CC chemokine receptor 8 potentiates donor T_reg survival and is critical for the prevention of murine graft-versus-host disease

Graft-versus-host disease (GVHD) remains the greatest barrier to successful hematopoietic stem cell transplant (HSCT) outcomes. Over the past several years, substantial animal model research has been devoted to the use of donor, thymically derived, “natural” regulatory T-cell (T_reg) infusions for the prevention of GVHD.^1-3  In addition, T_regs have recently been administered with both haploidentical and cord blood stem cell products in phase 1 clinical trials.^4,5 

Despite the growing use of therapeutic T_reg infusions in the HSCT setting, it is still unclear how donor T_regs prevent GVHD and which proteins are necessary for their immunosuppressive properties. Although there is indirect evidence that T_regs initially prevent GVHD by blocking early conventional T-cell (T_con) activation and expansion within recipient lymphoid tissue,^6,7  significant uncertainty exists as to how they function later. Understanding how T_regs function over time is critically important to optimizing their clinical use.

CC chemokine receptor (CCR) 4 and CCR8 are both expressed by T_regs and have been shown to mediate the chemotaxis of human T_regs toward dendritic cells (DCs) where they presumably suppress developing immune responses.⁸  In addition, both receptors have been implicated in the trafficking of T_regs^9,10  and T helper 2–polarized T_cons^11,12  into the skin, an important GVHD target organ. As a result, we set out to study the contributions of CCR4 and CCR8 to the respective abilities of donor T_cons and T_regs to induce and prevent GVHD.

Using CCR4 and CCR8 knockout mice, we found that neither chemokine receptor alone plays a critical role in GVHD induction by T_cons. Similarly, we found CCR4 to be dispensable for donor T_reg function after transplant. CCR8 knockout (CCR8^−/−) T_regs, however, were severely impaired in their ability to prevent GVHD because of increased cell death after the first transplant week. Collectively, our data suggest that targeting the CCR8 axis may represent a possible therapeutic approach for either enhancing or reducing T_reg numbers in vivo.

C57BL/6 (“B6”), B6xDBA/2 F1(“B6D2”), BALB/cJ, B6.PL-Thy1^a/CyJ, C.FVB-Tg(Itgax-DTR/EGFP)57Lan/J, and B6.FVB-Tg(Itgax-DTR/EGFP)57Lan/J mice were purchased from The Jackson Laboratory. Enhanced green fluorescent protein (eGFP)–expressing B6 mice were generated as described.¹³  CCR8^−/− mice backcrossed 8 generations onto a B6 background were obtained from the laboratory of Donald Cook.¹⁴  All experiments were performed in accordance with protocols approved by the Institutional Animal Care and Use Committee at the University of North Carolina.

Donor T-cell–depleted (TCD) bone marrow (BM) cells and whole CD25-depleted splenic T_cons were prepared as previously described.¹⁵  Donor T_regs were purified using a column purification method.⁷ 

CD4⁺CD25⁺ B6 T_regs and CD4⁺CD25^– B6 T_cons were isolated using column purification. Fifty percent of these cells were immediately frozen. The other 50% of the T_regs and T_cons were activated in vitro with plate-bound anti-CD3 and anti-CD28 antibody in the presence of supplemental interleukin (IL) 2 at 100 IU/mL. After 72 hours, these cells were frozen. Later, all cells were thawed, and their total RNA isolated for quantitative reverse-transcription polymerase chain reaction (qRT-PCR) as previously described.¹⁵ 

Intrinsic T_reg immunosuppressive function was assessed using an in vitro suppression assay as previously described.² 

Recipient mice were imaged with a Zeiss StereoLumar V12 microscope with eGFP band-pass filter at room temperature. eGFP intensities were determined with AxioVision (Carl Zeiss) software. For detailed exposure times and magnifications, refer to the supplemental Methods (see the Blood Web site).

Recipient organs were homogenized, and absolute eGFP levels measured using an enzyme-linked immunosorbent assay (ELISA) kit (Cell Biolabs).¹⁶  We verified that relative eGFP levels within host tissue after transplant correlate with actual T_reg numbers in vivo by performing a series of transplants using increasing doses of donor eGFP⁺ T_regs. As seen in supplemental Figure 1, eGFP levels within host lymphoid tissue as determined by ELISA correlated better with the initial T_reg dosing cohort (low, medium, and high) than traditional cell counting by flow cytometry.

B6 CD11c⁺CD14^– DCs were generated by culturing B6 BM cells in the presence of supplemental granulocyte macrophage–colony-stimulating factor, IL-4, and tumor necrosis factor as described previously.¹⁷  B6 macrophages were generated by culturing B6 BM cells for 7 days in conditioned media obtained from L929 cells (as a macrophage–colony-stimulating factor source) as previously described.¹⁸ 

Survival curves were constructed using the method of Kaplan and Meier. Overall survival was compared using Fisher’s exact test. For ELISAs, GVHD scoring, and other experiments where distributions were typically not Gaussian, means were compared using the nonparametric Mann-Whitney test. All other continuous variables were compared using the 2-tailed Student t test. Where indicated, T_reg ratios following coculture were compared using a 1-sample t test. For this work, the null hypothesis was as follows: the fold-difference results were sampled from a distribution with mean fold difference of 1.0. P values were adjusted for multiple testing according to the methods of Benjamini and Hochberg (false discovery rate control)¹⁹  and Bonferroni.²⁰ P values < .05 were considered significant. Error bars represent standard error of the mean.

Using qRT-PCR, we examined CCR4 and CCR8 expression by naïve and activated T_cons and T_regs. Compared with naïve CD4⁺CD25^– T cells, both CCR4 and CCR8 were produced at higher levels in unstimulated CD4⁺CD25⁺ T_regs (Figure 1A). Upon activation under nonpolarizing conditions, both populations of CD4⁺ T cells downregulated CCR7 and upregulated CCR4 and CCR8 to varying degrees. We found CCR4 expression to be strongest in activated T_cons, with CCR8 being upregulated preferentially by T_regs.

We next examined how CCR4 and CCR8 would affect the ability of donor T_cons to induce GVHD. For these experiments, we used a well-described C57BL/6 (H-2^b; “B6”) into B6xDBA/2 F1 (H-2^bxd; “B6D2”) haplotype matched murine transplant model and donor B6 mice knocked out at either the CCR4 (Figure 1B-C) or CCR8 locus (Figure 1D-E). Both CCR4^−/− and CCR8^−/− T_cons induced severe GVHD in wild-type (WT) B6D2 recipients that was virtually indistinguishable from that generated by WT T_cons.²¹ 

Because both CCR4 and CCR8 are expressed by naïve and in particular activated T_regs, we next examined the impact that these receptors would have on the ability of donor T_regs to prevent lethal GVHD. Somewhat unexpectedly, we found that CCR4^−/− T_regs were efficient at attenuating GVHD, with 100% of CCR4^−/− T_reg recipients surviving until the end of the study (Figure 1F-G). CCR8^−/− T_reg recipients, however, developed severe GVHD in a very reproducible pattern (Figure 1H-I). For the first 7 to 14 days posttransplant, recipients of CCR8^−/− T_regs exhibited low GVHD scores similar to those mice given WT T_regs. Thereafter, however, the GVHD curves between the CCR8^−/− and WT T_reg treatment groups diverged with recipients of CCR8^−/− T_regs developing acute GVHD that ultimately approximated that of control mice receiving T_cons alone (Figure 1I). In summary, although T_regs lacking CCR8 were initially effective in preventing GVHD, those animals given CCR8^−/− T_regs developed delayed, severe disease.

To elucidate a mechanism for the impaired capacity of CCR8^−/− T_regs to prevent GVHD lethality, we examined their activation and expansion early after transfer. Regardless of CCR8 expression, donor T_regs demonstrated a primarily activated phenotype by transplant day +7, with reduced levels of l-selectin and abundant CD44 (Figure 2A-B).

We next evaluated the ability of CCR8^−/− T_regs to proliferate in vivo using carboxyfluorescein diacetate succinimidyl ester (CFSE) labeling. For this experiment, freshly isolated WT or CCR8^−/− T_regs were labeled with CFSE and then transplanted into irradiated B6D2 recipients. As seen in Figure 2C, both WT and CCR8^−/− donor T_regs demonstrated a near total loss of CFSE, consistent with the very aggressive homeostatic T_reg expansion that occurs in the lymphopenic environment that develops after lethal conditioning. CFSE labeling controls using permeabilized cells demonstrated no loss of CFSE signal intensity as a result of the permeabilization step required for intracellular Foxp3 staining (supplemental Figure 2).

Next, we examined the ability of WT and CCR8^−/− T_regs to suppress CD4⁺ T-cell proliferation in a 1-way mixed lymphocyte reaction (Figure 2D). Using irradiated TCD B6D2 splenocytes as stimulator cells, we were unable to detect a difference in the in vitro suppressive ability of CCR8^−/− vs WT T_regs at the dilutions examined. Taken together, these data indicated that without CCR8, natural T_regs remain capable of normal activation in vivo and possess no intrinsic defects in their ability to suppress CD4⁺ T-cell proliferation.

Because chemokine receptors are known to contribute to lymphocyte trafficking in vivo, we hypothesized that CCR8^−/− T_regs might demonstrate abnormal homing after transplant. To address this question, irradiated B6D2 recipients were administered WT or CCR8^−/− eGFP⁺ T_regs. T_reg trafficking into recipient lymph nodes and Peyer’s patches was then assessed at multiple time points after transplant using fluorescence stereomicroscopy. Additionally, we used an anti-eGFP ELISA to compare relative donor T_reg numbers within GVHD target organs between WT and CCR8^−/− recipients.¹⁶ 

As seen in Figure 3A, by posttransplant day +7, WT and CCR8^−/− T_regs accumulated to a similar degree within the inguinal lymph nodes, mesenteric lymph nodes, and Peyer’s patches. In addition, similar eGFP levels were detected by ELISA within the spleen and peripheral GVHD target organs including the liver, lung, and colon. Notably, the strength of the eGFP signal at this relatively early time point was unexpectedly intense within the colon in both treatment groups, suggesting a particularly strong T_reg tropism for gastrointestinal tissues in the HSCT setting.

When we performed a similar set of experiments on transplant day +10, we began to observe differences between the 2 treatment groups. By microscopy, fewer CCR8^−/− T_regs were noted within recipient inguinal lymph nodes, mesenteric lymph nodes, and Peyer’s patches, and by ELISA significantly less eGFP was detected within the spleen and colon (Figure 3B). By day +14, a weaker eGFP signal was again visualized within recipient lymphatic tissue in those mice given CCR8^−/− T_regs, and less eGFP was detected within the lung and colon (Figure 3C). Collectively, these data indicated that despite a preserved ability to accumulate within host tissues early after transplant, CCR8^−/− T_regs failed to persist in vivo to the same extent as WT cells.

We next assessed how the premature loss of CCR8^−/− donor T_regs would affect donor T_con numbers within recipient tissues. For these experiments, WT or CCR8^−/− eGFP^– T_regs were transplanted into irradiated B6D2 recipients along with eGFP⁺ B6 T_cons. As seen in Figure 4A, recipients of WT and CCR8^−/− T_regs demonstrated similar T_con accumulation within peripheral GVHD target organs 2 weeks after transplant. Twenty-one days after transplantation, however, those mice receiving CCR8^−/− T_regs developed a delayed increase of T_cons within multiple sites by microscopy, and significantly greater eGFP levels within the liver and the colon by ELISA (Figure 4B). In summary, WT and CCR8^−/− T_regs allowed for a similar degree of donor T_con expansion over the first 1 to 2 weeks after transplant, which corresponded to the nearly identical clinical GVHD scores observed in the 2 groups over the same time period (Figure 1I). Thereafter, the accelerated decline in donor T_reg numbers seen in mice given CCR8^−/− cells was accompanied by an increase of T_cons in several GVHD target organs.

Based on our previous findings, we hypothesized that the accelerated in vivo decline in donor CCR8^−/− T_regs was the result of either a deficiency in their capacity to undergo extended cell division after transplant or an increased susceptibility to cell death. To evaluate for any differences in the late proliferative potential of WT vs CCR8^−/− donor T_regs after transplant, a 5-bromo-2’-deoxyuridine (BrdU) labeling approach was used. As seen in Figure 5A-B, no differences in BrdU incorporation were noted between recipients of WT and CCR8^−/− T_regs on transplant day +14.

Next, we looked for any differences in T_reg survival that might occur in the absence of CCR8. For this work, irradiated B6D2 mice were transplanted with T_regs from either WT or CCR8^−/− donors. On transplant day +10, donor T_regs were then examined for markers of apoptosis and cell viability. As seen in Figure 5C-D, CCR8^−/− donor T_regs were found to be more susceptible to apoptosis after adoptive transfer, with 18% of the CCR8^−/− cells undergoing cell death compared with only 7% of the WT T_regs at this time point.

Given the survival defect displayed by CCR8^−/− T_regs, we questioned whether either of CCR8’s known agonists, CC chemokine ligand (CCL) 1 or CCL8,²²  could directly affect T_reg survival in vitro. Previously, T_regs have been shown to be sensitive to apoptosis when cultured without supplemental IL-2.^23,24  Taking advantage of this, we cultured WT B6 T_regs on plates coated overnight with anti-CD3 and anti-CD28 antibody. In the setting of T-cell receptor and CD28 stimulation, more than half of the T_regs survived to 72 hours. Nevertheless, neither the addition of supplemental CCL1 nor CCL8 over a range of physiological concentrations in the absence of IL-2 had any direct effects on T_reg viability (supplemental Figure 3).

Previous studies have demonstrated a role for DCs in the survival of T_regs in nontransplanted mice.²⁵  Also, DCs have been shown to produce CCL1, a modest but highly specific chemoattractant for T_regs.⁸  As a result, we hypothesized that CCR8 signaling might enhance extended T_reg survival by facilitating crucial interactions with syngeneic donor DCs. To test this hypothesis, we generated activated CD11c⁺ Class II MHC^high CD14^– DCs from B6 BM cells.¹⁷  Using qRT-PCR, we found that B6 DCs produced both CCL1 and CCL8 (supplemental Figure 4A) and verified that CCL1 but not CCL8 was able to induce T_reg chemotaxis in a CCR8-dependent fashion (supplemental Figure 4B).

Following up on these data, we determined whether we could prevent B6 T_reg apoptosis in vitro with B6 DCs. For these experiments, we cocultured activated B6 DCs with freshly isolated B6 T_regs on uncoated plates in the absence of supplemental IL-2. As seen in Figure 6A, activated DCs were able to rescue syngeneic WT T_regs from cell death in a dose-dependent fashion. In addition, we examined whether this process was specific to DCs by culturing B6 T_regs with syngeneic B6 macrophages. As seen in Figure 6B, macrophages were modestly able to prevent T_reg apoptosis but were generally less efficient and less potent than DCs on a per cell basis. As a result, we elected to focus exclusively on T_reg:DC interactions for the remainder of our experiments.

To more definitively evaluate whether CCR8 would affect the ability of DCs to promote T_reg survival in vitro, we cocultured equal numbers of WT eGFP^– Thy1.1⁺ B6 T_regs and CCR8^−/− eGFP⁺ Thy1.2⁺ T_regs with variable numbers of eGFP^– activated B6 DCs. Under these circumstances, the WT and CCR8^−/− T_regs were present in the same culture conditions and therefore forced to compete with one another for access to prosurvival signals provided by the resident syngeneic DCs. After 72 hours, T_reg numbers were determined by flow cytometry. As seen in Figure 6C, no difference in the ratio of WT to CCR8^−/− T_regs was observed in the absence of DCs. However, when DCs were added to the culture conditions, we consistently found an increase in the number of WT T_regs relative to the number of CCR8^−/− T_regs, suggesting that CCR8 facilitates prosurvival interactions between T_regs and resident DCs. Notably, this effect was observed even in the presence of saturating amounts of supplemental IL-2, a cytokine critical for T_reg development and growth. Because the WT and CCR8^−/− T_regs were admixed under these culture conditions and therefore exposed to the same soluble signaling milieu, our data suggested that WT T_regs are better able to respond to DCs locally in an IL-2–independent fashion.

To further explore the mechanism by which syngeneic DCs are able to promote T_reg survival, we examined B6 T_reg signal transducer and activator of transcription (STAT) 5 phosphorylation following coincubation with B6 DCs. Stat5 is an important signaling intermediary downstream of both the T-cell receptor²⁶  and the IL-2 receptor,²⁷  and it is known to mediate antiapoptotic affects within T_regs via the induction of Bcl-xL.²⁷  In addition, Stat5 directly activates Foxp3 expression and inhibits T helper 17 differentiation in a dichotomous relationship with Stat3.²⁸  For this work, we began by examining baseline Stat5 phosphorylation in freshly isolated eGFP⁺ B6 T_regs. As seen in Figure 6D, naïve T_regs expressed phospho-Stat5 at low levels (left panel), which were modestly higher than a negative control (right panel). Next, we cultured WT or CCR8^−/− eGFP⁺ B6 T_regs with eGFP^– B6 DCs for 72 hours and examined the T_regs for the phosphorylated form of Stat5. As seen in Figure 6E, nearly all of the WT and CCR8^−/− T_regs that survived the incubation period and retained their ability to produce eGFP expressed phospho-Stat5 at higher levels than naïve T_regs, suggesting that this transcription factor may be involved in the prosurvival signals transmitted to T_regs by DCs. The frequency of T_regs expressing phospho-Stat5 did not vary as a function of CCR8. However, a major difference in absolute T_reg numbers was again noted between the 2 experimental groups despite exposure to identical B6 DC numbers (Figure 6F). Collectively, these results indicated that CCR8^−/− T_regs display no qualitative differences in their ability to activate Stat5 antiapoptotic machinery compared with WT cells but appear to demonstrate a quantitative impairment in their ability to physically interact with survival-promoting antigen-presenting cells (APCs) over time.

If indeed CCR8 promotes T_reg survival by promoting critical interactions with neighboring DCs, we reasoned that the survival defect displayed by CCR8^−/− T_regs compared with WT T_regs would be minimized in a scenario where the T_regs would be unable to establish direct contact with APCs. To definitively test this hypothesis, we made use of transwell plates. As seen in Figure 6G-H, no differences in WT or CCR8^−/− T_reg numbers were noted when the cells were physically separated from activated B6 DCs or when no DCs were present. However, when both the T_regs and DCs were cultured together and therefore capable of direct cellular contact, greater numbers of WT T_regs were once again noted (Figure 6I). In summary, our in vitro studies are consistent with a mechanism in which the expression of CCR8 by T_regs enhances their survival by promoting physical interactions with DCs.

Consistent with previous reports, our data indicated that syngeneic DCs are able to promote T_reg survival.²⁵  As a consequence, we hypothesized that in the HSCT setting, donor APCs derived from the donor BM play a role in maintaining donor T_regs posttransplant and that their depletion could recapitulate the transplant results observed with CCR8^−/− T_regs. To test this hypothesis, we performed a series of experiments using B6 mice transgenic for eGFP and the diphtheria toxin receptor (DTR) under control of the CD11c promoter, a cell surface marker expressed by APCs including classic and plasmacytoid DCs (“CD11c-DTR/eGFP” B6 mice). In brief, irradiated B6D2 recipients were transplanted with TCD BM from either WT B6 or CD11c-DTR/eGFP B6 mice along with WT Thy1.1⁺ B6 T_regs and Thy1.2⁺ B6 T_cons. All recipient mice were then administered diphtheria toxin (DT) by intraperitoneal injection. In this situation, CD11c-expressing APCs derived from the donor BM in those animals receiving CD11c-DTR/eGFP B6 BM but not WT B6 BM were selectively eliminated by the administration of DT. On transplant day +12, donor T_regs were quantified within recipient spleens using flow cytometry. Because only those donor T_regs arising from the initial T_reg inoculum were congenic for Thy1.1, we were able to distinguish these cells from any new Foxp3⁺ cells that may have arisen from the Thy1.2⁺ donor BM.

Across experiments, we were able to consistently deplete donor (K^d–) CD11c⁺ cell numbers by 70% to 80% in those mice given CD11c-DTR/eGFP BM compared with those receiving WT BM cells (supplemental Figure 5). As seen in Figure 7A, this reduction in donor APCs resulted in a 60% drop in Thy1.1⁺ donor T_regs in the recipient spleen by transplant day +12, suggesting that donor BM-derived APCs do contribute to the continued maintenance of donor T_regs after transplant.

Given our previous in vitro data suggesting a contribution by CCR8 to the ability of T_regs to colocalize to neighboring DCs, we reasoned that CCR8^−/− donor T_regs might not be affected by donor APC depletion, because they would fail to efficiently migrate to donor DCs. To test this hypothesis, we performed a similar transplantation experiment but substituted eGFP⁺Thy1.2⁺CCR8^−/− T_regs for WT Thy1.1⁺ T_regs. Consistent with our prior observations, donor CCR8^−/− T_regs were observed in lower numbers than WT donor T_regs within recipient spleens on transplant day +12 in both WT and CD11c-DTR/eGFP BM recipients (Figure 7A-B). Also, as predicted, the depletion of donor DCs did not have a statistically significant impact on the number of donor CCR8^−/− T_regs present. Collectively, these data indicated that donor DCs are much more critical for the maintenance of WT donor T_regs compared with CCR8^−/− donor T_regs.

Next, we turned our attention to the contribution of host APCs to donor T_reg expansion and survival. For this, we used a B6 into BALB/c transplant model system with either WT BALB/c or CD11c -DTR/eGFP BALB/c mice as our transplant recipients. In this scenario, host CD11c APCs can be depleted in the latter by the administration of DT. Unfortunately, DTR expression in commercially available CD11c-DTR/eGFP BALB/c mice is not restricted to the hematopoietic compartment, with DTR also appearing on pulmonary endothelial cells. As a result, CD11c-DTR/eGFP mice become extremely susceptible to DT toxicity following lethal irradiation. To circumvent this problem, we generated BALBc->BALB/c (“WT”) and CD11c-DTR/eGFP BALB/c-> BALB/c (“CD11c.DTR”) BM chimeras and then used these as secondary transplant recipients. For these second transplants, WT and CD11c.DTR chimeric mice were lethally irradiated and given DT by intraperitoneal injection. Recipient mice were administered TCD BM from Thy1.1⁺ B6 mice with T_cons and T_regs from Thy1.2⁺ WT B6 donors. Donor T_reg numbers within the recipient spleen were then quantified by flow cytometry on day +7. As seen in Figure 7C, we found that the depletion of host CD11c⁺ APCs did not have a statistically significant effect on donor T_reg numbers after transplant. In order to rule out any potentially confounding effects that host APC depletion might have on donor T_con expansion and overall IL-2 production, we performed an identical transplant in the absence of supplemental donor T_cons. As seen in Figure 7D, we again failed to observe a reduction in donor T_reg numbers in the setting of host APC depletion. Instead, donor T_reg numbers were actually found to be enhanced. Taken together, these data indicated that host CD11c⁺ APCs are at least partially dispensable for early donor T_reg engraftment and expansion after transplant.

Following the secondary transplantation of our BM chimera mice, we consistently observed varying degrees of splenomegaly in CD11c.DTR BALB/c chimeras following the administration of DT, as well as increased overall splenic cellularities compared with the WT BALB/c chimeras. As a result, we used flow cytometry to examine donor BM reconstitution in the setting of host CD11c⁺ APC depletion. Using a K^d– gate, we consistently found that both donor neutrophil and CD11c⁺ reconstitution were greatly enhanced in the setting of host CD11c⁺ APC depletion (Figure 7E-F, respectively). In contrast, the effects of host APC depletion on donor B-cell, natural killer–cell, and T_con numbers were more variable, with these numbers increasing modestly in some but not all of our transplantation experiments (data not shown).

In this study, we have used CCR8 knockout mice to examine the mechanisms by which donor T_regs prevent GVHD after transplant. Here, we show that extended T_reg survival is crucial for long-term GVHD protection and that this in turn depends at least in part on functional CCR8. Previously, emphasis has been placed on the importance of the very early suppression of donor T_con expansion within host secondary lymphoid tissue as the means by which T_regs prevent GVHD.^6,7  Although these effects are likely important for GVHD attenuation, our findings indicate that they represent only a portion of the overall mechanism(s) by which donor T_regs are able to improve transplant outcomes. Here, we show that CCR8^−/− T_regs ultimately fail to protect recipient mice against lethal GVHD despite demonstrating a normal capacity to accumulate within host secondary lymphoid tissue and a preserved ability to block early donor T_con expansion.

It is well known that chemokine receptors contribute to lymphocyte homing in vivo. Our data, however, highlight their more subtle functional effects. CCR8^−/− T_regs demonstrated in vivo trafficking patterns that were similar to WT cells. Despite this, T_regs lacking CCR8 were severely impaired in their ability to prevent GVHD because of an increased propensity to undergo cell death. To our knowledge, our data are the first to link a single chemokine receptor to T_reg survival. The mechanisms by which CCR8 potentiates donor T_reg survival in vivo are not entirely certain, although we suspect based on our in vitro data that CCR8 functions to direct donor T_regs toward donor DCs where they receive antiapoptotic signals. Although our data do not allow us to identify the specific DC subsets that mediate these effects in vivo, the prosurvival signals mediated by CCR8 appear to depend on cell contact because WT and CCR8^−/− T_regs were observed in similar numbers after being incubated with syngeneic DCs in transwell plates. The downstream signaling events following DC engagement are incompletely understood. However, we found that T_regs strongly phosphorylated Stat5 after coculture with syngeneic DCs, a protein that is known to mediate antiapoptotic effects in regulatory T cells. Notably, the in vitro stimulation of CCR8 with either of its known ligands was insufficient to directly alter T_reg survival and/or expansion. Furthermore, CCR8^−/− T_regs did not demonstrate any impairment in their ability to respond to soluble IL-2, as WT and CCR8^−/− persisted to a similar degree when cultured with saturating amounts of IL-2 in the absence of antigen presenting cells.

As described previously, CCR8’s prosurvival properties depended on the presence of DCs in vitro. In our transplantation experiments, however, we discovered that host- and donor-derived CD11c⁺ APCs produce substantially different effects on donor T_regs. As anticipated, donor BM-derived APCs played a role in maintaining the initial donor T_reg inoculum. Unexpectedly, however, the depletion of host BM-derived APCs failed to reduce donor T_reg numbers and even increased them in some experiments. Because we were not able to completely deplete all host CD11c⁺ cells, we cannot conclusively state that host CD11c⁺ APCs play no role in donor T_reg activation. However, our data would indicate that this population of cells is at least partially dispensable for early T_reg expansion.

Although initially unexpected, our APC depletion data are consistent with a previous report. In work by Tawara et al,²⁹  the authors examined the influence of host APCs on the expansion and maintenance of donor T_regs posttransplant using host animals unable to express class II MHC. There, class II MHC expression by recipient hematopoietic cells appeared to be required for T_reg-mediated GVHD protection but was dispensable for donor T_reg proliferation and maintenance. When donor BM-derived APCs were unable to express class II MHC, however, a significant drop in donor T_reg numbers was observed. Our own DTR chimeric transplants varied from the model system employed previously. However, our data would support their finding that donor APCs are primarily responsible for supporting donor T_reg numbers after transplant.

In several experiments, we observed splenomegaly and an increase in myeloid reconstitution following the administration of DT to CD11c.DTR chimeric mice. In previous reports, neutrophilia and monocytosis have been reported following the administration of DT to nontransplanted CD11c-DTR/eGFP mice.^30-32  In those studies, the increased myeloid numbers were attributed to an accelerated release of neutrophils from the BM in response to CXC chemokine ligand 1 and/or 2, and to enhanced granulopoiesis. Whether the increased myeloid engraftment observed in our CD11c.DTR chimeric recipients was a direct effect of the DT itself or secondary to the enhanced depletion of host APCs that occurs in these mice is uncertain. Notably, no neutrophilia was detected following the administration of DT to WT mice receiving donor CD11c.DTR BM cells, suggesting that host APCs in particular may exert myelosuppressive effects early after stem cell transplantation.

From a translational standpoint, our T_reg data suggest that the CCR8 signaling axis could be targeted as a means for manipulating natural T_reg numbers in vivo. Although T_regs can clearly produce beneficial effects in the HSCT setting, T_regs have been linked to a breakdown of antitumor immunity.^33-40  Therefore, it may at times be desirable to deplete T_reg numbers in vivo, a task that is often difficult with anti-CD25 antibody and/or cytotoxic chemotherapy.⁴¹  The chemical structures for small-molecule antagonists against human CCR8 have already been published⁴²  and could form the basis for future drug development. Furthermore, a previous study demonstrated the potential feasibility of this approach, with neutralizing antibodies against CCL1 leading to reduced numbers of intratumor T_regs and solid tumor regression in a murine breast cancer model.⁴³  In the hematologic malignancy setting, the inhibition of CCR8 signaling might be pursued as a means for boosting antileukemia immunity following autologous transplant or during standard acute myeloid leukemia induction.

In conclusion, we have demonstrated a critical role for CCR8 in the maintenance of donor T_regs after transplant. In so doing, we have identified a potential drug target for the manipulation of T_reg numbers in vivo and have shed new light on the contributions of both donor- and host-derived APCs to donor T_reg function after HSCT.

This work was supported by the following National Institutes of Health grants: National Cancer Institute 5K12CA120780-05 and National Heart, Lung, and Blood Institute 5K08HL111205-02 (J.M.C.); and National Heart, Lung, and Blood Institute R01HL115761, National Cancer Institute RO1CA166794, and National Institute of Allergy and Infectious Diseases R56 AI064363 (J.S.S.).

Contribution: J.M.C. performed experiments and wrote the manuscript; K.A.F., M.L.W., L.M.F., B.G.V., K.L., A.P.-M., K.P.M., and H.v.D. performed, designed, or interpreted experiments; D.N.C. derived the CCR8^−/− mice and reviewed the manuscript; B.R.B. reviewed the manuscript; and J.S.S. conceived the project and assisted in the writing of the manuscript.

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

Correspondence: James M. Coghill, Lineberger Comprehensive Cancer Center, 450 West Dr, The University of North Carolina at Chapel Hill, Chapel Hill, NC 27599-7305; e-mail: jcoghill@e-mail.unc.edu.