Graded levels of IRF4 regulate CD8⁺ T cell differentiation and expansion, but not attrition, in response to acute virus infection¹

Mice

Mice were housed in specific pathogen-free conditions at UMMS in accordance with IACUC guidelines. Irf4^fl/flxCD4-Cre and OT-IxRag1^−/− Irf4^fl/fl CD4-Cre have been previously described (26, 27). P14xTCRα^−/− were purchased from Taconic Farms (Germantown, New York). Irf4^+/+xCD4-Cre, Irf4^+/fl and Irf4^+/+ were used as WT controls.

Antibodies, H2D^b and H2K^b monomers and Staining

CD45.2-V500 and TNFα-APC-Cy7 were purchased from BD Biosciences (San Jose, California). KLRG1-FITC, Eomes-PE, CD107a-PE, CD107b-PE, CD27-PE, CD127-PE-Cy5, CD127-PerCP-Cy5.5, Tbet-PerCP-Cy5.5, IFNγ-PerCP-Cy5.5, Eomes-PerCP-efluor710, CD45.1-PECy7, KLRG1-PE-Cy7, Tbet-PE-Cy7, IRF4-AlexaFluor647, CD44-AlexaFluor700, CD62L-APC-eFluor780, CD44-eFluor450, KLRG1-eFluor450, IFNγ-eFluor450, CD90.2-APC-eFluor780, CD45.1-APC-eFluor780, IL-2-PerCP-Cy5.5 were purchased from eBioscience (San Deigo, California). CD8-PE-TexasRed, GranzymeB-PE, GranzymeB-APC, Live-Dead-Violet, Live-Dead-Aqua and goat-anti-rabbit IgG-AlexaFluor647 and -AlexaFluor488 were purchased from Life Technologies (Grand Island, New York). H2D^b-GP33 monomers were prepared at UMMS; LCMV-specific (H2D^b-NP396, H2D^b-GP276) and Influenza A PR8-OVA_I-specific (H2K^b-OVA257) monomers were obtained from the NIH Tetramer Core Facility (Atlanta, Georgia). Intracellular TCF1 staining was performed using rabbit-anti-mouse TCF1 (Cell Signaling Technology, Danvers, Massachusetts) followed by staining with goat-anti-rabbit secondary (Life Technologies). Samples were analyzed on an LSRII flow cytometer (Becton Dickinson), and data were analyzed using FlowJo (Tree Star).

Cell Culture

Lymph node cells from P14 WT and P14 Irf4^+/fl mice were mixed with equal numbers of WT CD45.1 splenocytes and stimulated with GP33 or F6L peptides for 24, 48 and 72 hr. Cells were harvested and analyzed for IRF4, Eomes, and TCF1 expression by intracellular staining. For cytokine production, splenocytes from infected mice were stimulated with GP33, GP276 and NP396 peptide for 5hr in the presence of 1μg/ml Golgi Stop and 1μg/ml Golgi Plug, and antibodies to CD107a and CD107b.

Viruses, infections and adoptive transfers

For virus infections, LCMV-Armstrong GP33 and F6L variants were injected intraperitoneally (IP) at 5×10⁴ PFU, unless otherwise specified. For adoptive transfers, splenocytes from P14 WT CD45.1⁺CD45.2⁺, P14 Irf4^+/fl CD45.2⁺, OT-I WT CD45.1⁺ or OT-I Irf4^+/fl CD45.2⁺ mice were stained with antibodies to CD8 and Vα2 to determine the proportions of P14 or OT I cells, and equal numbers of WT and Irf4^+/fl cells were mixed. 2,000, 20,000 or 1,000,0000 total P14 cells were transferred intravenously (IV) into WT or CD45.1⁺ hosts one day prior to infection. 6,000 total OT-I cells were transferred IV into CD90.1 hosts and infected with O.3 LD50 of influenza A PR8-OVA_I.

Plaque assay

Spleens were harvested at D8p.i., homogenized in media and stored at −80°C. Plaque assays were performed as previously described (28).

Statistical Analysis

All data are represented as mean±SEM. Statistical significance is indicated by ns (p>0.05), * (p≤0.05), ** (p≤0.01), *** (p≤0.001), **** (p≤0.0001) based on unpaired student T test.

The strength of TCR signaling regulates the levels and duration of transcription factor expression

The expression of IRF4 is upregulated in naïve T cells by TCR signaling (14). This response is dependent on the activation of the Tec kinase Itk (26). To determine if the levels of IRF4 were affected by the strength of TCR signaling to stimulation by natural ligands, P14 TCR transgenic TCRα^−/− (hereafter referred to as P14 WT) CD8⁺ T cells (29) were stimulated in vitro, and IRF4 levels were examined by intracellular staining. The P14 TCR recognizes the GP_33–41 epitope (GP33) of LCMV bound to H2-D^b. A single amino acid substitution from phenylalanine to leucine at position six generates a lower affinity peptide ligand, F6L (30). F6L–H2-D^b complexes display approximately 5-fold reduction in the equilibrium dissociation constant (K_D) for binding to the P14 TCR, and 100-1,000-fold reduction in functional avidity.

P14 T cells were stimulated with the high affinity GP33 peptide and the lower affinity F6L variant. At 24h, both populations of cells expressed similar amounts of IRF4. However, high IRF4 expression was sustained at 48h and 72h post-stimulation in cells stimulated with GP33, whereas cells stimulated with F6L showed declining IRF4 as early as 48h post-activation (Fig. 1A). Histograms showing IRF4 staining on IRF4-deficient P14 T cells (Irf4^fl/fl) stimulated with the GP33 peptide are included as negative controls. Alterations in peptide dose also impacted IRF4 expression. As shown, higher doses of GP33 peptide (1μM and 100nM) induced strong IRF4 expression at 24h and 48h post-stimulation relative to the 10nM stimulation condition (Fig. 1B). By 72h, differences in IRF4 levels were observed between each of the peptide doses, with the highest peptide dose leading to the most sustained IRF4 expression (Fig. 1B). These data indicate that expression of IRF4 is transient, and is regulated by the strength of TCR stimulation.

In CD8⁺ T cells, IRF4 negatively regulates the expression of the transcription factor, Eomesodermin (Eomes), that is required for the maintenance of memory cells post-infection (11, 26). As shown in Fig 1C, stimulation with the lower affinity F6L peptide resulted in higher Eomes expression, correlating with their reduced IRF4 expression. Similar results were seen with diminishing doses of GP33 (Fig. 1D). Eomes expression in CD8⁺ T cells is positively regulated by the transcription factor, TCF1 (10). As shown in Fig. 1E, stimulation with GP33 or F6L peptide resulted in similar TCF1 expression 24h post-activation; however, at later timepoints, TCF1 remained highest in cells stimulated with the lower affinity F6L ligand. A similar pattern was seen with the lowest dose of GP33 (Fig. 1F). Together, these data demonstrate that varying TCR signal strength, either by changes in TCR-MHC/peptide affinity or dose, leads to distinct expression patterns of three key transcription factors in CD8⁺ T cells.

To determine whether IRF4 regulated the expression of Eomes and/or TCF1, we utilized P14 T cells with one or two alleles of Irf4 deleted (Irf4^+/fl x CD4-Cre and Irf4^fl/fl x CD4-Cre, referred to as Irf4^+/fl and Irf4^fl/fl, respectively). For these studies, P14 WT, P14 Irf4^+/fl and P14 Irf4^fl/fl T cells were stimulated in vitro with GP33. As expected, WT cells expressed the highest levels of IRF4 while Irf4^+/fl cells expressed intermediate levels of IRF4 relative to Irf4^fl/fl and WT cells 24 and 48h timepoints (Fig 2A); furthermore, this pattern of expression showed a striking similarity to that seen following stimulation of WT P14 T cells with the lower affinity F6L ligand or with lower doses of GP33 peptide (Compare Fig 2A with Figs 1A and 1B). Eomes expression inversely correlated with IRF4 levels; P14 Irf4^fl/fl cells expressed the highest levels of Eomes, with Irf4^+/fl T cells expressing intermediate levels of Eomes compared to WT cells (Fig. 2B). TCF1 expression was elevated in Irf4^fl/fl and Irf4^+/fl cells at the 72h timepoint relative to the WT samples with Irf4^+/fl cells expressing the highest levels of TCF1 (Fig. 2C). These data indicate that a complete or heterozygous deficiency in Irf4 leads to lower expression of IRF4, and in turn, this alteration changes the expression patterns of Eomes and TCF1 in stimulated CD8⁺ T cells.

A heterozygous deficiency in Irf4 reduces virus-specific CD8⁺ T cell clonal expansion

To elucidate the role of graded IRF4 expression during polyclonal CD8⁺ T cell differentiation in vivo, WT, Irf4^+/fl and Irf4^fl/fl mice were infected with LCMV-Armstrong. Responses to three LCMV epitopes i.e. H2-D^b/GP_33–41, H2- D^b/NP_396–404, H2- D^b/GP_276–396 (hereafter referred to as GP33, NP396 and GP276) were examined using MHC-peptide tetramers. At the peak of the response, i.e. day 8 (D8) post-infection (p.i.), the magnitude of the CD8⁺ T cell response depended on the gene dosage of Irf4. WT CD8⁺ T cells mounted the most robust response, followed by Irf4^+/fl, and then Irf4^fl/fl cells (Fig. 3A and S1A and S1B). Enumeration of viral titers by plaque assay indicated that all WT (5/5) and Irf4^+/fl (3/3) mice had cleared the virus, while only 30% of Irf4^fl/fl (3/10) mice had cleared LCMV by D8 p.i. These data indicated that modest reductions in IRF4 expression did not interfere with viral clearance, but that a minimal level of IRF4 was required for sterilizing immunity to LCMV (Fig. 3B). As our in vitro studies showed reduced IRF4 expression levels following activation of WT versus Irf4^+/fl P14 T cells, these initial infection experiments indicated that even modest differences in the magnitude of IRF4 expression had a profound effect on the clonal expansion of virus-specific CD8⁺ T cells.

Reduced levels of IRF4 do not affect the kinetics of CD8⁺ T expansion or attrition

The reduced numbers of virus specific Irf4^+/fl cells at D8 p.i. could be attributed to delayed kinetics of Irf4^+/fl CD8⁺ T cell expansion relative to the WT cells. Based on previous data indicating a regulatory T cell defect in Irf4^fl/fl mice that disrupts normal T cell homeostasis (24, 26), studies of Irf4^fl/fl mice were not included in the subsequent analyses. Instead, we focused on a comparison of WT versus Irf4^+/fl T cell responses, to understand the effect of partial loss of IRF4 expression on CD8⁺ T cell differentiation. Examination of WT and Irf4^+/fl CD8⁺ T cell populations at later time points following LCMV-Armstrong infection indicated that the peak response for both WT and Irf4^+/fl CD8⁺ T cells was at D8 p.i; by D10 p.i., both populations had started to contract (Fig. 3C and S1C and S1D). Following antigen clearance, the majority of CD8⁺ T cells undergo attrition by apoptosis, and form a small but stable pool of memory cells (1, 2). Examination of virus-specific CD8⁺ T cells at D14 and D28 p.i. confirmed that this pattern was observed for both WT and Irf4^+/fl CD8⁺ T cells. However, by D28 p.i., the differences in the numbers of WT and the Irf4^+/fl cells were quite modest, and no longer significant for 1 of the 3 epitopes examined (Fig. 3C and S1C and S1D). Normalization of virus-specific CD8⁺ T cell numbers to the peak of the response indicated that loss of one allele of Irf4 did not change the kinetics of the CD8⁺ T response, nor did it affect the rate of CD8⁺ T cell contraction. Thus, in comparison to the numbers of virus-specific T cells present at the peak of the response, greater numbers of WT CD8⁺ T cells were lost between D8 and D28 p.i. relative to Irf4^+/fl CD8⁺ T cells (Fig. 3C and S1C and S1D). Therefore, not only did WT cells undergo more robust expansion than the Irf4^+/fl CD8⁺ T cells, the WT cells also underwent more extensive contraction.

We also examined the CD4⁺ T cell response to LCMV-Armstrong in infected WT and Irf4^+/fl mice. Similar to our findings for CD8⁺ T cells, analysis of GP61 epitope-specific CD4⁺ T cells at D14 and D28 p.i. indicated a defect in Irf4^+/fl CD4⁺ T cell expansion relative to the WT cells (Fig. 3D). Further, consistent with our analysis of the CD8⁺ T cell response, we found that greater numbers of WT GP61-epitope specific CD4⁺ T cells were lost between D14 and D28 p.i. relative to Irf4^+/fl CD4⁺ T cells (Fig. 3E).

To generalize these findings to a distinct infection model, we performed a second series of studies examining CD8⁺ T cell responses to the PR8-OVA_I strain of Influenza A. At D8 p.i., both WT and Irf4^+/fl mice mounted robust OVA_257–264-specific poly-clonal CD8⁺ T cell responses. As with LCMV, Irf4^+/fl CD8⁺ T cells showed a diminished response relative to WT cells. This reduction was observed in the draining mediastinal lymph node (DLN) as well as in the lungs of infected mice (Fig 4A, 4B). By D28 p.i., OVA_257–264-specific CD8⁺ T cell populations had undergone attrition in the DLN and lungs of both groups of mice; further, at this timepoint, no significant differences in the numbers of cells were observed when comparing WT and Irf4^+/fl mice. Consistent with this, we found that greater numbers of OVA-specific CD8⁺ T cell were lost between D8 and D28 p.i. in both DLN and lungs of WT relative to Irf4^+/fl mice following infection with PR8-OVA_I.

Together, these data suggest that the effects of reduced IRF4 expression are a general feature of CD8⁺ T cell responses to viral infections, and further, are impacting CD4⁺ T cell responses as well. We conclude that different amounts of IRF4 expression during T cell priming regulate the magnitude of the peak antiviral T cell response without affecting the kinetics of the response, or the rate of attrition following antigen clearance.

Reduced gene dosage of Irf4 regulates effector cytokine expression

To assess CD8⁺ T cell effector functions following virus infection, splenocytes from LCMV-Armstrong infected WT, Irf4^+/fl and Irf4^fl/fl mice were examined at D8 and D28 p.i. for IFNγ, TNFα and IL-2 expression. As expected, the gene dosage of Irf4 strongly correlated with the numbers of IFNγ-producing CD8⁺ T cells (Fig. 5, S2A and S2B). No differences in the frequencies of TNFα producing CD8⁺ T cells as a proportion of IFNγ⁺ CD8⁺ T cells were observed at D8 p.i. when comparing WT and Irf4^+/fl mice; however, Irf4^fl/fl mice showed a substantial reduction in the relative proportion of this double cytokine-producing subset. Furthermore, at D28 p.i, the median fluorescence intensity (MFI) of TNFα staining, and the frequencies of TNFα/IFNγ double-producers and IFNγ/TNFα/IL-2 triple-producers were significantly decreased in Irf4^+/fl mice compared to WT controls. Analyses of granzyme B expression and degranulation as assessed by CD107a and CD107b staining revealed no differences between any of the genotypes at either timepoint. Overall, these data indicate that reduced expression of IRF4 leads to an impaired ability of virus-specific CD8⁺ T cells to produce cytokines other than IFNγ as the cells transition into a long-term memory population.

Levels of IRF4 expression selectively impact the short-lived CD8⁺ effector cell population

In response to acute infections, CD8⁺ T cells undergo clonal expansion and differentiation to short-lived effector cells (SLEC, KLRG1^hiCD127^lo) and memory-precursor effector cells (MPEC, KLRG1^loCD127^hi). Examination of these populations revealed that reduced IRF4 expression had a more substantial impact on the numbers of virus-specific SLEC compared to MPEC for each population (Fig. 6A and S3A and S4A). Specifically, at D8 p.i., Irf4^+/fl mice had a 2.5–4.4 fold reduction in numbers of SLEC versus a 1.4–2.0 fold reduction in MPEC numbers over the three epitopes examined. Furthermore, in spite of the increase in MPEC percentages among virus-specific CD8⁺ T cells in Irf4^+/fl mice at early times after infection (D8, 10, and 14 p.i.), the absolute numbers of MPECs in these mice were decreased. Since the numbers of SLEC are much greater than the numbers of MPEC, these data indicate that diminished SLEC populations are largely responsible for the decrease in the total magnitude of the CD8⁺ T cell effector response in Irf4^+/fl mice. Interestingly, by D28 p.i., WT and Irf4^+/fl mice had comparable numbers of virus-specific CD8⁺ T cells, and no significant differences in the numbers of MPEC were observed for two of the three viral epitopes examined. Consistent with these data, examination of virus-specific effector (T_EM) and central (T_CM) memory populations at D28 p.i. showed only a modest reduction in the numbers of T_EM in Irf4^+/fl compared to WT mice, and no differences in the numbers of T_CM cells between the two genotypes (Fig. 6B and S3B and S4B). These data indicate that reduced IRF4 expression is regulating the expansion phase of the T cell response, but not impacting the generation of long-lived virus-specific CD8⁺ T cells.

Differential role for IRF4 in regulating TCF1, Eomes and T-bet expression

Virus-specific CD8⁺ T cell differentiation is regulated by the expression of transcription factors such as TCF1, Eomes, and T-bet at D8 p.i (8, 10–12, 31). Consistent with our in vitro stimulation data, LCMV-specific Irf4^fl/fl cells expressed the highest levels of TCF1 and Eomes, Irf4^+/fl cells had intermediate levels, and WT cells had the lowest levels of both factors. In contrast, the expression of T-bet was reduced in Irf4^fl/fl cells compared to WT, but no differences were observed in T-bet levels when comparing WT and Irf4^+/fl cells (Fig 6C, S3C and S4C). These data indicated that TCF1 and Eomes expression were more sensitive to modest changes in IRF4 levels than was T-bet expression, indicating an IRF4 dose-dependent variation in the regulation of these key transcription factors. At later timepoints post-infection, differences in Eomes and TCF1 levels in virus-specific Irf4^+/fl versus WT cells were less uniform across the three epitope-specific populations, although in general, Irf4^+/fl cells tended to express higher levels of these factors than WT cells at D10 and 14 p.i; however, by D28 p.i., no further differences were observed between Irf4^+/fl and WT cells. Consistent with the transient nature of IRF4 expression and the data presented above, these results confirm that variations in IRF4 expression levels have the greatest impact at the peak of the virus-specific CD8⁺ T cell response, and are not generally altering the long-lived population of virus-specific CD8⁺ T cells found at D28 p.i.

Cell-intrinsic role for IRF4 in regulating the magnitude of the CD8⁺ effector T cell response

To assess whether the altered virus-specific CD8⁺ T cell response seen in Irf4^+/fl versus WT mice was due to differences intrinsic to the CD8⁺ T cells, we performed adoptive transfer experiments. This approach also allowed us to examine whether activation of P14 T cells in vivo with an LCMV variant expressing the lower affinity F6L ligand would phenocopy the results of reducing IRF4 expression by a heterozygous deficiency in the Irf4 gene. We first established that activation of P14 cells with LCMV expressing the GP33 epitope results in higher IRF4 expression relative to P14 cells activated in response to LCMV-F6L infection. One million congenically-marked WT P14 (CD45.1⁺ CD45.2⁺) cells were transferred into naïve WT (CD45.2⁺) hosts. One day later, mice were infected with 1×10⁵ PFU LCMV-Armstrong expressing either the WT GP33 epitope (LCMV-GP33) or the mutant F6L epitope (LCMV-F6L). Previous studies showed that the single amino acid substitution in the LCMV-F6L virus has no impact on viral replication or viral clearance when compared to LCMV-GP33 (30). As expected, at D3 p.i., P14 cells activated with LCMV-GP33 expressed higher levels of IRF4 relative to P14 cells activated in response to the F6L epitope (Fig 7A). These data were consistent with the results seen upon in-vitro stimulation of WT P14 cells with the GP33 and F6L ligands (Fig 1A).

To assess the cell intrinsic role of reduced TCR stimulation, either alone or in combination with reduced IRF4 expression, in regulating the CD8⁺ T cell response, congenically-marked WT P14 (CD45.1⁺ CD45.2⁺) and Irf4^+/fl P14 (CD45.2⁺) cells were mixed 1:1 and transferred into naïve WT (CD45.1⁺) hosts (Fig 7B). When analyzed at D8 p.i., we observed a substantially greater proportion of WT P14 cells relative to the Irf4^+/fl P14 population, indicating differential expansion of the two populations (Fig 7C, D). This trend was observed when a total of 2,000 or 20,000 P14 cells were transferred as a 1:1 mixture of the two genotypes. Strikingly, infection with the LCMV-F6L virus reduced the overall expansion of both WT and Irf4^+/fl P14 cells, but maintained the competitive advantage of the WT over the Irf4^+/fl cells (Fig 7D). These findings indicate that the decreased expansion of P14 cells in response to LCMV-F6L relative to LCMV-GP33 is highly correlated with decreased IRF4 expression. Thus, these data suggest that variable upregulation of IRF4 in CD8⁺ T cells is responsible for the effect of TCR signal strength on the magnitude of the peak effector response, as reported (3).

More detailed analyses of the P14 populations were performed with mice receiving 20,000 transferred P14 cells prior to infection, as this provided a greater number of cells for analysis. Examination of SLEC versus MPEC subsets among the P14 populations indicated that the graded magnitude of the response seen among the four experimental groups could be largely attributed to differences in the expansion of P14 SLEC (Fig. 8A). This was affected by changes in IRF4 expression due to deletion of one Irf4 allele, or by infecting mice with an LCMV variant expressing only a lower affinity ligand for the P14 TCR, or both. In contrast, the MPEC populations were largely unaffected by loss of one functional Irf4 allele. Regardless of the virus used, Irf4^+/fl P14 cells expressed higher total levels TCF1 due to the increased proportions of TCF1⁺ cells among the Irf4^+/fl P14 populations (Fig. 8B). However, similar to our observations regarding the MPEC population, the absolute numbers of TCF1⁺ cells were not affected by a heterozygous deficiency in Irf4 (Fig. 8B). This is in contrast to a recent report showing a ~2-fold decrease in tcf7 mRNA in in-vitro activated Irf4-deficient OT-I CD8⁺ T cells (32). This discrepancy could arise from differential priming of CD8⁺ T cells in vivo in response to a virus infection versus that occurring when T cells are activated in-vitro; alternatively, the differences seen might arise from differences in the timepoints examined in our ex vivo analysis versus that used for the in vitro stimulation studies. Nonetheless, consistent with previous data (10), we found that higher TCF1 expression in Irf4^+/fl P14 cells correlated with enhanced Eomes expression in these cells (Fig. 8C). In contrast, we observed no differences in the expression of T-bet when comparing WT to Irf4^+/fl P14 cells (Fig. 8C).

The co-stimulatory molecule CD27 is important for survival of CD8⁺ effector and memory cells (33); interestingly, we found increased expression of CD27 on Irf4^+/fl P14 cells than WT P14 cells (Fig. 8C). This finding is consistent with our observation that, following the peak of the antiviral response, Irf4^+/fl cells undergo less cell death than their WT counterparts. To address this possibility further, we examined Bcl2 expression. As shown, the loss of one allele of Irf4 and/or infection with LCMV-F6L resulted in enhanced expression of the pro-survival molecule Bcl2 in P14 cells (Fig. 8D). This increase could be mainly attributed to higher Bcl2 expression in the Irf4^+/fl versus the WT P14 MPEC population (Fig. 8D). Together, these data are consistent with the greater survival potential of Irf4^+/fl P14 cells relative to WT, and provide an explanation for the reduced numerical attrition of virus-specific Irf4^+/fl CD8⁺ T cells observed (Figures 3C, S1C, S1D and 4A and 4B). Finally, these data also indicate that the differences in the clonal expansion and differentiation of virus specific effector CD8⁺ T cells observed upon infection of intact WT and Irf4^+/fl mice were due to a CD8⁺ T cell-intrinsic requirement for high levels of IRF4; further, these differences mainly arose from the variable expansion of the short-lived effector cells.

We next confirmed the cell intrinsic role of IRF4 in CD8⁺ T cell expansion using the Influenza A infection system. WT and Irf4^+/fl OT-I CD8⁺ T cells were mixed 1:1 and transferred into WT hosts, which were then infected with the PR8-OVA_I strain of Influenza A that expresses the epitope recognized by OT-I cells. Analysis of OT-I populations at the peak of the response (D8 p.i.) indicated a defect in the ability of Irf4^+/fl OT -I cells to clonally expand relative to WT OT-I populations in the spleen, draining mediastinal lymph node (DLN) and lung (Fig 9A). These data are in agreement with the findings from analysis of polyclonal OVA-specific CD8⁺ T cell responses in intact WT and Irf4^+/fl mice (Fig 4A and B), indicating the CD8⁺ T cell-intrinsic nature of this effect. We also observed that the expression of TCF1 and Eomes were higher in Irf4^+/fl OT-I cells compared to WT OT-I cells in the two lymphoid organs examined, whereas little difference was observed when comparing the two OT-I populations in the lung (Fig 9B). Interestingly, unlike our findings with acute LCMV infection, Influenza A infection revealed a reduction in T-bet levels in Irf4^+/fl OT-I cells compared to WT OT-I cells in the spleen and DLN (Fig 9B). The differences in T-bet expression seen following Influenza A, but not LCMV, infection might be due to a difference in the cytokine milieu between these two virus infections. Whereas high levels of IL-12 are induced during acute Influenza infection, LCMV-Armstrong infection induces little IL-12 (34, 35). As IL-12 is a potent inducer of T-bet expression (36), the presence of IL-12 might reveal a requirement for high levels of IRF4 to achieve the maximum expression of T-bet. In addition, these data confirm the critical role of high IRF4 expression in regulating the magnitude of the effector CD8⁺ T cell response in lymphoid organs, but indicate a lesser impact on the CD8⁺ T cells present at the site of infection in a non-lymphoid peripheral tissue.

Differential T cell expansion is driven by variations in the levels of IRF4 expressed in competing T cell populations

The data described above establish a CD8⁺ T cell-intrinsic requirement for IRF4. During polyclonal T cell responses, antigen-specific T cells compete for antigen and inflammatory cytokines. To test whether variations in IRF4 expression levels between competing T cell populations could result in variable T cell expansion, we transferred WT P14 T cells into either WT or Irf4^+/fl host mice, and infected with LCMV-GP33 or LCMV-F6L virus. Our findings thus far predicted that, in the case where WT P14 cells were responding in a host where all endogenous T cells are Irf4^+/fl, the WT P14 cells should show an enhanced response relative to their response in a WT host environment. In contrast, we reasoned that the WT P14 cells should show an impaired response following infection with LCMV-F6L in a WT host; however, we predicted that this response should improve if the WT P14 cells are transferred into Irf4^+/fl hosts, thereby providing the WT P14 cells with an advantage based on two functional alleles of Irf4.

As shown in Fig 10A, the data from these experiments supported our predictions. A single population of WT P14 T cells was transferred into either WT or Irf4^+/fl hosts, which were then infected with either LCMV-GP33 or LCMV-F6L. We found that WT P14 cells contributed more dominantly to the response to LCMV-GP33 when the endogenous T cell population was Irf4^+/fl cells, in contrast to their contribution when responding in a WT host. Alternatively, the WT P14 cells contributed little to the response to LCMV-F6L when present in a WT host, but this response could be greatly improved by transferring these cells into an Irf4^+/fl host, where the endogenous T cell response was handicapped in IRF4 expression. These data indicate that the contribution of an individual virus-specific CD8⁺ T cell population to the overall response is not simply regulated by the levels of IRF4 expressed in those cells, but is also determined by the levels of IRF4 expressed in competing T cell populations present in the same individual.

Based on these data, we speculated that the CD8⁺ T cells forming the most robust response to an acute infection would express the highest levels of IRF4, thus accounting for the predominance of these populations at the peak of the response. In acute LCMV infections, CD8⁺ T cell responses to GP33 and NP396 epitopes are immuno-dominant while those to GP276 are subdominant (37). Since the expression of IRF4 is transient in nature, we were unable to detect differences in IRF4 expression in these populations during the polyclonal response to LCMV-Armstrong. However, we did observe that the magnitude and kinetics of upregulation of TCF1, a target repressed by IRF4, were different between the epitope-specific populations. GP276-specific cells expressed the highest levels of TCF1, followed by GP33-specific cells, and then NP396-specific cells, suggesting lower IRF4 expression in GP276-specific cells and the highest expression in NP396-specific cells (Fig 9B). Since TCF1 also regulates the differentiation of Tem to Tcm (10), these data provide further mechanistic support for the observation that GP276-specific CD8⁺ T cells are the earliest subset to form Tcm in response to LCMV-Armstrong, while NP396-specific T cell population are the slowest (38). These results show that variations in IRF4 expression can modulate the relative proportions of different virus-specific T cell populations recognizing the same epitope, and suggest that differences observed in the expansion of polyclonal T cell responses to different epitopes may be driven by differential upregulation of IRF4.