Different T Cell Receptor Signals Determine CD8+ Memory Versus Effector Development

See allHide authors and affiliations

Science  23 Jan 2009:
Vol. 323, Issue 5913, pp. 502-505
DOI: 10.1126/science.1163612


Following infection, naïve CD8+ T cells bearing pathogen-specific T cell receptors (TCRs) differentiate into a mixed population of short-lived effector and long-lived memory T cells to mediate an adaptive immune response. How the TCR regulates memory T cell development has remained elusive. Using a mutant TCR transgenic model, we found that point mutations in the TCR β transmembrane domain (βTMD) impair the development and function of CD8+ memory T cells without affecting primary effector T cell responses. Mutant T cells are deficient in polarizing the TCR and in organizing the nuclear factor κB signal at the immunological synapse. Thus, effector and memory states of CD8+ T cells are separable fates, determined by differential TCR signaling.

It is unclear what determines whether a primed T cell will differentiate into an effector cell and die or will survive and become a long-lived memory cell. Several studies support a linear differentiation model, where memory T cells are descendents of effector cells that differentiate into a memory lineage after antigen has been cleared (1, 2). However, memory T cells have been shown to arise directly from naïve cells, bypassing the effector stage (35). This second group of studies suggests that signaling differences may determine the fate of naïve T cells upon infection.

Proinflammatory cytokine signals can favor the generation of effectors versus the development of memory T cells after infection (6, 7). It has also been suggested that cumulative TCR stimulation is required for the generation of memory (8). Other studies have shown that a short period of TCR stimulation is sufficient for programming CD8+ T cells to differentiate into memory cells (9, 10). Differentiation and longevity of memory CD4+ T cells are dependent on high functional avidities to complete the memory program (11). Several studies have considered the importance of the T cell–APC (antigen-presenting cell) interface in determining the fate of naïve T cells (5, 12). However, much less is known about how and which TCR signals contribute to T cell memory development.

The TCR β-chain transmembrane domain (βTMD) contains a conserved antigen receptor transmembrane (CART) motif (13). The CART motifs in membrane immunoglobulin and in the TCR β chain are involved in the segregation of B cell and CD4+ T cell functions (1315). We generated OT-1 TCR transgenic mice expressing a point mutation in the βTMD, where the most carboxy-terminal tyrosine residue of the CART motif was replaced by a leucine (CART15 Y→L) (16, 17). TCR expression on mutant T cells was slightly decreased, but the mutant TCR-CD3 complex composition was unaltered (fig. S1). T cell maturation and homeostasis in βTMD mutant mice (βTMDmut) were normal (fig. S1). The OT-1 and βTMDmut OT-1 TCRs recognize the ovalbumin peptide 257 to 264 (OVAp) bound to H–2Kb (18). No differences were found when wild-type (WT) and mutant T cells were compared for OVA tetramer (OVAtet) binding and OVAp responsiveness (fig. S1). Thus, the point mutation in βTMD did not affect the ability of the TCR to recognize and respond to ligand.

Next, we studied the capacity of mutant T cells to respond to antigen in vitro. WT and mutant T cells were similar for expression of CD25, CD69, and Fas. However, mutant cells were defective in Fas ligand (FasL) expression (Fig. 1A), similar to Jurkat and murine CD4+ T cells, which was linked to impaired nuclear factor–κB (NF-κB) signaling (14, 15). We tested βTMDmut T cells for defects in the activation of this or other TCR signaling pathways. Nuclear translocation and DNA binding of NF-κB were greatly diminished in mutant T cells (Fig. 1, B and C, and fig. S2). However, neither extracellular signal–regulated kinase (ERK), nor c-Jun N-terminal kinase (JNK) phosphorylation, nor Ca2+ mobilization was impaired (Fig. 1D).

Fig. 1.

βTMDmut naïve T cells are defective in NF-κB signaling. (A) T cells stimulated with 2 μM OVAp-pulsed APCs in vitro, showing CD69, CD25, Fas, and FasL (P = 0.0024) expression shown as mean fluorescence intensity (MFI). Data represent the means ± SD of three independent experiments. (B) T cells stimulated as in (A) for 30 min. Vesicular stomatitis virus peptide (VSVp) was a negative control. NF-κB nuclear translocation determined by confocal microscopy. RelA (green), DRAQ-5 nuclear dye (blue), and CD45.1 (red). Images are representative of n > 50 conjugates from three independent experiments. (C) Nuclear extracts from T cells stimulated with OVAp were tested by enzyme-linked immunosorbent assay (ELISA) for RelA-specific binding to NF-κB consensus sequence; P = 0.0003. (D) T cells stimulated with OVAp or VSVp (control). ERK and JNK phosphorylation and Ca2+ flux were determined by flow cytometry.

To study effector development of βTMDmut T cells, low numbers (50 to 500) of CD45.2+ WT or mutant naïve T cells were transferred into congenic CD45.1+ recipient mice, followed by infection with recombinant Listeria monocytogenes–expressing OVA (LM-OVA). WT and βTMDmut transgenic T cells responded equivalently to LM-OVA, in terms of phenotype, kinetics, and function; this response was also similar to that of the endogenous OVA-responding CD8+ T cell population (Fig. 2A). At the peak of the primary response, WT and mutant T cells exhibited identical effector phenotypes (CD62low, interleukin receptor IL-7Rlow, CD43high, CD27int, granzymeBhigh, IL-2low) (Fig. 2B). On restimulation, interferon-γ (IFN-γ) and tumor necrosis factor–α (TNF-α) expression were similar for both, which indicated that βTMDmut T cells were normal in their effector functions (Fig. 2C). Analogous results were obtained when higher numbers of WT or mutant naïve T cells were used and after OVAp-LPS (lipopolysaccharide) immunization. Mutant and WT T cells were equally efficient at killing in vivo (fig. S3). In addition, there were no differences in CD45.1+ endogenous responders, WT, or mutant numbers at the peak of the primary immune response in LM-OVA–infected mice (Fig. 2D).

Fig. 2.

Normal effector differentiation in βTMDmut T cells. (A) CD45.2+ naïve T cells were transferred into B6CD45.1+ mice and immunized with 2 × 104 colony-forming units (CFU) LM-OVA 1 day later. Frequencies and IL-7R, CD62L, and granzymeB expression measured at the peak of the response, day 7. (B) Comparison for expression of markers indicated at day 7 p.i. (C) At day 7 p.i., T cells were restimulated, and IFN-γ or TNF-α expression was measured. (NS, nonstimulated.) Graphs show the percentage (means ± SD) of WT or mutant cells; P < 0.004. (D) CD45.2+ naïve T cells treated as in (A). Representative plots show frequencies at day 7 p.i. Graph shows number of OVA-specific T cells (means ± SD). (A to D) Data are representative of more than three independent experiments; n = three mice per group.

To study the expansion, contraction, and memory phases of the CD8+ T cell response, we examined lymphocytes from mice that received low numbers of WT or mutant naïve cells and then were infected with LM-OVA (Fig. 3A). βTMDmut T cells expanded and contracted with kinetics resembling those of WT cells. However, by day 13 to 14, mutant T cells numbers were severely reduced, whereas WT cell numbers contracted more slowly (Fig. 3A). Stable numbers of WT memory T cells were detected by day 46. In contrast, mutant memory cell numbers were at the limit of detection, although similar numbers of endogenous OVA-specific memory CD8+ T cells were generated (Fig. 3B). Clonal competition can affect differentiation and survival of T cells (19). To check this possibility, we gave mice higher numbers (105 to 106) of mutant naïve cells and primed them with LM-OVA or OVAp-LPS. Neither condition induced efficient generation of mutant memory T cells (fig. S3). Thus, although mutant T cells differentiated into effectors normally, their memory development was defective. Interleukin receptors IL-15R and IL-7R, important for the survival of CD8+ memory (20), were equally expressed on both cell types (fig. S3), which indicated that the defect in generating memory T cells was not based solely on competition for survival signals. Thus, the induction of a distinct set of TCR signals, absent in the mutant T cells, is likely required for memory generation.

Fig. 3.

Memory generation is impaired in βTMDmut T cells. (A) Kinetic analysis of primary (1°) responses to 2 × 104 CFU LM-OVA in the blood of host mice previously injected with 500 or 50 naïve T cells. Numbers represent the frequency of OVA-specific CD8+ T cells pooled from three mice. (B) CD45.2+ naïve T cells were transferred and immunized as described in Fig. 2A. frequencies of T cells at day 46 p.i. from representative plots. Graph shows the numbers (means ± SD) of representative data from four independent experiments; n = three mice per group; for the mutant P = 0.0005. (C) CD45.2+ naïve T cells were treated as in (A). Kinetic analysis of the secondary (2°) response in mice reinfected with 9 × 104 CFU LM-OVA 46 days p.i. Numbers represent the frequency of OVA-specific CD8+ T cells pooled from three mice; P < 0.0005. (D) Memory T cells were restimulated and IFN-γ and TNF-α expression are shown; P = 0.004. (NS, nonstimulated.)

To compare the functional properties of WT and βTMDmut memory T cells, we tested their ability to mount recall responses. Mice into which low numbers of WT or mutant naïve T cells were previously transferred were reinfected 46 to 50 days post LM-OVA infection. Compared with WT, βTMDmut memory T cells responded poorly to secondary challenge (Fig. 3C). In experiments performed with higher numbers of transferred T cells, mutant memory T cells were deficient in CD25, leukocyte function–associated antigen–1 (LFA-1), granzyme B, and FasL expression; in CD62L down-regulation; and in IFN-γ and TNF-α expression and secretion (Fig. 3D and fig. S4). βTMDmut memory T cells were also impaired in their cytotoxicity, and they failed to generate a secondary memory pool (fig. S4). Therefore, mutant memory T cells were severely compromised, not only in their numbers but also in their ability to mount secondary responses.

The programming of memory T cell differentiation was tested during the final stages of the primary response. In experiments with LM-OVA, mutant T cells were almost undetectable from day 10 to day 13 after primary immunization (Fig. 3A). However, when we used higher precursor frequencies and immunization with OVAp-LPS, the development of mutant memory T cells could be followed throughout the contraction phase (days 11 and 20). At the peak of the response, WT and mutant cells responded similarly upon antigenic rechallenge. However, by day 11 and day 20, rechallenged mutant T cells were impaired in their expression of several activation, cytolytic, and antiapoptotic proteins (fig. S3). This, together with the fact that mutant cell numbers did not stabilize at the end of the contraction phase (Fig. 3A), strongly indicates that naïve mutant T cells were defective in memory differentiation. Furthermore, considering that mutant cells behaved normally during the primary response, these data suggest that the TCR signals required for generating effector and memory T cells are distinct.

We tested another mutation within the CART motif of the βTMD domain by generating OT-1 TCR transgenic mice with a βTMD mutation in which Ala16 was replaced by Asn (CART16A→N). βTMD(A→N) T cells were also selectively impaired in memory development, confirming the βTMD (Y→L) phenotype (fig. S5).

To better understand the TCR signals required for memory development, we analyzed some of the biochemical properties of the mutant TCR in memory T cells. Tetramer-binding assays showed that TCRs expressed on WT and mutant memory T cells have similar ligand affinities (Fig. 4A). When stimulated with OVAtet- or OVAp-pulsed APCs, both memory cell types were equally able to mobilize calcium (Fig. 4B). Consequently, the mutant TCR was not blunted in recognizing antigen or in delivering some of the early signals in the context of a memory T cell.

Fig. 4.

βTMDmut T cells are impaired in polarizing TCR and PKC-θ to the immunological synapse. Naïve CD45.2+ WT or βTMDmut T cells were transferred and immunized with OVAp-LPS. On day 80 post primary immunization, (A) OVAtet binding was measured by Geomean fluorescence intensity (gmfi), and (B) calcium flux was determined. Data are representative of two independent experiments of three mice each. (C) Naïve T cells were stimulated as in Fig. 1A. TCR and PKC-θ localization at 30 min. Images are representative of T cell–APC conjugates (n > 50). DIC, differential interference contrast; arrows indicate the IS. Graphs show the kinetics of TCR (P = 0.0007) or PKC-θ recruitment (P = 0.04) to the IS. Data represent the percentage of molecules in IS (means ± SD) of 30 to 60 cells from five independent experiments.

T cells interacting with APCs form conjugates and concentrate signaling molecules at the immunological synapse (IS), which dictates the fate of a naïve T cell (5, 12). It is noteworthy that WT and mutant T cells were similar in maintaining stable interactions with OVAp-pulsed APCs (figs. S7 to S11). Additionally, proteins important for conjugate formation like LFA-1 and Scribble were polarized to the IS (fig. S7). However, the enrichment of the mutant TCR in the IS (Fig. 4C) and the percentage of mutant T cells polarizing their TCR (fig. S6) were greatly reduced. Because mutant naïve T cells were impaired in NF-κB induction (Fig. 1C) and the βTMD mutant TCR was unable to properly polarize to the IS (Fig. 4C), we hypothesized that assembly of NF-κB signal at the synapse may be affected. As protein kinase C, isoform θ (PKC-θ), is required for NF-κB activity and colocalizes with the TCR in the IS, we studied the recruitment of PKC-θ to the IS. PKC-θ enrichment within the IS was evident by 15 min in WT cells. However, PKC-θ was not efficiently recruited to the IS until 60 min of stimulation in mutant cells (Fig. 4C). Nevertheless, this did not rescue the induction of NF-κB (Fig. 1C). βTMD(A→N) mutant T cells showed a similar phenotype (fig. S5). Together these data emphasize the role of βTMD in polarizing the TCR within the IS and in inducing NF-κB signals required to generate efficient memory T cells.

It has been unclear whether the TCR generates similar signals for the development of effector and memory T cells. Our results and others' (35) suggest that effector and memory differentiation require a different set of signals. Our data are consistent with a two-lineage model where memory or effector development is determined very early during the immune response by coordinating the recruitment of fate-determining proteins at the level of the IS.

Our studies suggest that different T cell programs are triggered by qualitatively distinct TCR signals, which implies that unique signaling pathways are important for T cell memory development. Several molecules, such as B cell lymphoma–6 (Bcl-6), the B and T lymphocyte attenuator (BTLA), and methyl-CpG binding domain protein 2 (MBD2), are selectively important for memory development but not for effector differentiation (2123). Along the same lines, mutant T cells are uniquely defective in memory development and NF-κB signaling. Several studies have reported a role for members of the NF-κB signaling pathway in memory development (24, 25).

Our studies emphasize the importance of the TCR in regulating the NF-κB signal required for memory development. We show here that effector and memory programming can be dissociated by the induction of a different arrangement of TCR signals in CD8+ T cells. Studying how these TCR signals are modulated by inflammatory signals or CD4+ help will be important in the design of better vaccination regimes.

Supporting Online Material

Materials and Methods

Figs. S1 to S7


Movies S1 to S4

References and Notes

View Abstract

Stay Connected to Science

Navigate This Article