N-ras couples antigen receptor signaling to Eomesodermin and to functional CD8⁺ T cell memory but not to effector differentiation

Several lines of evidence suggest that N-ras is a particularly important Ras isoform in T cell lines (Perez de Castro et al., 2004), thymocytes, and for polarization of mature CD4⁺ T lymphocytes. To test whether N-ras is necessary for mature CD8⁺ T lymphocyte activation after antigen stimulation, we generated N-ras^−/− mice expressing the transgenic OT-I TCR specific for OVA peptide ²⁵⁷SIINFEKL²⁶⁴ presented by H-2K^b. We first assessed early TCR signaling events after antigen stimulation in vitro. Similar to what has been previously shown for N-ras^−/− thymocytes (Pérez de Castro et al., 2003; Iborra et al., 2011), N-ras deficiency barely affected extracellular signal-regulated kinases (ERK) activation (Fig. 1 A). In contrast, phosphorylation of AKT at T306, a phosphoinositide 3-kinase (PI3K)–dependent event, was clearly deregulated in the absence of N-ras. Accordingly, nonstimulated N-ras^−/− cells exhibited increased basal levels of phospho-AKT compared with wild-type counterparts (Pérez de Castro et al., 2003), and upon exposure to antigen they were unable to sustain it (Fig. 1 A).

After stimulation with 10⁻¹⁰ M OVA peptide, N-ras^−/− OT-I CD8⁺ T lymphocytes were comparable to WT cells in terms of up-regulation of activation markers (CD69, CD25, and CD44) and down-regulation of CD62L, but less sensitive to decreasing concentrations of either OVA (Fig. 1, B and C) or the weaker altered peptides Q4R7 and Q4H7 (Daniels et al., 2006), derived from the original OVA peptide (Fig. 1 C). Also, when stimulated with OVA peptide, N-ras^−/− OT-I CD8⁺ T cells exhibited IL-2 production (Fig.1 D) and proliferation (Fig. 1 E) comparable to WT counterparts. Importantly, regardless of the abundance of antigen upon restimulation, both cell types displayed similar IFN-γ secretion (Fig. 1 F) and cytotoxic activity (Fig. 1 G). Thus, despite reduced sensitivity to antigen and impaired early AKT activation, N-ras^−/− CD8⁺ T cells were capable of activation and acquisition of hallmark effector functions.

Next, we sought to determine the role of N-ras in CD8⁺ T lymphocyte primary responses in vivo. We infected WT and N-ras^−/− mice with a recombinant vaccinia virus (VACV) expressing full length OVA protein (rVACV-OVA) and analyzed the CD8⁺ T lymphocyte responses to OVA and VACV immunodominant ²⁰TSYKFESV²⁷ (B8R) peptides. At the peak of the primary response, day 7, the frequency of antigen-specific CD8⁺ T cells in the spleen and blood was as high in the spleen and higher in blood in N-ras^−/− compared with WT mice when assessed by ex vivo K^bB8R pentamer staining (Fig. 2 A) or by intracellular IFN-γ expression (Fig. 2 B), with N-ras^−/− cells producing amounts of IFN-γ on a per cell basis equivalent to those of WT counterparts (Fig. 2 B, far right). The cytotoxic activity of CD8⁺ T lymphocytes in vivo was also unaffected by the absence of N-ras, as both type of mice cleared peptide-bearing cells with the same efficiency (Fig. 2 C). Similar results were obtained for the CD8⁺ T cell response to OVA peptide (not depicted). Furthermore, up-regulation of KLGR1 and down-regulation of CD62L, events associated with acquisition of a terminal effector phenotype, were moderately reduced or similar, respectively, in N-ras^−/− cells compared with WT counterparts (Fig. 2 D, left). Sustained expression of receptors for IL-15 and IL-7, a distinctive feature of putative memory precursors among antigen-responding cells (Williams and Bevan, 2007), was differentially affected by the N-ras deficiency. IL-2Rβ/IL-15Rβ was equally expressed on both cell types. On the contrary, the number of CD127⁺ cells was lower in N-ras^−/− mice at the peak of the primary response (Fig. 2 D, right), but they gradually reached the levels of wild-type cells in the weeks following primary infection (not depicted).

Both N-ras^+/+ and N-ras^−/− mice were able to similarly control VACV infection and were thus exposed to a similar antigen load (Fig. 2 E). Furthermore, after immunization with B8R and OVA peptide-loaded DC, which selectively induce the CD8⁺ compartment while providing a less inflammatory priming environment (Badovinac et al., 2005), both types of mice also generated a fully functional primary response (Fig. 2, F and G). Thus, N-ras seemed to be dispensable also in vivo for acquisition of the two main functional hallmarks of effector CD8⁺ T lymphocytes, namely IFN-γ production and cytotoxic activity.

Furthermore, we studied the memory response to VACV infection or DC immunization and found that the frequencies and numbers of B8R-specific IFN-γ–producing CD8⁺ T cells at the memory stage were mildly but consistently lower in infected (Fig. 3, A and B) or similar in DC-vaccinated N-ras^−/− mice (Fig. 3, D and E). In both settings, however, N-ras^−/− mice were clearly less efficient than WT animals in eliminating B8R-bearing splenocytes in vivo (Fig. 3, C and F).

Next, we assessed the ability of N-ras^−/− CD8⁺ T lymphocytes to mount rapid and potent recall responses against challenge infection, a hallmark of memory lymphocytes. A defective secondary response was found in N-ras^−/− mice immunized with peptide-coated DCs. When vaccinated WT mice were challenged with virus, the number of splenic and peritoneal B8R- and OVA-specific CD8⁺ T cells was increased ∼12–23-fold with respect to nonpulsed DC as controls. Remarkably, in N-ras^−/− mice, the number of CD8⁺ T cells specific for either antigen increased only two- to threefold as a consequence of priming (Fig. 4 A). Next, we tested the impact of N-ras deficiency on the protection against VACV challenge infection provided by memory CD8⁺ T lymphocytes. N-ras^+/+ and N-ras^−/− mice vaccinated with B8R-pulsed or nonpulsed DCs were infected in the ear pinnae with VACV (Tscharke and Smith, 1999). In unprimed mice, the development of dermal lesions and their resolution was not affected by N-ras deficiency (Fig. 4 B). In WT mice, vaccination with B8R-pulsed DCs significantly reduced dermal pathology, and virus titer in the infected ear was reduced ∼500-fold on day 6 (Fig. 4 C). In contrast, ear lesion development in B8R-DC–vaccinated N-ras^−/− mice was closer to that of the unprimed controls throughout the infection, and virus titer was reduced only 10-fold (Fig. 4, B and C). The reduced frequency of B8R-specific CD8⁺ T lymphocytes in the respective retromaxillar draining lymph nodes after VACV challenge (Fig. 4 C, bottom) correlated with the impaired protection caused by N-ras deficiency. Collectively, these data indicate that N-ras signaling is required for the secondary response of memory CD8⁺ T lymphocytes that mediate protective immunity against a challenge infection.

For the analysis of the primary response, small numbers of purified N-ras^+/+ and N-ras^−/− OT-I CD8⁺ T lymphocytes were cotransferred into the same congenic WT recipient. Upon primary infection, splenic N-ras^−/− OT-I cells expanded better (Fig. 5 A) but were similar to WT OT-I cells in differentiating into IFN-γ–producing cells (Fig. 5 B). These results would also indicate that homing to the spleen is not affected by N-ras deficiency, as previously described for naive CD8⁺ T cells in N-ras–deficient mice and in OT-I TCR transgenic mice lacking N-ras (Pérez de Castro et al., 2003; Iborra et al., 2011).

Next, for the analysis of the secondary response, we transferred purified N-ras^+/+ and N-ras^−/− OT-I cells into WT congenic recipients and immunized them with DCs pulsed with the OVA peptide (Fig. 5 C). As expected from Fig. 2 F, both types of OT-I cells exhibited a comparable primary response. DC priming in a noninflammatory environment accelerates CD8⁺ T lymphocyte memory differentiation (Badovinac et al., 2005), quickly promoting their ability to undergo secondary expansion upon rechallenge. In addition, this approach was used because DC immunization did not result in any differences in the number of CD8⁺ T lymphocytes homing to the spleen (at 35 d post infection [p.i.] in Fig. 3 E, and not depicted for earlier times after DC immunization). Therefore, 8 d later, N-ras^+/+ and N-ras^−/− OT-I cells were isolated and cotransferred into new congenic recipients that were subsequently infected with rVACV-OVA. 5 d after challenge, we found higher frequencies as well as numbers of WT OT-I splenic cells than of N-ras^−/− OT-I cells (Fig. 5 C). Therefore, the N-ras deficiency clearly caused a CD8⁺ T cell–intrinsic defect in the secondary response of memory-fated cells.

A role for mTOR in regulating effector versus memory CD8⁺ T cell development has been recently unveiled with the use of mTORC1 inhibitor rapamycin (Araki et al., 2009; Pearce et al., 2009). We therefore explored whether the memory-promoting effect of rapamycin in vivo could rescue the defect in generation of functional memory caused by the N-ras deficiency (Fig. 6). To this end, rapamycin-treated or untreated N-ras^+/+ and N-ras^−/− OT-I cells were adoptively transferred into the same recipient mice, which were then immunized with antigen-loaded DC in the presence or absence of rapamycin and later infected with virus (as depicted in Fig. 5 C). Rapamycin treatment consistently ameliorated the secondary expansion of antigen-specific N-ras^−/− cells, although only partly compared with WT counterparts treated with the inhibitor (Fig. 6 A). Nonetheless, while there was no differential effect from rapamycin on the antigen-driven primary expansion of OT-I CD8⁺ T cells of either genotype, rapamycin-treated N-ras^−/− cells expanded ∼2.5-fold more than wild-type cells after secondary antigen challenge (Fig. 6 B).

Next, we determined the effects of rapamycin treatment in vivo on the memory response to rVACV-OVA infection. The numbers of B8R-specific and OVA-specific IFN-γ–producing CD8⁺ T cells were significantly increased in rapamycin-treated compared with untreated N-ras^−/− mice, but to a lesser extent than in rapamycin-treated WT counterparts (Fig. 6 C). Similar results were obtained when the in vivo cytotoxic activity of memory cells was analyzed. Again, rapamycin ameliorated the capacity of N-ras^−/− mice to eliminate antigen-bearing splenocytes in vivo (Fig. 6 D, left). Of note, the proportional improvement by rapamycin of the memory cytotoxic response was superior in N-ras^−/− than in WT mice (Fig. 6 D, right).

We also tested the impact of rapamycin treatment on the recall response in DC-immunized mice. Also in this respect, rapamycin enhanced the expansion of responding memory CD8⁺ T cells (Fig. 6 E) and protective memory (Fig. 6 F) in N-ras^−/− mice, but only partly in comparison with WT mice.

In all settings where CD8⁺ memory was assessed, the enhancing effects of rapamycin were more prominent in N-ras^−/− than in WT mice or cells, but rapamycin-treated WT cells showed higher responses than their treated N-ras^−/− counterparts. This supports the notion that the N-ras deficiency could affect other key events involved in memory differentiation that are not negatively regulated by mTOR.

It has been proposed that the balance in the expression of transcription factors T-bet and Eomes impacts the effector versus memory T cell commitment (Intlekofer et al., 2005; Banerjee et al., 2010). To test the role of N-ras in the control of Eomes expression in differentiating CD8⁺ T lymphocytes in vivo, and the effect of inhibiting mTOR by rapamycin, naive N-ras^+/+ and N-ras^−/− OT-I cells, untreated or treated with rapamycin, were adoptively cotransferred into WT mice. Recipient mice were untreated or treated with the inhibitor and infected with rVACV-OVA. OT-I T cells were analyzed 7 d later ex vivo for expression of T-bet and Eomes (Fig. 7 A). T-bet expression was comparable in cells of both genotypes and mostly unaffected by rapamycin. In contrast, Eomes expression was significantly lower in N-ras^−/− OT-I cells, and partly rescued by rapamycin treatment, as the rapamycin effect was superior in WT OT-I cells (Fig. 7 A). Similarly, K^bB8R⁺ CD8⁺ T lymphocytes induced in N-ras^−/− mice by VACV infection showed a consistently lower level of Eomes expression than their WT counterparts after primary infection, which was also rescued by rapamycin (Fig. 7 B). Furthermore, after viral challenge of DC-immunized mice, Eomes levels remained significantly lower in N-ras^−/− compared with WT responding memory cells (Fig. 7 C). Together, these results suggest that N-ras^−/− CD8⁺ T lymphocytes harbor a cell-intrinsic defect affecting Eomes but not T-bet expression during distinct stages of the in vivo CD8⁺ response.

To further test whether N-ras is involved in regulating T-bet and Eomes in antigen-stimulated CD8⁺ T lymphocytes at earlier times, we analyzed induction of these transcription factors after exposure to antigen in vitro. As shown in Fig. 7 (D and E), the magnitude of T-bet induction was comparable in N-ras–sufficient and –deficient OT-I cells. In overt contrast, Eomes expression was consistently impaired in N-ras^−/− OT-I cells after stimulation with antigen, which correlated with the memory defect of N-ras^−/− mice.

Abundance of cognate antigen and the strength of TCR signals have been shown to impact effector versus memory fate decision (Zehn et al., 2009; Corse et al., 2011; Leignadier et al., 2011). However, the influence of these factors in the expression of fate-determining transcription factors, in particular that of T-bet and Eomes, had not been evaluated in detail. To test this, naive N-ras^+/+ and N-ras^−/− OT-I cells were stimulated with the strong OVA peptide agonist or with the weaker altered peptide ligands, Q4R7 and Q4H7. We tested concentrations promoting maximal (EC100) or half-maximal (EC50) CD69 induction response for each genotype and peptide (as determined in Fig. 1 C). By 72 h (Fig. 7 E), T-bet induction was comparable in N-ras–sufficient and –deficient OT-I cells irrespective of peptide affinity or concentration. In contrast, Eomes expression was always induced to a lesser extent in N-ras^−/− compared with N-ras–sufficient cells. Collectively, these results confirm the strong dependence of Eomes expression on N-ras–mediated signaling and independently of the nature or abundance of the antigen.

In T lymphocytes, early TCR-triggered activation of Ras results in downstream coupling to the ERK and PI3K pathways. To test the involvement of these pathways in the regulation of Eomes expression by the N-ras isoform, naive N-ras^+/+ OT-I cells were stimulated with antigen in the presence or absence of PD98059, an inhibitor of MEK-1, an upstream activator of ERK kinases, or of wortmannin, a general PI3K inhibitor (Fig. 8 A). Inhibition of ERK activation had no major effect on Eomes induction, consistent with the lack of effect of N-ras deficiency on ERK activation (Fig. 1 A).

In contrast, PI3K inhibition by wortmannin strongly suppressed antigen-mediated Eomes induction, mimicking the effect of the N-ras deficiency (Fig. 8 A). Together with data showing a deregulated antigen-induced AKT activity in N-ras^−/− cells (Fig. 1 A), these results suggest that N-ras acts as a critical inducer of early Eomes expression via the PI3K–AKT pathway. To gain further insight, we analyzed the effect of specific PI3K and AKT pharmacological inhibitors on Eomes induction after antigen exposure (Fig. 8 B). Inhibition of PI3Kγ or AKT resulted in a marked decrease in Eomes induction, which was comparable to the effect of LY294002, a general PI3K inhibitor. Furthermore, interfering PI3Kα activity affected Eomes induction but only at the highest dose of inhibitor tested, and to a lesser extent than did PI3Kγ and AKT inhibitors, whereas PI3Kβ inhibition and ERK inhibition by UO126 had no significant effect. Together, these results suggest that PI3Kγ, most likely via AKT, but not the ERK pathway, is involved in regulating early Eomes expression in antigen-activated CD8⁺ T cells.

It has been suggested that mTOR activity regulates the balance between T-bet and Eomes expression in CD8⁺ T cells in response to antigen plus IL-12 (Rao et al., 2010). Moreover, the Ras pathway has recently been involved in TCR-induced mTOR activation (Gorentla et al., 2011), although the relevant Ras isoform was not identified. These findings prompted us to carefully analyze the kinetics and magnitude of T-bet and Eomes induction, and the effect of rapamycin, in N-ras–sufficient and –deficient CD8⁺ T cells at early and late times after antigen exposure in vitro (Fig. 9).

Rapamycin had no effect on T-bet expression in WT or N-ras^−/− OT-I cells stimulated only with antigen (Fig. 9 A) and, as previously shown for WT cells (Rao et al., 2010), reverted the IL-12–induced enhancement of T-bet expression in both cases (Fig. 9 B). In addition, in WT OT-I cells, rapamycin treatment had no effect on Eomes expression except at 72 h after antigen exposure (Fig. 9 A), suggesting that early Eomes induction occurs in an mTOR-independent manner. Only when Eomes expression was poorly or not induced in N-ras^−/− OT-I cells (Fig. 9 A) or in either cell genotype stimulated in the presence of IL-12 (Fig. 9 B) did inhibition of mTOR consistently ameliorate Eomes expression at late time points (48–72 h). Thus, early control of Eomes induction in antigen-responding CD8⁺ T lymphocytes required N-ras and seemed to involve signaling pathways other than mTOR. These results correlate with the partial rescue of low Eomes expressing N-ras^−/− CD8⁺ T cells by rapamycin in vivo. However, later during CD8⁺ T cell priming, N-ras and mTOR seem to converge at the level of Eomes to regulate its expression in an opposite manner.

To test whether the defect in Eomes induction of N-ras^−/− CD8⁺ T cells was associated with their intrinsic defect in generating functional memory cells, expression of Eomes was forced in N-ras^−/− T lymphocytes. To this end, wild-type or N-ras^−/− OT-I cells were activated with OVA peptide through the TCR, transduced with a retrovirus expressing Eomes and GFP, and adoptively transferred into recipient mice that were then infected with rVACV-OVA. As shown in Fig. 10, during the primary response transduced GFP⁺ and transferred CD45.1⁺ CD8⁺ T cells could be detected among PEC, and there was no difference in percentage between wild-type and N-ras^−/− cells or between those transduced with control and with Eomes-expressing retroviruses (Fig. 10 A, top). After a secondary challenge with rVACV-OVA (Fig. 10 A, bottom), both transduced and nontransduced transferred wild-type OT-I cells expanded, as anticipated, and Eomes overexpression only conferred a minor benefit (Fig. 10 B, left). As also expected, N-ras^−/− OT-I CD8⁺ cells were not able to expand, whether untransduced or transduced with MigR1 control retrovirus. In sharp contrast, forced expression of Eomes in these antigen-specific N-ras^−/− cells allowed a highly significant secondary expansion in response to virus infection (Fig. 10, A and B, bottom). Similar results were obtained when responding cells in the spleen were analyzed. Again, forced expression of Eomes in N-ras^−/− CD8⁺ T cells allowed their otherwise defective secondary expansion in response to viral infection, with the proportional improvement being superior in N-ras^−/− cells to that in WT counterparts during the secondary but not the primary response (Fig. 10 B, right). Collectively, this data strongly supports a role for N-ras in coupling antigen receptor signaling to regulation of Eomes expression and to memory programming in CD8⁺ T lymphocytes.

During viral infections, antigen-specific naive CD8⁺ T lymphocytes give rise to both short-lived effectors and long-lived memory lymphocytes that contribute to the protective adaptive response. Many questions remain, however, as to the underlying molecular mechanisms and the temporal window when the effector versus memory fate decision is made. In particular, the nature and timing of TCR signals specifically required for programming the memory fate of CD8⁺ T lymphocytes have remained elusive.

Evidence exists that signaling differences may determine the fate of naive CD8⁺ T lymphocytes upon activation (Lauvau et al., 2001; Badovinac et al., 2005; Daniels et al., 2006; Tewari et al., 2006; Chang et al., 2007; Teixeiro et al., 2009). Several such fate-determining molecules regulating memory but dispensable for effector CD8⁺ T lymphocyte differentiation have been identified, and most of them are distal or unrelated to TCR-triggered pathways. These include surface receptor BTLA (Krieg et al., 2007), intracellular signaling molecules such as TRAF-6 and the mTOR kinase (Araki et al., 2009; Pearce et al., 2009), and transcriptional regulators such as Bcl-6, Eomes, MBD2, Id2, and TCF-1 (Ichii et al., 2002; Intlekofer et al., 2005; Cannarile et al., 2006; Kersh, 2006; Zhou et al., 2010).

In the present work, we report that naive N-ras^−/− antigen-responding CD8⁺ T lymphocytes were able to efficiently expand and differentiate into cytolytic effectors but not into memory cells with protective function. As previously shown for CD4⁺ T cells (Iborra et al., 2011), CD8⁺ T cells from N-ras–deficient mice exhibited reduced sensitivity to antigen. However, this was seemingly without effect on antigen-mediated primary T cell expansion and acquisition of effector functions, either in vitro or in vivo. This is in agreement with previous reports (Zehn et al., 2009; Leignadier and Labrecque, 2010) and supports the existence of compensatory mechanisms enhancing antigen sensitivity in antigen-experienced T cells (Kumar et al., 2011). N-ras deficiency impinged instead on both the quantity and functional quality of memory cells, but more prominently on the latter, and severely compromised their protective antiviral function. This is consistent with previous work emphasizing the role of quality of memory T lymphocytes for efficient protective immunity (Lauvau et al., 2001). N-ras would thus partake in a CD8⁺ T lymphocyte memory differentiation program as a TCR-proximal and specific transducer of early signals.

The balance of expression of transcription factors Eomes and T-bet has been reported to influence whether CD8⁺ T lymphocytes commit to memory or effector cells (Intlekofer et al., 2005). High T-bet expression seems to favor effectors (Matsuda et al., 2007; Yeo and Fearon, 2011), whereas persistent Eomes expression is proposed to enhance memory lymphocyte differentiation (Joshi et al., 2007; Rao et al., 2010). This has been determined in response to IL-2 (Pipkin et al., 2010) or in contexts with designed antigen-presenting cells and with cytokines providing inflammatory (IL-12) or survival (IL-7 and IL-15) signals that may predominate over and potentially obscure early TCR signals (Intlekofer et al., 2005; Rao et al., 2010). When we focused on early antigen-mediated activation of naive CD8⁺ T cells in our in vitro system without added cytokines or costimulatory signals, we found that TCR-mediated induction of T-bet was independent of N-ras. Of interest, unlike antigen-exposed CD8⁺ T cells, activated CD4⁺ T lymphocytes required N-ras for T-bet induction and IFN-γ production (Iborra et al., 2011), suggesting that N-ras could be differentially required for TCR-elicited functions in CD4⁺ and CD8⁺ T cell lineages or depend on other concurrent signals.

With regard to Eomes, TCR-mediated signals inducing its expression and their relevance to effector versus memory fate choice had remained unclear. TCR engagement results in activation of the PI3K–AKT and ERK pathways downstream of active Ras (Smith-Garvin et al., 2009), although which Ras isoforms are specifically involved was unknown (Mor and Philips, 2006). We now report that antigen-responding CD8⁺ T cells lacking N-ras have a major defect in Eomes induction in vitro and in vivo. A similar defect is revealed when PI3K, particularly PI3Kγ, and AKT are pharmacologically inhibited in WT cells in vitro. Because antigen-stimulated N-ras^−/− CD8⁺ T cells also showed an intrinsic impairment in sustaining PI3K–AKT activation, our data suggest that N-ras, through PI3Kγ–AKT, mediates TCR signals required for up-regulation of Eomes, and thus for concomitant programming of the memory phenotype. Nonetheless, the involvement of other PI3K family members (Delgado et al., 2009) in Eomes regulation cannot be presently excluded. In contrast, in antigen-responding CD8⁺ T cells, we observed that N-ras deficiency had no major effect on ERK activation, suggesting that the ERK pathway is dispensable for Eomes induction. Thus, N-ras would clearly stand out among other Ras isoforms, having much less impact on ERK than on the PI3K–AKT pathway, and thereby dominating the induction of Eomes and of memory in TCR-activated CD8⁺ T cells.

A critical role for mTOR in determining whether CD8⁺ T lymphocytes adopt an effector versus memory cell fate was identified using the mTORC1 inhibitor rapamycin (Araki et al., 2009; Pearce et al., 2009). Furthermore, rapamycin has been reported to inversely affect Eomes and T-bet levels in CD8⁺ T cells following antigen stimulation in the presence of IL-12, with the inhibitor favoring the persistence of Eomes over T-bet (Rao et al., 2010; Li et al., 2011). We found that rapamycin promoted Eomes expression in vitro while leaving T-bet unaffected, in both WT and N-ras^−/− CD8⁺ T cells, but importantly, this happened only at late time points after exposure to antigen. In vivo, rapamycin treatment also enhanced Eomes expression in WT and N-ras^−/− CD8⁺ T cells. However, Eomes expression in treated N-ras^−/− cells rarely reached the levels of treated WT cells. This resulted in the partial rescue by rapamycin of each of the CD8⁺ T cell defects associated with N-ras deficiency in vivo. These concordant results may reflect a late downstream convergence of signaling pathways involving N-ras and mTOR at the level of Eomes, with N-ras and mTOR acting in opposite manners on Eomes expression. In this scenario, rapamycin treatment during priming would allow N-ras^−/− CD8⁺ T cells to reach a minimal threshold of Eomes expression for memory commitment. However, because of the earlier, mTOR-independent defect on Eomes induction impinged by the N-ras deficiency, the mTOR inhibitor would only partially rescue the memory defect of N-ras^−/− mice. Alternatively, N-ras could critically regulate, apart from Eomes, other memory fate-determining molecules in CD8⁺ T cells.

To date, only a handful of positive regulators of Eomes have been identified in CD8⁺ T cells, including Runx3 (Cruz-Guilloty et al., 2009), Notch (Cho et al., 2009), and TCF-1 (Zhou et al., 2010). Our study identified N-ras as a novel inducer of Eomes expression, but in contrast to the aforementioned ones, N-ras was distinctively required for programming memory but not effector CD8⁺ T cell fates. Furthermore, among the positive regulators of Eomes, N-ras would be the most proximally linked to the TCR. Suppression of Eomes has recently been shown to enhance IL-17 expression and Th17 differentiation (Ichiyama et al., 2011), which resembles the phenotype of TCR-activated CD4⁺ T lymphocytes in N-ras^−/− mice (Iborra et al., 2011; unpublished data). This further supports the notion that N-ras, but not the other Ras isoforms, specifically couples TCR signaling and induction of Eomes.

A recent study in a fluorescent Eomes reporter mouse (Paley et al., 2013) showed that expression of Eomes was very stable along the effector-to-memory CD8⁺ T cell differentiation after acute viral infection. Notably, in agreement with previous reports (Intlekofer et al., 2005; Banerjee et al., 2010), Eomes was found to be equally expressed in memory precursors and terminal effectors. However, Eomes expression correlated with improved central memory formation but not with enhanced primary effector response. Consistent with this, we found that retrovirally forced expression of Eomes in N-ras^−/− antigen-specific CD8⁺ T cells rescued their defective secondary expansion in response to viral infection while having no effect on their primary response. Collectively, these data support a direct, nonredundant and specific role of N-ras in coupling antigen receptor signaling to early Eomes induction and thus to memory development in CD8⁺ T lymphocytes. Our data also further support Eomes as a marker to reliably identify the memory potential of CD8⁺ T cells early after their response to antigen.

Because N-ras was dispensable for the generation and function of primary effector cells, our findings collectively support the concept that the TCR recruits distinct signaling pathways for programming effector and memory fates. Furthermore, they suggest that isoform-specific Ras inhibitors could be of value for specific control of pathogenic memory T lymphocytes, or alternatively as anti-cancer therapeutics without detrimental side effects on patients’ immunity.

N-ras^+/+ mice (C57BL/6 strain) were purchased from Charles River. N-ras^−/− mice (H-2^b haplotype; Umanoff et al., 1995) were bred with C57BL/6 for six generations and subsequently crossed with OT-I TCR transgenic mice to generate OT-I TCR transgenic N-ras^−/− mice. All animal studies were approved by CSIC and Instituto de Salud Carlos III’s Review Boards and were performed in accordance with national regulations.

Peptides ²⁵⁷SIINFEKL²⁶⁴ from OVA and ²⁰TSYKFESV²⁷ (B8R) from VACV soluble IFN-γR homologue (Moutaftsi et al., 2006) were synthesized in an Applera peptide synthesizer model 433A, purified, and determined to be homogeneous by HPLC analysis. OVA variant peptides Q4R7 (SIIQFERL) and Q4H7 (SIIQFEHL) were a gift of H. Van Santen (Centro de Biología Molecular Severo Ochoa, Madrid, Spain).

The Pro5 MHC class I H-2K OVA and H-2K^bB8R pentamers and FITC–anti-CD8α and anti-CD19 were obtained from ProImmune. PE–anti-CD8α, PE–anti-CD69, APC–anti-CD25, PE–anti-CD44, APC–anti-CD62L, PE–anti-IL-2, PE–anti–IFN-γ, biotinylated anti-CD45.1, APC–anti-CD45.1, PerCP-Cy5.5–anti-CD45.2, PE–anti-AKT phosphorylated at Thr 308, and PE–anti-ERK1/2 (pT202/pY204) were all anti–murine proteins antibodies (BD). PE–anti–mouse Eomes and PerCP-Cy5.5–anti–mouse/human T-bet were obtained from eBioscience. Antibody to mTOR phosphorylated at Ser 2481 was from Cell Signaling Technology.

Intracellular staining for IL-2 after primary activation of OT-I CD8⁺ T lymphocytes was performed after a 3-h stimulation with peptide-pulsed DCs. The co-cultures were then incubated overnight in the presence of brefeldin A before staining for FACS according to conventional procedures. In the ex vivo assays, cells were stimulated with an excess of peptide (1 µM) for up to 2 h and stimulated for a further 4 h in the presence of brefeldin A. Cells were then stained with FITC–anti-CD8α, fixed, and incubated with PE–anti–IFN-γ during permeabilization (Dako; Johnstone et al., 2004).

The intracellular staining for phosphorylated ERK, AKT, and mTOR, was performed as previously described (Krutzik and Nolan, 2003). In brief, after surface staining, the cells were fixed with formaldehyde, permeabilized with methanol, and stained with phosphospecific antibodies. The intracellular staining of T-bet and Eomes was performed using the Foxp3 fixation/permeabilization buffer (eBioscience), according to the protocol recommended for detecting nuclear antigens.

Events were acquired using a FACSCanto flow cytometer and the data were analyzed using FACSDiva software (BD). Representative dot plots or histograms of an individual mouse per group are shown in the figures. Percentage of stained cells was calculated and is indicated within dot plots. Percentage and mean fluorescence intensity (MFI) data from sets of experiments are graphed as the mean ± SEM. An average of 10,000 CD8⁺ cells was analyzed in each sample. Background activation obtained with nonpulsed cells (0–0.3%) was subtracted.

DCs were generated from BM progenitors. Freshly prepared BM cells were cultured in the presence of 200 U/ml GM-CSF (PeproTech) and fed with GM-CSF on days 3 and 6. After 7 d, nonadherent cells with a typical DC morphology and a myeloid DC phenotype (MHC class II⁺, CD11c⁺, and CD8⁻) were collected (Medina et al., 2009). Maturation of DCs was performed by the addition of 0.5 µg/ml LPS to cultures on day 7. Mature DCs (mDCs) showed a mature morphology and up-regulated CD40 and CD86 costimulatory molecules and were collected 1 h later for immunizations or 16 h later for other purposes.

CD8⁺ T lymphocytes were isolated from the spleens of OT-I TCR transgenic N-ras^+/+ and N-ras^−/− mice by negative selection using Miltenyi magnetic beads. 1 × 10⁶ CD8⁺ T cells were stimulated with either 5 × 10⁶ WT DCs pulsed with titrated concentrations of OVA or agonist peptides or with 6 × 10⁶ CD45.1⁻ CD45.2⁺ WT splenocytes pulsed with 10⁻⁹ M OVA peptide. For some experiments, 2 ng/ml IL-12 (R&D Systems) was added to the cultures. When indicated, cells were pretreated 20 min before stimulation and kept thereafter with 20 nM PI3K inhibitor wortmannin (Sigma-Aldrich), 50 µM ERK kinase (MEK) inhibitor PD98059 (EMD Millipore), 10, 5, and 2.5 µM ERK kinase (MEK) inhibitor UO126 (Cell Signaling Technology), 10, 5, and 2.5 µM PI3K inhibitor LY294002 (Cell Signaling Technology), 250, 125, and 62.5 nM PI3Kα inhibitor VII (EMD Millipore), 1, 0.5, and 0.25 µM μM PI3Kβ inhibitor VI (EMD Millipore), 10, 5, and 2.5 µM PI3Kγ inhibitor (EMD Millipore), 1, 0.5, and 0.25 µM AKT inhibitor VIII (EMD Millipore), or 2.2 nM mTORC1 inhibitor rapamycin (Sigma-Aldrich).

Proliferation was assessed by labeling OT-I CD8⁺ T lymphocytes with 5 µM CFSE (Molecular Probes) and coculturing them with mDC prepulsed with 10⁻⁷ M OVA peptide. CFSE dilution was analyzed 72 h later.

CTL lines were generated by culturing purified OT-I CD8⁺ cells with 10⁻⁹ M SIINFEKL-pulsed DCs for 72 h at an E/T ratio of 5/1 before adding human rIL-2 (NCI Preclinical Repository). The CTL lines were then used 2 d later for in vitro ⁵¹Cr-release cytotoxicity assays at an E/T ratio of 5/1 (Samino et al., 2004).

VACV strain WR and the recombinant virus encoding full-length OVA (rVACV-OVA) based on the WR strain were provided by J.W. Yewdell and J. Bennink (National Institutes of Health, Bethesda, MD). Stocks were grown in CV-1 monolayers and consisted of clarified sonicated cell extracts.

Mice were infected either i.p. with 10⁶ PFU of rVACV-OVA or intradermally in the ears with 5 × 10⁴ PFU of VACV WR (Tscharke and Smith, 1999). In this case, the evolution of the infection was monitored by measuring the diameter of the ear lesion with a digital caliper. For DC immunizations, mice were inoculated i.p. with a mixture of 5 × 10⁵ mDCs pulsed with 1 µM B8R peptide and 5 × 10⁵ mDCs pulsed with 1 µM OVA peptide, or with mDCs pulsed with only one of the peptides.

For virus titration ex vivo, the ventral and dorsal dermal sheets of infected mouse ears were separated using forceps and incubated with 50 µg/ml liberase CI (Sigma-Aldrich) for 1 h at 37°C to prepare ear homogenates. After five freeze-thaw cycles, ear and mechanically prepared ovary homogenates were serially diluted, inoculated onto CV-1 cells, and stained 24 h later with crystal violet. Each dot in the figures represents the virus titer in each ear or ovary from individual mice and the thick horizontal bars represent the mean values for each group.

Rapamycin (Sigma-Aldrich) was administered in 5% Tween 80 and 5% polyethylene glycol 400 (Sigma-Aldrich) in saline solution (Dehay et al., 2010). Rapamycin or vehicle was inoculated i.p. daily at a low dose (75 mg/kg per mouse; Araki et al., 2009) 1 d before immunization or infection and for an additional 7 d during the T lymphocyte expansion phase.

In vivo cytotoxicity assays with N-ras^+/+ and N-ras^−/− splenocytes were performed as previously described (Medina et al., 2009). N-ras^+/+ and N-ras^−/− splenocytes were each split in two populations, labeled with either a high or a low concentration of CFSE, and washed. N-ras^+/+ and N-ras^−/− CFSE^hi cells were pulsed with OVA peptide, mixed with unpulsed CFSE^lo cells, their actual relative ratio in the mixture measured by cytometry, and injected i.p. into infected, immunized, or control syngeneic recipients. The peritoneal cavity was lavaged 16 h later. The cells were then analyzed by FACS to measure in vivo killing. Specific lysis was calculated using to the formula: [1 − (ratio unprimed/ratio primed) × 100], where the ratio unprimed is %CFSE^lo/%CFSE^hi cells remaining in the control recipients and the ratio primed is %CFSE^lo/%CFSE^hi cells remaining in the experimental recipients. Representative histograms of an individual mouse per group are shown in the figures. The numbers in the histograms indicate the percentage of specific lysis. Mean specific lysis data from sets of experiments are graphed as the mean ± SEM.

Around 200 CD8α⁺ purified OT-I T cells from N-ras^+/+ mice (CD45.1⁺ CD45.2⁺) and/or from N-ras^−/− mice (CD45.1⁺ CD45.2⁻) were mixed, their actual relative ratio in the mixture measured by cytometry, and adoptively transferred i.v. into the same intact congenic N-ras^+/+ recipients (CD45.1⁻ CD45.2⁺). In secondary response experiments, OT-I cells were first transferred into separate CD45.1⁻ CD45.2⁺ recipients that were then immunized with mDCs pulsed with OVA peptide. Transgenic OT-I cells were purified ex vivo 8 d after priming in two steps. CD8⁺ T cells were first negatively selected and then enriched in CD45.1⁺ OT-I cells by positive selection using magnetic beads (Miltenyi Biotec). Equal numbers of primed N-ras^+/+ and N-ras^−/− OT-I cells (∼200) were then cotransferred into the same congenic N-ras^+/+ recipients that were subsequently infected with rVACV-OVA. The number of splenic N-ras^+/+ and N-ras^−/− transgenic OT-I cells was determined 5 d after challenge. For some experiments, OT-I cells and recipients were treated with rapamycin. The number of CD45.1⁺ OT-I cells was determined after staining and FACS evaluation. The total number of adoptively transferred N-ras^+/+ and N-ras^−/− cells was calculated by multiplying the total cell count by the fraction of CD8⁺ CD45.1⁺ CD45.2⁺ or CD8⁺ CD45.1⁺ CD45.2⁻ cell gates, respectively.

Retroviral vector MigR1 expressing GFP and bicistronic retrovirus expressing GFP and Eomes were obtained from S. Reiner (Columbia University Medical Center, New York, NY; Intlekofer et al., 2005). Retroviruses were packaged by transient transfection of 293T cells with the retroviral vector along with pCL^eco and retroviral transduction of antigen-specific CD8⁺ T cells was performed following a previously described procedure (García-Peydró et al., 2003). In brief, splenocytes from CD45.1⁺ N-ras^+/+ or N-ras^−/− OT-I TCR transgenic mice were incubated with 10⁻⁹ M OVA peptide for 24 h, and CD8⁺ T cells were purified and then spin-infected with control or Eomes-expressing retrovirus-containing supernatants in the presence of 8 µg/ml polybrene. The retrovirally infected cells were transferred into N-ras^+/+ CD45.2⁺ recipients (10⁵ cells/mouse), which were infected with 10⁶ PFU rVACV-OVA. To study the primary response, PECs were analyzed 8 d after first exposure to antigen. Some recipients were infected 14 d later and the secondary response was similarly analyzed 5 d p.i.

Data are represented as means and SD or SEM. Statistical significance of differences between the means of experimental groups was determined using an unpaired two-tailed Student’s t test.

We thank S. Reiner and H.-H. Xue for the retrovirus constructs, L. López-Ferreras, M.L. Toribio, and M. García-Peydró for help with retroviral protocols, J.L. Rodríguez Fernández for the PI3K inhibitors, P. Boya for help with rapamycin, A. Anel and J. Pardo for helpful discussion, B. Alarcón for his initial insight, and Y. Laó, C. Mir, and S. Sánchez for technical assistance.

This work was supported by Spanish Ministerio de Ciencia e Innovación, Red Temática de Investigación Cooperativa en SIDA from Instituto de Salud Carlos III (ISCIII) and CSIC (M. Del Val); by Comunidad Autónoma de Madrid, Universidad Complutense de Madrid and Fondo de Investigaciones Sanitarias (E. Fernández-Malavé); by Fondo de Investigaciones Sanitarias and by Red Temática de Investigación Cooperativa en Cáncer from ISCIII (E. Santos); by ISCIII (S. Iborra); and by Ministerio de Ciencia e Innovación (S. Lázaro).

The authors declare that they have no conflicting financial interests.

S. Iborra, M. Ramos, S. Lázaro, F. Aguilar, D. López, E. Fernández-Malavé, and M. Del Val designed research and analyzed data; S. Iborra, M. Ramos, S. Lázaro, F. Aguilar, and E. Fernández-Malavé performed research; E. Santos and E. Fernández-Malavé provided mice; and S. Iborra, E. Fernández-Malavé, and M. Del Val wrote the paper.