TCR signal strength controls the differentiation of CD4+ effector and memory T cells

See allHide authors and affiliations

Science Immunology  20 Jul 2018:
Vol. 3, Issue 25, eaas9103
DOI: 10.1126/sciimmunol.aas9103

Signal strength seals fate

T cell differentiation into effector and memory T cell subsets is influenced by T cell receptor (TCR) signals. Snook et al. examine how TCR signals influence CD4+ T cell differentiation using a panel of cloned TCRs that recognize the same MHC II–restricted epitope of lymphocytic choriomeningitis virus. Strong TCR signals were associated with TH1 differentiation, whereas lower TCR signal strength corresponded with follicular helper T cell and memory T cell differentiation. Low CD25 expression by early effector T cells also predicted memory differentiation, although CD25 expression levels were not predictive of recall responses. Enhanced TCR signaling via knockdown of SHP-1 favored TH1 over Tfh and memory T cell differentiation. These results indicated that stronger TCR signaling promotes terminal effector TH1 differentiation.


CD4+ T cell responses are composed of heterogeneous T cell receptor (TCR) signals that influence the acquisition of effector and memory characteristics. We sought to define early TCR-dependent activation events that control T cell differentiation. A polyclonal panel of TCRs specific for the same viral antigen demonstrated substantial variability in TCR signal strength, expression of CD25, and activation of nuclear factor of activated T cells and nuclear factor κB. After viral infection, strong TCR signals corresponded to T helper cell (TH1) differentiation, whereas T follicular helper cell and memory T cell differentiation were most efficient when TCR signals were comparatively lower. We observed substantial heterogeneity in TCR-dependent CD25 expression in vivo, and the vast majority of CD4+ memory T cells were derived from CD25lo effector cells that displayed decreased TCR signaling in vivo. Nevertheless, memory T cells derived from either CD25lo or CD25hi effector cells responded vigorously to rechallenge, indicating that, although early clonal differences in CD25 expression predicted memory T cell numbers, they did not predict memory T cell function on a per cell basis. Gene transcription analysis demonstrated expression clustering based on CD25 expression and enrichment of transcripts associated with enhanced T follicular helper cell and memory development within CD25lo effector cells. Direct enhancement of TCR signaling via knockdown of Src homology region 2 domain–containing phosphatase 1, a tyrosine phosphatase that suppresses early TCR signaling events, favored the differentiation of TH1 effector and memory cells. We conclude that strong TCR signals during early T cell activation favor terminal TH1 differentiation over long-term TH1 and T follicular helper cell memory responses.


The induction of memory T cells is a key focus in the development of vaccines and immunotherapies directed toward infectious pathogens and tumors (1, 2). The primary T cell response to acutely infecting pathogens is marked by rapid proliferation and the development of key effector functions. After pathogen clearance, 90 to 95% of effector T cells die, leaving behind a long-lived population of memory T cells (3, 4). CD8+ effector T cells that are memory precursors can be identified by the expression of cell surface markers such as interleukin-7Rα (IL-7Rα) (5), and some progress has been made in identifying memory precursor CD4+ T cells (6). However, the specific signals and mechanisms that dictate CD4+ memory T cell fate commitment during the effector response remain elusive.

External differentiation cues, such as cytokines, play a well-known role in controlling T helper subset effector and memory differentiation. However, cell-intrinsic signals mediated by the T cell receptor (TCR) also control many aspects of CD4+ T cell differentiation. CD4+ T cells require multiple interactions with their cognate antigen to successfully differentiate into competent effector (7, 8) and memory (9) T cells. Several lines of evidence indicate that strong TCR signals favor T helper 1 cell (TH1) differentiation both in vitro (10) and in vivo (11). In addition, TH1 differentiation is associated with enhanced CD25 expression (12), an early activation marker driven by TCR signaling. In contrast, T follicular helper cell (Tfh) specification has been associated, in separate studies, with high-affinity TCRs or TCRs with long dwell times (13, 14), and occupation of multiple immunoreceptor tyrosine-based activation motifs on a single CD3ζ is also required for Tfh differentiation (15). Monoclonal T cell populations responding to the same epitope can also produce heterogeneous TCR signals, leading to differential effector fates (11, 12). The TCR-dependent early activation molecules IL-2 and IL-2Rα (CD25) are also implicated in T helper differentiation. Exogenous IL-2 treatments (16) or analysis of early CD25 expression profiles (12) has highlighted a key temporal role for IL-2 signaling in T helper differentiation. A key downstream transcription factor of IL-2 signaling, signal transducer and activator of transcription 5, has been shown to drive TH1 development (17), and IL-2 and IL-21 have been shown to promote TH1 and Tfh differentiation, respectively, although it is not clear whether the effect is paracrine or autocrine (18, 19).

Because TCR molecules are themselves highly variable, the antigen-specific response to an infection is marked by a high level of clonal diversity (20, 21). However, this diversity is subject to a process of selection, as shown by our previous finding that not all T cell clones give rise to memory cells with equal efficiency after acute infection with lymphocytic choriomeningitis virus (LCMV) or Listeria monocytogenes (22). The goal of the current study is to acquire a better understanding of the TCR signals propagated by “memory-biased” versus “effector-biased” T cell clones during the polyclonal response.

We analyzed a panel of previously cloned TCRs, all recognizing the same major histocompatibility complex (MHC) class II–restricted epitope, glycoprotein 61–80 (GP61–80) of LCMV, and each with a previously defined contribution to the CD4+ memory T cell pool during an in vivo polyclonal response. We found that overall TCR signal strength inversely corresponded to the contribution of each TCR to the formation of T cell memory. During in vivo infection with LCMV, the extent of both ζ chain–associated protein kinase 70 (ZAP-70) phosphorylation and CD25 expression at early effector time points inversely corresponded to memory potential. Heterogeneous CD25 expression predicted a bias in the formation of TH1 and Tfh populations. CD25lo effector cells gave rise to a mix of TH1 and Tfh effector cells, as well as most TH1-like and Tfh-like memory cells, whereas CD25hi early effector cells gave rise almost exclusively to terminally differentiated effector TH1 cells. This differential T cell fate was further supported through global transcriptional analysis. Direct modulation of TCR signaling via the short hairpin RNA (shRNA)–mediated knockdown (KD) of the Src homology region 2 domain–containing phosphatase 1 (SHP-1) additionally biased the response toward the differentiation of effector TH1 cells, indicating that TCR signal strength shapes the differential formation of both effector and memory CD4+ T cells with Tfh or TH1 characteristics.


Heterogeneous induction of TCR signals in vitro corresponds to in vivo fate

We investigated a panel of natively arising TCRs specific for the immunodominant MHC class II–restricted epitope of LCMV, GP61–80 (fig. S1A). All TCRs in the panel were derived from SMα mice, single-chain TCR transgenic mice expressing the TCRα of the SMARTA TCR (GP61–80-specific) paired to an endogenous TCRβ repertoire (20). Because each cloned TCR has a defined contribution to memory in the setting of in vivo viral infection (20), we used this panel to assess the consequences of differential signaling initiated by memory- and effector-biased TCRs. We first created cell lines expressing each TCR by transducing a parent hybridoma T cell line with recombinant retroviruses expressing a bicistronic TCR construct and an mCherry reporter (fig. S1B) (23). The parent hybridoma line did not express an endogenous TCR and contained a green fluorescent protein (GFP) reporter under the control of a minimal consensus nuclear factor of activated T cells (NFAT)–sensitive promoter (24). We further transduced each hybridoma line with an additional retrovirus containing a cyan fluorescent protein (CFP) reporter under the control of a nuclear factor κB (NFκB) response element (25), thus allowing us to simultaneously detect NFAT and NFκB activity (fig. S1B). Each line expressed similar levels of surface TCR (fig. S1A) and GFP after stimulation with phorbol 12-myristate 13-acetate (PMA)/ionomycin (Fig. 1A). Hybridomas were coincubated for 24 hours with dendritic cells (DCs) presenting GP61–80 peptide (pepDCs) on MHC class II. GFP production by each hybridoma cell line was measured by flow cytometry (Fig. 1A) and confirmed by Western blot (fig. S1C). The TCR-dependent NFAT activity, as measured by GFP induced by each cell line, was highly heterogeneous and inversely corresponded to our previous assessment of the memory potential of each TCR (Fig. 1, A and B, and fig. S1C). TCRs previously shown to have lower representation in the memory compartment, as compared with the peak of the effector response (MemLo; fig. S1A), induced significantly higher levels of GFP than TCRs previously shown to have equal or higher representation in the memory compartment, as compared with the peak of the effector response (MemHi; fig. S1A). Similar results were found when we assessed NFκB-induced CFP expression (Fig. 1B). We also measured TCR-dependent gene expression. After 24 hours of stimulation, induction of CD25 surface expression corresponded to GFP expression (Fig. 1C).

Fig. 1 Memory-biased TCRs induce weaker TCR signals than effector-biased TCRs in vitro.

(A) Eleven T cell hybridoma lines, each expressing a unique GP61–80-specific TCR, an NFAT GFP reporter, and an NFκB CFP reporter were stimulated with GP61–80-pulsed DCs (pepDCs) or PMA/ionomycin (ION) for 24 hours. Representative flow plots show GFP expression for three of the cell lines. Bar graphs show GFP expression after stimulation with PMA/ionomycin or pepDCs for 24 hours. (B) Bar graphs depict GFP and CFP mean fluorescence intensity (MFI) after 24 hours of stimulation, comparing TCRs that are present at reduced frequency at memory time points in vivo (Memlo) to TCRs that were present at equal or increased frequencies at memory time points (Memhi), as compared with the peak of the effector response. (C) Plot indicates the correlation of GFP expression to either CD25 surface expression after 24 hours of stimulation with pepDCs, as determined by Pearson’s correlation. Throughout the study, error bars indicate the SEM, and statistical significance was determined by Student’s t test. *P < 0.05 (n = 3 biological replicates per group, representative of three independent experiments).

To further explore differences in TCR signaling in the context of primary T cell activation, we created two transgenic mouse lines expressing TCRs that recognize GP61–80 of LCMV (C7 and C26). C7 CD4+ T cells displayed diminished phosphorylation of ZAP-70 and CD3ζ after 1 hour of coincubation with pepDCs, as compared with C26 cells (fig. S2, A and B). In addition, C7 T cells expressed lower levels of the TCR-dependent activation marker CD25 than did C26 T cells after 24 hours of stimulation (fig. S2C). We also performed reverse transcription polymerase chain reaction (RT-PCR) on RNA extracted from C7 or C26 splenocyte cultures that had been stimulated with 0.1 μM GP61–80 for 1 to 3 days, with particular focus on genes immediately up-regulated after TCR activation (Il2 and Nfatc1) or involved in effector function (Tbx21 and Ifng). C26 T cells demonstrated increased expression of Nfatc1 and Il2 by day 3 of culture (fig. S2D). C26 T cells also expressed higher levels of Ifng and Tbx21 transcripts at days 1 and 3 of culture (fig. S2D). Previous studies have suggested that strong TCR signals promote enhanced TH1 differentiation, and our findings are consistent with that premise (11, 26). These findings confirm heterogeneous TCR signaling and TCR-dependent activation induced by two TCRs that recognize the same immunodominant epitope.

Interclonal differences in TCR signal strength and CD25 expression predict effector and memory differentiation

In our previous studies, we used SMα mice to establish a role for the TCR in regulating CD4+ memory T cell differentiation. We observed that memory-biased TCRs (represented at higher frequencies at memory time points than at effector time points) were enriched for Vβ14+ T cells, whereas effector-biased TCRs (represented at higher frequencies at effector time points than at memory time points) were enriched for Vβ7+ T cells (20). We took advantage of this observation to assess the expression of CD25 on effector T cells that were more or less likely to give rise to memory T cells. We infected SMα mice with LCMV, followed by detection of GP66–77 tetramer-binding CD4+ T cells in the spleen at day 5 post-infection. We found that responding Vβ14+ effector cells were less likely to express CD25 than the tetramer-binding population as a whole, whereas Vβ7+ effector cells were more likely to express CD25 (Fig. 2A). T-bet expression was also higher in the Vβ7+ effector cells (Fig. 2B), leading to the conclusion that, within a monoclonal population, CD25lo effector cells were more likely to give rise to memory T cells, whereas CD25hi effector cells were enriched for terminally differentiated TH1 cells. We additionally assessed the expression of CD25 in early effector cells derived from a polyclonal T cell repertoire in wild-type (WT) mice. At day 5 post-infection, we observed highly variable CD25 expression within activated (IAb-GP66–77 tetramer+, CD44+) CD4+ T cells (fig. S3), indicating that CD25 expression is broadly heterogeneous during a physiologic T cell response in vivo.

Fig. 2 TCR signal strength and CD25 expression correspond to CD4+ Tfh effector differentiation and memory T cell formation in vivo.

(A) At day 5 post-infection with LCMV, CD4+ splenocytes from SMα mice were stained with I-Ab/GP66–77 tetramer (Tet), CD25, and either Vβ7 or Vβ14. Representative plots show tetramer staining (gated on CD4+) and CD25 staining (gated on CD4+tetramer+). The bar graph indicates the ratio frequency of Vβ7+ or Vβ14+ cells within the CD25hi versus CD25lo tetramer+ cells. (B) Bar graph shows the T-bet MFI for Vβ7+ and Vβ14+ tetramer-binding cells. (C) Four retrogenic (GFP+) CD4+ T cell lines were adoptively transferred (1 × 105 for analysis at day 3 and 1 to 3 × 104 for analysis at days 8 and 42) into B6 hosts that were subsequently infected with LCMV. At day 3 post-infection, GFP+ T cells were analyzed for the presence of pZAP-70 and expression of CD25 by flow cytometry. The induction of ZAP-70 phosphorylation was calculated by subtracting the pZAP-70 MFI of the total CD4+ T cell population from the pZAP-70 MFI of the GFP+ Rg T cells (ΔpZAP-70). (D) The numbers of GFP+ retrogenic T cells in the spleen were calculated at days 8 and 42 post-infection, and percent survival between these two time points was calculated (% survival). The plots indicate the correlation of ΔpZAP-70 MFI or CD25 expression (%CD25hi) at day 3 post-infection to % survival for each Rg TCR, as determined by Pearson’s correlation. Error bars indicate the SEM, and statistical significance was determined by Student’s t test (n = 3 to 5 mice per group, representative of at least two independent experiments). *P < 0.05, **P < 0.01.

We next asked whether clonal differences in TCR signal strength predicted memory formation in vivo. We performed a clone-by-clone analysis by generating several TCR retrogenic T cell lines as previously described (fig. S1A) (20, 27, 28), adoptively transferring them into B6 mice and infecting them with LCMV 1 day later. For each of the four clones tested (5, 7, 26, 27), we observed heterogeneity in the phosphorylation of ZAP-70 and the expression of CD25 by day 3 post-infection (Fig. 2C). We further measured the number of peak effector (day 8) and memory cells (day 42) for each clone. We found that reduced expression of phosphorylated ZAP-70 (pZAP-70) at day 3 in clones 5 and 7, as compared with clones 26 and 27, significantly corresponded to the proportion of resulting peak effector cells that gave rise to memory cells (Fig. 2D). A significantly smaller proportion of clone 5 T cells expressed high levels of CD25 as compared with clones 26 and 27, whereas clone 7 displayed an intermediate phenotype (Fig. 2C). CD25 also corresponded to memory formation, although in this case, the differences indicated a trend only (Fig. 2D).

Differences in CD25 expression correspond to differences in TCR signal strength in vivo

We next used an adoptive transfer model in which WT mice (Thy1.2+) received an intravenous injection of SMARTA TCR transgenic CD4+ T cells (Thy1.1+), followed by infection with LCMV 1 day later. SMARTA T cells showed uniform patterns of CD25 expression through day 2 post-infection, with almost all activated T cells expressing high levels of CD25 by day 2. However, by day 3, a proportion of SMARTA T cells expressed low levels of CD25, and this bimodal expression persisted through day 5. We found similar results when assessing the expression of CD25 by C7 and C26 T cells after LCMV infection. In addition, the proportion of CD25lo early effector cells was significantly different when comparing C7 and C26 at day 3 post-infection (fig. S4A), similar to the clonal differences in CD25 expression observed for retrogenic T cell clones (Fig. 2). Differences in CD25 expression did not coincide with differences in the expression of classical activation markers CD44 and CD62L or secretion of the effector cytokines interferon-γ (IFN-γ), tumor necrosis factor–α, and IL-2 (fig. S4B).

CD25 surface expression predicted differences in effector differentiation. CD25hi early SMARTA effector cells (day 3 post-infection) in the spleen expressed phenotypic markers indicative of TH1 differentiation [lymphocyte antigen 6 complex, locus C1 (Ly6C) and T cell immunoglobulin and mucin-domain containing-3 (Tim-3)]that were largely absent on CD25lo effector cells (Fig. 3B). CD25hi SMARTA early effector cells additionally expressed higher levels of T-bet (Fig. 3C), whereas CD25lo effector cells expressed increased levels of the Tfh markers CXCR5 and T cell factor 1 (TCF-1; Fig. 3C). Furthermore, CD25hi effector cells in the spleen expressed higher levels of pZAP-70 at days 3 and 5 post-infection (Fig. 3D).

Fig. 3 CD25 surface expression and TCR signal strength predict T helper differentiation and memory potential of early effector T cells in vivo.

(A) SMARTA T cells (Thy1.1+) were adoptively transferred (3 × 104) into B6 hosts (Thy1.2+), followed by LCMV infection. Representative flow histograms indicate the CD25 surface expression by SMARTA T cells at days 0 to 5 (d0 to d5) after LCMV infection. (B) The representative flow histogram shows expression of CD25 by SMARTA cells at day 3 post-infection and expression of Tim-3 and Ly6C by CD25hi and CD25lo subsets. (C) The bar graphs indicate the MFI of CXCR5, T-bet, TCF-1, and Bcl-6 in CD25hi and CD25lo SMARTA CD4+ T cells in the spleen at day 3 post-infection. (D) The bar graph indicates ΔpZAP-70 MFI of CD25hi (“High”) and CD25lo (“Low”) SMARTA CD4+ T cells in the spleen at days 3 and 5 post-infection. Error bars indicate the SEM, and statistical significance was determined by Student’s t test (n = 4 mice per group, representative of four independent experiments). *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

CD25 surface expression predicts memory potential

Given the heterogeneity of TCR signaling and CD25 expression even within monoclonal populations, we tested whether CD25 surface expression by early effector cells predicted effector and memory differentiation. Several lines of evidence suggested this possibility. First, CD25 expression distinguished CD8+ effector T cells likely to undergo terminal effector differentiation from those that give rise to memory T cells (29). Second, Blimp-1 expression by early CD4+ effector T cells inversely corresponded to memory potential, and CD25 was strongly coexpressed with Blimp-1 in those studies (12). Third, in vitro induction of CD25 driven by TCR signal strength strongly corresponded to increasing TH1 polarization (Fig. 3, C and D, and fig. S2D).

We isolated CD25hi and CD25lo SMARTA effector cells at days 3 or 5 post-infection and transferred them into separate infection-matched B6 hosts in equal numbers (Fig. 4A). Although both populations continued to expand, the vast majority of circulating and spleen-residing memory cells were derived from the CD25lo effector population as early as day 3 post-infection (Fig. 4, B and C). In contrast, both CD25lo and CD25hi early effector cells gave rise to liver-residing memory T cells (Fig. 4C) with similar efficiency. This may reflect the reported role for IL-2 in the establishment of tissue-residing CD4+ memory T cells (30, 31), although the interpretation is made complex by the temporal changes that we observed in high-affinity IL-2R expression (Fig. 3A). We used an additional retrogenic T cell line (clone 18), generated as previously described (fig. S1), to perform similar adoptive transfer experiments. As with SMARTA T cells, CD25lo clone 18 early effector cells similarly gave rise to most memory T cells (fig. S5, A and B), indicating that CD25 expression predicts memory differentiation across multiple clones.

Fig. 4 CD25 expression predicts effector and memory differentiation.

(A) Schematic depicts isolation and transfer of CD25hi and CD25lo early effector SMARTA CD4+ T cells infection-matched secondary hosts. (B) Graph indicates the frequency of SMARTA cells (Thy1.1+) in the blood of secondary hosts at the indicated time points after transfer of CD25hi (“High”) and CD25lo (“Low”) SMARTA cells at day 5 post-infection (p.i.). (C) Bar graph indicates the total number of SMARTA memory cells (day 42 post-infection) in the spleen and liver after transfer of equal numbers of CD25hi or CD25lo early effector cells into infection-matched hosts at either days 3 or 5 post-infection. (D) Plots indicate the frequency of CXCR5hiPD-1hi Tfh effector cells derived from CD25lo and CD25hi effector SMARTA that were transferred at day 5 and analyzed at day 8. (E) Bar graphs depict the frequency of Tfh (CXCR5hiPD-1hi) and TH1 (Ly6Chi) day 8 effector cells derived from CD25hi and CD25lo early effector cells isolated and transferred at day 5 post-infection. Remaining bar graphs depict the MFI of Bcl-6 and T-bet at day 8. (F) Bar graphs show the frequency of CXCR5+ SMARTA cells in the spleen and the MFI after intracellular antibody staining for the presence of Bcl-6 and T-bet. Error bars indicate SEM, and statistical significance was determined by Student’s t test (n = 3 to 5 mice per group, representative of at least three independent experiments). *P < 0.05, **P < 0.01, ***P < 0.001.

CD25 expression at early effector time points further predicted T helper differentiation at the peak of the effector response (day 8). CD25hi early effector cells gave rise to mostly TH1 cells at day 8 post-infection, as measured by increased expression of Ly6C and T-bet. In contrast, CD25lo cells gave rise to mostly Tfh effector cells at day 8 post-infection, as measured by expression of CXCR5, programmed cell death protein 1 (PD-1), and B cell lymphoma 6 protein (Bcl-6) (Fig. 4, D and E). Similar results were obtained after transfer of CD25hi or CD25lo clone 18 early effector cells (fig. S5C). Memory cells derived from CD25hi effector cells expressed increased T-bet (Fig. 4F), consistent with the differentiation of effector memory cells. These results indicate a role for TCR signaling in driving both memory and effector CD4+ T cell differentiation.

CD25 expression during the primary response does not predict memory T cell function

To test the memory function of CD25hi- or CD25lo-derived SMARTA memory cells, we isolated SMARTA cells at day 42 after primary LCMV infection, transferred them into naïve hosts, and rechallenged them with LCMV (Fig. 5A). At the peak of their secondary response (day 5 post-infection), both CD25lo- and CD25hi-derived SMARTA memory cells had undergone similar levels of clonal expansion (Fig. 5B). We observed no differences in secondary T helper differentiation, as shown by expression of CXCR5, PD-1, Ly6C, Bcl-6, and T-bet (Fig. 5, C and D). We concluded that, although CD25lo and CD25hi early effector cells gave rise to different numbers of memory cells, secondary expansion and differentiation of those memory cells were similar. Further, we determined that memory cells were highly functional regardless of whether they were derived from CD25hi or CD25lo effector cells, as shown by high levels of effector cytokine secretion upon ex vivo restimulation (Fig. 5E).

Fig. 5 CD4+ memory T cells derived from either CD25hi or CD25lo effector cells respond robustly to secondary challenge.

(A) Schematic depicts the isolation and transfer of SMARTA memory cells (day 42) derived from either day 5 CD25hi or CD25lo effector populations into naïve B6 hosts, followed by secondary challenge with LCMV. (B) Bar graph indicates the fold expansion of each SMARTA population, as measured by splenic cell numbers at day 5 postsecondary challenge. (C) Representative flow plots show the expression of PD-1 and CXCR5 on SMARTA T cells 5 days after LCMV infection. (D) Bar graphs indicate the expression levels of CXCR5, PD-1, Ly6C, Bcl-6, and T-bet in either frequency or MFI via flow cytometry. (E) Bar graphs show the frequency of single and multicytokine-producing SMARTA T cells (day 5 post-infection) after ex vivo peptide restimulation. Error bars indicate SEM, and statistical significance was determined by Student’s t test (n = 3 mice per group).

CD25 expression identifies two transcriptionally distinct subsets of early effector cells

We used RNA sequencing (RNA-seq) to compare the gene expression profiles of CD25hi and CD25lo SMARTA effector cells undergoing different levels of TCR signaling at day 5 post-infection. Each population had a unique transcriptional signature, as shown by cluster analysis of genes that showed significant differential expression (Fig. 6A). Among these genes was CD25 itself, which indicated that differences in CD25 expression were transcriptionally regulated and served as an internal control for the validity of the analysis (Fig. 6B). CD25hi early effector cells had increased expression of a number of NFAT-inducible genes, including Runx3 (32), Ifng (32), and Ppp3ca (33), and other TCR-inducible genes, including Bhlhe40 (34) and Dusp22 (35). Furthermore, gene expression in CD25hi cells was indicative of enhanced TH1 differentiation, as determined by expression of Prf1, Il12rb2, Tbx21, and Prdm1 (Blimp-1) (19, 36, 37). IL12rb2 expression has also been associated with TCR signal strength (38). In contrast, CD25lo effector cells had increased expression of genes related to the regulation of T cell activation such as Btla (39), Egr2, and Egr3 (40); genes associated with memory T cell formation such as Tcf7 (TCF-1) (41, 42), Pou2af1 (OCA-B) (43), and Cd27 (44); and genes associated with Tfh differentiation such as Cebpa, Tcf7, Il6st, Id3, and Il6ra (Fig. 5B) (18, 41, 45). Differential expression of key genes was confirmed via RT-PCR (Fig. 6C).

Fig. 6 CD25 expression identifies two transcriptionally distinct subsets of very early effector cells.

(A) CD25hi and CD25lo SMARTA cells were isolated at day 5 post-infection, followed by RNA-seq (n = 3). Hierarchical clustering indicated unique gene expression patterns and significantly up-regulated (red) or down-regulated (blue) genes. (B) Bar graph indicates a list of selected genes with significantly increased expression in CD25hi (red) and CD25lo (blue) populations. The x axis indicates the difference in the number of transcripts on a log2 scale. (C) Bar graphs show RT-PCR–based confirmation of differences in gene expression for the selected genes between day 5 CD25hi (“High”) and CD25lo (“Low”) SMARTA CD4+ T cells. SEM (n = 3 samples per group). *P < 0.05, **P < 0.01, ***P < 0.001.

Decreased SHP-1, a key TCR signaling modulator, reduces Tfh differentiation

To modulate TCR signal strength in primary CD4+ T cells, we targeted the protein tyrosine phosphatase SHP-1 via shRNA. Because SHP-1 is a key regulator of the activity of TCR proximal tyrosine kinases, including ZAP-70 (46), we hypothesized that SHP-1 KD would result in enhanced and/or sustained TCR signaling. We expressed two different mir30-flanked shRNAs in retroviral expression plasmids, both specific for SHP-1, that had been previously described and displayed significant SHP-1 KD (30 to 80%) when expressed in EL-4 cells (fig. S6A) (47). We then generated bone marrow chimeras by transducing SMARTA bone marrow with SHP-1 shRNA retroviral vectors (SHP-1 KD), or an empty vector (EV) control, and transplanting it into irradiated Rag−/− recipients. Eight to 10 weeks later, cells were stimulated in vitro with DCs presenting GP61–80 and tested for the presence of pZAP-70. For both shRNA constructs, SHP-1 KD resulted in more rapid induction and sustained maintenance of pZAP-70 (fig. S6B), indicating an impact on TCR signal strength.

GFP+ (SHP-1 KD or EV) and GFP (WT) cells from the chimeras were adoptively transferred into B6 recipient mice, followed by LCMV infection. This allowed us to compare SHP-1 KD and EV SMARTA responses in different hosts and responses by transduced (GFP+) and nontransduced (GFP) SMARTA cells in the same host. As early as day 3 post-infection, a TH1 bias was present in SHP-1 KD SMARTA cells, evidenced by increased levels of CD25 and Tim-3 compared with WT and EV controls (Fig. 7A). SHP-1 KD did not affect the overall activation of the SMARTA CD4+ T cells, as measured by CD44 and CD62L expression (Fig. 7B). Day 8 effector cells also evidenced a TH1 bias, as determined by a decrease in the proportion of effector cells expressing CXCR5 and an increase in the proportion of effector cells expressing Ly6C (Fig. 7, C and D). The bias away from Tfh differentiation persisted into memory, as SHP-1 KD resulted in significantly fewer Tfh-like memory cells (Fig. 7C). However, at both effector and memory time points, the overall number of SMARTA was not significantly altered by SHP-1 KD, indicating that the decrease in Tfh was compensated by an increase in TH1. In support of this, SHP-1 KD cells produced the TH1 cytokine IFN-γ at a higher frequency and an increased level on a per cell basis than their WT counterparts (Fig. 7E). We concluded that effector and memory CD4+ T cell differentiation is governed, at least in part, by TCR signal strength.

Fig. 7 SHP-1 KD induces a bias toward effector and memory TH1 cells.

We generated SMARTA bone marrow chimeras expressing either a SHP-1–specific shRNA or an EV control, along with a GFP reporter. At 8 to 10 weeks, SMARTA T cells (Thy1.1+) were adoptively transferred into B6 recipient and infected with LCMV. GFP+ (SHP-1 KD or EV) and GFP (nontransduced, WT) effector SMARTA cells from the spleen were analyzed. (A) Representative flow plots show CD25 and Tim-3 expression on SHP-1 KD (black) and WT (gray) SMARTA T cells 3 days after LCMV infection. Line graphs show the difference in the frequency of GFP+ and GFP SMARTA expressing each marker within the same mouse. (B) Representative flow plots show the surface expression of CD44 and CD62L on SMARTA T cells at day 8 post-infection. (C) Representative flow histograms show CXCR5 expression by SHP-1 KD (GFP+) and WT (GFP) SMARTA cells within the same mouse. Line graphs depict the differences in the frequency of CXCR5-expressing SMARTAs at days 8 and 42 post-infection between GFP+ and GFP SMARTA cells in the spleen in both SHP-1 KD and EV recipients. (D) Representative flow plots show the expression of CXCR5 and Ly6C by SHP-1 KD (GFP+) and WT (GFP) SMARTA cells in the spleen at day 8 post-infection. Numbers indicate the frequency of CXCR5+Ly6C (Tfh) and the CXCR5Ly6C+ (TH1) SMARTA effector cells. The line graph shows the change of frequency of CXCR5+Ly6C SMARTA cells in the spleen when comparing SHP-1 KD (GFP+) and WT (GFP) SMARTA in the same host. (E) The line graph shows the change of frequency of IFN-γ–secreting SMARTA cells in the spleen. The bar graph shows the IFN-γ MFI of SHP-1 KD (GFP+) and WT (GFP) SMARTA IFN-γ–producing cells from the spleen. Error bars indicate the SEM. Pairwise comparisons and statistics were performed on GFP+ and GFP SMARTA cells that were analyzed in the same recipient mouse. Results are representative of at least two independent experiments (n = 3 to 4 mice per group). *P < 0.05, **P < 0.01, ****P < 0.0001.


Our results find a key role for TCR signal strength, as regulated by SHP-1, in determining clonal differences in both T helper differentiation (TH1 versus Tfh) and memory formation. The TCR has previously been shown to influence T helper cell differentiation (11, 13, 15). Strong TCR signals favor TH1 over TH2 differentiation (11), and the extent of TH1 effector function and polarization is dependent on TCR signal strength (26). The TCR also plays a role in the differentiation of TH1 and Tfh cells. In one study, high-affinity TCRs favored the differentiation of Tfh (13), whereas in a second study, Tfh differentiation corresponded to long TCR/MHC dwell times (14). More recently, it was found that limiting TCR signaling in T cells selectively impaired Tfh differentiation while leaving TH1 responses relatively untouched (15). In contrast, our findings demonstrate a correlation of stronger TCR signals and terminal TH1 differentiation, whereas the differentiation of both Tfh and TH1 with memory potential required comparatively weaker TCR signals. Several possibilities may explain these differences. First, although we show that TCR signaling distinguishes terminal effector cells from memory precursors, it is not clear how it might influence the differentiation of TH1 and Tfh derived from CD25lo effector cells. Second, the role of the TCR may be influenced by antigen availability, antigen localization, or the infection-dependent inflammatory environment. Third, TCR signaling is unlikely to be uniform throughout the primary response, and the impact of altering TCR signaling may depend on the approach.

IL-2 has been shown to play an important role in the generation of TH1 cells (15) and effector and memory CD4+ memory T cells that home to tertiary organs (30, 31). In contrast to the preferential survival of CD25lo effector T cell in circulation and in secondary lymphoid organs, CD25hi effector cells gave rise to liver-residing memory T cells with equal efficiency to CD25lo effector cells. This may reflect the variable role of IL-2 in driving the formation of these memory populations. CD25 is expressed by virtually all T cells during early activation. It is not known whether IL-2 signaling is required for tissue-resident memory T cell development only during the early phases of the effector response or whether sustained IL-2 signaling is required throughout the effector response. Given the heterogeneity of the expression of the high-affinity IL-2 receptor, it will be of interest to define these differences and to determine whether the TCR mechanistically controls T cell fate, at least in part, by controlling CD25 expression.

The fate of individual clones within the primary immune response is highly heterogeneous. Single antigen-specific CD4+ precursor T cells can give rise to a clonally uniform T helper phenotype but with a high amount of variability from clone to clone (14). This suggests that T helper fate decisions occur very early in the immune response but that naïve precursors T cells activated under similar in vivo conditions can give rise to highly distinct differentiation programs. Fate tracking of single CD4+ T cell precursors and their progeny revealed heterogeneity in T helper differentiation even between precursor cells that expressed the same TCR (48). These results support our finding that differences in CD25 expression and ZAP-70 phosphorylation reflect heterogeneity in TCR signal strength even within a monoclonal T cell population. We speculate that the activity of TCR-mediated differentiation may be affected and shaped by a number of environmental factors, including cytokines, costimulatory molecules, the antigen presenting cell, and antigen dose. Future studies are required to determine the factors that can give rise to differential activation events among T cells even when the TCR is the same.

There are several limitations to the interpretation of this study. Although we report differences in TCR signal strength in vitro, it is important to note that the in vitro initiation of TCR signals may not fully predict the heterogeneity of the in vivo response. It is likely that in vivo activation results from multiple or prolonged contacts with antigen. Furthermore, in vivo activation may occur in microenvironments that have variable concentrations of cytokines and other accessory signals. Our study does not distinguish between TCR signal strength and signal duration, and differences in antigen recognition by naïve T cells during the earliest phases of the immune response may have a distinct influence on memory formation as compared with antigen recognition by effector T cells during the peak of the infection or in the late phases of the effector response. In addition, TCR signaling is highly complex, and qualitatively distinct TCR signals may result from heterogeneity in antigen recognition during the polyclonal response. Future studies are needed to determine the impact of modulating multiple components of TCR signaling on CD4+ memory T cell differentiation.



The objective of this study was to explore the role of TCR signal strength in determining the differentiation of effector and memory CD4+ T cells in vivo. Flow cytometry was used to assess the activation, differentiation, and subsequent survival of CD4+ T cells after acute viral infection within laboratory mice. Cellular analysis was performed during the early T cell effector phase at the peak of the effector response and after memory formation. Techniques for the modulation of gene expression (shRNA) were used to confirm the results of observational studies. The sample size (n = 3 to 5) for the in vivo experiments was determined to be the optimal size for statistical analysis while using an appropriate number of laboratory mice and allowing for independent repeats. The investigators were not blinded when conducting or analyzing the experiments outlined in this study, the mice were randomly assigned to the different treatments, and repeat experiments were carried out in both male and female cohorts, with no apparent sex differences.

Mice and infections

C57BL/6 (6 to 8 weeks old) mice were purchased from the Jackson Laboratories. SMARTA (49), Rag1-deficient, and SMα mice were maintained in our colony at the University of Utah. C7 and C26 mice were generated at the University of Utah Transgenic Core Facility by standard microinjection techniques using a T cell–specific expression vector, VA-hCD2, in which the Tcrb gene was placed under the control of the human CD2 promoter and a 3′ locus control region of the Cd2 gene (50). They were then bred with the previously generated SMα line (20) to produce a full TCR transgenic line on a Tcra−/− background. Mice are currently being backcrossed to a C57BL/6 background. LCMV Armstrong 53b was grown in baby hamster kidney (BHK) cells, titered in Vero cells, and injected intraperitoneally into recipient mice at a dose of 2 × 105 plaque-forming units. All mouse experiments were performed in accordance with protocols approved by the Institutional Animal Care and Use Committee at the University of Utah.

Cell lines and retroviral transductions

We previously generated a panel of TCRs (20) cloned into MigR1. In this construct, the TCRa and TCRb sequences are separated by a cis-acting hydrolyzing element, P2A, that allows bicistronic expression (27) and has an internal ribosomal entry site–dependent mCherry reporter. Using previously described methods (28), replication-incompetent retroviruses were used to transduce the 58αβ hybridoma cell line (23) expressing an NFAT reporter [hCD4-pA-GFP-NFAT-RV; provided by K. Murphy (Washington University, St. Louis, MO)] (24). The hybridoma lines were additionally transduced with a MigR1 retroviral construct expressing CFP under the control of NFκB minimal promoter [construct provided by P. Steinberger (Medical University of Vienna, Austria)]. Hybridoma lines were purified by fluorescence-activated cell sorting (FACS).

Cell preparations and flow cytometry

Splenocyte and liver single-cell suspensions were generated as previously described (8) and placed in cell culture media. Untouched CD4+ T cells from SMARTA, C7, and C26 mice were isolated via magnetic beads (Miltenyi Biotec) and injected intravenously into B6 mice 1 day before LCMV infection. For cell surface stains, single-cell suspensions were incubated with fluorescently conjugated antibodies diluted in antibody-staining buffer [phosphate-buffered saline (PBS) containing 1% fetal bovine serum (FBS)] at 4°C for 30 to 45 min. For intracellular cytokine assays, splenocytes were restimulated for 4 hours with 1 μM GP61–80 peptide from LCMV (GLKGPDIYKGVYQFKSVEFD) at 37°C in the presence of Brefeldin A (1 μl/ml; GolgiPlug), permeabilized with a kit (BD Biosciences), and stained with fluorescently labeled antibodies specific to the indicated cytokines. Transcription factor analysis was performed using the Foxp3 Fixation/Permeabilization Buffer and accompanying protocol (eBioscience). To probe for phosphorylation events via flow cytometry, cells were immediately fixed after extraction from the animal and formation of a single-cell suspension using prewarmed (37°C) 1.5% paraformaldehyde for 10 min at 37°C, followed by permeabilization in ice-cold 100% MeOH for 10 min. Cells were then incubated with antibodies to surface and intracellular targets from 45 to 60 min. Tetramer staining was done for 1 hour at room temperature in RPMI 1640 containing 2% FBS and 0.1% sodium azide, followed by cell surface staining.

Retrogenic bone marrow chimeras

To generate TCR retrogenic bone marrow chimeras, we used the above-described TCR expression constructs to generate retrovirus and then transduced TCR-expressing retroviruses into Rag1-deficient bone marrow cells using the described methods (28). We then injected bone marrow cells intravenously into irradiated (450 rads) Rag1−/− hosts and monitored for the presence of GFP+TCR+CD4+ T cells in the blood 8 to 10 weeks later.

Isolation of CD25hi and CD25lo early effector cells

C57BL/6 mice received 1 to 10 × 104 SMARTA CD4+ T cells, followed by LCMV infection 1 day later. At days 3 or 5 post-infection, single-cell splenocyte suspensions were stained with a nondepleting biotinylated anti-CD25 antibody (eBio7D4, eBioscience) (51) for 20 min on ice in a magnetic-activated cell sorting staining buffer (PBS with 0.5% bovine serum albumin and 1 mM EDTA), followed by incubation with anti-Biotin MicroBeads (Miltenyi Biotec) for an additional 20 min on ice. CD25hi and CD25lo CD4+ T cells were separated via magnetic sorting columns (Miltenyi Biotec).

RNA sequencing

CD25hi and CD25lo SMARTA CD4+ T cells (Thy1.1+) were FACS-sorted from splenocytes 5 days after LCMV infection. RNA was then extracted using a kit (miRNeasy, Qiagen). Library preparation and RNA-seq were performed by the University of Utah DNA Sequencing Core Facility (NuGEN Ovation RNA-Seq System v2, HiSeq 50 Cycle Single Read) on a HiSeq 2000 (Illumina). A real-time analysis software (RTA v1.18.61) performed base calling and assigned a quality score to each base for each cycle. Reads were aligned to hg19 + splice junctions using NovoAlign. Spliced alignments were converted back to genomic space, sorted, and indexed using Useq 8.8.9 SamTranscriptomeParser. Differentially expressed genes were determined using DefinedRegionDifferentialSeq. Sequencing results were analyzed, and differences in gene expression were calculated by the University of Utah Bioinformatics Shared Resource. Sequencing raw data are available in the Gene Expression Omnibus (GEO) repository (GSE114884).

Reverse transcription polymerase chain reaction

RNA was isolated using TRIzol (Life Technologies) and converted to complementary DNA using SuperScript III First-Strand Synthesis System (Thermo Fisher Scientific). Semiquantitative RT-PCR reactions were carried out using Power SYBR Green PCR Master Mix (Life Technologies) and run on the LightCycler 480 (Roche).


Retroviral vectors (pMig-R1) were used to express shRNA KD constructs specific for SHP-1 (SHP-1 KD) as previously described (47). A human microRNA (mir30) flanking sequence allowed for optimal expression and processing of siRNA (52). KD was confirmed in an EL-4 thymoma cell line. KD in primary CD4+ T cells was accomplished by transducing SMARTA bone marrow with SHP-1 KD or EV retrovirus and transplanting into irradiated Rag1−/− mice. After reconstitution (8 to 10 weeks later), SMARTA CD4+ T cells (GFP+ and GFP) were then isolated from the spleen and transferred (1 to 2 × 104) into B6 recipient mice that were subsequently infected with LCMV the next day (53).

Statistical analysis

Data analysis was performed using Prism (GraphPad) software.


Materials and Methods

Fig. S1. Creation and stimulation of hybridoma T cell lines with NFAT and NFκB reporters.

Fig. S2. Stimulation and characterization of C7 and C26 transgenic CD4+ T cells in vitro.

Fig. S3. Analysis of CD25 expression in an endogenous CD4+ T cell repertoire early after viral infection.

Fig. S4. Characterization of C7 and C26 transgenic CD4+ T cells via adoptive transfer and LCMV infection.

Fig. S5. Early expression of CD25 by clone 18 retrogenic CD4+ T cells predicts effector and memory differentiation.

Fig. S6. SHP-1 KD validation and effect on TCR signaling in vitro.

Fig. S7. Flow cytometry gating strategies.

Table S1. Primary source data.

Table S2. Complete list of genes with significantly changed expression when comparing CD25lo to CD25hi day 5 SMARTA effector cells, as determined by RNA-seq.


Acknowledgments: We acknowledge T. Mosbruger and the Bioinformatics Shared Resource at the Huntsman Cancer Institute for assistance with deep sequencing and gene expression analysis. We acknowledge J. Marvin and the University of Utah Flow Cytometry Core Facility for assistance with the design and execution of cell purification by FACS. We acknowledge the NIH Tetramer Core Facility for providing reagents. Funding: Financial support for these studies was provided by the NIH (R01AI080830 and R01AI137248), the American Association of Immunologists Careers in Immunology Fellowship, and the University of Utah. Author contributions: All authors formulated the experimental design and performed major experiments and data analysis. J.P.S. performed the statistical analysis. J.P.S. and M.A.W. wrote and revised the manuscript and conceptualized results. Competing interests: The authors declare that they have no competing financial interests. Data and materials availability: Sequencing data have been placed in the GEO repository (GSE114884).

Stay Connected to Science Immunology

Navigate This Article