GL7 ligand expression defines a novel subset of CD4⁺ T_RM cells in lungs recovered from pneumococcus

Abstract

Streptococcus pneumoniae is the most common etiology of bacterial pneumonia, one of the leading causes of death in children and the elderly worldwide. During non-lethal infections with S. pneumoniae, lymphocytes accumulate in the lungs and protect against reinfection with serotype-mismatched strains. Cluster of differentiation CD4⁺ resident memory T (T_RM) cells are known to be crucial for this protection, but the diversity of lung CD4⁺ T_RM cells has yet to be fully delineated. We aimed to identify unique subsets and their contributions to lung immunity. After recovery from pneumococcal infections, we identified a distinct subset of CD4⁺ T cells defined by the phenotype CD11a^hiCD69⁺GL7⁺ in mouse lungs. Phenotypic analyses for markers of lymphocyte memory and residence demonstrated that GL7⁺ T cells are a subset of CD4⁺ T_RM cells. Functional studies revealed that unlike GL7⁻ T_RM subsets that were mostly (RAR-related Orphan Receptor gamma T) RORγT⁺, GL7⁺ T_RM cells exhibited higher levels of (T-box expressed in T cells) T-bet and Gata-3, corresponding with increased synthesis of interferon-γ, interleukin-13, and interleukin-5, inherent to both T helper 1 (T_H1) and T_H2 functions. Thus, we propose that these cells provide novel contributions during pneumococcal pneumonia, serving as important determinants of lung immunity.

INTRODUCTION

Lower respiratory infections impart one of the highest global burdens of disease worldwide and are a leading cause of death, especially in children and the elderly^1–3. The most common bacterial source of pneumonia is Streptococcus pneumoniae (Spn). Over 100 different serotypes have been identified based on differences in the Spn capsule⁴. While effective serotype-specific vaccines have reduced mortality from invasive pneumococcal disease worldwide^5,6, infections with non-vaccine serotypes continue to increase^7,8. Our group and others have demonstrated that self-limiting experimental pneumonia or immunization leads to an accumulation of resident memory lymphocytes that provide protection across multiple Spn serotypes, termed heterotypic immunity^9–11. Leveraging our knowledge about these resident lymphocytes will be crucial in developing strain-independent Spn vaccines.

Resident memory T cells (T_RM) localize to non-lymphoid tissues and deliver local and fast effector functions at the time of reinfection^9–13. Lung T_RM cells have been shown to provide crucial long-term protection against many bacterial and viral agents^10,12–15. T_RM cells deliver essential cues to their environment and cooperate with epithelial cells, B cells, and other T cells to choreograph important functions needed for immune protection, such as lymphocyte and neutrophil recruitment^16–18. T_RM cells have been characterized by expression of the surface markers cluster of differentiation CD11a and CD69 at rest^10,12, although heterogeneity within this population is gaining recognition^18,19. Previous work by our group demonstrates the presence of diverse and novel subsets of CD4⁺ and CD8⁺ T_RM and B resident memory cells in mouse lungs after pneumococcal exposures^20,21, but the extent of lung T_RM cell diversity and their roles in immune protection are still unclear.

In this work, we used full-spectrum flow cytometry (FSFC) to study CD4⁺ T cells in the lungs of Spn-experienced mice. We have identified previously undescribed subsets of CD4⁺ T_RM cells that express the GL7 antibody ligand, known as a marker of germinal center B cells. GL7 is a monoclonal antibody that recognizes surface glycans sialylated with N-Acetylneuraminic Acid (Neu5Ac)²². Throughout this study, we will denote cells expressing the ligand of GL7 (Neu5Ac) as GL7⁺. Our analyses revealed that this novel subset of GL7⁺ T cells represents a non-recirculating resident population of CD4⁺ T_RM cells with a unique transcription factor profile that secretes a distinct combination of cytokines. Our study introduces GL7⁺ T_RM cells as a discrete population of lung-resident lymphocytes arising after recovery from pneumococcal pneumonia.

RESULTS

Recovery from pneumonia leads to subpopulations of CD4⁺ T cells arising in the lungs

CD4⁺ T cells mediate heterotypic immune protection against pneumococcal pneumonia^10,16. Comprehensive phenotyping of Spn-experienced lungs revealed a previously unappreciated heterogeneity of resident lymphocytes²⁰. Following this approach, we combined FSFC using a 22-color immunopheno-typing panel²⁰ to further investigate lung CD4⁺ T_RM cells in Spn-experienced lungs. Adult mice received two intratracheal (i.t.) instillations of Spn serotype 19F (Sp19F) directed to the left lobes (LLs) (Fig. 1A). The two infections result in a self-limiting pneumonia that promotes heterotypic immune remodeling¹⁰. After a resting period of 30 days, we collected the lungs and processed them for further studies. We used a panel of 21 surface markers to characterize and distinguish lymphocyte subsets²⁰, concatenated data from six experienced mice, subsetted live extravascular (i.v.CD45.2⁻) CD4⁺ T cells (Supplementary Fig. 1), and projected the multiparameter data into 2-dimensional space using optimized parameter T-distributed stochastic neighborhood embedding (opt-SNE)²³ (Fig. 1B). To identify the cells subsets on the resulting map, we overlaid heatmaps of normalized expression of each marker over the opt-SNE visualization (Fig. 1C, Supplementary Fig. 1B). We observed that three markers, CD11a, CD69, and GL7, could separate CD4⁺ T cells into distinct populations. Following heatmap expression, we designed a supervised gating strategy to categorize lung CD4⁺ T cells which confirmed four major subsets: CD11a^hiCD69⁺GL7⁻, CD11a^hiCD69⁺GL7⁺, CD11a^hiCD69⁻GL7⁻ and CD11a^lo (Figs. 1D and 1E, Supplementary Fig. 2A), referred to henceforth as CD69⁺GL7⁻, CD69⁺GL7⁺, CD69⁻GL7⁻ and CD11a^lo respectively. All but the CD11a^lo subsets were CD44^hi and CD62L^lo, characteristic of memory T cells, while the CD11a^lo cells were CD44^loCD62L^hi, characteristic of naive T cells^24,25 (Figs. 1F and 1G). These phenotypes suggest that both CD69⁺ populations were T_RM cells^12–14, with expression of GL7 ligand differentiating a distinct subset (Figs. 1F and 1G). The GL7-reactive CD4⁺ cells were observed in loose aggregates within the interstitial connective tissue of bronchovascular bundles after recovery from pneumococcal pneumonia, in association with other CD4⁺ cells that were not GL7-reactive (Fig. 1H).

Fig. 1.

Lungs from pneumococcus-experienced mice contain multiple subsets of memory CD4⁺ T cells. (A) Model of heterotypic immunity. Adult female mice are infected intratracheally into the left lobe (LL) of the lungs with Streptococcus pneumoniae serotype 19F (Sp19F) on day 0 and 7 then rested until day 35, referred to as experienced. Three minutes before LL collection, mice were injected Intravenously. with mouse anti-CD45.2 antibody then processed for flow cytometry. (B) Optimized parameter T-distributed stochastic neighborhood embedding (opt-SNE) visualization depicting live lung (ivCD45.2⁻) CD45⁺CD3⁺CD4⁺ T cells of concatenated data from n = 6 mice gathered over two independent experiments. (C) Heat map depiction of expression level of surface markers overlayed on opt-SNE map. (D) Representative manual gating of flow cytometry data of lung CD4⁺ T cells, n = 6 mice, two independent experiments. (E) Manual gates overlayed on opt-SNE map.(F) Representative histograms of CD62L and CD44 in lung CD4⁺ T cells, n = 6, 2 independent experiments. (G) Heatmap of average mean fluorescence intensity (MFI) normalized to fluorescence minus one (FMO) controls or minimal MFI and maximal MFI, n = 6, two independent experiments. Each group was compared to CD11a⁻ for statistical comparisons, one-way ANOVA *p < 0.002. (H) Representative immunofluorescent micrograph showing anatomical distribution of CD4⁺ (yellow) and GL7⁺⁶ cells in experienced mouse lungs at 35 dpi, with white arrows indicating some double-positive cells. (I) Fold change of Cmah mRNA in GL7⁺ over GL7⁻ sorted, unstimulated, lung CD4⁺CD11a⁺CD69⁺ T cells, n = 3, 6 LL pooled per sample, Welch’s T-test, *p value = 0.0210.

GL7 detects surface glycans sialylated with Neu5Ac^22,26. In resting lymphocytes, the enzyme cytidine monophospho-Neu5Ac hydroxylase (CMAH) catalyzes the hydroxylation of Neu5Ac into N-glycolylneuraminic acid (Neu5Gc)²⁶. During lymphocyte activation, Cmah is downregulated, resulting in glycan sialylation with Neu5Ac^22,26–29. GL7 ligand expression occurs in germinal center B cells (B_GC) and GC T follicular helper (T_FH) cells, but such phenotype has not been reported in memory T cells^28,30. To examine if CD69⁺GL7⁺ cells have a reduced level of CMAH, we measured Cmah transcripts in CD69⁺GL7⁺ and CD69⁺GL7⁻ cells sorted from experienced mouse lungs. CD69⁺-GL7⁺ cells demonstrated a reduced level of Cmah transcripts, suggesting the expression of GL7 ligand associates with a reduced level of the enzyme and an additional phenotypic difference between T_RM cell subsets (Fig. 1I, Supplementary Fig. 2B). Together, these results suggest CD4⁺ T_RM cells in pneumococcal experienced mouse lungs can be separated into subpopulations based on their differential expression of surface markers including the GL7 ligand Neu5Ac.

The GL7⁺ T cells in recovered lungs are not circulating, proliferating, follicular, or restricted to pneumococcal infections

In agreement with previous reports by our group and others^10,11,16,21, we observe that all four subsets of lung T cells, CD11a^hiCD69⁺GL7⁻, CD11a^hiCD69⁺GL7⁺, CD11a^hiCD69⁻GL7⁻ and CD11a^lo, increased in number after pneumococcal experience (Fig. 2A). The largest fraction of CD4⁺ T cells in naïve lungs corresponded to CD11a^lo cells; in experienced lungs, CD11a^hi subsets predominated, including the more resident-like CD69⁺-GL7⁻ (Fig. 2A), supporting immune remodeling after infection. To further understand the dynamics of these subsets after Spn infections, we collected the left lung lobes at eight different timepoints from 0 to 35 days after exposure (Fig. 2B). The second Sp19F infection, designed to facilitate local antigen presentation in the lung, triggered acute accumulation of T cells (Supplementary Fig. 3A). The numbers of cells from all four subsets reached a maximum at 3 days after the second infection (10 dpi) (Fig. 2C). From 24 to 35 dpi both CD69⁺GL7⁺ and CD69⁺GL7⁻ subsets plateau and their cell numbers stabilized, while CD69⁻ and CD11a^lo cells continuously decreased over these timepoints (Fig. 2C). The proportion of CD4⁺ T cells that are CD69⁺GL7⁺ increased in the lungs after the second infection, and it became progressively greater over successive timepoints (Fig. 2D, Supplementary Fig. 3A).

Fig. 2.

CD69⁺GL7⁺ subsets of CD4⁺ T cells have distinct dynamic patterns of accumulation in mouse lungs. (A) Numbers of cells (above) and percentage of lung CD4⁺ T cells in LL (below) of each subset from saline or Spn experienced mice at 35dpi, n = 6, two independent experiments. Bar graphs of mean and error bars of standard error of mean⁵, 2-way ANOVA with Bonferroni multiple comparisons test, ***p < 0.0003, ****p < 0.0001. (B) Diagram of timeline of lung collection. Sp19F intratracheally. on days 0 and 7. (C) Average number of cells in LL at every collection timepoint, n = 6 per timepoint, 2 independent experiments, error bars are SEM. (D) Average percent of CD4⁺ T cells in LL from CD69⁺GL7⁺ subset, n = 6 per timepoint, two independent experiments, error bars are SEM. One-way ANOVA with Bonferroni multiple comparisons test, ***p = 0.0004 ****p < 0.0001. FTY720 or vehicle Intraperitonealy to experienced mice at 1mg/kg every other day for 12 days. (E) Representative flow cytometry plots of lung CD4⁺T cells from LL, n = 10 per group, two independent experiments. (F) Quantification of percent of CD4⁺ T cells in LL, error bars are SEM. Two-way ANOVA with Bonferroni multiple comparisons test, **p = 0.0013 ****p < 0.0001.

T_RM cells reside in the lung after infections rather than regularly recirculating through blood and secondary lymphoid organs^10–15. To evaluate if the CD69⁺GL7⁺ and CD69⁺GL7⁻ subsets were extravasating from the lungs, we treated experienced mice with FTY720 every alternate day for 12 days prior to lung collection (Supplementary Figs. 3B and 3C). FTY720 prevents sphingosine-1-phosphate (S1P)-mediated T cell egress from lymph nodes and has been widely utilized to evaluate T cell migration in multiple tissues^31–33. We observed the proportion of neither CD69⁺GL7⁺ nor CD69⁺GL7⁻ subsets of CD4⁺ T cells were affected by FTY720 treatment, while CD11a^lo subsets were reduced significantly (Figs. 2E and 2F). These data suggest that the GL7⁺ subset, like other T_RM cells^31–33, are not in constant S1P-mediated circulation but instead, get retained in the lungs long-term.

To determine whether CD69⁺GL7⁺ cells that accumulated in Spn-experienced lungs were in a state of proliferation, we measured their expression of Ki67¹⁹. Although Ki67⁺ cycling cells were abundant in the first days after infection, both CD69⁺GL7⁺ and CD69⁺GL7⁻ subsets were equally low in Ki67 expression at the late timepoints (Fig. 3A, Supplementary Fig. 4A). Thus, neither of the two T_RM cell subsets were proliferating in the resting lungs. Despite the absence of actively proliferating cells, these GL7⁺ subsets of memory CD4⁺ T cells could persist in the lungs for at least 6 months (Figs. 3B and 3C, Supplementary Fig. 5B). Together, these data suggest CD69⁺GL7⁺ T cells in the Spn-experienced lung have resident memory-like characteristics resembling T_RM cells¹⁹. Hereafter, CD69⁺GL7⁺ and CD69⁺GL7⁻-subsets from these lungs will be referred to as GL7⁺ and GL7⁻ T_RM cells.

Fig. 3.

CD69+GL7+ subsets of CD4+ T cells have characteristics of resident memory T cells. (A) Flow cytometry of single cell suspensions from LL collected at 10, 14 and 35dpi. Representative plots of Ki67 staining in GL7⁺ or GL7⁻ of CD69⁺CD11a⁺ CD4⁺ T cells in LL. Mean +/− standard deviation of percent of parent, n = 5 to 6 per dpi, two independent experiments. LL of Sp19F experienced mice collected at 6, 18 and 24 wpi. (B) Representative flow cytometry plots of GL7 staining, n = 4 to 6 per group, two independent experiments. (C) Quantification of percent of GL7⁺ o of CD69⁺CD11a⁺ CD4⁺ T cells in LL, error bars are SEM, ordinary one-way ANOVA with Tukey multiple comparisons test. Mice infected intratracheally to the LL with SP19F, killed Klebsiella pneumoniae or Influenza A/PR8 on day 0 and 7 and rested until day 40. LL collected for flow cytometry. (D) Representative flow cytometry plots of GL7 staining, n = 6 per group, two independent experiments. (E) , Quantification of percent of T_RM cells of CD4⁺ T cells (right) and percent of GL7⁺ in CD4⁺ T_RM cells in LL of each infection model, error bars are SEM, ordinary one-way ANOVA with Tukey multiple comparisons test, *p = 0.0464, ****p < 0.0001. LL from Sp19 experienced mice processed with manual digestion method after anti-ARTC2 nanobody Intravenous injection for 30min. (F) Quantification of percent of T_RH and T_RM cells of parent (CD44⁺CD4⁺ T cells), error bars are SEM, paired T-test, ****p < 0.0001. (G) Representative flow cytometry plots of of T_RH and T_RM cells (right) and GL7 staining (left) of parent population, n = 6, two independent experiments. (H) Quantification of percent of GL7⁺ percent of parent (T_RM or T_RH), error bars are SEM, paired T-test, **p = 0.0015.

GL7 ligand can be expressed on T_FH cells in germinal centers^28,30. A subset of lung-resident cells with T_FH-like biology, T resident helper⁷ cells, arises in the lungs of mice recovered from influenza infection^17,18. Influenza infections as well as Klebsiella pneumoniae preparations elicit tertiary lymphoid structures in the lungs^34–36, which are not observed after recovery from pneumococcus^10,11. Although the numbers of CD4+ TRM cells varied across microbial challenges, CD69⁺GL7⁺ memory CD4⁺ cells were identified in all settings and at similar proportions (Figs. 4C and 4D). To evaluate if the GL7⁺ T_RM cells in lungs recovered from pneumococcus resemble T_RH cells that have been described after influenza, we evaluated the expression of T_RH-associated markers^17,18. Using a mechanical digestion method reported to conserve T_RH cell recovery from the lungs^17,18, we observed very few PSGL1^lowFR4^high T_RH cells (Fig. 3G, Supplementary Fig. 5A). GL7 ⁺ cells were found among the PSGL1^high-FR4^low T_RM cells, not the PSGL1^lowFR4^high T_RH cells (Figs. 3H and 3J, Supplementary Fig. 5A). Furthermore, BCL6⁺ and CXCR5⁺ cells were rare among the memory CD4⁺ T cells in lungs from mice recovered from pneumococcus, and when found these cells tended to be GL7⁻ (Supplementary Fig. 5B). We conclude that GL7⁺ T_RM cells are distinct from follicular-like lung T_RH cells.

Fig. 4.

GL7⁺ T_RM cells secrete similar cytokines reflective of their transcription factor profiles. Intracellular cytokine staining (ICS) of lung (Intravenous injection CD45.2⁻) CD4⁺T_RM cells stimulated ex vivo with PMA/Ionomycin in presence of golgi plug and golgi block. (A) Representative plot of IL-17A (right) and quantification of frequency of IL-17A⁺ cells (left) in parent population (GL7⁺ or GL7⁻ CD4⁺T_RM cells). (B) Representative plot of IFN-γ (right) and quantification of frequency of IFN-γ⁺ cells (left) in parent population (GL7⁺ or GL7⁻ CD4⁺T_RMcells). (C) Representative plots of IL-5/IL-13 (right) and quantification of percent of IL-13⁺ and/or IL-5⁺ cells in parent population (GL7⁺ or GL7⁻ CD4⁺T_RM cells). Paired t test, ***p = 0.0002, ****p < 0.0001. n = 13, two independent experiments.

Cytokine expression by GL7⁺ T_RM subsets reveals functional diversity

A cardinal function of T_RM cells in mucosal tissues is cytokine secretion^{10,12,14,15,37}. Our group and others have shown that cytokines secreted by lung CD4⁺ T_RM cells provide important signals to neighboring cells to orchestrate immune protection^16–18,21. To determine if GL7 expression highlighted a functionally distinct subset of lung CD4⁺ T cells, we designed an FSFC panel that included four cytokines and major T_RM lineage markers. We combined intracellular cytokine staining (ICS) with preliminary antibody labeling to track GL7 and CD69 surface marker expression throughout ex vivo stimuli with PMA/Ionomycin. As expected, the majority of lung CD4⁺ T cells from Spn-recovered lungs expressed IL-17A upon stimulation, and this applied to both the GL7⁺ and GL7⁻ subsets (Fig. 4A, Supplementary Fig. 6A). However, the distribution of cytokine-secreting cells differed between the T cell subsets. Compared to GL7⁻ T_RM cells from the same lung, the GL7⁺ T_RM cells were less frequently IL-17A-producers, and more frequently IFN-γ or IL-5 and/or IL-13 producers (Figs. 4A–4C). This insinuates a potential functional difference in this subset of T cells, less T_H17-like and more prone to T_H1 and T_H2-like responses. These data suggest GL7⁺ T_RM cells have a functional difference that results in a reduced output of IL-17A and an increase in IFN-γ, IL-5, and/or IL-13 after stimulation.

GL7⁺ T_RM subsets have distinct transcription factor profiles

Cytokine expression from helper T cells is typically dependent on cellular phenotypes dictated by lineage-defining transcription factors (LDTF)³⁸. We used a flow cytometry panel with T-bet, Gata-3, RORγT, and FoxP3, to help us discriminate the T_H1, T_H2, and T_H17 helper and T_reg subsets of T cells, respectively^21,38. Within the GL7⁺ T_RM subset, there was a lower frequency of RORγT⁺ and more Gata-3⁺ cells in comparison to the GL7⁻ counterparts (Figs. 5A and 5B, Supplementary Fig. 7A). FOXP3⁺ cells were rare among both GL7⁺ and GL7⁻ T_RM cells, but a little less so in the former (Supplementary Fig. 7B) This implies that the GL7⁺ T cells may be more predisposed to the T_H2-like lineage and less predisposed to the T_H17-like lineage compared to the GL7⁻ cells from the same lungs, consistent with the cytokine expression data.

Fig. 5.

GL7⁺ T_RM cells have unique transcription factor profiles. (A) Representative flow cytometry plots of intracellular lineage-defining transcription factor (LDTF) staining in live lung CD4⁺T_RM cells and (B) Frequency of LDTF⁺ cells in parent population (GL7⁺ and GL7⁻ CD4⁺T_RM). Paired t test, **p = 0.005. (C) Relative multipotency index²¹ of GL7⁺ and GL7⁻ lung (Intravenous injection CD45.2⁻) CD4⁺T_RM cells. Bar graphs of mean and error bars of SEM. Ratio paired t test, **p = 0.0044. n = 9, two independent experiments. (D) Frequency of GL7⁺ or GL7⁻ CD4⁺T_RM from each lineage in parent population (GL7⁺ and GL7⁻ CD4⁺T_RM). Bar graph of mean and error bars of SEM. Two-way ANOVA with Bonferroni multiple comparisons test, ***p = 0.0008

While expression of a single transcription factor has been widely used to define T_H cell lineages³⁸, multi-lineage CD4⁺ T cell plasticity is known^21,39,40. Studies by our group recently demonstrated that many T_RM cells in Spn-experienced lungs express multiple lineage-defining transcription factors simultaneously^20,21. To evaluate if GL7⁺ T_RM cells represented a subset with unique combinations of transcription factors, we combined LDTF data from Fig. 4 with a nested manual gating strategy that measured the expression of one, two, or three of the factors simultaneously in each T_RM subset among FOXP3⁻ cells (Supplementary Fig. 7C)²¹. While cytokine analyses suggested that multifunctionality was rare (Supplementary Fig. 6B), lung T cells often exhibited multiple lineage-defining transcription factors (Fig. 5D) and the relative multipotency index²¹ revealed that the GL7⁺ T_RM cells tend to express fewer compared to the GL7⁻ counterparts (Fig. 5C). Examining the co-expression phenotypes more specifically, we identified eight subpopulations, each of which was present in both T_RM subsets. The most common phenotype was Gata-3⁺RORγT⁺ cells, which may reflect a skewing toward Th2/Th17 (Fig. 5D). Within GL7⁻ T_RM cells, the second most common subtype was a triple-negative population that did not express any of the lineage-defining transcription factors included in our study. In GL7⁺ T_RM cells, the second most common subtype was Gata-3⁺ cells lacking the other transcription factors, which were significantly less abundant within the GL7⁻ population (Fig. 5D, Supplementary Fig. 6B).

These data further support the concept that GL7⁺ cells represent a distinct subphenotype of lung T_RM cells. While subsets exist among both GL7⁺ and GL7⁻ T_RM cells, the differences in distribution suggest the former tilt away from T_H17 and toward T_H2 compared to the latter, consistent across transcription factor content and cytokine expression.

Upon further evaluation, we observed different levels of expression of GL7 within GL7+ T_RM cells, which we defined as GL7^hi and GL7^lo. Inspection of the GL7 vs. transcription factor data suggested that cells with lower or higher amounts of GL7 (Fig. 6A) may have different expression levels of LDTFs. The GL7^lo subsets had frequencies of RORγT⁺ cells and T-bet⁺ cells that matched GL7⁻ T_RM cells, but Gata-3⁺ cells were enriched in GL7^lo T cells (Fig. 6B–C). In contrast, the GL7^hi subsets had decreased frequencies of RORγT⁺ cells and increased frequencies of T-bet⁺ cells compared to the GL7⁻ counterparts, while Gata-3⁺ cells were unchanged (Fig. 6B–C). Altogether, these transcription factor results match the cytokine expression data well, providing consistent evidence for decreased T_H17 but increased T_H1 and T_H2 skewing in the GL7⁺ lung CD4⁺ T_RM cells compared to the GL7⁻ lung CD4+ T_RM cells.

Fig. 6.

GL7⁺ T_RM cells are a bimodal population with different roles. (A) Manual gating of GL7⁻, GL7^Low and GL7^High in lung (Intravenous injection CD45.2⁻) CD4⁺T_RM cells. (B) Representative flow cytometry plots of intracellular LDTF staining in lung (Intravenous injection CD45.2⁻) CD4⁺T_RM with GL7⁻, GL7^Low and GL7^High gates. (C) Frequency of RORγT⁺, T-bet⁺ and Gata-3⁺ cells in parent population (GL7⁻, GL7^Low and GL7^HighCD4⁺T_RM). Bar graphs of mean and error bars of SEM. One-way ANOVA with Tukey test of multiple comparisons. *p = 0.02, **p = 0.003, ****p < 0.0001. n = 9, two independent experiments.

DISCUSSION

Our study highlights a novel population of resting CD4⁺ T cells residing in lungs that have recovered from pneumonia, distinguished by expression of Neu5Ac that is recognized by the GL7 monoclonal antibody. These cells share many characteristics with GL7⁻ T cells from these lungs, including a stable and long-lived CD11a^HiCD69⁺CD62L^LoCD44^HiKi67⁻ phenotype, which led us to conclude they are lung-resident memory lymphocytes, further supported by FTY720 studies. However, the GL7⁺ T cells in these lungs differ from their GL7⁻ neighbors. The former expresses lower levels of the CMAH enzyme contain a different balance of LDTFs at rest and produce a different mixture of cytokines when activated. We propose that GL7 marks a previously unrecognized subset of lung T_RM cells.

GL7 reactivity is not unprecedented for mouse lymphocytes. GL7 ligand expression has been detailed in cells from mice in primary and secondary lymphoid organs, including B_GC cells^22,26,27,41 as well as a subset of T_FH cells whose surface markers include CD4⁺CXCR5^HiGL7^+28,30. While CXCR5⁺ and BCL6⁺ memory CD4⁺ T cells could be recovered from Spn-experienced lungs using manual digestion methods, they were mostly GL7⁻. The GL7⁺ subsets also did not express markers associated with the TFH-like lung T_RH cells^17,18, suggesting that these GL7⁺ T_RM may not be the T_FH phenotype or the newly reported and related T_RH cells^17,18. GL7 ligand expression is acutely upregulated on B cells and T cells due to in vitro stimulation^22,26,27,29. The GL7⁺ phenotype in these activated naïve cells is due to a transient decrease in CMAH, the reduction of which results in the surface glycan sialylation with Neu5Ac^27,29.

We demonstrate that GL7⁺ T_RM cells have low CMAH-like activated naïve cells but with several important distinctions: 1) they are found in resting lungs that have recovered from bacterial or viral pneumonia, but without any ongoing infection or inflammation; 2) they exhibit markers of resident memory cells, 3) they form a stable population that persists long-term; 4) they do not circulate out of the lungs; 5) and they are not proliferating. We believe this may be the first evidence for GL7 ligand expression on resting or memory T cells.

We postulate that GL7⁺ cells are a distinct subset of lung CD4 + T_RM cells that may influence the immune tone of lungs responding to a heterotypic respiratory infection. Most memory T_RM cells after a Spn lung infection are of the T_H17 lineage that predominantly secretes IL-17A during activation or subsequent reinfection^10,21. Previous work from our group and others revealed T_RM cells that expressed transcription factors from multiple lineages and could secrete diverse cytokines after ex vivo stimulation^20,35,42. Our current work supports these claims and shows such plasticity applies to both GL7⁻ and GL7⁺ subsets of T_RM cells. However, the GL7⁺ subset appears to favor T_H1 and T_H2 phenotypes more and the T_H17 phenotype less, in comparison to the more numerous GL7⁻ lung T_RM cells. This conclusion is based on balances of both the transcription factors and the cytokines in these cells; and suggests T_RM cells in Spn-recovered lungs share many similarities, but the GL7⁺ subset is more poised to provide T_H1 and T_H2 responses than are GL7⁻ T_RM cells, which more robustly favor T_H17 responses.

There is still much to learn about these CD4⁺ T_RM cell subsets, but these first results suggest a possible model related to T cell contributions to immune resistance and tissue resilience in acutely infected lungs that have a relevant heterotypic infection history^3,43. T_H17 responses are key to many aspects of immune resistance against extracellular microbes like pneumococcus^34,44; T_H1 responses foster antimicrobial activity against intracellular organisms, and also instruct the local immune remodeling of phagocytes and possibly other lung cells⁴⁵; T_H2 responses promote tissue repair and regeneration after injury^42,46. Thus, it is possible that the more numerous GL7⁻ T_RM cells are most critical to the immune resistance and the elimination of pneumococcus from acutely infected lungs^10,16, whereas the scarcer GL7⁺ T_RM cells affect immunomodulation and repair during the tissue recovery stage after pneumonia.

Humans have an inactive CMAH gene and lack a functional CMAH enzyme, therefore surface glycans on human lymphocytes are uniformly sialylated with Neu5Ac^47,48. Human lung T_RM cells therefore should not exhibit the surface sialylation variations seen here in mouse lung T_RM cells. However, humans use a different pathway that functions analogously to sialylation in mice. Humans differentially sulfate lymphocyte surface glycans, and the sulfation state toggles CD22 activation in human cells like glycan sialylation toggles CD22 in mouse lymphocytes^47–50. CD22 interaction with Neu5Gc or sulfated Neu5Ac inhibits B cells⁴⁹. A separation of T_RM cells by CD22 ligand expression may create subsets of cells that distinctly interact with and regulate CD22⁺ B cells, advancing another potential biological significance for these lung CD4+ T_RM cell subsets. The present results raise new questions about whether human lungs contain two CD4+ T_RM cell subpopulations which differ in Neu5Ac sulfation.

As the roles of CD4⁺ T_RM cells in lung immunity continue to be elucidated, we begin to understand that T_RM cells are a diverse community with myriad functions that contribute to lung protection. We propose GL7⁺ T_RM cells as one such subset, whose functions may involve differential cytokine secretion, contributing to the overall immune roles of lung T_RM cells. Future efforts to develop mucosal vaccines that trigger a robust immune environment in the lungs should consider how the diverse subsets of resident lymphocytes provide mechanisms of protection that result in more effective elimination of microbes and better protection of tissues against damage.

METHODS

Mice

We obtained 6-week-old C57BL/6J^{51 000664} mice from The Jackson Laboratory (Bar Harbor, ME, USA). All mice were housed in specific pathogen-free conditions with a 12-hour light-dark cycle, and ad libitum access to standard chow and water. Then, 7–8-week-old mice that were adequately rested for 1 week after transfer were used for experiments. Mice were euthanized using isoflurane overdose and death was confirmed using pneumothorax before organ collections. Animal procedures were approved by Institutional Animal Care and Use Committee of Boston University (IACUC).

Streptococcus pneumoniae infections

Spn-specific lung-resident adaptive immunity was generated as previously described^{10,12,13,15,37}. Mice were anesthetized with intraperitoneal (i.p.) administration of ketamine (50mg/kg) and xylazine (5mg/kg) before infection. After surgical exposure of the trachea, sterile saline, or Spn was administered intratracheally via a 24-gauge angiocatheter directed to the LL. Mice received 50μL saline containing 1–5 × 10⁶ CFU of serotype 19F Spn (Strain EF3030) or saline. Seven days later, mice were exposed again to either saline or Sp19F and then allowed to recover for 28–35 days.

Influenza A infection

Mice were anesthetized with i.p. administration of ketamine (50mg/kg) and xylazine (5mg/kg) before infection. After surgical exposure of the trachea, Influenza A/PR8/34 (IAV) was administered intratracheally via a 24-gauge angiocatheter directed to the LL. Mice received 50μL saline containing ~15 plaque-forming units per mouse and allowed to recover for 40 days.

Heat-killed Klebsiella pneumoniae exposure

Klebsiella pneumoniae (KPN) ATCC strain 43816, serotype 2, was prepared as previously described³⁵. It was expanded overnight at 37°C, 5% CO₂ in blood agar plates, and replated for 4 hours to reach log phase. Bacteria were diluted to 0.3 optical density measured at 600 nM, pelleted, and washed twice in sterile phosphate buffered saline (PBS) before resuspension in same. Bacteria were heat-killed by incubating at 70°C for 2 hours, and cultures confirmed no remaining bacterial viability. Mice were anesthetized with i.p. administration of ketamine (50mg/kg) and xylazine (5mg/kg). After surgical exposure of the trachea, heat-killed KPN was administered intratracheally via a 24-gauge angiocatheter directed to the LL. Mice received 50 uL containing 2.8 × 10⁶ CFU-equivalents of heat-killed KPN. Seven days later, mice were re-exposed via similar heat-killed KPN administration, and then allowed to recover for 40 days.

Enzymatic digestion of mouse lungs for flow cytometry

To allow enumeration of extravascular (tissue) versus intravascular (circulating) fraction of lymphocytes, mice were anesthetized with i.p. administration of ketamine (50mg/kg) and xylazine (5mg/kg). Three minutes prior to euthanasia, 2 μg fluorochrome-labeled anti-CD45.2 antibody in 100 μL saline were retro-orbitally administered. Right and left lobes of lungs were collected separately in RPMI 1640 with 10% FBS before processing for flow cytometry. Lungs were digested in type 2 collagenase (Worthington Biochemicals, Lakewood, NJ, USA), DNase I (Sigma-Aldrich, St. Louis, MO, USA) and 2.5 mM CaCl2 in PBS for 1 hour at 250–300 rpm at 37°C. Digested lungs were filtered through a 70 μm cell strainer, followed by RBC lysis (Sigma-Aldrich, St. Louis, MO, USA). Cells were blocked with TruStain αCD16/CD32 Fc-Block (BioLegend, San Diego, CA, USA) before staining with optimized dilutions of antibodies (Supplementary Table 1) in 100 uL of FACS buffer and Brilliant Violet Buffer (BD Biosciences, Franklin Lakes, NJ, USA). Conventional flow cytometry was performed on LSR II Flow Cytometer (BD Biosciences, Franklin Lakes, NJ, USA). High-dimensional full-spectrum flow cytometry was performed on Cytek Aurora (Cytek). SpectraFlo (Cytek) software was utilized for spectral unmixing of the raw full-spectrum data and unmixed data were analyzed with FlowJo software v10.7 (BD Biosciences, Franklin Lakes, NJ, USA). Gating strategies are provided in the Supplemental figures and were based on use of Fluorescence minus one (FMO) controls. For spectral unmixing and conventional compensation, UltraComp compensation beads (Thermo Fisher, Waltham, MA, USA) were used for all fluorochromes except when indicated. Total cell numbers were measured by multiplying the percentage fraction of specified leukocyte subset (in percentage of total) with the total number of cells recovered from each mouse lung.

Manual digestion of mouse lungs for flow cytometry

Experienced mice were anesthetized with i.p. administration of ketamine (50 mg/kg) and xylazine (5 mg/kg). Thirty minutes prior to euthanasia, 2.5 μg of Treg-protector^™ anti-ARTC2 Nanobody (Biolegend, San Diego, CA, USA) was retro-orbitally administered administered. Then, 3–5 minutes prior to euthanasia, 2 μg fluorochrome-labeled anti-CD45.2 antibody in 100 μL saline were retro-orbitally administered. Right and left lobes of lungs were collected separately in cold PBS and manually filtered through a 70 μm cell strainer followed by RBC lysis with Ammonium-Chloride-Potassium (ACK) lysis buffer (Thermo Fisher, Waltham, MA, USA). Cells were blocked with TruStain αCD16/CD32 Fc-Block (BioLegend, San Diego, CA, USA) before staining with optimized dilutions of antibodies (Supplementary Table 1) in 100 uL of FACS buffer and Brilliant Violet Buffer (BD Biosciences, Franklin Lakes, NJ, USA).

Heterotypic immune timeline infection

Adult mice were anesthetized via i.p injection of ketamine and xylazine before infection. Spn suspended in sterile saline was instilled via a 24-gauge angiocatheter directed to the left lobe after surgical exposure of the trachea. Mice were infected i.t. with 50 μL saline containing 1–5 × 10⁶ CFU of serotype 19F Spn (Strain EF3030) or saline. Seven days later, only some mice were exposed again to Sp19F. Control mice received sterile saline on days 0 and 7. Lungs were collected and evaluated on day 0, 1, 3, 10, 14, 24 and 35 and processed for high-dimensional FSFC as previously described^20,21. Data were analyzed using FlowJo v10.7 (BD Biosciences, Franklin Lakes, NJ, USA).

FTY720 Dosage

Experienced mice received 1mg/kg of FTY720 (Sigma-Aldrich, St. Louis, MO, USA) i.p. every other day for 12 days. Lungs were collected and processed for flow cytometry staining. Flow cytometry was performed on BD LSR II Flow Cytometer (BD Biosciences, Franklin Lakes, NJ, USA).

Intracellular staining for transcription factors

To evaluate transcription factor and nuclear factors, eBioscience Foxp3/Transcription Factor Staining Buffer Set (Thermo Fisher, Waltham, MA, USA) was used following manufacturer’s protocols.

Ex vivo Stimulation and Intracellular cytokine staining

2×10⁶ cells per 1 × 10⁶ cells were blocked with TruStain αCD16/CD32 Fc-Block (Biolegend, San Diego, CA, USA) for 5 minutes before staining with optimized dilutions of mouse anti-CD69 (PE) and GL7 (APC) in FACS buffer for 35 min on ice. Cells were then washed and resuspended and plated in T cell media with 250 ng/mL of Phorbol Myristate Acetate (PMA) (LC Laboratories, Woburn, MA, USA), 1.5 ug/mL Ionomycin (Sigma-Aldrich, St. Louis, MO, USA), and Monensin (Biolegend, San Diego, CA, USA), and Brefeldin A (Biolegend, San Diego, CA, USA) for 6 hours at 37°C and 5% CO₂. Cells were then processed for ICS with Intracellular Fixation & Permeabilization Buffer Set (Thermo Fisher, Waltham, MA, USA) following manufacturer’s instructions.

Flow cytometric assisted sorting (FACS)

Single-cell suspensions prepared from digested lungs were prepared with RBC lysis (Sigma-Aldrich, St. Louis, MO, USA) and blocked with TruStain αCD16/CD32 Fc-Block (BioLegend, San Diego, CA, USA) for 5 minutes. Cells were stained in FACS buffer with optimized dilutions of antibodies. GL7⁺ and GL7⁻ subsets of lung (i.v.CD45.2⁻) CD4⁺ T_RM cells were sorted into 20% FBS RPMI 1640 on ice with FACS-Aria II SORP (BD Biosciences, Franklin Lakes, NJ, USA). After sorting, cells were cultured for 6 hours with T cell media with or without PMA/Ionomycin stimulation at 37°C and 5% CO₂. Immediately after, cells were washed and pelleted into (RNA) ribonucleic acid Lysis Buffer (Qiagen, Germantown, MD, USA) and stored at −80°C until RNA extraction.

RNA Extraction and Quantitative Real-time PCR

RNA from sorted T cells was extracted using the RNAeasy Micro Kit (Qiagen, Germantown, MD, USA) following manufacturer’s protocols and stored at −80°C. qRT-PCR was performed using the RNA-to-Ct kit (Thermo Fisher, Waltham, MA, USA). Primers for Cmah were commercially available TaqMan gene expression assays (Mm00483341_m1). We used 18s rRNA (Thermo Fisher, Waltham, MA, USA) to normalize expression levels of mRNA. Fold change of mRNA was calculated from all samples over unstimulated GL7⁻ samples.

Duplex immunohistochemistry (IHC)

Mouse lung samples were embedded in optimal cutting temperature compound and subsequently sectioned at 5 μm using a CM3050 S cryostat (Leica, Wetzlar, Germany), with samples transferred to positively charged slides. Slides were allowed to air dry for 1 hour, before a 15-minute incubation in 10% neutral buffered formalin before progression to immunostaining. IHC was conducted using a Ventana Discovery Ultra (Roche, Basel, Switzerland), with detailed assay parameters outlined in Supplementary Table 2. Secondary (HRP) horse radish peroxidase conjugated species-specific antibodies were utilized to develop primary antibodies (Vector Laboratories, Burlingame, CA), with signal detection through TSA conjugated Opal fluorophores (Akoya Biosciences, Marlborough, MA, USA). Spectral (DAPI) 4′,6-diamidino-2-phenylindole was utilized to label nuclei (Akoya Biosciences, Marlborough, MA, USA). For figure preparation, 200x total magnification digitalized whole slide images of the duplex IHC assay were generated using a multispectral Vectra Polaris Automated Quantitative Pathology Image System (Akoya Biosciences, Marlborough, MA, USA).

T cell Media

T cell media was prepared with RPMI 1640 (Thermo Fisher, Waltham, MA, USA) supplemented with 10% heat-inactivated fetal bovine serum, Penicillin/Streptomycin (100 U/mL, Thermo Fisher, Waltham, MA, USA), 1X Non-essential amino acids (Thermo Fisher, Waltham, MA, USA), 2 mM L-Glutamine (Glutamax, Thermo Fisher, Waltham, MA, USA), 1 nM Na-pyruvate (Thermo Fisher, Waltham, MA, USA), 10 mM HEPES (Thermo Fisher, Waltham, MA, USA) and 0.2 uM b-mercaptoethanol.

Calculation of relative multipotency index

We calculated the Relative Multipotency Index as previously described²¹, using the following formula:

$$\mathit{\text{RelativeMultipotencyIndex}}=\text{In}\left(\frac{\%\mathit{\text{ofmemoryCD}}4+\mathit{\text{Tcellsexpressing}}\ge 2\mathit{\text{LDTFs}}}{\%\mathit{\text{ofmemoryCD}}4+\mathit{\text{Tcellsexpressing}}1\mathit{\text{LDTF}}}\right)$$

Statistical analyses

Statistical analyses were performed using Prism Version 9.2.0 (GraphPad, Boston, MA, USA). Differences were deemed statistically significant if the p value was ≤0.05. Each figure legend communicates the number of mice used, the experiment replicates performed, and the statistical tests used to make comparisons. For all figures, data are represented as mean ± (SEM) standard error of the mean.

DATA AVAILABILITY

Data generated during and/or analyzed during the current study are available from the corresponding author on reasonable request.