Memory CD4⁺ T-cell–mediated protection depends on secondary effectors that are distinct from and superior to primary effectors

Abstract

Whether differences between naive cell-derived primary (1°) and memory cell-derived secondary (2°) CD4⁺ T-cell effectors contribute to protective recall responses is unclear. Here, we compare these effectors directly after influenza A virus infection. Both develop with similar kinetics, but 2° effectors accumulate in greater number in the infected lung and are the critical component of memory CD4⁺ T-cell–mediated protection against influenza A virus, independent of earlier-acting memory-cell helper functions. Phenotypic, functional, and transcriptome analyses indicate that 2° effectors share organ-specific expression patterns with 1° effectors but are more multifunctional, with more multicytokine (IFN-γ⁺/IL-2⁺/TNF⁺)-producing cells and contain follicular helper T-cell populations not only in the spleen and draining lymph nodes but also in the lung. In addition, they express more CD127 and NKG2A but less ICOS and Lag-3 than 1° effectors and express higher levels of several genes associated with survival and migration. Targeting two differentially expressed molecules, NKG2A and Lag-3, reveals differential regulation of 1° and 2° effector functions during pathogen challenge.

Analyses of the mechanisms underlying memory CD4⁺ T-cell–mediated protection have focused largely on their earlier provision of help as compared with naive cells (1), although it also is appreciated that highly activated secondary CD4⁺ T-cell (hereafter, 2°) effectors develop after the re-expansion of memory populations (2). Studies also suggest that optimal protection provided by T helper type 1 (T_H1)-like memory CD4⁺ T cells correlates with the capacity to produce multiple cytokines, including IFN-γ, TNF, and IL-2, rather than IFN-γ alone (3). Whether 2° effectors derived from protective memory CD4⁺ T cells retain this phenotype, the extent to which 2° effectors contribute to the protection mediated by memory CD4⁺ T cells, and whether and how 2° effectors differ from primary CD4⁺ T-cell (hereafter, 1°) effectors are unclear.

Comparison of 1° and 2° CD4⁺ T-cell effectors within mixed populations is difficult. The higher proportion of antigen-specific memory cells as compared with naive cells complicates quantitative analysis, and the maintenance of very few antigen-specific cells in general (4) precludes rigorous analysis ex vivo of the effectors to which they give rise. Polyclonal naive and memory T-cell populations also may differ in repertoire and T-cell receptor (TcR) affinity (5, 6), further complicating comparisons. Finally, phenotypic discrimination alone between highly activated effectors and cells that have divided only once or twice is not reliable, because effectors responding in different organs can express different surface phenotypes (7). To overcome these obstacles, we transferred equal numbers of naive and memory HNT TcR transgenic CD4⁺ T cells recognizing the A/PR8/34 (PR8) strain of influenza A virus (IAV) to unprimed hosts and then challenged with PR8 to compare directly the generation, function, and protective capacity of 1° and 2° effectors.

IAV infection presents a compelling model for addressing these questions. The transfer of 1° effectors to unprimed mice can protect against lethal challenge (8–10), and studies demonstrating memory CD4⁺ T-cell protection in mice deficient for CD8⁺ T and B cells suggest that an important helper-independent protective contribution of memory CD4⁺ T cells may be mediated directly by the 2° effectors (11–13). In addition, IAV-specific 1° effector (7, 10) and memory (14, 15) CD4⁺ T cells isolated from the lung and from secondary lymphoid organs (SLO) display distinct functional and phenotypic characteristics. Whether 2° effectors comprise a similarly heterogeneous population is unknown. This issue is important, because recent studies show that lung-resident memory CD4⁺ T cells provide greater protection against IAV than SLO-resident memory cells (16). Thus, a more comprehensive understanding of organ-specific heterogeneity within responding CD4⁺ T-cell pools may provide clues about the critical attributes of the most protective CD4⁺ T cells that could be generated by vaccination.

We find that although both populations develop and peak with similar kinetics, the 2° effectors accumulate in greater numbers in the lung, the primary site of infection. We show that the generation of the 2° effectors is the critical component of protective immunity mediated by memory CD4⁺ T cells against IAV and that 2° effectors are superior to 1° effectors in mediating viral clearance. We demonstrate that 2° effectors contain more cells producing TNF and/or IL-2 together with IFN-γ and fewer cells producing IL-10 than do 1° effectors. In addition, we identify several phenotypic markers that distinguish the two effector populations from each another.

To define the differences between 1° and 2° effectors further, we analyzed gene expression by microarray. The 1° and 2° effectors recovered from both lung and SLO display a high degree of shared, organ-specific specialization. However, 2° effectors are less compartmentalized, as evidenced by a wider distribution of follicular helper T (T_FH) cells. Furthermore, we identify a short list of genes that distinguish 1° and 2° effectors and that could be involved in controlling the greater expansion and more pluripotent functions of 2° effectors. Finally, we demonstrate the specific regulation of 1° or 2° effector cytokine production by blocking the differentially expressed surface proteins NKG2A and Lag-3. These findings define pathways that explain, in part, the functional superiority of the memory CD4⁺ T-cell response.

Results

Generation of 1° and 2° Effectors in Vivo.

To generate comparable effectors from naive and memory precursors, we transferred equal numbers of naive or memory HNT CD4⁺ T cells to Thy-disparate hosts and infected the hosts with PR8. We used both in vivo-primed and reisolated (in vivo PR8 memory) and in vitro-generated T_H1 memory cells, allowing the investigation of effectors arising from heterogeneous memory cells resulting from in vivo priming and from populations with defined polarization (17). Upon transfer to uninfected mice, similar numbers of naive and memory HNT CD4⁺ T cells were initially present in all organs analyzed and both decayed with identical kinetics (Fig. S1), arguing against differences in initial trafficking or in survival of donor cells after adoptive transfer contributing to the results reported here.

Unlike naive cells, memory CD4⁺ T cells are poised for rapid secretion of cytokines (18). In agreement with studies using in vitro-generated memory cells (19), in vivo PR8 memory cells up-regulated CD69 1–2 d earlier in the lung and draining lymph nodes (dLN) than did naive cells (Fig. 1A), indicating a more rapid activation of memory cells, but 4 d postinfection (dpi) the expansion of naive and memory donors was similar (Fig. 1B). All donors reached similar peak numbers in SLO by 7 dpi, but those arising from memory precursors accumulated in greater number in the lung (Fig. 1B). A similar pattern was observed when precursors were reduced up to 100-fold, with in vivo- and in vitro-derived memory cells giving rise to about five times and 10 times more cells in the lung, respectively, than naive donors (Fig. 1C); this result confirms that the kinetics and magnitude of expansion are independent of precursor frequency.

Fig. 1.

Generation of 1° and 2° effectors in vivo. Naive or memory HNT cells (5 × 10⁶) were transferred to Thy-disparate hosts that then were infected with 500 EID₅₀ PR8. Spleen, dLN, and lungs were harvested on stated days (n = 5 mice per group on each day) and stained to visualize donor cells. (A) CD69 expression on donor cells. (B) Number of donor cells detected. *P < 0.05. (C) Lung-resident donor cells at 7 dpi from mice receiving indicated number of donor cells. n = 4 mice per group. *P < 0.05, ***P < 0.001, and ****P < 0.0001. Upper asterisks represent in vitro T_H1 memory cells vs. naive cells; lower asterisks represent in vivo PR8 memory cells vs. naive cells). (D) Proportion of 1° and 2° effectors present in the dLN at 7 dpi (as determined by loss of CFSE) derived from mice receiving the indicated number of precursors. n = 3–5 mice per group. (E) CD4⁺ T cells were isolated from the SLO and lung of unprimed or PR8-primed mice and were labeled with CFSE. Then 1 × 10⁷ cells were transferred to Thy-disparate hosts, and the hosts then were challenged with 500 EID₅₀ PR8. (F) At 7 dpi, donors that had divided at least five times were enumerated. n = 5 mice per group. All error bars represent the SD.

By 7 dpi, virtually all donor cells recovered from the lung were effectors as defined by their having undergone five or more divisions based on loss of carboxyfluorescein succinimidyl ester(CFSE), as were >80% of cells arising from naive and >90% of cells arising from memory donors in SLO. We titrated the number of donor cells and found that at lower numbers transferred (≤2 × 10⁶) nearly all naive and memory donors developed into effectors in SLO and lung (Fig. 1D). Together Fig. 1 C and D indicate that both naive and memory cells give rise to highly divided effectors (Fig. 1D) whose number is proportional to input (Fig. 1C). It is notable that, despite the earlier activation of memory cells (Fig. 1A), the kinetics of development in response to PR8 in SLO were similar in 1° and 2° effectors, but the magnitude of the 2° effector response increased markedly in the lung. This enhanced accumulation of 2° effectors in the lung was seen in all experiments.

To determine whether polyclonal memory CD4⁺ T cells would show a similar enhanced effector response in the lung, we transferred CFSE-labeled polyclonal CD4⁺ T cells from unprimed or PR8-primed mice (enriched for PR8 memory) to new hosts and infected the hosts with PR8 (Fig. 1E). At 7 dpi, effectors were enumerated by gating on donors that had divided five times or more. Effectors derived from donors containing memory cells reached levels about fourfold higher in SLO, likely reflecting the increased proportion of precursor IAV-specific cells (Fig. 1F). Strikingly, cells from primed animals gave rise to more than 20-fold more effectors in the lung, about five times higher than the ratio in SLO (Fig. 1F). This result is consistent with observations using equal numbers of naive and memory monoclonal CD4⁺ T cells showing that the response in the lung is five- to 10-fold higher for 2° effectors than for 1° effectors. The enhanced ability of memory CD4⁺ T cells to give rise to effectors in the lung following IAV infection suggests that this feature of memory may contribute to enhanced protection.

Secondary Effectors Are Critical for Memory CD4⁺ T-Cell Protection.

We next analyzed the protective efficacy of the 2° effectors. Memory CD4⁺ T cells regulate innate immunity during the initial days of IAV infection (19) and provide help for the enhanced antibody production that is evident by 7 dpi (20). We reasoned that these functions would occur by 5 dpi, but, as described above, we found that 2° effectors enter the lung only at 6 dpi and peak at 7–8 dpi (Fig. 1). Therefore we transferred 5 × 10⁶ memory cells [the number of memory cells found to protect reliably against lethal PR8 challenge (12)] or an equal number of control naive cells to unprimed Thy-disparate hosts and then selectively depleted donors at 5 dpi, leaving earlier functions of memory cells intact but depleting 2° effectors before they finished differentiating and trafficked to the lung (Fig. 2A). Treatment led to efficient removal of donor cells (Fig. 2A). Mice receiving memory cells but not naive HNT cells and treated with control or Thy1.2-depleting antibody had equivalently enhanced germinal center B-cell formation (Fig. 2B) and PR8-specific IgG (Fig. 2C) at 8 dpi, indicating that depletion at 5 dpi did not effect helper function. However, the recipients depleted of memory HNT cells did not survive lethal challenge (Fig. 2D), although they survived longer than the recipients of naive HNT cells (Fig. 2D). This result stresses the necessity of late-acting events and also is consistent with some protective contribution from earlier-acting memory cell functions (19). These results suggest that 2° effectors are key contributors to memory CD4⁺ T-cell–mediated protection against IAV.

Fig. 2.

The 2° effectors are critical for memory CD4⁺ T-cell–mediated protection. (A) Naive or in vitro-generated memory HNT cells (5 × 10⁶) were transferred to Thy-disparate hosts, and the hosts then were infected with a 10,000 EID₅₀ of PR8 and treated with either donor cell depleting or isotype control antibody on day 5 with representative staining of dLN on day 6. (B) Absolute numbers of germinal center (GC) B cells present in spleen and dLN at 8 dpi. (C) Serum PR8-specific IgG levels were determined from three to five mice per group. (D) Survival of experimental groups described in A. n = 5 per group. Results are shown for one of two independent experiments. All error bars represent the SD.

Secondary Effectors Are Superior to 1° Effectors in Mediating Protection Against IAV.

The protection provided by 2° but not 1° effectors described above could be caused largely by the differences in the magnitude of the response, as suggested by studies correlating increased protection against IAV with increasing numbers of transferred 1° HNT effectors (8). Alternatively, 2° effectors may mediate superior protection through functional differences that distinguish them from 1° effectors. To establish that 2° effectors indeed are capable of enhanced protection as compared with 1° effectors and to investigate the possible reasons for the difference, we isolated effectors from recipients of naive or memory HNT cells on 7 dpi that had been infected with PR8. We transferred equal numbers of isolated 1° or 2° effectors directly to unprimed hosts and challenged the hosts with a lethal dose of PR8 (Fig. 3A) against which 5 × 10⁶ in vitro-generated 1° HNT effectors are required to protect unprimed hosts and to promote enhanced viral clearance at 4 dpi (8). Consistent with previous studies, 5 × 10⁶ 1° or 2° in vivo-generated effectors rescued hosts (Fig. 3B). However, when 2.5 × 10⁶ cells were transferred, only 2° effectors provided protection and enhanced viral control (Fig. 3 B and C). These results demonstrate that, in addition to their enhanced representation in the lung, 2° effectors are superior to 1° effectors in mediating protection as evaluated per cell input.

Fig. 3.

Secondary effectors are superior to 1° effectors in mediating protection against lethal infection. (A) Naive or in vitro-generated memory HNT cells were transferred to Thy-disparate hosts that then were infected with 500 EID₅₀ PR8 to generate 1° or 2° effectors, respectively. The 1° and 2° effectors were reisolated from SLO and lung at 7 dpi and were pooled, and equal numbers of each (5 or 2.5 × 10⁶) were transferred to unprimed BALB/c hosts. The hosts then were infected with 10,000 EID₅₀ PR8. (B) Survival (n = 10 per group) and (C) viral titers from mice receiving no cells or 2.5 × 10⁶ donor cells at 4 dpi (n = 5 per group). Results are shown for one of two independent experiments. (D) Survival of mice receiving no cells or 2.5 × 10⁶ 1° or 2° effectors isolated from the lung only. n = 10 per group. (E) At 4 dpi, donor cells in the lung were enumerated. n = 5 per group. (F) Representative staining of donor cells at 4 dpi for forward scatter (FSC), CD69, and CD25 expression as compared with host CD4⁺ T cells (results are shown for one of two independent experiments). All error bars represent the SD.

The previous experiments utilized effectors pooled from lung and SLO. Recent studies suggest that CD4⁺ T cells isolated from the lung provide improved protection compared with those in SLO (16). As the pooled 2° effectors contain proportionally more cells isolated from the lung compared with 1° effectors (see Fig. 1B), we transferred 2.5 × 10⁶ 1° or 2° effectors isolated only from the lung to unprimed hosts and challenged the hosts with a lethal dose of PR8. We again observed that 2° effector protection was significantly enhanced compared with 1° effector protection (Fig. 3D), although 1° effectors from the lung provided far greater protection than the pooled 1° effectors (compare Fig. 3 D and B). Perhaps surprisingly, equivalent numbers of 1° and 2° effectors were observed in the lung at 4 dpi (Fig. 3E), and both populations exhibited a highly activated phenotype (Fig. 3F). This result suggests that 2° effectors have enhanced per cell function as compared with 1° effectors responding in the same organ.

Cytokine Profiles of 1° and 2° Effectors.

Many T-cell effector functions are mediated by cytokines made after TcR triggering. To examine whether 1° and 2° effectors make different patterns or levels of cytokines, we used intracellular cytokine staining (ICCS) to assess the coproduction of the dominant cytokines seen in the IAV response, namely, IL-2, IFN-γ, and TNF. The 2° effectors derived from both in vitro and in vivo memory cells and recovered from either SLO or lung at 7 dpi contained higher proportions of IL-2⁺IFN-γ⁺ (Fig. 4A), TNF⁺IFN-γ⁺, and IL-2⁺IFN-γ⁺TNF⁺ (Fig. S2) cells than did 1° effectors. Dilution of precursors up to 100-fold did not alter the proportion of IL-2⁺IFN-γ⁺ 1° or 2° effectors substantially (Fig. 4B). Higher frequencies of IL-2⁺IFN-γ⁺ donors were also detected in effectors derived from polyclonal CD4 T cells from PR8-primed (enriched for 2° effectors) versus from naive mice (Fig. 4C), for which loss of CFSE was used to gate IAV-specific effectors (Fig. S3). These observations imply that enhanced multicytokine production is a general feature of 2° effectors.

Fig. 4.

Cytokine production and differential surface marker expression by 1° and 2° effectors. Naive or memory HNT cells (2 × 10⁶) were transferred to Thy-disparate hosts, and the hosts then were infected with 500 EID₅₀ PR8. (A) The percent of effectors coproducing IFN-γ and IL-2 as determined by ICCS at 7 dpi. (B) Coproduction of IFN-γ and IL-2 from effectors in dLNs at 7 dpi from mice initially receiving the indicated number of donor cells. n = 3 mice per group. *P < 0.05, **P < 0.005, ***P < 0.001. (C) Percentage of IFN-γ⁺IL-2⁺ effectors at 8 dpi from mice receiving naive or PR8-primed bulk polyclonal CD4 T cells. n = 4 per group. (D) Viral titers (n = 5 per group) and (E) percentage of IFN-γ⁺IL-2⁺ HNT effectors following infection with indicated doses of PR8 (n = 4 per group). (F) Dual cytokine-producing 1° and 2° effectors responding in the dLN were determined on stated days. (G) IL-10 and IL-17 production from lung-resident effectors. n = 3 mice per group. **P < 0.005, ***P < 0.001. (H) At 7 dpi, 1° and 2° effectors from the dLN and lung were analyzed for the indicated surface markers. Representative histograms are shown as well as MFI from five mice per group. *P < 0.05, ***P < 0.001. All error bars represent the SD.

We next infected mice with 50, 500, or 5,000 egg infectious doses (EID₅₀) PR8 to determine whether the magnitude of infection would impact effector cytokine production. No differences in viral burden were observed in mice initially receiving 2 × 10⁶ naive or memory cells and infected with the same dose of PR8. Pulmonary titers at 7 dpi were proportional to the challenge dose (Fig. 4D), consistent with previous studies (21). Interestingly, the proportion of IL-2⁺IFN-γ⁺ cells did decrease significantly with increasing challenge dose (Fig. 4E). This was most notable for 2° effectors in the dLN, where the frequency of IL-2⁺IFN-γ⁺ cells detected at 50 > 500 > 5,000 EID₅₀ challenge (P < 0.005 and 0.05, respectively). However, regardless of the challenge dose, 2° effectors contained more IL-2⁺IFN-γ⁺ cells than did 1° effectors. We analyzed cytokine production throughout the peak effector phase (6–8 dpi) to rule out the possibility that the difference was only transient or that kinetic patterns in cytokine production differed in 1° and 2° effectors. We observed similar IFN-γ⁺IL2⁺ and IFN-γ⁺TNF⁺ cells throughout (Fig. 4F). Thus, 2° effectors consistently contain a higher proportion of double and triple cytokine producing cells.

To evaluate other key cytokines produced by T cells during pathogen challenge, we next assessed IL-10 and IL-17. We have shown that IL-10 production by 1° CD4⁺ T-cell effectors can impede protection against IAV, whereas IL-17⁺ T cells can contribute to viral clearance (9, 22). We concentrated on lung-resident effectors, because the expression of both IL-10 and IL-17 from CD4⁺ T cells is restricted largely to the lung during IAV infection, with production of both IL-10 and IL-17 peaking at 7 dpi (9). A significantly lower fraction of 2° effectors, generated from either in vitro or in vivo memory cells, produced IL-10 (Fig. 4G), consistent with the lack of IL-10 observed in our earlier studies of memory CD4⁺ T-cell responses against IAV (9). A reciprocal pattern was found for IL-17 production. The 2° effectors generated from in vivo PR8-primed memory cells produced IL-17. However, virtually no IL-17 was detected from 1° effectors or from 2° effectors generated from in vitro T_H1 memory cells (Fig. 4G), suggesting that the T_H1-polarizing conditions in vitro suppress subsequent T_H17 differentiation. Together, as is consistent with the superior protective capacity of 2° effectors, these results indicate that 2° effectors are more capable of producing a set of cytokines implicated in viral protection and produce less of the inhibitory cytokine IL-10.

Phenotypic Profiles of 1° and 2° Effectors.

We next screened effectors for expression of a broad panel of surface proteins to see if we could identify any reproducible differences between 1° and 2° effectors. Although cells in the lung and in the SLO differentially expressed many markers associated with functional potential, consistent with previous studies (7), only a few key markers distinguished 1° from 2° effectors within the organs (Fig. 4H and Fig. S4A). The 2° effectors expressed increased CD127 (IL-7Rα) [a marker that is up-regulated on memory as compared with naive precursors (17) and that is necessary for CD4⁺ T-cell survival] and higher levels of NKG2A/C/E [a marker that has been associated with enhanced T-cell proliferative capacity and cytokine production (23, 24)]. In contrast, ICOS, whose higher expression has been linked with IL-10 production (25), was decreased on 2° effectors in the lung, correlating well with enhanced IL-10 production by 1° effectors. The 1° effectors found in the lung also expressed more of the inhibitory receptor Lag-3. These phenotypic distinctions were not affected by challenge dose or by whether the 2° effectors were derived from in vitro- or in vivo-generated memory cells (Fig. S4 B and C). These shifts in the expression of phenotypic markers that have been associated with function provide additional clues about which cellular pathways may be regulated differentially in 1° and 2° effectors subsets; however, the changes in levels are sufficiently modest that they alone are not likely to be useful for definitively distinguishing 1° from 2° effectors in mixed populations in vivo.

Organ-Specific Gene-Expression Analysis of 1° and 2° Effectors.

To assess further the magnitude of differences between 1° and 2° effectors and to look for unexpected genes that might be expressed differentially, we compared transcriptomes using whole-genome microarrays. Effectors in the lung express a distinct and apparently more highly differentiated phenotype than those in SLO, suggesting that organ-specific differences could obscure a global analysis. Therefore we separately analyzed effectors isolated from the spleen, dLN, and lung. We sort-purified 1° and 2° effectors from recipients of naive or in vivo PR8-primed memory HNT cells, respectively (Fig. S5). As predicted, and as is consistent with phenotypic and functional distinctions among organs, a number of genes that were shared by 1° and 2° effectors were expressed differentially among organs. For example, 1° and 2° effectors recovered from the lung shared about 200 genes that were differentially expressed in effectors in SLO (Fig. 5A and Dataset S1). This cohort of shared lung-enriched effector genes is enriched for immune-response, cytokine, chemokine, and defense-response pathways, as shown by Database for Annotation, Visualization and Integrated Discovery (DAVID) analysis (Fig. 5A), suggesting that 1° and 2° effectors in the lung share a spectrum of functions (likely including direct antiviral effector functions) and that 1° and 2° effectors in SLO have other discrete functions (e.g., helper activities).

Fig. 5.

Differential gene expression by 1° and 2° effectors in various organs. Organ-resident 1° and 2° effectors were sort-purified, and mRNA was prepared for microarray analysis as described. (A) (Upper) Venn diagram of the number of genes differentially expressed in 1° and 2° effectors in the lung and in the dLN and spleen combined (SLO). (Lower) DAVID functional annotation enrichment on genes shared by 1° and 2° effectors in the lung or in SLO. (B) T_FH-associated mRNA expression in 1° effectors relative to naive cells isolated from stated organs. Naive or in vitro memory HNT cells were transferred to Thy-disparate hosts; then the hosts were infected with 500 EID₅₀ PR8, and at 7 dpi effectors were stained for the T_FH-associated markers PD-1, CXCR5, and Bcl-6. (C) Representative staining and gating strategy. (D) The percentage and (E) the absolute number of T_FH in 1° and 2° effectors in stated organs. n = 3 per group. Results are shown for one of three independent experiments. **P < 0.005, ***P < 0.001. All error bars represent the SD.

To test this hypothesis, we analyzed whether SLO-resident effectors were enriched for T_FH-associated genes compared with lung effectors. Interestingly, only 1° effectors fit this pattern (Fig. 5B), suggesting either that 2° effectors in SLO do not express these T_FH-associated genes or that 2° effectors in all organs express T_FH signature genes. To test these two possibilities, we analyzed 1° and 2° effectors at 7 dpi for T_FH cells by FACS-based expression of CXCR5, PD-1, and Bcl-6 (Fig. 5C). We used in vitro T_H1 memory precursors to rule out the possibility that in vivo-primed PR8 memory cells used for microarray analysis contained preexisting T_FH cells that could account for the differing patterns of T_FH gene expression. Both 1° and 2° effectors contained phenotypically defined T_FH cells in SLO, but only 2° effectors contained significant T_FH cells in the lung (Fig. 5 D and E). Similarly, we observed T_FH in polyclonal 2° effector populations in the lung during heterosubtypic challenge, but no T_FH were observed in 1° polyclonal lung effectors (Fig. S6). Thus, although 1° and 2° effectors in different organs express transcriptomes indicative of organ-specific heterogeneity, 2° effectors express unique gene patterns that, together with enhanced multicytokine production, indicate broader and less-localized functional capacity.

Primary vs. 2° Effector Gene Expression.

We next focused on the genes that were expressed differentially in 1° and 2° effectors. A heatmap summarizing these genes by ANOVA analysis is shown in Fig. 6A. Pair-wise comparisons of 1° and 2° effector transcriptomes within each organ revealed differential expression of only a modest number of genes (about 150). Strikingly, very few of the genes expressed differentially between 1° and 2° effectors were common between the spleen, dLN, and lung, and no genes were differentially regulated in all three organs (Fig. 6B), again indicating a high level of organ-specific specialization. When subjected to DAVID analysis, the genes that show significant (at least twofold) changes between 1° and 2° effectors fall into functional annotation pathways such as regulation of transcription, metabolism, cell growth, and migration (Fig. 6C). In Dataset S2, we show the genes that define the greatest differences between 1° and 2° effectors. A cohort of some of the most compelling of these genes and their functional pathways are shown in Fig. 6D. Their differential expression was validated by RT-PCR (Fig. 6E). These genes could provide clues to the mechanisms responsible for the unique phenotype and function of 2° effectors responding to IAV (see below and Discussion).

Fig. 6.

Genes expressed differentially in 1° and 2° effectors. (A) Heatmap showing signal strength for individual probes for all organs, n = 2–3 organs per group. (B) Venn diagram of the number of genes differentially expressed in 1° and 2° effectors within spleen, dLN, and lung and differentially expressed genes shared by different organs. (C) DAVID functional annotation enrichment on genes differentially expressed in 1° and 2° effectors within organs. (D) Select genes within the enriched pathways. (E) PCR validation of differential gene expression in 1° and 2° effectors. n = 2–3 organs per group. Error bars represent the SD.

Differential Regulation of Function in 1° and 2° Effectors.

To evaluate whether some of the distinctions identified above correlate with functions that are likely to be relevant in protective efficacy, we tested whether the multicytokine production patterns of 1° and 2° effectors could be modified by targeting two differentially expressed molecules for which blocking antibodies are available. We first tested whether blocking a target preferentially expressed by 2° effectors would alter their cytokine production but not that of 1° effectors. We chose Klrc1, encoding NKG2A, because its expression was up-regulated by 2° effectors in the lung and dLN, mirroring surface expression of NKG2A/C/E (Fig. 4G). We confirmed higher expression of Klrc1 in 2° effectors by RT-PCR (Fig. 6E) but found equal expression of Klrc2-3, encoding NKG2C/E, in both effector populations (Fig. 6E), indicating that increased NKG2A/C/E staining on 2° effectors is caused largely by changes in the expression of NKG2A.

Administration of blocking antibody against NKG2A/C/E (20d5) on 4–6 dpi to mice that had received memory HNT cells (Fig. 7A) resulted in efficient in vivo blockade at 7 dpi as compared with administration of an isotype control (Fig. 7B). At 7 dpi, donor cells were enumerated and analyzed for cytokine production by ICCS. To help rule out an indirect effect of 20d5 treatment on CD4⁺ T-cell responses through the regulation of other cell types expressing NKG2A, we also analyzed 1° effector responses from recipients of naive HNT cells, which express reduced levels of NKG2A (Fig. 6E), that were treated with 20d5 or isotype antibody. The absolute numbers of 1° or 2° effectors present in the dLN and lung was not affected by 20d5 treatment (Fig. 7C). However, treatment did reduce the frequency and number of TNF⁺IFN-γ⁺ 2° effectors in both the dLN and lung (Fig. 7D) and significantly reduced the mean fluorescence intensity (MFI) of TNF in 2° effectors (Fig. 7 E and F). Treatment had no effect on IL-2 production by 2° effectors (Fig. 7E). Importantly, 20d5 treatment had no effect on cytokine production by 1° effectors (Fig. 7D); this observation supports a selective role for NKG2A signaling in regulating 2° but not 1° effector function in vivo.

Fig. 7.

NKG2A and Lag3 blockade affect the cytokine production potential of 1° and 2° effectors. Naive or in vitro-generated memory HNT cells (2 × 10⁶) were transferred to Thy-disparate hosts; then hosts were infected with 500 EID₅₀ PR8, and at 4–6 dpi mice were treated with isotype antibody or 20d5 as depicted (A). (B) Blockade of lung donor cell NKG2A/C/E expression. (C) The ratio of 1° and 2° effectors from the dLN and lungs at 7 dpi. n = 4 per group. (D) The number of IFN-γ⁺TNF⁺ effectors in the dLN and lung. n = 4 per group. Results are shown for one of two independent experiments. **P < 0.005. (E) Representative ICCS staining of dLN 2° effectors for IFN-γ⁺TNF⁺ and IFN-γ⁺IL-2⁺ cells. (F) Summary of 2° effector TNF⁺ MFI. **P < 0.005. (G) Mice received cells as above and were treated with isotype antibody or C9B7W at 4–6 dpi. (H) Blockade of Lag3 expression on 1° effectors. (I) Number of effectors. *P < 0.05. (J) The ratio of IFN-γ⁺IL-2⁺ (filled circles) and IFN-γ⁺TNF⁺ cells (open circles) in the dLN and lung with or without C9B7W at 7 dpi. (K) Representative ICCS staining of dLN 1° effectors. n = 5 per group. Results are shown for one of two independent experiments. All error bars represent the SD.

We used the same approach to investigate if targeting a molecule preferentially expressed by 1° effectors could affect their function selectively. We focused on Lag3 because of its role in modulating T-cell responses in vivo (26). Although Lag3 expression was higher on 1° than on 2° effectors, as determined by FACS (Fig. 4G), it was not expressed differentially in our microarray analysis. This result might reflect differential posttranscriptional regulation in 1° and 2° effectors, an hypothesis supported by the intracellular stores of Lag3 in activated CD4⁺ T cells and by its rapid and tightly controlled surface translocation (27).

Mice receiving naive or memory HNT cells were treated at 4–6 dpi with Lag3-blocking antibody (C9B7W) (Fig. 7G), resulting in efficient blockade at 7 dpi (Fig. 7H). Strikingly, C9B7W-treated mice contained higher absolute numbers of 1° effectors at 7 dpi (Fig. 7I), and roughly twofold higher numbers of IL-2⁺IFN-γ⁺ (Fig. 7 J and K) cells in the lung and dLN and a similar increase in TNF⁺IFN-γ⁺ cells (Fig. 7 J and K). In contrast, 2° effector responses were not affected by treatment (Fig. 7 I and J). These results confirm the functional relevance of these two molecules identified in our analyses and suggest that 1° and 2° effector responses are regulated differently. Not surprisingly, these treatments show only partial effects on modulating functional potential, suggesting that the enhanced properties of 2° effectors also are regulated by changes in genes other than NKG2A and Lag3.

Discussion

Overall, these studies lead to three striking conclusions. First, 2° CD4⁺ T-cell effectors are distinct from and are functionally superior to 1° effectors and mediate better protection against lethal IAV infection. The 2° effector response was temporally separable from helper activities mediated by memory cells, occurring after 5 dpi, when help already had been delivered. The loss of protection observed upon depletion of transferred memory CD4⁺ T cells, even though help was unchanged, suggests that the enhanced abilities of 2° effectors as compared with 1° effectors are a major component of the protection conferred by memory CD4⁺ T cells. Our results thus provide a basis for understanding the protection mediated by memory CD4⁺ T cells in B-cell–deficient, CD8⁺ T-cell–deficient, and even lymphocyte-deficient hosts challenged with IAV (11, 12). Second, we find that effectors in the spleen, dLN, and lung are strikingly different from one another, suggesting that they are specialized to perform unique functions at different sites. We also find that 2° effectors are more multifunctional, regardless of location, as evidenced by their enhanced production of multiple inflammatory cytokines and their wider distribution of T_FH cells, which are seen in the lung in the 2° but not the 1° effector response. This finding indicates that CD4⁺ T cells responding to pathogen challenge follow multiple, separate developmental patterns and raises the possibility that unique cues in distinct environments might coordinate effector specialization. Third, we demonstrate distinct regulation of 1° and 2° effector function in vivo. Our results underscore the unique character of 2° effectors but also suggest that the regulation of the enhanced function of 2° effectors is complex, involving multiple pathways.

It perhaps is unexpected that the array of cytokines produced by individual CD4⁺ T cells broadens instead of becoming more specialized with further exposure to antigen and with division. We had considered the possibility that 2° effectors might be more uniform among organs, because the process of epigenetic remodeling is thought to change future expression by increasing the ease with which certain cytokines are expressed and by silencing others (28). Instead, in agreement with studies addressing gene expression in repeatedly stimulated memory T cells (29), our results suggest that even well-polarized T_H1 memory cells making a restricted cytokine profile (30) can differentiate further to make multiple cytokines at higher levels. These findings suggest that the transition to the resting memory state may reset certain aspects of a CD4⁺ T cell’s response potential, a hypothesis supported by the nearly identical gene expression in naive and memory cells (17).

Our analysis of 1° and 2° effectors recovered from different organs revealed striking similarities, as well as differences, in their transcriptomes. For example, 1° and 2° effectors isolated from the lung both expressed high levels of genes associated with antiviral responses, such as IFN-γ, and numerous chemokines and chemokine receptors. Our data are consistent with the hypothesis that effectors in the lung are important for protection against IAV through direct mediation of viral clearance. How CD4⁺ T-cell effectors combat IAV is not fully understood, but our studies suggest that individual protective mechanisms, including perforin-dependent killing of infected cells and production of IFN-γ, become more or less important depending on the context of infection (8, 12). In contrast, the expression of T_FH-associated genes in 1° effectors was restricted to SLO, but 2° effectors contained substantial T_FH populations in all organs tested. It is interesting to speculate that T_FH activity from 2° effectors in the lung during recall responses against IAV might be important in providing efficient help for local B-cell antibody production when inducible bronchus-associated lymphoid tissue is present (31). The dramatic distinctions seen at the gene-expression level among effectors in lung, spleen, and dLN suggest that the distribution of the T_FH subset in 1° effectors likely represents only one set of organ-specific distinctions and that other subsets also might be found preferentially in different sites. Indeed we find that CD4⁺ T cells with cytotoxic activity generated by IAV are restricted almost exclusively to the lung (10). Thus, correlates of protective T cells may differ in different organs as a result of the simultaneous generation of effector subsets specialized for different roles at different sites. Moreover the depletion studies indicate that the helper functions occur earlier in SLO, whereas the peak of effector responses, which correlates with viral clearance, occurs later in the lung. Thus, a temporal distinction also must be considered in identifying correlates of protection.

Perhaps it is not surprising that we have identified only a relatively small number of genes, about 450, that are expressed differentially in 1° and 2° effectors. This cohort of genes includes many that are involved in regulating apoptosis, signaling pathways, translation, cell migration, chemokine signaling, and metabolism. For example, 2° effectors expressed lower levels of genes associated with the suppression of DNA replication (setd8) (32) and dampening of cellular proliferation (lrig3) (33), as is consistent with the increased magnitude of 2° effector responses. Similarly, the higher expression in 1° effectors of genes such as satb2, a negative regulator of CD127 expression (34), and pcdh10, which can potentiate apoptosis (35), also is consistent with increased numbers of 2° effectors at the site of infection. In addition, 2° effectors express higher levels of gpr12, gpr63, rgs3, and tiam1, all of which are associated with chemotaxis and migration (36–38) and also might be involved in the greater accumulation of 2° effectors in the lung. Other genes, such as eif4gI [a translation initiation factor (39)], araf [a potential modulator of TcR signaling (40)], and rab18 [involved in the ER secretory pathway (41)] also could regulate or mediate aspects of the superior functional capabilities of 2° effectors. Further studies will be needed to evaluate how each of these pathways affects the generation, migration, and function of CD4⁺ T-cell effectors.

Despite these indications that multiple genes contribute to the more potent 2° effector response, we found that blocking NKG2A on 2° effectors decreased the number of cells that secrete both TNF and IFN-γ, but the same treatment had little effect on 1° effector responses. Similarly, blockade of Lag3 specifically enhanced cytokine production by 1° effectors but not by 2° effectors. Although these studies cannot formally rule out indirect effects of treatment through other cellular populations expressing NKG2A or Lag3 on CD4⁺ T-cell function, the comprehensive analysis presented here provides insights into the differential regulation 1° and 2° effector responses and provides a compelling integrated and unbiased picture of organ-specific differences in T-cell effector function. We suggest that the differences between 1° and 2° effectors identified here bear further investigation to determine which genes are most important for the superior protection mediated by memory CD4⁺ T cells. Defining the functionally relevant molecules and pathways will allow the definition of mechanisms that contribute to the superior efficacy of T-cell memory. In addition, the genes we identified as being differentially regulated in 1° and 2° effectors are good candidates for targets that might be manipulated to increase the potency of T-cell effectors in IAV and also in other pathogen and therapeutic settings.

Our studies also provide an integrated and unbiased snapshot of organ-specific differences in T-cell effector function that deserve further analysis. How this specialization is achieved is not yet known. It will be interesting to investigate whether some of the determination occurs because of organ-specific microenvironments, cells, and factors and how much is a result of the selective recruitment of predestined subsets that develop in the same peripheral location.

Often, the most important contribution from memory CD4⁺ T cells during recall challenge is the provision of help, leading to accelerated B-cell and CD8⁺ T-cell responses (42). However, under certain circumstances, such as those presented here, the critical protective contribution from memory CD4⁺ T cells results from the action of 2° effectors. Thus, the further definition of the protective contributions from 2° effectors and the pathways responsible for their efficacy could provide important new correlates of vaccine-induced protection against important human pathogens.

Materials and Methods

Mice.

Naive CD4⁺ T cells were obtained from 5- to 8-wk-old HNT.Thy1.1/Thy1.2 mice on a BALB/c background recognizing amino acids 126–138 (HNTNGVTAACSHE) of PR8 HA (43). Recipients of cell transfers were BALB/c or BALB/c.Thy1.1 mice that were at least 8 wk old. Mice were obtained from the breeding facility at Trudeau Institute or the University of Massachusetts Medical School. All experimental animal procedures were conducted in accordance with the Trudeau Institute or the University of Massachusetts Medical School Animal Care and Use Committee guidelines.

Naive CD4⁺ T-Cell Isolation, Memory Generation, and Adoptive Transfer.

Naive CD4⁺ T cells were obtained from pooled spleen and lymph nodes as previously described (7). Resulting cells were routinely >97% Vβ8.3⁺ and expressed a characteristic naive phenotype (small size, CD62L^hi, CD44^lo, and CD25^lo). In some experiments, CD4⁺ T cells were CFSE labeled, as previously described (44).

T_H1-polarized memory CD4⁺ T cells were generated in vitro as previously described (17). In vivo PR8-primed memory cells were generated and reisolated as described (19) by transferring naive HNT cells to nude hosts that then were infected with PR8 and allowed to recover at least 30 d before reisolation of donor CD4⁺ T cells from SLO and lung. Similarly, polyclonal memory CD4⁺ T cells were isolated from SLO and lungs of mice that had been primed with PR8 at least 30 d previously.

All donor CD4⁺ T cells were adoptively transferred in 200 μL PBS by i.v. injection. In some experiments mice were treated i.p. with 1 mg of either anti–Thy1.2-depleting antibody (30-H12) (Bio X Cell) or with an isotype control. In further experiments, mice were treated i.p. as indicated with 0.5 mg of Rat IgG2a (20d5; eBioscience) directed against NKG2A/C/E or 0.5 mg of Rat IgG1 (C9B7W; Bio X Cell) or with the appropriate isotype controls.

Virus and Infections.

PR8 virus was produced in the allantoic cavity of embryonated hen eggs from virus stocks originating at St. Jude Children’s Hospital. Mice were infected intranasally under light isoflurane anesthesia (Webster Veterinary Supply) with stated doses of virus in 50 μL PBS (500 EID₅₀ = 0.1 LD₅₀ and 10,000 EID₅₀ = 2 LD₅₀). Viral infection was performed on the same day as cell transfer. Viral titer was assessed by PA copy number, and PR8-specific serum IgG titers were assessed as previously described (45).

Tissue Preparation, Flow Cytometry, and Microarray Data Analysis.

Mice were euthanized at different time points after virus infection, and lungs were perfused by injecting 10 mL of PBS into the left ventricle of the heart. Lungs, spleen, and dLN were prepared into single-cell suspensions by mechanical disruption of organs and passage through a nylon membrane. Cell suspensions were washed, resuspended in FACS buffer (PBS plus 0.5% BSA and 0.02% NaN₃), and incubated on ice with 1 μg anti-FcR (2.4G2) followed by saturating concentrations of fluorochrome-labeled antibodies. Further details can be found in SI Materials and Methods. Intracellular staining for cytokine expression was performed as previously described (7). FACS analysis was performed using a FACS Scan or LSR II (BD Biosciences) and FlowJo (Tree Star) software.

1° and 2° effectors that had undergone at least five divisions by day 7 postinfection were sort-purified from the spleen, dLN, and lung of recipients, and total mRNA was isolated (Qiagen) for microarray analysis. Further details can be found in SI Materials and Methods.

Statistical Analysis.

Unpaired, two-tailed Student’s t tests, ∝ = 0.05, were used to assess whether the means of two normally distributed groups differed significantly. The Welch correction was applied when variances were found to differ. One-way ANOVA with Bonferroni’s multiple comparison posttest was used to compare multiple means.