The basis of distinctive IL-2- and IL-15-dependent signaling: Weak CD122-dependent signaling favors CD8⁺ T_CM survival but not T_EM development

Mice

CD122^−/− mice (23) and CD122^−/− mice expressing T cell-targeted transgenic CD122 with mutations (Y1 and Y3) in cytoplasmic tyrosine residues (22) have been previously described. These mice were crossed to OT-I mice to yield OT-I CD122^−/− and OT-I CD122^−/− IL-2Rβ mutant mice. CD45.1-congenic C57BL/6, OT-I and IL-15^−/− mice were obtained from Taconic Farms. C57BL/6, human Bcl-2 transgenic mice (C57BL/6Tg (Bcl-2)36Wehi), and IL-15Rα^−/− mice were obtained from the Jackson Laboratory. OT-I CD122^−/− mice were crossed to Bcl-2 transgenic mice to yield OT-I WT^Bcl-2Tg and OTI CD122^{−/−Bcl2Tg} mice. CD45.1 was also crossed onto OT-I background and provide by E. Gilboa (University of Miami). Animal studies were approved by the Institutional Animal Care and Use Committee at the University of Miami.

Adoptive Transfer and immunization of OT-I

OT-I T cells were purified (typically ≥90%) by magnetic based positive selection using anti-CD8 beads (Miltenyi Biotec). For adoptive transfer, naïve CD45.2 OT-I T cells (0.5 × 10⁶) from the indicated mouse models were injected i.v. into naïve CD45.1 C57BL/6 recipient mice. 24 hr after adoptive transfer, mice were immunized by i.v. injection of OVA_257–264 (10 µg) (synthesized by AnaSpec, San Jose, CA) and LPS (10 µg) (from E.coli 055:B5, Sigma, Saint Louis, Missouri). At day 12 or day 27 after priming, some mice were re-challenged with OVA _257–264 (10 µg) and LPS (10 µg) i.v. and the resulting secondary T_EFF cells were analyzed three days later.

In vitro culture

OT-I spleen cells (1 × 10⁶ cells/well) were cultured in 24 well flat bottom plates in RPMI 1640 complete medium (CM) with OVA_257–264 peptide (0.1nM) for 72 hours (24). Normal C57BL/6 spleen cells were cultured (2 × 10⁶ cells/well) with anti-CD3 (5% supernatant) or LPS (1ug/ml) in 24-well flat bottom plates in CM for 48 hr (25) to generate T or B cell blasts, respectively. For intracellular signaling experiments, in vitro-derived OT-I and anti-CD3 activated T cells were washed 3 times, and re-cultured in CM at (1 × 10⁶ cells/well) in 24 well plates without cytokines for 4 hours. For ex vivo derived OT-I cells from immunized mice, they were cultured in CM for 30 min. Cells were stimulated with mouse IL-2 (10 ng/ml) and IL-15 (10 ng/ml) for 15–180 min unless otherwise indicated. For growth and survival experiments, anti-CD3 activated T cells were washed 3 times, and re-cultured in 5 ml of CM at (1 × 10⁵ cells/ml) in T25 flasks in absence or presence of IL-2 (10 ng/ml) or IL-15 (10 ng/ml) for 48 hrs.

FACS analysis

Antibodies used in flow cytometry for cell surface or intracellular cytokine staining were purchased from Biolegend (San Diego, CA), eBioscience (San Diego, CA) or BD Bioscience (San Diego, CA). Other purchased antibodies are IL-15Rα (R&D Systems Minneapolis, MN), Granzyme B (Invitrogen, Carlsbad, CA), pStat5 (pY694, BD Bioscience) or pS6 kinase (Cell Signaling, Denver, MA). Intracellular staining for mouse Bcl-2 (clone 3F11) and human Bcl-2 (clone 100) was performed according to manufacturer’s instructions (BD Biosience). Cells were analyzed with a Becton Dickson LSR2 and FACS Diva or FlowJO software. Typically 100,000–300,000 events were collected per sample.

Cell surface staining was performed as previously described (26). Intracellular staining for pStat5 or pS6 kinase was performed as previously described (27, 28). To measure intracellular cytokine expression, spleen cells from the different mouse models were cultured (2.5 × 10⁶ cells/well) in 24 well plates in 1 ml of CM for 5 hrs with OVA_257–264 (0.1 nM), in the presence of brefeldin A (GolgiPlug, BD-Biosciences). Granzyme B was determined after 24 hr ex vivo stimulation with OVA_257–264 where brefaldin A was added during the last 4 hrs of culture. The cells were subjected to surface staining with the markers of interest and then permeabilized using Cytofix Cytoperm (BD Pharmingen) prior to staining with mAbs to IFNγ, IL-2, TNF-α or granzyme B. For intracellular staining for Ki67 and Bcl-2, cells were permeabilized using Fixation/Permeabilization buffer from Foxp3 staining kit (eBiosience).

Statistical Analysis

Data were analyzed by unpaired Student-T-test or by one-way ANOVA with Tukey’s multiple comparison test. Statistically significant differences (p<0.05) for comparison to WT controls are designated by an asterisk in the graphs.

IL-2 and IL-15 induce quantitative distinct receptor proximal signaling

Both IL-2 and IL-15 activate Stat5 and the PI-3K/Akt signaling pathways (2, 3). To study receptor proximal signal transduction by these two cytokines, we examined the activation of Stat5 and the S6 kinase, the latter is a downstream target of the PI-3K/Akt/mTORC1 pathway (29), by flow cytometry using mAbs to their activation intermediates, pStat5 and pS6. Antigen-activated OT-I T cells were cultured in medium for 4 hr to return activation of these pathways to basal levels and then were stimulated with IL-2 or IL-15 for the indicated times. Representative FACS profiles for time course (Fig. 1A) and results from multiple time-course (Fig. 1B, left) and dose-response experiments (Fig. 1B, right) revealed that IL-2 induced more sustained signaling than IL-15. Both cytokines initially induced comparable and maximal levels of pStat5 by 15 min. However, IL-2-dependent pStat5 activation remained relatively high for 2 hr whereas IL-15-dependent pStat5 was more transient. As expected, the peak of pS6 activation occurred later. Notably, IL-15 supported much lower levels of activated pS6. As previously noted (22), IL-2 induced an obviously bimodal distribution of pS6^hi and pS6^lo cells (Fig. 1A) and these were strikingly reduced for the IL-15 stimulated CD8⁺ T cells. As the Stat5 and PI-3K pathways cooperate to promote T cell proliferation, this result might explain the capacity to drive more extensive T cell expansion by IL-2 when compared to IL-15 (Fig. 3B and (22, 30, 31). Dose-response curves at the initiation of pStat5 and pS6 activation (15 min) for IL-2 and IL-15 were nearly identical (Fig. 1B, right), consistent with effective ligand capture by IL-2Rα and IL-15Rα and the similar high affinity (K_d~10⁻¹¹M) ligand binding by the IL-2R and IL-15R.

IL-2Rα has a short cytoplasmic tail (11 amino acids) with no known signaling kinase activity but TRAF and Syk associates with IL-15Rα and induces signaling in fibroblasts and mast cells (32). One possible explanation, therefore, for the lower signal transduction by IL-15 is that activity associated with IL-15Rα down-regulates signal transduction through CD122 and γc. However, this possibility is highly unlikely for CD8⁺ T cells because domain swapping of the IL-2Rα cytoplasmic tail with that of IL-15Rα in knock-in mice revealed that IL-15Rα acts to enhance activation of Stat5 rather than attenuate the signaling (33).

Alternatively, the inability for IL-15 to sustain signaling might reflect exhaustion of key signaling intermediates, either through the capacity of IL-15Rα to enhance signaling or through distinctive orientation of the cytoplasmic tails of CD122 and γc in the context of the IL-15R. These possibilities were functionally tested by evaluating the extent IL-15 dampens IL-2-dependent signaling. When activated OT-I T cells were stimulated by a mixture of IL-2 and IL-15 or the cells were first stimulated with IL-15 before adding IL-2, the activation of pStat5 resembles cells solely stimulated with IL-2, rather than less sustained IL-15-dependent pStat5 activation (Fig. 1C). Similar results were obtained for pS6 (Fig. S1A). Furthermore, if activated T cells were stimulated with IL-15 for 60 min, when pStat5 returned to a low level, the addition of lL-2 rapidly led to substantial sustained pStat5 activation (Fig. 1D). Thus, IL-15 does not deplete or desensitize signal transduction dependent upon CD122 and γc.

Another possibility for the lower signaling through the IL-15R is that the IL-15R depleted IL-15. However, when IL-15 was re-added to activated OT-I T cells 45 min after the initial induction of pStat5 (Fig. 1E) or pS6 (Fig. S1B), these signaling molecules were not reactivated. This finding shows that transient IL-15-dependent signaling is not due to inadequate cytokine levels during the test period. Notably, when activated OT-I T cells were pretreated with IL-2 or IL-15 and washed prior to culture at 37° C, activation of pStat5 (Fig. 1F) and pS6 (Fig. S1C) by IL-2 and IL-15 were very similar and resembled the transient activation of these signaling molecules by continuous stimulation with IL-15. This finding suggests that quantitatively lower levels of signaling associated with the IL-15R may be due to a limiting component at the initiation of signaling, perhaps cytokine receptor subunits themselves.

IL-15Rα is rapidly depleted by IL-15

Both IL-2 and IL-15 cause rapid receptor-mediated endocytosis of the cytokine complexed with its high affinity receptor (34–36). Thus, signal transduction is potentially limited by depletion of any one of the receptor subunits. To investigate this issue, the levels of CD25, IL-15Rα, CD122, and γc were followed after anti-CD3 activated CD8 T cells were stimulated with IL-2 or IL-15. First, we confirmed for these activated T cells that IL-2 and IL-15 induced pStat5 in a manner very similar to OVA_257–264-activated OT-I T cells, including more transient pStat5 activation by IL-15 (Fig. 2A). FACS analyses of the surface levels of cytokine receptor subunits for anti-CD3 activated CD8⁺ T cells prior to stimulation by IL-2 or IL-15 revealed very high levels of CD25, but relatively low levels of IL-15Rα, CD122, and γc (Fig. 2B, left). Upon stimulation with IL-2, levels of CD25 and IL-15Rα were at best minimally reduced, but the levels of CD122 and γc markedly decreased (Fig. 2B, right). For IL-2R-dependent signaling, CD122 is the limiting component, but these levels are sufficient for sustained signal transduction.

After stimulation with IL-15, no reduction was noted for CD25, while CD122 and γc levels remained at ≥75% of that found for unstimulated cells. Thus, reduced IL-15-dependent signaling cannot be due to limiting levels of CD122 and γc. However, the levels of IL-15Rα were reduced by approximately 90% (Fig. 2B, right). Thus, for IL-15R-dependent signaling, IL-15Rα is the limiting subunit. This rapid loss of IL-15Rα prevents high affinity binding of IL-15 and provides a ready explanation for the more transient signal transduction associated with the IL-15R.

IL-15 induces signaling by both cis- and trans-presentation of IL-15/IL-15Rα to CD122 and γc (37). We assessed the extent cis versus trans-presentation accounts for activation of pStat5 by CD8⁺ T cells. First, when activated CD45.1-congenic wild-type (WT) and IL-15Rα^−/− T cells were co-cultured with soluble IL-15, pStat5 was only activated in the WT T cells (Fig. 2C). Thus, cis presentation of IL-15 to CD122 and γc dominates the response to exogenous IL-15. Second, when WT T and B blasts were pre-treated with IL-15 and used to trans-present IL-15 to IL-15Rα ^−/− T blasts, only a minimal increase in pStat5 was noted for the IL-15Rα^−/− T cells whereas pStat5 activation was readily noted by cis presentation for the pre-treated WT T blasts (Fig. 2D). These experiments demonstrate little trans-presentation occurs in vitro even when using activated B cells that express IL-15Rα, but do not respond to this cytokine due to the absence of CD122. In addition, increased levels of IL-15 supported strong pStat5 induction by WT, but not IL-15Rα^−/−, CD8⁺ activated T cells, indicating that high IL-15 does not overcome the loss of cis-expressed IL-15Rα (Fig. 2E). Thus, trans-presentation of IL-15 to the IL-15R does not support strong sustained signal transduction as seen with IL-2R. Collectively, when compared to the IL-2/IL-2R interaction, our results indicate that IL-15R deliver quantitatively lower and distinguishable signal transduction in activated CD8⁺ T cells due to the limited expression of IL-15Rα.

Signal transduction by T_EFF cells bearing mutant CD122

To further explore the consequences of varied CD122-dependent signaling, pStat5 and pS6 activation by anti-CD3 activated T cells bearing defined mutations within the cytoplasmic tail of CD122 were examined. These experiments utilized three lines of previously characterized transgenic mice on the CD122^−/− genetic background whose T cells expressed CD122 where key tyrosine residues were mutated to phenylalanine. These mutations were in only Y341, (Y1), required for association of the adaptor protein Shc, in both Y395 and Y498 (Y2), required for optimal association of STAT5, or in all three of these tyrosine residues Y341, Y395, and Y498 (Y3). Past work showed that transgenic expression was restricted to T lineage cells, with levels similar to that expressed by control mice (22). In response to IL-2, signal transduction was attenuated, but not abrogated, with the least signaling associated with the Y3 mutations, leading to increasingly impaired IL-2-dependent T cell proliferation (22). Additionally, T cells from CD122^−/− mice that expressed WT transgenic CD122 developed IL-2-dependent signaling, biological responses, and gene expression profiles that closely paralleled those by control mice (22) (and unpublished data). Therefore, normal B6 mice (designated as WT) served as the controls for these and all subsequent studies.

First, IL-2- versus IL-15-dependent signal transduction was compared for activated CD8⁺ T cells bearing these mutant CD122 molecules (Fig. 3A). Each CD122 mutant activated pStat5. For the Y1 mutation, initial pStat5 activation 15 min post-stimulation by IL-2 and IL-15 were very similar to each other and to that induced in activated CD8⁺ WT T cells whereas pStat5 activation by CD122 with the Y2 and Y3 mutations was similarly lower. Moreover, all 3 mutant CD122 molecules failed to sustain IL-2-dependent pStat5 activation and poorly activated pS6 to IL-2 or IL-15. This latter result is predicted for the Y1 and Y3 mutations, as Y341 is required for Shc association to CD122 that leads to activation of PI-3K/Akt and downstream activation of S6 (38). The poor activation of pS6 by the Y2 mutation is consistent with a requirement for optimal activation of Stat5 for induction of the PI-3K/Akt pathway, as previously reported (39). Overall, the Y1 mutation leads to signal transduction by IL-2 and IL-15 that approximates that induced in activated CD8⁺ T cells by IL-15. In contrast, signal transduction by the Y2 and Y3 mutations were similar, but lower than supported by IL-2 or IL-15.

Given the roles of IL-2 and IL-15 in promoting T cell growth and survival, we tested the extent these mutant CD122 molecules supported these activities in activated CD8⁺ T cells. Anti-CD3-activated T cells were prepared from mice expressing WT and mutant CD122 molecules and these T cell blasts were re-cultured with IL-2 or IL-15 for 48 hr. Cell recovery (Fig. 3B) and CD25 expression (Fig. 3C) decreased with increased number of tyrosine mutations. For the recovered cells, only the Y1 mutation supported protection from apoptosis (Fig. 3D) and increased levels of Bcl-2 (Fig. 3E) that were comparable to that found for WT T cells. Thus, survival, but not clonal expansion, of activated CD8⁺ T cells is readily supported by weak IL-15-like signal transduction associated with the Y1 mutation of CD122.

Primary and recall responses by CD122-deficient CD8⁺ T cells

The distinctive activation of pSTAT5 and pS6 by CD8⁺ T cells bearing mutant CD122 provides a direct approach to assess the contribution of varied CD122-dependent signaling strength for a CD8-dependent immune response in vivo. However, we first established the requirement for CD122 signaling for primary and recall responses to OVA_257–264 by CD122^−/− OT-I T cells. To prevent autoimmunity that is associated with CD122^−/− mice, the OT-I CD122^−/− mice received WT Treg cells at birth (40). WT or CD122^−/− OT-I T cells were adoptively transferred to syngeneic CD45.1 B6 mice, to facilitate identification of the donor cells, and 1 day later the mice were immunized with OVA_257–264 and LPS.

The CD122^−/− OT-I T cells expanded and contracted (Fig. 4A) and developed into IFNγ- and IL-2-secreting T_EFF cells (Fig. 4B) comparable to WT OT-I T cells. However, granzyme B expression was markedly lower (Fig. 4C). In addition, WT and CD122^−/− OT-I T cells exhibited a largely similar cell surface phenotype on day 2, 3 and 7 post-challenge, including similar proliferative status, as assessed by Ki67, and expression of pro-survival Bcl-2 (Fig. 4D). This analysis also showed that CD25 was detected at a low level only on day 2 post-immunization, suggesting that the WT OT-I T cells can only respond to IL-2 at this time. In addition, on day 3 WT OT-I T cells were characterized by increased CD27^lo cells, consistent with somewhat greater activation.

Although CD122^−/− CD8 T cells readily generated a largely normal primary response, they were not maintained long-term (Fig. 4A), most likely because they did not receive survival and homeostatic signals through IL-15 (7). By day 7 post-challenge after initial contraction, several phenotypic changes were apparent between WT and CD122^−/− OT-I T cells, where the latter cells were more prominently CD62L^hi, Ly-6C^lo, and Ki67^neg (Fig. 4D). The lower percentage of Ki67⁺ CD122^−/− OT-I T cells is consistent with a lower proliferative rate with time post-challenge and represents a contributory factor for the sharper reduction in these antigen-activated T cells. These findings suggest potential intrinsic problems with CD122^−/− OT-I T cells.

Since CD122^−/− T cells were not maintained long-term, i.e. day 30, properties of WT and CD122^−/− OT-I cells were assessed at an intermediate time point after initial contraction. By day 15 post-challenge, the relative numbers of CD122^−/− OT-I T cells were severely reduced and barely detectable, but after re-challenge, these few CD122^−/− OT-I T cells readily expanded (Fig. 5A). Although the absolute relative numbers remained lower than WT cells, the overall extent of expansion by CD122^−/− OT-I T cells was significantly greater than WT OT-I T cells (Fig. 5B). However, the function and phenotype of the recalled CD122^−/− OT-I T cells were not entirely normal. WT OT-I T cells were characterized by increased percentage of IFNγ producing cells (Fig. 5C) and CD62L^lo, CD127^lo, and Klrg1⁺ cells (Fig. 5D), markers of highly activated T_EFF cells, and. Overall, these results demonstrate that CD122^−/− OT-I T cells are not impaired in their capacity to elicit recall responses characterized by extensive expansion, but may harbor intrinsic defects related to development of highly activated T_EFF cells. The expression of Ly-6C^hi is a marker of memory CD8⁺ T cells (41). CD122^−/− OT-I T cells were distinctively characterized by a higher proportion of Ly6C^lo cells (Fig. 5D). This result raises the possibility that expression of memory cell properties require CD122 signaling.

CD122 signaling thresholds and the survival of CD8⁺ T memory cells

Two immediate questions arise concerning the lack of CD122^−/− T cells by 30 day post primary antigenic challenge. First, to what extent is this abnormality simply due to the lack of a CD122-dependent survival signals? Since Bcl-2 is an important target of IL-2 and IL-15 signaling (14, 42), this issue was directly examined by evaluating the extent transgenic Bcl-2 expression in CD122^−/− OT-I T cells promoted persistent T cells. The second question is to what extent does varied CD122 signaling support CD8⁺ memory development/survival? This was approached by determining the extent that OT-I T cells that bear mutant CD122 supported persistent OT-I memory cells. Here we utilized the transgenic Y1 and Y3 CD122 mutants on the CD122^−/− genetic background as these varied in Bcl-2 expression and CD8 T cell survival as well as the levels of IL-2- and IL-15-driven signaling in vitro (Fig. 3). We confirmed that at the peak of the primary response, the expansion each of these types of OT-I T cells was similar to each other and to WT and CD122^−/− OT-I T cells (Fig. 6A).

30 days after the primary challenge, similar relative numbers of persistent Y1, Y3, and CD122^{−/−Bcl2Tg} OT-I T cells were found that were much higher (p<0.05, ANOVA) than detected for CD122^−/− OT-I T cells. These numbers, however, for Y3 and CD122^{−/−Bcl2Tg} OT-I T cells were significantly lower (p<0.05, ANOVA) than detected for WT OT-I T cells (Fig. 6B). Thus, either weak CD122-dependent signaling or pro-survival Bcl-2 in the absence of IL-2 or IL-15 signaling supported retention of antigen-activated OT-I T cells, although it was not always entirely normal. Upon secondary recall, the relative numbers of WT, Y1, Y3, and CD122^{−/−Bcl2Tg} OT-I T cells each increased by 5 to 10 fold (Fig. 6C, D). Only CD122^−/− OT-I T cells showed significantly lower overall expansion. The lower expansion by CD122^{−/−Bcl2Tg} OT-I T cells was not secondary to transgenic human Bcl-2 because expansion by WT^Bcl2Tg OT-I T cells was comparable to non-transgenic control WT OT-I T cells.

To evaluate the basis by which mutant CD122 and ectopic Bcl-2 increased the persistence and recall of OT-I T cells when compared to CD122^−/− OT-I T cells, the fraction of cells in cell cycle was assessed by determining the expression of the proliferative marker Ki67. For the persistent cells prior to recall, only a small fraction (1–3%) of the cell were Ki67⁺ (Fig. 6E). However, this homeostatic proliferation was significantly lower for Y3 and CD122^{−/−Bcl2Tg} OT-I T cells and a comparable low percentage of Ki67⁺ cells was noted for one Y1 mouse that was similarly analyzed. The detection sometimes of fewer persistent cells from OT-I T cells with impaired or absent CD122 signaling may in part be related to lower homeostatic proliferation. In marked contrast, all groups of OT-I T cells exhibited substantial and similar proliferation upon re-challenge (Fig. 6F). Collectively, these results indicate that CD122 signaling importantly contributes to homeostatic proliferation but is not required to drive proliferation of memory cells after secondary antigen encounter.

Given the importance of CD122 signaling in contributing to expression of Bcl-2 (14, 42), we also evaluated the levels of expression of endogenous mouse Bcl-2 by the persistent OT-I T cells 30 days post-primary immunization. Y1 and Y3 OT-I T cells expressed endogenous Bcl-2 similar to WT cells (Fig. 6G). It was not possible to directly assess reliably Bcl-2 expression by the persistent CD122^−/− OT-I T cells due to their few numbers. However, mouse-Bcl-2 expression is reduced by 50% in CD122^{−/−Bcl-2Tg} OT-I T cells. These results confirm the requirement for CD122 signaling for optimal expression of pro-survival Bcl-2. In CD122^{−/−Bcl-2Tg} OT-I T cells, the human transgenic Bcl-2 was readily detected in these cells (Fig. S2). This finding and the normal levels of mouse Bcl-2 in Y1 and Y3 OT-I T cells demonstrate another critical role for CD122 signaling for memory T cells is to support proper expression of Bcl-2, which is achieved by relatively weak CD122 signaling.

The development of T_CM and T_EM depends on varied CD122 signaling

To determine whether the lack of proper CD122 signaling affected other activities associated with the persistent cells, memory phenotype and function were tested before and after secondary recall. Representative FACS histograms used to quantify the persistent and recalled cells are shown for these phenotypic and functional analyses. For the persistent cells, in comparison to WT OT-I T cells, Y1, Y3, and CD122^{−/−Bcl2Tg} OT-I T cells all showed a significantly higher and dominant percentage of CD62L^hi cells and these were essentially all CD127^hi, a phenotype characteristic of T_CM cells (Fig. 7A). Therefore, using these markers, WT OT-I cells were a near equal mix of T_CM and T_EM cells whereas with no (CD122^{−/−Bcl2Tg}) or the weak (Y3) CD122 signaling, nearly all the persistent cells were T_CM cells. The Y1 mutant, which exhibits CD122-dependent signaling similar to that induced by IL-15, still favor T_CM cells (Fig. 7C). These results indicate that strong CD122 signaling is required for full T_EM development while low to no CD122 signaling support T_CM traits.

The Ly-6C^hi memory marker was also somewhat lower on persistent CD122^{−/−Bcl2Tg} and Y3 OT-I T cells, suggesting some defect in full memory development (Fig. 7A). There were very few Klrg1⁺ Y1, Y3, and CD122^{−/−Bcl2Tg} OT-I T cells, associating a requirement for strong CD122 signaling for this subpopulation of antigen-experienced cells, where low expression of Klrg1 is related to memory precursor cells (11). The secondary recalled T_EFF cells mirrored these phenotypic properties as WT OT-I T cells were characterized by a greater fractions of CD62L^lo, Ly-6C^hi, Klrg1⁺ cells (Fig. 7B). These differences likely reflect some intrinsic defects in memory development as a consequence of impaired CD122 signaling.

When examined 5 hr after ex vivo re-stimulation, the persistent Y1, Y3, CD122^{−/−Bcl2Tg} OT-I T cells contained fractions of cytokine producing cells that were largely similar to WT OT-I T cells (Fig. 8A), including the levels of IFNγ as assessed by the MFI. However, when compared to WT OT-I T cells, the secondary recalled response was somewhat attenuated, especially for TNFα by Y3 and CD122^{−/−Bcl2Tg} OT-I T cells (Fig. 8B). Moreover, although the % IFNγ⁺ cells were not significantly different, all recalled OT-I T cells with impaired CD122 signaling produced lower levels of IFNγ based on the MFI of the IFNγ⁺ cells. Thus, in the absence of full CD122 signaling, the persistent cells contain some defects in the expression of surface markers and functional activities.

IL-2 and IL-15 induced signaling by primary and memory OT-I T cells

To relate the developmental differences in T_EM and T_CM upon varied CD122 signaling, we examined the expression of IL-2R and IL-15R subunits and pStat5 activation upon stimulation with IL-2 or IL-15 at various times after antigen challenge. At 2 days post challenge, OT-I T cells exhibited increased levels of CD25, CD122, γc and IL-15Rα (Fig. 9A). However, only a fraction (approximately 40%) of the OT-I cells expressed CD25 while the other subunits were uniformly increased on all cells. At 3 days post-challenge, this induction return to baseline for CD25 and γc and was reduced for IL-15Rα, while high levels of CD122 remained. For the persistent cells 30 days post-challenge, increased levels of CD122 and IL-15Rα were observed.

When these same cells were activated with IL-2 or IL-15 directly ex vivo, substantial IL-2 induced pStat5 activation was only noted at day 2 post challenge whereas IL-15 activated pStat5 at all time points (Fig. 9B). For IL-2-induced signaling at day 2, bimodal IL-2-dependent pStat5^lo and Stat5^hi activation was observed, which corresponded to the CD25^neg/^lo and CD25^hi cells. These results directly correlate pStat5 activation with the level of expression of the high affinity IL-2R and IL-15R. When comparing total pStat5 activation by IL-2 and IL-15 on day 2, IL-15 supports greater signaling. However, if only pStat5^hi signaling by IL-2 is considered, these cells show more sustained pStat5 than induced by IL-15 (Fig. 9D). These latter finding are consistent with our in vitro findings that showed that the initiation of pStat5 activation (15min) for IL-2 and IL-15 were almost identical whereas after more prolonger cytokine signaling (60 min) IL-2 dependent pStat5 activation remained high but L-15 dependent pStat5 was more transient. Furthermore, similar to responses in vitro (Fig. 3A), at day 2 post-challenge, OT-I T cells from the CD122 Y3 mutant showed reduced capacity to sustain IL-2 and IL-15-dependent pStat5 activation (Fig. 9C) when compared to WT OT-I T cells (Fig. 9B). Collectively, these data demonstrate that in vivo activated OT-I T cells have a restricted period of time to respond to IL-2, but not IL-15. For cells expressing readily detectable CD25, IL-2-induced pStat5 activation is more sustained than IL-15-induced pStat5. Moreover, the minimal pStat5 activation by cells expressing undetectable CD25 further indicates the strength of IL-2- and IL-15-dependent signaling is regulated by levels of a limiting receptor subunit. These latter points are in agreement with results shown for IL-2R and IL-15R on in vitro activated CD8⁺ T cells.

T_EM development is supported by IL-2

The sustained expression of high affinity IL-15R and induction of pStat5 by IL-15 at all times post-antigen challenge is in line with the role in supporting CD8 memory survival and homeostasis. However, this observation raises the point that the development of T_EM might also depend on IL-15 rather than IL-2 signaling. To directly test this point, we examined OT-I memory development in IL-15-deficient recipients. At 17 day post challenge, fewer OT-I T cells were detected in IL-15-deficient hosts, as expected, due to the requirement for IL-15 in CD8 memory survival (Fig. 10A). However, for those cells detected, a normal distribution of T_EM and T_CM were seen as assessed by expression of CD62L and CD127 (Fig. 10B, C). However, there was a significant decrease in the persistence of KLRG1⁺ cells (Fig. 10 B and C) which were also CD62L^lo CD127^lo cells (not shown) in the absence of IL-15. These findings demonstrate that IL-2 is sufficient for the development of T_EM cells, whereas IL-15 is required for production of persistent Klrg1⁺ OT-I T_EFF cells after challenge with OVA_257–264 and LPS.