Collaboration between interleukin-7 and -15 enables adaptation of tissue-resident and circulating memory CD8⁺ T cells to cytokine deficiency

SUMMARY

Interleukin-7 (IL-7) is considered a critical regulator of memory CD8⁺ T cell homeostasis. However, this is primarily based on circulating memory populations, and the cell-intrinsic requirement for IL-7 signaling during memory homeostasis has not been directly tested. Here, we addressed the role for IL-7Rα in circulating and resident memory CD8⁺ T cells (Trm) after their establishment. We found that inducible Il7ra deletion had only a modest effect on persistence of circulating memory and Trm subsets, causing reduced basal proliferation. Loss of IL-15 signaling imposed heightened IL-7Rα dependence on memory CD8⁺ T cells, including Trm cells described as IL-15-independent. In the absence of IL-15 signaling, IL-7Rα was elevated, and loss of IL-7Rα signaling reduced IL-15-elicited proliferation, suggesting crosstalk between these pathways in memory CD8⁺ T cells. Thus, across subsets and tissues, IL-7 and IL-15 act in concert to support memory CD8⁺ T cells, conferring resilience to altered availability of either cytokine.

eTOC Blurb

Circulating memory CD8⁺ T cells are thought to be highly dependent on interleukin (IL)-7 signaling. Using inducible Il7ra deletion in circulating and tissue-resident memory CD8⁺ T cell subsets, Jarjour et al. reveal resilience to loss of IL-7 signaling and collaboration between the IL-7 and IL-15 signaling pathways.

Graphical Abstract

INTRODUCTION

Immunological memory is formed in the wake of a successful immune response and establishes stable cellular and humoral immunity against the eliciting pathogen. Together these form the basis for a potent and protective recall response upon re-exposure to the pathogen, mediating long-term defense of the organism. CD8⁺ T cell memory was first characterized in circulating populations found in the blood and lymphoid organs, but more recently abundant tissue-resident memory CD8⁺ T cells (Trm) have been identified in diverse organs^1–3. These populations are locally maintained independent of contribution from the circulation^4,5, acquire tissue-specific gene expression programs^6–12, and are capable of local expansion and control of infection and cancer at the point of origin^13–17. Perhaps in part because Trm cells adapt to tissue environments, their requirements for generation and maintenance differ from those of circulating memory populations^2,3.

The common gamma chain-(γC)-dependent cytokines interleukin-7 (IL-7) and IL-15 are each thought to be critical for maintenance of CD8⁺ T cell memory, based largely on early studies of circulating populations^18–25. With regard to Trm requirements for IL-7 and IL-15 signaling, levels of the receptor components IL-2 receptor beta (IL-2Rβ) and IL-7 receptor alpha (IL-7Rα) are reduced on some Trm cells^26,27, which could indicate lessened dependence on tonic levels of these cytokines. Indeed, the requirement for IL-15 is in fact variable for memory CD8⁺ T cells (particularly for some Trm cells), depending on the memory subset, route of infection, and number of previous recall responses^28,29. However, IL-15 responsiveness is a conserved feature of circulating and resident memory CD8⁺ T cell subsets, even memory populations that do not depend on IL-15 for normal homeostasis²⁷. IL-7 is considered a crucial regulator of CD8⁺ T cell memory^30–33, though there is limited assessment of this requirement for Trm cells³⁴. Notably, seminal early work used transfer of T cells derived from germline Il7ra^−/− mice and antibody blockade of IL-7 signaling to assess requirements for memory CD8⁺ T cells in the blood and lymphoid tissues^20,22. As noted by the authors, development of T cells is severely compromised in Il7ra^−/− mice, and IL-7Rα blockade may elicit lymphopenia. Global and lifelong deficiency in IL-7Rα could result in confounding defects in naïve CD8⁺ T cells that exaggerate the requirement for IL-7Rα in memory CD8⁺ T cells or, alternatively, adaptation to become IL-7Rα-independent, underestimating the true severity of IL-7Rα deficiency^20,35. Other studies have assessed IL-7 dependence in a variety of ways, with mild to severe defects in the absence of IL-7 signaling during and after memory differentiation^35–41. In light of these concerns, Carrette and Surh³¹ highlighted the importance of cell-intrinsically ablating IL-7 signaling during T cell memory homeostasis, without globally inhibiting IL-7 signaling. Intriguingly, IL-7Rα is regulated by IL-7 and IL-15 signaling in primarily naïve T cell populations, suggesting that loss of one signaling modality could alter the other⁴². Taken together, how IL-7 and IL-15 regulate memory CD8⁺ T cells (and especially the Trm compartment) needs reassessment.

Here, we aimed to specifically address the role for IL-7Rα in homeostasis of circulating and tissue-resident memory CD8⁺ T cells and to resolve how IL-7 and IL-15 jointly regulate CD8⁺ T cell memory. Inducible deletion of Il7ra caused relatively modest impairment of maintenance for circulating and resident memory CD8⁺ T cell subsets formed in response to prior systemic and localized infections. This included Trm subsets resilient to loss of IL-15^28,29. IL-7Rα was required for normal basal proliferation of memory cells during homeostasis in most locales, but IL-7Rα-deficient memory CD8⁺ T cells were capable of antigen-driven proliferation and substantial (though reduced) expansion during recall. Deficiency in IL-15 elevated IL-7Rα expression and greatly augmented IL-7Rα-dependence, suggesting that IL-15-independent CD8⁺ T cell memory is explained by compensatory IL-7 signaling. Loss of signaling for both cytokines caused a pronounced defect in memory cells in all sites tested. Furthermore, loss of IL-7Rα during memory homeostasis resulted in reduced proliferation in response to IL-15 treatment, suggesting cooperativity between the IL-7 and IL-15 signaling pathways. We propose a model for cytokine control of CD8⁺ T cell memory based on integrated regulation by IL-7 and IL-15 with capacity for adaptation in the absence of either cytokine.

RESULTS

Inducible deletion of Il7ra in established memory CD8⁺ T cells has modest effects on maintenance but alters proliferation.

To assess whether IL-7Rα was stringently required for established memory CD8⁺ T cell cells, Il7ra^fl/fl mice were backcrossed and intercrossed with the tamoxifen-inducible Ubc-cre^ERT2 and P14 T cell receptor (TCR) transgenic lines. Congenically distinct control (Ubc-cre^ERT2 Il7ra^+/+ or Cre-negative Il7ra^fl/fl) and Ubc-cre^ERT2 Il7ra^fl/fl P14 T cells were adoptively co-transferred to recipients, followed by infection with lymphocytic choriomeningitis virus (LCMV) Armstrong to generate memory cells. Tamoxifen could then be given at a memory timepoint to ablate Il7ra in memory P14 cells after their differentiation (Figure 1A) and specifically interrogate the importance of IL-7Rα during memory homeostasis. Co-transfer was used so that (after administration of tamoxifen) IL-7Rα-sufficient and -deficient P14 T cells could compete within the same recipient, increasing sensitivity to detect an advantage for IL-7Rα-sufficient memory cells. First, deletion was confirmed via staining for CD127 (IL-7Rα), revealing robust tamoxifen-inducible loss of CD127 on Ubc-cre^ERT2 Il7ra^fl/fl P14 T cells across tissues, as well as generally lower CD127 expression by control P14 T cells in non-lymphoid tissues (most clearly for small intestine intraepithelial lymphocytes [IEL], salivary gland [SG], and female reproductive tract [FRT]) (Figures 1B–1D)²⁶. Loss of CD127 was essentially completed within a week of starting tamoxifen treatment (data not shown). While a severe defect might have been predicted in the absence of IL-7Rα, at least for circulating memory populations²⁰, lack of Il7ra only conferred a minor competitive disadvantage (~2–3 fold by ~2 months post-tamoxifen) and did not result in selective loss of lymphoid or nonlymphoid tissue (NLT) memory populations (Figures 1E, S1A, and S1B). This is a notable contrast with IL-15 deficiency, which has been reported to cause substantial defects in memory CD8⁺ T cell homeostasis across lymphoid tissues and several non-lymphoid tissue sites, although loss of IL-15 has very little effect on Trm in other sites (such as the small intestine and FRT)²⁸. It was possible that memory cells within these IL-15-independent sites (IEL, FRT) were highly IL-7-dependent. However, this simplistic explanation did not hold true in our hands, indicating that cytokine regulation of CD8⁺ T cell memory has an additional layer of complexity.

Figure 1: Circulating and resident memory P14 T cells are resilient to loss of Il7ra during memory homeostasis.

(A) Generation of congenic control (Ubc-cre^ERT2 Il7ra^+/+ or Il7ra^fl/fl) and Ubc-cre^ERT2 Il7ra^fl/fl memory P14 T cells and treatment with tamoxifen. (B) Representative flow cytometry for CD127 (IL-7Rα) on control and Ubc-cre^ERT2 Il7ra^fl/fl memory P14 T cells and quantitation for (C) percent CD127⁺ and (D) CD127 gMFI as in (B) from untreated and tamoxifen-treated mice. (E) Quantitation of the ratio of control and Ubc-cre^ERT2 Il7ra^fl/fl P14 T cells from untreated and tamoxifen-treated mice. (F) Quantitation of the proportions of NLT memory P14 subsets (DN, CD69⁻ CD103⁻. DP CD69⁺ CD103⁺) from each donor from untreated and tamoxifen-treated mice. (G) Representative flow cytometry for CD62L and KLRG1 for splenic memory P14 T cells and (H) quantitation of the proportion of long-lived effector cells (LLEC, KLRG1⁺ CD62L⁻), T effector memory (Tem, KLRG1⁻ CD62L⁻), and T central memory (Tcm, KLRG1⁻ CD62L⁺) from each donor from untreated and tamoxifen-treated mice as in (G). LN, inguinal LN. IEL, small intestine intraepithelial lymphocytes. SG, salivary gland. FRT, female reproductive tract. Data are (B,G) representative of 3 experiments (n=6–9/group), (C,E,F,H) compiled from 2–5 experiments (n=4–10/group), or (D) compiled from 1–4 experiments (n=3–9/group). Error is expressed as ± S.E.M. Unpaired (C,E) and paired Student’s t tests (D,F,H).* p<0.05. ** p<0.01. *** p<0.001. **** p<0.0001. See also Figures S1 and S2.

To compare IL-7Rα dependence of established memory CD8⁺ T cells to that of a population known to be highly IL-7Rα-dependent^20,43,44, naïve control and Ubc-cre^ERT2 Il7ra^fl/fl P14 T cells were co-transferred into naïve recipients (Figure S2A). In tamoxifen-treated mice, induced Il7ra-deficient naïve P14 T cells were rapidly lost and exhibited a profound competitive disadvantage (>25 fold by ~2 weeks post-tamoxifen), as previously described⁴⁴ and in stark contrast with memory P14 T cells (Figures 1E versus S2B and S2C). By two weeks after tamoxifen treatment, remaining Ubc-cre^ERT2 Il7ra^fl/fl naïve CD8⁺ T cells were largely CD127-positive escapees of Cre-mediated recombination, in contrast to memory P14 T cells, which largely remained CD127-negative (Figures 1C and 1D versus S2D). Therefore, memory CD8⁺ T cells exhibited a modest dependence on IL-7Rα across subsets and far greater resilience than their naïve counterparts to loss of IL-7Rα.

After administration of tamoxifen, Il7ra-deficient P14 T cells showed relatively normal proportions of Trm phenotype subsets, but a slightly higher proportion of KLRG1⁺ long-lived effector cells (LLEC) and a slightly lower proportion of CD62L⁺ central memory cells (Tcm) (Figures 1F–1H, S1C, and S1D). In all experiments, intravenous (i.v.) labelling was used to discriminate vascular and tissue-localized memory cells⁴⁵, while CD69 and CD103 expression was also used to identify Trm^4,27. No clear differences were observed when assessing necrotic and apoptotic cells via Annexin V (Figure S1E). However, consistent with previous reports^20,39,40, Bcl2 staining was reduced for circulating memory cells in the absence of IL-7Rα, though Bcl2 levels of IL-7Rα-deficient memory cells remained higher than those of naïve T cells (Figures S1F and S1G)⁴⁶. Taken together, induced ablation of Il7ra in established, mature memory CD8⁺ T cells did not result in a strong predisposition towards cell death.

However, when proliferation of memory P14 T cells was assessed after ablation of Il7ra, we observed a consistent reduction in the proportion of cycling cells across circulating memory populations using both Ki67 staining and bromodeoxyuridine (BrdU) incorporation (as an indicator of DNA synthesis during S phase) (Figures 2A–2D, S3A, and S3B). Proliferation was also reduced for most NLT populations, but memory P14 T cells in the SG and IEL were only modestly affected (Fig. 2A–2D, S3A, and S3B). With regard to proliferation of specific memory subsets, induced loss of Il7ra affected all circulating memory populations, but again had a more complex effect on NLT-localized populations (Figures S3C–F). However, generally consistent trends were observed for different subsets within the same tissue. Thus, IL-7Rα regulates self-renewal of memory CD8⁺ T cells, with the most pronounced effect on circulating memory populations.

Figure 2: Loss of Il7ra alters proliferation of memory CD8⁺ T cell subsets.

(A) Representative flow cytometry for Ki67 expression and (B) quantitation of the proportion of Ki67-expressing control and Ubc-cre^ERT2 Il7ra^fl/fl memory P14 T cells as in (A) from untreated and tamoxifen-treated mice. (C) Representative flow cytometry for BrdU incorporation and (D) quantitation of the proportion of BrdU-incorporating control and Ubc-cre^ERT2 Il7ra^fl/fl memory P14 T cells as in (C) from untreated and tamoxifen-treated mice. Data are (A,C) representative of 3 experiments (n=5–9/group) or (B,D) compiled from 3–5 experiments (n=5–9/group). Error is expressed as ± S.E.M. Paired Student’s t tests. * p<0.05. ** p<0.01. *** p<0.001. **** p<0.0001. See also Figure S3.

Oligoclonal tetramer-binding memory CD8⁺ T cells are resilient to loss of Il7ra during memory homeostasis

As P14 TCR transgenic T cells could be unusual in their IL-7-dependence, we also assessed the impact of induced Il7ra loss from LCMV-specific polyclonal CD8⁺ T cells. Mixed bone marrow chimeras were generated using congenically distinct control (Il7ra^fl/fl) and Ubc-cre^ERT2 Il7ra^fl/fl marrow transferred into irradiated wildtype recipients. After reconstitution, these animals were infected with LCMV to elicit antigen-specific memory populations trackable with tetramers for the D^b-restricted gp33, gp276, and NP396 epitopes of LCMV (Figure 3A). After establishment of memory populations, tamoxifen was given to delete Il7ra, allowing us to track the effects of IL-7Rα deficiency for these oligoclonal populations of different specificities. Employing mixed chimeras allowed us to assess direct competition between IL-7Rα-sufficient and -deficient naïve and memory CD8⁺ T cells as well as to mitigate the impact of potential lymphopenia (due to loss of IL-7Rα) via control bone marrow cells.

Figure 3: Oligoclonal memory CD8⁺ T cells of defined antigen specificity are resilient to loss of Il7ra during memory homeostasis.

(A) Generation of Il7ra^fl/fl Ubc-cre^ERT2-negative (control) and -positive bone marrow chimeras, elicitation of LCMV-specific memory CD8⁺ T cells, and tamoxifen treatment. (B-D) Quantitation of control and Ubc-cre^ERT2 Il7ra^fl/fl naïve (CD44^lo CD62L⁺) CD8⁺ T cells for (B) ratio, (C) proportion CD127⁺, and (D) ratio of control to CD127⁻ Ubc-cre^ERT2 Il7ra^fl/fl naïve cells from untreated and tamoxifen-treated bone marrow chimeras. (E) Quantitation of the ratio of control and Ubc-cre^ERT2 Il7ra^fl/fl gp276/D^b and NP396/D^b tetramer-binding memory cells from untreated and tamoxifen-treated bone marrow chimeras. (F-H) Quantitation of the (F) proportion CD127⁺, (G) CD127 gMFI, and (H) proportion of Ki67-expressing cells for control and Ubc-cre^ERT2 Il7ra^fl/fl gp276/D^b and NP396/D^b tetramer-binding memory cells from untreated and tamoxifen-treated bone marrow chimeras. Data are compiled from 2–7 experiments (n=5–12/group). Error is expressed as ± S.E.M. Unpaired (B-G) and paired Student’s t tests (H). * p<0.05. ** p<0.01. *** p<0.001. **** p<0.0001. See also Figure S4.

As expected^20,43,44 and as we observed in the P14 system, loss of Il7ra conferred a disadvantage to naïve CD44^lo CD62L⁺ polyclonal CD8⁺ T cells, which also experienced strong selection for IL-7Rα (CD127)-expressing escapees of Il7ra deletion (Figures 3B and 3C). When the ratio of control Il7ra^fl/fl cells to CD127-negative Ubc-cre^ERT2 Il7ra^fl/fl cells was calculated to account for CD127-positive escapees, favoritism for IL-7Rα-sufficient naïve phenotype CD8⁺ T cells became even more stark (Figures 3D). In contrast, for memory CD8⁺ T cells binding pooled gp276/D^b and NP396/D^b tetramers or gp33/D^b tetramer, inducible loss of Il7ra again had only a relatively modest effect (Figures 3E and S4A–S4C). Furthermore, the proportion of CD127-positive, LCMV-specific escapee memory cells was relatively stable over time, in contrast to the strong selection observed for IL-7Rα-sufficient naïve CD8⁺ T cells (Figures 3F, 3G, S4D, and S4E). Tetramer-binding memory populations showed reduced proliferation in the absence of Il7ra, especially for circulating populations (Figures 3H, S4F, and S4G). While trends were generally comparable to the P14 system, some differences in IL-7Rα dependence of proliferation were observed, particularly for the salivary gland (which exhibited greater IL-7Rα dependence in the chimera system). Among other explanations, it is possible that loss of IL-7Rα from half of the hematopoietic compartment in bone marrow chimeras resulted in a degree of lymphopenia, or alternatively that precursor frequency or TCR specificity may alter the degree of IL-7Rα dependence for proliferation. Taken together, resilience to loss of IL-7Rα was a conserved property across the P14 TCR transgenic and oligoclonal antigen-specific populations, with a conserved role for IL-7Rα in maintaining a normal proportion of proliferating memory CD8⁺ T cells. Loss of IL-7Rα did not result in rapid and severe attrition of CD8⁺ T cell memory.

IL-7Rα dependence is modest across multiple models of CD8⁺ T cell memory

We next assessed whether our findings were due to unique features of memory cells elicited by LCMV or were also representative of CD8⁺ T cell memory generated in response to other pathogens. We investigated systemic infection with the bacterial pathogen Listeria monocytogenes (Lm-gp33, expressing the gp33 antigen recognized by P14 T cells⁴⁷), as well as lung-localized infection with influenza virus (PR8-gp33, expressing gp33⁴⁸) given intranasally. CD8⁺ T cell memory induced by intranasal influenza infection has been described as IL-15-independent²⁹.

Much as for systemic LCMV infection, Lm-gp33-elicited memory CD8⁺ T cells only showed modest IL-7Rα dependence for maintenance and comparable alterations in proliferation, Bcl2 expression, and circulating memory subsets (Figures 4A–4E and S5A–S5C). As influenza virus-elicited lung Trm have been described to more rapidly wane over time than is typical for other Trm^49–52, we wondered whether this population might be especially vulnerable to loss of IL-7 signaling. However, in the setting of intranasal PR8-gp33 infection we found comparable IL-7Rα dependence of circulating and lung-resident memory subsets to that observed for LCMV- and Lm-gp33-elicited memory (Figures 4F–4I). When we specifically focused on i.v. label low (IV^lo) CD69⁺ CD103⁺ lung Trm, this population was equally or less IL-7Rα-dependent than circulating (IV^hi) and other tissue-localized (IV^lo) memory CD8⁺ T cell subsets found in the lung (Figures S5D and S5E). While the proportion of IL-7Rα-deficient Tcm was again reduced in this model, the proportion of IL-7Rα-deficient T effector memory (Tem) was more clearly expanded than in the LCMV and Lm-gp33 models, perhaps due to scarcity of PR8-gp33-elicited LLEC at this timepoint. Similarly to memory CD8⁺ T cells generated in response to LCMV and Lm-gp33, PR8-gp33-elicited memory CD8⁺ T cells exhibited a degree of IL-7Rα dependence for proliferation and Bcl2 expression (Figures 4J and S5F–S5H). Just as for memory CD8⁺ T cells formed during LCMV infection, influenza virus-elicited memory populations previously reported as IL-15-independent²⁹ did not show a more profound dependence on the IL-7Rα than that observed for IL-15-dependent memory subsets in other models. Thus, LCMV-induced memory CD8⁺ T cells were not unique in their resilience to loss of IL-7Rα. Loss of IL-7Rα from established memory cells resulted in generally comparable effects across memory subsets generated in response to different pathogens.

Figure 4: Circulating and resident memory P14 T cells share resilience to loss of Il7ra across multiple models of pathogen-elicited memory.

(A-E) Lm-gp33-elicited memory. (A) Quantitation of the ratio of control and Ubc-cre^ERT2 Il7ra^fl/fl P14 T cells from untreated and tamoxifen-treated mice. (B) Quantitation for (Left) percent CD127⁺ and (Right) CD127 gMFI from untreated and tamoxifen-treated mice. (C and D) Quantitation of the proportions of (C) circulating and (D) NLT memory P14 subsets from each donor from untreated and tamoxifen-treated mice. (E) Quantitation of the proportion of Ki67-expressing control and Ubc-cre^ERT2 Il7ra^fl/fl memory P14 T cells from untreated and tamoxifen-treated mice. (F-J) PR8-gp33-elicited memory. (F) Quantitation of the ratio of control and Ubc-cre^ERT2 Il7ra^fl/fl P14 T cells from untreated and tamoxifen-treated mice. (G) Quantitation for (Left) percent CD127⁺ and (Right) CD127 gMFI from untreated and tamoxifen-treated mice. (H and I) Quantitation of the proportions of (H) circulating and (I) lung memory P14 subsets (DN, CD69⁻ CD103⁻. DP CD69⁺ CD103⁺) from each donor from untreated and tamoxifen-treated mice. (J) Quantitation of the proportion of Ki67-expressing control and Ubc-cre^ERT2 Il7ra^fl/fl memory P14 T cells from untreated and tamoxifen-treated mice. MedLN, mediastinal lymph node. IV, i.v. label. Data are (A-E) compiled from 2 experiments (n=2–6/group) or (F-J) compiled from 2 experiments (n=6–8/group). Error is expressed as ± S.E.M. Unpaired (A,F) and paired Student’s t tests (B-E,G-J). * p<0.05. ** p<0.01. *** p<0.001. **** p<0.0001. See also Figure S5.

Efficient recall of memory CD8⁺ T cells can occur in the absence of IL-7Rα

While CD8⁺ T cell memory was relatively resilient to inducible deletion of Il7ra, capacity to recall in response to antigen might still be impaired. For example, IL-2Rα is required for normal programming of circulating memory CD8⁺ T cells during memory differentiation. In its absence, memory CD8⁺ T cell homeostasis is largely normal, with a severe defect appearing only upon recall⁵³. Therefore, we deleted Il7ra from established memory P14 T cells, waited >4 weeks, and then subjected mice to heterologous challenge with Lm-gp33 (Figure 5A). Lm-gp33 was used to specifically recall gp33-specific P14 T cells and avoid confounding effects of antibodies against other LCMV antigens. To track the kinetics of the recall response, animals were bled over time. In the blood, we observed a robust increase in control and Ubc-cre^ERT2 Il7ra^fl/fl P14 T cells after recall, near universal acquisition of Ki67 and granzyme B, and a profound shift from CD62L⁺ central memory phenotype cells to KLRG1⁺ effector phenotype cells (Figures 5B–5H). Nevertheless, we observed a slight increase in the ratio of control to IL-7Rα-deficient cells, potentially consistent with a modest defect in recall of memory CD8⁺ T cells lacking IL-7Rα.

Figure 5: IL-7Rα is required during memory homeostasis for maximal antigen-elicited recall of circulating memory.

(A) Recall of control (Ubc-cre^ERT2 Il7ra^+/+ or Il7ra^fl/fl) and Ubc-cre^ERT2 Il7ra^fl/fl memory P14 T cells in untreated and tamoxifen-treated mice infected intravenously with Lm-gp33. (B) Control and Ubc-cre^ERT2 Il7ra^fl/fl P14 T cells were quantitated per microliter (μL) blood from untreated and tamoxifen-treated Lm-gp33-reactivated mice and Lm-gp33-naïve controls. (C and D) Quantitation of the (C) percent (out of all CD8⁺ T cells) and (D) ratio of control and Ubc-cre^ERT2 Il7ra^fl/fl P14 T cells from the blood of untreated and tamoxifen-treated Lm-gp33-reactivated mice. (E) Quantitation of the proportion of Ki67-expressing (left) and granzyme B-expressing (right) control and Ubc-cre^ERT2 Il7ra^fl/fl P14 T cells from the blood of untreated and tamoxifen-treated Lm-gp33-reactivated mice. (F) Representative flow cytometry for KLRG1 and CD62L expression, (G) quantitation of the proportion of KLRG1⁺ CD62L⁻ P14 T cells as in (F), and (H) quantitation of the proportion of KLRG1⁻ CD62L⁺ P14 T cells as in (F) from each donor from untreated and tamoxifen-treated Lm-gp33-reactivated mice. Data are compiled from (B-E,G,H) 3 experiments (n=6–9/group), except granzyme B data (1 experiment, n=2–3/group), (F) Data are representative of 3 experiments (n=6–7/group). Error is expressed as ± S.E.M. Unpaired (D) and paired Student’s t tests (B-C,E,G,H). * p<0.05. ** p<0.01. *** p<0.001. **** p<0.0001. See also Figure S5.

It was possible that this moderate impairment in recall expansion was solely a consequence of the increased LLEC to Tcm ratio of IL-7Rα-deficient memory cells (Figures 5G and 5H), since recall proliferation of LLEC is limited⁵⁴. Accordingly, we considered whether IL-7Rα-deficient LLEC and Tcm were defective in recall on a per-cell basis. We harvested recipient mice that had previously experienced LCMV infection and subsequent tamoxifen treatment to sort control and Ubc-cre^ERT2 Il7ra^fl/fl P14 LLEC and Tcm. For Tcm, we additionally used CD127 expression to specifically sort on CD127-negative Ubc-cre^ERT2 Il7ra^fl/fl cells that had successfully recombined Il7ra, excluding escapees of deletion from transfer. We recombined the sorted cells (either LLEC+LLEC or Tcm+Tcm) in a defined ratio (varied from ~2:1 to 1:1, control to Ubc-cre^ERT2 Il7ra^fl/fl) and transferred them into separate groups of naïve recipients, followed by rechallenge with LCMV (Figure S5I). We could then assess whether control cells were favored for recall. Indeed, control Tcm exhibited a modest competitive advantage over time compared to IL-7Rα-deficient Tcm (Figure S5J). In contrast, Ubc-cre^ERT2 Il7ra^fl/fl LLEC expanded equivalently (or even slightly better) than control LLEC (Figure S5J), though we could not distinguish Il7ra deleters from escapees because of low CD127 expression on LLEC. In both cases, loss of Il7ra did not impair acquisition of Ki67 or granzyme B (Figure S5K). Thus, our data are most consistent with a modest Tcm defect in recall in the absence of IL-7Rα, which is due to both altered proportions of circulating memory subsets and slightly weaker recall of Tcm on a per-cell basis. Overall, however, loss of IL-7Rα during memory homeostasis did not profoundly compromise the ability of circulating memory CD8⁺ T cells to recall.

Deficiency in IL-15 signaling confers heightened dependence on IL-7Rα

While our data did not support the notion that IL-15-independent memory populations^28,29 were especially IL-7Rα-dependent, we considered the possibility that memory CD8⁺ T cells are instead regulated by the combination of IL-7 and IL-15 signaling. The lack of exclusive IL-15 dependence for some memory CD8⁺ T cell subsets does not preclude a role for IL-15 in concert with other cytokines in regulating such memory populations. As mentioned above, IL-15-independent memory CD8⁺ T cells respond robustly to IL-15 therapy²⁷, supporting that memory cells adapt to γC cytokine availability irrespective of homeostatic requirements. Furthermore, enhanced severity of a combinatorial defect in IL-7 and IL-15 signaling has been proposed for circulating memory in the setting of a lymphopenic host or after transfer of day 10 effector cells^22,35,36,55. Such redundancy could explain why IL-7 and IL-15 are individually less critical to CD8⁺ T cell memory than predicted and would have significant implications for their stability in the face of altered cytokine availability. We therefore considered whether IL-7Rα-deficient memory cells were more severely affected in the absence of IL-15.

To address this, control and Ubc-cre^ERT2 Il7ra^fl/fl P14 T cells were transferred to IL-15-sufficient (Il15^+/+ or Il15^+/−) and Il15^−/− recipient mice followed by LCMV infection to induce memory differentiation. First, by assessing control memory P14 T cell populations in IL-15-sufficient and - deficient recipients, we confirmed IL-15 independence of LCMV-elicited P14 memory cells in the IEL and FRT²⁸, while also establishing that liver memory cells have a moderate IL-15 dependence (Figure S6A). To determine the combinatorial effects of IL-7Rα and IL-15 deficiency, tamoxifen-treated IL-15-sufficient recipients were compared with untreated- and tamoxifen-treated Il15^−/− recipients. When tracking memory P14 populations in the blood after administration of tamoxifen, a greatly enhanced competitive advantage was observed for IL-7Rα-sufficient P14 cells in Il15^−/− recipients (Figure 6A). This led to profound selection for Ubc-cre^ERT2 Il7ra^fl/fl CD127-expressing escapees of deletion over time; both features are consistent with a heightened requirement for IL-7Rα in the absence of IL-15 signaling (Figure 6B). When an adjusted ratio of control to genuinely CD127-negative Ubc-cre^ERT2 Il7ra^fl/fl P14 cells was calculated, this revealed a nearly 100:1 competitive advantage for control cells in IL-15-deficient hosts (Figure 6C). In light of the outgrowth of CD127-expressing escapees of deletion, we took advantage of variation in the efficiency of Il7ra deletion using three independent cohorts, one with high efficiency deletion of Il7ra (Cohort 1, estimated at ~90–95% for blood Tcm from concurrent wildtype recipients) and two with moderate efficiency deletion of Il7ra (Cohorts 2 and 3, estimated at ~70–75% in the same manner) to address two related points (Figures S6B and S6C). In Cohort 1, escapees of deletion were quite rare, allowing clear assessment of the severe competitive disadvantage of lacking Il7ra across tissues and memory subsets in Il15^−/− recipients by ratio (Figures 6D–6F and S6D). However, it was challenging to accurately assess the phenotype of the remaining Ubc-cre^ERT2 Il7ra^fl/fl P14 cells, especially in NLT, because of their extreme scarcity, though it appeared that a significant majority of these cells were CD127-positive escapees of Cre activity (Figure 6E, 6F, and S6E). However, in Cohorts 2 and 3, while the higher proportion of escapees limited the ratio that could be observed between donors (Figure S6F), it allowed observation of competition within the Ubc-cre^ERT2 Il7ra^fl/fl P14 population after tamoxifen. Across all cohorts combined deficiency in IL-7Rα and IL-15 resulted in reductions in P14 counts (Figure S6G). While CD127-positive escapees in B6 recipients had only a small advantage, persisting Ubc-cre^ERT2 Il7ra^fl/fl P14 cells in Il15^−/− hosts became nearly 100% CD127-positive in circulating populations and exhibited nearly normal gMFI for CD127 in lymphoid and NLT sites, consistent with selection for IL-7Rα-sufficient escapees (Figures 6G, S6H, and S7A–C). Therefore, lack of IL-15 imposed a heightened requirement for IL-7Rα and revealed cooperativity between γC cytokines which conferred resilience to altered cytokine availability.

Figure 6: Redundancy between IL-7 and IL-15 affords resilience to loss of individual cytokines.

Control (Ubc-cre^ERT2 Il7ra^+/+ or Il7ra^fl/fl) and Ubc-cre^ERT2 Il7ra^fl/fl memory P14 T cells were generated in IL-15-sufficient (Il15^+/+ or Il15^+/−) or -deficient recipients, followed by tamoxifen treatment. (A-C) Control and Ubc-cre^ERT2 Il7ra^fl/fl memory P14 T cells were quantitated for the (A) ratio, (B) proportion CD127⁺, and (C) adjusted ratio of control to CD127-negative Ubc-cre^ERT2 Il7ra^fl/fl memory P14 T cells from the blood of untreated and tamoxifen-treated mice. (D-F) Cohort 1 control and Ubc-cre^ERT2 Il7ra^fl/fl P14 T cells were quantitated for the (D) ratio, (E) proportion CD127⁺, and (F) adjusted ratio of control and CD127-negative Ubc-cre^ERT2 Il7ra^fl/fl memory P14 T cells from untreated and tamoxifen-treated mice. (G) Representative flow cytometry for CD127 expression on Cohort 2 control and Ubc-cre^ERT2 Il7ra^fl/fl P14 T cells from Il15^+/− or Il15^−/− recipients. Data are (A-C) compiled from 3 experiments (6–9/group), (D-F) representative of/derived from 1 experiment (n=2–3/group, except SG, No tamoxifen [n=1]), or (G) representative of 2 experiments (n=3–6/group). Error is expressed as ± S.E.M. Unpaired (A-D,F) or paired (E) Student’s t tests. * p<0.05. ** p<0.01. *** p<0.001. **** p<0.0001. See also Figures S6 and S7.

Integrated regulation of IL-7 and IL-15 signaling modulates CD8⁺ T cell memory

In naïve T cells, IL-7 (and other cytokines including IL-15) decrease mRNA and protein for IL-7Rα, a potential mechanism to conserve IL-7 and allow signaling for as many cells as possible⁴². However, it has not been determined whether a similar mechanism holds for memory cells. CD127 expression appeared to be increased on control P14 T cells in Il15^−/− recipient mice (Figure 6G), so we quantitated this across IL-15-sufficient and -deficient animals. Strikingly, CD127 was elevated on control memory P14 T cells from most tissues of Il15^−/− recipients (Figures 7A). When we assessed CD127 levels across circulating and resident memory CD8⁺ T cell subsets, we likewise found that lack of IL-15 in the recipient mouse correlated with elevated CD127 across subsets (Figure 7B). In vivo treatment with either IL-7 or IL-15 complexes (IL-7c and IL-15c, a more potent and long-lasting method to administer cytokine^27,56–58) decreased CD127 on circulating memory CD8⁺ T cells (Figures 7C–7F). Reduced IL-7Rα in response to IL-15c treatment suggested that the observed increase in CD127 staining across memory subsets in Il15^−/− recipients could be due to a lack of IL-15 signaling and may act as a compensatory mechanism to allow enhanced IL-7 sensitivity. While our results with cytokine therapy were consistent with memory cells modulating IL-7Rα levels acutely in response to cytokine availability, this did not address whether altered CD127 on memory P14 T cells in Il15^−/− mice was adaptive and reversible. To interrogate this, we generated P14 memory T cells in IL-15-sufficient and - deficient mice, followed by harvest of splenocytes and enrichment for CD8⁺ T cells. We then transferred ~1.5–2M CD8⁺ T cells into naïve recipients that were either sufficient or deficient for IL-15. We allowed transferred cells to rest without rechallenge to determine how CD127 levels on P14 memory cells changed if their access to IL-15 was altered. We generated three groups of recipient mice. Two used memory CD8⁺ T cells formed in a wildtype host and transferred to either wildtype or Il15^−/− recipients. The third used memory CD8⁺ T cells generated in an Il15^−/− host and transferred to wildtype recipients (Figure S7D). We found that even after memory differentiation, transfer of memory cells into a new recipient could alter CD127 expression (Figures S7E–G). Memory cells generated in wildtype hosts and transferred to Il15^−/− hosts exhibited elevated CD127 expression compared to memory cells generated in wildtype and Il15^−/− hosts and transferred to wildtype recipients. Thus, elevated CD127 expression on memory CD8⁺ T cells in Il15^−/− mice was at least in part due to reversible adaptation.

Figure 7: Crosstalk between the IL-7 and IL-15 signaling pathways modulates CD8⁺ T cell memory.

(A and B) Control (Ubc-cre^ERT2 Il7ra^+/+ or Il7ra^fl/fl) and Ubc-cre^ERT2 Il7ra^fl/fl memory P14 T cells as in Figure 6. Quantitation of CD127 gMFI for (A) Left-lymphoid tissue- and Right-non-lymphoid tissue-localized control memory P14 T cells (some data points are also shown in Figs. S6E, S7B, and S7C) and (B) Left-circulating and Right-NLT-localized control memory P14 subsets from untreated and tamoxifen-treated mice. (C-F) Wildtype memory P14 T cells were generated and (after >28 days), recipient mice were given PBS and (C and D) IL-7c or (E and F) IL-15c on day 0 and 2 with sacrifice on day 4. (C and E) Representative flow cytometry for CD127 expression on memory P14 T cells from PBS- and cytokine complex-treated mice and (D and F) quantitation of CD127 gMFI as in (C and E). (G, H) Control (Il7ra^fl/fl) and Ubc-cre^ERT2 Il7ra^fl/fl memory P14 T cells were generated as in Figure 1. Recipient mice were then treated with PBS or 2 μg IL-15. (G) Quantitation of the proportion of Ki67-expressing control and Ubc-cre^ERT2 Il7ra^fl/fl memory P14 T cells from PBS or IL-15-treated mice. (H) Quantitation of the proportion of Ki67-expressing control, Ubc-cre^ERT2 Il7ra^fl/fl CD127⁺, and Ubc-cre^ERT2 Il7ra^fl/fl CD127⁻ memory P14 T cells from PBS or IL-15-treated mice. Data are (A,B,D,F-H) compiled from 2–4 experiments (n=4–18/group) or (C,E) representative of 2–3 experiments (n=5–11/group). Error is expressed as ± S.E.M. Unpaired (A,B,D,F) or paired (G,H) Student’s t tests. * p<0.05. ** p<0.01. *** p<0.001. **** p<0.0001. See also Figure S7.

As IL-7Rα was regulated by IL-15 signaling, we then asked whether loss of IL-7Rα from homeostatic memory P14 T cells altered IL-15 responsiveness. We first assessed whether inducible deletion of Il7ra resulted in increased IL-2Rβ (CD122). However, there was no clear evidence of this (Figures S7H and S7I), suggesting that any altered sensitivity of IL-7Rα-deficient memory CD8⁺ T cells to IL-15 did not occur at the level of receptor expression. We next treated mice with a low dose of IL-15 (2 μg) to induce an intermediate response⁵⁹ of memory CD8⁺ T cells and thereby assess whether IL-7Rα-sufficient and -deficient memory cells differed in IL-15 sensitivity. Indeed, loss of IL-7Rα from homeostatic memory cells impaired proliferation in response to low dose IL-15 for circulating memory P14 T cells as well as some resident populations to a more modest extent (Figures 7G and S7J). This phenotype was enhanced when specifically gating on CD127-negative Ubc-cre^ERT2 Il7ra^fl/fl P14 T cells in the blood and lymphoid tissues (Figure 7H). Taken together, the IL-7 and IL-15 signaling pathways appear to regulate each other in memory CD8⁺ T cells, consistent with both compensatory and cooperative relationships between these cytokines to support memory CD8⁺ T cells

DISCUSSION

The notion that IL-7 and IL-15 each fulfill exclusive roles essential for memory CD8⁺ T cell maintenance has been widely accepted in the field, despite reports potentially inconsistent with this model^{22,28,29,38,41}. Previous experimental approaches to address the role of IL-7 signaling are difficult to interpret because of potential confounding effects on hematopoiesis, T cell development, naïve T cell homeostasis, and memory differentiation, with substantial T-lymphopenia when IL-7 or its receptor are ablated or blocked. Using inducible deletion of a floxed Il7ra allele, we rigorously assessed the cell-intrinsic role of IL-7Rα in homeostasis of defined, antigen-specific populations of memory CD8⁺ T cells in lymphoreplete hosts. Counter to prevailing models, IL-7Rα played a nuanced role in maintenance of memory CD8⁺ T cells, with ablation leading to impaired proliferation of memory CD8⁺ T cells in most sites and gradual attrition. However, loss of IL-15 greatly augmented dependence on IL-7Rα, even among Trm populations that are classified as IL-15-independent. These data, together with the observation that loss of either IL-7 or IL-15 signaling affected the signaling pathway for the other cytokine, support that memory CD8⁺ T cells dynamically adapt to available homeostatic cytokines, and thus that IL-7 or IL-15 “dependence” is not fixed.

A major focus of this study was to investigate the roles of IL-7 and IL-15 in Trm. Because these populations adapt to the tissues in which they reside^6–12, it would not be surprising if Trm in distinct sites exhibited different homeostatic requirements, consistent with reduced IL-2Rβ and IL-7Rα expression on some Trm^26,27. Indeed, Trm populations in some sites are IL-15-independent^28,29, while Trm in other tissues stringently require IL-15 much like circulating memory, as confirmed in our study. It was unclear whether IL-15-independent Trm would exhibit heightened sensitivity to IL-7 or, alternatively, no requirement for either IL-7 or IL-15 at all. Instead, induced Il7ra ablation caused gradual decline of memory CD8⁺ T cells in all sites, a resilience that is presumably acquired during memory differentiation as it is not shared with naïve CD8⁺ T cells⁴⁴. Interestingly, the most notable impact of IL-7Rα loss was impaired basal proliferation of most memory CD8⁺ T cells (although this was muted among Trm in some sites), in contrast to the prevailing notion that IL-7 predominantly regulates memory cell survival in normal (i.e. non-irradiated) hosts^20,22. Goldrath et al noted that blockade of IL-7Rα may elicit lymphopenia-induced proliferation (i.e. via IL-15). This likely confounded previous analyses, but using inducible deletion of Il7ra, we now describe a role for IL-7Rα in basal proliferation of most memory CD8⁺ T cell subsets. While changes were not observed in cell death, loss of IL-7Rα during memory homeostasis did reduce Bcl2 levels on circulating memory cells, as predicted based on previous work^20,39,40. Therefore, future studies should account for the impact of IL-7Rα on both self-renewal and survival of memory CD8⁺ T cells. Despite their defect in basal proliferation, IL-7Rα-deficient circulating memory CD8⁺ T cells were capable of robust antigen-elicited proliferation and acquisition of granzyme B during recall. This suggests that cells which are quiescent prior to TCR engagement are not excluded from recall in favor of memory cells actively undergoing basal proliferation. Despite undergoing strong proliferation in response to recall, IL-7Rα-deficient memory CD8⁺ T cells were not able to maximally expand to the same extent as IL-7Rα-sufficient memory cells. This defect seemed to be caused by a subtly reduced resting proportion of IL-7Rα-deficient Tcm and a modest competitive disadvantage in their expansion during recall compared to control Tcm. Interestingly, even though resting IL-7Rα-deficient Tem and LLEC were modestly disfavored compared to IL-7Rα-sufficient Tem and LLEC, they appeared to be somewhat less affected by IL-7Rα deficiency than were Tcm in homeostasis. Furthermore, recall of IL-7Rα-deficient LLEC was at least equal to that of control LLEC. Thus, while our data support a shared resilience to IL-7Rα loss across memory CD8⁺ T cell subsets, there may be even greater capacity to support loss of IL-7 signaling for select subsets like LLEC.

While loss of IL-7Rα did perturb basal proliferation, memory CD8⁺ T cell subsets experienced only gradual attrition over time (2–3 fold up to 2.5 months after tamoxifen), likely due to low requirements for self-renewal. In contrast, CD8⁺ T cells lacking normal IL-7Rα signaling during activation and memory differentiation show a competitive defect in excess of 10:1 approximately one month after transfer^20,39. However, our study focused on the role of IL-7Ra in mediating homeostasis of established memory CD8⁺ T cells, not the differentiation of such cells. Indeed, IL-7Rα supports the generation of CD8⁺ memory-precursors and establishment of a durable memory CD8⁺ T cell pool^35,36,39,41. By inducibly ablating Il7ra after memory CD8⁺ T cell differentiation, we could selectively interrogate the contribution of IL-7Rα to memory CD8⁺ T cell homeostasis. Our results focus on pathogen-elicited circulating and resident memory CD8⁺ T cell populations across multiple NLT and pathogens, including influenza virus-elicited lung Trm, which wane over time^49–52. We had speculated that this relatively short-lived population would be especially vulnerable to loss of IL-7Rα. However, CD69⁺ CD103⁺ lung Trm were equally (and perhaps more) resilient to IL-7Rα loss than other memory CD8⁺ T cell subsets induced by influenza virus infection. While we defined broadly conserved levels of IL-7Rα dependence across heterogenous memory subsets induced by different pathogens and in different tissues, we did not assess the recently described skin Trm17 subset, which is induced by the commensal Staphylococcus epidermidis and exhibits non-classical MHC restriction^60,61. Park, Christo, et al showed that antibody blockade of IL-7Rα for 1 week results in a 2–3 fold reduction in commensal-elicited Trm17, but not Trm1⁶¹. This suggests that higher IL-7 dependence can be observed for some memory CD8⁺ T cell subsets and for distinct forms of CD8⁺ T cell memory. The capacity of pathogen-elicited memory CD8⁺ T cells to persist after loss of IL-7 signaling may be critical for their stability during temporary alterations in IL-7 availability in vivo. This could potentially occur during inflammation or tissue damage affecting IL-7-producing stromal cells, though little is known regarding physiological perturbations in IL-7 expression^20,62–65.

Our data also indicated that CD8⁺ T cell memory populations in an IL-15-deficient environment adopted a heightened reliance on IL-7 signaling, likely resolving the enigma of IL-15-independent memory subsets. Furthermore, the IL-7 and IL-15 signaling pathways appeared to modulate each other in memory CD8⁺ T cells. This involved repression of IL-7Rα levels by IL-15 signaling, as well as reduced responsiveness to IL-15 in the absence of IL-7Rα by mechanisms still to be determined. We collectively refer to these connections between the two signaling modalities as “crosstalk” in this study. IL-7Rα expression in naïve T cells is regulated by physiological cytokine levels and reduced by stimulation with diverse cytokines (including IL-7 itself and IL-15)⁴². These findings were confirmed in vivo for memory CD8⁺ T cells, and we also showed that surface levels of IL-7Rα are enhanced on memory P14 T cells in Il15^−/− hosts. As repression of IL-7Rα has been proposed as an IL-7 conservation mechanism to allow efficient distribution of IL-7 amongst a population of cells⁴², it may be that defects in CD8⁺ T cell memory in Il15^−/− mice^{19,21–25,28,29} are not solely due to IL-15 deficiency, but also to inefficient use of IL-7 at the population level. In turn, the modest defect we describe after loss of IL-7Rα from homeostatic memory cells may not be exclusively due to loss of IL-7 sensitivity. We found that loss of IL-7Rα can impair proliferation of memory CD8⁺ T cells in response to IL-15, suggesting that intact IL-7 signaling also has a secondary effect on IL-15 responsiveness. Whether this is due to direct action of the IL-7Rα signaling cascade on IL-15 signal transduction is unknown. In a physiological context, these mechanisms could potentially conserve IL-7 while maximizing IL-7 and IL-15 signal strength across a memory population, thereby moderating the effect of local changes in cytokine availability due to consumption and competition. We propose that IL-7 and IL-15 collaborate to support memory CD8⁺ T cells in homeostasis, not only via redundancy (as both strongly activate STAT5 as well as other shared pathways, including PI3K-Akt and MAP kinases), but also by crosstalk between the two pathways. This affords a measure of resilience and adaptability in the absence of signaling for either cytokine, but results in a severe defect across homeostatic memory populations when both signals are absent. During inflammation, cytokine availability undergoes major alterations, including marked increases in IL-15 during viral infection due to type 1 interferon⁵⁹ and also altered cytokine turnover. The flexibility of memory CD8⁺ T cells to adapt to individual loss (or gain) of IL-7 and IL-15 signaling may help to ensure their stability during inflammation. In our study, we artificially imposed sudden and total loss of IL-7 signaling. Even so, memory CD8⁺ T cells were maintained with minimal loss for weeks afterwards, arguing that in a more physiological setting of reduced IL-7 availability for a few days, memory cells would likely emerge unscathed. Furthermore, potential loss of homeostatic signals during inflammation could be counterbalanced by elevated inflammatory signals. This could include not only other γC cytokines, but also members of other cytokine families or cell-surface ligands. It is intriguing to speculate that in an inflammatory setting, memory CD8⁺ T cells may rely on an expanded cadre of trophic factors, with important implications for their stability and long-term maintenance.

LIMITATIONS OF THE STUDY

The present study primarily employed memory CD8⁺ T cells which differentiated in Il15^−/− recipient mice and therefore lacked IL-15 throughout both differentiation and memory homeostasis, which may overrepresent the requirement for IL-15 during memory homeostasis. Future work should address the specific role for IL-15 signaling in memory CD8⁺ T cell homeostasis. The mechanism by which IL-7Rα licenses full IL-15 responsiveness is not clear. Preliminary data suggest that strong IL-15 signals (i.e. IL-15 complexes) can overcome the requirement for homeostatic IL-7Rα, suggesting that IL-7Rα acts to enhance the response to IL-15 at more physiological levels.

Our study used multiple models of pathogen-elicited CD8⁺ T cell memory, primarily employing TCR transgenic T cells. However, a skin Trm17 subset⁶¹ forms during the endogenous response to the commensal S. epidermidis, exhibiting restriction to the non-classical MHC H2-M3. Thus, our study does not capture the entire diversity of T cell memory, and it should be considered that requirements for IL-7 and 15 signaling may change for other types of memory in different models.

RESOURCE AVAILABILITY

Lead contact

Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, Stephen Jameson (james024@umn.edu).

Materials availability

This study did not generate unique reagents.

Data and code availability

All data reported in this paper will be shared by the lead contact upon request.

This paper does not report original code.

Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.

STAR METHODS

EXPERIMENTAL MODEL AND STUDY PARTICIPANT DETAILS

Mice

See also the Key Resources Table. C57BL/6J and JAXboy mice were purchased from Jackson Laboratory for use as Il7ra^fl/fl P14 transfer recipients. Il15^−/− mice¹⁹ were backcrossed to C57BL/6J and bred in house as heterozygous by heterozygous and heterozygous by homozygous for use as Il7ra^fl/fl P14 transfer recipients; in some experiments C57BL/6J were also used as Il15^+/+ recipients. For all lines backcrossed, speed congenics from Transnetyx were used. CD45.1/.1 and CD45.2/.2 C57BL/6N (purchased from Charles River) were used as recipients for mixed bone marrow chimera experiments and cytokine complex therapy experiments. Il7ra^fl/fl mice⁶⁶ with a floxed exon 3 of the Il7ra gene were purchased from Jackson Laboratory on a mixed background (used for BM chimera experiments) and were backcrossed 5–6 generations to C57BL/6J (used for adoptive transfer experiments). Ubc-cre^ERT2 mice were purchased from Jackson Laboratory. LCMV-gp33/D^b-specific P14 TCR transgenic mice⁶⁷ were backcrossed and maintained on the C57BL/6J and C57BL/6N backgrounds as separate lines. C57BL/6J P14 mice were intercrossed with Il7ra^fl/fl, Ubc-cre^ERT2, and Jaxboy mice to generate congenically distinct donor mice. Ubc-cre^ERT2-positive male breeders were crossed to female Il7ra^fl/fl mice lacking the Cre transgene and offspring were typed for the floxed allele and the recombined version to identify spontaneous germline deletion in the absence of tamoxifen⁶⁸, which was rare for this strain. Donors (control and experimental derived from the same breeding colony, siblings and littermates used when possible) and recipients were used between 6–14 weeks of age. Female animals were used as recipients (and donors) in our study, due to limitations on housing males for long-term memory experiments due to aggression. Animals were housed under specific pathogen-free conditions at the University of Minnesota; infections were performed in a BSL2 animal facility. All animal procedures were approved by the institutional animal care and use committee of the University of Minnesota.

Key resources table.

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
Anti-mIL-7 antibody (M25)	BioXCell	BE0048, RRID:AB_1107711
FITC anti-CD45.1 (A20)	BioLegend	110706, RRID:AB_313494
PE anti-CD45.1 (A20)	BioLegend	110708, RRID:AB_313497
violetFluor450 anti-CD45.1 (A20)	Tonbo	75-0453-U100, RRID:AB_2621949
BUV737 anti-CD45.1 (A20)	BD	565212, RRID:AB_2722493
FITC anti-CD45.2 (104)	Tonbo	35-0454-U500, RRID:AB_2621692
PE anti-CD45.2 (104)	Tonbo	50-0454-U100, RRID:AB_2621766
BUV737 anti-CD45.2 (104)	BD	612778, RRID:AB_2870107
PerCP/Cy5.5 anti-CD8α (53-6.7) i.v. labelling	Tonbo	65-0081-U100, RRID:AB_2621882
PE/Cy7 anti-CD8α (53-6.7)	Tonbo	60-0081-U100, RRID:AB_2621832
BV786 anti-CD8α (53-6.7)	BD	563332, RRID:AB_2721167
BUV395 anti-CD8α (53-6.7)	BD	563786, RRID:AB_2732919
PE anti-CD122 (TM-β1)	BD, BioLegend	553362 (RRID:AB_394809), 123210 (RRID:AB_940615)
BV510 anti-CD44 (IM7)	BD	563114, RRID:AB_2738011
redFluor710 anti-CD44 (IM7)	Tonbo	80-0441-U100, RRID:AB_2621985
BV510 anti-CD103 (M290)	BD	563087, RRID:AB_2721775
BV421 anti-CD69 (H1.2F3)	BioLegend	104528, RRID:AB_10900250
BV421 anti-CD127 (A7R34)	BioLegend	135024, RRID:AB_10897948
BUV737 anti-CD127 (A7R34)	BD	612841, RRID:AB_2870163
BV605 anti-TCRβ (H57-597)	BD	562840, RRID:AB_2687544
BV786 anti-CD62L (MEL-14)	BD	564109, RRID:AB_2738598
BV711 anti-KLRG1 (2F1)	BD	564014, RRID:AB_2738542
PerCP/Cy5.5 I-a/I-e (M5/114.15.2)	Tonbo	65-5321-U100, RRID:AB_3665172
PerCP/Cy5.5 F4/80 (BM8.1)	Tonbo	65-4801-U025, RRID:AB_2621908
PerCP/Cy5.5 CD19 (1D3)	Tonbo	65-0193-U100, RRID:AB_2621887
FITC anti-Bcl2 (10C4)	Thermo Fisher Scientific	11-6992-42, RRID:AB_10734060
PE anti-Granzyme B (GB11)	Thermo Fisher Scientific	GRB04 RRID:AB_2536538
AF647 anti-Granzyme B (GB11)	BioLegend	515405, RRID:AB_2294995
APC anti-BrdU (Bu20a)	Thermo Fisher Scientific	17-5071-42, RRID:AB_11040534
APC anti-Ki67 (SolA15)	Thermo Fisher Scientific	17-5698-82, RRID:AB_2688057
Mouse Fc block	BD	553142, RRID:AB_394657
Bacterial and virus strains
Lymphocytic choriomeningitis virus-Armstrong	Rafi Ahmed, Emory	NA
Listeria monocytogenes-gp33	Rafi Ahmed, Emory	Generated by Dr. Hao Shen (UPenn) first reported PMID: 11323695
Influenza virus, PR8-gp33	Ryan Langlois, Minnesota	Generated by Dr. Ryan Langlois, first reported PMID: 31235953
Chemicals, peptides, and recombinant proteins
Collagenase I	Worthington Biochemical	LS004197
Collagenase IV	Thermo Fisher Scientific	17104019
Dithioerythritol	EMD Millipore	233152
Percoll	GE Healthcare	17-0891-09
Mouse Interleukin 7	Shenandoah Biotechnologies	200-03AF
Mouse Interleukin 15	BioLegend	566304
Mouse Interleukin 15	Tonbo	21-8153-U500
IL-15Rα-Fc chimera	R&D Systems	551-MR
5-Bromo-2’-deoxyuridine	Sigma Aldrich	B5002
gp33/Db monomer (H-2Db KAVYNFATM)	NIH Tetramer Core	NA
gp276/Db monomer (H-2Db SGVENPGGYCL)	NIH Tetramer Core	NA
NP296/Db monomer (H-2Db FQPQNGQFI)	NIH Tetramer Core	NA
Polymyxin B sulfate salt	Sigma Aldrich	P4932
Neomycin trisulfate salt hydrate	Sigma Aldrich	N6386
Tamoxifen	Sigma Aldrich	T5648
Corn oil	Sigma Aldrich	C8267
Sucrose	Sigma Aldrich	S1888
RPMI 1640	Corning	10-040-CV
10x HBSS	Corning	20-020-CF
FBS	Atlas Biologicals	FS-0050-AD
DNase I	Sigma Aldrich	D5025
Normal Rat Serum	Stem Cell Technologies	13551
Critical commercial assays
CD8a+ T cell isolation kit	Miltenyi Biotec	130-104-075
Cytofix/Cytoperm Fixation/Permeabilization Solution Kit	BD	554714
FoxP3/Transcription Factor Staining Buffer Kit	Tonbo	TNB-0607-KIT
Experimental models: Organisms/strains
Mouse: C57BL/6J	Jackson Laboratories	000664 RRID:IMSR_JAX:000664
Mouse: C57BL/6J-Ptprcem6Lutzy/J	Jackson Laboratories	033076 RRID:IMSR_JAX:033076
Mouse: NCI C57BL/6NCr	Charles River	556
Mouse: NCI B6-Ly5.1/Cr	Charles River	564, RRID:IMSR_CRL:564
Mouse: Il7rtm1.1Asin/J	McCaughtry et al/Jackson Laboratories	022143 RRID:IMSR_JAX:022143
Mouse: Il15tm1Imx	Kennedy et al/David Masopust, Minnesota	RRID:MGI:3590155
Mouse: P14 TCR transgenic	Pircher et al/Rafi Ahmed, Emory	NA
Mouse: B6.Cg-Ndor1Tg(UBC-cre/ERT2)1Ejb/1J	Jackson Laboratories	007001 RRID:IMSR_JAX:007001
Software and algorithms
Prism v9.5.0	Graphpad	RRID:SCR_002798
Illustrator 2023 27.2	Adobe	RRID:SCR_010279
Flowjo v10.8.2	Treestar/BD	RRID:SCR_008520
BD FACSDiva	BD	RRID:SCR_001456
Other
Ghost Dye Red e780	Tonbo	13-0865-T500
R-PE-streptavidin	Thermo Fisher Scientific	S21388
PE/Cy7-streptavidin	Thermo Fisher Scientific	25-4317-82
LS columns	Miltenyi Biotec	130-042-401
QuadroMACS Separator	Miltenyi Biotec	130-091-051
GentleMACS Dissociator	Miltenyi Biotec	130-093-235, RRID:SCR_020267
GentleMACS C Tubes	Miltenyi Biotec	130-093-237, RRID:SCR_020270
RS 2000 irradiator	Rad Source
Variomag Poly 15	Thermo Fisher Scientific	50094595
CountBright Plus counting beads	Invitrogen	C36995

Pathogens and infections

Infections and treatments: Mice were infected intraperitoneally (i.p.) with LCMV Armstrong at 2*10⁵ PFU. Mice were infected intravenously with ~10,000 colony forming units (CFU) of Listeria monocytogenes expressing gp33 (Lm-gp33) or rechallenged intravenously with ~500,000 CFU Lm-gp33⁴⁷. Mice under anesthesia were infected intranasally with 500 PFU influenza virus expressing gp33 (PR8-gp33)^48,69.

METHOD DETAILS

Tamoxifen treatment

Tamoxifen was dissolved in corn oil at 20 mg/mL for >5 hours at 37C with 250 rpm shaking (New Brunswick Scientific, Excella E24 Incubator Shaker Series) and vortexing, with protection from light. Tamoxifen in oil was prepared fresh for each course of treatment. Mice were given 100 uL (2 mg) i.p. on 5 consecutive days, per the Jackson Laboratory (https://www.jax.org/research-and-faculty/resources/cre-repository/tamoxifen). For P14 cohorts, high efficiency or moderate efficiency deletion was confirmed to be a reproducible feature of each cohort across different batches of tamoxifen and for both naïve and memory cells derived from the same donor. Variability in Cre activity, even between littermates, has been previously reported⁶⁸.

BrdU treatment

1 mg BrdU was given i.p. at the start of labelling, with continuous BrdU administration via drinking water ad libitum thereafter for 8 days (0.8 mg/mL in 2% sucrose water with protection from light). BrdU water was changed every other day.

Cytokine treatment

For low dose IL-15 treatment, 2 μg mIL-15 were given i.p. as published⁵⁹ on days 0 and 2 before harvest on day 3. For IL-7 and IL-15 complex treatment, mice were treated on days 0 and 2 before harvest on day 4. IL-7c were formed with 1.5 μg mIL-7 and 7.5 μg anti-IL-7 antibody per mouse, combined in equal volume and incubated for 2–3 minutes at RT, followed by dilution in PBS to 200 uL, and were placed on ice before injection i.p.⁵⁸. IL-15c were prepared as previously described^27,70.

CD8⁺ T cell adoptive transfers

CD8⁺ T cells were isolated from the spleens of donor P14 TCR transgenic mice with the Miltenyi Biotec CD8a⁺ T cell isolation kit, mouse and Miltenyi Biotec LS columns. 50,000 cells from each donor were transferred to naive recipient mice by retro-orbital or tail vein injection before pathogen infection to elicit memory. To track naïve P14 T cells, ~1.5–2M P14 T cells from each donor were co-transferred, followed by tamoxifen treatment as above within a week (generally ~4 days post-transfer). To assess adaptability of CD127 expression on memory P14 T cells, LCMV-elicited P14 memory mice were generated as above. At >4 weeks after LCMV infection, mice were sacrificed, splenic CD8⁺ T cells were isolated, and 1.5–2M total CD8⁺ T cells were transferred i.v. to each recipient (averaging ~18% memory P14 T cells from WT hosts and ~13% memory P14 T cells from Il15^−/− hosts). Animals were sacrificed approximately two weeks later. For recall of sorted control and Ubc-cre^ERT2 Il7ra^fl/fl memory P14 T cells, splenic CD8⁺ T cells were isolated from previously tamoxifen-treated memory mice (>3 weeks post-tamoxifen) followed by sorting on a BD FACSAria II. P14 T cells were sorted as singlet lymphocytes gated as live lineage-negative (CD19, F4/80, I-a/I-e), CD8α⁺ and on congenic markers to avoid antibody-mediated crosslinking of the TCR or CD3. LLEC and Tcm were gated on KLRG1 and CD62L. Ubc-cre^ERT2 Il7ra^fl/fl P14 T cells were additionally gated as CD127-negative. 50,000 Tcm (Tcm+Tcm) or 30,000 LLEC (LLEC+LLEC) from each donor were cotransferred to naïve recipients and the transfer ratio was confirmed via flow cytometry of the mix. Recipients were challenged with LCMV on the following day and were tracked by bleeds over time.

Mixed BM chimeras

Recipient mice were given split-dose irradiation (2 doses of 500 rads, RS 2000 irradiator [Rad Source]) on two consecutive days, followed by intravenous transfer of 4M BM cells from each donor (isolated from hind limb femurs and tibias without red blood cell lysis). Mice were given two weeks of antibiotic-treated water (polymyxin B sulfate salt 15 mg/L, neomycin trisulfate salt hydrate 40 mg/L) during reconstitution. After >8 weeks, animals were infected with LCMV-Armstrong as above and rested for >4 weeks to establish memory. Animals were then treated with tamoxifen as above.

Tissue harvests

5 minutes before harvest, mice were given 3 μg anti-CD8α PerCP/Cy5.5 by retro-orbital injection⁴⁵. Animals were then cheek bled into heparin and sacrificed for tissue harvest. The inguinal LNs (LCMV, Lm-gp33), mediastinal LNs (PR8-gp33), and the spleen were collected into harvest media (either RPMI 1640 supplemented with 5% fetal bovine serum [heat inactivated before use] or 1x Hanks’ balanced salt solution [HBSS] supplemented with 2.38 g/L Hepes], 2.1 g/L sodium bicarbonate, and 5% FBS) and passed through a 70-μm cell strainer. For IEL, the small intestine (SI) was excised and divested of fat and doused in IEL media (1x HBSS supplemented with Hepes, sodium bicarbonate, and 2% FBS) to keep moist. The Peyer’s patches were then removed and fecal contents were extruded from the lumen, which was then cut open. The sample was cut into 3–4 longitudinal sections, vortexed, and left on ice in 20 mL IEL medium. IEL medium was decanted, and tissue was washed with 30 mL of fresh IEL medium. Tissue was then transferred to 50 mL Erlenmeyer flasks with stir bars and 30 mL of IEL dithioerythritol (DTE) media supplemented with additional FBS to 5% and 154 mg/L dithioerythritol. After 30 min of stirring at 37 °C on a Variomag Poly 15, the supernatant was decanted through a 70 μm filter. 20 mL of IEL DTE media was added, followed by vortexing. After allowing tissue to settle, the supernatant was filtered and combined with the previous supernatant fraction. For liver, the organ was excised, avoiding the gallbladder, and placed in 5 mL harvest media on ice. It was then disrupted using a GentleMACS C tube on a GentleMACS Dissociator (m_spleen_01.01 twice) and filtered. For SG, kidney, FRT, lung, skin. SGs (submandibular, part of the sublingual) were excised, and cervical LNs were removed if present. The parotid was excluded to avoid lymphoid contamination. The kidney capsule was removed during isolation. The FRT was collected inclusive of the ovaries to the vagina and was bisected open. The two lungs were excised. All tissues were placed in harvest media on ice. SG, kidney, FRT, and lung were minced with scissors and transferred to Erlenmeyer flasks with stir bars. 30 mL of collagenase solution (RPMI 1640 supplemented with 1 mM MgCl2, 1 mM CaCl2, 111.6 mg/L Hepes, 292 mg/L L-glutamine, and 5% FBS) containing 0.364 mg/mL collagenase I (SG, kidney, lung) or 0.5 mg/mL collagenase IV (FRT) was added. These tissues were then incubated at 37 °C with stirring for 45–55 min (SG, kidney) or 60–70 min (lung, FRT). After digestion, supernatants were filtered, and the remaining tissue was transferred to a GentleMACS C tube as above for liver. GentleMACS contents were then filtered and combined with the previous fraction. SG, kidney, and FRT were pelleted and resuspended in 5 mL of room temperature (RT) 44% Percoll (diluted with RPMI 1640), before underlay with 3 mL of RT 67% Percoll (diluted with PBS). Percoll was mixed with 10x PBS before use. Samples were centrifuged for 20 min at 800 g at RT, with minimum acceleration and deceleration. The interface was collected, diluted with harvest media, and used for downstream analysis. Red blood cells in blood, spleen, and lung samples were lysed with ACK buffer (150 mM ammonium chloride, 1 mM potassium bicarbonate, and 0.1 mM EDTA [ethylenediaminetetraacetic acid] in water) before staining. Analysis of lung memory was performed between days 64–82 post-PR8-gp33 to avoid loss of lung Trm as previously reported^49–52.

Flow Cytometry

Samples and splenic single stain controls were washed in fluorescence-activated cell sorter (FACS) buffer (2% FBS, 2 mM EDTA in 1x PBS), followed by Fc blocking for 5 min. Antibodies and viability dye for staining were then added for 20 min at 4 °C with concurrent tetramer staining when used. Samples were then washed before fixation, washed after fixation, and stored in FACS buffer at 4 °C before intracellular staining 1 to 2 d later. Antibodies and viability dye are listed in the Key Resources Table. All surface antibodies were used at 1/200, except anti-CD69 and anti-CD127 (1/100). Viability dye was used at 1/1,000. Intracellular Staining. For Ki67 (1/200), Bcl2 (1/50), and granzyme B staining (1/200), the Tonbo Foxp3/Transcription Factor Staining Buffer Kit was used according to the standard protocol, with an additional 15-min incubation in 1x Perm buffer plus 2% Normal Rat Serum before antibody staining for 45–60 min at RT. BrdU staining was done using a modified protocol for the BD Cytofix/Cytoperm Fixation/Permeabilization Solution Kit. Briefly, cells were washed in 1x Perm/Wash (P/W) buffer and then incubated for 10 min at 4 °C in 1x P/W buffer plus dimethyl sulfoxide (1 part 10x P/W, 1 part DMSO, and 8 parts water). Samples were then washed in 1x P/W buffer and refixed in Cytofix/Cytoperm buffer for 5 min at RT, followed by washing in 1x P/W buffer. Cells were then digested with deoxyribonuclease (DNase) I at 37 °C for at least 50 min. DNase was stored at −80 °C as a 2 mg/mL stock in PBS and diluted to 600 μg/mL in 1x PBS immediately before use. After digestion, cells were washed in 1x P/W buffer and stained with anti-BrdU antibody (1/100) for 50–60 min at RT. Cells were washed with 1x P/W buffer and then with FACS buffer. All NLT samples were filtered, and CountBright Plus counting beads were added to flow tubes before analysis. Samples were acquired using BD LSRII, LSRFortessa X20 and X30, and LSRFortessa flow cytometers and FACSDiva software. Analysis was done using FlowJo v7 and v10. Singlet lymphocytes were gated by forward scatter area (FSC-A)/side scatter area (SSC-A) and FSC-A/FSC-width (FSC-W). Live cells were then gated according to i.v. labeling status (blood, i.v.-positive; LN, IEL, SG, and FRT i.v.-negative; spleen, kidney, and lung, i.v.-low; liver, not gated by i.v. labeling status). CD8α⁺/TCRβ⁺ cells were then divided by CD45 congenic staining into host and donor populations, with gp33/D^b tetramer staining to exclude the host for Il15^−/− and naïve CD8⁺ T cell experiments. Tetramer binding cells were gated as tet⁺ CD44^hi. To validate BrdU gating, each experiment included a control animal not given BrdU, tissues from which were processed and stained alongside the other samples to serve as negative controls. For subsetting of memory, blood, spleen, and LN P14 T cells were gated as KLRG1⁺ CD62L⁻ (long-lived effector cells), KRG1⁻ CD62L⁻ (T effector memory), and KLRG1⁻ CD62L⁺ (T central memory). For the adjusted ratio of control to CD127-negative Ubc-cre^ERT2 Il7ra^fl/fl P14 T cells, adjusted ratios were only calculated for recipients that had received tamoxifen and could therefore have lost the IL-7Rα. For animals not receiving tamoxifen, the simple ratio of donor to donor was used, meaning the ratios shown for no tamoxifen recipients are identical for the simple and adjusted ratio graphs. For salivary gland, P14 T cells were split into two CD69⁺ resident subsets, differing in expression of CD103. In rare cases, dominance of CD69⁻ CD103⁻ cells in SG samples revealed lymphoid tissue contamination, and these samples were excluded. For kidney and liver, P14 T cells were split into CD69⁺ resident and CD69⁻ circulating memory. For lung, P14 T cells were split by i.v. labelling, with high labelling representing vascular cells and low labelling representing cells within the parenchyma. i.v. label low cells were furthermore split into CD69⁻ CD103⁻, CD69⁺ CD103⁻, and CD69⁺ CD103⁺ subsets.

Tetramer reagents

Tetramers were made using biotinylated monomers (H-2D^b KAVYNFATM [gp33/D^b]; H-2D^b SGVENPGGYCL [gp276/D^b]; H-2D^b FQPQNGQFI [NP396/D^b]) from the NIH Tetramer Core at Emory University. For gp33/D^b monomer, PE/Cy7-streptavidin was used. For gp276/D^b and NP396/D^b, R-PE-streptavidin was used. Fluorophore-conjugated streptavidin was added to 20 μg of monomer, in 10 additions of 3.18 μg, each 10 min after the other (at room temperature). Tetramer was then stored at 4 °C before use.

Software

Flow cytometry data was acquired in FACSDiva and analyzed in FlowJo v7 and v10. Statistical calculations and graphing were performed using GraphPad Prism v9. Figures were generated in Adobe Illustrator 2023.

QUANTIFICATION AND STATISTICAL ANALYSIS

All statistical analyses were performed in Prism, as specified in the figure legends. Unpaired, two-tailed Student’s t tests were applied to determine the difference between two independent groups. Paired, two-tailed Student’s t tests were applied to determine the difference between two related groups (i.e. co-transferred populations within the same recipient animal). Where no p value is provided for a relevant comparison, the result was not significant. Infrequently, a ratio could not be calculated due to no cells in the Il7ra-deficient group remaining (i.e. 100 Il7ra-sufficient cells divided by 0 Il7ra-deficient treated cells). In this case, the ratio was set conservatively at the minimum possible value (i.e. 100:0 becomes 100:1, or a ratio of 100). Infrequently, too few events (less than five events) were captured for accurate quantitation of downstream parameters (i.e. Ki67, BrdU, CD127 gMFI). In such cases, these values were excluded for these parameters. In rare cases of sickness, animals were excluded. Error bars represent the SEM.

Supplementary Material

Supplemental information can be found online at

Figures S1–S7

Highlights.

Tissue-resident and circulating memory CD8⁺ T cells slowly decline after IL-7Rα loss

IL-7Rα is required for normal self-renewal of memory CD8⁺ T cells

Loss of IL-7 and IL-15 causes a profound defect across memory CD8⁺ T cell subsets

Crosstalk between the IL-7 and IL-15 signaling pathways occurs in memory CD8⁺ T cells