A minority of Th1 and Tfh effector cells express survival genes shared by memory cell progeny that require IL-7 or TCR signaling to persist

SUMMARY

It is not clear how CD4⁺ memory T cells are formed from a much larger pool of earlier effector cells. We found that transient systemic bacterial infection rapidly generates several antigen-specific T helper (Th)1 and T follicular helper (Tfh) cell populations with different tissue residence behaviors. Although most cells of all varieties had transcriptomes indicative of cell stress and death at the peak of the response, some had already acquired a memory cell signature characterized by expression of genes involved in cell survival. Each Th1 and Tfh cell type was maintained long term by interleukin (IL)-7, except germinal center Tfh cells, which depended on a T cell antigen receptor (TCR) signal. The results indicate that acute infection induces rapid differentiation of Th1 and Tfh cells, a minority of which quickly adopt the gene expression profile of memory cells and survive by signals from the IL-7 receptor or TCR.

In brief

Osum et al. show that bacterial infection expands CD4⁺ T cell subsets with diverse properties. Their results suggest that death-prone effector and survival-prone memory cells coexist in each subset at the peak of the response, after which the memory cells live and the effector cells perish.

Graphical abstract

INTRODUCTION

CD4⁺ memory T cells are critical for immune protection from intracellular pathogens.¹ Memory cells are formed during infection from a process that begins when naive CD4⁺ T cells with T cell receptors (TCRs) specific for major histocompatibility complex class II (MHCII)-bound microbial peptides proliferate and differentiate into effector cells that help clear the pathogen.² Although most of the effector T cells die by apoptosis during a contraction phase, some survive to become long-lived memory cells.^3,4 The leading paradigm for how memory cells are formed is based on CD8⁺ T cells and posits that the early pool of expanded T cells contains both interleukin (IL)-7 receptor (IL-7R)⁺ memory precursor effector cells (MPECs) that are destined to survive the contraction phase and transcriptionally distinct,⁵ IL-7R⁻ short-lived effector cells (SLECs) that are destined to die.⁶ This model, however, may not apply to CD4⁺ T cells because adoptively transferred IL-7R⁺ and IL-7R⁻ CD4⁺ T cells from the early pool produce memory cells equally well.⁷

Understanding CD4⁺ memory T cell formation has been complicated by the many CD4⁺ T cell subsets. Naive CD4⁺ T cells can differentiate into T helper (Th)1, Th2, Th9, Th17, Th22, peripheral regulatory T (Treg) cells, or T follicular helper (Tfh) cells,⁸ and each of these cell types may have memory cell versions.² CD4⁺ memory T cells can also exist as central (T_CM), effector (T_EM), or resident (T_RM) cells that recirculate through secondary lymphoid organs (SLOs), non-lymphoid organs, or reside permanently in tissues.^9,10 CD4⁺ memory T cells also have complicated survival requirements with IL-7 and IL-15 playing dominant roles.^11–13 It is unclear if all CD4⁺ T cell subsets are capable of forming memory cells, and, if so, whether CD4⁺ T cell memory formation conforms to the MPEC/SLEC paradigm of CD8⁺ T cells.

Here, we address these knowledge gaps using a type 1 immune response to acute bacterial infection in mice as a model system. We identified several Th1 and Tfh cell varieties that existed at the peak of expansion and after the contraction phase of the response. The post-contraction subsets had different trafficking patterns and dependence on IL-7, IL-15, or TCR but expressed a similar set of pro-survival genes. Although most cells in the population at the peak of expansion had transcriptomes indicative of cell stress and death, some had already acquired the pro-survival gene signature. Thus, acute bacterial infection induces the clonal expansion of Th1 and Tfh cell populations, which at their peak contain mostly cells with death-prone transcriptomes and a minority of cells with transcriptomes indistinguishable from memory cells.

RESULTS

The primary CD4⁺ T cell response has expansion, contraction, and memory phases

We used a transient bacterial infection that induces type 1 immunity to study CD4⁺ memory T cell formation. Mice were infected with an attenuated strain of Listeria monocytes (aLm)¹⁴ engineered to express an immunogenic peptide, 2W,¹⁵ which binds to the I-A^b MHCII molecule and is recognized by about 300 naive T cells per uninfected C57BL/6 (B6) mouse.¹⁶ We used a fluorophore-conjugated 2W:I-A^b tetramer-based cell-enrichment method and flow cytometry¹⁷ to identify when and where memory cells are formed. 2W:I-A^b tetramer-binding CD4⁺ T cells (Figure S1) underwent rapid proliferation after intravenous aLm-2W infection, peaking in the SLOs on days 5–7 at about 200,000 cells and then falling to 20,000 cells by day 15 (Figure 1A). The expanded population was dominated by CD44^hi Foxp3⁻ conventional T cells and contained very few Foxp3⁺ Treg cells¹⁸ (Figure S1) as observed in another type 1 immune response.¹⁹ 2W:I-A^b tetramer-binding CD4⁺ T cells also peaked in the liver and blood at 20,000 and 10,000 cells, respectively, on days 5–7, and then fell 10-fold by day 15 (Figure 1A). The number of cells in each location did not change significantly between days 15 and 30, as described in other studies.^7,20,21 These results suggest that the peak of effector T cell generation occurs after intravenous aLm-2W infection around day 6 and is followed by a contraction phase in all sampled body sites that ends around day 15, after which memory cells emerge. As noted below, the transcriptomes of aLm-induced CD4⁺ cells did not change between days 21 and 60. Therefore, day 21 aLm-induced CD4⁺ T cells are suitable for analysis of the memory state.

Figure 1. Naive 2W:I-A^b-specific T cells undergo expansion after intravenous infection with aLm-2W bacteria.

(A) Mean numbers (±SD) of 2W:I-A^b tetramer-binding cells in the indicated organs of B6 mice (n ≥ 3 at each time point from two independent experiments) at the indicated times after aLm-2W infection.

(B) Flow cytometry plots of staining of the indicated proteins on 2W:I-A^b tetramer⁺ cells from the SLOs of mice, 7 days after aLm-2W infection.

(C) Mean fluorescence intensities (±SD) of antibody staining of the indicated proteins on the indicated subsets identified as in (B). CD44^lo naive CD4⁺ T cells are shown for comparison.

(D) Numbers (top) and percentages (bottom) of 2W:I-A^b tetramer-binding cells of the indicated subsets from the SLOs of individual wild-type (filled circles) or Tbx21^−/− (open circles) mice (n ≥ 6 from two independent experiments) on day 7 after aLm-2W infection. Mean values are indicated with a horizontal bar. p values are from multiple unpaired t tests.

(E) Numbers (top) and percentages (bottom) of 2W:I-A^b tetramer-binding cells of the indicated subsets from the SLOs of individual Lck^WT/WT Bcl6^fl/fl (filled circles) or Lck^Cre/WT Bcl6^fl/fl (open circles) mice (n ≥ 5 from two independent experiments) on day 7 after aLm-2W infection. Mean values are indicated with a horizontal bar. p values are from multiple unpaired t tests.

(F) Histograms of CXCR5 staining on 2W:I-A^b tetramer-binding pre-Tfh cells (blue) or naive 2W:I-A^b tetramer⁻ CD4⁺ T cells (black) from mice of the indicated genotypes on day 7 after aLm-2W infection.

aLm-2W infection drives formation of many effector cell varieties

Previous research showed that aLm infection induces a mixture of Th1 and Tfh effector and memory cells.^7,20,22–26 We used a flow-cytometry strategy described by Marshall et al.⁷ based on Ly-6C, a member of the Ly-6 superfamily,²⁷ and P-selectin glycoprotein ligand-1 (PSGL1), a receptor for P-selectin,²⁸ and added the Tfh marker CXCR5,^20,22,23 the Th1-associated marker CXCR6,^25,29 and the T_CM marker CD62L.^9,10

Day 7 after aLm-2W infection was chosen as a time near the peak of expansion expected to be dominated by effector cells. As described for lymphocytic choriomeningitis virus (LCMV)-specific CD4⁺ T cells,^7,24,26 the 2W:I-A^b tetramer-binding conventional CD4⁺ T cell populations in aLm-2W-infected mice consisted of Ly-6C^hi PSGL1⁺, Ly-6C^lo PSGL1⁺, and Ly-6C⁻ PSGL1⁻ varieties (Figure 1B). The Ly-6C^hi PSGL1⁺ cells expressed CXCR6 and the Th1 master transcription factor T-bet³⁰ but lacked CD62L and CXCR5 (Figure 1C). These cells were Th1 cells because they were absent in aLm-2W-infected mice lacking Tbx21 encoding T-bet (Figure 1D). The Ly-6C⁻ PSGL1⁻ cells expressed the largest amounts of CXCR5 and PD-1, less T-bet than the Ly-6C^hi Th1 cells, and lacked CXCR6 and CD62L (Figure 1C). These cells were greatly reduced in aLm-2W-infected mice lacking Bcl6 in T cells (Figure 1E), which established their identify as Tfh cells. Expression of relatively large amounts of PD-1 marked these cells as germinal center Tfh (GC-Tfh) cells.³¹

The Ly-6C^lo PSGL1⁺ population contained CXCR6⁺ CD62L⁻, CXCR6⁻ CD62L⁺, and CXCR6⁻ CD62L⁻ subsets (Figure 1B). The CXCR6⁺ CD62L⁻ cells lacked CXCR5 and expressed the same amount of T-bet as the Ly-6C^hi Th1 cells (Figure 1C) but underwent a minimal and statistically insignificant reduction in T-bet-deficient mice (Figure 1D). As described below, however, these cells are transcriptionally very similar to the Ly-6C^hi Th1 cells and will therefore be referred to as Ly-6C^lo Th1 cells.

The 2 Ly-6C^lo PSGL1⁺ CXCR6⁻ populations expressed less CXCR5 and PD-1 than GC-Tfh cells and less T-bet than the Th1 cells (Figure 1C). The CXCR6⁻ CD62L⁻ cells were Tfh cells as shown by reduction in aLm-2W-infected mice with T cell-specific deletion of Bcl6 (Figure 1E). The absolute number of these cells but not their percentage in the population was slightly reduced in T-bet-deficient mice (Figure 1D). The lower level of PD-1 expression (Figure 1C) indicated that the CXCR6⁻ CD62L⁻ cells were likely the follicular mantle zone Tfh (M-Tfh) cells identified in imaging studies.³¹ Surprisingly, the CXCR6⁻ CD62L⁺ cells formed normally in the absence of Bcl-6 (Figure 1E) but lacked CXCR5 in this case (Figure 1F). These cells, which resemble a subset identified by Feng et al.,³² will be referred to as pre-Tfh cells based on cell-transfer experiments described below. Together, these results indicate that acute systemic bacterial infection induces Ly-6C^hi Th1, Ly-6C^lo Th1, pre-Tfh, M-Tfh, and GC-Tfh cells.

Memory cell populations are phenotypically like effector cell populations

We then assessed the 2W:I-A^b tetramer-binding memory cells that survived the contraction phase 21 days after aLm-2W infection. Ly-6C^hi PSGL1⁺, Ly-6C^lo PSGL1⁺, and Ly-6C⁻ PSGL1⁻ varieties were present and the Ly-6C^hi PSGL1⁺ cells expressed CXCR6 and T-bet but lacked CD62L and CXCR5 (Figure 2A and 2B). The Ly-6C⁻ PSGL1⁻ cells expressed CXCR5 and PD-1, less T-bet than the PSGL1⁺ Th1 cells, and lacked CXCR6 and CD62L. The Ly-6C^lo PSGL1⁺ population contained CXCR6⁺ CD62L⁻, CXCR6⁻ CD62L⁺, and CXCR6⁻ CD62L⁻ varieties and the CXCR6⁺ CD62L⁻ cells lacked CXCR5 and expressed the same amount of T-bet as the Ly-6C^hi Th1 cells. The CXCR6⁻ CD62L⁻ and CD62L⁺ populations expressed CXCR5 but less than at day 7, less T-bet than the Th1 cells, and lacked PD-1. Thus, the post-contraction 2W:I-A^b tetramer-binding memory T cell population contained varieties that resembled the Ly-6C^hi Th1, Ly-6C^lo Th1, pre-Tfh, M-Tfh, and GC-Tfh cells present on day 7. Indeed, the percentages of cells with these phenotypes were nearly identical at days 7 and 21, indicating that each subset underwent the same amount of contraction (Figure 2C).

Figure 2. Phenotype and location of 2W:I-A^b-specific memory T cells after intravenous infection with aLm-2W bacteria.

(A) Flow cytometry plots of staining of the indicated molecules on 2W:I-A^b tetramer⁺ cells from the SLOs of mice 21 days after aLm-2W infection.

(B) Mean fluorescence intensities (±SD) of antibody staining of the indicated proteins on the indicated subsets identified as in (A). CD44^lo naive CD4⁺ T cells are shown for comparison.

(C) Percentages of 2W:I-A^b tetramer-binding cells of the indicated subsets from the SLOs, blood, or livers of mice (n = 4 from two experiments) on day 7 (circles) or day 21 (squares) after aLm-2W infection. Mean values are indicated with a horizontal bar. The p value is from a multiple unpaired t test.

(D) Mean percentages (±SD, n = 9–10 from three independent experiments) of the indicated 2W:I-A^b tetramer-binding memory cell subsets that did not recirculate between the SLOs of the parabionts in a 21-day period (% resident). p values were derived from a one-way ANOVA with a Tukey’s multiple-comparisons test. One outlying value was removed from the Ly-6C^hi Th1 cell group based on the robust regression and outlier removal method with a Q value of 1%.

Cells with the phenotypes of Ly-6C^hi Th1, Ly-6C^lo Th1, pre-Tfh, and M-Tfh subsets were also present in the blood on days 7 and 21 after aLm-2W infection, whereas GC-Tfh cells were difficult to detect in this location. The only significant change in the blood was that Ly-6C^lo Th1 cells accounted for a smaller fraction of the day-21 than the day-7 population, consistent with progressive residency in tissues. 2W:I-A^b tetramer-binding cells also migrated to the liver, a tissue with Lm tropism,³³ and the population was dominated by Ly-6C^hi Th1 and Ly-6C^lo Th1 cells at both days 7 and 21. Thus, the phenotypes and localization of memory cell subsets were like the earlier effector cell subsets.

The low representation of Ly-6C^lo Th1 and GC-Tfh cells in the blood on day 21 suggested that these cells resided in tissues longer than the other memory cell types. We performed a parabiosis experiment to test this possibility. Mice with different congenic markers were infected with aLm-2W bacteria and surgically joined 60 days later such that their blood supplies were shared.³⁴ CD4⁺ T cells that bound 2W:I-A^b tetramer or an I-A^b tetramer containing a peptide (LLOp) from the listeriolysin O protein that is naturally expressed by Lm bacteria³⁵ were analyzed to increase the number of aLm-specific cells in the experiment. Tetramer-binding CD4⁺ memory T cells from each parabiont were assessed 21 days after joining to allow time for recirculating populations to reach equilibrium. The Ly-6C^hi Th1, M-Tfh, and pre-Tfh memory cell populations moved equally between the SLOs of the parabionts, whereas Ly-6C^lo Th1 and GC-Tfh cells were more likely to remain in the parabiont in which they were generated (Figure 2D). These results and those from the blood and liver (Figure 2C) indicate that M-Tfh and pre-Tfh cells are T_CM cells that recirculate through SLOs, whereas Ly-6C^hi Th1 cells are T_EM cells that recirculate through SLOs and liver. Ly-6C^lo Th1 cells are T_RM cells in the SLOs and may also reside in the liver. GC-Tfh cells are SLO T_RM cells, as described by others.^36,37

Th1 or Tfh subsets from the peak of the response tend to produce similar post-contraction memory cells

Cell-transfer experiments were then performed to assess the capacity of each effector cell subset to generate memory cells. 2W:I-A^b and LLOp:I-A^b tetramer-enriched Ly-6C^hi Th1, Ly-6C^lo Th1, pre-Tfh, M-Tfh, and GC-Tfh effector cells were sorted from the SLOs of day-7 aLm-2W-infected mice (Figure 3A). The purity of the post-sort Ly-6C^hi Th1, Ly-6C^lo Th1, M-Tfh, and GC-Tfh cell populations was greater than 95% and the pre-Tfh population was greater than 85% (Figure 3B). The sorted cells were injected intravenously into mice expressing a different congenic marker, which had been infected 7 days earlier with aLm-2W bacteria. The phenotypes of the transferred 2W:I-A^b tetramer-binding T cells were then determined 3 weeks later. Ly-6C^hi Th1 effector cells mainly produced Ly-6C^hi and Ly-6C^lo Th1 memory cells (Figure 3C, 3D) and Ly-6C^lo Th1 effector cells produced mostly Ly-6C^lo Th1 memory cells. M-Tfh effector cells mainly produced M-Tfh memory cells, whereas GC-Tfh effector cells mainly produced GC-Tfh and M-Tfh memory cells. It should be noted, however, that the sorting strategy did not produce baseline separation between M-Tfh and GC-Tfh cells (Figure 3A); thus, it is possible that some of the GC-Tfh cells generated from sorted M-Tfh effector cells came from contaminating GC-Tfh effector cells and vice versa. The pre-Tfh effector cell donor population, which consisted of about 90% pre-Tfh and 10% M-Tfh cells at the time of transfer (Figure 3B), generated a memory cell population of 25% pre-Tfh and 40% M-Tfh memory cells with smaller populations of Ly-6C^lo Th1 and GC-Tfh cells (Figures 3C and 3D). Thus, Ly-6C^hi Th1, Ly-6C^lo Th1, pre-Tfh, M-Tfh, and GC-Tfh effector cells each tended to produce memory cells with the matching phenotype. Each effector cell type also had some capacity to produce other memory cell types, although sort contamination could have contributed to this result.

Figure 3. CD4⁺ effector T cell subsets generate phenotypically similar memory T cell subsets.

(A) Flow cytometry plots of the indicated molecules on 2W:I-A^b tetramer⁺ cells from the SLOs of CD45.2⁺ mice 7 days after aLm-2W infection from a sample later subjected to fluorescence-activated cell sorting (FACS).

(B) Flow cytometry plots of the indicated molecules on 2W:I-A^b tetramer⁺ effector cells of the indicated subsets from SLOs of day 7 aLm-2W-infected CD45.2+ mice after FACS based on the gates shown in (A).

(C) Mean percentages (±SEM, n ≥ 3 from four independent experiments) of 2W:I-A^b tetramer-binding memory cell subsets of CD45.2⁺ donor origin in CD45.1⁺ mice that received the indicated FACS-sorted CD45.2⁺ effector cell populations 21 days earlier.

(D) Flow cytometry plots of the indicated molecules on 2W:I-A^b tetramer⁺ memory cells derived from the indicated FACS-sorted effector cell subsets 21 days after transfer into day 7 aLm-2W-infected recipients.

Transcriptionally distinct Th1 and Tfh effector and memory cells coexist at the peak of the response

We then performed an unbiased single-cell transcriptomics study to determine whether MPECs were present in each Th1 and Tfh variety. 2W:I-A^b and LLOp:I-A^b tetramer-binding cells were enriched and sorted from SLOs on days 6, 21, or 60 after aLm-2W infection and single cells were subjected to an RNA sequencing protocol. The Seurat R package was used to analyze the integrated sequencing data from the time points (11,128 cells from day 6, 5,234 from day 21, and 4,560 from day 60) and generate uniform manifold approximation and projection (UMAP) plots from each time containing cells clustered by transcriptional state.³⁸

The effector to memory cell transition was readily apparent when the FindClusters function in Seurat was applied to the data at a low resolution. Three cell clusters were identified (Figure 4A). The blue and much larger red cluster contained 96% of the cells from day 6 and very few cells from days 21 and 60, whereas the green cluster contained 93% and 98% of the cells from days 21 and 60% and 4% from day 6 (Figure 4B). The cells in the blue cluster expressed the cell cycle marker Mki67,³⁹ whereas most cells in the red and green clusters did not (Figure 4C). In contrast, most of the cells in the green cluster expressed the memory cell marker Il7r,⁴⁰ whereas most cells in the red and green clusters did not (Figure 4D). The abundance of blue and red cluster cells at the pre-contraction stage indicated that these were effector cells, while the abundance of the green cluster cells at the post-contraction stage indicated that these were memory cells. The cells in the blue cluster will be referred to as proliferating effector cells, the cells in the red cluster as non-proliferating effector cells, and the cells in the green cluster as memory cells.

Figure 4. Identification of CD4⁺ effector and memory T cells by single-cell RNA sequencing.

(A) Low-resolution dimension-reduction plots of single-cell RNA sequencing data from 2W:I-A^b and LLOp:I-A^b tetramer⁺ cells FACS sorted at the indicated times after aLm-2W infection.

(B) Number and percentages of cells in the three clusters identified in (A).

(C and D) Feature plots of Mki67 (C) or Il7r (D) expression.

(E) Volcano plot of genes over-expressed in non-proliferating effector cells (left half) or memory cells (right half) with a subset of genes labeled.

(F) Lists of the top 20 genes over-expressed in the non-proliferating effector cell cluster compared to the memory cell cluster or vice versa and with p values <10⁻³⁰⁰. The full list is in Table S1. The proliferating effector cell cluster was excluded to prevent highly expressed cell-cycle genes from skewing the analysis.

(G and H) Feature plots of expression of a module of Th1 (G) or Tfh (H) genes.

Many of the genes over-expressed by the non-proliferating effector cells were involved in apoptosis (S100a11, Gstt2, Pycard, Rilpl2, Bcl2a1b, Bcl2a1d, Morrbid, Anxa2, Ifi27L2a, Lsp1, Lgals1, Sh2d1a)^41–51 or cell stress (S100a4, S100a6, S100a10, S100a11, Hmgb2, E2f2)^52–54 (Figures 4E and 4F; Table S1), consistent with the possibility that these cells were SLECs destined to die. In contrast, many of the genes over-expressed by memory cells were involved in cell survival (Il7r, Prkce, Scml4),^55–58 DNA repair or genome stability (Parp8, Ssbp2),^59,60 protection from oxidative stress or apoptosis (Txnip, Lncpint),^61,62 or tumor suppression or quiescence (Zbtb20, Foxp1, Zfp36L2)^63–65 as expected for long-lived cells. The 3.6% of the cells in the day-6 population that were present in the memory cell cluster and thus expressed these genes were likely MPECs poised to survive the contraction phase.

Genes known to be over-expressed in Th1 or Tfh cells were then used to identify these subsets. A module of Th1 genes including Tbx21 (encoding T-bet), Ifngr1, Nkg7, and Ccl5 (Figure 4G) identified cells on the upper half of the UMAP at all time points, whereas a module of Tfh genes including Bcl6, Cxcr5, Il6st, Il6ra, and Btla identified cells on the lower half (Figure 4H). Notably, cells in the proliferating effector cell population expressed the Th1 module but not the Tfh module (Figure 4A, 4G, and 4H), consistent with the work of Ray et al.⁶⁶ showing that Tfh cells exit the cell cycle earlier than Th1 cells.

Th1 and Tfh cell identity is largely conserved from the effector to memory phases

The FindClusters function in Seurat was then applied to the data with a higher resolution to resolve Th1 and Tfh cell subsets. This setting generated 15 cell clusters with unique expression signatures at days 6, 21, and 60 (Figure 5A). Only a small number of Foxp3⁺ Treg cells were detected as expected from the flow cytometry results (Figure S1) and clustered with Foxp3⁻ cells in cluster 8 (Figure 5A). Cells in cluster 12 had a type I interferon-stimulated signature exemplified by expression of Isg15⁶⁷ (Figure 5B), did not change with time after aLm-2W infection, and were present among naive cells (K.C.O., unpublished data). Treg cells and cluster 12 cells were not analyzed further.

Figure 5. Identification of Th1 and Tfh effector and memory cell subsets by single-cell RNA sequencing.

(A) High-resolution dimension-reduction plot of single-cell RNA sequencing data from 2W:I-A^b and LLOp:I-A^b tetramer⁺ cells FACS sorted from the indicated times after aLm-2W infection with identity calls on each cluster listed in the legend.

(B) Violin plots of expression of the indicated genes by the indicated Th cell subsets.

(C) Flow cytometry plots of the indicated molecules on tetramer⁺ cells from aLm-2W-infected mice demonstrating the presence of CX₃CR1⁺ cells in the Ly-6C^hi Th1 cell subset.

(D) 2W:I-A^b plus LLOp:I-A^b tetramer-binding T cells sorted from the spleens of day-21 and -60 aLm-2W-infected mice were analyzed by single-cell RNA sequencing or by flow cytometry and memory cells subsets were identified as described in (A) and Figure 2A. The values in the figure indicate the percentage of each subset in the tetramer-binding population identified by each method.

The Th1 clusters were then assessed. Clusters 13 and 14 contained proliferating Th1 cells, whereas cluster 11 cells expressed Cx3cr1 typical of terminally differentiated Th1 cells (Figure 5A and 5B).⁶⁸ We confirmed by flow cytometry that cells expressing CX₃CR1 indeed accounted for about 10% of the Ly-6C^hi Th1 population detected by flow cytometry (Figure 5C). Cells in non-proliferating effector cell cluster 5 and memory cell cluster 1 expressed Ly6c2, Cxcr6, Selplg, Tbx21, and Ccl5 (Figure 5B) and thus corresponded to the Ly-6C^hi Th1 effector and memory cells identified by flow cytometry. Likewise, cells in non-proliferating effector cell cluster 7 and memory cell cluster 8 expressed Cxcr6, Selplg, Tbx21, and Ccl5 but less Ly6c2 than cells in clusters 5 and 1 and were likely the Ly-6C^lo Th1 effector and memory cells, respectively. The memory cells in cluster 8 expressed T_RM genes such as Itga1⁶⁹ and Rgs1,⁷⁰ while the memory cells in cluster 1 expressed Klf2 and S1pr1 involved in egress from lymphoid tissue.⁷¹ These results align with the parabiosis results indicating that Ly-6C^lo Th1 memory cells are tissue resident while Ly-6C^hi Th1 memory cells recirculate (Figure 2D).

Heterogeneity in the Tfh cell population was also assessed. Cells in non-proliferating effector cell cluster 4 and memory cell cluster 10 expressed the Tfh genes Bcl6, Cxcr5,⁷² and Il6st and the GC-Tfh genes Tox2⁷³ and Izumo1r⁷⁴ and lacked Klf2 and S1pr1⁷¹ (Figure 4A; 5B). Cells in cluster 10 expressed more Il7r and less Bcl6 and Cxcr5 than cells in cluster 4 (Figure 5B). The cells in cluster 4 were therefore likely the SLO-resident GC-Tfh effector cells detected on day 6 by flow cytometry (Figure 1), whereas those in cluster 10 were the GC-Tfh memory cells present on days 21 and 60 (Figure 2).

Cells in clusters 3 and 6 were candidates for pre-Tfh or M-Tfh effector cells and cells in clusters 0 and 2 for the memory cell versions. Cells in all four clusters did not express Ly6c2 or Cxcr6 (Figure 5B) and had lower expression of Cxcr5, Tox2, and Izumo1r and higher expression of Klf2 and S1pr1 than the GC-Tfh cells in clusters 4 and 10. Although Sell was poorly detected in general, clusters 0, 2, 3, and 6 were the only clusters that expressed it (K.C.O., unpublished data) and contained cells that expressed Ccr7, which is under the same transcriptional control as Sell.⁷¹ Therefore, clusters 3 and 6 likely contained a mixture of pre-Tfh and M-Tfh non-proliferating effector cells and clusters 0 and 2 a mixture of pre-Tfh and M-Tfh memory cells. This conclusion was bolstered by the fact that the sum of the frequencies of pre-Tfh and M-Tfh memory cells among the tetramer⁺ population identified by flow cytometry was about the sum of the frequencies of cells in clusters 0 and 2 (Figure 5D). The fact that pre-Tfh and M-Tfh memory cells did not resolve into separate clusters at this resolution is further evidence of the relatedness of these cell types.

Cells in cluster 9 had similar amounts of Cxcr5, Tox2, and Izumo1r to cluster 3 and 6 cells but low amounts of Klf2 and S1pr1 like GC-Tfh cells (Figure 5B). Cluster 9 cells were therefore categorized as transitional Tfh cells.

Notably, the aLm-induced CD4⁺ memory T cells did not preferentially express Tcf7, Foxo1, or Myb described to mark CD8⁺ memory T cells.^75–78 Foxo1 was uniformly expressed in all CD4⁺ effector and memory cell subsets and Tcf7 and Myb were expressed in all effector and memory cell subsets with highest expression in Tfh cells. Rather, each Th1 or Tfh effector and memory cell pair, for example cells in clusters 5 and 1, were segregated on the UMAP because the effector cells expressed the effector cell module typified by Rilpl2 and Gstt2, whereas the memory cells expressed the memory module typified by Il7r, Vps37b, and Sspb2. Importantly, cells from each of the Th1 and Tfh clusters were present in the day-6 memory cell population (Figures 4A and 5A).

CD4⁺ memory T cell subsets vary in their dependence on IL-7, IL-15, and TCR signaling for survival

The increase in Il7r expression in the memory cell populations was consistent with the model that IL-7R is required for CD4⁺ memory T cell survival.¹¹ Given, however, that this model was established by analyzing bulk CD4⁺ memory T cells, it was of interest to test if IL-7R is required for the survival of all memory subsets identified in this study. The role of IL-7R was assessed using Cd4^Cre-ERT2 Il7r^fl/fl mice in which administration of tamoxifen results in deletion of the Il7r gene encoding the IL-7R alpha chain. Cd4^WT/Cre-ERT2 Il7r^fl/fl mice or Cd4^WT/WT Il7r^fl/fl controls were infected with aLm-2W bacteria and then treated at a post-contraction time with tamoxifen for 1 week. One month later, the Ly-6C^hi Th1, Ly-6C^lo Th1, M-Tfh, and pre-Tfh memory cell populations cells were significantly smaller in Cd4^WT/Cre-ERT2 Il7r^fl/fl mice than in the Cd4^WT/WT Il7r^fl/fl controls (Figure 6A). The GC-Tfh cells were diminished but not to the point of statistical significance. The minimal effect of IL-7R loss on post-contraction GC-Tfh cells may have been related to these cells expressing less Il7r than other memory cells or the presence of IL-7R⁻ cluster 9 transitional Tfh cells in the population (Figures 5A and 5B).

Figure 6. Memory T cell subsets depend on IL-7 and/or IL-15 or CD4 for survival.

(A) Numbers of 2W:I-A^b tetramer-binding cells of the indicated memory subsets from the SLOs of individual Cd4^WT/WT Il7r^fl/fl (filled circles) or Cd4^Cre-ERT2/WT Il7r^fl/fl (open circles) mice (n ≥ 5 from four independent experiments) 50 days after aLm-2W infection and 30 days after tamoxifen treatment. Mean values are indicated with a horizontal bar. p values from a multiple unpaired t test.

(B) Numbers of 2W:I-A^b tetramer-binding cells of the indicated memory subsets from the SLOs of individual wild-type (filled circles) or Il15^−/− (open circles) mice (n ≥ 8 from four independent experiments) 30 days after aLm-2W infection. Mean values are indicated with a horizontal bar. p values from a multiple unpaired t test.

(C) Normalized numbers of 2W:I-A^b plus LLOp:I-A^b tetramer-binding cells of the indicated memory subsets from mice infected with aLm-2W bacteria and then 40 days later, injected twice intraperitoneally with YTS177 CD4 antibody or control immunoglobulin (Ig)G, 7 days apart and analyzed 7 days after the second antibody injection. Mean values are indicated with a horizontal bar. p values from a multiple unpaired t test.

(D) Fluorescence microscope images of spleens from uninfected mice or mice infected intravenously with aLm-2W bacteria 14 or 60 days before analysis. T cell areas were identified as dense collections of CD90.2⁺ cells (yellow). Follicles were identified as dense collections of IgD⁺ (purple) CD19⁺ (blue) cells and GCs as foci of CD19⁺ IgD⁻ GL7⁺ cells (bluish green) within follicles. GCs are indicated with arrows. A bar representing 300 μm is shown in the left panel.

(E) Mean percentages (±SD, n = 2 sections averaged from three mice per group) of follicles containing a GL7⁺ GC in the spleens of mice from the indicated groups. p values were derived from a one-way ANOVA with a Tukey’s multiple-comparisons test.

The lack of a significant effect of IL-7R ablation on post-contraction GC-Tfh cells warranted assessment of IL-15, another cytokine implicated in the survival of CD4⁺ memory T cells.¹² Mice lacking IL-15, which is sensed by a receptor consisting of the IL-15 receptor alpha chain, IL-2/IL-15 receptor beta chain, and the common gamma chain,⁷⁹ were used for this purpose. Ly-6C^lo Th1 memory cells were the only subset to be markedly reduced in aLm-2W-infected IL-15-deficient mice compared to controls (Figure 6B). Ly-6C^hi Th1 memory cells did not rely on IL-15 for survival despite expression of Il2rb, which encodes the IL-2/IL-15 receptor beta chain, at similar levels to Ly-6C^lo Th1 memory cells (Figure 5B).

The single-cell RNA sequencing suggested that post-contraction GC-Tfh cells might be maintained by TCR signaling. Although very few non-proliferating effector cells were present on days 21 and 60, most of the few that were present at these times were transitional Tfh cells in cluster 9 (Figure 5A). These cells expressed the TCR-induced gene, Egr3,⁸⁰ and lacked Il7r,¹³ indicative of ongoing TCR signaling. It was therefore possible that a source of aLm-derived peptide:MHCII complexes, perhaps GC B cells,²⁴ was present even on day 60 to cause TCR signaling in the GC-Tfh population. This possibility was addressed with the YTS177 CD4 monoclonal antibody, which blocks CD4-dependent TCR activation without depleting all CD4⁺ T cells.⁸¹ Mice were infected with aLm-2W bacteria and treated on day 30 with YTS177 or isotype-control antibody for 2 weeks. YTS177 antibody treatment resulted in a minor reduction of total CD4⁺ T cells. Normalizing the number of tetramer-binding cells to account for this effect showed a reduced frequency of GC-Tfh but no other memory cell subset in the 2W:I-A^b or LLOp:I-A^b tetramer-binding population (Figure 6C). Therefore, CD4-assisted TCR activation was required for long-term maintenance of GC-Tfh cells or their phenotype.

One potential explanation for this phenomena was that GCs were present even at day 60.⁷² As expected, about 30% of the follicles in the spleens of mice infected intravenously with aLm-2W 14 days earlier contained GCs marked by expression of GL7, whereas GCs were rare in the spleens of uninfected mice (Figure 6D and 6E). Remarkably, however, about 20% of the splenic follicles still contained GCs 60 days after aLm-2W infection. Thus, it is possible that post-contraction GC-Tfh cells were maintained by TCR stimulation in response to aLm-2W-derived peptide:MHCII complexes in GCs.

DISCUSSION

A goal of this research was to better understand the heterogeneity of CD4⁺ T cells generated by acute infection. We identified six phenotypically distinct subsets (CX₃CR1⁺ Th1, Ly-6C^hi Th1, Ly-6C^lo Th1, pre-Tfh, M-Tfh, and GC-Tfh cells) at the peak of the response to an attenuated bacterial infection and after contraction. Similar cell types have been identified in the antigen-specific CD4⁺ T cell populations generated by LCMV^{7,24,67,74,82–84} or Plasmodium,⁸⁵ suggesting that this level of specification is a common feature of type 1 immune responses.

Another goal was to improve understanding of CD4⁺ memory T cell formation. Although most cells in the 6 aLm-induced Th subsets expressed genes involved in cell stress and death at the peak of the response, about 4% expressed genes involved in cell quiescence and survival and shared by later memory cells. These results are consistent with the possibility that the cells with stress and death-prone transcriptomes in each subset on day 6 were SLECs that died during the contraction phase, while the cells with quiescence and survival-prone transcriptomes were MPECs that lived to become memory cells. The cell-transfer data indicated that most of the MPECs retained their cell identity as they transitioned to the memory cell state; for example, Ly-6C^hi Th1 MPECs became Ly-6C^hi Th1 memory cells. The fact, however, that the day-6 Ly-6C^hi Th1 cell population also yielded some Ly-6C^lo memory cells indicates that some plasticity is possible, as noted by Hu et al.⁸⁶ The day-6 pre-Tfh, M-Tfh, and GC-Tfh populations also each had some capacity to form the other types of Tfh memory cells, indicating some degree of plasticity in this subset as well.

Research on CD8⁺ T cells suggests that MPECs become memory cells because of early expression of the IL-7R, perhaps guided by transcription factors such as Tcf7, Foxo1, and Myb, and are maintained by a small subset of Tcf7 ⁺ stem-like cells.⁸⁷ As shown by Marshall et al.,⁷ however, IL-7R expression is not the determining factor for CD4⁺ memory T cell formation. In addition, our data show that aLm-induced CD4⁺ memory T cells do not preferentially express Tcf7, Foxo1, or Myb and do not contain a small subset of Tcf7 ⁺ stem-like cells. In addition, the fact that the CD4⁺ memory T cells present on day 21 or 60 had very similar transcriptomes suggests that these cells do not undergo the progressive differentiation described for CD8⁺ T cells.⁸⁸ CD4⁺ MPECs may derive from cells that exit the cell cycle due to early disengagement from antigen-presenting cells and then revert to the quiescent state typical of naive T cells. Indeed, naive T cells express many of the genes that we found to be over-expressed in CD4⁺ memory T cells.

This research also shed light on the functions and migration behaviors of the Th cells induced by acute systemic bacterial infection. The Ly-6C^hi Th1 cells circulated through SLOs and migrated to the liver. These cells therefore fit the classical definition of T_EM cells⁹ and likely function as macrophage activators that can be quickly recruited to infected tissues. We also identified an Ly-6C^lo population that expressed T-bet but formed in a relatively T-bet-independent manner and resembled cells described in several studies on the immune response to Toxoplasma gondii.^89,90 These cells were transcriptionally similar to the Ly-6C^hi Th1 cells but differed by lacking Klf2 and S1pr1, which are needed for egress from tissues,⁷¹ and by expressing Itga1⁶⁹ (encoding the alpha chain of CD49a) and Rgs1,⁷⁰ which are associated with tissue residence. These facts, together with the observations that Ly-6C^lo Th1 cells were progressively lost from the blood and migrated slowly into joined parabionts, suggest that Ly-6C^lo Th1 cells are T_RM cells in the SLOs and probably the liver. Our phenotyping and parabiosis data also indicate that pre-Tfh and M-Tfh cells are the major T_CM populations induced by aLM infection.

Although CD4⁺ effector T cells may not require IL-7R expression to become memory cells, our data show that Ly-6C^hi Th1, Ly-6C^lo Th1, pre-Tfh, and M-Tfh memory cells use it to survive. Ly-6C^lo Th1 T_RM cells also required IL-15 for optimal survival, perhaps because of a need for more STAT5 signals, which are transducedby both IL-7R and IL-15R,⁹¹ than other memory cell types. IL-15 is also required for the survival of some types of CD8⁺ T_RM cells,⁹² suggesting that T_RM cells may live in IL-15-rich niches. The post-contraction GC-Tfh cells that persisted on day 60²⁴ expressed memory cell genes, including Il7r, and lacked the cell stress/death genes associated with effector cells. These features indicate that post-contraction GC-Tfh cells are memory cells. On the other hand, the post-contraction GC-Tfh cells expressed the memory cell genes to a lesser extent than other memory cells, were less dependent on IL-7, and required TCR signals to survive or maintain their phenotype. TCR signaling in these cells may be driven by self-peptide:MHCII ligands of the tonic variety.^93,94 Given that GCs are still present 2 months after infection, however, it is possible that GC B cells specific for aLm antigens acquire this material from follicular dendritic cells, which can store injected proteins bound to antibodies or complement fragments for months^95–97 and present small numbers of aLm peptide:MHCII complexes to the GC-Tfh cells. The ensuing weak TCR signals may then promote the survival of the GC-Tfh cells without causing cell division, perhaps in concert with signals from the IL-7R. Although it is debatable whether T cells maintained by an antigen-dependent process can be called memory cells, such T cells would be poised to respond rapidly to a new infection and thus meet an important criterion of immune memory.

Limitations of the study

Although it is likely that the small population of cells with the memory cell transcriptome at the peak of expansion on day 6 survive the contraction phase to become memory cells, this point was not proved. This issue may be addressed in the future by production of transgenic mice expressing a fluorescent reporter controlled by the promoter elements of one of the memory cell genes discovered here. Such a mouse strain would allow isolation of the putative MPECs on day 6 and assessment of their survival capacity after adoptive transfer. Another limitation of this study is that it does not involve human CD4⁺ memory T cells. The murine transcriptomic data provided here, however, may help identify human CD4⁺ memory T cell varieties.

RESOURCE AVAILABILITY

Lead contact

Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, Marc K. Jenkins (jenki002@umn.edu).

Materials availability

Mouse strains used in this study can be purchased, produced by breeding commercially available strains, or obtained from the donating lab. Non-commercially available reagents used in this study will be made available by the lead contact.

Data and code availability

• Data: the single-cell RNA sequencing results are deposited in the Gene Expression Omnibus (GEO) repository with accession number GSE281547.

• Code: no new code was generated here.

STAR★METHODS

EXPERIMENTAL MODEL AND STUDY PARTICIPANT DETAILS

Mice of the strains listed in the key resources table and method details were housed in a specific pathogen-free facility at the University of Minnesota and experiments were conducted according to federal and institutional guidelines and with the approval of the University of Minnesota Institutional Animal Care and Use Committee. Six-12-week-old female mice were used for all experiments for the sake of consistency and to eliminate the possibility of adoptively transferred male cells being rejected by female recipients. This research may therefore not be generalizable to male mice. The aLm-2W bacterial strain was provided by Sing Sing Way.

KEY RESOURCES TABLE.

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
FC block	Stemcell	Cat#18731
L/D Ghost Dye 780 Viability Dye	TONBObiosciences	Cat#13-0865-T100
CXCR5PE	BioLegend	Cat#145504
CXCR6 BV421	BioLegend	Cat#151109
CX3CR1 BV605	BioLegend	Cat#149027
CD4 Buv395	BD Horizon	Cat#563790
B220 APCef780	eBioscience	Cat#47-0452-82
CD11c APCef780	Invitrogen	Cat#47-0114-82
CD11b APCef780	Invitrogen	Cat#47-0112-82
CD8a APCef780	Invitrogen	Cat#47-0081-82
CD44 BV510	BD Horizon	Cat#563114
Ly-6C BV711	BioLegend	Cat#128037
CD127 PEcy5	Invitrogen	Cat#15-1271-83
CD45-1 PE	BioLegend	Cat#110708
CD45-2 PE	BioLegend	Cat#109808
PSGL1 BUV737	BD Optibuild	Cat#741796
CD62L PE-AF610	Invitrogen	Cat#61-0621-82
PD-1 BV786	BioLegend	Cat#135225
CD45-1 FITC	BioLegend	Cat#110706
CD45-2 FITC	ebioscience	Cat#11-0454-85
FR4 PEcy7	BioLegend	Cat#125012
FOXP3 AF700	Invitrogen	Cat#56-5773-82
Tbet PEcy7	Invitrogen	Cat#25-5825-82
GL7 FITC	BioLegend	Cat#144603
IgD APC	BioLegend	Cat#405714
CD90.2 PE	Invitrogen	Cat#12-0902-83
CD19 BV421	BD Horizon	Cat#562701
PSGL1 BV605	BD Optibuild	Cat#740384
Ly-6C FITC	BD Pharmigen	Cat#553104
CD62L PE	Invitrogen	Cat#12-0621-82
CD4 BV786	BD Optibuild	Cat#740844
CD4	Bio-X-Cell	RRID AB_1107642
Bacterial and virus strains
aLm-2W, attenuated Listeria monocytogenes bacteria	-	Erteltetal.14
Chemicals, peptides, and recombinant proteins
Treg-Protector (anti-ARTC2 Nanobody)	Biolegend	Cat#149801
Biotin-conjugated 2W:I-Ab molecules	-	Moon et al.16
Biotin-conjugated LLOp:I-Ab molecules	-	Pepper et al.20
Allophycocyanin-conjugated streptavidin	Agilent Technologies	Cat#PJ25S-1
Critical commercial assays
FOXP3/Transcription Factor Staining Buffer kit	TONBObiosciences	TNB0607KIT
Deposited data
D6 D21 and D60 scRNA-seq processed and raw data files	This study	Gene Expression Omnibus (GEO) Accession code: GSE281547
Experimental models: Organisms/strains
C57BL/6J mice	Jackson Lab	RRID:IMSR_JAX:000664
B6.SJL-Ptprca Pep3b/BoyJ mice	Jackson Lab	RRID:IMSR_JAX:002014
B6.129S6-Tbx21tm1Glm/J mice	Jackson Lab	RRID:IMSR_JAX:004648
B6(Cg)-Il15tm1.2Nsl/J mice	Jackson Lab	RRID:IMSR_JAX:034239
B6.129S(FVB)-Bcl6tm1.1Dent/J mice	Jackson Lab	RRID:IMSR_JAX:023727
B6.Cg-Tg(Lck-icre)3779Nik/J mice	Jackson Lab	RRID:IMSR_JAX:012837
B6(129X1)-Tg(Cd4-cre/ERT2)11Gnri/J mice	Jackson Lab	RRID:IMSR_JAX:022356
Il7rfl/fl B6 mice	Alfred Singer	McCaughtry et al.98
Software and algorithms
Cell Ranger Aggr v7.2.0	10X Genomics	-
Seurat V5.0.0	-	Satija et al.99
Other
IMDM (1x) + GlutaMAX-I Iscoves Modified Dulbecco’s Medium	Gibco	Cat#31980-030
Chromium Next GEM Single Cell 3’ Kit v3.1, 4 rxns	10x Genomics	Cat#PN-1000269
Anti-APC microbeads	Miltenyi	Cat#130-090-855
EasySep APC selection cocktail	Stemcell	Cat#18451C
EasySep Dextran RapidSpheres	Stemcell	Cat#50100

METHOD DETAILS

Mice

Female C57BL/6J (B6), B6.SJL-Ptprca Pep3b/BoyJ (CD45.1⁺), B6.129S6-Tbx21tm1Glm/J (Tbx21 ^−/−),¹⁰⁰ B6(Cg)-Il15tm1.2Nsl/J (Il15 ^−/−),¹⁰¹ (B6.129S(FVB)-Bcl6tm1.1Dent/J (Bcl6 ^fl/fl),¹⁰² B6.Cg-Tg(Lck-icre)3779Nik/J (Lck ^Cre/Cre),¹⁰³ and B6(129X1)-Tg(Cd4-cre/ERT2)11Gnri/J (Cd4 ^Cre-ERT2)¹⁰⁴ mice were purchased from The Jackson Laboratory. Bcl6^fl/fl mice were bred with Lck ^Cre/Cre mice to generate Lck ^Cre/WT Bcl6 ^fl/fl and Lck ^WT//WT Bcl6 ^fl/fl mice. Il7r ^fl/fl mice⁹⁸ were obtained from Alfred Singer and crossed to Cd4 ^Cre-ERT2/WT mice to produce Cd4 ^Cre-ERT2/WT Il7r ^fl/fl and Cd4 ^WT/WT Il7r ^fl/fl mice.

Parabiotic surgeries

Parabiotic surgery was performed as described.¹⁰⁵ Briefly, mice to be joined were anesthetized and their sides to be joined were shaved from about 1 cm below the shoulder and about 1 cm above the hip. Matching skin incisions were made in the shaved areas and closed with 3–0 Prolene stitches and overlying surgical wound clips. Parabionts were then allowed to rest for 21 days before experiments. Percent residence was calculated as follows: percent resident = (1-(number of cells of a Th subset from the joined mouse × 2)/(number of cells of that subset from the joined mouse + number of cells of that subset from the parent mouse) × 100)). The number of cells of a Th subset from the joined mouse and the parent mouse for a joined pair were normalized to the number of 2W:I-A^b tetramer⁻ naive CD4⁺ T cells from the joined mouse and the parent mouse with the assumption that naive CD4⁺ T cells should distribute equally to each member of the pair.

Single-cell RNA sequencing

2W:I-A^b and LLOp:I-A^b tetramer-binding T cells were purified by FACS sorting from the SLOs of 3–5 mice per group infected intravenously 6, 21, or 60 days earlier with 10⁷ aLm-2W bacteria. Cells from each sample were loaded into individual ports of a 10X Genomics chip at 1,500 cells/μl. Libraries were prepared for each sample with Next GEM single cell 3′ gene expression kits and sequenced on one 25B lane using a Novaseq X Plus with V1.0 sequencing kits (28 × 150 nt reads) at the University of Illinois. Features were enumerated using Cellranger (Cell Ranger Aggr v7.2.0, 10X Genomics) and gene expression values were analyzed using Seurat V5.0.0.⁹⁹ Single cells from each sample were identified based on barcodes and normalized within each pool using the sctransform method in Seurat. Different clusters were visualized using UMAP dimensional reduction.³⁸

Cell transfer

For the experiments shown in Figure 3, 2W:I-A^b and LLOp:I-A^b tetramer-binding cells were enriched from the spleen and lymph nodes of CD45.2⁺ mice 7 days after aLm-2W infection using Miltenyi positive selection kits according to the manufacturer’s instructions. The cells were stained with phenotyping antibodies, purified by sorting on a FACSAria-II (BD) flow cytometer based on gates shown in Figure 3A, and injected intravenously into separate mice expressing CD45.1 and had been infected 7 days earlier with aLm-2W bacteria. 2W:I-A^b tetramer-binding cells were enriched from the spleen and lymph nodes of the recipient mice 21 days later, stained with phenotyping antibodies, and analyzed for the various memory cell subsets as in Figure 2A.

Infections and injections

The ActA-deficient aLm-2W bacterial strain was described previously.¹⁴ Mice were injected intravenously with 10⁷ bacteria. For the experiments in Figure 6D, mice were infected with aLm-2W bacteria and then 40 days later, twice intraperitoneally with 1 mg of YTS177 CD4 antibody (Bio-X-Cell) or control IgG, 7 days apart. Subsets of 2W:I-A^b and LLOp:I-A^b tetramer-binding memory cells were analyzed 7 days after the second antibody injection. YTS177 treatment caused a modest 1.5-fold reduction in the total number of all CD4⁺ T cells. This effect was corrected by multiplying the number of tetramer-binding cells in the YTS177-treated animals by 1.5.

Tetramers

Biotin-labeled I-A^b molecules containing the 2W (EAWGALANWAVDSA) or LLOp (NEKYAQAYPNVS) peptides covalently attached to the I-A^b beta chain were produced with I-A^b alpha chains in Drosophila melanogaster S2 cells,^16,20,106 then purified and tetramerized with streptavidin allophycocyanin (APC) as described previously.¹⁶

Cell enrichment and flow cytometry

Livers were perfused with Dulbecco’s PBS via the portal vein until they changed color. Intrahepatic lymphocytes were isolated by gentleMACS (Miltenyi) disruption using the liver 01.01 program and incubated for 30 minutes at 37°C with shaking in 0.4 mg/mL collagenase IV and 25 μg/mL DNAse I. Single cell suspensions were spun at 300 rpm for 3 minutes to remove hepatocytes. Lymphocytes were purified on a 21% Histodenz gradient after centrifugation at 2,200 rpm slow acceleration for 20 minutes without braking and at 27°C. Single cell suspensions of pooled spleens and lymph nodes or liver preparations were stained for 1 hour at room temperature with APC-conjugated peptide:I-A^b tetramers¹⁷ and BV650- or PE-conjugated CXCR5 antibodies and BV421- or BV711-conjugated CXCR6 antibodies. Cells were enriched using the EasySep Mouse APC Positive Selection (STEMCELL Technologies) or Miltenyi Biotec systems according to the manufacturer’s instructions. Enriched samples were stained for 20 minutes at 4°C with GhostDye Red 780 and fluorophore-conjugated antibodies specific for: CD4, B220, CD11b, CD11c, F4/80, CD44, CXCR5, PD-1, PSGL1, CD62L, CXCR6, and Ly-6C and in some cases CD45.2 and CD45.1. Stained cells were fixed and permeabilized with the FOXP3/Transcription Factor Staining Buffer Set (TONBO) according to the manufacturer’s instructions and stained overnight at 4°C with fluorophore-conjugated Foxp3 and T-bet antibodies. Cells were counted and analyzed by flow cytometry with counting beads on a Fortessa (BD) flow cytometer. Data were analyzed using FlowJo software.

Immunofluorescence

Spleens were embedded in OCT and frozen in a 2-methyl butane liquid bath. Seven-micron sections were cut on a Leica cryostat and stained with BV421-conjugated CD19 (BD, 1D3), APC-conjugated IgD (BioLegend, 11–26c.2a), PE-conjugated CD90.2 (eBioscience, 53–2.1), or FITC-conjugated GL7 (BioLegend, GL7) antibodies at 4°C overnight. Whole spleen images were taken with a Leica Thunder microscope and a 20×0.5 NA objective. CD19⁺ IgD⁺ follicles and CD19⁺ IgD⁻ GL7⁺ GCs were quantified from these images using Imaris 8 software. Higher resolution images shown in Figure 6D were acquired with a Leica Stellaris 8 confocal microscope using a 20×0.75NA glycerol immersion objective. A 300 μm scale bar was added for reference.

QUANTIFICATION AND STATISTICAL ANALYSIS

Statistical tests were performed using Prism software. The exact values of n (number of animals) and dispersion and precision measures (mean, SD, or SEM) are indicated in the figure legends. p values, derived from multiple unpaired t tests or one-way ANOVA with a Tukey’s multiple comparisons test, are indicated on the figures.

Highlights.

• Six Th1 and Tfh varieties are induced during a primary response to acute bacterial infection

• All varieties contain cells with death- or survival-prone transcriptomes at the peak

• Cells with survival-prone transcriptomes persist past peak of infection

• Antigen-experienced CD4⁺ T cells use IL-7 or TCR signals to survive