increased induction of T_FH cells ¹¹. To test whether the distinct CD4⁺ T cell differentiation during the VSV and LCMV infection was due solely to different TCR binding affinities, we infected C57BL/6 mice subcutaneously with recombinant VSV (rVSV) and LCMV (rLCMV) expressing glycoproteins containing the LCMV gp61-80 epitope recognized by SMARTA CD4⁺ T cells^12,13. On day 5, >40% of SMARTA CD4⁺ T cells differentiated into T_FH cells upon infection with rVSV, while >60% of cells became T_H1 cells upon infection with rLCMV (Fig. 1b, c and Extended Data 1f, g), indicating that features of the viral backbone dictated the cell fate of the CD4⁺ T cells, independently of the TCR affinity.

We next tested the localization and dynamic behaviour of CD4⁺ T cells in the LN by confocal immunofluorescence histology and intravital multiphoton microscopy (IVM). We imaged the draining LNs of wild-type mice transferred with CD4⁺ T cells from Tg7 x β-actin-GFP or SMARTA x β-actin-GFP mice on day 3 post-infection with VSV Ind, rVSV or rLCMV. At this time point, we detected Tg7 or SMARTA T cells in the B cell follicles following infection with either VSV Ind or rVSV, respectively (Fig. 1d, e); by contrast, SMARTA T cells in rLCMV-infected mice were confined outside the B cell follicles (Fig. 1d, e). The imaged Tg7 or SMARTA T cells had a higher velocity and higher meandering index, an indicator of track straightness, in mice infected with either VSV Ind or rVSV, respectively, than SMARTA cells in mice infected with rLCMV (Fig. 1f, g, Extended Data 2a, b and Supplementary Movie 1). These data indicated that VSV and LCMV induced distinct antiviral CD4⁺ T cell polarization and in vivo dynamic behaviour, independently of viral strain, viral dose, infection route and TCR signal strength.

Characterization of the antiviral CD4⁺ T cell priming niche

Next, we characterized the precise location in the LN where priming and differentiation of the antiviral CD4⁺ T cell took place. CD4⁺ T cell priming is thought to happen within the first two days upon antigen administration in the outer paracortex and interfollicular areas of LNs^14–17. Consistent with this, 48 h post-infection with rVSV or rLCMV, SMARTA CD4⁺ T cells had upregulated the activation markers CD69 and CD25 (Extended Data 2c, d) and migrated to the outer paracortex and interfollicular areas of the infected LNs (Fig. 2a, b, Extended Data 2e and Supplementary Movie 2), where they started to form clusters (Fig. 2a, c, Extended Data 2f and Supplementary Movie 2). We refer to these areas as CD4⁺ T cell ‘priming niches’ hereafter. To test whether the cellular and molecular composition of the niche supporting CD4⁺ T cell priming differed between the rVSV or rLCMV infections, we used NICHE-seq, which combines photoactivation and single-cell RNA-sequencing (scRNA-seq) to spatially reconstruct immune niches¹⁸. We intravenously transferred CD4⁺ T cells from SMARTA x β-actin-CFP mice into transgenic mice ubiquitously expressing a photoactivatable green fluorescent protein (PA-GFP)¹⁹; we then photoactivated areas in the dLNs containing SMARTA CD4⁺ T cell clusters (i.e. in the outer paracortex and interfollicular areas) on day 2 post-infection with rVSV or rLCMV and single-cell sorted 2406 photoconverted total GFP⁺ endogenous cells (Fig. 2d, e). scRNA-seq indicated that B cells and NK cells were over-represented in the samples from rVSV-infected mice, whereas CD8⁺ T cells and CCR2⁺ inflammatory monocytes were enriched in the samples from rLCMV-infected mice (Fig. 2e, f; cell subsets were identified based on their gene expression signatures, see Methods). Confocal microscopy indicated the presence of Ag-specific B cells, CX3CR1⁺CCR2⁺ cells, Ag-specific CD8⁺ T cells and NKp46⁺ cells in the CD4⁺ T cell-priming niches (Extended Data 3a-d).

Fig. 2. Characterization of the antiviral CD4⁺ T cell priming niche.

(a) Confocal micrographs of dLNs in mice injected with 5 x 10⁶ purified GFP⁺ Ag-specific (SMARTA) CD4⁺ T cells (green) and 3 x 10⁷ purified deep red-labelled polyclonal B cells (grey), 1 day before injection with PBS, rVSV or rLCMV. Images were acquired 2 days after infection and are representative of at least 2 independent experiments. The top panels depict the positioning of GFP⁺ Ag-specific (SMARTA) T cells whereas the bottom panels indicate the clustered Ag-specific (SMARTA) T cells (purple). A T cell cluster was defined as a minimum of 3 T cells within a distance of 15 μm measured from the centroid of each cell (see Materials and Methods). Scale bars represent 200 μm. See also Supplementary Movie 2. (b, c) Quantification of the normalized distance of Ag-specific (SMARTA) T cells from the T cell area centre (b) and percentages of clustered T cells / section (c) in the mice described in a. Mean+/- SEM is shown. PBS n =772, rVSV n=1025, rLCMV n=1127 (b); PBS n = 3, rVSV n=5, rLCMV n=4 (c). Results are pooled from 2 independent experiments. A one-way Anova with Bonferroni’s post-test was applied. * p < 0.05; **** p < 0.0001. (d) Multiphoton intravital micrographs depicting the photoactivation of the CD4⁺ T cell priming niche upon rVSV or rLCMV infection. Images were acquired 2 days after infection. Ag-specific (SMARTA) CFP⁺ CD4⁺ T cells are depicted in green and PA-GFP⁺ in blue. Top panels, before photoactivation; bottom panels, after photoactivation. Scale bars represent 50 μm. Images are representative of at least 2 independent experiments. (e) Gene expression profiles of 2406 single cells from photoactivated priming niches, grouped in 7 clusters. The colour bar on the bottom indicates each cell’s origin (blue: photoactivated cells from rVSV infected mice; red: photoactivated cells from rLCMV infected mice). (f) Relative abundance of different cell clusters within CD4⁺ T cell priming niches. Data represent cell counts from two biological replicates. Two-sided FDR-adjusted Fisher’s exact test. * p < 0.05; *** p < 0.001; **** p < 0.0001. (g-i) 1 x 10⁶ purified CD45.1⁺ Ag-specific (SMARTA) CD4⁺ T cells were injected into CD45.2⁺ WT recipients and into either VI10YEN (g, n=6), Cor93 Tg TCR (h, n=4 or 6), or Ccr2^-/- (I, n=4 or 6) mice. dLNs were collected 3 (g) or 5 (h-i) days after rVSV (blue) or rLCMV (red) infection. Percentages of T_FH (g-i, left) and T_H1 (h-i, right) Ag-specific CD4⁺ T cells out of total transferred cells in dLNs were quantified by flow cytometry. Data are representative of at least 2 independent experiments. Mean+/- SEM is shown. A one-way Anova with Bonferroni’s post-test was applied. ** p < 0.01; *** p < 0.001

B cells are required for the later stages of differentiation into T_FH cells ³. To test whether a higher number of B cells in the CD4⁺ T cell-priming niche in the VSV-infected mice drove differentiation into T_FH cells, we used VI10Yen mice, in which B cells express a transgenic BCR specific for an irrelevant antigen (VSV Ind glycoprotein)²⁰. Adoptively transferred SMARTA CD4⁺ T cells differentiated into T_FH cells and homed to the B cell follicles similarly in wild-type or VI10Yen mice after infection with rVSV (Fig. 2g and Extended Data 4a), indicating that Ag-specific B cells were not needed for the initial differentiation of CD4⁺ T cells. NK cells were reported to curtail virus-specific CD4⁺ T cell responses, including T_FH cell responses, during chronic virus infection²¹. However, depletion of NKp46⁺ cells, including NK cells, in NKp46-Cre x Rosa26-iDTR mice ^22,23 injected with diphtheria toxin, indicated NK cells were dispensable for early SMARTA CD4⁺ T cell differentiation upon rVSV or rLCMV infection (Extended Data 5a-c). Immunopathological changes in secondary lymphoid organs or direct killing of B cells by CD8⁺ T cells can subvert B cell responses^8,24–26. The use of Cor93 Tg recipient mice, in which T cells bear a TCR specific for an irrelevant antigen (the hepatitis B virus core protein)²⁷, indicated that Ag-specific CD8⁺ T cells were not required for the early polarization of SMARTA CD4⁺ T cells upon rVSV or rLCMV infection (Fig. 2h and Extended Data 4b). The recruitment of CCR2⁺ inflammatory monocytes to the LCMV-infected LNs was shown to hinder LCMV-specific B cell responses^8,28. However, Ccr2^-/- recipient mice, in which monocytes cannot migrate to infected LNs^28,29, were indistinguishable from wild-type mice in their ability to support early SMARTA CD4⁺ T cell differentiation upon rVSV or rLCMV infection (Fig. 2i and Extended Data 4c).

Priming of naïve CD4⁺ T cells within LNs is mediated by mature conventional dendritic cells (cDC), which present antigen and deliver costimulatory stimuli and cytokines^14,15,17,30. cDCs were detected in the T_FH cell and T_H1 cell priming niches by NICHE-seq (Fig. 2e, f), but the relative paucity of these cells within the photoactivated cells prevented their detailed phenotypic characterization. The membrane receptor ICOSL is sensitive to ICOS-induced shedding during interactions with activated ICOS^hi T cells³¹. To determine the DC subsets that interacted with the CD4⁺ T cells upon VSV or LCMV infection, we measured the expression of ICOSL on CD11c⁺MHCII^hi DCs in rVSV- or rLCMV-infected mice treated with a blocking antibody against ICOS, as an indirect indicator of the interaction between LN DCs and cognate T cells. In both rVSV- and rLCMV-infected mice, CD11c⁺MHCII^hiCD11b⁺ cDC2, but not CD11c⁺ MHCII^hiCD8⁺ cDC1, had increased expression of ICOSL (Extended Data 5d, e), suggesting that interactions between cDC2 and cognate CD4⁺ T cells drive the differentiation of both T_FH and T_H1 cells. In addition, SMARTA CD4⁺ T cells transferred into XCR1-DTR mice, in which cDC1s were depleted by diphtheria toxin injection starting 2 days before infection, differentiated normally into T_FH or T_H1 cells upon rVSV or rLCMV infection, respectively (Extended Data 5f). Thus, the differences in helper T cell polarization observed during the infection with VSV and LCMV were due to priming by DC subsets other than cDC1, possibly exposed to a different local cytokine milieu.

Spatiotemporal regulation of type I IFN expression determines CD4⁺ T cell polarization

Next, we analysed the transcriptome of the cells in the photoactivated CD4⁺ T cell priming niche of rVSV- and rLCMV-infected mice. NICHE-seq analysis indicated an increased type I IFN signature in cells isolated from the CD4⁺ T cell-priming niche of rLCMV-infected mice compared to cells from the CD4⁺ T cell-priming niche of rVSV-infected mice on day 2 post-infection (Fig. 3a). Although sensing of type I IFN is known to influence CD4⁺ T cell polarization, the effect of type I IFN on the activation and differentiation of antiviral CD4⁺ T cells remains ill defined, with conflicting reports, possibly depending on the context of infection. Kinetic analysis of IFN-α2, IFN-α5, IFN-α6, IFN-α7, IFN-α9, IFN-α12, IFN-α13, IFN-α14, IFN-β, and two representative IFN-stimulated genes (ISGs), ISG15 and OAS2, in total RNA isolated from LNs at 4, 8, 16, 24 and 48 h post-infection with rVSV or rLCMV indicated that although the magnitude of type I IFN induction did not significantly differ between the rVSV or rLCMV infection, rVSV induced an earlier wave of type I IFN, which peaked at 8 h after infection, whereas rLCMV induced a delayed (24 h) and prolonged peak (Fig. 3b and Extended Data 6), possibly reflecting the different kinetics of the viral replication in vivo ²⁸. REX3 reporter mice express the red fluorescent protein (RFP) and the blue fluorescent protein (BFP) under the control of the promoters of the ISGs Cxcl9 and Cxcl10, respectively ¹⁷. We detected stronger expression of CXCL9-RFP and CXCL10-BFP in the outer paracortex, cortical ridge and interfollicular areas in the LNs of rVSV-infected REX3 mice compared to rLCMV-infected mice 12 h post-infection, corresponding to areas where the SMARTA CD4⁺ T cells accumulated (Fig. 3c).

Fig. 3. Spatiotemporal regulation of type I IFN expression determines antiviral CD4⁺ T cell polarization.

(a) Expression profile of selected ISGs (Irf7, Cxcl9, Cxcl10, Oasl1, Ifitm3, Oas2, Isg15) in 2406 single cells from the photoactivated CD4⁺ T cell priming niches described in Fig. 2 d-f. Data are pooled from 2 independent experiments. Kolmogorov–Smirnov test was applied. **** p < 0.0001. (b) Analysis of Ifna4, Ifnb (top) and Isg15, Oas2 (bottom) gene expression in dLN at 0, 4, 8, 16, 24 and 48 hours after rVSV (blue) or rLCMV (red) infection by qPCR. 0 hours n = 3; 4 hours n = 4; 8 hours n = 4; 16 hours n = 4; 24 hours n = 3 (rLCMV), 4 (rVSV); 48 hours n = 4. Data are pooled from 2 independent experiments. Mean+/- SEM is shown. A two-way Anova with LSD post-test was applied. * p < 0.05; ** p < 0.01; *** p < 0.001. The same sample was measured repeatedly for the 4 genes. (c) Confocal micrographs of dLNs in REX3 reporter mice injected with 1 x 10⁷ purified GFP⁺ Ag-specific (SMARTA) CD4⁺ T cells (green) and 3 x 10⁷ purified deep-labelled polyclonal B cells (grey), 12 hours after rVSV or rLCMV infection. CXCL10⁺ (blue) and CXCL9⁺ (red) cells are depicted. Data are representative of at least 2 independent experiments. (d) Representative flow cytometry plots show T_FH and T_H1 within Ag-specific CD4⁺ T cells, 5 days after infection of CD45.2⁺ WT recipients injected with 1 x 10⁶ purified CD45.1⁺ SMARTA CD4⁺ T cells and treated with anti-IFNAR1 blocking antibody (or isotype control) 1 day prior to rVSV (blue, left) or rLCMV (red, right) infection. Numbers represent the percentage of cells within the indicated gate. (e) Quantification of T_FH (top) and T_H1 (right) – expressed as percentages of Ag-specific CD4⁺ T cells out of total transferred cells – in dLNs of mice described in d. Mean+/- SEM is shown. n=6 (rVSV), 8 (rLCMV). Data are representative of at least 2 independent experiments. A one-way Anova with Bonferroni’s post-test was applied. *** p < 0.001. (f) Quantification of the percentages of T_FH (left) and T_H1 (right) Ag-specific CD4⁺ T cells (out of total transferred cells) in dLNs 5 days after infection of CD45.2⁺ WT recipients injected with 1 x 10⁶ purified CD45.1⁺ Ag-specific (SMARTA) CD4⁺ T cells and treated anti-IFNAR1 blocking antibody either 1 day prior to (light blue) or 1 day after (yellow) rVSV infection. Data are representative of at least 2 independent experiments. Mean +/- SEM is shown. n=4. A one-way Anova with Bonferroni’s post-test was applied. ** p < 0.01; *** p < 0.001. (g) Quantification of the percentages of T_FH (left) and T_H1 (right) Ag-specific CD4⁺ T cells (out of total transferred cells) in dLNs 5 days after infection (bottom) of CD45.2⁺ WT recipients injected with 1 x 10⁶ purified CD45.1⁺ Ag-specific (SMARTA) CD4⁺ T cells, infected with rLCMV and treated or not with Poly(I:C).. Data are representative of at least 2 independent experiments. PBS n=7, Poly(I:C) n=9. An unpaired two-tailed t test was applied. ** p < 0.01.

To test the effect of the two profiles of type I IFN induction on the differentiation of CD4⁺ T cells, we transferred SMARTA CD4⁺ T cells into wild-type mice and treated them with an antibody that blocks signalling through the type I IFN receptor (IFNAR1) 1 day before infection with rVSV or rLCMV. Whereas the CD4⁺ T cells transferred in IFNAR1-blocked rLCMV-infected mice differentiated into T_H1 cells, induction of T_FH cells in the rVSV-infected mice was severely compromised, with transferred CD4⁺ T cells differentiating into T-bet⁺ CXCR5^- T_H1 cell, which were excluded from B cell follicles and induced fewer germinal centre B cells and GP-specific IgG1 antibodies (Fig. 3d, e, Extended Data 7a, b, and Supplementary Movie 3), suggesting that type I IFN affected CD4⁺ T cell polarization at early time points post-infection. Administration of the blocking Abs against IFNAR1 on day 1 post-rVSV infection did not affect the differentiation of SMARTA cells into T_FH cells compared to untreated controls (Fig. 3f, Extended Data 7c), indicating that type I IFN sensing within the first 24 h post-infection was essential for the induction of T_FH cells. To assess whether an early wave of type I IFN increased the differentiation of T_FH cells in LCMV-infected mice, we injected poly(I:C), which binds TLR3 to induce type I IFN, simultaneously with rLCMV. On day 5 post-infection, we observed a significant increase in T_FH cell differentiation and a reduction in T_H1 cells in the LNs of poly(I:C)-treated mice compared to untreated mice (Fig. 3g). Together, these results indicate that early (<24 h) type I IFN sensing drives T_FH cell differentiation.

Early type I IFN sensing by DC and IL-6 are essential for T_FH cell differentiation

Type I IFN sensing by different cell types has been reported to influence CD4⁺ T cell differentiation in a variety of experimental settings, with data pointing to either a stimulatory or inhibitory role in the polarization towards either T_FH and/or T_H1 cells ³². To investigate the role of type I IFN receptor signalling in CD4⁺ T cells, we adoptively transferred wild-type or Ifnar1^-/- SMARTA CD4⁺ T cells into wild-type recipients 1 day before rVSV infection. On day 5 post-infection, we observed no differences between wild-type and Ifnar1^-/- SMARTA cell differentiation into T_FH or T_H1 cells (Fig. 4a), indicating that intrinsic type I IFN sensing by CD4⁺ T cells is dispensable for CD4⁺ T cell differentiation. We next investigated the role of type I IFN sensing in DCs, which had the most upregulated expression of ISGs in the CD4⁺ T cell priming niche (Extended Data 8). Wild-type SMARTA CD4⁺ T cells adoptively transferred into irradiated wild-type mice reconstituted with CD11c-Cre x Ifnar1^fl/fl bone marrow 8 weeks before CD4⁺ T cell transfer showed significantly impaired differentiation into T_FH cells on day 5 post rVSV infection (Fig. 4b), indicating that type I IFN sensing by DCs was essential for antiviral T_FH induction.

Fig. 4. Early type I IFN sensing by DC and IL-6 are essential for antiviral T_FH differentiation.

(a) Quantification of the percentages of T_FH and T_H1 Ag-specific CD4⁺ T cells (out of total transferred cells) in dLNs 5 days after rVSV infection of CD45.2⁺ WT mice injected with 1 x 10⁶ purified CD45.1⁺ WT or Ifnar1^-/- Ag-specific (SMARTA) CD4⁺ T cells. Mean +/- SEM is shown. n=3 (b) Quantification of the percentages of T_FH and T_H1 Ag-specific (SMARTA) CD4⁺ T cells (out of total transferred cells) in dLNs 5 days after rVSV infection of either CD45.2⁺ CD11c-Cre or CD45.2⁺ CD11c-Cre x Ifnar1^fl/fl recipients injected with 1 x 10⁶ purified CD45.1⁺ Ag-specific (SMARTA) CD4⁺ T cells. dLNs were collected 5 days after rVSV infection. Data are representative of 3 independent experiments. Mean+/- SEM is shown. CD11c-Cre n=3, CD11c-Cre x Ifnar1^fl/fl n=4. An unpaired two-tailed t test was applied. *** p < 0.001; **** p < 0.0001. (c) qPCR analysis of Il6 gene expression profile at 0, 4, 8, 16, 24 and 48 hours in dLNs of rVSV (blue)- or rLCMV (red)-infected mice. Mean+/- SEM is shown. 0 hours n = 4; 4 hours n = 4; 8 hours n = 4; 16 hours n = 4; 24 hours n = 3 (rLCMV), 4 (rVSV); 48 hours n = 4. Data are pooled from 2 independent experiments. A two-way Anova with LSD post-test was applied. * p < 0.05; ** p < 0.01. (d) qPCR analysis of Il6 gene expression in either wild-type or Ifnar1^-/- mice in dLNs 8 hours after rVSV infection. Data are pooled from 2 independent experiments. Mean +/- SEM is shown. An unpaired two-tailed t test was applied. * p < 0.05. (e) Quantification of the percentages of T_FH (left) and T_H1 (right) Ag-specific CD4⁺ T cells (out of total transferred cells) in CD45.2⁺ wild-type recipients injected with 1 x 10⁶ purified CD45.1⁺ Ag-specific (SMARTA) CD4⁺ T cells and treated with anti-IL-6 blocking antibody starting either 1 day prior to (yellow) or 1 day after rVSV infection (orange). Data are representative of at least 2 independent experiments. Mean +/- SEM is shown. No Ab n=5, IL-6 Ab d -1 n=5, IL-6 Ab d +1 n=3. A one-way Anova with Bonferroni’s post-test was applied. * p < 0.05; ** p < 0.01.

IL-6 is known to promote early T_FH differentiation^33,34 and its induction to be dependent on type I IFN³⁵. Kinetic analysis of Il6 expression in LNs of wild-type mice infected with either rVSV or rLCMV indicated that induction of Il6 mRNA had an early peak (8 h post-infection) during VSV infection and a delayed (>16 h) and lower peak during LCMV infection (Fig. 4c), mirroring the type I IFN signature. Ifnar1^-/- mice had no upregulation of Il6 mRNA 8 h after rVSV infection (Fig. 4d), indicating the VSV-induced early expression of Il6 required type I IFN sensing. Blocking IL-6 prior to, but not 24 hours after rVSV infection significantly impaired the induction of T_FH cells and the virus-specific antibody titres and increased the induction of T_H1 cells (Fig. 4e, Extended Data 9a, b), while it did not affect T helper cell polarization upon rLCMV infection (data not shown). These observations identified IL-6 as a critical early determinant of the differentiation of antiviral T_FH cells.

To determine whether type I IFN induced the expression of IL-6 on DCs, we assessed the composition and transcriptional state of DC subsets after early or late type I IFN sensing. scRNA-seq on 2179 quality-controlled (QC)-positive CD11c⁺MHC-II^hi DCs sorted from the lymph nodes of wild-type or Ifnar1^-/- mice at 8 h or 48 h post-infection with rVSV or rLCMV and unbiased analysis using the Metacell package³⁶ indicated that the CD11c⁺MHC-II^hi DCs could be subsetted into migratory cDC2, cDC1, cDC2, monocyte-derived DCs (moDCs) and a small subset of contaminant macrophages (Fig. 5a). Experimental conditions where type I IFN signalling is maximal (i.e. 8 h after rVSV infection and 48 h after rLCMV infection of wild-type mice compared to the same time points in Ifnar1^-/- mice or 8 h after rLCMV infection) showed an enrichment of DC subsets (i.e. migratory cDC2 and moDCs) that can support T_FH cell differentiation (Fig. 5b)³⁷. Furthermore, ISGs induced by type I IFN were upregulated in DCs from wild-type, but not from Ifnar1^-/- mice, particularly in experimental conditions (8 h post-rVSV and 48 h post-rLCMV infection) where type I IFN signalling is maximal (Fig. 5c, d and Extended Data 10), indicating that DCs responded to type I IFN. Importantly, migratory cDC2 and moDCs produced Il6 at 8 hours after rVSV infection of wild-type but not Ifnar1^-/- mice (Fig. 5e); by contrast, after rLCMV infection, Il6 mRNA had very low expression in migratory cDC2 and moDCs from wild-type and Ifnar1^-/- mice even at 48 h post-infection, when other ISGs were maximally induced (Fig. 5e). Thus, DCs produced Il6 and drove T_FH cell polarization in response to early (rVSV) but not late (rLCMV) type I IFN signalling.

Fig. 5. scRNA-seq analysis of lymph node DCs upon viral infection.

(a) Gene expression profiles of 2179 single QC-positive lymph node CD11c⁺ MHC-II^high cells grouped in 5 clusters. The colour bars on the bottom indicates each cell’s origin (blue: cells from the popliteal lymph node of mice 8 hours after intrafootpad rVSV infection; pink: cells from the popliteal lymph node of mice 8 hours after intrafootpad rLCMV infection; red: cells from the popliteal lymph node of mice 48 hours after intrafootpad rLCMV infection; light grey: WT mice; dark grey: Ifnar1^-/- mice). (b) Relative abundance of different DC clusters 8 hours after rVSV infection, 8 hours after rLCMV infection and 48 hours after rLCMV infection. Dashed lines show changes in cluster abundances between WT and Ifnar1^-/- mice under the same infection condition. Significant changes in abundances are highlighted. Two-sided FDR-adjusted Fisher’s exact test. * p < 0.05; *** p < 0.001 (c) Colour bar shows total size-normalized expression of interferon stimulated genes (ISGs) in cells in (a). Detailed expression of individual ISG genes is shown in Extended Data Fig. 10. (d) Distribution of total size-normalized expression of ISGs in CD11c⁺ MHC-II^high cells grouped by experimental condition. (e) Total expression of the ISG Ifit1, and the T_FH cell-promoting cytokine Il6 in different experimental conditions. Values represent size-normalized total transcripts per 100 cells. Colours represent relative contribution from the five different DC subsets.

Discussion

Here, we identified the tight spatiotemporal regulation of the expression of type I IFN as a critical determinant of CD4⁺ T cell fate upon viral infection. When DCs were exposed to an early (<24 h) wave of type I IFN, they made IL-6, thus promoting the differentiation of T_FH cells and, consequently, enhancing humoral immunity; in contrast, when DCs were exposed to late type I interferon (>24 h), they no longer produced IL-6, and CD4⁺ T cells differentiated into a non-T_FH cell fate. These results might explain why many non-cytopathic, slow-replicating viruses fail to induce or interfere with the generation of neutralizing antibodies^7,8.

The notion that type I IFN influences the differentiation of CD4⁺ T cells is not without precedent ³². However, its role in CD4⁺ T cell polarization has remained controversial. While some studies have shown type I IFN promoted T_FH cell differentiation, others reported that type I IFN suppressed the formation of T_FH cells or induced T_H1 cells³². Our results suggested that the spatial and temporal regulation of type I IFN was critical for its effect on DCs, thus providing a potential explanation for the abovementioned conflicting results. The cellular source of type I IFN in VSV and LCMV infections is an interesting question, but unfortunately, one that is technically difficult to address because of the many isoforms of type I IFN and the lack of a sensitive, robust type I IFN reporter mouse that would allow for the unambiguous identification of cells producing small amounts of these cytokines^38,39. VSV induces production of type I IFN in subcapsular sinus macrophages and plasmacytoid dendritic cells⁴⁰. However, the relative contribution of each abovementioned lymph node cell type to the production of type I IFN after subcutaneous infection with LCMV or the recombinant viral strains used in this study are unknown⁴¹. Nevertheless, it is tempting to speculate that subcapsular sinus macrophages might be the critical cellular source of the early type I IFN response necessary for T_FH cell differentiation. Pertinent to this, it is worth noting that viral infections induce a relocalisation of subcapsular sinus macrophages to the inner follicular areas and towards the anatomical niche where antiviral CD4⁺ T cell priming takes place ⁴².

Our data identify DCs as the necessary platform that integrates early type I IFN signalling to promote T_FH cell differentiation. Mechanistically, type I IFN is necessary for the induction of the T_FH cell-promoting cytokine IL-6³⁵. Our scRNA-seq data identified subsets of DCs (migratory cDC2 and moDCs) that produced Il6 mRNA in response to early, but not late, type I IFN. One experimental caveat is that we analysed DCs from the entire lymph node, rather than from the CD4⁺ T cell priming niche, thus potentially underestimating the localized production of IL-6 by DCs. The relative contribution of DC-derived IL-6, compared to IL-6 from other cellular sources, in shaping antiviral CD4⁺ T cell polarization remains to be determined.

We found that CD4⁺ T cells that differentiated into T_FH cells after rVSV infection showed relatively high velocity and high meandering index, whereas CD4⁺ T cells that differentiated into T_H1 cells after rLCMV infection were characterized by reduced mean speed and lower meandering index. The mechanistic explanations for these different motility behaviours remain elusive, and might potentially include intrinsic differences in motility between T_FH cells and T_H1 cells, differences in spatial constrains provided by the different anatomical niche they occupy (B cell follicles versus interfollicular and T cell areas), different adhesion molecule expression and different antigen levels.

Blockade of early type I IFN signalling or blocking of IL-6 resulted in reduced generation of T_FH cells and, consequently, in reduced antiviral antibody responses. Of note, whereas the main effect of type I IFN blockade was observed on antiviral IgG1 titres, IL-6 affected mostly the levels of antiviral IgG2b. The reason for this discrepancy is unknown but might lie with additional T-independent roles of IL-6 that affect B cell differentiation ⁴³.

Although the data presented in this study linked T_FH cell differentiation to early type I IFN sensing by DCs, they do not fully explain the strong T_H1 cell differentiation observed during LCMV infection. In addition to a lack of early type I IFN and IL-6, as reported here, LCMV might promote T_H1 cell differentiation through the induction of T_H1 cell-polarizing cytokines, such as IL-12 and IFN-γ^44–46. Although IL-12 is dispensable for T_H1 cell polarization upon LCMV infection^47,48, the role of other cytokines such as IFN-γ on antiviral CD4⁺ T cell differentiation, its cellular source and its potential interference with IL-6 induction warrants further investigation.

In conclusion, we have characterized the cellular and molecular composition of the LN niches where CD4⁺ T cell differentiation occurs, delineated the spatiotemporal dynamics of the ensuing T_H cell subsets and identified the tight spatiotemporal regulation of type I IFN expression as a critical regulator of antiviral CD4⁺ T cell polarization. Our model predicts that if viruses induce an early wave of type I IFN (a situation that typically occurs with fast-replicating, highly cytopathic viruses that are controlled by neutralizing antibodies), this cytokine acts on migratory cDC2 and/or monocyte-derived dendritic cells to induce IL-6 (and possibly other T_FH cell-promoting cytokines), drive T_FH cell differentiation and ultimately enhance humoral immunity; by contrast, if type I IFN is induced later upon viral infection (a situation that is typical of slow-replicating, non-cytopathic viruses that are controlled by the CD8⁺ T cell response), DCs no longer make IL-6 and T_H cell polarization is biased towards non-T_FH cell fates such as T_H1 cells, which favour cellular immunity at the expense of B cell responses.

Methods

Mice

C57BL/6, CD45.1 (inbred C57BL/6), β-actin-CFP (B6.129(ICR)-Tg(CAG- ECFP)CK6Nagy/J), β -actin-GFP [C57BL/6-Tg(CAG-EGFP)1Osb/J], Ccr2^-/- (B6.129S4-Ccr2tm1Ifc/J), Rosa26-ZsGreen [B6.Cg-Gt(ROSA)26Sor^{tm6(CAG-ZsGreen1)Hze}/J]⁴⁹ and PA-GFP¹⁹ mice were purchased from Charles River or from the Jackson Laboratory. LCMV-P14⁵⁰, SMARTA¹⁰, Tg7⁹, and Ifnar1^-/-51 mice were obtained through the Swiss Immunological Mouse Repository (SwImMR, Zurich, Switzerland). Heavy chain knock-in and light chain BCR transgenic mice specific for VSV Indiana (VI10YEN²⁰) were obtained through the European Virus Archive. BCR transgenic mice specific for LCMV WE (KL25) bred against b-actin–DsRed were described in ²⁸. Cor93 TCR transgenic mice (lineage BC10.3, inbred CD45.1), in which > 98% of the splenic CD8+ T cells recognize a Kb-restricted epitope located between residues 93–100 in the HBV core protein (MGLKFRQL), have been previously described²⁷. XCR1-DTR-Venus, CD11c-cre and Ifnar1^fl/fl have been previously described^52–54. NKp46-Cre ²² mice were obtained from Eric Vivier. Rosa26-iDTR²³ mice were obtained from Marco Bacigaluppi. CX3CR1^{GFP/+ 55} and CCR2^{RFP/+ 56} mice were provided by I. Charo (University of California, San Francisco) by way of B. Engelhardt (University of Bern). Bone marrow chimeras were generated by irradiation of C57BL/6 mice with ~900 rad and reconstitution with the indicated bone marrow; mice were allowed to reconstitute for at least 8 weeks prior to use. Mice were housed under specific pathogen-free conditions and used at 8-10 weeks of age, unless otherwise indicated. In all experiments, mice were matched for age and sex before experimental manipulation. All experimental animal procedures were approved by the Institutional Animal Committee of the San Raffaele Scientific Institute.

Infections and immunizations

Unless otherwise indicated, mice were infected intrafootpad with 1x10⁵ plaque-forming units (pfu) of VSV serotype Indiana (Ind), VSV serotype New Jersey (NJ), rVSV (a recombinant VSV expressing a LCMV glycoprotein recognized by both SMARTA TCR transgenic and KL25 BCR transgenic cells instead of the VSV glycoprotein), 1x10⁵ focus forming units (ffu) of LCMV Armstrong (LCMV Arm), LCMV WE, LCMV clone 13 (LCMV Cl13), rLCMV (a recombinant LCMV Clone 13 expressing a LCMV glycoprotein recognized by both SMARTA TCR transgenic and KL25 BCR transgenic cells instead of the LCMV Cl13 glycoprotein)^12,28.

In indicated experiment mice were infected with 1x10², 1x10⁵ or 1x10⁷ pfu of VSV Ind or 1x10², 1x10⁵ or 1x10⁷ ffu of LCMV WE. In indicated experiments mice were infected intravenously with 1x10⁶ ffu of LCMV Arm or LCMV Cl13, or 1x10⁶ pfu of VSV Ind. Viruses were propagated and quantified as described^12,28, and diluted in 25 μl of PBS prior to footpad injection or in 200 μl of PBS prior to intravenous injection.

Mice were retro-orbitally bled at the indicated time points for GP-specific Abs measured by GP-1–binding enzyme-linked immunosorbent assay (ELISA)⁵⁷. Anti-mouse IgG1 and IgG2b antibodies of the SBA Clonotyping System-C57BL/6-HRP kit (SouthernBiotech) were used 1:1000 to detect GP-binding Abs.

All infectious work was performed in designated Biosafety Level 2 (BSL-2) and BSL-3 workspaces in accordance with institutional guidelines.

T and B cell isolation, adoptive transfer and in vivo treatments

Naïve CD4⁺ T cells from spleens of SMARTA CD45.1⁺, SMARTA GFP⁺, SMARTA CFP⁺, SMARTA Ifnar1^-/- transgenic mice, CD8⁺ T cells from spleens of LCMV-P14-GFP mice and naïve B cells from the spleen of WT mice and KL25-DsRed mice were negatively selected by magnetic isolation (Miltenyi Biotec), with purity always above 98% as determined by flow cytometry. Unless otherwise indicated, 1 x 10⁶ SMARTA CD45.1⁺ T cells were injected intravenously into indicated recipients one day prior to intrafootpad infection. In all imaging experiments, 3 x 10⁷ purified naïve B cells were labelled with 1 μM Deep Red (Life Technologies) for 20 min at 37°C in plain RPMI prior to adoptive transfer. In selected imaging experiments (12 hours and 48 hours after infection), 1 x 10⁷ or 5 x 10⁶ SMARTA GFP⁺ or CFP⁺, 5 x 10⁶ P14 GFP⁺, 5 x 10⁶ KL25 DsRed⁺ cells were injected, respectively. In indicated experiments, mice were treated with: the InVivoMab anti-IFNAR (MAR-1 5A3, BioXcell; 1 mg i.v. one day prior to or one day after infection) or mouse IgG1 isotype control (MOPC-21, BioXcell); anti-IL-6 (MP5-20F3, BioXcell; 0.5 mg i.v. one day prior to or one day after infection and 0.25 mg every other day) or rat IgG1 isotype control (BioXcell); the InVivoMab anti-ICOS (7E.17G9, BioXcell, 0.5mg i.v. two and one days prior to infection) or with rat IgG2b isotype control (LTF-2, BioXcell); 25 μg of Poly(I:C) HMW (InvivoGen, intrafootpad injection the day of the infection and one day after). To deplete cDC1 and NK cells, 500 ng of diphtheria toxin (DT, Millipore) diluted in 200 μl of PBS was administered intraperitoneally 1 day before the infection and every other day thereafter to XCR1-DTR and NKp46-DTR mice, respectively.

Cell isolation and flow cytometry

Single-cell suspensions of spleens and LNs were generated as described^28,58. All flow cytometry stainings of surface-expressed and intracellular molecules were performed as described²⁸. Antibodies used included: Bcl-6 (K112-91), T-bet (4B10), CXCR5 (2G8), CD4 (RM4-5), CD45.1 (A20), CD45.2 (K104), CD44 (IM7), IgD (11-26C.2A), NK1.1 (PK136), TCRβ (H57-597), CD8 (K53-6.7), CD11c (N418), CD11b (M1/70), MHCII (AF6-120.1), IgG1 (A85-1) CD95 (JO2), B220 (RA3-6B2), CD69 (H1.2F3) and CD25 (PC61). Fluorochrome-conjugated Abs were purchased from BioLegend, eBioscience or BD Pharmingen. All flow cytometry analyses were performed in FACS buffer containing PBS with 2 mM EDTA and 2% FBS on a FACS CANTO (BD Pharmingen) and analysed with FlowJo software (Treestar).

Confocal immunofluorescence histology

Confocal microscopy analysis of popliteal LNs was performed as previously described²⁸. The following primary Abs were used for staining: rat anti-B220 (RA3-6B2), rabbit anti-GFP (Invitrogen). Images were acquired on an inverted Leica microscope (SP8, Leica Microsystems) with a motorized stage for tiled imaging using a HC PL APO CS2 20X objective (NA 0.75). To minimize fluorophore spectral spill over, we used the Leica sequential laser excitation and detection modality. B cell follicles were defined based on the positioning of polyclonal B cells or by B220 staining. For 3D imaging acquisition, 10 xy stacks (1024 x 1024 pixel) sampled with 2 μm z spacing were acquired to provide image volumes that were 20 μm in depth.

Confocal imaging analysis

Antigen-specific CD4⁺ T cells were localized using Imaris built-in spot detection function and follicular T cell were identified based on polyclonal B cell positioning. T cell area centre was geometrically identified in Imaris 9.0.2 (Bitplane). Distances from T cell area centre were calculated and normalized for T cell area volumes. A T cell cluster was defined as a minimum of 3 T cells within a distance of 15 μm measured from the centroid of each cell, as previously reported⁵⁹. Spot-to-spot closest distance was calculated to detect ‘clustered’ spots (spot-to-spot closest distance ≤ 15um). Cell clusters of less than 3 cells were manually removed.

Intravital multiphoton microscopy

Mice were anesthetized with 5% isoflurane (Abbot) through a nose cone also delivering oxygen at 1L/min. Follow-up surgery and popliteal LN intravital imaging were carried out with lower concentrations of isoflurane (between 0.5% and 0.8% adjusted based on breathing rate). Surgical preparation of the popliteal LN was performed as described^28,60. Images were acquired with a LaVision BioTec TriMScope II coupled to an Olympus BX51 up-right microscope enclosed in a custom-built environmental chamber (Life Imaging Services) that was maintained at 37°C with heated air. Continuous body temperature monitoring through a rectal probe was performed to ensure that a narrow range of 37– 38°C was maintained at all times. Fluorescence excitation was provided by two tuneable fs-pulsed Ti:Sa lasers (680–1080 nm, Ultra II, Coherent) and one optical parameter oscillator (1000-1600 nm, APE Chameleon OPO, Coherent). The setup includes four photomultiplier tubes (3 Hamamatsu H7422-40 GaAsP High Sensitivity PMTs and 1 Hamamatsu H7422-50 GaAsP High Sensitivity red- extended PMT) with four emission filters (BS495, ET525/50, ET595/40, BS624) and a high working distance water-immersion 25x objective (NA=1.05, Olympus) or 20x objective (NA=1.0, Zeiss). For 4D analysis of cell migration, stacks of 10 square xy sections (512 x 512 pixel) sampled with 5 μm z spacing were acquired every 10 seconds for up to 1.5 hours to provide image volumes that were 50 μm in depth and with an xy field of view variable between 356 x 356 μm² and 409 x 409 μm². Sequences of image stacks were transformed into volume-rendered, 4D time-lapse movies with Imaris 9.0.2 (Bitplane). The track speed mean and the meandering index (the ratio of a migrating cell’s linear displacement to the total path length) were calculated by semi-automated cell tracking algorithm of Imaris.

Intravital cytometry

Intravital cytometry was performed by adapting a method for histocytometry⁶¹. Briefly, T cell and B cell surfaces were created by the surface object creation wizard in Imaris 9.0.2 (Bitplane). T and B cell surfaces were then merged into T-B surfaces using the merge function in Imaris. Channel statistics from T-B-merged surfaces were exported and merged into combined CSV files (Python) that were uploaded and analysed on FlowJo X⁶¹.

Niche-seq

Niche-seq was performed as previously described¹⁸. Briefly, Olympus BX51 upright microscope fitted with tuneable fs-pulsed Ti:Sa lasers was used to photoactivate and image niches of interest (680–1080 nm, Ultra II, Coherent). Images were acquired with a two- photon laser tuned to 940 nm. Regions within the outer paracortex of draining LNs (with Ag-specific CD4⁺ T cell clusters) were photoactivated at 820 nm¹⁹. Tile images were acquired as a 20-30 μm Z-stacks with 5 μm steps between each Z-plane. Images were acquired at 512 x 512 x-y resolution. The volume and depth along the Z axis of the photoactivated zones were consistent among experiments.

Immediately after photoactivation, lymph nodes were forced through a 70 μm mesh into ice cold FACS buffer (EDTA pH8.0 2mM, BSA 0,5% in PBS). Cell populations were sorted with BD FACSAria fusion (BD Biosciences). Because in individual photoactivated cells only a fraction of the inactive-PA-GFP is converted to active-PA-GFP, photoactivated PA-GFP cells express both the activated PA-GFP (detectable upon 488 nm excitation) as well as the inactive PA-GFP (detectable upon 405 nm excitation) by flow cytometry. After doublets exclusion, isolated cells were single-cell sorted into 384-well cell capture plates containing 2 μL of lysis solution and barcoded poly(T) reverse-transcription (RT) primers for single-cell RNA-seq. Four empty wells were kept in each 384-well plate as a no-cell control during data analysis. Immediately after sorting, each plate was spun down to ensure cell immersion into the lysis solution, snap frozen on dry ice, and stored at –80°C until processed.

Single-cell RNA-seq

Dendritic cells were sorted with BD FACSAria fusion (BD Biosciences). After doublets exclusion, cells positive for CD11c and MHC-II but negative for CD3 and B220 were single-cell sorted into 384-well cell capture plates containing 2 μL of lysis solution and barcoded poly(T) reverse-transcription (RT) primers for single-cell RNA-seq. Immediately after sorting, each plate was spun down to ensure cell immersion into the lysis solution, snap frozen on dry ice, and stored at –80°C until processed.

Massively Parallel Single-Cell RNA-seq library preparation (MARS-seq)

Single-cell libraries were prepared as previously described⁶². Briefly, mRNA from cell sorted into cell capture plates were barcoded and converted into cDNA and pooled using an automated pipeline. The pooled sample was then linearly amplified by T7 in vitro transcription, and the resulting RNA was fragmented and converted into a sequencing-ready library by tagging the samples with pool barcodes and Illumina sequences during ligation, RT, and PCR. Each pool of cells was tested for library quality and concentration was assessed as described⁶².

All RNA-Seq libraries (pooled at equimolar concentration) were sequenced using Illumina NextSeq 500 at a median sequencing depth of 38,323 reads per cell. Sequences were mapped to mouse genome (mm9), demultiplexed, and filtered as described⁶², extracting a set of unique molecular identifiers (UMI) that define distinct transcripts in single cells for further processing. We estimated the level of spurious UMIs in the data using statistics on empty MARS-seq wells and excluded all plates with estimated noise > 5%. Mapping of reads was done using HISAT (version 0.1.6); reads with multiple mapping positions were excluded. Reads were associated with genes if they were mapped to an exon, using the UCSC genome browser for reference. Exons of different genes that shared genomic position on the same strand were considered a single gene with a concatenated gene symbol. Cells with less than 500 UMIs were discarded from the analysis.

Clustering of infected lymph nodes

For clustering of scRNA-seq of infected lymph nodes we used the R package MetaCell³⁶. MetaCell was used to derive informative genes and compute cell-to-cell similarity, to compute K-nn graph covers and derive distribution of RNA in cohesive groups of cells (or meta-cells), and to derive strongly separated clusters using bootstrap analysis and computation of graph covers on resampled data. Default parameters were used unless otherwise stated. Cells from rVSV- and rLCMV-infected lymph nodes were clustered together. Meta-cells were annotated by pooled differential expression of marker genes, using the FP_gene,mc metric, which signifies for each gene and meta-cell the fold change between the geometric mean of this gene within the meta-cell and the median geometric mean across all meta-cells. Each gene was given a FP threshold and a priority index – such that colouring for Activated T cells by Gzma is favoured over colouring for general T cells by Trbc2. The selected genes, priority, and fold change threshold parameters are as follows:

group	gene	priority	fold change
T	Trbc2	1	2
T	Cd3e	1	2
CD8T	Cd8a	2	2.5
NK	Klrb1c	3	2
actT	Gzma	4	2
unknown	Gm23935	3	2
CD4T	Cd4	3	1.5
DC	Cxcl16	5	2
Mono	Fcer1g	2	3
B	Cd79b	3	0.6

Clustering of dendritic cells

For clustering of scRNA-seq of CD11c⁺ MHC-II^high DCs, we used the R package MetaCell³⁶, as previously described. Cells from rVSV- and rLCMV-infected lymph nodes, from all time points and genetic backgrounds were clustered together. Meta-cells were annotated by pooled differential expression of marker genes, as described before. The selected genes, priority, and fold change threshold parameters are as follows:

group	gene	priority	fold change
Migratory cDC2	Fscn1	1	5
Migratory cDC2	Il12b	1	2
cDC1	Naaa	1	3
cDC1	Cd24a	1	2.5
cDC2	Cd209a	2	5
cDC2	Cd209d	2	1.8
Mo-DC	Csf1r	4	3
Mo-DC	Tgfbi	4	2
Mo-DC	Fcer1g	4	2
Macrophages	C1qb	5	4

qPCR

Total RNA was isolated from frozen LNs with ReliaPrep RNA Miniprep system (Promega), following the manufacturer’s instructions. 1 μg of total RNA was reverse transcribed prior to qPCR analyses for Ifna2 (Mm00833961_S1, Thermo Fisher), Ifna4 (Mm00833969_S1, Thermo Fisher), Ifna5 (Mm00833976_S1, Thermo Fisher), Ifna6 (Mm01703458_S1, Thermo Fisher), Ifna7 (Mm02525960_S1, Thermo Fisher), Ifna9 (Mm00833983_S1, Thermo Fisher), Ifna12 (Mm00616656_S1, Thermo Fisher), Ifna13 (Mm01731013_S1, Thermo Fisher), Ifna14 (Mm01703465_S1, Thermo Fisher), Ifnb (Mm00439552_S1, Thermo Fisher), Isg15 (Mm 01705338_S1,hermo Fisher), 2’5’Oas (Mm00460961_m1, Thermo Fisher) and IL-6 (Mm00446190_M1, Thermo Fisher) in a QuantStudio 5 Real-Time PCR System (Thermo Fisher Scientific). All experiments were done in duplicate and normalized to the housekeeping gene GAPDH (Mm99999915_g1, Thermo Fisher).

Statistical analyses

Results are expressed as mean ± SEM. All statistical analyses were performed in Prism 5 (GraphPad Software). Means between two groups were compared with unpaired two-tailed t test. Means among three or more groups were compared with one-way or two-way ANOVA. Bonferroni post-test was used to correct for multiple comparisons, and in some experiments the Fisher’s Least Significant Difference (LSD) post-test was used when correction for multiple comparisons was not necessary.

Extended Data

Extended Data Fig. 1. VSV and LCMV infections result in distinct antiviral CD4⁺ T cell polarization and in vivo dynamic behavior, independently of viral strain, viral dose, infection route and TCR signal strength.

(a) Schematic representation of experimental procedure for the results described in Fig. 1a-c. 1 x 10⁶ purified Ag-specific (Tg7 when VSV-Ind was used, SMARTA cells in all other cases) CD45.1⁺ CD4⁺ T cells were injected into CD45.2⁺ WT recipients 1 day before intrafootpad infection. dLNs were collected at the indicated time points after infection and analyzed by flow cytometry. T_FH were defined as either Bcl-6⁺ CXCR5⁺ or CXCR5⁺ T-bet^- cells (in the latter case we always verified that cells were also Bcl-6⁺); T_H1 were defined as T-bet⁺ CXCR5^- cells. (b) Representative flow cytometry plots showing T_FH and T_H1 cells (out of CD44^high endogenous CD4⁺ T cells) in dLNs 7 days after footpad infection with the indicated virus. Numbers indicated the percentage of cells within the indicated gate. Results are representative of at least 2 independent experiments. (c) Quantification of T_FH (top) and T_H1 cells (bottom) – expressed as percentages of CD44^high endogenous CD4⁺ T cells – in dLNs of mice described in b. Results are representative of at least 2 independent experiments. Mean+/- SEM is shown. PBS n=2, all other conditions n=3. A one-way Anova test was applied; *** p value < 0.001. (d) Quantification of T_FH (left) and T_H1 (right) cells (expressed as percentages out of total transferred cells) in the spleens of CD45.2⁺ WT recipients injected with 1 x 10⁶ Ag-specific (Tg7 when VSV-Ind was used, SMARTA in all other cases) CD45.1⁺ T cells one day prior to intravenous infection with VSV-Ind (left), LCMV-Arm (right) or LCMV-Cl13 (not shown), respectively. Mean+/- SEM is shown. An unpaired two-tailed t test was applied. *** p value < 0.001. (e) Quantification of T_FH (left) and T_H1 (right) cells (expressed as percentages out of total transferred cells) in dLNs 5 days after infection of CD45.2⁺ WT recipients injected with 1 x 10⁶ Ag-specific (Tg7 for VSV-Ind, SMARTA for LCMV-WE) CD45.1⁺ T cells one day prior to intrafootpad infection with the indicated doses of VSV-Ind (black) or LCMV-WE (red). Results are pooled from 2 independent experiments. Mean+/- SEM is shown.n=6. A one-way Anova test was applied. *** p value < 0.001. (f-g) Quantification of T_FH (left) and T_H1 (right) cell absolute numbers (f) or T_FH/T_H1 absolute number ratios (g) at 0,3,5,7 and 14 days after VSV-Ind (black), rVSV (blue) or rLCMV (red) infection. Mean+/- SEM is shown. Day 0 n = 3 (VSV), 4 (rVSV and rLCMV); Day 3 n = 5 (VSV), 7 (rVSV and rLCMV); Day 5 n = 5 (VSV), 9 (rVSV), 10 (rLCMV); Day 7 n = 3; Day 14 n = 3 (VSV and rVSV), 6 (rLCMV). Black and blue stars indicate significance of respectively VSV and rVSV samples towards LCMV samples. A two-way Anova with LSD post-test was applied. * p value < 0.05; ** p value < 0.01; *** p value < 0.001; **** p value < 0.0001.

Extended Data Fig. 2. Spatiotemporal dynamics and activation of Ag-specific CD4⁺ T cells within dLNs upon VSV or LCMV infection.

(a, b) Track speed mean (a) and meandering index (b) of GFP⁺ Ag-specific (Tg7 when VSV-Ind was used, SMARTA in all other cases) CD4⁺ T cells in the mice described in Fig. 1d-g and Supplementary Movie 1, 3 days after PBS, VSV-Ind, rVSV or rLCMV injection. Data are pooled from 2 independent experiments. PBS, n=395; VSV-Ind n=11219; rVSV n=6692; rLCMV n=3537. One-way Anova test was applied. **** p value < 0.0001; (c, d) Mean fluorescent intensity of CD69 (c) and CD25 (d) within Ag-specific (SMARTA) CD4⁺ T cells in dLNs, 2 days after PBS, rVSV or rLCMV injection. Data are representative of 2 independent experiments. Mean+/- SEM is shown. n=2. One-way Anova test was applied. ** p value < 0.01; *** p value < 0.001 (e, f) Methods used to determine the normalized distance from T area centre (e) and percentages of clustered T cells / section (f). (e) T cell area volume was defined based on polyclonal B cell positioning and its centre was geometrically identified in Imaris. Ag-specific CD4⁺ T cells were localized using Imaris built-in spot detection function and distance from T cell area centre was calculated and normalized for T cell are volumes. (f) A T cell cluster was defined as a minimum of 3 T cells aggregating within closest distance of 15 μm measured from cell centroids (see Materials and Methods). Cell clusters of less than 3 cells were manually removed.

Extended Data Fig. 3. Confocal analysis of the CD4⁺ T cell priming niche.

Confocal imaging of murine dLNs collected 2 days after rVSV or rLCMV infection. Dashed lines represent the edges of B cell follicles and were depicted based on B220 staining. (a) Ag-specific GFP⁺ CD4⁺ T cells (SMARTA, depicted in purple) and Ag-specific CFP⁺ B cells (KL25, depicted in cyan) were adoptively transferred into WT mice. (b) Ag-specific CFP⁺ CD4⁺ T cells (SMARTA, depicted in purple) were adoptively transferred into CX3CR1-GFP x CCR2-RFP mice. A colocalization channel for GFP and RFP was used to depict inflammatory monocytes (cells positive for both CX3CR1 and CCR2, in green). (c) Ag-specific CFP⁺ CD4⁺ T cells (SMARTA, depicted in purple) and Ag-specific GFP⁺ CD8⁺ T cells (P14, depicted in green) were adoptively transferred into WT mice. (d) Ag-specific CFP⁺ CD4⁺ T cells (SMARTA, depicted in purple) were adoptively transferred into NKp46-ZsGreen mice. Scale bars represent 50 μm or 30 μm (zoom). The dotted square represents the zoomed area in the IFA where CD4⁺ T cell clusters are found. All images are representative of at least 2 independent experiments.

Extended Data Fig. 4. Early antiviral CD4⁺ T cell localization is independent of Ag-specific B cells, Ag-specific CD8⁺ T cells and CCR2+ monocytes.

Confocal imaging of dLNs of VI10YEN (a), Cor93 Tg TCR (b), and CCR2^-/- (c) mice collected either 3 (a) or 5 (b, c) days after rVSV (left) or rLCMV (right) infection. Ag-specific CD4⁺ T cells are depicted in green. Dashed lines represent the edges of B cell follicles and were depicted based on polyclonal B cell positioning (b, c) or B220 staining (a) (both in grey). Scale bars represent 50 μm. Results are representative of at least 2 independent experiments.

Extended Data Fig. 5. Antiviral CD4⁺ T cell are primed by cDC2 cells and differentiate independently of NK cells.

(a) Schematic representation of the experimental procedure for the results described in panels b and c. 1 x 10⁶ purified CD45.1⁺ Ag-specific (SMARTA) CD4⁺ T cells were injected into NKp46-DTR mice treated with PBS or DT as indicated. dLNs were collected 5 days after rVSV (blue) or rLCMV (red) infection. Percentages of NK cells (b), T_FH (c, left) and T_H1 (c, right) in dLNs were quantified by flow cytometry. Data are representative of 2 independent experiments. Mean+/- SEM is shown. n=3. A one-way Anova with Bonferroni’s post-test was applied. * p value < 0.05; **** p value < 0.0001 (d) Schematics of the experimental setup for the results described in panel e. 1 x 10⁶ purified CD45.1⁺ Ag-specific (SMARTA) CD4⁺ T cells were transferred to WT mice treated with anti-ICOS blocking antibody or with isotype control, as indicated, prior to rVSV (blue) or rLCMV (red) infection. dLNs were collected 3 days after infection. (e) ICOSL expression (mean fluorescent intensity) within CD11c⁺ MHC-II^high CD8⁺ (cDC1) and CD11c⁺ MHC-II^high CD11b⁺ (cDC2) cell subsets in dLNs of the mice described in d. Data are representative of 2 independent experiments. Mean+/- SEM is shown. PBS conditions n=2, all other conditions n=3. A one-way Anova with Bonferroni’s post-test was applied. ** p value < 0.01; **** p value < 0.0001. (f) 1 x 10⁶ purified CD45.1⁺ Ag-specific (SMARTA) CD4⁺ T cells were transferred to WT and DT-treated XCR1-DTR mice prior to rVSV (blue, left) or rLCMV (red, right) infection. Quantification of T_FH (left) and T_H1 (right) – expressed as percentages of the total transferred cells – in dLNs 5 days after infection is shown. Mean+/- SEM is shown. Data are representative of 2 independent experiments. n=4-6.

Extended Data Fig. 6. Measurement of IFN-α isoforms upon rVSV and rLCMV infection.

Analysis of Ifna2, Ifna5, Ifna6, Ifna7, Ifna9, Ifna12, Ifna13, Ifna14 gene expression in dLN at 0, 4, 8, 16, 24 and 48 hours after rVSV (blue) or rLCMV (red) infection by qPCR. Data are pooled from 2 independent experiments. Mean+/- SEM is shown. 0 hours n = 3; 4 hours n = 4; 8 hours n = 3 (rLCMV), 4 (rVSV); 16 hours n = 3 (rLCMV), 4 (rVSV); 24 hours n = 2 (rLCMV), 4 (rVSV); 48 hours n = 4. A two-way Anova with LSD post-test was applied. * p value < 0.05. The same sample was measured repeatedly for the 4 genes.

Extended Data Fig. 7. Early type I IFN signalling promotes germinal centre B cells and antiviral antibody responses.

a) Quantification of IgD^- CD95⁺ germinal centre (GC) B cells (left) – expressed as percentage of B220⁺ cells – and of IgG1⁺ cells (right) – expressed as percentage of B220⁺ B cells – in the dLNs of mice treated with anti-IFNAR blocking antibody (or isotype control), and infected with rVSV 14 days earlier. Mean+/- SEM is shown. n=3. An unpaired two-tailed t test was applied. *** p value < 0.001. (b) GP–binding IgG1 Abs (expressed as fold induction over uninfected controls) were measured in the sera of mice described in panel A, 14 days after rVSV infection. Data are pooled from 2 independent experiments. Mean+/- SEM is shown. n=7. An unpaired two-tailed t test was applied. * p value < 0.05. (c) Schematic representation of experimental procedure for the results described in Fig 3d-f. 1 x 10⁶ purified CD45.1⁺ Ag-specific (SMARTA) CD4⁺ T cells were transferred to CD45.2⁺ WT recipients and treated anti-IFNAR1 blocking antibody either 1 day prior to (light blue) or 1 day after (yellow) rVSV infection.

Extended Data Fig. 8. Expression of interferon stimulated genes within the cellular components of the CD4⁺ T cell priming niche.

Expression of the indicated interferon-stimulated genes (ISGs) within the cellular components of the photoactivated CD4⁺ T cell priming niches of the mice described in Fig. 2d-f. The colour bar on the bottom indicates each cell’s origin (blue: photoactivated cells from rVSV; red: photoactivated cells from rLCMV).

Extended Data Fig. 9. Blocking IL-6 impairs antiviral antibody responses.

(a) WT mice were treated with anti-IL-6 blocking antibody (or isotype control) and sera were collected 14 days after rVSV infection. GP–binding IgG2b Abs were measured in the sera and expressed as fold induction over uninfected controls. Data are representative of 2 independent experiments. Mean+/- SEM is shown. An unpaired two-tailed t test was applied. *** p value < 0.001. (b) Schematic representation of experimental procedure for the results described in Fig. 4e. 1 x 10⁶ purified CD45.1⁺ Ag-specific (SMARTA) CD4⁺ T cells were transferred to CD45.2⁺ WT recipients and treated with anti-IL-6 blocking antibody starting either 1 day prior to (yellow) or 1 day after rVSV infection (orange).

Extended Data Fig. 10. Expression of interferon stimulated genes in dendritic cells.

Expression of different ISGs across 2179 single QC positive CD11c⁺ MHC-II^high cells grouped in 5 clusters as described in the legend to Fig. 5A.

Supplementary Material

Supplementary Video 3

Multiphoton intravital microscopy in dLNs of WT mice injected with GFP⁺ Ag-specific CD4⁺ T cells (green) and deep red-labeled polyclonal B cells (gray) and treated with anti-IFNAR blocking antibody (or isotype control), 3 days after infection with rVSV. Dashed lines denote B cell follicles based on polyclonal B cell positioning. Scale bars represent 50 μm. Time in min:s. For four-dimensional analysis of cell migration, 10 z-stacks (spacing 5 μm) of 50 μm were acquired every 10s for 25-50 minutes. Data are representative of at least 3 independent experiments.

Supplementary Video 1

Multiphoton intravital microscopy in dLNs of WT mice injected with GFP⁺ Ag-specific CD4⁺ T cells (green) and deep red-labeled polyclonal B cells (gray) 3 days after intrafootpad infection with VSV-Ind, rVSV and rLCMV. Dashed lines denote B cell follicles based on polyclonal B cell positioning. Scale bars represent 50 μm. Time in min:s. For four-dimensional analysis of cell migration, 10 z-stacks (spacing 5 μm) of 50 μm were acquired every 10s for 25-50 minutes. Data are representative of at least 3 independent experiments.

Supplementary Video 2

Multiphoton intravital microscopy in dLNs of WT mice injected with GFP⁺ Ag-specific CD4⁺ T cells (green) and deep red-labeled polyclonal B cells (gray) 2 days after infection with rVSV and rLCMV. Dashed lines denote B cell follicles based on polyclonal B cell positioning. Scale bars represent 50 μm. Time in min:s. For four-dimensional analysis of cell migration, 10 z-stacks (spacing 5 μm) of 50 μm were acquired every 10s for 25-30 minutes. Data are representative of at least 3 independent experiments.