Expression of IL-7 receptor α is necessary but not sufficient for the formation of memory CD8 T cells during viral infection

Constitutive Expression of IL-7Rα Does Not Affect Virus-Specific Effector CD8 T Cell Differentiation.

To test whether the formation of virus-specific effector and memory CD8 T cells was affected by the constitutive expression of IL-7Rα, we bred IL-7Rα transgenic (IL-7Rαtg) mice that constitutively express IL-7Rα in all T cells via the CD2 promoter to the P14 TCR transgenic mice, which contain CD8 T cells specific for the GP_33–41 epitope of lymphocytic choriomeningitis virus (LCMV) (16). As expected, the naïve IL-7Rαtg P14 CD8 T cells expressed higher than normal levels of IL-7Rα (Fig. 1a). Naïve P14 CD8 T cells isolated from both IL-7Rαtg or non-IL-7Rαtg littermates (referred to as WT) displayed a typical naïve T cell phenotype and expressed low amounts of CD44, killer cell lectin-like receptor G1 (KLRG1), CD43, IL-2Rα (CD25), IL-2/15Rβ (CD122), and high amounts of L-selectin (CD62L) and CD27 [supporting information (SI) Fig. 6]. Increased IL-7Rα did not alter levels of thymic stromal-derived lymphopoetin receptor or the common-γ chain (CD132), which bind to IL-7Rα to form the TSLP and IL-7 receptor complexes, respectively (SI Fig. 6) (17).

To compare the development of effector and memory CD8 T cells during LCMV infection, equal numbers of naïve WT or IL-7Rαtg P14 CD8 T cells were adoptively transferred together into C57BL/6 mice that were subsequently infected with LCMV. Congenic markers were used to distinguish the two P14 CD8 T populations; the WT cells were homozygous for Thy1.1 (1.1/1.1), and the IL-7Rαtg cells were heterozygous (1.1/1.2). Eight days postinfection (pi), the frequencies and numbers of LCMV-specific IL-7Rαtg and WT P14 CD8 T cells were calculated, and both effector cell populations were found to be of nearly equivalent size, ranging between 20 and 30 × 10⁶ CD8 T cells (Fig. 1 b and c). This result indicated that constitutive IL-7Rα expression did not enhance or impair effector CD8 T cell expansion. The IL-7Rαtg P14 CD8 T cells maintained high levels of IL-7Rα during infection (≈99% IL-7Rα positive), whereas the majority (≈87%) of the WT CD8 T cells down-regulated IL-7Rα (Fig. 1d). To ensure that early after activation, the IL-7Rαtg P14 CD8 T cells did not down-regulate IL-7Rα, as has been reported for WT cells (18), we examined these cells for IL-7Rα expression at days 1–5 pi and found little change (SI. Fig. 7). The effector functions of the IL-7Rαtg P14 CD8 T cells appeared normal, and both the number and frequency of IFN-γ, TNF-α, and IL-2-producing cells were nearly identical to WT P14 CD8 T cell population (Fig. 1e). Both effector cell populations expressed similar amounts of Granzyme B and killed GP_33–41 peptide-labeled targets equally in vivo, indicating their cytotoxicity was unaffected by IL-7Rαtg (Fig. 1d and data not shown). Furthermore, the IL-7Rαtg and WT P14 effector CD8 T cells expressed nearly identical amounts of the differentiation markers CD44, CD62L, and CD27 and the cytokine receptors CD25, CD122, CD132, and thymic stromal-derived lymphopoetin receptor (Fig. 1d). Together, these findings show that enforcement of IL-7Rα expression has little to no effect on the expansion, differentiation, or function of effector CD8 T cells in response to LCMV.

Forced Expression of IL-7Rα Does Not Affect Memory CD8 T Cell Formation or Function.

Next, we examined the ability of the IL-7Rαtg effector P14 CD8 T cells to develop into long-lived memory CD8 T cells after LCMV infection. Surprisingly, the IL-7Rαtg P14 CD8 T cells underwent normal effector cell contraction, and both their numbers and frequencies were similar to that of the WT population in the spleen, lymph nodes, liver, and lungs at days 15, 45, 60, and 90+ pi (Fig. 1 b and c and SI Fig. 8). Similar results were obtained after infection with recombinant Listeria monocytogenes (rLM-33) expressing the LCMV GP_33–41 epitope (data not shown). These findings indicated normal numbers of antigen-specific memory CD8 T cells formed and were maintained long-term, regardless of IL-7Rα overexpression after acute viral or bacterial infection.

We also tested whether there were any significant differences in the ability of the IL-7Rαtg and WT P14 memory CD8 T cells to persist or undergo homeostatic turnover by labeling WT and IL-7Rαtg T cells from day 45-infected mice with carboxy fluoro scein succinimidyl ester (CFSE), transferring equal numbers of each population into naïve C57BL/6 mice and analyzing them ≈2 months later. Similar numbers of donor IL-7Rαtg and WT CD8 T cells were recovered from the hosts, indicating that their overall rates of survival were comparable (data not shown), and both populations had divided to a similar extent based on dilution of CFSE (Fig. 2a). Interestingly, the cells with highest expression of IL-2/15Rβ (CD122) or IL-7Rα and lowest expression of KLRG1 divided the most, indicating that memory CD8 T cells with these characteristics have the greatest self-renewal capacity and presumably longevity (Fig. 2b). IL-7Rαtg and WT CD8 T cells also expressed similar amounts of Bcl-2 at memory time points (Fig. 2e). Thus, it does not appear that the survival or long-term maintenance of memory CD8 T cells was significantly improved by overexpression of IL-7Rα.

Another cardinal property of memory T cells is their ability to quickly develop effector functions and expand in response to secondary antigenic stimulation or infection. We found that both the IL-7Rαtg and WT P14 memory CD8 T cells produced the cytokines IFN-γ, IL-2, and TNF-α equivalently when restimulated ex vivo (Fig. 2c and data not shown). To compare their ability to respond to secondary infection, IL-7Rαtg or WT P14 memory CD8 T cells (≈7.5 × 10³) were isolated from animals infected 45 days previously and transferred into naïve C57BL/6 mice that were subsequently infected with recombinant vaccinia virus expressing GP_33–41 peptide. The donor IL-7Rαtg and WT CD8 T cells expanded and contracted equally, making up ≈25% of the CD8 T cells 8 days after secondary infection and generating similar numbers of secondary memory CD8 T cells 45 days later (Fig. 2d). Together, these data show that constitutive expression of IL-7Rα does not create memory CD8 T cells that persist or protect better than WT memory CD8 T cells.

Last, no significant differences were noted between IL-7Rαtg and WT P14 memory CD8 T cells in the expression of other memory cell markers such as CD62L and CD122, suggesting that IL-7Rαtg did not alter the formation of other memory cell subsets such as “central” or “effector” memory T cells (Fig. 2e).

Overexpression of IL-7Rα Does Not Rescue SLECs That Would Normally Die After Infection.

A major aim of our study was to determine whether IL-7Rα overexpression could specifically rescue IL-7Rα^lo SLECs from death and permit these cells to develop into long-lived memory CD8 T cells. If this was occurring, then a large fraction of the naturally arising IL-7Rα^hi MPECs would have been “replaced” by the IL-7Rα^lo SLECs, because similar numbers of memory CD8 T cells formed when IL-7Rα was overexpressed. Unfortunately, this question could not be addressed directly by using surface expression of IL-7Rα, because the constitutive expression of the murine IL-7Rα transgene masked the ability to distinguish the naturally arising IL-7Rα^lo effector CD8 T cells within this cell population. However, our prior work has shown that the natural killer cell inhibitory receptor KLRG1 is an additional marker of IL-7Rα^lo SLECs, because at days 8–10 after LCMV infection, ≈70% of the LCMV-specific effector CD8 T cells express KLRG1, and ≈90% of these KLRG1^hi cells are IL-7Rα^lo (Fig. 3a) (7). To evaluate whether the IL-7Rα genomic locus is repressed in IL-7Rαtg KLRG1^hi effector CD8 T cells, as it is in WT KLRG1^hi cells, we sorted KLRG1^hi and KLRG1^lo P14 CD8 T cells from both WT and IL-7Rαtg groups at days 8–10 pi and used real-time PCR to measure the levels of endogenous IL-7Rα mRNA. The results showed that the IL-7Rαtg KLRG1^hi effector CD8 T cells had significantly reduced levels of endogenous IL-7Rα mRNA, like their WT counterparts (Fig. 3b).

KLRG1^hi IL-7Rα^lo effector CD8 T cells progressively die over the next several weeks to months pi and show minimal conversion to an IL-7R^hi phenotype (Fig. 3a; N. S. Joshi and S.M.K., unpublished data). However, the small fraction of KLRG1^hi IL-7Rα^hi effector CD8 T cells present at days 8–10 pi can persist and constitute ≈20–30% of the memory CD8 T cells found ≈60–100 days pi (Fig. 3a). Their homeostatic turnover rates are low and consequently, these cells are gradually outpopulated by the self-renewing KLRG1^lo IL-7Rα^hi memory cells over time (Fig. 2b; N. S. Joshi and S.M.K., unpublished data). Thus, KLRG1 is an additional marker that can distinguish the majority (≈90%) of the naturally arising IL-7Rα^lo effector CD8 T cells within the IL-7Rαtg cell population. If the IL-7Rαtg could rescue the death of the naturally arising IL-7Rα^lo cells, then a greater percentage and number of KLRG1^hi cells would be found in the IL-7Rαtg P14 memory CD8 T cell population compared with the WT population. However, this was not the case, because both groups contained ≈65–70% KLRG1^hi effector CD8 T cells (days 8–10 pi) and ≈20–30% KLRG1^hi memory cells (at days 60–100 pi) (Figs. 1d, 2e, and 3). In some experiments, the IL-7Rαtg memory CD8 T cell population contained ≈5–10% more KLRG1^hi cells compared with the WT population, but this small difference did not achieve statistical significance. That similar numbers of KLRG1^hi (presumptive IL-7Rα^lo) effector CD8 T cells died in both the WT and IL-7Rαtg P14 CD8 T cell population showed that enforced expression of IL-7Rα was incapable of saving those effector cells naturally destined for death. This demonstrated that, after infection, the bulk of effector CD8 T cell contraction occurs in an IL-7Rα-independent manner.

KLRG1^hi IL-7Rαtg Effector CD8 T Cells Exhibit Decreased Responsiveness to IL-7 in Vivo.

To further investigate why the IL-7Rαtg was incapable of preventing the death of effector CD8 T cells death, we tested whether increasing IL-7 levels during the contraction phase could offset the proapoptotic pathways of IL-7Rαtg KLRG1^hi effector CD8 T cells. Mice containing equal numbers of IL-7Rαtg and WT P14 effector CD8 T cells were treated in vivo with recombinant human IL-7 (rhIL-7) or PBS between days 8 and 15 pi. The IL-7Rαtg P14 effector CD8 T cells essentially did not contract during the treatment period, and their numbers were elevated ≈12-fold compared with the PBS control animals and ≈6-fold compared with the WT P14 CD8 T cell population found in rhIL-7 treated animals (Fig. 4a). This result showed that additional IL-7 can enhance survival of the IL-7Rαtg effector CD8 T cells. In contrast, despite the additional rhIL-7, the number of WT P14 CD8 T cells declined considerably and only ≈2-fold more WT effector cells existed compared with controls. This result was observed, because only the IL-7Rα^hi effector CD8 T cells (which comprised ≈15% of the WT population at day 8) responded to rhIL-7, and their frequency rose to 60% in treated animals compared with 36% in PBS controls (SI Fig. 9).

The above data showed that the IL-7Rαtg can reduce effector CD8 T cell contraction when IL-7 is in excess; however, further experimentation showed this was not solely due to enhanced survival, because these cells also proliferated in response to rhIL-7 (Fig. 4b). Equal numbers of CFSE-labeled IL-7Rαtg or WT P14 effector CD8 T cells from day 8 pi were transferred into C57BL/6 hosts that were subsequently treated with rhIL-7 for 7 days. The IL-7Rαtg P14 CD8 T cells proliferated extensively, whereas their WT counterparts divided some, but to a much lesser extent (Fig. 4b). Little to no cell division was observed in the PBS controls.

Because the IL-7Rαtg had little effect on KLRG1^hi effector CD8 T cell survival at physiological IL-7 levels, we next examined whether IL-7Rαtg KLRG1^hi and KLRG1^lo effector CD8 T cells responded equivalently to rhIL-7 in vivo. At day 8 pi, the IL-7Rαtg effector P14 CD8 T cells were sorted into KLRG1^hi and KLRG1^lo subsets, labeled with CFSE, and transferred into C57BL/6 animals that were subsequently treated with rhIL-7 in vivo. Seven days later, we found that the KLRG1^lo cells proliferated extensively, whereas the KLRG1^hi cells did not (Fig. 4c). The functional difference between the KLRG1^hi and KLRG1^lo subsets was particularly interesting, given that the expression of IL-7Rα on both subsets was nearly identical (Fig. 4c). Correlating with the increased proliferation, the recovery of IL-7Rαtg KLRG1^lo cells was increased relative to the KLRG1^hi cells (Fig. 4d). This result demonstrated that, even when the KLRG1^hi express high levels of IL-7Rα ectopically, they could not respond to exogenous IL-7 as robustly as the KLRG1^lo effector cells. Thus, the KLRG1^hi and KLRG1^lo effector cells intrinsically differ in their ability to transduce IL-7Rα signals, and this likely provides an explanation for the failure of the IL-7Rαtg to induce greater numbers of memory CD8 T cells.

IL-7Rαtg KLRG1^hi SLECs Can Transmit IL-7 Signals to STAT5 But Have Increased Levels of p27^kip.

IL-7Rα activates two primary downstream signaling pathways: the PI3K/AKT and JAK/STAT pathways (17). To directly investigate whether IL-7Rαtg and WT KLRG1^hi and KLRG1^lo effector CD8 T cells intrinsically differ in IL-7Rα signal transduction, we tested their ability to activate the JAK3/STAT5 pathway by examining STAT5 phosphorylation (pSTAT5) after IL-7 treatment in vitro by flow cytometry. Most of the day 10 WT KLRG1^lo P14 effector CD8 T cells, of which ≈75–80% were IL-7Rα^hi, dramatically increased pSTAT5 levels after IL-7 stimulation (Fig. 5a). In contrast, the WT KLRG1^hi cells, of which ≈80–90% were IL-7Rα^lo, could not phosphorylate STAT5 very efficiently (Fig. 5a). Perhaps, as expected, these data showed that effector CD8 T cells require detectable levels of IL-7Rα to transduce IL-7 signals and phosphorylate STAT5. On the other hand, the IL-7Rαtg KLRG1^hi and KLRG1^lo effector CD8 T cells were equally capable of phosphorylating STAT5 when stimulated with IL-7 (Fig. 5a). This showed that this early step in IL-7Rα signaling was not impaired in IL-7Rαtg KLRG1^hi effector CD8 T cells. Moreover, both subsets of WT and IL-7Rαtg effector CD8 T cells could phosphorylate STAT5 equally in response to IL-15, demonstrating no global suppression of STAT5 signaling in KLRG1^hi cells (Fig. 5a). We also observed that both KLRG1^hi and KLRG1^lo IL-7Rαtg effector CD8 T cells increased bcl-2 and granzyme B expression, two downstream targets of STAT5 signaling, after treatment with rhIL-7 in vivo (Fig. 5b) (19). Combined with the above data, this finding showed that although the IL-7Rαtg KLRG1^hi effector cells could activate STAT5 efficiently when stimulated with excess IL-7, they could not proliferate or survive as well as KLRG1^lo cells.

To further investigate a potential mechanism for the differential responses of IL-7Rαtg KLRG1^hi and KLRG1^lo effector CD8 T cells to IL-7, we measured the amounts of the cell cycle inhibitor p27^kip in these cells. Western blotting indicated that the IL-7Rαtg KLRG1^hi effector CD8 T cells contained ≈8-fold more p27^kip than the KLRG1^lo cells when isolated directly ex vivo. The same result was found when WT KLRG1^hi and KLRG1^lo effector CD8 T cells were compared (data not shown). This considerable increase in p27^kip levels could account for the impaired proliferative responses of IL-7Rα KLRG1^hi effector cells to IL-7. Previous work has shown that IL-7 treatment can reduce the levels of this cell cycle inhibitor via activation of the PI3K/AKT pathway (20, 21) and therefore, this result may suggest that, even though the IL-7Rαtg KLRG1^hi effector CD8 T cells have the ability to “see” IL-7, they cannot effectively transduce signals through the PI3K/AKT pathway to degrade p27^kip.

Mice and Viral and Bacterial Infection.

C57BL/6Ncr mice were purchased from the National Cancer Institute (Frederick, MD). IL-7Rαtg mice were a kind gift from A. Singer (National Institutes of Health, Bethesda, MD) (16). IL-7Rαtg mice were bred to our laboratory strain of P14 TCR transgenic mice that express the congenic Thy1.1 marker and have been described (7). To make P14 chimeric mice, small numbers (≈5,000) of WT or IL-7Rαtg naïve P14 CD8 T cells were adoptively transferred into C57BL/6 mice that were subsequently infected with 2 × 10⁵ pfu LCMV Armstrong i.p. or with 2 × 10⁴ cfu recombinant L. monocytogenes i.v. that expresses the LCMV GP_33–41 epitope (strain XFL203, but referred to as rLM-33) (7). For secondary infection, C57BL/6 mice containing ≈7,500 WT or IL-7Rαtg P14 memory CD8 T cells were infected with 2 × 10⁶ recombinant vaccinia virus (vaccinia virus expressing GP_33–41 peptide) that expresses the LCMV GP_33–41 epitope. All experiments were done with approved Institutional Animal Care and Use Committee protocols.

Antibodies.

All antibodies used for flow cytometry were purchased from E-biosciences (San Diego, CA) except anti-Granzyme B-PE (Caltag, Burlingame, CA), anti-p27^kip (Cell Signaling, Danvers, MA), anti-Grp94 (StressGen, Victoria, BC, Canada), and pTyr647 STAT5 and Bcl-2 PE (BD, San Jose, CA). KLRG1 (2F1) hybridoma was obtained from D. Raulet (University of California, Berkeley, CA) and was conjugated to Alexa-647 (Invitrogen, Carlsbad, CA) according to the manufacturer's instructions.

Cell Sorting, CFSE Labeling, and Adoptive Transfer.

Splenocytes containing WT or IL-7Rαtg P14 CD8 T cells were isolated from LCMV-infected mice on the indicated days, stained with antibodies to CD8 and KLRG1, and then sorted using either FACSAria (BD) or MoFlo (DakoCytomation, Carpinteria, CA) instruments. In some experiments, splenocytes were depleted of non-CD8 T cells by magnetic separation before sorting/adoptive transfer. In all experiments, equal numbers of IL-7Rαtg and WT P14 CD8 T cells were transferred i.v. Cells were labeled with CFSE (Invitrogen), as described (7).

In Vitro Stimulations and Surface and Intracellular Staining.

To assay cytokine production, splenocytes were stimulated with GP33–41 peptide for 5–7 h followed by surface staining and intracellular staining for IFNγ and IL-2, as described (4). To assay STAT5 phosphorylation, splenocytes were stimulated for 20 min with 5 ng/ml rIL-7 or rIL-15 at 37°C, fixed in 4% formaldehyde, permeabilized in 90% methanol at 4°C, and then stained according to the manufacturer's protocol. Cells were examined by flow cytometry by using the FACSCalibur or LSRII instruments (BD) and analyzed with FlowJo software (Tree Star, San Carlos, CA).

rhIL-7 Treatment.

rhIL-7 was provided by Cytheris, Inc.. Animals containing WT or IL-7Rαtg P14 CD8 T cells were treated with 5 μg of rhIL-7 per day i.p. from days 8–15 pi.

Real-Time PCR.

Real-time PCR of genomic IL-7Rα was carried out as described (36) by using the following primers with an annealing temperature of 60°C: Exon 7 primer sequence, 5′-gcctagtctccccgatcata-3′; 3′-UTR primer sequence, 5′-ttcttcctgcattcaccaca-3′.

Western Blotting.

Cells were washed twice in PBS and lysed in RIPA buffer (50 mM Tris, pH 8.8/150 mM NaCl/1% Nonidet P-40/0.5% sodium deoxycholate/0.5% SDS/2 mM EDTA/10% glycerol) supplemented with protease and phosphatase inhibitor mixture (Sigma, St. Louis, MO). Thirty micrograms of protein was analyzed by using SDS/PAGE followed by Western blotting using anti-p27kip (1:1,000) and anti-Grp94 (1:5,000) antibodies. Signals were detected by using SuperSignal West Pico CL substrate (Pierce, Rockford, IL) and visualized by autoradiography.

Statistical Analyses.

Where indicated, P values were determined by using a two-tailed unpaired Student's t test. P values <0.05 were considered significant. All graphs show averages of the mean ± SEM.