Inflammatory Signals Regulate IL-15 in Response to Lymphodepletion

This article is featured in In This Issue, p.4421

Interleukin-15 responses by T cells are elevated after the depletion of lymphocytes by total body irradiation (TBI) or high-dose chemotherapy regimens and in lymphocyte-deficient RAG^−/− mice (1–3). These enhanced responses to IL-15 during lymphopenia have been exploited therapeutically to improve the immune responses to cancer therapies and vaccinations (2, 4–6). For instance, lymphodepletion is used as a conditioning regimen for adoptive T cell therapies of melanoma in mouse models and in human patients (2, 3). Lymphopenia induces the proliferation of adoptively transferred T cells as well as promotes a loss of tolerance against self-Ags, leading to enhanced antitumor responses (7); all of these responses are dependent on IL-15. Despite the clear importance of IL-15 in enhancing T cell responses during lymphopenia, the mechanisms regulating IL-15 during lymphopenia are not clear, particularly with respect to the realization that IL-15 can mediate responses through multiple mechanisms, including transpresentation and soluble cytokine–receptor complexes. Furthermore, the cellular source for either form of IL-15 during lymphopenia has not been elucidated.

The mechanism of lymphopenia-enhanced IL-15 expression was originally considered a passive process, whereby the availability of a constant low level of IL-15 is enhanced simply due to the loss or lack of endogenous lymphocytes utilizing this IL-15; this theory is often referred to as the cytokine sink (3). However, a later report demonstrated that the antitumor effect induced by TBI was instead mediated by the presence of microbial components that leak through the intestines (8). More recent studies found that soluble IL-15/IL-15Rα complexes (sIL-15 complexes) are transiently increased in mice treated with TBI or cyclophosphamide (CTX) and in the sera of patients undergoing lymphodepletion regimens (9). These transiently induced sIL-15 complexes are one potential factor enhancing T cell responses after lymphodepletion as recombinant sIL-15 complexes are robust mediators of T cell and NK cell proliferation (10, 11). Interestingly, the transient induction of sIL-15 complexes in mice after TBI precedes the presence of microbial components and LPS (8), suggesting the early induction of sIL-15 complexes is mediated by an additional mechanism. Our recent studies have found that type I IFNs are an important pathway stimulating the production of sIL-15 complexes after TBI (12), providing further support for the idea that active inflammatory signals upregulate sIL-15 complexes. Therefore, the objective of this study was to further elucidate the mechanisms regulating IL-15 during lymphopenia. Using multiple mouse models of lymphopenia, we investigated the factors and cell types required for the generation of lymphopenia-induced IL-15 and sIL-15 complexes.

C57BL/6 (wild-type [WT]) mice were purchased from National Cancer Institute/Charles River (Frederick, MD). RAG1^−/− mice, IL-15Rα^fl/fl (13), CD11cCre (14), LysM-Cre (15), and Tmem173^−/− mice (16) were purchased from The Jackson Laboratory (Bar Harbor, ME). IL-15Rα^−/− knockout (Rko) mice (17) were originally generated and obtained by A. Ma through L. Lefrancois (Department of Immunology, University of Connecticut, Farmington, CT) and backcrossed to the C57BL/6 line. Thy1.1⁺ Pmel-1 TCR transgenic (specific for gp100 peptide 25–33 in the context of H2-D^b) mice were provided by W. Overwijk (Melanoma Medical Oncology, University of Texas MD Anderson Cancer Center) (18). IFNAR1^−/− mice were provided by P. W. Dempsey (Department Of Microbiology and Molecular Genetics, University of California, Los Angeles, CA) and T. Taniguchi (Department of Immunology, Tokyo University, Tokyo, Japan) to W. Overwijk and crossed to the C57BL/6 background (19). IL-15 transcriptional reporter mice were generated by L. Lefrancois (20); experiments utilizing these mice were performed at the University of Connecticut Health Science Center. All other mice described were maintained under specific pathogen-free conditions at the institutional animal facility. The animal facility is fully accredited by the Association of Assessment and Accreditation of Laboratory Animal Care International. All animal procedures were conducted on mice between 6 and 10 wk of age, in accordance with the animal care and use protocols (100409934) approved by the Institutional Animal Care and Use Committee at the University of Texas MD Anderson Cancer Center.

To generate bone marrow (BM) chimeras, BM was collected from the tibia and femurs of IL-15Rα^−/− (CD45.2) and WT (CD45.1) mice and depleted of T cells, as previously described (21). IL-15Rα^−/− (CD45.2) and WT (CD45.2) recipients were irradiated with 1000 rads and injected i.v. with 3 × 10⁶ BM cells. After complete BM reconstitution (8–12 wk later), serum was collected from peripheral blood and analyzed for levels of sIL-15 complexes.

For poly I:C stimulation, mice were administered poly I:C i.p. (150 μg; Sigma-Aldrich, St. Louis, MO). For stimulation of the stimulator of IFN genes (STING) pathway, mice were administered c-di-GMP i.v. (Invivogen, San Diego, CA) at the indicated doses. For chemotherapy treatments, mice were administered cyclophosphamide dissolved in saline i.p. (200 mg/kg). For TBI, mice were exposed to a cesium irradiation source at the indicated doses. Complement-mediated depletion of T cells was performed by injecting i.p. anti-Thy1.2/CD90.2 (300 μg, clone 30H12), anti-CD8α (clone 2.43), or anti-CD4 (GK1.5) (all from BioXcell, Lebanon, NH). Depletion was confirmed with flow cytometric analysis of the spleen. For serum isolations, peripheral blood was collected from sacrificed mice via cardiac puncture and on some occasions from the retro-orbital cavity prior to treatment from the same mice. Blood was allowed to clot and centrifuged to separate serum.

ELISA specific for murine sIL-15Rα/IL-15 complexes (eBioscience, San Diego, CA) was performed according to manufacturer’s recommendations. The limit of detection of this ELISA in our hands was 1 pg/ml. Limulus amebocyte lysate chromogenic endotoxin quantitation assay (ThermoScientific, Rockford, IL) was performed according to manufacturer’s recommendations. Cell surface IL-15 was detected in splenic myeloid cells isolated directly ex vivo, as previously described (22). Briefly, cell surface IL-15 was detected with polyclonal rabbit anti–IL-15 biotin (PeproTech, Rocky Hill, NJ), followed by streptavidin-allophycocyanin (Jackson ImmunoResearch Laboratories). Background staining was determined by staining analogous populations with a biotinylated Ig control (Jackson ImmunoResearch Laboratories). The following mAbs were purchased from BD Biosciences (San Jose, CA), eBiosciences, or BioLegend: CD19, CD3, DX5, CD11b, CD11c, B220, CD19, CD90.1, CD90.2, CD8, NK1.1, and CD44. Lineage⁺ cells were identified as CD19⁺, CD3⁺, or NK1.1⁺. For analysis of the IL-15 transcriptional reporter mice, the following cell populations were gated based on forward light scatter, side light scatter, and exclusion of a Live/Dead dye and phenotypically defined as follows: CD11b⁺ dendritic cells (DCs) (lin⁻CD45⁺F4/80⁻Ly6C⁻CD11c⁺MHC class II⁺CD11b⁺), monocytes (lin⁻CD45⁺CD11b⁺Ly6G⁻Ly6C⁺), CD8⁺DCs (lin⁻CD45⁺F4/80⁻Ly6C⁻CD11c⁺MHC class II⁺CD8⁺), neutrophils (lin⁻CD45⁺CD11b⁺Ly6G⁺), and CD103⁺ DCs (lin⁻CD45⁺F4/80⁻Ly6C⁻CD11c⁺MHC class II^highCD103⁺). At >30 d postinfection, the expression of CD8α, CD90.1, CD62L, and high CD44 expression was used to define memory Pmel-1 T cells. Flow cytometric data were acquired with a LSRII (BD Biosciences) or LSR Fortessa (BD Biosciences) and analyzed with Flowjo software version 9.7.6 (Flowjo). RNA was isolated from sorted cell populations using TRIzol (Invitrogen, Carlsbad, CA), according to the manufacturer’s protocol. cDNA was synthesized using the Superscript II reverse transcriptase (Invitrogen). iQ SYBR supermix (Bio-Rad, Hercules, CA) was used to conduct quantitative PCR assays in triplicate. Expression of IL-15 was determined by the following formula, applied to each reaction: 1.8 (C_T β-actin − C_T IL-15) × 10⁶.

BM-derived DCs (BMDCs) were generated using Flt3L, as previously described (23). Briefly, BM cells were flushed from femurs of indicated mice, dissociated, and treated with Tris ammonium chloride to lyse RBCs. BM cells were then cultured in RPMI 1640 complete medium at a concentration of 1 × 10⁶ cells/ml. Complete medium RPMI 1640 contains 2.5 mM HEPES, 5.5 × 10⁻⁵ M 2-ME, 100 U/ml penicillin, 100 μg/ml streptomycin, 5 mM glutamine, and 10% FCS supplemented with 200 ng/ml FLT-3L (R&D Systems, Minneapolis, MN) at 37°C with 5% CO₂ for 9 d. BMDCs were seeded at 1–2.5 × 10⁶ cells/ml and stimulated with IFN-α (300 U/ml; PBL Laboratories, Piscataway, NJ), or the STING agonist c-di-GMP (25 μg/ml; Invivogen, San Diego, CA). Supernatants were collected after 24 h and analyzed for sIL-15 complexes using ELISA.

A total of 4 × 10⁵ naive (CD44⁻) Thy1.1⁺ (CD90.1) Pmel-1 TCR transgenic CD8 T cells was adoptively transferred to Thy1.2⁺ WT recipients and 1 d later infected with recombinant vesicular stomatitis virus expressing gp100 (18, 24). Greater than 35 d postinfection, total CD8 T cells (containing Thy1.1⁺ memory Pmel-1 T cells) were negatively enriched using the Dynal CD8 enrichment kit (Invitrogen, Carlsbad, CA), labeled with 2 mM CFSE, as previously described (22, 25), and injected i.v. into Thy1.2⁺ WT or IL-15Rα^−/− mice (between 0.1 and 0.5 × 10⁶ Pmel-1 CD8 T cells/mouse). Isolated splenic memory Pmel-1 T cells were CD44hi, with ∼40–60% expressing CD62L, suggesting this population contained a mixture of effector and central memory T cells. Five days after αThy1.2 injection, lymphocytes in spleen and lymph nodes were analyzed for the presence of Pmel-1 T cells and CFSE dilution.

Statistical differences were determined by a two-tailed Student t test: *p < 0.05. Analyses were performed using GraphPad Prism, version 6 (GraphPad Software, San Diego, CA), or Microsoft Excel 2010 (Redmond, WA).

IL-15 responses by T cells and NK cells are increased during lymphopenic conditions (3, 9). Because IL-15 responses can be mediated by both transpresentation and sIL-15 complexes, we examined the expression of sIL-15 complexes and cell surface IL-15 during lymphodepletion. Similar to previous reports, TBI increased the levels of sIL-15 complexes in peripheral blood serum 24 h posttreatment, as previously described (9, 12) (Fig. 1A). In contrast to previous reports (9), significant increases in serum sIL-15 complexes were not observed 72 h after CTX treatment (Fig. 1A) in multiple experiments as well as during earlier time points (Fig. 1B). This lack of upregulation of sIL-15 complexes after CTX treatment in our hands is not clear but was not due to insufficient drug delivery, as abundant lymphopenia was observed, similar to the level of depletion observed in previous studies (Supplemental Fig. 1) (9). Despite RAG^−/− mice being endogenously lymphopenic, the levels of sIL-15 complexes were equivalent to those of untreated WT mice (Fig. 1A). To determine whether baseline levels of sIL-15 complexes in untreated WT and RAG^−/− mice were above the level of detection of the ELISA, serum from IL-15Rα^−/− mice was analyzed and found to contain no detectable levels of sIL-15 complexes (Fig. 1A). These results reveal sIL-15 complexes are circulating at low levels during the steady state and are not upregulated in all forms of lymphopenia.

Because previous studies provided evidence that antitumor therapies involving TBI can lead to damage of the intestinal lining resulting in presence of commensal bacteria in the mesenteric lymph nodes and enhanced levels of circulating LPS (8), we wanted to examine the levels of LPS present at early times post-TBI when sIL-15 complexes are generated. Baseline LPS levels in untreated mice were equivalent to those observed in the aforementioned study (8). Whereas we observed sIL-15 complexes as early as 16 h post-TBI (data not shown), levels of serum LPS levels were not increased at this time, or at 24, 72, or 168 h post-TBI (Fig. 1C). Therefore, these findings suggest the early induction of sIL-15 complexes is not mediated by the presence of commensal bacteria in the circulation.

We also examined how transpresentation of IL-15 is affected by varying types of lymphopenia. In general, cell surface IL-15 expression by myeloid cells was enhanced, relative to untreated WT mice, under all lymphopenic conditions examined; however, differential cell-type expression was noted (Fig. 1D, 1E). DCs (lin⁻CD11b⁺CD11c⁺) from RAG^−/− mice and CTX-treated mice, but not mice subjected to TBI, displayed a significant increase in cell surface IL-15 expression (Fig. 1D, 1E). In contrast, IL-15 expression by monocytes/macrophages (lin⁻CD11b⁺CD11c⁻) was increased after TBI and in RAG^−/− mice, but not CTX treatment (Fig. 1D, 1E). To determine whether increased levels of cell surface IL-15 by DCs of RAG^−/− mice coincide with increased IL-15 transcription, splenic DCs were sorted and IL-15 transcription was measured using quantitative PCR. Levels of IL-15 transcripts were not significantly different in DCs isolated from WT and RAG^−/− mice (Fig. 1F), indicating increased surface IL-15 in RAG^−/− mice occurs independent of transcription. The differential induction of sIL-15 complexes and increases in cell surface IL-15 suggest that distinct mechanisms regulate IL-15 after TBI versus chemotherapy. Furthermore, these results suggest increased surface IL-15 is common to lymphopenia, whereas sIL-15 complexes are induced by inflammatory signals.

Because we found sIL-15 complexes are produced during the steady state, we examined the cellular source of sIL-15 complexes during homeostasis in various combinations of IL-15Rα BM chimeras. After complete reconstitution of the hematopoietic compartment (∼8–12 wk after BM transfer), WT control BM chimeras (WT BM→WT recipients) expressed similar levels of sIL-15 complexes as unmanipulated WT mice, whereas control IL-15Rα^−/− BM chimeras (Rko BM→Rko recipients) expressed undetectable levels of sIL-15 complexes (Fig. 2A). The levels of sIL-15 complexes were significantly decreased in BM chimeras lacking IL-15Rα in the hematopoietic compartment (Rko BM→WT recipients), but not in chimeras lacking expression of IL-15Rα by radiation-resistant parenchymal cells (WT BM→Rko recipients), indicating most of the sIL-15 complexes produced during the steady state is derived from hematopoietic cells (Fig. 2A).

To determine the cellular requirements of TBI-induced sIL-15 complexes, we examined cell-specific IL-15Rα–deficient mice. Serum was collected from CD11cCre^+/+ × IL-15Rα^fl/fl, LysMCre^+/+ × IL-15Rα^fl/fl, and IL-15Rα^fl/fl mice prior to and 24 h after TBI (1000 rads), and levels of sIL-15 complexes were examined by ELISA. During the steady state, modest decreases in sIL-15 complexes were observed in both CD11c-Cre × IL-15Rα^fl/fl mice and LysM-Cre × IL-15Rα^fl/fl mice (Fig. 2B); however, these trends were not statistically significant. In response to TBI, induction of sIL-15 complexes was substantially hampered in both the CD11cCre^+/+ × IL-15Rα^fl/fl and LysMCre^+/+ × IL-15Rα^fl/fl mice than in the control IL-15Rα^fl/fl mice (Fig. 2B). Overall, these experiments demonstrate that both CD11c⁺ and LysM⁺ cells, which consist mostly as DCs and macrophages, respectively, contribute minimally to the production of sIL-15 complexes during the steady state, but are substantial sources in response to TBI.

To specifically identify the type of myeloid cells expressing IL-15 after TBI as well as identify additional cell types that may be contributing to IL-15 production, we used an IL-15 transcriptional reporter mouse (developed in the laboratory of L. Lefrancois) that expresses GFP under the control of the IL-15 promoter (20). Cells were isolated 24 h after TBI from spleens and peripheral lymph nodes of WT and transgenic mice. TBI upregulated GFP expression in CD11b⁺ DCs, monocytes, CD8α⁺ DCs, and even in CD45⁻ cells (Fig. 2C). Unfortunately, GFP expression analysis of macrophages after TBI was inconclusive as changes in GFP were inconsistent (data not shown). Although neutrophils have been previously shown to express IL-15 mRNA at baseline and low levels of IL-15 protein upon in vitro stimulation (26), IL-15 reporter expression by neutrophils was not increased in response to TBI (Fig. 2C). Furthermore, reporter expression was not increased by CD103⁺DCs in the peripheral lymph nodes (Fig. 2C), showing that specific subsets of DCs are responsive to this type of stimulation. These results indicate that IL-15 transcriptional activity is induced by TBI in specific cells among both the hematopoietic and parenchymal compartments.

Because Ab-mediated lymphocyte depletion is more specific to lymphocytes than chemotherapy or irradiation and should not affect the integrity of the intestinal lining, we next examined how this model of lymphodepletion impacts production of sIL-15 complexes. Hence, mice were depleted of T cells using complement-fixing Ab, anti-Thy1 mAb (30H12, 300 μg, i.p.). After Ab treatment, levels of serum sIL-15 complexes significantly increased at 24 h, but, unlike after TBI, decreased slowly over time (Fig. 3A). Treatment with either depleting anti-CD4 Ab (GK1.5) or anti-CD8 Ab (2.43) alone had little effect on levels of sIL-15 complexes (Fig. 3B), indicating the amount of lymphocyte death occurring with these treatments is insufficient to induce sIL-15 complexes. Because complement-mediated depletion is an active process that leads to systemic inflammation (27), we examined the role of type I IFN signaling in the αThy1-mediated induction of sIL-15 complexes. In mice that are unable to respond to type I IFN (IFNAR^−/−), induction of sIL-15 complexes in response to αThy1 treatment was significantly reduced, corresponding to ∼50% reduction (Fig. 3B). Hence, whereas both TBI and complement-mediated depletion of T cells lead to a substantial amount of cell death, their reliance on type I IFN signaling to mediate induction of sIL-15 complexes varied (12). To examine cellular sources of sIL-15 complexes produced in response to αThy1-mediated T cell depletion, we again used cell-specific IL-15Rα–deficient mice. Levels of sIL-15 complexes were measured in serum of IL-15Rα^fl/fl, CD11cCre^+/+ × IL-15Rα^fl/fl, and LysMCre^+/+ × IL-15Rα^fl/fl mice prior to and 24 h after treatment with αThy1 Ab. A significant induction of sIL-15 complexes was observed in response to αThy1 Ab in all groups (Fig. 3C); however, the induction was only slightly reduced in the absence of IL-15 derived from DCs or monocyte/macrophage compared with WT controls (Fig. 3C). These results indicate that DCs, macrophages, and monocytes contribute to the production, but are not major sources of sIL-15 complexes in response to αThy1-mediated T cell depletion.

To determine how the cell surface IL-15 expression changed with Ab-mediated T cell depletion, splenocytes isolated at different times post-Thy1 Ab treatment were stained for surface IL-15 along with other cell surface markers to delineate different myeloid cell populations, as previously described. Increases in cell surface IL-15 by CD8⁺ DCs, CD11b⁺ monocytes, and macrophages was observed after Thy1 Ab treatment, but increases were not present in CD11b⁺ DCs at any time point (Fig. 3D). In the monocytes and macrophages, increases in surface IL-15 were apparent 1, 2, and 4 d after Ab treatment, whereas increases by CD8⁺ DCs were not evident until 2 d posttreatment (Fig. 3D). These results demonstrate that increases in IL-15 transpresentation can be concurrent with increases in sIL-15 complexes, similar to that observed after TBI. In addition, treatment with αThy1 Ab leads to a specific array of myeloid cells transpresenting IL-15. Altogether, these results directly show that specific depletion of T cells leads to the generation of sIL-15 complexes, which is partially driven by type I IFNs, as well as the upregulation of surface IL-15 by distinct myeloid cells.

As both complement activation and TBI lead to a significant amount of cell death, we next asked whether immunogenic cell death may be an important signal leading to induction of sIL-15 complexes. STING serves as an important recognition of cytoplasmic DNA, leading to the induction of type I IFNs (28–31). As type I IFN signaling is an important regulator of sIL-15 complexes during lymphopenia (12), we investigated the role of the STING pathway in the induction of sIL-15 complexes. Injection of STING agonist, c-di-GMP (25 μg, i.v.), induced high levels of circulating sIL-15 complexes as early as 12 h posttreatment, remained elevated at 24 h, and returned to almost baseline levels at 48 h (Fig. 4A). To determine whether STING agonist could directly induce sIL-15 complexes in DCs, Flt3L-derived BMDCs were treated with c-di-GMP. Indeed, STING agonist robustly induced production of sIL-15 complexes by DCs, which was more potent than IFN-α alone (Fig. 4B). Furthermore, in vivo induction of sIL-15 complexes by STING agonist was significantly impaired in both CD11cCre^+/+ × IL-15Rα^fl/fl and LysMCre^+/+ × IL-15Rα^fl/fl compared with control mice (Fig. 4C), showing both DCs and macrophages are major sources of sIL-15 complexes in response to this type of stimulation. Type I IFNs are a required intermediate in STING-induced sIL-15 complexes as serum sIL-15 complexes failed to be induced in IFNAR^−/− mice in response to treatment with the STING agonist, c-di-GMP (Fig. 4D). Collectively, to our knowledge, these findings are the first demonstration that stimulation of the STING pathway, via the induction of type I IFNs, leads to the induction of sIL-15 complexes.

As our previous studies have shown that type I IFNs are important for induction of sIL-15 complexes after TBI (12) and substantial cell death is induced by TBI, we investigated the requirement for STING signaling in TBI-induced sIL-15 complexes. We obtained commercially available Tmem173^−/− mice that are incapable of stimulating the STING pathway (16) and examined the ability of TBI to induce sIL-15 complexes in the absence of STING signaling. In Tmem173^−/− mice, induction of sIL-15 complexes by TBI was reduced by ∼50% (Fig. 4E), indicating optimal induction of sIL-15 complexes by TBI requires STING signaling. Because type I IFN signaling is completely required for TBI-induced sIL-15 complexes (12), these results also suggest other STING-independent pathways are driving production of type I IFNs after TBI. In response to αThy1-mediated T cell depletion, the early induction of sIL-15 complexes was unaffected in Tmem173^−/− mice, whereas the production of sIL-15 complexes (2 d after Thy1 Ab) was impaired (Fig. 4F), suggesting that STING signaling is required for the sustained generation of sIL-15 complexes in response to αThy1 treatment. Overall, these results show that stimulation of the STING pathway significantly contributes to the induction of sIL-15 complexes in response to TBI and Ab-mediated T cell depletion.

To date, we have examined the regulation of lymphopenia-induced IL-15 expression at the levels of transcription, transpresentation, and production of sIL-15 complexes. These findings provide evidence that a complex set of inflammatory signals is involved in the upregulation of IL-15 that is specifically hallmarked by the production of sIL-15 complexes. We suspect that the sIL-15 complexes produced after lymphodepletion contribute to enhanced responses by IL-15–responsive lymphocytes, such as memory CD8 T cells and NK cells. Because presently there is no model to distinguish the effects of sIL-15 complexes from those mediated by transpresented IL-15, we are unable to conclusively determine the functions of lymphopenia-induced sIL-15 complexes on responding lymphocytes. Nevertheless, we can address the role of IL-15 induced by inflammatory signals after αThy1 Ab-mediated T cell depletion, which is dominated by the expression of sIL-15 complexes. Specifically, we investigated the effects of αThy1-induced inflammatory IL-15 on memory CD8 T cell proliferation.

To generate memory CD8 T cells that are resistant to αThy1.2-mediated depletion, naive Thy1.1⁺ TCR transgenic Pmel-1 CD8 T cells (specific for the melanocyte differentiation Ag gp100 peptide 25–33 in the context of H2-D^b) were adoptively transferred to WT recipients (Thy1.2⁺), followed by infection with recombinant vesicular stomatitis virus expressing gp100 1 d later (18, 24). After the Pmel-1 T cells were allowed to differentiate into memory T cells (>30 d postinfection), total CD8 T cells (containing Thy1.1⁺ memory CD8 T cells) from the spleen were enriched via negative selection, CFSE labeled, and adoptively transferred to Thy1.2⁺ WT or IL-15Rα^−/− mice. One day posttransfer, all mice were injected with αThy1.2 to deplete host T cells and induce inflammatory IL-15 expression. Because the donor Thy1.1⁺ memory CD8 T cells are resistant to depletion by Thy1.2 Ab treatment, they can respond to the IL-15 upregulated within the environment. As expected, 1 d post-Ab treatment, sIL-15 complexes were elevated in blood from WT mice, but not in IL-15Rα^−/− mice (data not shown). After 5 d, the majority of Thy1.1⁺ T cells (between 75 and 95%) proliferated at least one to two divisions in the spleen and lymph nodes of WT mice (Fig. 5A, 5B). This depletion-induced T cell proliferation was fully dependent upon inflammatory IL-15, as few of the memory CD8 T cells (<25%) underwent proliferation in the IL-15Rα^−/− mice (Fig. 5A, 5B). The increase in proliferation also translated to an increase in the number of donor memory CD8 T cells (Fig. 5C); although the increased cell numbers were consistent, this effect was not statistically significant. These data directly confirm that inflammatory signals induced by Ab-mediated T cell depletion lead to the proliferation and accumulation of memory CD8 T cells in an IL-15–dependent manner. These findings reveal an unappreciated role of inflammatory signals in the regulation of IL-15 after lymphodepletion.

Induction of lymphopenia is an effective strategy to enhance T cell responses in various types of immunotherapies that work in part by increasing responses to homeostatic cytokines, such as IL-15 and IL-7 (1, 3, 32). Surprisingly, the mechanism(s) responsible for enhancing IL-15 responses during lymphopenia, until now, was unclear. This is attributed to the unique and recently appreciated mechanisms that are used by IL-15 to induce responses, which include transpresentation and sIL-15 complexes. Although transpresentation is believed to be responsible for driving IL-15 responses during homeostasis (13, 25, 33, 34), the production of sIL-15 complexes is associated with immune activation or inflammatory signals (12, 35). Herein, this study demonstrates that both transpresentation and sIL-15 complexes are distinctly regulated during lymphopenia, with each differentially affected by the manner in which lymphopenia is induced. Moreover, we provide clear evidence that inflammatory signals are an integral component in the regulation of IL-15, particularly in the induction of sIL-15 complexes after TBI and Ab-mediated lymphodepletion. These inflammatory signals are heterogeneous among the different methods of inducing lymphodepletion, with pathways utilizing STING and type I IFNs, but not completely dependent on any one pathway. Lastly, we provide evidence that elevated sIL-15 complexes are contributing to enhanced T cell responses after lymphodepletion. Our discovery mirrors the results described by the Harty Lab showing an important role for inflammatory IL-15 in the selective stimulation of memory CD8 T cells (36, 37).

Our results showing expression of sIL-15 complexes is limited to TBI- and Thy1-treated mice have led us to conclude that induction of sIL-15 complexes is related to the amount of inflammatory signals present in each model. Previous studies have reported that the cytotoxicity of CTX is due to the induction of apoptosis (38, 39), which is a type of cell death that avoids the release of cellular contents. We suspect that sIL-15 complexes were not generated in CTX-treated mice, despite the abundant amount of cell death because the type of cell death involved (apoptotic versus nonapoptotic) did not sufficiently ignite an inflammatory response. In the context of a weak inflammatory response, mouse colonies housed in different locations and thus harboring a different composition of microbiota could possess a higher or lower threshold of immune activation; this could be an explanation for why other groups clearly observed an induction of sIL-15 complexes with CTX (9) and we did not. In contrast to cell death induced by CTX, apoptosis is not the main process of cell death induced by irradiation (40) and complement-mediated cell death is clearly associated with an inflammatory response (27). This knowledge, together with our observation that STING signaling contributes to the production of sIL-15 complexes, to our knowledge, provides the first evidence that products of cell death induced by TBI or αThy1 Ab directly regulate IL-15 expression. Our inability to correlate increases in sIL-15 complexes with systemic LPS provides additional evidence that the presence of microbial products is not required for the upregulation of sIL-15 complexes. We do not discount the role of LPS in potentially driving production of sIL-15 complexes in other models, such as HIV (41) and some tumor models, which use numerous approaches to induce immune stimulation and/or disrupt immune homeostasis (8). Overall, our findings support the idea that abundant cell death and the immune system’s subsequent response to cellular byproducts is an important inflammatory pathway regulating IL-15 after lymphodepletion.

Although sIL-15 complexes were not increased in RAG^−/− mice, we and others (42) find cell surface IL-15 expression by DCs is increased, suggesting lymphopenia alone is sufficient to upregulate IL-15. Because there is no overt immune stimulation in RAG^−/− mice, we postulate the upregulated surface IL-15 is due to a passive mechanism, such as the presence of a cytokine sink. This theory is supported by the lack of upregulated IL-15 transcription that we observe in DCs isolated from RAG^−/− mice. Cell surface IL-15 is also increased in the other models of lymphopenia, which could be mediated by both active inflammatory signals and passive mechanisms. At least with TBI, our observation that IL-15 transcription is increased in multiple cell types indicates the presence of an activating signal. Signaling via type I IFNs is at least one potential inflammatory mechanism increasing surface IL-15 expression levels, as type I IFNs are produced after TBI and directly increase IL-15 transcription and cell surface expression (12, 43–45). Nevertheless, some increased surface IL-15 could also be due to the presence of a cytokine sink as these mice are lymphocyte deficient. The increase in surface IL-15 along with the absence of sIL-15 complexes in RAG^−/− mice suggests the enhanced IL-15 responses reported in these mice are driven by enhanced transpresentation of IL-15. As such, increased levels of IL-15 transpresentation in any model of lymphopenia have the potential to modulate lymphocyte responses. Alternatively, increasing the levels of surface IL-15 could be an additional mechanism to increase IL-15 expression prior to cleavage, thus enhancing the production of sIL-15 complexes. Unfortunately, for circumstances in which both transpresentation and sIL-15 complexes are increased, there currently are no model systems that can be used to discern the contribution of each mechanism in driving IL-15 responses.

Whereas increases in IL-15 transpresentation are observed to some degree in each model of lymphopenia, the type of cells transpresenting IL-15 under these different conditions varied. This differential cellular expression most likely reflects the diverse signals regulating IL-15 among the different model systems, which again could involve both passive and active forces. A previous study determined that the loss of DC-specific IL-15 expression predominantly affected CD8 T cells with a central memory phenotype, whereas loss of macrophage-specific IL-15 expression resulted in a specific loss of effector memory CD8 T cells (13). If similar effects were observed during adoptive T cell therapy, then one would predict that CTX treatment would better support central memory T cells, whereas TBI treatment would better support effector memory T cells.

In addition to differences in cellular IL-15 expression, we revealed unique kinetics in the production of sIL-15 complexes among models of lymphopenia that could influence T cell responses in adoptive T cell therapy. Studies investigating differences in the signaling of IL-2 and IL-15 have found evidence that the duration of cytokine signal can differentially affect T cell responses (46, 47). Specifically in infection models, IL-15 signaling promotes differentiation of short-lived effector T cells over memory precursor T cells (48, 49), whereas later in the response, it supports the maintenance of memory CD8 T cells. Hence, similar effects on differentiation may be observed during adoptive T cell therapy. Alternatively, with the chronic stimulation associated with a tumor setting, a more continuous exposure to sIL-15 complexes may contribute to the exhaustion of tumor-specific T cells. Future studies should be carried out to better characterize the different kinetics of IL-15 after lymphodepletion regimens in human patients while also taking into consideration how treatment with recombinant formulations of IL-15 or IL-15 agonists may complement the changing levels of endogenous sIL-15 complexes.

In summary, this study reveals that the regulation of IL-15 after lymphodepletion is far more complex than previously appreciated and how expression of sIL-15 complexes and transpresented IL-15 are each independently regulated by a variety of mechanisms and/or pathways. By showing that both transpresented IL-15 and sIL-15 complexes are upregulated during lymphopenia, we provide evidence that IL-15 responses mediated during lymphopenic conditions could be elicited by transpresented IL-15, sIL-15 complexes, or the combination of both. Much of the regulation of IL-15 during lymphodepletion is driven by inflammatory signals that involve the immune system’s response to products of cell death and type I IFNs. During our investigations, we also identified the STING pathway as another inflammatory pathway regulating sIL-15 complexes, which can contribute to enhanced T cell responses in circumstances associated with release of cytosolic DNA. Because STING agonists are currently being pursued as adjuvants in antitumor immune responses, it is likely that IL-15 induced by this treatment is contributing to enhanced CD8 T cell responses. Overall, the unique expression profile of IL-15 in each model could impact IL-15–responsive lymphocytes in a distinct way, which provides a rationale to better investigate the effects of lymphodepletion on T cell differentiation and its corresponding effects on T cell effector functions and persistence. Understanding these differences and their impact could be one approach toward fine-tuning T cell responses during immunotherapy for enhanced antitumor responses.

We thank Dr. Willem Overwijk for helpful comments and suggestions and for sharing IFNAR1^−/− and Pmel-1 Tg mice and Dr. Lynn Puddington for mentoring support.

The authors have no financial conflicts of interest.