Effect of anti-IL-15 administration on T cell and NK cell homeostasis in rhesus macaques

Abstract

IL-15 has been implicated as a key regulator of T and NK cell homeostasis in multiple systems; however, its specific role in maintaining peripheral T and NK cell populations relative to other gamma-chain (γc) cytokines has not been fully defined in primates. Here, we address this question by determining the effect of IL-15 inhibition with a rhesusized, anti-IL-15 mAb on T and NK cell dynamics in rhesus macaques. Strikingly, anti-IL-15 treatment resulted in rapid depletion of NK cells, and both CD4⁺ and CD8⁺ effector memory T cells (T_EM) in blood and tissues, with little to no effect on naïve or central memory T cells. Importantly, whereas depletion of NK cells was nearly complete and maintained as long as anti-IL-15 treatment was given, T_EM depletion was countered by the onset of massive T_EM proliferation, which almost completely restored circulating T_EM numbers. Tissue T_EM, however, remained significantly reduced, and most T_EM maintained very high turnover throughout anti-IL-15 treatment. In the presence of IL-15 inhibition, T_EM became increasingly more sensitive to IL-7 stimulation in vivo, and transcriptional analysis of T_EM in IL-15-inhibited monkeys revealed engagement of the JAK/STAT signaling pathway, suggesting alternative γc cytokine signaling may support T_EM homeostasis in the absence of IL-15. Thus, IL-15 plays a major role in peripheral maintenance of NK cells and T_EM. However, whereas most NK cell populations collapse in the absence of IL-15, T_EM can be maintained in the face of IL-15 inhibition by the activity of other homeostatic regulators, most likely IL-7.

INTRODUCTION

Lymphocyte homeostasis and function are tightly regulated by the activities of the common gamma chain (γc) cytokines, in particular IL-2, IL-7 and IL-15. While these 3 cytokines share the γc receptor (CD132) their activity on various lymphocyte populations are often different, differences that are, in part, mediated by differential expression of the α-chain component of the receptor (1–7). For instance, IL-2 is mainly produced by CD4⁺ T cells, and to a lesser degree CD8⁺ T cells, NK cells and NKT cells, and it acts to generally potentiate the expansion of activated T and NK cells (8, 9). However, IL-2 has a unique primary role in regulating immune tolerance by promoting the production and maintenance of CD4⁺ CD25⁺ Foxp3⁺ T regulatory (Treg) cells, which constitutively express IL-2Rα (CD25). Indeed, this was demonstrated in early models of IL-2^−/− and IL-2Rα^−/−/IL-2Rβ^−/− knockout (KO) mice, which manifested rapid lethal autoimmune diseases likely resulting from the failure of Treg development and homeostasis (10). IL-7 is produced by non-hematopoietic cells such as stromal and epithelial cells and is important for thymocyte development and peripheral T cell homeostasis. IL-7 signals via the receptor IL-7R, which is a heterodimer consisting of the IL-7Rα (CD127) and CD132. IL-7 is particularly important for promoting the proliferative expansion and survival of naïve T cells (T_N) and central memory T cells (T_CM), which maintain high levels of CD127 expression (11). IL-15, on the other hand, has been shown to regulate the homeostasis and activation of many cell types throughout the body, including memory T cells, NK cells, invariant NKT cells, γδ T cells and intestinal intraepithelial lymphocytes (12–22). As such, IL-15Rα^−/− and IL-15^−/− KO mice typically manifest severe deficiencies in these cell subsets (23). IL-15 signals via interaction with heterotrimeric receptor complex composed of IL-15Rα (CD215), IL-2/15Rβ (CD122) and CD132 (16, 24–27). IL-15Rα is expressed on antigen-presenting cells (macrophages, monocytes and dendritic cells) and binding of IL-15 to IL-15Rα enables trans-presentation to a responding cell expressing CD122 and CD132 (28). Because biologically active IL-15 has been shown to exist in a soluble form in complex with IL-15Rα (29), it has been questioned whether IL-15Rα is part of the receptor for IL-15 or part of a heterodimeric cytokine that interacts with the CD122/CD132 receptor. In either scenario, the IL-15Rα protein is thought to be a critical determinant of IL-15 specificity and function.

Although each of these γc cytokines has unique characteristics, their in vivo activity often manifests considerable overlap. For instance, IL-2 and IL-15 share the same β receptor (CD122) and are both involved in the initial amplification of antigen-specific T cell responses, and the regulation of memory T cell development, differentiation, and maintenance (30–32). In addition, both IL-2 and IL-15 induce the activation and proliferation of NK cells and enhance NK cell cytolytic activity by inducing the up-regulation of effector molecules such as perforin and granzyme B (33–35). Similarly, IL-7 and IL-15 both seem to play major, albeit non-exclusive, roles in maintaining peripheral T_M homeostasis, supporting both T_M proliferation and survival (31). Thus, the specific non-redundant roles these γc cytokines play in controlling various lymphocyte population dynamics in vivo are not completely characterized, a lack of understanding that complicates efforts to rationally develop therapeutic strategies based on their specific biologic activities to enhance immune responses to cancer or microbial agents, to promote immune reconstitution after conditions of lymphopenia (HIV infection, chemotherapy, aging), or to counter pathologic immune responses in the various autoimmune/inflammatory disorders (rheumatoid arthritis, celiac disease, inflammatory bowel disease, multiple sclerosis and type 1 diabetes) linked to dysregulation of these cytokines (36–40).

Due to its activity on NK cells and antigen-specific cytotoxic T cells, IL-15 is in clinical trials for the treatment of metastatic malignancies (41). Previous studies have shown that IL-15 can increase the production of long-lived antigen-specific T_M (32, 42, 43), and can also induce the migration and redistribution of T_M from circulation into tissues (44, 45). In nonhuman primates (NHP), provision of exogenous IL-15 typically induces an initial brief period of lymphopenia followed by lymphocytosis (45–47). Lymphocytosis is associated with the expansion of NK cells and T_M (41, 44). However, the T_M compartment is quite heterogeneous and comprises the T_CM subset, which is responsible for anamnestic T cell responses and primarily recirculates between secondary lymphoid tissues, and the effector-differentiated memory subsets – transitional memory (T_TrM) and T_EM - which can also migrate to extra-lymphoid effector sites (48). In NHP, T_EM and T_TrM are very responsive to IL-15 in vivo, and studies from our lab and others have shown that exogenous IL-15 administration can dramatically increase the proliferative fraction and absolute numbers of T_EM and T_TrM in peripheral blood (44, 46, 47). In contrast, T_N and T_CM are less responsive to IL-15 administration, and it is still unclear whether IL-15 plays a direct role in vivo in regulating their homeostasis. Most of these studies have focused on CD8⁺ T cells, and in general, IL-15 has been more closely associated with regulation of CD8⁺ T_M than with CD4⁺ T_M. However, CD4⁺ T_EM and T_TrM are also highly responsive to IL-15 in vivo, suggesting that this discrepancy is largely attributable to the fact that the highly IL-15-responsive T_EM subset comprises a much larger fraction of total circulating T_M for CD8⁺ than for CD4⁺ lineage cells (44).

In this study, we sought to determine the specific, non-redundant role(s) of IL-15 in the regulation of CD4⁺ and CD8⁺ T and NK cell population dynamics in NHP. As indicated above, most previous studies addressing IL-15 activity in NHP have assessed the effect of therapeutic administration of IL-15, which results in unnaturally high levels of cytokine immediately after administration. Since such pharmacologic (supra-physiologic) IL-15 levels may not accurately reflect physiologic cytokine function in vivo, we have taken the alternative approach of determining the impact of inhibiting IL-15 activity in vivo with a newly developed rhesusized anti-IL-15 monoclonal antibody (mAb) on T cell and NK cell homeostasis in rhesus macaques (RM). We demonstrate that this rhesusized anti-IL-15 can be repeatedly administered to RM and is highly effective at long-term inhibition of IL-15 activity in vivo. We further demonstrate that in vivo inhibition of IL-15 activity resulted in a near complete depletion of NK cells and a significant decrease in the numbers of circulating CD4⁺ and CD8⁺ T_EM with negligible effects on the T_CM or T_N subsets. Strikingly, however, T_EM, but not NK cell numbers, rebounded by proliferative expansion, and in the absence of IL-15 signaling, T_EM became increasingly more sensitive to IL-7 signaling. These data suggest that whereas IL-15 signaling is required for NK cell homeostasis, T_EM can be maintained by other cytokines, most likely IL-7, when IL-15 signaling is not available.

MATERIALS AND METHODS

Animals

A total of 41 purpose-bred RM (Macaca mulatta) of Indian genetic background and free of Macacine herpesvirus 1, D type simian retrovirus, simian T-lymphotrophic virus type 1, and SIV infection were used in this study. A group of 30 RM were administered the rhesus recombinant anti-IL-15 mAb, clone M111 (n=17) or rhesus recombinant IgG control antibody (n=13), i.v. once every two weeks at 20mg/Kg on day 0 and 10mg/Kg on days 14 and 21. The neutralizing anti-IL-15 antibody was constructed by grafting the complementarity determining regions of mouse anti-human IL-15 mAb, M111 (American Type Culture Collection) into rhesus variable region frameworks. MAb was expressed in Chinese hamster ovary cells as full-length Ig with rhesus IgG1 and kappa constant regions. Two RM were dosed with an immunologically silenced form of the same antibody (anti–IL-15 LALA) which contained the L235A, L236A mutation in the heavy chain constant region. RM dosed with the anti-IL-15 LALA mAb displayed similar phenotypes to the monkeys treated with the non-mutated anti-IL-15 mAb for all parameters assayed, and were therefore grouped as anti-IL-15 mAb-treated RM. BrdU (Sigma-Aldrich) was prepared as previously described (49), and administered i.v. in three separate doses of 30mg/Kg body weight over a 24-hour period prior to terminal necropsy at ≥35 days post-first anti-IL-15 treatment. A separate group of 11 RM received 6 biweekly doses of rhesus recombinant anti-IL-15 mAb (n=6) or rhesus recombinant IgG control (n=5) at 20mg/Kg on day 0, followed by 10mg/Kg on days 14, 28, 42, 56 and 77, concurrently, with subcutaneous administration of rhesus recombinant IL-7 at 30μg/Kg on days 35 and 42. All RM were housed at the Oregon National Primate Research Center in accordance with standards of the Center’s Institutional Animal Care and Use Committee and the National Institutes of Health Guide for the Care and Use of Laboratory Animals.

Measurement of plasma IL-7 concentration

The concentration of IL-7 in the plasma was measured by ELISA using the IL-7 Quantikine HS kit (R & D Systems) according to manufacturer’s instructions.

Flow cytometric analysis and cell sorting

Whole blood and mononuclear cells isolated from lymph nodes (LN), bronchoalveolar lavage (BAL), bone marrow, spleen, kidney, lung, liver, tonsil, and vaginal and intestinal mucosa were obtained and stained for flow cytometric analysis as described previously (49, 50). Polychromatic (8–12 parameter) flow cytometric analysis was performed on an LSR II instrument using Pacific blue, AmCyan, FITC, PE, PE–Texas red, PE-Cy7, PerCP-Cy5.5, Allophycocyanin, Allophycocyanin-Cy7, and Alexa Fluor 700 as the available fluorescent parameters. Instrument set-up and data acquisition procedures were performed as previously described (51). List mode multiparameter data files were analyzed using FlowJo software. Delineation of naive and memory T cell subsets and criteria for setting + versus − markers for CCR5 and Ki-67 expression have been previously described (49, 50, 52–54). In brief, T_N constitute a uniform cluster of cells with a CD28^moderate, CCR7⁺, CCR5⁻, CD95^low phenotype, which is clearly distinguishable from the phenotypically diverse memory population that is CD95^high and may display one or more of the following non-naive phenotypic features: CD28⁻, CCR7⁻, or CCR5⁺. The T_CM, T_TrM, and T_EM components of the memory subset in the blood were further delineated based on the following phenotypic criteria: T_CM (CD28⁺, CCR7⁺ and CCR5⁻), T_TrM (CD28⁺, CCR7^+/− and CCR5⁺), and T_EM (CD28⁻, CCR7⁻ and CCR5^dim). For delineating NK cells in blood, small lymphocytes were gated to obtain CD3⁻, CD8α⁺, NKG2a⁺ cells that were CD20⁻ and CD14⁻. NK cell subsets were further delineated based on CD16 and CD56 expression as previously described (55). For each subset to be quantified, the percentages of the subset within the overall small lymphocyte populations were determined. For quantification of peripheral blood subsets, absolute small lymphocyte counts were obtained using an AcT5diff cell counter (Beckman Coulter) and from these values, absolute counts for the relevant subset were calculated based on the subset percentages within the light scatter–defined small lymphocyte population on the flow cytometer. Baseline values were determined as the average of values at days −14, −7 and 0. Results are presented as percentage of baseline, with baseline shown as 100%, or changes in proliferative fraction, indicated as the difference in the %Ki-67⁺ (Δ%Ki-67⁺) measured at the designated time points from baseline (0% = no change).

For analysis of pSTAT5 expression, whole blood (100μL) was added to polystyrene flow tubes and stained with fluorochrome-conjugated mAbs against CD3, CD4, CD8, CD28, CCR5, CCR7 and CD95, at room temperature (RT) for 30 minutes. Next, tubes were either left unstimulated or stimulated with increasing concentrations of IL-7 or IL-15 (ranging from 0.5 – 32ng/mL) for 15 min at 37°C/5% CO₂. Detection of pSTAT5 was assessed with the BD Phosflow staining protocol according to the manufacturer’s instructions. Briefly, cells were fixed with the BD Phosflow Lyse/Fix buffer for 5 min at RT and permeabilized in ice-cold BD Phosflow perm buffer IV for 5 min at RT. After washing, cells were stained for intracellular markers with fluorochrome-conjugated anti-pSTAT5 and anti-Ki-67 for 45 min at RT. Cells were washed and flow cytometric analysis was performed on the LSR II flow cytometer.

In vitro cytokine-induced expansion assay

PBMCs were sort purified using a FACS Aria II (BD Biosciences) based on defined phenotypic markers as described above and plated in 48-well plates in 1mL of R10 media [RPMI (HyClone), 10% Fetal Bovine Serum (FBS), 100units/mL Penicillin, 10mg/mL Streptomycin (Sigma-Aldrich), 200μM L-glutamine (Sigma-Aldrich)] at a density of 150,000 to 300,000 cells/mL. IL-7 or IL-15 were added at a concentration of 50ng/mL to the cultures and incubated at 37°C/5% CO₂ for 14 days alone or in the presence of 10% sort purified CD14⁺ monocytes. After 7 days, the culture was resuspended and 0.5mL was removed for phenotypic analysis by flow cytometry. An equal amount of fresh R10 was added back to the remaining culture and incubated at 37°C/5% CO₂ for a further 7 days. On day 14, the entire culture was harvested for phenotypic analysis by flow cytometry.

Antibodies and cytokines

The following antibodies were used for flow cytometry: CD3 Alexa 700 (SP34-2 BD Biosciences), CD4 AmCyan (L200 BD Biosciences), CD8 PerCP-Cy5.5 (SKI eBiosciences), CD8 AmCyan (SKI BD Biosciences), CD28 PE-Texas Red (CD28.2 Beckman Coulter, BD Biosciences), CD95 PE (DX2 BD Biosciences, eBiosciences), CCR5 Allophycocyanin (3A9 BD Biosciences), Ki-67 FITC (B56 BD Biosciences), CD56 PerCP-Cy5.5 (MEM-188 Invitrogen), CD16 Pacific Blue (3G8 BD Biosciences, Biolegend), CD20 Allophycocyanin-Cy7 (L27 BD Biosciences), HLA-DR PE-Texas Red (TU36 Invitrogen, Immu357 Beckman Coulter), NKG2A PE (Z199 Beckman Coulter), CD14 FITC (M5E2 BD Biosciences, R&D Systems), STAT5 PE (47/Stat5 (pY6) BD Biosciences), BrdU FITC (B44 BD Biosciences), BrdU Allophycocyanin (B44 BD Biosciences). Anti-CCR7 (150503) was purchased as purified immunoglobulin from R&D Systems, conjugated to biotin using a Pierce Chemical Co. biotinylation kit, and visualized with streptavidin–Pacific Blue (Invitrogen). Rhesus recombinant anti-IL-15 and rhesus recombinant control IgG1 mAb were provided through the National Institutes of Health’s Nonhuman Primate Reagent Resource Program. Simian recombinant IL-7 was provided by Cytheris SA (Issy-Les-Moulineaux, France). Rhesus recombinant IL-15 was provided by Francois Villinger (Emory University) through the Resource for Nonhuman Primate Immune Reagents.

Immunohistochemistry

Immunohistochemistry was performed as previously described (56). Antibodies used in this study were mouse monoclonal anti-human IL-15 (Antibodies Online; clone BDI150), rabbit monoclonal anti-active caspase-3 (Cell Signaling Technologies; clone 5A1E), rabbit monoclonal anti-phosphorylated STAT5 (1:100; clone C11C5; Cell Signaling Technologies, Inc.), and rabbit monoclonal anti-human CD3 (clone SP7; Labvision/Thermo Fisher Scientific). All stained slides were scanned at high magnification (x200) using the ScanScope CS System (Aperio Technologies) yielding high-resolution data from the entire tissue section. Representative regions of interest (ROIs; 250 to 500 mm²) were identified and high-resolution images extracted from these whole-tissue scans. The percent area of the lymph node T cell zone and lamina propria (colon) that stained for each protein/cell type of interest were quantified using Photoshop CS5 using Fovea tools.

Microarray analysis

For transcriptional analysis, sort-purified CD8⁺ T_EM and CD8⁺ T_CM from PBMC, obtained 28–49 days post-first anti-IL-15 or control IgG mAb administration, were resuspended in RLT lysis buffer (Qiagen) and stored at −80°C until use. RNA was isolated using RNeasy Micro Kits (Qiagen), and the quantity and quality of the RNA was confirmed using a NanoDrop 2000c (Thermo Fisher Scientific) and an Experion Electrophoresis System. Samples (50ng) were amplified using Illumina TotalPrep RNA amplification kits (Ambion). The microarray analysis was conducted using 750ng of biotinylated complementary RNA hybridized to HumanHT-12_V4 BeadChips (Illumina) at 58°C for 20h. The arrays were scanned using Illumina’s iSCAN and quantified using Genome Studio (Illumina). The analysis of the Genome Studio output data was conducted using the R and Bioconductor software packages. Quantile normalization was applied, followed by a log₂ transformation performed using the Bioconductor LIMMA package (57). Outlier samples with abnormalities in gene expression based on hierarchical clustering methods and multidimensional scaling analysis as a dimensionality reduction method for the evaluation of similarities or dissimilarities between samples were identified and removed. The LIMMA package was used to fit a linear model to each probe and perform (moderated) t tests or F tests on the groups being compared. A Gene Set Enrichment Analysis (GSEA) (58) using 1000 permutations was performed on the genes differentially expressed between the groups compared in the subsets pre-ranked by the decreasing order of the absolute T-statistic. The canonical pathways of Ingenuity Pathway Analysis software (IPA, Ingenuity Systems) were used as the database to perform GSEA. This was followed by building modules of related pathways based on at least 25% gene overlap [Jaccard index (59) > 25%] between pathways using the enrichment map (60) strategy and representing the genes present in at least 25% of the pathways in the module. To control the expected proportions of false positives, the FDR for p-values was calculated using the Benjamini and Hochberg method implemented in LIMMA. The complete dataset is available at the Gene Expression Omnibus microarray repository (http://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?token=mbqtaouqztynzsz&acc=GSE76797), accession number GSE76797.

Statistics

To investigate if the administration of anti-IL-15 mAb blocks signaling in circulating lymphocytes, we estimated the EC₅₀ (the dose where the response is the midpoint between the maximum and minimum) for each RM at given time points (days post-mAb) using the four-parameter logistic model often referred to as an E_max model, followed by repeated measures ANOVA to evaluate the effect of anti-IL-15 mAb with EC₅₀ as a response variable and anti-IL-15 mAb-treated status as between group factor, and days post-antibody as within group factors. Longitudinal analysis of STAT5 phosphorylation, peripheral blood T_N and T_M (including T_CM, T_TrM and T_EM subsets) counts and proliferation, peripheral blood NK cell counts and plasma IL-7 levels were evaluated using repeated-measures ANOVA with anti-IL-15 mAb and IgG control mAb-treated groups as between group factors and time points as within group factors; since in a typical experiment using repeated measures, two measurements taken at adjacent times are more highly correlated than two measurements taken several time points apart. Due to the limited sample size, a simpler covariance structure, first order auto-regressive (AR1), was used as correlation within each animal. Tukey-Kramer adjustment was used to control for multiple comparisons. Differences in the number of pSTAT5⁺ cells/mm² in peripheral lymph nodes between pre-treatment and post-treatment were evaluated using Wilcoxon Signed rank tests. The difference in CD3⁺ caspase-3+ in colon and peripheral LN via immunostaining and BrdU incorporation in CD4⁺ and CD8⁺ T cell subsets between anti-IL-15 and IgG control mAb-treated groups was compared using Mann-Whitney U test. Data was analyzed using SAS 9.4. The p-values <0.05 were considered to be statistically significant.

RESULTS

Differential signaling of IL-15 and IL-7 on NK and T cell subsets in vitro

As a first step in characterizing the specific, non-redundant in vivo role(s) of IL-15 in controlling NHP T cell and NK cell population dynamics, we determined the extent to which RM CD4⁺ and CD8⁺ T cell and NK cell subsets, studied directly ex vivo in peripheral blood, could respond to IL-15 and IL-7. Since common γc cytokines mediate signal transduction via the JAK/STAT signaling pathway leading to the phosphorylation of the transcription factor STAT5 (61), we quantified the fraction of cells within each phenotypically defined subset that phosphorylated STAT5 (e.g., became pSTAT5⁺) within 15 min after exposure to increasing concentrations of these two γc cytokines. As shown in Figs. 1A and 1B, both IL-15 and IL-7 induced pSTAT5 in naïve and all memory T cell subsets; however, there were clear differences among these subsets with respect to their dose response to these cytokines, and the fraction of responding cells at the optimal dose. T_N and T_CM (both CD4⁺ and CD8⁺) were more responsive to IL-7 than to IL-15 with over 60% of cells in these subsets expressing pSTAT5 at 4ng of IL-7 vs. less than 12% of T_N and T_CM at the same dose of IL-15. In contrast, both CD4⁺ and CD8⁺ T_EM showed increased responsiveness to IL-15 compared to IL-7 with pSTAT5 expression between 60–80% at 8ng, in contrast to less than 40% with the same dose of IL-7. This reduced response to IL-7 is likely due to the low expression of CD127 on T_EM (62). Interestingly, CD4⁺ and CD8⁺ T_TrM responded to IL-7 and IL-15 at similar levels, reflecting their intermediate phenotype between T_CM precursors and effector-differentiated T_EM [Supplementary Figure 1; (54)]. NK cells showed an even larger difference in responsiveness to IL-7 vs. IL-15. Although all circulating NK cell subsets [defined by CD16 and CD56; (55)] responded robustly to IL-15 with induction of pSTAT5 expression in >80% of cells at 16ng (Fig. 1C), only the CD56⁺ CD16⁻ “regulatory” NK cell subset showed any response to IL-7 stimulation at the tested doses, and this response required a 4-fold higher dose of IL-7 than the response of the same cells to IL-15.

Figure 1.

Comparative analysis of IL-7 and IL-15 signaling in T cell and NK cell populations. (A) Representative histograms showing the induction of pSTAT5 in the indicated peripheral blood T cell subsets after ex vivo stimulation with 8ng/mL of IL-7 (blue), 8ng/mL of IL-15 (red) or no stimulation (gray). (B) Dose response curves showing the induction of pSTAT5 in the indicated peripheral blood T cell subsets after ex vivo stimulation with 0, 0.5, 1, 2, 4, 8, 16, 32 ng/mL of IL-7 (n=14) or IL-15 (n=14). Results (mean + SEM) are shown as change from baseline. (C) Induction of pSTAT5 in the indicated peripheral blood NK cell subsets after ex vivo stimulation with 0, 0.5, 1, 2, 4, 8, 16, 32 ng/mL of IL-7 (n=8) or IL-15 (n=8). Results (mean + SEM) are shown as change from baseline.

These results demonstrate that lymphocyte populations differ in their response to early signal transduction by IL-7 vs. IL-15, and suggest that IL-7 and IL-15 preferentially signal pre-effector and effector populations, respectively. We next sought to determine the degree to which this differential signaling translates to downstream gene expression and homeostatic regulation by examining the ability of these cytokines to support survival and proliferative expansion of sort-purified T_N, T_CM and T_EM during 7 and 14 days of in vitro culture. Since the pattern of STAT5 phosphorylation to IL-7 and IL-15 among these differentiation-defined subsets was similar between the CD4⁺ and CD8⁺ lineages (Fig. 1A–B), we primarily focused on CD4⁺ T cells for these studies. In this in vitro CD4⁺ T cell model system, responses to IL-15 were generally more robust than to IL-7, but responsiveness to both cytokines was differentiation dependent. T_EM were again found to be highly responsive to IL-15 compared to IL-7, with over 60% of IL-15-treated cells expressing the proliferation antigen Ki-67 by day 7 of culture, compared to background Ki-67 expression after IL-7 treatment (Fig. 2A). IL-7 and IL-15 induced similar, modest increases in Ki67 expression by CD4⁺ T_CM, whereas IL-7 was more effective than IL-15 in inducing T_N proliferation. Next, based on the concept that IL-15 might be more efficiently presented in trans, we cultured the same cells with IL-15 or IL-7 in the presence of sort-purified CD14⁺ monocytes (63, 64). In the presence of monocytes, we saw modestly enhanced responsiveness of T_N, T_CM, and T_EM to IL-7, and profoundly increased responsiveness of T_N and T_CM to IL-15 (Fig. 2B). Interestingly the CD4⁺ T_EM proliferative response to IL-15 was similar with or without monocytes, suggesting that trans-presentation or other monocyte-derived signals are not required for maximal T_EM responsiveness. Similar IL-15 responsiveness was demonstrated for CD8⁺ T_EM (Supplementary Figure 2A).

Figure 2.

Representative dot plots demonstrative of 8 separate experiments, showing induction of Ki-67 expression on CD4⁺ T_N, T_CM and T_EM. CD4⁺ T cell subsets were purified from PBMC by multi-parameter cell sorting and cultured in media containing (A) recombinant IL-15 or recombinant IL-7 at 50ng/mL, alone or (B) in the presence of 10% highly purified CD14⁺ monocytes. After 7 and 14 days, cultures were analyzed for Ki-67 expression.

Collectively, these in vitro studies demonstrate that T_N and T_CM are more responsive to IL-7 than T_EM, whereas T_EM are more responsive to IL-15. However, IL-15 can induce robust proliferation in T_N and T_CM in the presence of monocytes, suggesting a more pleotropic role for this cytokine. Interestingly, culture of T_N and T_CM with IL-15 in the presence of monocytes induces up-regulation of CCR5 and down-regulation of CCR7 (Supplementary Figure 2B), consistent with induction of effector memory differentiation (50, 65). Thus, IL-15 may both maintain the homeostasis of pre-existing T_EM and, under certain conditions, drive T_EM differentiation from noneffector-differentiated precursors.

Anti-IL-15 specifically blocks IL-15 signal transduction in vivo

To investigate the role of IL-15 in T cell and NK cell homeostasis in vivo, we developed a “rhesusized” anti-IL-15 mAb that is suitable for repeated administration to RM. This mAb was based on the mouse, anti-human IL-15 mAb clone, M111, which is cross-reactive with RM IL-15 (66). Rhesusization was accomplished by exchanging the mouse amino acid sequences in the constant regions and variable binding surfaces for rhesus amino acid sequences, leaving variable region sequences responsible for IL-15 recognition unchanged (67, 68). We hypothesized that repeated administrations of this rhesusized M111 mAb would block IL-15 activity in vivo and result in changes to T and/or NK cell population dynamics that would reflect physiologic IL-15 function. To initially test this approach, we administered rhesus anti-IL-15 or a rhesus IgG control mAb to RM every other week for 6 weeks prior to necropsy, as shown in Fig. 3A. We assessed IL-15 signaling inhibition by performing pSTAT induction analysis with rhesus recombinant IL-15 on whole blood at various time points post-treatment. From day 1 through day 35 post-treatment, the ability of IL-15 to induce pSTAT5 expression in CD4⁺ and CD8⁺ T cells ex vivo in anti-IL-15-treated RM was significantly reduced (essentially abrogated) compared to baseline and to control IgG-treated RM (p<0.0001; Fig. 3B and Supplementary Figure 3A–B). To further assess the in vivo activity of our anti-IL-15 mAb treatment, we performed immunohistochemical analysis of LN sections from both control IgG mAb-treated and anti-IL-15 mAb-treated RM using the original mouse anti-IL-15 mAb M111. As expected, there was a substantial reduction in IL-15 immunoreactivity in LN of anti-IL-15-treated RM compared with control RM (Fig. 3C), demonstrating either absence of IL-15 post-treatment or blocking of immunostaining by the administered function-abrogating mAb (consistent with in situ neutralization). As a final confirmation of the in vivo effectiveness of our anti-IL-15 mAb, we determined whether anti-IL-15 treatment reduced JAK/STAT signaling in situ by measuring pSTAT5 expression in LN sections. As shown in Fig. 3D and Supplementary Figure 3C, we observed that RM treated with the anti-IL-15 mAb, but not the IgG control mAb, had a significant reduction in pSTAT5 detection post-treatment (untreated animals: p=0.38, anti-IL-15-treated RM: p<0.0001). This decline in pSTAT5 expression indicates a reduction in JAK/STAT signal transduction specific to IL-15. A complete inhibition of pSTAT5 in tissues was not expected as other γc cytokines such as IL-2 and IL-7, which are not inhibited by M111, also signal via the JAK/STAT pathway. Taken together, these results clearly demonstrate that the rhesusized anti-IL-15 mAb can effectively inhibit IL-15 signaling in vivo.

Figure 3.

Anti-IL-15 administration specifically blocks IL-15 signaling in RM. (A) Schematic representation of the anti-IL-15 treatment schedule used in this study. Healthy RM received 20mg/Kg of anti-IL-15 or IgG control antibody on day 0 and 10mg/Kg on days 14 and 28. 24 hours prior to necropsy, all RM received 3 doses of BrdU at 30mg/Kg. (B) Showing the change in pSTAT5 in peripheral blood T_N, T_CM, T_TrM and T_EM, after ex vivo stimulation with 8ng/mL of IL-15 in anti-IL-15 mAb-treated (n=9) or IgG control mAb-treated (n=5) RM. Results (mean + SEM) are shown as percentage of baseline and significance was assessed as described in materials and methods. (C) Representative images from immunohistochemical analysis performed on lymph node (LN) sections obtained from IgG control mAb-treated (top panel) and anti-IL-15 mAb-treated (bottom panel) RM using a mouse mAb which is cross-reactive with rhesus IL-15 (original magnification 200X). (D) Quantification of the number of pSTAT5⁺ cells/mm² in peripheral LN of RM treated with anti-IL-15 mAb (n=15) or IgG control mAb (n=8). Significance was evaluated using Wilcoxon Signed rank tests.

IL-15 blockade results in near complete NK cell depletion

Having demonstrated the in vivo activity of the rhesusized anti-IL-15 mAb, we next examined the effect of this IL-15 blockade on peripheral lymphocyte population dynamics. As shown in Fig. 4A, compared to control mAb treatment, anti-IL-15 mAb administration resulted in a near complete loss of NK cells in the blood in the first 2 weeks of treatment. Importantly, the absolute numbers of total NK cells in anti-IL-15-treated RM were maintained at <5% of baseline through day 35. Since the cytotoxic CD16⁺ CD56⁻ NK cell subset constitutes the most abundant population in the peripheral blood of RM (55), it is not surprising that the post-treatment dynamics of this subset closely followed the overall NK cell population with absolute counts declining to <14% of baseline by day 7, and <3% of baseline at day 35. The depletion of the other NK subsets was substantial but less complete. The CD16⁻ CD56⁻ NK subset declined to ~10% of baseline by day 14, but recovered to <30% of baseline by day 35, despite ongoing therapy. Interestingly, the largely regulatory CD16⁻ CD56⁺ NK subset was least affected by IL-15 blockade, showing a maximal depletion to ~25% of baseline by day 28. We would note that CD16⁻ CD56⁺ NK cells were the only NK subset to respond to IL-7 signaling in our in vitro assay (Fig. 1C), suggesting the possibility that IL-7 may compensate for IL-15 signaling inhibition to partially maintain CD16⁻ CD56⁺ NK cell homeostasis (see below).

Figure 4.

IL-15 blockade induces profound depletion of NK cells in blood and tissues. (A) Quantification of absolute NK cell counts, including CD16⁺ CD56⁻, CD16 ⁻CD56⁺ and CD16⁻ CD56⁻ subsets in blood after anti-IL-15 mAb (n=17) or IgG control mAb (n=13) treatment. Significance was assessed as described in materials and methods. Results (mean + SEM) are shown as percentage of baseline. (B) Dot plots showing CD3⁻ NKG2A⁺ NK cells in peripheral blood, spleen, liver, lung, kidney and tracheobronchial lymph node (LN TB) of two representative RM that received either anti-IL-15 mAb or IgG control mAb prior to necropsy. (C) Lymphocytes isolated from the indicated tissues were analyzed to determine frequencies of total NK cells (CD3⁻, CD8α⁺, NKG2A⁺) within the overall small lymphocyte population in anti-IL-15 mAb-treated (n=9) or IgG control mAb-treated (n=5) RM. Histograms show mean frequencies of the NK cells as a fraction of total lymphocytes in each tissue, error bars show SEM.

To more comprehensively assess the extent of NK cell depletion after anti-IL-15 treatment we isolated cells from peripheral lymphoid and extra-lymphoid tissues at necropsy of the anti-IL-15-treated vs. control-treated RM, and determined the fraction of NK cells (CD3⁻ CD8α⁺ NKG2A⁺ lymphocytes) among total lymphocytes in each tissue. As shown in Fig. 4B–C, NK cells were dramatically reduced in anti-IL-15 mAb-treated RM compared with controls. Taken together, these data demonstrate that anti-IL-15 mAb administration can induce systemic depletion of NK cells in RM, with the implication that both circulating and tissue resident NK cells in RM are highly dependent on IL-15 signaling for their homeostatic stability.

IL-15 blockade results in selective perturbation of T_EM homeostasis

As expected based on the above-described differential sensitivity of T cells to IL-15 signaling according to differentiation status, the impact of IL-15 blockade on T cell population dynamics varied by differentiation defined subset. Anti-IL-15 mAb treatment had a negligible effect on circulating CD4⁺ and CD8⁺ T_N, as absolute counts and proliferative responses in anti-IL-15 mAb-treated RM were not significantly different from what was observed in control RM (Fig. 5A). A similar pattern was observed with the T_CM subsets, as IL-15 blockade had no significant effect on CD4⁺ or CD8⁺ T_CM absolute counts and CD8⁺ T_CM proliferation, and resulted in only marginal enhancement of CD4⁺ T_CM proliferation. However, IL-15 blockade had a profound effect on T_EM dynamics, with anti-IL-15 administration resulting in a rapid reduction in CD4⁺ and CD8⁺ T_EM absolute counts to ~30% of baseline by day 14 post-treatment (CD4⁺: p=0.0006, CD8⁺: p<0.0001). Strikingly, starting about day 14, this depletion was countered by a dramatic increase in the fraction of proliferating T_TrM and T_EM, such that absolute CD4⁺ and CD8⁺ T_EM counts in blood were restored to pretreatment (CD4⁺), or near-pretreatment levels (CD8⁺) by day 35. While anti-IL-15 treatment significantly increased the proliferative fraction of CD4⁺ and CD8⁺ T_TrM (CD4⁺: p<0.0001, CD8⁺: p<0.0001), absolute T_TrM counts remained unchanged, possibly reflecting a proportion of T_TrM differentiating to T_EM. Indeed, we would note that the relative proportions of all memory T cell subsets in the blood prior to anti-IL-15 administration returned to baseline by day 35 post-treatment (Fig. 5B). Taken together, these data demonstrate a key, specific role for IL-15 signaling for T_EM homeostasis, but also indicate that a significant proportion of T_EM, both CD4⁺ and CD8⁺, can persist in the presence of IL-15 signaling inhibition (similar to the CD16⁻ CD56⁺ NK cell subset), suggesting other regulatory molecules could play a role in modulating peripheral T_EM regeneration and stability in the absence of IL-15.

Figure 5.

Comparison of T cell dynamics in blood of anti-IL-15-treated RM. (A) Absolute counts and proliferative fraction of CD4⁺ and CD8⁺ T_M, including the T_CM, T_TrM, and T_EM subsets in blood of RM treated with anti-IL-15 mAb (n=17) or IgG control mAb (n=13). Results (mean + SEM) are shown as percentage of baseline, or for percentage of Ki-67+, change (Δ) from baseline with significant p-values shown as: *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001. Significance of difference between treatment groups was assessed as described in materials and methods. (B) Pie charts showing the relative proportions of CD4⁺ and CD8⁺ T_CM, T_TrM and T_EM subsets as fraction of total memory T cells in the blood of anti-IL-15-treated RM (n=16) prior to treatment, and days 14 and 35 post-anti-IL-15 mAb treatment.

We next determined whether the robust increases in CD4⁺ and CD8⁺ T_EM and T_TrM proliferation observed in response to anti-IL15 mAb treatment was confined to cells in the blood or a reflection of cells proliferating in tissues prior to entering into peripheral circulation. To address this question we used in vivo BrdU labeling to label all cells in S phase around the time of peak proliferation (see Fig. 3A). BrdU incorporation offers a more accurate indication of the location and timing of cell proliferation (specifically delineating S-phase of the mitotic cycle) than Ki-67 expression, which is maintained for several days following cell division during which time cell migration might occur (50). We administered three doses of BrdU within 24 hours prior to necropsy to label cells synthesizing DNA in this 24-hour period. We observed that the frequencies of BrdU-labeled CD4⁺ and CD8⁺ T_CM in all tissues were unchanged in anti-IL-15 mAb-treated RM compared to controls (Fig. 6A), substantiating our previous conclusion that anti-IL-15 treatment has negligible effects on steady state T_CM homeostasis (Fig. 5A). There was also no difference between control mAb- and anti-IL-15-treated RM in the frequencies of BrdU-labeled CD8⁺ T_TrM, although we did observe significant differences in the %BrdU⁺ CD4⁺ T_TrM in the blood (p=0.018), spleen (p=0.044), lung (p=0.024) and liver (p=0.018) with anti-IL-15 mAb treatment. Strikingly, however, the T_EM compartment saw the most dramatic increases in cell turnover in the presence of IL-15 signaling inhibition, which was apparent for both the CD4⁺ and CD8⁺ T cell lineages. Anti-IL-15-treated RM showed increased frequencies of BrdU⁺ CD8⁺ T_EM within most peripheral LN (iliosacral: p=0.024 and tracheobronchial: p=0.018) as well as within the lung (p=0.045), whereas frequencies of BrdU⁺ CD4⁺ T_EM were significantly higher in blood (p=0.044), lung (p=0.024), kidney (p=0.045) and liver (p=0.018) of anti-IL-15-treated animals. Interestingly, the frequencies of BrdU-labeled CD4⁺ and CD8⁺ T_EM in the intestinal lamina propria (including the colon, ileum, jejunum) and vaginal mucosa were not significantly different between anti-IL-15 mAb- and IgG control mAb-treated RM.

Figure 6.

IL-15 blockade increases T_EM turnover in vivo. (A) RM treated with anti-IL-15 mAb (n=7) or IgG1 control mAb (n=5) were intravenously administered three doses (30 mg/Kg) of BrdU 24 hours prior to necropsy. Lymphocytes were isolated from the indicated tissues and further analyzed for T cell markers and BrdU positivity. Results (mean + SEM) are shown as the percentage of BrdU⁺ cells (CD4⁺ or CD8⁺ T_CM, T_TrM and T_EM) determined for each tissue. Significance was evaluated using Mann-Whitney U test (*p<0.05). (B–C) Quantification of the T cell population within the colon lamina propria (LP) as determined by (B) the percent area of the LP that stained CD3⁺ and (C) extent of apoptosis measured by caspase-3 positive within the colon and (D) lymph node (LN) via immunostaining in RM following anti-IL-15 mAb or IgG control mAb treatment. The significance in these parameters between groups was evaluated with Mann-Whitney U test. (E–F) Images show active caspase-3 expression (red, original magnification 200X) in the colon (E) and LN (F) of three representative RM treated with either the IgG control mAb (top images) or anti-IL-15 mAb (bottom images).

Collectively, this data suggests that in the presence of IL-15 blockade, a significant proportion of T_EM initially undergo apoptosis due to abrupt withdrawal of IL-15 signaling, but subsequently are replaced by T_EM cells generated by a homeostatic proliferative response that is IL-15-independent (e.g., occurs despite ongoing IL-15 blockade). Although this homeostatic response returns the numbers of circulating T_EM back to baseline levels, it does not appear to restore the T_EM compartment to normal as we observed a significant reduction in CD3⁺ T cell density in the colonic lamina propria (p<0.0001) in anti-IL-15 mAb-treated RM compared with control RM (Fig. 6B). In should be noted that the blood T_EM compartment is very small compared to the tissue effector site T_EM compartment represented by the colonic lamina propria, and thus the latter is a much more accurate indicator of the status of the overall T_EM population than the former (which can be “reconstituted” by relatively few cells). In addition, we observed a significant increase in caspase-3 expression within lymphocytes of both colonic lamina propria (p=0.0087) and LN T cell zones (p = 0.02) (Fig. 6C–F), suggesting increased and sustained levels of T cell apoptosis. Thus, the compensatory mechanism that restores circulating T_EM numbers in the absence of IL-15 results in a high turnover state that is unable to completely reestablish normal T_EM homeostasis.

IL-15 blockade disrupts pathways that regulate T cell activation, cell cycle entry and survival

To further characterize the impact of IL-15 signaling inhibition on T_EM homeostasis, we performed transcriptomic profiling on sort-purified CD8⁺ T_EM obtained at the approximate peak of T_EM proliferation in the blood. These studies demonstrated that CD8⁺ T_EM from anti-IL-15 mAb-treated RM showed significant differences with control IgG mAb-treated RM with respect to gene expression in several signal transduction pathways and transcriptional nodes that regulate proximal T cell activation, cell cycle entry and cell survival/cell death (Fig. 7A–B). With regards to T cell activation, components of TCR signaling (CD3D, CD3E), activation markers such as MHC molecules, and transcription factors downstream of TCR signaling (REL, CIITA) (69), were all up-regulated in CD8⁺ T_EM from anti-IL-15 mAb-treated RM (Fig. 7A). Genes associated with cell cycle progression (ATM, E2F2, PIK3CB, PIK3CG, PIK3R4, PIK3C2A) were also up-regulated in these cells, while the cyclin-dependent kinase inhibitor, CDKN2B/p15, which inhibits cell cycle G1 progression, was down-regulated (Fig. 7B). In addition, genes associated with the mitochondrial respiration machinery (MRM) pathway, including NDUFV1, NDUFA2, UQCR1, UQCRB and ATP5C1, were significantly up-regulated in CD8⁺ T_EM from the anti-IL-15 mAb-treated RM (Fig. 7C). This was further supported by the up-regulation of genes for enzymes with antioxidant activity, such as PRDX3 and GPX4, which protect cells from death by their effect on lipid peroxidation (70). All of these changes in T_EM from anti-IL-15 mAb-treated RM are consistent with their increased proliferation observed in vivo (Fig. 5A). However, we also noted concomitant changes in gene pathways associated with T cell survival, in particular the upregulation of cells that promote cell death, including Bax, Fas, and Casp6 (p53 pathway) and Cabin1 and Mef2 (Nur77 signaling pathway) (71) (Supplementary Table 1). In addition, disruption of cell survival mechanisms was further shown by the down-regulation of BCL-2 (Fig. 7B). Collectively, these observations suggest that IL-15 signaling inhibition results in induction of T_EM activation and proliferation, but at the same time, results in transcriptional changes that may limit the long-term T_EM survival.

Figure 7.

Transcriptomic profiling on sort-purified CD8⁺ T_EM and CD8⁺ T_CM obtained from anti-IL-15 mAb-treated RM and compared with samples taken from RM treated with the IgG control mAb. GSEA was performed to identify pathways that are enriched at a p-value <0.05 and for grouping pathways into functional modules. The genes represented in the heatmaps are enriched in at least 50% of the gene sets of the module (if the number of gene sets in the module is less than or equal to 4), and genes enriched in at least 25% of the gene sets of the module (if the number of gene sets in the module is greater than 4). The color scale of the pathways represent the −log10(p-value) of the pathway (A, B, and D). (A) Module 1 enriched in CD8⁺ T_EM is representative of T cell activation. (B) Module 2 enriched in CD8⁺ T_EM is representative of cell cycle progression. (C) Representation of the mitochondrial respiration machinery pathway in CD8⁺ T_EM up-regulated in the anti-IL-15 mAb-treated RM compared to IgG control mAb-treated RM. (D) Module enriched in CD8⁺ T_CM is representative of up-regulation of pro-apoptotic signaling and down-regulation of Type I Interferon activity with blockade of IL-15.

Transcriptional profiling of CD8⁺ T_CM also revealed differences in gene expression pathways between anti-IL-15 mAb-treated and IgG control mAb-treated RM (Fig. 7D); however, these differences involved pathways and genes distinct from those observed in T_EM. In addition, the number of genes differentially expressed between anti-IL-15 mAb-treated RM vs. IgG- control mAb-treated RM in CD8⁺ T_EM (2808 genes) was significantly greater than those differentially expressed in the CD8⁺ T_CM (2225 genes) (p-value < 2.2e-16, using a Pearson’s chi-squared test for equality of proportions). Strikingly, IL-15 blockade did not induce changes in the expression of genes associated with the MRM pathway in T_CM, suggesting no disruption to their steady state mitochondrial activity. However, as shown by the module enrichment in T_CM, we did observe an up-regulation of pro-apoptotic genes (BID, APAF1, CASP8, CASP3) in this cell type in anti-IL-15 mAb-treated RM, possibly reflecting an increase in cell death in a fraction of T_CM induced to differentiate in the absence of IL-15 signaling. Interestingly, we also observed the enrichment of the S1P1 pathway in T_CM that regulates lymphocyte egress from the lymph node (72) (Supplemental Table 1). However, these differences did not translate to significant changes in T_CM population dynamics in vivo, as absolute counts and proliferative fractions of CD8⁺ T_CM were statistically indistinguishable between anti-IL-15 mAb-treated RMs and controls (Fig. 5A).

T_EM become increasingly sensitive to IL-7 in response to IL-15 blockade

Collectively, gene expression profiling demonstrates the dramatic effect IL-15 blockade has on peripheral T_EM, in comparison to the T_CM. Somewhat paradoxically, in the absence of IL-15 signaling, circulating CD8⁺ T_EM predominantly display the induction of genes involved with T cell activation, cell cycle progression and mitochondrial activity, raising the possibility that in the periphery, T_EM are responding to another pro-proliferative cytokine. In this regard, we asked whether IL-7 might be supporting T_EM homeostasis in the absence of IL-15. IL-7 is the most likely candidate to act as a compensatory cytokine for IL-15 because of its bioavailability in the blood and LN (73–75). Notably, IL-15 blockade did not change plasma levels of IL-7 (Fig. 8A) suggesting that anti-IL-15 mAb treatment did not increase overall production of IL-7. We then determined whether T_EM proliferating in response to anti-IL-15 mAb treatment manifest increased sensitivity to IL-7. To address this question, we performed ex vivo pSTAT induction analysis on whole blood at various time points post-anti-IL-15 mAb treatment with 4ng of IL-7 (this dose was selected because it reflected the minimum dose that elicited a maximal response in most T cell subsets in our in vitro assay; Fig. 1B). Interestingly, we observed a significant increase in pSTAT5 expression in response to this dose of IL-7 in both CD4⁺ and CD8⁺ T_EM from anti-IL-15 mAb-treated RM compared with controls (CD4⁺: p=0.0073, CD8⁺: p=0.0016) (Fig. 8B). Importantly, this increase in IL-7 responsiveness was only observed in the T_EM subset, as the Δ pSTAT5 expression in T_N, T_CM or T_TrM after IL-7 stimulation was statistically indistinguishable between both groups.

Figure 8.

T_EM are more responsive to IL-7 in the presence of IL-15 signaling inhibition. (A) Plasma collected at indicated time points post-anti-IL-15 mAb (n=6) or IgG control mAb (n=4) treatment was analyzed for IL-7 levels by ELISA. Results (mean + SEM) are shown as percent change from baseline in IL-7 concentration (NS = not significant). (B) Showing change in pSTAT5 expression in peripheral blood T_N, T_CM, T_TrM and T_EM, after ex vivo stimulation with 4ng of IL-7 in RM treated with anti-IL-15 mAb (n=9) or IgG control mAb (n=5). Significance of difference in all parameters was assessed as described in materials and methods (*p<0.05).

To determine whether anti-IL-15 treatment increases IL-7 responsiveness in T_EM in vivo, we designed an in vivo experiment in which anti-IL-15 mAb-treated or IgG control mAb-treated RM were treated with recombinant simian IL-7 (rsIL-7) (11, 62). RM received three doses of anti-IL-15 or IgG control mAb bi-weekly followed by two doses of rsIL-7 (30μg/Kg) one week apart and then a further three doses of anti-IL-15. To ensure a steady state of IL-15 neutralization at the time of IL-7 dosing, rsIL-7 was administered between the 3^rd and 4^th doses of anti-IL-15. As shown in Fig. 9A, rsIL-7 induced a 2-fold increase in the absolute numbers of CD4⁺ and CD8⁺ T_CM in the blood, which was similar between anti-IL-15 mAb-treated RM and controls, further supporting our previous observations that IL-15 blockade has negligible effects on steady state T_CM homeostasis (Fig. 5A and 6A). In contrast, there was a greater increase in both CD4⁺ and CD8⁺ T_EM absolute counts in the blood of anti-IL-15 mAb-treated RM in response to IL-7, which was significant on days 10 (p=0.016), 12 (p=0.003) and 14 (p<0.001) for CD4⁺ and on days 10 (p=0.002) and 12 (p=0.0069) for CD8⁺. This increase in absolute T_EM counts peaked by day 14, before declining to levels similar to controls by day 28. These data strongly support the concept that T_EM become more responsive to IL-7 in vivo in the absence of IL-15 signaling. This increased sensitivity to IL-7 could help support the rapid T_EM proliferative expansion observed after anti-IL-15 treatment. Consistent with this interpretation, the transcriptional analysis of peripheral blood CD8⁺ T_EM described above also showed an up-regulation of genes associated with the JAK/STAT pathway in anti-IL-15-treated RM, highlighted by the up-regulation of STAT1 and SOCS3 (known to be downstream of IL-7 and other cytokines), suggesting engagement of this pathway (Fig. 9B). Altogether these data support the involvement of secondary γc receptor cytokine signaling in supporting T_EM homeostasis in the presence of IL-15 signaling inhibition.

Figure 9.

The T_EM response to IL-7 in vivo is increased with anti-IL-15 treatment. (A) RM were administered anti-IL-15 mAb (n=6) or IgG control mAb (n=5) biweekly over 8 weeks for a total of 5 doses. On days 35 and 42 post-treatment, both groups received 2 subcutaneous injections of IL-7 at 30ug/Kg. Results (mean + SEM) are shown as percentage of baseline absolute counts of the indicated T cell subsets. The significance in parameters between RM groups was evaluated using rANOVA with Tukey-Kramer adjustment (significant p-values shown as: *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001). (B) Transcriptomic profiling of sort purified CD8⁺ T_EM showing genes associated with the JAK/STAT pathway (source gene set from MSigDB) at p<0.05 in anti-IL-15 mAb-treated RM compared with IgG control mAb-treated RM.

DISCUSSION

IL-2, IL-7 and IL-15 share a common γc receptor and mediate most of their immunoregulatory function through activation of the JAK/STAT signaling pathway. Because of these common features, the downstream activities of these cytokines often overlap, but nevertheless, each has its own biologic role. In this study, we sought to characterize the distinct role IL-15 plays in maintaining lymphocyte population dynamics in NHP. To achieve this we developed an RM model of in vivo IL-15 blockade that involved repeated administration of a rhesusized anti-IL-15 mAb. We hypothesized that blocking IL-15 signaling would result in substantial changes in lymphocyte population dynamics specific to IL-15 function. Similar to previous reports in humans and cynomolgus macaques, anti-IL-15 was well tolerated, with no adverse side effects observed with repeated dosing. We directly demonstrated that anti-IL-15 administration was effective at inhibiting IL-15 signaling in vivo.

A major consequence of IL-15 blockade was the systemic, nearly complete depletion of NK cells, demonstrating the essential role of IL-15 in the development and survival of most NK cells in RM. A recent report by Lebrec et al., suggests that IL-15 may not be required for NK cell homeostasis in humans, as a blocking monoclonal antibody directed against human IL-15 did not decrease circulating NK cell counts (66). While it is unclear why anti-IL-15 administration did not affect NK cell homeostasis in that study, our data confirms previous observations in mice and NHP that IL-15 blockade significantly diminishes NK cell populations in vivo (23, 34, 76–79). In our study, the CD16⁺ CD56⁻ NK cell subset, which constitutes the major NK population in the blood of RM was severely depleted by IL-15 blockade. In contrast, the minor NK cell subset (CD16⁻ CD56⁺) was less sensitive to IL-15 signaling inhibition. Considering CD16⁻ CD56⁺ NK cells can induce pSTAT5 expression in response to both IL-7 and IL-15 in vitro, whereas CD16⁺ CD56⁻ NK cells respond only to IL-15 (Fig. 1C), it is possible that this less efficient depletion of the CD16⁻ CD56⁺ NK cell subset is due to the ability of IL-7 to compensate for IL-15 signaling inhibition, partially supporting their homeostatic maintenance.

The impact of IL-15 blockade on both CD4⁺ and CD8⁺ T_EM was also both dramatic and specific, revealing a unique aspect of T_EM homeostasis. In the absence of IL-15 signaling, absolute CD4⁺ and CD8⁺ T_EM (but not T_N, T_CM or T_TrM) counts in blood rapidly declined to less than 70% of baseline levels. However, starting between 7 and 14 days after treatment initiation, this initial T_EM decline was countered by a burst of sustained proliferation in both T_EM and T_TrM that corresponded with an increase in T_EM counts in blood back to baseline or near baseline levels. Although some anti-γc cytokine antibodies that display blocking effects in vitro sometimes become activating (pro-proliferative) when administered in vivo (80–85), we feel that this mechanism is unlikely in our study as the effects of anti-IL-15 included (1) the ex vivo inhibition of IL-15 signaling as measured by the induction of pSTAT5 expression, (2) the systemic depletion of NK cells and (3) the selective effects of anti-IL-15 mAb treatment on T_TrM and T_EM, but not T_N or T_CM populations. However, we did experimentally address the issue of whether T_EM were indirectly induced to proliferate in response to IL-15/anti-immune complexes via an Fc-dependent mechanism. To test this we treated 2 RM with a variant of the rhesus anti-IL-15 mAb that had an inactivating LALA mutation in the Fc receptor-binding site (86). We found that this Fc inactivation had no effect on the in vivo activity of the anti-IL-15 mAb on T cell and NK cell dynamics (data not shown), and we thus concluded that the T_EM proliferative response following anti-IL-15 mAb treatment was most likely not an immune complex mediated artifact, but rather a compensatory, IL-15-independent homeostatic process.

This conclusion is also supported by the observations that T cell densities in colonic lamina propria [an extra-lymphoid effector site with predominant populations of resident memory T cells with T_EM or T_TrM phenotypes; (48, 87)] remained significantly depleted well after normalization of absolute T_EM counts in blood, and that high T_EM proliferation was maintained for the duration of treatment (7 weeks), associated with increased levels of active caspase-3, consistent with apoptosis and a high turnover state (88), which wad widely (although not homogeneously) distributed in peripheral lymphoid and extra-lymphoid tissues. Transcriptional profiling of CD8⁺ T_EM in the blood confirmed a predominant up-regulation of genes associated with T cell activation and cell cycle progression in anti-IL-15 mAb-treated RM. In particular, genes involved with mitochondrial respiration were up-regulated in CD8⁺ T_EM, indicating that in the periphery, T_EM can receive secondary signals that support proliferative expansion and an active metabolic state in the absence of IL-15. The cell substrate upon which this homeostatic mechanism depends remains an open question with 3 non-mutually exclusive possibilities. First, the homeostatic mechanism may simply operate on T_EM cells that are normally maintained by IL-15, but can respond to other cytokines in the absence of IL-15. Although, in this scenario, we would posit the availability or activity of these putative non-IL-15 cytokines is suboptimal, accounting for the incomplete T_EM reconstitution and failure to achieve (or approach) quiescence during the 7 weeks of observation. Second, since in mice there is clear heterogeneity in the IL-15 dependence of effector memory CD8⁺ T cell populations in various sites (89), it is possible the rebounding population in anti-IL-15 mAb-treated RM derives from the IL-15-independent subsets. In this scenario, the incomplete reconstitution observed in the IL-15-deficient RM might be due to an inability of these IL-15-independent subsets to effectively fill T_EM niches usually occupied by IL-15-dependent cells. Finally, it is possible that the T_EM regeneration that occurs in the absence of IL-15 derives from T_TrM or even T_CM precursors that are induced by non-IL-15 signals to both proliferate and differentiate towards T_EM. As with possibility #2, the observed incomplete T_EM reconstitution in this situation might be due to inability of these non-IL-15 regulators to induce full differentiation into the panoply of T_EM cells needed to fill all T_EM niches.

The next question to be addressed concerns the nature of homeostatic factor(s) responsible for stimulating the T_EM proliferative expansion in the presence of IL-15 signaling inhibition. In this regard, it is interesting that treatment of monkeys with the JAK3 inhibitor CP-690550, which would inhibit all γc cytokines (90) results in depletion of NK cells and CD8⁺ memory T cells (particular T_EM) similar to what we report here for anti-IL-15 mAb treatment; however, with JAK3 inhibition CD8⁺ T_M rebound occurs only after drug cessation and there is no evidence of the T_EM regenerative process described in this report (43). These differences suggest that the T_EM regenerative process observed with anti-IL-15 depends upon the JAK3 signaling, and therefore might involve a non-IL-15 member or members of the γc cytokine family.

Although expression of the IL-7 receptor CD127 is maintained at low levels on T_EM, the relative abundance of IL-7 in peripheral circulation as well as constitutive expression within lymphoid microenvironments suggests IL-7 as a primary candidate for this role. Initial analysis of plasma cytokine levels showed no changes in IL-7 or other immune-modulating cytokines such as TNFα, and IL-12 (data not shown). Although this indicated that anti-IL-15 mAb treatment did not increase IL-7 production, it did lead us to ask whether T_EM might acquire increased sensitivity to endogenous IL-7 in the absence of IL-15 signaling. In line with this, we observed a significant increase in the fraction of both CD4⁺ and CD8⁺ T_EM with pSTAT5 expression in response to IL-7 stimulation. Interestingly, this increase in IL-7 responsiveness was not observed in other T cell subsets (T_N, T_CM or T_TrM). This effect was further demonstrated in vivo by administering IL-7 to monkeys that were being treated with anti-IL-15 vs. control mAbs. Notably, absolute numbers of CD4⁺ and CD8⁺ T_EM, but not T_CM, were significantly increased in response to IL-7 in anti-IL-15 mAb-treated RM compared to controls. While it’s important to note that we did not examine whether T_EM were more responsive to other γc cytokines (i.e., IL-2), our data strongly supports the concept that T_EM generated in the absence of IL-15 signaling become increasingly more responsive to JAK/STAT signaling via at least one other γc cytokine (IL-7). Transcriptional profiling also supported this conclusion by revealing an increased expression of the transcription factor, STAT1 in T_EM. STAT1 is up-regulated during conditions of lymphopenia and plays a role in regulating the expansion and survival of CD8⁺ T cell responses, particularly after viral infection (91, 92). Since IL-7 can activate STAT1 and JAK1/STAT1 signaling has been linked with supporting T cell survival (93), it is possible that increases in STAT1 expression contributed to T_EM becoming more responsive to IL-7 signaling in animals treated with anti-IL-15. Other non-mutually exclusive possibilities are that in the absence of IL-15, expanding and differentiating T_TrM retain IL-7 sensitivity as they differentiate into T_EM and/or that IL-15-independent T_EM subsets have the ability to more efficiently respond to IL-7. Although elucidation of these mechanisms will require further study, the data presented here strongly support the concept that even when γc cytokines have primary roles in the homeostatic regulation of specific subsets, there is often (T_EM; CD56⁺ NK cells), but not always (CD16⁺ NK cells), plasticity in this regulation such that other members of this cytokine family can provide “back-up” support for population maintenance.

As previously discussed, compensatory γc signaling does not appear to be sufficient to normalize peripheral T_EM homeostasis, and this compensation is likely to become increasingly inadequate over the long-term. In this regard, IL-15 appears to be required for optimal development of T_EM and effector T cells to new Ags, and for the emigration of these cells to tissue effector sites (44, 45). In keeping with this, we have observed that anti-IL-15 mAb administration during acute SIV infection severely diminishes the magnitude of SIV-specific T cell responses in mucosal compartments (i.e., BAL) but not in the peripheral blood (Okoye et al., manuscript in preparation). Thus, inadequate tissue homing may play a role in the reduced T_EM in tissue effector sites such as the colonic lamina propria. The decrease in T cell density observed in the colonic lamina propria may also reflect the inability of T_EM resident within such effector sites, or T_TrM recently emigrating into these sites, to effectively compete with other lymphocyte populations (such as B cells, γδ T cells, intraepithelial lymphocytes, CD56⁺ NK cells) for IL-7 or other compensatory γc cytokines when IL-15 is not available. Such competition would decrease T_EM population stability in tissues where IL-7 and/or other compensatory γc cytokines are limiting. Indeed, the precise role that IL-15 plays in regulating the migration and subsequent maintenance of T_EM in such tissues still needs to be more clearly defined in primates.

In summary, this study unequivocally establishes a major in vivo role for IL-15 in the development and survival of NK cells in an NHP model. In addition, this study demonstrates the important role IL-15 plays in maintaining peripheral CD4⁺ and CD8⁺ T_EM homeostasis, and the complex interaction of IL-15 with other γc cytokines in the regulation of the T_EM compartment. IL-15 is overexpressed in several autoimmune disorders, where it is thought to drive the activation and proliferation of self-reactive memory T cells (94). Based on this, blockade of IL-15 activity is being evaluated as a therapeutic strategy for these diseases (95, 96). We would propose that in vivo IL-15 blockade of IL-15 in RM provides a useful model for exploring the implications of this therapeutic strategy. As shown here, anti-IL-15 therapy will likely not uniformly deplete T and NK cell effectors, but rather will change their balance, potentially resulting in unexpected, off-target immune reactions. These results suggest that further study will be necessary to fully characterize the therapeutic potential of anti-IL-15 mAbs in the treatment of autoimmune disorders.

Abbreviations used in this article

RM

rhesus macaque

T_N

naïve T cell

T_M

memory T cell

T_CM

central memory T cell

T_TrM

transitional memory T cell

T_EM

effector memory T cell

pSTAT5

phosphorylated STAT5

NHP

nonhuman primates

mAb

monoclonal antibody