IL-15 produced and trans-presented by DCs underlies homeostatic competition between CD8 and γδ T cells in vivo

Peripheral lymphocyte homeostasis is an actively regulated process during which the proportions of different lymphocyte subsets remain constant.¹  One potential mechanism to achieve such maintenance includes a competition for homeostatic resources between lymphocytes.^2,3  Homeostatic cytokines, interleukin-7 (IL-7) and IL-15, play dominant roles in regulating the size of peripheral lymphocytes by controlling proliferation and survival.^4,5  IL-7, produced mainly by stromal cells, plays a central role in generating lymphocytes in the primary lymphoid tissues.^6,7  In the periphery, IL-7 controls the size⁸  as well as the survival of T cells.^9,10  On the other hand, IL-15 plays a major role in inducing generation of natural killer (NK) cells, NKT cells, memory CD8 T cells, and γδ T cells.¹¹  Overexpression of IL-15 dramatically increases memory phenotype CD8 T cells in the lymphoid tissues,¹²  while mice deficient in IL-15 display defects in NKT cells, memory CD8 T cells, and subpopulations of intraepithelial lymphocytes.¹³  IL-15 also plays a role in limiting proliferation of certain T-cell subset under lymphopenic settings.^14,15  γδ T cells undergo homeostatic proliferation after the transfer into lymphopenic hosts, which is greatly diminished by CD8 T cells and γδ T cells themselves.^16,17  Competing IL-15 seems a likely mechanism underlying the competition as anti–IL-15Rβ blocking antibody (Ab) treatment significantly reduces the degree of proliferation while anti–IL-2Rα/IL-2 Ab–mediated inhibition is only marginal.¹⁶ 

Unlike IL-7, which is delivered as a soluble form that binds to the IL-7 receptor complex on responding cells, IL-15 has drawn much attention by its unique mechanism of trans-presentation.¹⁸  IL-15 made mostly by activated monocytes and dendritic cells (DCs) is preassembled with the IL-15Rα subunit within the endoplasmic reticulum (ER)/Golgi compartment and transported to the cell membrane,¹⁹  where membrane bound IL-15/IL-15Rα complexes are trans-presented to responding cells that express dimeric IL-15Rβ/γ_c complex.²⁰  The trans-presentation mechanism of IL-15 has been shown to be critical during CD8 memory T-cell survival,²¹  NK cell development, as well as NK cell activation.¹⁹  Whether IL-15–dependent homeostatic competition between T-cell subsets is operated by similar mechanism remains obscure.

Using 2 widely used lymphopenic models: Rag1^−/− and TCRβ^−/− mice, we revisited naive CD4 and CD8 T cells responses to lymphopenia associated homeostatic signals and found that the expansion (and probably the survival) but not the initial proliferation of the transferred T-cell subsets greatly differs depending on the types of recipients. Both naive CD4 and CD8 T cells equally expanded in γδ T cell–deficient Rag1^−/− recipients. By contrast, CD8 T-cell expansion was severely compromised in γδ T-cell–bearing TCRβ^−/− recipients. The presence of γδ T cells in TCRβ^−/− recipients was responsible for the diminished CD8 T-cell expansion as eliminating γδ T cells restored the expansion. γδ T-cell–dependent inhibition of CD8 T-cell expansion was correlated with IL-15 trans-presented on the membrane of CD11c⁺ DCs. Importantly, DC production of IL-15 was critical to enhance CD8 T-cell expansion. Collectively, these results provide in vivo evidence that IL-15 production and trans-presentation by DCs controls the expansion and the survival of different T-cell subsets, particularly γδ and CD8 T cells.

Thy1.1 C57BL/6, Ly5.1 C57BL/6, B6 Rag1^−/−, B6 TCRβ/δ^−/−, and B6 TCRβ^−/− mice were purchased from The Jackson Laboratory (Bar Harbor, ME). B6 IL-15^−/− and wild-type B6 mice were purchased from Taconic Farms (Germantown, NY). All experimental procedures were conducted according to the guidelines of the Cleveland Clinic Foundation Institutional Animal Care and Use Committee.

T cells were purified as previously reported.²²  Lymph node (LN) cells were incubated with fluorescein isothiocyanate (FITC)–labeled antibodies (anti-B220, anti–I-A^b, anti-CD16/32, and anti-NK1.1), followed by anti-FITC–conjugated microbeads (Miltenyi Biotec, Auburn, CA). Magnetic separation was performed using an LS column (Miltenyi Biotec). Purity of T cells was typically greater than 98%. Naive (CD44^low) T cells were further sorted using a FACSAria (Becton Dickinson, San Jose, CA). In some experiment, naive T cells were labeled with carboxyfluorescein succinimidyl ester (CFSE; Molecular Probes, Invitrogen, Carlsbad, CA) and transferred intravenously into recipient mice (10⁶ cells per recipient). For γδ T-cell sorting, spleen, LN, and mesenteric LN cells from TCRβ^−/− mice were combined, incubated with Fc blocker, and stained for γδTCR (anti-γδ TCR, clone GL3). γδ T cells (3 × 10⁶) were adoptively transferred intravenously into Rag1^−/− recipients. All the antibodies were purchased from eBioscience (San Diego, CA).

In some experiments that require γδ T-cell depletion, TCRβ^−/− mice were injected intraperitoneally with 500 μg anti-Thy1.2 Ab (clone 53-2.1). The injection was performed at days −2, 1, and 3 days of Thy1.1 naive T-cell transfer.

A miniosmotic pump (Durect, Cupertino, CA) containing 5 μg murine IL-15 (Peprotech, Rocky Hill, NJ) was used to deliver the cytokine in vivo.⁸  Mice were killed 14 days after implantation.

Spleen, LN, and mesenteric LN cells were separately harvested at the indicated times after transfer and analyzed. The following antibodies were used: biotinylated anti-Ly5.1 (A20), anti–mouse IL-15Rα (BAF551; R&D Systems, Minneapolis, MN), anti–mouse IL-15 (BAF447, R&D Systems), phycoerythrin (PE)–anti-CD11c (N418), anti-CD90.1 (HIS51), anti-CD44 (IM7), anti-CD19 (1D3), strepatvidin-PE, allophycocyanin (APC)–anti-CD4 (RM4-5), anti-CD8 (53-6.7), FITC–anti-γδ TCR (UC7-13D5), and FITC–anti-bcl2 (3F11) Abs. Antibodies were purchased from eBioscience, PharMingen (BD Biosciences, San Diego, CA), and R&D Systems. Flow cytometry was performed on a FACSCalibur, and data were analyzed with a FlowJo software (TreeStar, Ashland, OR). In some experiments, 1 mg BrdU (Sigma-Aldrich, St Louis, MO) was injected intraperitoneally to measure in vivo proliferation. The mice were killed 6 hours after the injection, and BrdU incorporation was determined by fluorescence-activated cell sorting (FACS) analysis using a FITC-labeled anti-BrdU Ab (PharMingen). For experiments determining membrane bound IL-15, harvested tissue cells were immediately stained for CD11c and IL-15. IL-15 staining was validated after washing cells with acidic phosphate-buffered saline (PBS), pH 3.5, to strip cell-bound IL-15 as previously reported.²¹ 

B-cell depleting 18B12 (anti-mCD20) and isotype matched control 2B8 Abs were injected intravenously at 250 μg per mouse 7 days before T-cell transfer. CFSE-labeled 10⁶ naive T cells were then transferred into Ab-treated mice. For B-cell cotransfer, splenic B cells were purified by MACS (Miltenyi Biotec, Bergisch Gladbach, Germany). Purity of B cells was greater than 98%. Purified 5 × 10⁶ B cells were cotransferred with CFSE-labeled Thy1.1 naive T cells into Rag1^−/− recipients.

Bone marrow–derived DCs (BMDCs) were generated by culturing BM cells from C57BL/6 mice or IL-15^−/− mice with granulocyte macrophage colony-stimulating factor (GM-CSF) and IL-4. Purity of BMDC was determined by CD11c expression (> 85% purity). BMDCs (5 × 10⁶) were cotransferred with naive T cells in the indicated experiments.

Statistical significance was determined by the Student t test using the SigmaPlot 9.0 (SPSS, Chicago, IL). A P value less than .05 was considered to indicate a significant difference.

After transfer into lymphopenic hosts, naive T cells undergo proliferation.²²  To examine whether types of lymphopenia influence T-cell responses, purified naive T cells were adoptively transferred into either Rag1^−/− or TCRβ^−/− recipients, widely used T-cell-deficient lymphopenic models. Groups of Rag1^−/− and TCRβ^−/− mice received 10⁶ FACS-sorted, CFSE-labeled naive Ly5.1 T cells, and the proliferation within the peripheral LN (axillary and cervical LN), the mesenteric LN, and the spleen was separately examined 7 days after transfer. Transferring sorted CD44^low total T cells, which typically yielded greater than 99% pure naive T cells of 50% to approximately 55% CD4 and 45% to approximately 50% CD8 T cells (Figure S1, available on the Blood website; see the Supplemental Materials link at the top of the online article), allowed us to measure responses of both subsets within the same recipients. Both naive CD4 and CD8 T cells underwent robust proliferation and fully diluted the CFSE within 7 days of transfer (Figure 1A). Although no significant difference in CFSE profiles was found between Rag1^−/− and TCRβ^−/− recipients, the recovery of donor T cells was markedly different in these recipients. As demonstrated in Figure 1B, CD4 and CD8 T cells were similarly recovered in all tested tissues of Rag1^−/− recipients. The ratio of CD4/CD8 T cells based on the recovery was calculated at approximately 2 (Figure 1C). By contrast, in TCRβ^−/− recipients CD4 T-cell recovery dramatically increased, while CD8 T-cell recovery was reduced (Figure 1B). Because the CFSE dilution was comparable (Figure 1A), the poor recovery of CD8 T cells within the TCRβ^−/− recipients was not the result of their inability to proliferate. The ratio of CD4/CD8 T cells recovered from the indicated tissues was calculated at approximately 9, approximately 4, and approximately 15 in the spleen, the peripheral LN, and the mesenteric LN, respectively (Figure 1C). These results indicate that although the proliferation may equally be induced, the capacity of proliferating T cells to expand and probably to survive greatly differs depending on the types of lymphopenia, suggesting that different mechanisms might be involved.

We next examined the T-cell levels every week after the transfer. CD8 and CD4 T-cell recovery was comparable in Rag1^−/− recipients; the ratio of CD4/CD8 T cells remains at approximately 1 over the next 4 weeks of transfer (Figures 2A,B, S2). By contrast, CD4 T cells continue to outnumber CD8 T cells in all tissues of TCRβ^−/− recipients including blood (Figures 2A, S2). The ratio of CD4/CD8 T cells in TCRβ^−/− recipients remained at approximately 10 up to 4 weeks after transfer (Figure 2B). Differences in cell recovery were further supported by proliferation kinetics of T-cell subsets measured by in vivo BrdU incorporation. Early proliferation measured before 7 days of transfer (2, 3, 4, 5, and 7 days in TCRβ^−/−, and 4, 5, and 7 days in Rag1^−/− mice) demonstrates that both CD4 and CD8 T cells proliferate similarly in Rag1^−/− mice, but that CD8 T-cell proliferation was severely impaired in TCRβ^−/− mice (Figure 2C). Therefore, the differences in early proliferation kinetics of CD4 and CD8 T cells in TCRβ^−/− recipients appear to result in dissimilar expansion. To our surprise, we observed that CD8 T-cell recovery in the spleen of TCRβ^−/− recipients was significantly greater than that in the spleen of Rag1^−/− recipients, particularly at the very early time point (4 days after transfer; Figure 2C). The differences in cell recovery were also noticed in other tissues to a lesser extent in other lymphoid tissues. The analysis was further extended to the next 4 weeks of transfer (Figure 2D). While CD8 T-cell proliferation was relatively comparable to that of CD4 T cells in Rag1^−/− mice, CD8 T-cell proliferation was substantially diminished particularly in the mesenteric LN of TCRβ^−/− recipients up to 2 weeks after transfer (Figure 2D). Therefore, the dissimilar expansion behavior of T-cell subsets appears to be determined during early time after the transfer. Overall, these results strongly suggest that a homeostatic mechanism(s) that controls CD4 and CD8 T-cell expansion is different in Rag1^−/− and TCRβ^−/− recipients.

Earlier studies demonstrated that in the absence of either CD4 or CD8 T cells, the expansion of the remaining T-cell subset increases and compensates the deficiency, thus the total T-cell number remains unchanged compared with that of normal mice.^23,24  Therefore, CD4 and CD8 T cells might compete each other for homeostatic needs. If the differential expansion of CD4 and CD8 T cells in TCRβ^−/− recipients is the result of such competition, CD8 T cells transferred without CD4 T cells are expected to expand to the level of CD4 T cells in TCRβ^−/− recipients. This possibility was tested by transferring 0.5 × 10⁶ naive Ly5.1 CD4 or CD8 T cells into Rag1^−/− and TCRβ^−/− recipients. Both CD4 and CD8 T cells fully diluted the CFSE in these recipients, whether they were cotransferred or separately transferred (data not shown). When the total cell recovery was determined 7 days after transfer, increased CD4 T-cell recovery was still observed in TCRβ^−/− recipients regardless of the transfer method (Figure 3). CD8 T-cell recovery was still diminished in TCRβ^−/− recipients even after transfer without CD4 T cells. Relative recovery of T-cell subsets was comparable in Rag1^−/− recipients when they were cotransferred or separately transferred. Interestingly, the absolute numbers of T cells recovered from Rag1^−/− recipients were higher when they were cotransferred than when each subset was separately transferred (Figure 3). A mechanism underlying the discrepancy made by the cell transfer mode is unclear and remains to be investigated. Overall, these results suggest that the impaired CD8 T-cell expansion seen in TCRβ^−/− recipients is not due to a homeostatic competition between the T-cell subsets, and there appears to be an independent mechanism leading to the unequal expansion of T-cell subsets seen in TCRβ^−/− but not in Rag1^−/− recipients.

As obvious difference between Rag1^−/− and TCRβ^−/− mice is B cells, we next tested if B cells in TCRβ^−/− recipients are responsible for the differential expansion of T-cell subsets. TCRβ^−/− recipients were injected with B cell–depleting anti–mouse CD20 Ab, 18B12²⁵  or isotype control Ab, 2B8. Figure 4A shows that B-cell depletion by 18B12 Ab injection is efficient (> 95%). Naive T cells were then transferred into these recipients. CFSE dilution demonstrated that B-cell depletion in TCRβ^−/− mice did not affect the proliferation (Figure 4D). However, CD8 T-cell recovery still remained low compared with CD4 T cells in the absence of B cells (Figure 4B). The ratio of CD4/CD8 T cells was similar between untreated, isotype Ab treated, and B cell–depleted TCRβ^−/− recipients (Figure 4C). Interestingly, the overall recovery of T cells seemed slightly reduced when B cells were depleted. Consistent with this finding, cotransfer of splenic B cells into Rag1^−/− recipients did not alter T-cell proliferation (data not shown) and the CD4/CD8 T-cell ratio remained the same by the presence of exogenous B cells (Figure 4E). It should be noted, however, that TCRβ^−/− recipients acutely depleted of B cells could be different from Rag1^−/− mice that permanently lack B cells (ie, lymphoid architectures) and that the different lymphoid structures might still affect the homeostatic behavior of transferred T cells. Nevertheless, the data strongly suggest that B cells play little or no role in the proliferation and the homeostatic maintenance of CD4 and CD8 T cells under lymphopenic conditions.

It was previously reported that γδ T-cell homeostatic proliferation within sublethally irradiated animals is controlled by both γδ and CD8 T cells via IL-7/IL-15–dependent mechanism.¹⁷  Thus, γδ T cells may control the size of proliferating CD8 T cells in TCRβ^−/− mice. γδ T cells were abundant in TCRβ^−/− recipients, ranging from 4% to approximately 15% depending on lymphoid tissues (Figure S3). The unusual high numbers of γδ T cells in TCRβ^−/− compared with wild-type mice seem to reflect the absence of αβ T cells. To test whether γδ T cells control CD8 T-cell expansion, we depleted γδ T cells by injecting anti-Thy1.2 Ab. The depletion by Ab injection was highly efficient (∼ 100% depletion; Figure S3). Naive Thy1.1 T cells were subsequently transferred into these mice. As shown in Figure 5A, CD8 T-cell expansion was markedly elevated in TCRβ^−/− mice injected with anti-Thy1.2 Ab. At least 2-fold to approximately 4-fold increases in CD8 T-cell recovery were observed in lymphoid tissues, while CD4 T-cell recovery was not significantly different except CD4 T cells in the spleen (Figure 5B). When the CD4/CD8 ratio was calculated based on the recovery, the ratio decreased from 9.3 to approximately 3 after the depletion (Figure 5C), suggesting that the presence of γδ T cells may negatively controls CD8 T-cell expansion. Elevated CD8 T-cell expansion was more pronounced after the first week of transfer. As shown in Figure 5D, the ratio of CD4/CD8 T cells in the blood dramatically decreased. When the cell recovery was examined at 4 weeks after transfer, 3-fold to approximately 8-fold increases in CD8 T-cell recovery were observed in every lymphoid tissues while CD4 T-cell recovery remained unchanged (Figure S4). The CD4/CD8 T-cell ratio in the tissues dramatically decreased from 10 to approximately 20 to 2 to approximately 3 by the γδ T-cell depletion (Figure 5E). Collectively, these results suggest that γδ T cells within TCRβ^−/− recipients are critically involved in limiting CD8 T-cell expansion.

Although Thy1-expressing NK cells are expected to be depleted by anti-Thy1.2 Ab and NK cells are known to compete γδ T cells through a mechanism involving IL-15,¹⁶  the involvement of NK cells in the regulation of CD8 T-cell expansion in this setting is unlikely because NK cells do exist in both Rag1^−/− and TCRβ^−/− recipients. To further confirm competition between CD8 T cells and γδ T cells and to exclude a potential involvement of NK cells, we adoptively transferred purified γδ T cells into Rag1^−/− recipients. The recipients were then transferred with naive Ly5.1 T cells 2 weeks after the γδ T-cell transfer. As shown in Figure 5F, the transfer of γδ T cells into Rag1^−/− recipients dramatically suppressed CD8 T-cell expansion; 2-fold to approximately 3-fold reduction of donor CD8 T-cell recovery was found while CD4 T-cell recovery was slightly enhanced. As a result, the CD4/CD8 T-cell ratio increased 3-fold to approximately 6-fold (Figure 5G). These results suggest that γδ T-cells in TCRβ^−/− recipients are primarily responsible for the defects in CD8 T-cell expansion.

Consistent with these results, CD8 T-cell expansion significantly increased when transferred into TCRβ/δ^−/− recipients. CD8 T-cell recovery within TCRβ/δ^−/− recipients increased 2-fold to approximately 6-fold compared with that in TCRβ^−/− recipients (Figure 6). Particularly interesting is that CD4 T-cell recovery dramatically decreased in TCRβ/δ^−/− compared with TCRβ^−/− recipients. The CD4/CD8 T-cell ratio was 1 to approximately 2 in TCRβ/δ^−/− recipients, while the ratio was higher than 10 in TCRβ^−/− recipients. Overall, T-cell expansion in TCRβ/δ^−/− recipients was similar to that found in Rag1^−/− recipients, suggesting that γδ T cells inhibit CD8 T-cell expansion and may promote CD4 T-cell expansion.

IL-15 has been demonstrated to be the prime factor underlying homeostatic competition between CD8 and γδ T cells in sublethally irradiated hosts.¹⁶  To examine if IL-15 is the liming factor of diminished CD8 T-cell expansion in TCRβ^−/− recipients we administered recombinant IL-15 into TCRβ^−/− mice that receive naive Thy1.1 T cells. Provision of exogenous IL-15 increased CD8 T-cell recovery in the lymphoid tissues without altering CD4 T cells (Figure 7A). Similar findings were also observed when Rag1^−/− recipients that have received γδ T cells and rIL-15 before T-cell transfer. As shown in Figure 7B, CD8 T-cell recovery was enhanced when IL-15 was given to the recipients. Of note, γδ T-cell recovery dramatically increased after the IL-15 treatment (Figure 7C). The efficient γδ T-cell recovery after the IL-15 treatment implies that γδ T cells are highly efficient in competing for IL-15. Indeed, IL-15Rα expression on γδ T cells was found to be higher than that on CD8 T cells (data not shown). However, γδ T cells do not make IL-15 and IL-15 production is necessary for transpresentation. Therefore, it is possible that IL-15Rα on γδ T cells is needed for γδ T cell to better use IL-15. We also noticed that IL-15–mediated CD8 T-cell expansion was less pronounced in Rag1^−/− recipients (Figure 7B) compared with that in TCRβ^−/− recipients (Figure 7A). Given that γδ T cells adoptively transferred into Rag1^−/− recipients are activated and induced to proliferate by homeostatic mechanism,¹⁶  these “activated” γδ T cells may compete better for IL-15 than “resting” γδ T cells within TCRβ^−/− recipients.

IL-15/IL-15Rα complexes trans-presented on DC membrane sustained IL-15 activity in vivo, ie, survival of memory CD8 T cells as well as NK cell activation, through IL-15Rβ/γ_c subunits on the responding cells. We thus examined whether γδ T cells limit CD8 T-cell expansion through a mechanism that involves membrane bound IL-15 competition. We measured membrane IL-15 on splenic CD11c⁺ DCs by FACS analysis.²¹  Surface IL-15 expression was significantly higher on DCs in Rag1^−/− mice compared with those in TCRβ^−/− mice (Figure 8A). The IL-15 staining was validated by staining the same DC population after stripping off the membrane IL-15.²¹  As shown in Figure 8A, stripping off the membrane IL-15 eliminated the surface IL-15. Importantly, increases in surface IL-15 expression were reproducibly found on DCs after γδ T-cell depletion in TCRβ^−/− mice (TCRβ^−/− mice with anti-Thy1.2 Ab, Figure 8A), suggesting that γδ T cells might actively down-regulate the surface IL-15 on DCs and that elevated surface IL-15 expression may be responsible for the elevated CD8 T-cell expansion seen in Rag1^−/− or in γδ-depleted TCRβ^−/− recipients. Interestingly, increase in membrane IL-15 after γδ T-cell depletion was more pronounced on the LN DCs than any other tissues tested, although there seemed to be a consistent increase in IL-15 staining as well (Figure 8A). Different γδ T-cell levels within the tissues might result in different surface IL-15 levels in different tissues.

As IL-15 is trans-presented by IL-15Rα, we also measured expression level of IL-15Rα on DCs. Surface IL-15Rα expression was found higher on the LN DCs of Rag1^−/− mice compared with DCs from TCRβ^−/− mice (Figure S5). Consistent with membrane bound IL-15, γδ T-cell depletion increased IL-15Rα expression on DCs (Figure S5). When IL-15 mRNA expression of FACS sorted CD11c⁺ DCs was determined by quantitative PCR, IL-15 expression was similar between these cells (data not shown). Expression of IL-15Rβ/γ_c on responding CD8 and γδ T cells was comparable (data not shown). Therefore, surface IL-15 expression seems to correlate with the presence of γδ T cells in the periphery, suggesting that trans-presentation of IL-15Rα/IL-15 may underlie a competition mechanism between γδ and CD8 T cells.

It was recently demonstrated that both trans-presentation and production of IL-15 occur at the same cells.²⁶  To directly test whether DC production of IL-15 is needed to enhance CD8 T-cell expansion in TCRβ^−/− recipients, we adoptively transferred BMDCs into TCRβ^−/− mice together with Thy1.1 naive T cells. As shown in Figures 8B and C, adoptive transfer of wild-type BMDCs greatly enhanced expansion of both CD4 and CD8 T cells. By contrast, the transfer of IL-15^−/− BMDCs only supported CD4 T cell but CD8 T-cell expansion was marginal. Impaired CD8 T-cell expansion after IL-15^−/− BMDC transfer was more pronounced 14 days after transfer (Figure S6). Diminished CD8 T-cell recovery found after IL-15^−/− BMDC transfer seems partly due to poor survival of CD8 T cells as bcl-2 expression was significantly lower in CD8 T cells in TCRβ^−/− recipients of IL-15^−/− BMDCs compared with those in TCRβ^−/− recipients of wild-type BMDCs (Figure 8D). Therefore, these results suggest that IL-15 production as well as membrane IL-15 presentation by DCs directly control CD8 T-cell expansion and survival within lymphopenic conditions.

Homeostatic competition between CD8 and γδ T cells has been reported. Baccala et al demonstrated that γδ T-cell homeostatic proliferation in sublethally irradiated hosts is mediated through competition of homeostatic cytokine, IL-15, and that γδ T-cell proliferation is inhibited both by γδ T cells themselves and by αβ T cells.¹⁷  French et al also reported that IL-15 competition is the major mechanism that limits γδ T-cell proliferation by CD8 T cells within TCRβ/δ^−/− recipient.¹⁶  However, cellular mechanism underlying IL-15–dependent homeostatic competition between lymphocyte subsets has not been addressed. The current study demonstrates that γδ T cells play a major role in limiting CD8 T-cell expansion within lymphopenic settings and that the competition between these subsets is mediated by IL-15 trans-presented on the membrane of CD11c⁺ DCs.

A key finding that led us to the conclusion came from unexpected dissimilar expansion of T-cell subsets, CD4 and CD8 T cells, when transferred into different lymphopenic recipients: Rag1^−/− and TCRβ^−/−. In lymphocyte-deficient Rag1^−/− recipients, both T-cell subsets underwent rapid proliferation via an IL-7–independent mechanism,⁸  and expanded CD4 and CD8 T cells were equally recovered in the secondary lymphoid tissues. By contrast, in TCRβ^−/− recipients, these T-cell subsets, although the initial proliferation burst was indistinguishable from Rag1^−/− recipients, expanded dissimilarly; CD4 T-cell recovery was enhanced, while CD8 T-cell recovery was relatively diminished. Poor expansion was not due to the presence of cotransferred CD4 T cells or B cells in TCRβ^−/− recipients. Instead, depletion of γδ T cells in TCRβ^−/− mice greatly enhanced CD8 T cell expansion. Similarly, adoptive transfer of γδ T cells into Rag1^−/− recipients significantly diminished CD8 T-cell expansion to the level found in TCRβ^−/− recipients. The surface IL-15 and IL-15Rα level on DCs was found higher without γδ T cells or after γδ T-cell depletion. Importantly, IL-15–sufficient DC cotransfer enhanced both CD4 and CD8 T-cell expansion, whereas IL-15–deficient DC cotransfer only enhanced CD4 T-cell expansion. Collectively, these results demonstrate that the availability of IL-15 produced and trans-presented by DCs underlies homeostatic competitive mechanism between γδ and CD8 T cells in vivo.

Recent studies demonstrated that IL-15 trans-presentation requires expression of both IL-15 and IL-15Rα on the same cells and that IL-15Rα expression on CD8 T cells is dispensable for CD8 T-cell IL-15 responsiveness.^26-28  Both DCs and activated monocytes coexpress IL-15 and IL-15Rα,¹⁸  suggesting that these cells are potential candidates of IL-15 trans-presentation in vivo. Although IL-15Rα subunit displays a wide cellular distribution, including lymphocytes, macrophages, stromal cell lines, and tissue cells such as liver, heart, lung, and endothelial cells,²⁹  a study by Mortier et al indicated that the expression of IL-15Rα specifically on DCs is critical for IL-15 trans-presentation and for IL-15–mediated NK cell activation in vivo¹⁹  as well as the survival of memory CD8 T cells.^21,26,30  Consistent with these reports, Stonier et al recently reported that DCs play a crucial role in trans-presenting IL-15, driving memory CD8 T-cell proliferation in vivo.³¹  The present study provides additional evidence that membrane bound IL-15 trans-presented by DCs also mediates homeostatic competition between γδ and CD8 T-cell subsets. It is important to point out that both CD8 and γδ T cells increase upon IL-15 administration, suggesting that γδ T-cell-mediated inhibition of CD8 T-cell expansion is not operated by a mechanism involving direct suppression by γδ T cells.

IL-15, once made, was shown to preassemble with the IL-15Rα within the ER/Golgi of stimulated DCs before being transported to the cell membrane.¹⁹  This assembly process was reported to be critical for its stability as well as bioactivity.³²  This finding implies that IL-15^−/− cells used in this study might be equivalently defective in IL-15 trans-presentation functions to either IL-15Rα^−/− or IL-15/IL-15Rα^−/− cells.

A study by Jameson and colleagues using mixed bone marrow chimeric mice elegantly demonstrated that IL-15 trans-presentation is limited to the cells that produce the IL-15,²⁶  suggesting that the mechanism would be contact-mediated between cells that present and cells that respond. However, soluble IL-15/IL-15Rα complexes can be released from the activated DCs and found in the sera of polyinosinic:polycytidylic acid (poly I:C)–challenged mice.¹⁹  It is unclear whether the soluble complexes are important during maintenance of T-cell homeostasis. However, only the membrane-bound complexes rather than the soluble complexes were shown to support the long survival of memory CD8 T cells in vivo.²¹  IL-15Rα also recycles leading to the persistent presentation of the complexes, by which IL-15 promotes the survival of T cells after IL-15 withdrawal, and sustained lymphopenia-driven proliferation and importantly accumulation of CD8 T cells in vivo.³³ 

Is IL-15 the sole factor mediating homeostatic competition between CD8 and γδ T cells? IL-7 has been demonstrated to be the prime factor involved in T-cell proliferation under lymphopenic conditions as well as survival.¹⁴  When T-cell expansion was examined in IL-7-administered TCRβ^−/− recipients, expansion of all T-cell subset (CD4, CD8, and γδ) increased probably by enhanced cell survival (data not shown). Thus, IL-15 may play a unique role in mediating homeostatic competition between CD8 and γδ T-cell subsets in vivo.

It is interesting to note that CD8 T-cell recovery at the very early time point (4 days after transfer) is very different in Rag1^−/− and TCRβ^−/− recipients (Figure 2C). More specifically, approximately 10- to 1000-fold greater numbers of CD8 T cells were found in the lymphoid tissues of TCRβ^−/− recipients, while CD4 T-cell recovery was comparable. These results raise an interesting possibility that the relatively large numbers of CD8 T cells within TCRβ^−/− recipients might lead to poor CD8 T-cell expansion observed in these recipients. Given that γδ T cells control CD8 T-cell expansion/survival via IL-15–dependent mechanism, it will be interesting to test whether γδ T cells affect CD8 T-cell recovery, and then CD8 T-cell expansion at early time points.

It is also interesting to note that relative expansion of T-cell subsets is determined relatively early after transfer into lymphopenic conditions. Experiments measuring BrdU incorporation (1-4 weeks after transfer) revealed that a burst-like CD8 T-cell proliferation and subsequent expansion occurs relatively early after transfer into Rag1^−/− mice (ie, before 7 days in all tissues and before 14 days in mesenteric LN). Expanded T cells then appear to reach a homeostatic equilibrium after which constant T-cell levels are maintained. Furthermore, it also seems that CD4 T cells expand better in TCRβ^−/− than in Rag1^−/− recipients. Consistent with this finding, CD4 T-cell expansion was found to be significantly compromised in TCRβ/δ^−/− recipients, the level of which is similar to that found in Rag1^−/− recipients. These results raise an interesting possibility that γδ T cells may enhance CD4 T-cell expansion and/or survival. Indeed, interplay between αβ and γδ T-cell subsets have been reported during γδ T-cell development in the thymus³⁴  and after γδ T-cell modulation with anti-γδ T-cell Ab, GL3 injection.³⁵  Whether γδ T cells contribute to maintaining CD4 T-cell homeostasis will require further investigation. One alternative possibility is that APCs, particularly DCs within TCRβ^−/− mice, may enhance proliferating CD4 T-cell survival, although noticeable differences in phenotypes (based on major histocompatibility complex [MHC] and costimulatory molecules expression) of DCs were not found in TCRβ^−/− and Rag1^−/− mice (data not shown).

In conclusion, we have demonstrated that the presence of γδ T cells results in dramatic impacts on expansion as well as on survival of CD8 αβ T-cell subset. The mechanism that competes for IL-15 produced and trans-presented by the CD11c⁺ DCs directly underlies homeostatic maintenance within lymphopenic conditions between γδ and CD8 T cells in vivo. Identifying cellular interaction between these cells in vivo will further provide critical insights into understanding T-cell responses associated with dysregulated homeostasis found in patients with HIV infection or with chemotherapy.

We thank Biogen Idec for kindly providing anti–mouse CD20 Ab (18B12) and isotype-matched control Ab (2B8). We also thank Jennifer Power for excellent cell sorting.

This work was supported by startup funds from the Cleveland Clinic Foundation, by the Scientist Development grant (0730139N) from the American Heart Association (New York, NY), and in part by National Institutes of Health (NIH; Bethesda, MD) grant no. R01-AI074932 (to B.M.).

National Institutes of Health

Contribution: J.-s.D. designed experiments, performed all experiments, analyzed data, and prepared part of the manuscript; and B.M. designed experiments, analyzed data, and wrote the manuscript.

Conflict-of-interest disclosure: The authors declare no competing financial interests.

Correspondence: Booki Min, Department of Immunology/NB30, Lerner Research Institute, Cleveland Clinic Foundation, 9500 Euclid Ave, Cleveland, OH 44195; e-mail: minb@ccf.org.