CD8+ T cells play a critical role in adaptive immunity, differentiating into CD8+ memory T cells that form the basis of protective cellular immunity. Vaccine efficacy is attributed to long-term protective immunity, and understanding the parameters that regulate development of CD8+ T cells is critical to the design of T cell–mediated vaccines. We show in this study using mouse models that two distinct parameters, TCR signal strength (regulated by the tyrosine kinase ITK) and Ag affinity, play important but separate roles in modulating the development of memory CD8+ T cells. Unexpectedly, our data reveal that reducing TCR signal strength along with reducing Ag affinity for the TCR leads to enhanced and accelerated development of CD8+ memory T cells. Additionally, TCR signal strength is able to regulate CD8+ T cell effector cytokine R production independent of TCR Ag affinity. Analysis of RNA-sequencing data reveals that genes for inflammatory cytokines/cytokine receptors are significantly altered upon changes in Ag affinity and TCR signal strength. Furthermore, our findings show that the inflammatory milieu is critical in regulating this TCR signal strength–mediated increase in memory development, as both CpG oligonucleotide treatment or cotransfer of wild-type and Itk−/− T cells eliminates the observed increase in memory cell formation. These findings suggest that TCR signal strength and Ag affinity independently contribute to CD8+ memory T cell development, which is modulated by inflammation, and suggest that manipulating TCR signal strength along with Ag affinity, may be used to tune the development of CD8+ memory T cells during vaccine development.
Effective vaccination relies on the formation of proper adaptive immune responses, requiring immune memory cell development (1). Although humoral immunity (B cell mediated) provides the basis for most classical vaccines, there are still problematic pathogens, such as HIV, Mycobacterium tuberculosis, and the malaria parasite, among others, for which humoral-mediated vaccines either do not work or do not exist, and that may require harnessing of both B and T cell immunity (2). CD8+ T cells play a key role in cell-mediated immunity and are important in the clearance of intracellular pathogens. During an infection or vaccination, naive CD8+ T cells are activated by Ags and pass through several characteristic phases before becoming mature, long-lived memory cells. These phases have been well characterized and defined by the differential expression of cell surface markers: IL-7R α-chain (IL-7Rα, also known as CD127) and killer cell lectin-like receptor subfamily G member 1 (KLRG1). During the initial phase, Ag-stimulated naive CD8+ T cells expand and differentiate into a heterogeneous population of effector cells (3, 4). The majority of the effector cell population is comprised of short-lived effector cells (SLECs), identified as CD127lo KLRG1hi. These SLECs are responsible for mediating pathogen clearance and do so by secreting effector cytokines such as IFN-γ and TNF-α (3). Once these SLECs successfully clear the pathogen, the T cell population contracts, and the remaining 5–10% of surviving cells are known as memory precursor effector cells (MPECs), identified as CD127hi KLRG1lo (3, 5). Importantly, MPECs are the effector cells that eventually give rise to long-term memory cells (6–8). The CD8+ memory T cell pool consists of diverse subsets of memory cells with distinct homing properties (9), defined by the differential expression of trafficking/migration molecules such as CD62L and CD44. Effector memory T cells (TEM; KLRG1loCD127hiCD44hiCD62Llo) recirculate in the periphery, whereas central memory T cells (TCM; KRLG1loCD127hiCD44hiCD62Lhi) and long-lived effector cells (LLECs; KLRG1hiCD27loTbethiEomeslo) reside in secondary lymphoid organs (10–12).
Several different determinants have been reported to influence the magnitude of the primary T cell response, including inflammatory cytokines, costimulatory signals, Ag abundance, and tissue microenvironment (13–16). The signal strength theory proposes that the strength of the signal from the TCR is important in CD8+ T cell differentiation of effector and memory cells (3). Although Ag affinity has been associated with TCR signal strength, both parameters have been suggested to make separate contributions to T cell activation (17–20). Furthermore, whereas low-affinity TCR-ligand interactions are sufficient in activating CD8+ T cells (21–24), it remains unclear whether TCR signal strength and Ag affinity intersect to regulate the CD8+ T cell response.
IL-2 inducible tyrosine kinase (ITK) is a Tec family kinase that acts downstream of the TCR (25–27). ITK has been shown to regulate the strength of the TCR signal during T cell activation (28–31). We have previously shown that reducing TCR signal strength via deletion of ITK leads to an increase in the proportion of Ag-specific CD8+ MPECs (32). This data support the theory that TCR signal strength inversely regulates the development of memory T cells. In this study, by using ITK-deficient OT-1 TCR transgenic mice, in which CD8+ T cells are engineered to recognize the OVA protein but exhibit reduced TCR signaling, we were able to examine the intersection between TCR signal strength and Ag affinity to determine their influence on the development of CD8+ memory T cells during infection with Listeria monocytogenes. We found that TCR signal strength and Ag affinity independently contribute to CD8+ memory T cell development and that reducing both leads to enhanced development of MPECs.
Materials and Methods
All mice were on a C57BL/6 background. OT-1/Rag−/− mice were from Taconic Biosciences, and Itk−/−/OT-1/Rag−/− mice were previously described (32). CD45.1 (B6.SJL-Ptprca Pepcb/BoyJ) mice were from The Jackson Laboratory and crossed to OT-1/Rag−/− to generate CD45.1+CD45.2+ OT-1/Rag−/− mice. Congenically marked CD45.2+ OT-1/Rag−/− mice were used in single adoptive transfer experiments, whereas CD45.1+CD45.2+ congenically marked mice were used for cotransfer purposes. CD45.1 mice were used as recipients in transfer experiments. Both female and male mice were employed in all experiments. All experiments were reviewed and approved by the Cornell University Institutional Animal Care and Use Committee.
Adoptive transfer of naive CD8+ T cells and in vivo infection with L. monocytogenes
A total of 1 × 105 sorted naive CD8+ T cells per mouse (OT-1/Rag−/−, Itk−/−/OT-1/Rag−/−) were transferred i.v. into CD45.1 recipient mice. Twenty-four hours following the transfer of naive CD8+ T cells, L. monocytogenes expressing either the N4 OVA epitope (wild-type [WT] or referred to as high-affinity condition) or T4 OVA epitope variant (lower-affinity condition) (22) (a gift from Dr. Michael Bevan, University of Washington) were administered i.p. at a dose of 5 × 105 CFU/mouse. Mice were bled once per week for a month to track the primary immune response, and cells were analyzed using flow cytometry. After a month, some mice were reinfected with 5 × 106 CFU/mouse of L. monocytogenes expressing the N4 epitope to examine the secondary immune response. On day 7 following the reinfection, spleens were harvested and analyzed using flow cytometry. A similar protocol was followed for the cotransfer experiment, in which naive CD8+ T cells were sorted from OT-1/Rag−/− (CD45.1+CD45.2+) and Itk−/−/OT-1/Rag−/− (CD45.1−CD45.2+) mice and mixed at a 1:1 ratio before i.v. injecting the cells into CD45.1+CD45.2− recipient mice. For inflammation experiments, CpG oligonucleotide was used to induce inflammation/signal 3 cytokine production. Mice were infected with L. monocytogenes expressing N4-OVA at the same dose indicated above, and subsequently injected (i.p.) with 100 μg of CpG oligonucleotide 1826 (Invivogen) (33).
In vitro cultures and stimulation
Complete RPMI 1640 was used for all cell culture experiments. For functional analysis of cells isolated from infected animals, cells were cultured with either 1 μM SIINFEKL (N4) peptide or PMA (20 ng/ml)/ionomycin (2 μM) in presence of brefeldin A (100 μg/ml; Sigma-Aldrich) for 4–6 h followed by intracellular staining and analysis by flow cytometry. Cells were analyzed for production of proinflammatory cytokines (TNF-α, IFN-γ) and proliferation (Ki67) using specific Abs. For analysis of the expression of IFN regulatory factor (IRF) 4 and CD69 during in vitro stimulation, splenocytes from OT-1/Rag−/− or Itk−/−/OT-1/Rag−/− mice were cultured in vitro with 1 μM SIINFEKL (N4) or SIITFEKL (T4) for 4 d, then collected, and T cells were analyzed by flow cytometry.
Abs and flow cytometric staining
Blood was collected in 50 U/ml heparin (Sigma-Aldrich) to prevent clotting, and ammonium–chloride—potassium (ACK) lysis was performed to lyse RBCs before surface, cytokine, and nuclear staining. The following Abs were used for staining and FACS analysis: Pacific Blue–anti-CD45.2, PECy7–anti-KLRG1, eF506-viability dye, PerCP-Cy5.5-anti-TNF-α, PECy7-anti-IFN-γ, allophycocyanin–anti-CD27, FITC–anti-KLRG1, PeCy7–anti-Ki67, PE–anti-Nurr77, AF647–anti-IRF4 (eBioscience), allophycocyanin–anti-IRF4, AF700–anti-CD45.1, PerCP-Cy5.5-anti-CD127, PECy7–anti-CD62L, PE–anti-CD44, PE-Cy7-anti-CD69, AF700–anti-CD8α, allophycocyanin-Cy7-anti-CD45.1, allophycocyanin-Cy7-anti-Vα2 (BioLegend), and PE-CF594-anti-CD8α (BD Biosciences). All transferred cells (donor cells) were gated based on the expression of CD45.1, CD45.2, CD8α, and Vα2 markers. SLECs and MPECs were identified using expression of CD127 and KLRG1 (SLECs: CD127loKLRG1hi; MPECs: CD127hiKLRG1lo). Distinct memory cell subsets were further identified by the markers CD44, CD62L, and CD27 (TCM: CD44hiCD62Lhi; TEM: CD44hiCD62Llo, LLEC KLRG1hiCD27lo). Mean fluorescence intensity (MFI) was determined by gating on the donor population in FlowJo. The Foxp3 Transcription Factor Staining Buffer Set (eBioscience) was used to detect nuclear proteins. To determine cytokine production, cells were stimulated with the N4 (WT) OVA peptide, or PMA/ionomycin (P/I) and analyzed as described above.
CD8+ T cells were sorted from spleens of mice as indicated above prior to infection (day 0) and following infection on day 7 with L. monocytogenes expressing the N4 OVA (LM-N4) or T4 OVA (LM-T4) epitope. RNA-sequencing data from day 0 cells were generated as previously described (W. Huang, J. Luo, and A. August, manuscript posted on bioRxiv). For day 7 cells, RNA-sequence libraries were prepared and subjected to Illumina sequencing by the RNA sequencing Core Facility in the College of Veterinary Medicine at Cornell University. Sequencing results were mapped to the mm10 genome. Copy numbers were normalized, and fragments per kb of transcript per million mapped reads were used for analyses. Differentially expressed genes were identified using GeneSpring, with a fold change of at least two along with a p value of <0.05, to generate volcano plots and principal component analysis (PCA) plot. Gene set analysis was performed using Gene Set Enrichment Analysis (GSEA) software from the Broad Institute, and genes were deemed significant based on a false discovery rate cutoff of <0.05. GSEA was performed using the Hallmarks referencing dataset, and the classic enrichment statistic was employed with 1000 gene set permutations. To analyze and compare the data from Scott-Browne et al. (34), a custom gene set (CGS) was generated (gmx format) for all up- and downregulated genes from the effector/naive (CGS) condition after values were transformed to log2, averaged, and the corresponding human orthologs identified (34) (Supplemental Fig. 3). The RNA-sequencing data have been deposited in the National Center for Biotechnology Information's Gene Expression Omnibus under accession number GSE127406 (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE137406).
Statistical and data analysis
Representative experiments were chosen for data depiction. In each experiment, three to five mice were used per group, and experiments were repeated two to three times as indicated. A Student t test and ANOVA statistical tests were performed using GraphPad Prism v5.00, and p values <0.05 were deemed statistically significant.
Reduced TCR signal strength and lower Ag affinity increase memory cell development during a primary immune response
To probe the effect of Ag affinity on the development of CD8+ memory T cells during a primary immune response, we used a recombinant L. monocytogenes strain carrying either the OVA SIINFEKL (N4) epitope, described in this study as the high-affinity peptide, or an altered SIITFEKL variant (T4), described in this study as the low-affinity peptide (LM-N4 or LM-T4). The T4 variant has been shown to have a 70.7-fold lower EC50 compared with the WT epitope for activating T cells bearing the OVA-specific OT-1 TCR (22). To probe the effect of TCR signal strength, we took advantage of the fact that the absence of ITK reduces TCR signal strength during T cell activation (28, 32, 35). There are multiple reports that ITK plays a critical role as a regulator of TCR signal strength [e.g., see (29, 30, 32, 35–40)], and our studies support this conclusion as well because the absence of ITK in OT-1 transgenic T cells results in reduced levels of expression of IRF4 and of CD69 without affecting the proportion of cells that actually express CD69 (Supplemental Fig. 1). We performed adoptive transfer experiments using naive CD8+ T cells isolated from CD45.2+ OT-1/Rag−/− (WT) and Itk−/−/OT-1/Rag−/− (Itk−/−) mice. Note that crossing the Itk−/− mice to the OT-1/Rag−/− transgenic mouse system eliminates the memory phenotype (41–43) of the resulting CD8+ T cells, and these cells have a naive phenotype similar to the WT OT-1/Rag−/− T cells (Fig. 1A) (32). Reducing TCR signal strength in the absence of ITK does not affect the affinity of the OT-1 TCR for its cognate Ag SIINFEKL as determined by tetramer binding assays (data not shown). Following the transfer of these T cells into CD45.1 recipient mice, we infected recipients with L. monocytogenes expressing the high-affinity (LM-N4) or low-affinity (LM-T4) OVA peptide to assess how Ag affinity influences the immune response in the face of reduced TCR signal strength (Fig. 1A). As previously reported, we found that there was significantly less expansion of WT OT-1 T cells in mice infected with the L. monocytogenes carrying the low-affinity peptide (T4) compared with those infected with L. monocytogenes carrying the high-affinity peptide (N4) (Fig. 1B) (22, 44). However, there was no difference in the percentages of either SLECs (CD127loKLRG1hi) or MPECs (CD127hiKLRG1lo) over the course of the primary response, regardless of Ag affinity (Fig. 1C), suggesting these differences in Ag affinity alone do not affect this process.
By contrast, we found that reducing TCR signal strength in the absence of ITK did not affect the expansion of responding Itk−/− OT-1 T cells (Fig. 1D). However, reducing the TCR signal strength and lowering Ag affinity led to significantly reduced percentages of SLECs following expansion on days 13, 20, and 26, whereas the percentages of MPECs were significantly increased by day 20 (Fig. 1E). To illustrate this finding more clearly, we plotted the data from all four conditions on the same graphs (Fig. 1F, 1G). In this study, it can be more clearly appreciated that OT-1 T cells that receive reduced TCR signal strength (in the absence of Itk) and are responding to lower-affinity Ag respond significantly better, with peak cell expansion on day 13. Similarly, this combination of reduced TCR signal strength and reduced Ag affinity also resulted in enhanced MPEC development along with reduced SLEC development (Fig. 1E, 1G). Consistent with our previous report, our data in this study revealed that reducing TCR signal strength (via deletion of ITK) leads to accelerated MPEC development starting on day 7, regardless of Ag affinity (Fig. 1G; MPEC Itk−/− N4 versus WT N4) (32). However, surprisingly, combining reduced TCR signal strength (via deletion of ITK) with reduced Ag affinity (via T4 variant) led to further enhanced development of MPECs (Fig. 1G; Itk−/− T4 versus all other groups). In agreement with the conclusion that ITK regulates the strength of signals that the cells receive, analysis of responding WT and Itk−/− T cells in vivo for the expression of Nurr77, a key readout for TCR signal strength (45), revealed lower expression in the absence of ITK and in response to infection with LM-T4 (lower-affinity Ag) (Supplemental Fig. 1).
Next, we determined which memory cell subsets were present on day 30 postinfection. The best-characterized memory cell subsets are TCM (CD44hiCD62Lhi), which are known to undergo robust proliferation upon recall and localize to lymphoid tissue, and TEM (CD44hiCD62Llo), which are present mostly in peripheral tissue with limited recall proliferative capacity (12). On day 30 there was no significant difference in the percentage of TCM cells, whereas the percentage of TEM was enhanced when the TCR signal was attenuated or Ag affinity was reduced, although not when both TCR signal strength and Ag affinity were reduced (Fig. 1H). By contrast, reducing both TCR signal strength and Ag affinity resulted in a decrease in the development of the LLEC (KLRG1hi CD27lo) subset known to play a protective role against infections such as vaccinia virus and L. monocytogenes (11, 46). Altogether, our data suggest that TCR signal strength and Ag affinity are distinct parameters that differentially modulate memory development.
Effector cytokine response during primary infection is regulated by TCR signal strength independent of Ag affinity
To assess whether cytokine production is affected by TCR signal strength and Ag affinity over the course of the primary immune response, mice were bled each week following the adoptive transfer of naive CD8+ T cells and infection with L. monocytogenes expressing either high-affinity (LM-N4) or low-affinity (LM-T4) OVA peptide. After bleeding, primary CD8+ T cells were cultured with either OVA-N4 peptide or P/I as control, and production of proinflammatory cytokines IFN-γ and TNF-α were examined by flow cytometry. We found that, whereas the proportion of cells responding to OVA peptide stimulation increases over the course of the immune response, the percentage of cells producing TNF-α, and the levels of TNF-α (as determined by the MFI of the staining), was significantly reduced in Itk−/− T cells, independent of Ag affinity (Fig. 2A). Similar results were observed for the production of IFN-γ, with both percentages and MFI reduced in Itk−/− T cells, regardless of the primary Ag affinity (Fig. 2B, 2D for double producers). By contrast, there was no difference in TNF-α and/or IFN-γ production when the TCR was bypassed by stimulation with P/I (day 7 data shown in Fig. 2C, 2E). This data suggests that TCR signal strength may be more important than Ag affinity in regulating the effector cytokine response of CD8+ T cells. It is also important to note that, whereas the Itk−/− CD8+ T cells make significantly less proinflammatory cytokine, they are still able to successfully clear the L. monocytogenes infection by day 7 (32).
TCR signal strength and Ag affinity regulate effector cell development and cytokine production upon reinfection
To determine how these two parameters influence the recall response, mice were reinfected with a high dose of L. monocytogenes expressing the high-affinity (N4) OVA peptide (10 times the dose used for the primary infection in Fig. 1). Upon the secondary challenge, whereas proliferation (as determined by Ki67 staining) remained similar in all groups, regardless of TCR signal strength or Ag affinity, effector cytokine production was significantly altered (Fig. 3A). WT cells that previously responded to infection with low-affinity Ag produced less IFN-γ and TNF-α compared with those that previously responded to infection with high-affinity Ag (Fig. 3B–D). Furthermore, as in the primary response, Itk−/− cells that previously responded to infection with either high- or low-affinity Ag produced less IFN-γ and TNF-α, compared with WT cells, regardless of Ag affinity. By contrast, no change in the MFI of TNF-α was observed when TCR signal strength was attenuated (Fig. 3E). These results further support the regulatory role of TCR signal strength in mediating cytokine production during primary and secondary response.
Infection and Ag affinity drive the largest changes in transcriptome of responding CD8+ T cells
To further understand the effects of changes in Ag affinity and TCR signal strength during the initial portion of the primary immune response, RNA sequencing was carried out to compare sorted CD8+ T cells prior to infection (at day 0) and 7 d postinfection with L. monocytogenes expressing either the high-affinity (N4) or the low-affinity (T4) OVA peptide. PCA of the expressed genes in each condition revealed distinct clustering driven primarily by infection and Ag affinity, with little apparent effect of TCR signal strength (Fig. 4A). Note that WT and Itk−/− cells cluster closely together prior to infection (day 0), suggesting that, irrespective of differences in TCR signal strength, the transcriptome of these CD8+ T cells are closely related prior to activation with Ag (Fig. 4A). Following infection on day 7, differences in Ag affinity led to distinct clustering, despite altered TCR signal strength, such that the transcriptome of these CD8+ T cells diverged significantly, largely based on their perception of the differences in Ag affinity during infection.
We first analyzed the data to understand the effect of Ag affinity alone on the T cell response by comparing WT cells responding to either the LM-N4 or LM-T4 infection. Among the differentially expressed genes, we found that the following pathways and genes were significantly upregulated in WT cells responding to both LM-N4 and LM-T4 compared with WT day 0 cells (WTN4:WT0 and WTT4:WT0): TNF-α signaling (TNFAIP2, TNFSF9, CXCL10, TNF, TNFAIP3, CCL5), IL-2_STAT5 signaling (IFNGR1, TNFSF10, IL-2RB, TNFRSF1B, IL-10RA, IL-18R1, IL-2RA), and the inflammatory response (CXCR6, IL-18RAP, CCRL2, CCL5) (Supplemental Fig. 2A, 2B). However, WT cells responding to LM-T4 compared with WT day 0 cells (WTT4:WT0) also exhibited upregulation of the IL-6_JAK_STAT3 pathway (IL-2RA, IL-15RA, IL-2RB1, IL-1B, TNF, CXCL10, TNF, IL-18R1, TNFRSF1B, IL-3RA, TNFRSF1A, TNFRAF12A, IFNGR1) (Supplemental Fig. 2B). Importantly, gene sets for the metabolic pathways, including glycolysis, MTORC1 signaling, and oxidative phosphorylation were upregulated in the lower-affinity (WTT4) condition only compared with WT day 0 cells. This suggests that, whereas the inflammatory response and cytokine signaling pathways/genes are not affected when Ag affinity is reduced, the metabolic profile of the cells may be playing an important role in the response to Ag of lower affinity.
Next, we compared our GSEA findings to the work of Scott-Browne and colleagues (34) by examining the pathways/genes that were significantly upregulated in effector (day 8) cells responding to lymphocytic choriomeningitis virus infection compared with naive (day 0) CD8+ T cells (referred to in this study as a CGS) (Supplemental Fig. 3). We found that similar pathways were significantly upregulated in day 8 effector cells compared with naive cells. Comparison of our data in GSEA (WTN4:WT day 0 and WTT4:WT day 0) revealed certain gene sets that are enriched (IL-2-STAT5 signaling, PI3K-AKT-MTOR signaling, KRAS signaling, TNF-α signaling, and inflammatory response) to be identical to those enriched in the CGS derived from the Scott-Browne study (Supplemental Figs. 2, 3).
Using GSEA, we then assessed the role of Ag affinity on the transcriptome of the responding WT CD8+ T cells, comparing those responding to either LM-N4 or LM-T4 infection (WTN4:WTT4). Enriched genes included Egr1 and Egr2, which were upregulated in WT-N4 compared with WT-T4, supporting the difference in signaling by TCR because of reduced Ag affinity (Fig. 4B). In addition, the IFN-γ and IFN-α response (IRF7, IRF1, IRF9, IRF2, IL-4R), TNF-α signaling (IFNGR2, TNFRSF9, IRF1) and IL-6_JAK_STAT3 signaling (IFNGR2, IL-3R, IRF1, TNFRSF12A, TGFB1, IL-2R, IRF9, TNFRSF1A, IL-4R), and cytokine receptors (IFN-γR2, IL-3RA, TNFSF11, TNFRSF9) were significantly upregulated in WT-T4 compared with WT-N4 control (Supplemental Fig. 4A). Genes involved in oxidative phosphorylation, E2F, MYC1, MYC2, and MTORC1 signaling were upregulated, as well suggesting Ag affinity may also be acting to alter the metabolic state of the cells (Supplemental Fig. 4A).
To investigate the effect of TCR signal strength, we explored gene expression profiles of Itk−/− CD8+ T cells responding to either infection with LM-N4 or LM-T4 (Itk−/− N4:Itk−/− T4). In this condition, similar to what was observed in responding WT cells, genes involved in oxidative phosphorylation (E2F, MYC1, and MTORC1) were upregulated; although different from WT cells, those involved in glycolysis and fatty acid metabolism were also upregulated in the Itk−/− T cells responding to low-affinity Ag (T4) compared with control Ag (N4) (Supplemental Fig. 4B). E2f1, IFNGR2, IL-1B, IL-3RA, ID3, TNFRSF9, TNAIP2, and TNFSF12 are upregulated in Itk−/− T cells responding to low-affinity Ag (T4) compared with those responding to the higher-affinity Ag (N4). By contrast, EGR1, EIF2S37, CD127 (IL-7R), and TRAT1 were downregulated in Itk−/− T cells responding to low-affinity Ag (T4) compared with control Ag (N4) (Fig. 4C).
In hopes of gaining a better functional understanding of how TCR signal strength and Ag affinity intersect, we examined the gene sets significantly changed in our Itk−/−: WT (IWR) cells. We determined that genes involved in the inflammatory response, TNF-α signaling (IL-7R, IFNGR2, IRF1, IL-15R), WNT β-catenin signaling, KRAS signaling, and TGFβ (IFNGR2, TGFBR1) were upregulated, whereas E2F, MYC1, oxidative phosphorylation, MTORC1, glycolysis, cholesterol homeostasis, and fatty acid metabolism were downregulated (Fig. 4D, Supplemental Fig. 4C) in Itk−/− cells compared with WT. A heatmap of the enriched genes involved in inflammation indicates that cytokine receptors IL-7R, IL-4RA, IL-15RA, IFNGR2, and IFNAR1, chemokines CXCL10 and CCL5, and transcription factors IRF7 and IRF1 are upregulated in Itk−/− cells regardless of Ag affinity (Fig. 4E). This data suggests that reducing TCR signal strength potentially alters the cells response to cytokine and inflammatory signals.
Transcriptome analysis reveals changes in cytokine and metabolic gene sets driven by difference in TCR signal strength and Ag affinity
Next, we examined the changes occurring during infection with either the high-affinity (N4) or low-affinity (T4) Ag. In the high-affinity condition (LM-N4), GSEA analysis of the Itk−/−:WT ratio at D7 revealed that the genes associated with the inflammatory response were upregulated in the Itk−/− cells compared with WT cells, in particular, cytokine receptor genes IL-7R, IFNGR2, IFNAR1, TNFRSF9, TNFSF10, IL-18R1, and IL-4RA and cytokines IL-1B and IL-18 (Fig. 5A, 5C). Genes associated with E2F targets, TNF-α signaling, oxidative phosphorylation, PI3K-AKT-MTOR, glycolysis, hypoxia, cholesterol homeostasis, and mTORC1 were all downregulated in Itk−/− compared with WT cells (Supplemental Fig. 4D). A volcano plot shows that the transcription factors EGR1 and EGR2 are downregulated, whereas IL-7R and Trat1 are upregulated in the Itk−/− cells. Contrary to N4, the low-affinity (LM-T4) infection revealed that oxidative phosphorylation, glycolysis, MTORC1 signaling, MYC1, and cholesterol homeostasis are upregulated in the Itk−/− cells, whereas TNF-α signaling, hypoxia, WNT β-catenin signaling, KRAS signaling, and the inflammatory response are downregulated compared with WT cells (Supplemental Fig. 4E). Enriched genes include those for cytokine receptors IL-18RAP, IL-15RA, and IL-4RA (Fig. 5D). The volcano plot indicates that transcription factors FOSL1, EGR1, and NRA41 (Nur77) are downregulated, whereas LYN, SYK, TGFBI, LRP1, E2F1, and TNFRSF25 are upregulated (Fig. 5B).
Notably, E2F1, E2F7, FOXM1, E2F8, TNFRSF9, SYK, LYN, IL-1B, IL-3RA, IFNGR2, TNFS12, were upregulated in both WT and Itk−/− T cells responding to low-affinity Ag (LM-T4) compared with those responding to the higher-affinity Ag (LM-N4), whereas EGR1, MAP3K2, and EIF2S3Y, were downregulated in both WT and Itk−/− T cells responding to low-affinity Ag (LM-T4) compared with those responding to the higher-affinity Ag (LMM-N4), suggesting that these genes are regulated by Ag affinity regardless of TCR signal strength (not all genes are depicted on volcano plots) (Fig. 4B, 4C).
This data suggests that reducing TCR signal strength along with reducing Ag affinity leads to the upregulation of various metabolic pathways, whereas the cytokine receptor genes IL-18RAP, IL-15RA, IRF1, IL-4RA, and IRF7 are downregulated compared with WT cells (Fig. 5D, Supplemental Fig. 4E). Furthermore, infection (regardless of Ag affinity) accounts for the changes in transcriptome we observe as both the high-affinity (N4) and IWR conditions led to upregulation of genes involved in the inflammatory response, whereas the low-affinity condition led to downregulation.
Cotransfer of WT and Itk−/− T cells eliminates the differential increase in MPECs regardless of Ag affinity
The production of inflammatory cytokines has been reported as a key determinant in the development of SLECs and MPECs, with greater levels of inflammation diminishing MPEC potential (47–49). Our data reveal that WT cells produce significantly more inflammatory cytokine compared with Itk−/− cells during the primary response (Fig. 2), whereas GSEA revealed that the IFN-γ–, IFN-α–, and TNF-α–associated gene sets were upregulated in the WT cells responding to low-affinity infection; this was not observed when TCR signal strength was reduced (Supplemental Fig. 4A, 4B). Furthermore, WT cells responding to Ag, regardless of affinity, upregulated the TNF-α signaling, IL-2_STAT5 signaling, and the inflammatory response gene sets compared with day 0 cells (Supplemental Fig. 2). To determine whether this difference in cytokine production between WT and Itk−/− T cells affects the response of the Itk−/− T cells, we cotransferred WT and Itk−/− cells into the same animals (mixing congenic donor cells, WT: CD45.2+CD45.1+ and Itk−/−: CD45.2+CD45.1− at a 1:1 ratio), into the same recipient animal (CD45.2−CD45.1+). We then determined whether the response of the Itk−/− T cells persisted in the presence of accompanying WT T cells and based on Ag affinity. Following infection, the number of transferred WT and Itk−/− T cells was unaffected by Ag affinity (Fig. 6A). In addition, the proportion of SLECs and MPECs was not significantly different, despite differences in TCR signal strength and/or Ag affinity (Fig. 6B, 6C). This suggests that the local inflammatory environment generated by the cotransferred WT cells may be able to influence cell expansion and differentiation and prevent the predilection of Itk−/− cells for memory formation. Analysis of cytokine production upon Ag stimulation (by restimulation with the WT N4 peptide) revealed that Itk−/− T cells continued to secrete significantly less proinflammatory cytokine (single IFN-γ or TNF-α or double producers) independent of Ag affinity, further supporting the view that TCR signal strength is a major regulator of cytokine production (Fig. 6D–G). This finding suggests that the reduced inflammatory environment generated by Itk−/− T cells may play a role in enhanced MPEC development upon reduction of TCR signal strength, as well as on the effects of the intersection between TCR signal strength and Ag affinity.
Inflammation negatively regulates MPEC development
Our results suggest that the inflammatory environment mediated by the cytokines secreted by the cotransferred WT T cells may be able to influence the development of MPECs in the Itk−/− T cells (Fig. 6). To determine more directly whether the MPEC trajectory of Itk−/− T cells is affected by the presence of inflammatory cytokines, we used CpG oligonucleotide 1826 to induce systemic inflammation in mice during infection with LM-N4 (high-affinity peptide) (33, 50–52). Upon infection, unlike what was observed in the absence of CpG oligonucleotide, there was no difference in development of MPECs between WT and Itk−/− T cells in the presence of CpG oligonucleotide (Fig. 7A–C; CpG oligonucleotide also did not affect the expansion of cells), further supporting the idea that inflammation modulates memory precursor development and may function to suppress memory development. It is worth noting that CpG oligonucleotide exposure did not affect the reduced production of the proinflammatory cytokine exhibited by Itk−/− T cells (Fig. 7D–G). Our data suggest that attenuated TCR signal strength/Ag affinity can enhance MPECs, which is normally inhibited by strong inflammation.
In this study, we have examined how TCR signal strength and Ag affinity tune CD8+ memory T cell development to understand memory formation during infection and vaccination. We showed that reducing TCR signal strength leads to an accelerated development of memory effector precursor cells and that reducing TCR signal strength along with reducing Ag affinity for the TCR results in a further increase in the proportion of Ag-specific memory precursors. Furthermore, transcriptomic analysis by RNA sequencing suggests that the inflammatory response, specifically cytokine receptor expression, along with different metabolic pathways (glycolysis, oxidative phosphorylation, MTORC1 signaling, MYC1, MYC2) are significantly altered when TCR signal strength and Ag affinity are modulated. In this work, we have relied on multiple reports that ITK plays a critical role as a regulator of TCR signal strength [e.g., see (29, 30, 32, 35–40)]. Indeed, other investigators have reported on other molecules downstream of the TCR that also regulate signal strength [e.g., see (29, 30, 32, 35–40)]. Taken together, our data supports the idea that there is an inverse relationship between TCR strength, Ag affinity, and memory development.
Notably, we found that a reduction in Ag affinity alone did not lead to changes in MPEC development, although there was less cell expansion. This finding supports the finding of Zehn et al. (22), who reported that T cells activated by low-affinity Ag underwent less expansion, yet they were still able to differentiate into CD8+ memory T cells and maintain a recall response upon infection. Paradoxically, however, we found that reducing TCR signal strength along with reducing Ag affinity led to significantly more cell expansion, along with lower SLEC and greater MPEC percentages. However, analysis of the transcriptome of responding T cells, comparing signal strength and Ag affinity conditions by PCA, revealed that the primary influence of the transcriptome of the responding T cells at D7 is Ag affinity and not TCR signal strength. Further analysis by GSEA revealed that reducing Ag affinity alone led to the upregulation of key cytokine receptors such as IFNGR2, TNFSF1A, TNFRSF9, TNFAF12A, and TGFB1. Reducing TCR signal strength, regardless of Ag affinity, revealed that a number of metabolic pathways were upregulated, including MYC1, oxidative phosphorylation, MTORC1, glycolysis, cholesterol homeostasis, MYC2, and adipogenesis.
TCR signal strength also seems to be pivotal in mediating cytokine production, as attenuating signal strength resulted in significantly less proinflammatory cytokine production. Other groups have reported a similar finding: that TCR signal strength plays a role in regulating CD8+ T cell effector functions (53). Our data suggest that this direct relationship between TCR signal strength and cytokine production seems to be independent of Ag affinity, as cytokine production is similar, regardless of whether the cells have been initially stimulated with high-affinity (N4) or low-affinity (T4) Ag.
Aside from the signal strength, inflammation has been identified as another parameter that influences the CD8+ T cell response (47, 50, 54, 55). Our cotransfer experiments, in which a mixture of both WT and Itk−/− cells were allowed to respond in the same animals, suggest that inflammatory cytokines produced by WT cells alter the behavior of the Itk−/− cells, eliminating their advantage in developing MPECs, regardless of TCR signal strength and Ag affinity. Supporting this conclusion, we also found that the enhanced development of MPECs by Itk−/− cells was reverted by inducing systemic inflammation (using CpG oligonucleotide), suggesting that reduced inflammation was associated with a better MPEC phenotype in Itk−/− cells. Importantly, and by contrast, the decrease in cytokine production observed in Itk−/− cells was retained, despite the cotransfer with WT cells. Our data suggest that cell-intrinsic differences in how the cells respond to inflammation may account for why reducing TCR signal strength and lowering Ag affinity leads to greater cell expansion and MPEC formation. Indeed, several different types of inflammatory molecules, including IL-12, IFN-γ, and type 1 IFNs, are known to inhibit the acquisition of memory characteristics (50, 51, 56). RNA sequencing confirmed that reducing TCR signal strength led to the upregulation of the receptors for several proinflammatory cytokines (IL-7R, IL-4R, IL-15R, Ifnar1, Irf7, Irf1, Ifngr2, TGFBR1), suggesting that Itk−/− cells may be more sensitive to the surrounding inflammatory milieu compared with WT cells. The differential behavior of the Itk−/− cells, dependent on whether WT cells were present, suggests a more nuanced explanation for the influence of the inflammatory milieu on their response. It is more likely that both the intrinsic and extrinsic differences account for why the Itk−/− cells behave differently during the response, dependent on the environment in which they are responding. It would be of considerable interest to determine the behavior of Itk−/− cells that lack the ability to respond to inflammatory cytokines. Reducing TCR signal strength also led to a number of downregulated gene sets, including E2F targets, MYC1 targets, oxidative phosphorylation, MTORC1 signaling, glycolysis, cholesterol homeostasis, fatty acid metabolism, and adipogenesis (IWR comparison), suggesting that the metabolic profile of the cells is altered when TCR signal strength is attenuated.
Our IWR comparison revealed that the cytokine receptor gene IL-4R was positively enriched in Itk−/− T cells compared with WT cells, regardless of Ag affinity. IL-4R is known to play a role in CD8+ T cell memory responses (57–60), and IL-4 may play a role in Itk−/− cells in the observed increase in memory development. Other cytokine receptors that could also play such roles include IL-15R and IL-7R, which were also found to be upregulated in Itk−/− cells compared with WT cells. One possibility is that reducing TCR signal strength makes the cells more sensitive to the inflammatory milieu, leading to better ability to receive strong memory-inducing cytokine signals (IL-7, IL-15, IL-21), allowing them to survive and further develop into MPECs. Finally, MTORC1 signaling, as well as a number of other metabolic pathways, were downregulated when TCR signal strength was reduced. Indeed, changes in the metabolic profiles occur during T cell differentiation, and inhibiting the mTOR pathway with rapamycin has been previously shown to enhance memory CD8+ T cell development (61). We do note that for RNA sequencing we used cells collected at day 7 of the response, the peak of the WT T cell response to LM-N4 and LM-T4 and of the Itk−/− T cell response to the LM-N4 infection. However, the peak of the Itk−/− T cell response to the LM-T4 is at day 14. We balanced this differential peak response with our interest in identifying pathways that are acting prior to the time that we start to see a difference in MPEC development, which we observe at day 14. It would be of significant interest to be able to do RNA sequencing from responding cells at multiple time points for all four conditions. We also used 1 × 105 for naive cell transfers, which increases the precursor frequency for the Ag beyond what would normally be the case for a nontransgenic system. This cell number is within the range used by others for similar experiments; however, we are aware of studies reporting that the initial precursor frequency affects the nature of the subsequent response. We also cannot, of course, rule out other contributions that regulate a process that takes multiple days to evolve.
In conclusion, we have shown that both TCR signal strength and Ag affinity tune the CD8+ T cell response upon infection and that both parameters are promising for vaccine development purposes. Given that Btk/Itk inhibitors, such as ibrutinib, are currently being explored for their potential to treat a variety of cancers, autoimmune, and inflammatory disease, their use in vaccine development may also be advantageous (62, 63). Ag affinity has also emerged as another desirable lever with which to modulate the T cell response, as low-affinity TCRs can preferentially mediate tumor killing while remaining tolerant against self-antigens (64, 65). Given that a polyclonal repertoire of various TCR affinities exists, choosing the ideal affinity for vaccination purposes will be important for eliciting protective immunity. Additionally, although we showed that reduced TCR signal strength/Ag affinity resulted in a strong MPEC eliciting effect, it also resulted in reduced inflammation, which may be a potential benefit in the development of vaccines. Hence, our finding suggests that both TCR signal strength and Ag affinity are parameters that may lead to promising use in the design of T cell–mediated vaccines.
We thank Amie Redko for animal care, Ling Zhang for technical support, Dr. Jennifer Grenier for help in RNA-sequencing data collection and analyses, and members in the August laboratory for helpful discussions.
This work was supported by National Institutes of Health (NIH)/National Institute of Allergy and Infectious Diseases Grants AI120701 and AI138570 (to A.A.) and AI129422 (to A.A. and W.H.), NIH Grant R35ES028244 (to A.A. and Gary Perdew, principal investigator), NIH/National Institute of General Medical Sciences Grant GM130555 Sub-6610 (to W.H.), NIH/National Institute of Child Health and Human Development Grant HD076210 (to the Cornell RNA Sequencing Core Facility), and a Howard Hughes Medical Institute Professorship (to A.A.). J.E. was supported in part by NIH/National Institute of Biomedical Imaging and Bioengineering Training Grant T32EB023860. C.L. is a Cornell Sloan Fellow.
The RNA-sequencing data presented in this article have been submitted to the National Center for Biotechnology Information's Gene Expression Omnibus (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE137406) under accession number GSE127406.
The online version of this article contains supplemental material.
Abbreviations used in this article:
custom gene set
Gene Set Enrichment Analysis
IFN regulatory factor
IL-2 inducible tyrosine kinase
killer cell lectin-like receptor subfamily G member 1
long-lived effector cell
L. monocytogenes expressing the N4 OVA
L. monocytogenes expressing the T4 OVA
mean fluorescence intensity
memory precursor effector cell
principal component analysis
short-lived effector cell
central memory T cell
effector memory T cell
A.A. receives research funding from the 3M Company. The other authors have no financial conflicts of interest.