CCR6 regulates CD4⁺ T-cell–mediated acute graft-versus-host disease responses

Graft-versus-host disease (GvHD) is a common, debilitating complication of allogeneic bone marrow transplantation.¹  The disease is a multistep process involving lymphocyte activation, proliferation, and target cell killing. Donor T cells recognize the major histocompatibility complex (MHC) and associated peptides on host-derived antigen-presenting cells. This leads to cell activation and differentiation into various effector and memory T cells with selective tropism.^2,3  In acute GvHD, activated T cells infiltrate mainly the skin, liver, and gastrointestinal tract where they initiate a T-helper 1–mediated immune response.⁴  Interferon γ (IFN-γ) collaborates in recruitment of other T-cell subsets, monocytes, and macrophages that further damage affected tissues.^5,6 

Adhesion and chemoattractant receptors expressed by effector/memory T cells control their homing specificity.⁷  The CCL20/CCR6 axis has a central role in chemoattracting specific T-, B-, and dendritic cell subsets to peripheral tissues and lymphoid organs associated with epithelial surfaces (for a recent review, see Schutyser et al⁸ ). CCR6 expression is restricted to the CLA⁺,

⁠, or

\(\mathrm{CLA}^{-}{\alpha}_{4}{\beta}_{7}^{-}\)

cell subsets of the effector/memory T-cell pool, which home to skin, gut, or other nonlymphoid peripheral tissues, respectively,⁹  suggesting that CCR6 is induced on T cells that are activated in diverse environments. CCR6-deficient T-cell proliferation is not impaired in response to T-cell receptor (TCR) stimulation in vitro or to alloantigens in a mixed leukocyte reaction (MLR) assay,¹⁰  in which CCR6 is up-regulated in CD4⁺ T cells.¹¹  Mucosal immunity is impaired in CCR6-deficient mice,^10,12  which also have altered immune responses to cutaneous and intestinal inflammations.^10,13  CCL20 is expressed constitutively in normal skin and mucosa-associated tissues; inflammatory stimuli up-regulate these basal levels.⁸  The data available indicate that CCL20 acts through CCR6 to recruit differentiated memory cells to inflammation sites in skin and mucosa.

We analyzed the role of CCR6 in GvHD development in MHC-mismatched recipients. Results showed delayed disease onset and decreased GvHD incidence and mortality in MHC class II–mismatched recipients of CD4⁺ T cells from CCR6-deficient donors compared with wild-type (WT) donors. Conversely, CD8⁺ T cells from CCR6-deficient and WT donors provoked similar GvHD symptoms in MHC class I–mismatched recipients. CCR6 was also critical for GvHD development across minor histocompatibility complex (mhc)–mismatched barriers. Onset of skin lesions and intestinal inflammation was delayed and symptoms were milder in MHC-matched/mhc-mismatched recipients of naive CD4⁺CD45RB^high T cells from CCR6-deficient donors. We observed significant delay in donor cell infiltration to skin and intestine in recipients of naive CCR6-deficient T cells. We found lower IFN-γ and interleukin 10 (IL-10) levels in skin and gut of these mice, and levels of the chemokines that control activated T-cell homing to skin and gut were also decreased. Our data suggest that by facilitating recently activated alloreactive effector CD4⁺ T-cell recruitment to target tissues, CCR6 plays a major role in GvHD development.

Mice with a targeted deletion of the CCR6 gene were generated¹⁰  and backcrossed onto a C57BL/6J background for 8 generations. Control animals were C57BL/6J WT littermates. B6.C-H2^bm1/By (bm1) and B6.C-H2-2^bm12/KhEg (bm12) C57BL/6 (H-2^b) mice were obtained from Jackson Laboratories (Bar Harbor, ME); CD1, C.B17/lcr-SCID, BALB/c, and BALB/cJ-SCID mice were from Harlan (Bicester, United Kingdom; and Gannat, France). CD1 WT or CCR6-deficient mice (generated in our laboratory by backcrossing onto a CD1 background for 8 generations) were analyzed to select H-2^d haplotype mice, using the following antimouse antibodies from BD PharMingen (San Diego, CA): H-2D^b (clone 28-14-8), H-2D^k (clone 15-5-5), H-2D^d (clone 34-2-12), H-2K^b (clone AF6-88.5), H-2K^k (clone AF3-12.1), H-2K^d (clone SF1-1.1), I-A^b (clone AF6-120.1), I-A^k (clone 11-5-2), and I-A^d (clone AMS-32). Animal procedures were approved by the Centro Nacional de Biotecnología Animal Care and Use Committee.

Spleens were removed aseptically, disrupted, and filtered through nylon mesh to separate aggregates. After erythrocyte lysis in buffered 0.83% (wt/vol) NH₄Cl solution (pH 7.2), samples were enriched for CD4⁺ or CD8⁺ cells by depleting B220⁺ and CD8⁺ or CD4⁺ cells, respectively, with Dynabeads (Dynal Biotech, Oslo, Norway). After staining with antimouse CD4 or CD8 and CD45RB antibodies (BD PharMingen), cells were positively selected as CD4⁺CD45RB^high and CD8⁺CD45RB^high populations (> 98% pure) in an Epics Altra sorter (Beckman Coulter, Hialeah, FL). CD8⁺ T-cell preparations (> 98% pure) were positively selected with anti-CD8 Dynabeads. T-cell–depleted (TCD) bone marrow (BM) cells were prepared from single-cell suspensions from femora and tibiae, which were depleted of CD4⁺ and CD8⁺ cells with specific Dynabeads.

For GvHD induction across an MHC class I or II point mismatch, CD4⁺CD45RB^high (3.5 × 10⁵) or CD8⁺ (2.5 × 10⁶) T cells from WT or CCR6-deficient C57BL/6J mice were injected via the tail vein into sublethally irradiated (6 Gy) age- and sex-matched C57BL/6J bm1 or bm12 recipients. For GvHD induction across MHC mismatches, CD4⁺CD45RB^high or CD8⁺CD45RB^high T cells (3.5 × 10⁵) from WT or CCR6-deficient C57BL/6J mice were injected via the tail vein into lethally irradiated (8 Gy) age- and sex-matched BALB/c mice. Mice received 4 × 10⁶ isogeneic TCD BM cells for reconstitution, alone or with allogeneic T cells from donors as indicated in “Lack of CCR6 activity in donor T cells protects MHC class II–mismatched recipient mice from GvHD without affecting GvHD development across an MHC class I barrier” and Figure 1. For GvHD induction across mhc mismatches, CD4⁺CD45RB^high (3 × 10⁵) T cells from WT or CCR6-deficient mice (H-2^d haplotype–selected CD1 strain) were injected into the tail vein of CB17-Prkdc^SCID/J (CB17/SCID) recipients. In all cases, recipients were maintained on acidified antibiotic water for the first 4 weeks after transplantation.

Animals were graded every 2 days for skin and hair changes, ear inflammation, hunched posture, diarrhea, and weight loss. As a measure of skin inflammation, ear thickness was determined using a dial thickness gauge (Mitutoyo, Kawasaki, Japan). Skin lesion development was assessed using a clinical score: 0, no lesions; 1, erythema and scaling on ear/head; 2, severe erythema and scaling on head and trunk; 3, pustules plus severe erythema at 2 or more sites.

CD4 T cells labeled with CMFDA (5-chloromethylfluorescein diacetate), CMTMR (4-chloromethyl benzoyl amino tetramethyl rhodamine), or CFSE (carboxyfluoresceindiacetate succinimidyl ester; Molecular Probes, Eugene, OR) were injected into recipient mice and their distribution was determined by flow cytometry in peripheral and mesenteric lymph nodes, as well as in spleen cell suspensions at different times after transfer. CFSE staining was also used to assess cell division as measured by sequential halving of fluorescence intensity with each generation.¹⁴  Flow cytometric analysis was performed in an EPICS XL cytometer (Beckman Coulter, Fullerton, CA), using fluorescein isothiocyanate (FITC)– and phycoerythrin (PE)–conjugated antimouse CD45, CD4, and CD8 monoclonal antibodies (BD PharMingen).

Ear skin was split into dorsal and ventral portions and cultured (24 h) in complete RPMI 1640 medium supplemented with IL-2. For histopathology, tissue samples were placed in 10% neutral-buffered formalin, paraffin embedded, sectioned with a microtome, and hematoxylin-eosin stained. Images were captured at × 25 magnification.

Real-time quantitative reverse transcription–polymerase chain reaction (RT-PCR) assays were used to determine cytokine and chemokine transcript amounts. RNA extraction and cDNA synthesis were performed as described.¹³  PCR primers were designed to amplify sequences spanning different exons using Primer Express software (PE Applied Biosystems, Foster City, CA) (Table 1). cDNA samples were amplified and their cycle threshold values determined on the ABI PRISM 7700 Detection System (PE Applied Biosystems) using SYBR green PCR core reagents. All cycle threshold values were normalized to β-actin and glyceraldehyde 3 phosphate dehydrogenase (GAPDH) housekeeping gene expression.¹⁵  Data were expressed as the x-fold induction of gene expression in treated mice compared with control mice.

We used 2 parallel models to study the role of CCR6 in GvHD development across MHC barriers. CD4⁺ or CD8⁺ T cells from C57BL/6J WT or CCR6-deficient donors were transferred to recipients that differed from donor mice at a single MHC class I (bm1) or class II (bm12) allele or to BALB/c recipients (MHC class I and II mismatched). Injection of WT CD4⁺CD45RB^high (naive) lymphocytes into bm12 mice produced acute GvHD and all mice had died by day 18 (Figure 1A). When naive T cells from CCR6-deficient donors were transferred, however, mortality onset was delayed and its incidence decreased significantly in recipient bm12 mice, as 37.5% remained alive 70 days after cell transplantation (ACT; Figure 1A; not shown). Weight loss at day 15 ACT was significantly lower in bm12 mice that received CCR6-deficient T cells than in WT cell recipients (Figure 1D).

In contrast, 100% of bm1 mice that received transplants of CD8⁺ T cells from WT or CCR6-deficient donors died 15 to 16 days ACT (Figure 1B). Similar weight loss was observed in both bm1 mouse groups by day 13 ACT (Figure 1E). Following control cell transfer of CD4⁺ naive T cells from WT or CCR6-deficient donors to bm1 mice, no alterations were detected in body weight or survival for 70 days ACT (Figure 1C,F).

The results were corroborated in a model of GvHD across MHC class I and II mismatches in which CD4⁺CD45RB^high or CD8⁺CD45RB^high T cells from C57BL/6J WT or CCR6-deficient donors were transferred into BALB/c mice, which also received 4 × 10⁶ isogeneic TCD BM cells for reconstitution. Animals were graded according to lesion development as follows: 1, hair loss; 2, hair loss and skin lesions in head and trunk; 3, hair loss, skin lesions, and diarrhea or weight loss of greater than 10%. Most mice (90%) that received naive WT CD4⁺ T cells developed GvHD symptoms (Figure 1G). Recipients of CCR6-deficient naive CD4⁺ T cells had significantly lower GvHD incidence (Figure 1G), with only 50% of mice showing severe GvHD symptoms during 4 months ACT (score 1.5 or 3) whereas the remainder were asymptomatic. Survival analysis confirmed these findings, as recipients of CCR6-deficient naive CD4⁺ T cells showed a statistically significant increase in survival compared with mice that received WT naive CD4⁺ T cells (Figure 1I).

In similar experiments, we transplanted naive CD8⁺ T cells from WT or CCR6-deficient donors but observed no significant differences between the 2 mouse recipient groups in survival (Figure 1H) or weight loss (Figure 1J). Both models thus confirmed impairment of the ability of CCR6-deficient naive T cells to cause GvHD across an MHC class II but not a class I barrier.

We used a nonconditioned model to prevent irradiation-induced alterations in cytokine expression and to facilitate tracking of CD4⁺ T-cell trafficking and proliferation. CD4⁺CD45RB^high T cells from WT or CCR6-deficient mice were transferred to MHC-matched/mhc-mismatched CB17/SCID recipient mice, which developed a wasting syndrome associated with splenomegaly, erythrosquamous skin lesions, and intestinal inflammation, as reported.¹⁶  Ear thickness in recipient mice was evaluated at several time points as an overall measure of skin inflammation, and skin lesions were scored clinically. Weight loss and diarrhea were monitored as an indication of intestinal damage. Survival curves were established including animals that died and those killed when they reached final GvHD stages (weight loss > 20%, grade 3 in skin lesions and inflammation, and cachexia and/or diarrhea).

As early as 19 days ACT, ear inflammation increased significantly (46%) in WT naive T-cell recipients (0.370 ± 0.022 mm) compared with controls that did not receive transplants (0.261 ± 0.006 mm) (Figure 2A). Subsequent damage to ear tissue in WT naive T-cell recipients prevented registering further increases in ear inflammation. By 19 days ACT, only a slight increase in ear thickness was observed in T-cell recipients from CCR6-deficient donors (0.305 ± 0.024 mm), 17% higher than that of controls (Figure 2A). Inflammation increased with time, peaking at days 30 to 35 ACT.

Skin lesions were first detected on ears and head, extending to trunk and feet; progression was rapid and homogeneous in WT T-cell recipients (Figure 2B). Lesion onset was apparent 3 weeks ACT (day 23-24); at 5 weeks, all mice showed the maximum clinical score, with ears, head, trunk, and feet affected. Pustules were observed on the trunk at later stages. By 3 weeks ACT, T-cell recipients from CCR6-deficient donors had only developed mild ear inflammation. Skin lesions were not observed until 6 weeks ACT; furthermore, only 50% of the population was affected (Figure 2B). Disease was progressive thereafter and 80% of the population reached the maximum score by 9 weeks ACT (day 62). Trunk lesions were less severe in this group than those developed by WT T-cell recipients. Histopathologic analysis of dorsal skin samples confirmed that leukocyte infiltrate and tissue damage were more prominent in T-cell recipients from WT donors (Figure 2H).

Differences were also observed in gut damage. By day 28 ACT 50% of WT T-cell recipients showed diarrhea, whereas symptoms of intestinal inflammation were not observed in recipients from CCR6-deficient donors until day 34 ACT; even by day 51, only 50% were affected (Figure 2C). WT T-cell recipients showed a mean 10% weight loss at 3 weeks ACT, whereas similar loss was observed in T-cell recipients from CCR6-deficient donors only at 6 weeks ACT (Figure 2D). Survival analysis data confirmed the results from the conditioned animal models, as 85% of WT T-cell recipients were dead by day 40 ACT, whereas 56% of CCR6-deficient T-cell recipients remained alive approximately 3 months ACT (Figure 2E).

In control experiments in which BALB/cJ donor naive T cells were transferred to CB17/SCID recipients (strains that differ only in the immunoglobulin heavy chain complex¹⁷ ), there was no apparent disease progression for 7 weeks, as we observed no weight loss (Figure 2F), diarrhea, or skin lesions (not shown). Recipients showed discrete ear inflammation by day 21 ACT (0.306 ± 0.03 mm; 17.9% increase over controls that did not receive transplants), with a maximum at days 30 to 35 ACT (0.343 ± 0.015 mm; 30% increase; Figure 2G).

All together, the results show that symptoms appeared later and were milder when donor T cells were derived from CCR6-deficient mice.

The more severe pathology in WT T-cell recipients could be due to a simple mass effect, as increasing the donor T-cell inoculum in a transplantation leads to more severe GvHD, with greater morbidity and mortality (Liu et al¹⁸  and R.V.'s unpublished observations, April 2004). As similar numbers of WT or CCR6-deficient donor T cells were used, mass differences may have been generated by more rapid in vivo expansion of WT cells. To test the CCR6 requirement for CD4⁺ T cells to reach peripheral lymphoid organs, we cotransferred CMFDA-labeled WT and CMTMR-labeled CCR6-deficient CD4⁺ T cells (3 × 10⁶ cells each) to BALB/c-SCID mice. Flow cytometric analysis of splenocytes at 19 hours ACT showed similar percentages of WT and CCR6-deficient CD4⁺ T cells had reached the spleen; by 44 hours, most cells from both donors had progressed through at least one division (Figure 3A).

In another experiment, CFSE-labeled CD4⁺ T cells from WT or CCR6-deficient donors showed robust cell division in BALB/c-SCID mice 4 days ACT, as indicated by the sequential halving of the fluorescence intensity with each cell generation observed in spleen and lymph nodes from recipients of both types of allogeneic CD4⁺ T cells (Figure 3B). These data suggest that neither CD4⁺ T-cell infiltration of lymphoid organs nor allogeneic stimuli-driven lymphocyte proliferation is CCR6 dependent.

Engraftment kinetics of CD4⁺ T cells in recipient mice differed depending on donor genotype. Similar spleen engraftment was observed in week 1 ACT, independent of donor CD4⁺ T-cell origin (Figure 4A). Subsequently, the spleen CD4⁺ T-cell pool increased rapidly in WT T-cell recipients (Figure 4A) and was maximal approximately 3 weeks ACT (12- to 14-fold over control mice that received isogeneic transplants). In agreement with the acute wasting GvHD effects observed, 4 weeks ACT the cell-pool size had already decreased. In contrast, CCR6-deficient T-cell recipients showed a 2- to 4-fold cell-pool increase during approximately 1 month ACT and, in surviving mice, spleen CD4⁺ pool size was 10-fold over that in controls. Percentages and absolute numbers of blood CD4⁺ T cells were also higher in WT T-cell recipients (Figure 4B). The data suggest that although proliferative responses of CD4⁺ T cells from CCR6-deficient donors are unimpaired, CCR6 interactions improve the long-term expansion of these donor-derived cells in allogeneic recipient mice. The spleen cytokine pattern was similar in recipients of WT and CCR6-deficient T cells during the first month ACT. We observed increases in IFN-γ (8- to 16-fold) and IL-10 (4- to 8-fold) transcript levels, whereas IL-4 and IL-12 levels were as in controls (Figure 4C). By day 50 ACT, the CCR6-deficient T-cell recipients still alive (64%) showed increases in IL-4 (32- to 64-fold) and IL-12 (16-fold); IFN-γ and IL-10 levels were also increased in these animals. This new cytokine equilibrium in these mice may account for their attenuated GvHD symptoms and prolonged survival.

We studied the time course of alloreactive T-cell infiltration in skin, small intestine, liver, and colon. CD45⁺ and CD4⁺ T cells were increased in the skin of mice that received transplants of naive T cells from WT or CCR6-deficient donors compared with controls that did not receive transplants (Figure 5A). During the early period of acute GvHD there was a progressive increase in the number of CD45⁺ cells infiltrating recipient mouse ears, which was more rapid in WT T-cell recipients. Infiltration of CD4⁺ T cells from CCR6-deficient donors was significantly delayed, as a 100-fold increase was observed at 11 days ACT in explants from recipients of WT T cells compared with mice that did not receive transplants, whereas a similar increase was not found in T-cell recipients from CCR6-deficient donors until 4 weeks ACT (Figure 5A). During the asymptomatic period (∼3 weeks for WT T-cell recipients), approximately 65% of CD45⁺ cells were CD4⁺ T cells; by day 28 ACT, when disease symptoms were observed, this population represented only 30% of total CD45⁺ cells, although the absolute number of CD4⁺ T cells did not differ significantly. This is probably a consequence of later infiltration of other leukocyte subpopulations to ear skin, as described for monocytes in scleroderma GvHD models.¹⁹  In T-cell recipients from CCR6-deficient donors, CD4⁺ T-cell ear infiltrates were delayed and were not detected until day 19 ACT; by day 28, when no symptoms other than ear inflammation were observed, CD4⁺ T cells represented approximately 56% of total CD45⁺ cells (Figure 5A). This difference in the degree of leukocyte infiltration concurs with observations in dorsal skin (Figure 2H).

We analyzed alloreactive CD4 T-cell infiltration in small intestine by estimating CD4 transcript levels using real-time quantitative RT-PCR. CD4 mRNA levels increased rapidly in the gut of WT T-cell recipients compared with T-cell recipients from CCR6-deficient donors; gut CD4 mRNA levels were not equivalent until 4 weeks ACT (Figure 5B).

CD4 mRNA levels in recipient liver showed a 6-fold increase in WT T-cell recipients by day 8 ACT, which increased to 12-fold by 19 days ACT. Conversely, at 28 days ACT only a 2-fold increase in CD4 mRNA was observed in CCR6-deficient T-cell recipients. In WT T-cell recipient colon, we observed a 10-fold increase in CD4 mRNA by day 19 ACT. No variations were detected in colons of CCR6-deficient T-cell recipients until days 25 to 28 ACT, when we found only a 3-fold increase.

Cytokines such as IFN-γ and IL-10 have a central role in murine GvHD models and in human patients.^4,20  IFN-γ mRNA expression in the skin of WT T-cell recipients was 23-fold and 200-fold higher by days 8 and 21 ACT, respectively, compared with controls that did not receive transplants. In contrast, only an approximately 16-fold increase was observed in skin IFN-γ transcript levels in T-cell recipients from CCR6-deficient donors at 4 weeks ACT (Figure 6A). Skin IL-10 mRNA expression followed a similar pattern; an 8-fold increase was seen 8 days ACT in WT T-cell recipients, which increased to 20-fold by 3 weeks ACT, whereas similar IL-10 mRNA levels were not found in recipients of CCR6-deficient T cells until 5 weeks ACT (Figure 6A).

Chemokines contribute to alloantigen-primed CD4⁺ T-cell homing to target tissues, leading to cell activation and cytokine production; this enhances expression of other chemokines that attract activated lymphocytes and other leukocytes to target sites as the immune response expands. The β-chemokines CCL17, CCL20, and CCL27 are expressed in pathologic and nonpathologic skin and are considered important elements controlling memory CD4⁺CLA⁺ T-cell homing to skin.^8,21,22  CCL20 and CCL27 were modulated during the GvHD development period; skin CCL20 mRNA expression was higher in recipients of WT T cells than in controls (∼16-fold by day 8; ∼200-fold by day 19 ACT), after which expression decreased as skin lesions appeared and their severity augmented (Figure 6A). A discreet CCL27 mRNA expression increase was detected in WT T-cell recipients during very early stages (days 1-8), which was subsequently down-regulated (Figure 6A). CCL17 mRNA expression was unaltered during GvHD development (not shown).

IFN-γ stimulates different cell types to produce the α-chemokines CXCL9, CXCL10, and CXCL11,²³  which are potent chemoattractants for activated T cells and may thus be critical for directing T-cell infiltration into target sites. Increased levels of CXCL10 and the β-chemokine CCL3 were reported in autologous GvHD skin lesions.¹⁹  Compared with controls, skin CXCL9, CXCL10, and CXCL11 mRNA levels were up-regulated rapidly in WT T-cell recipients (∼80- to 90-fold) by 8 days ACT and reached 200- to 300-fold levels by 3 weeks ACT (Figure 6A). mRNA levels of CCL3 and CCL5, 2 activated T-cell– and monocyte-attractant chemokines, were up-regulated similarly in these mice (∼20-fold by day 8 ACT; ∼200-fold by 4 weeks ACT; Figure 6A).

The situation clearly differed in T-cell recipients from CCR6-deficient donors, as skin chemokine mRNA increases were lower and delayed. For CCL20, a moderate increase was observed (1.5- to 9-fold from days 8-34 ACT), clearly lower than the levels seen in recipients of WT T cells (Figure 6A). The discreet increase in CCL27 transcripts was similar in T-cell recipients from WT or CCR6-deficient donors, although peak expression appeared 2 weeks later in the latter group (Figure 6A). Compared with animals that did not receive transplants, T-cell recipients from CCR6-deficient donors showed only an approximately 10-fold increase in CXCL9, CXCL10, and CXCL11 mRNA levels at 4 weeks ACT, at which time CCL3 and CCL5 transcript levels had increased approximately 50-fold and approximately 5-fold, respectively, clearly lower than the levels in WT T-cell recipients (Figure 6A).

We analyzed cytokine and chemokine transcript levels in small intestine. Gut IFN-γ, IL-10, CCL3, and CXCL10 mRNA expression was up-regulated after GvHD induction in both T-cell–transferred mouse groups compared with controls. Gut mRNA increases during early GvHD were faster and reached higher values in WT T-cell recipients (Figure 6B). Transcript levels of CCL25, another chemokine involved in T-cell homing to small intestine,²⁴  were nonetheless unaltered by GvHD progression (not shown). We detected no increase in IL-4 transcript levels over controls, either in skin or small intestine (Table 2).

Similar results were observed in liver and colon. In WT T-cell recipient liver, IFN-γ mRNA levels increased progressively (4-fold and 8-fold by days 5 and 19 ACT, respectively). Consequently, 2- to 6-fold increases in CXCL9, CXCL10, and CXCL11 transcript levels were also observed in these livers soon after cell transplantation (day 5 ACT). Conversely, no increase in the levels of IFN-γ or the 3 CXC chemokines were observed in livers from CCR6-deficient T-cell recipients in the first month ACT. Colon from WT T-cell recipients showed increased IFN-γ mRNA levels ranging from 3-fold to 30-fold by days 5 and 19 ACT, respectively. In these colons, CXCL9, CXCL10, and CXCL11 mRNA levels also increased from 2- to 4-fold (day 5) to 20-fold (day 19). Again, the situation differed in CCR6-deficient T-cell recipient colon, in which a 9-fold increase in IFN-γ levels was not observed until days 25 to 28 ACT; this was paralleled by a 3- to 4-fold increase in levels of the 3 CXC chemokines.

Involvement of chemokines and their receptors in GvHD was reported for CCL3, CCL5, CCL21, CCR5, and CXCR3.^19,25-29  Here we show a critical role for CCR6 in GvHD development across an MHC class II but not a class I barrier, as demonstrated by prolonged survival of bm12 and BALB/c mouse recipients of naive CD4⁺ T cells from CCR6-deficient donors compared with WT donors. In contrast, CD8⁺ T cells from CCR6-deficient donors did not reduce recipient mortality. The importance of CCR6 in the context of an MHC class II but not a class I mismatch concurs with preferential CCR6 expression in CD4⁺ but not CD8⁺ T cells in both man and mouse.^9,30,31  Certain effector/memory T-cell subsets, including skin- and gut-homing T cells (almost all of which express CCR6), are also the best responders to CCL20-mediated migration and Ca²⁺ mobilization.⁹ 

Our data also show that transfer of CD4⁺ T cells from CCR6-deficient donors to MHC class I– and II–matched mhc-mismatched recipients markedly ameliorates GvHD symptoms in skin and gut. Weight loss, ear inflammation, skin lesions, and diarrhea appeared later and were less severe in recipients of CD4⁺ T cells from CCR6-deficient donors than from WT donors. Differences in the size of the donor alloreactive CD4⁺ T-cell pool generated in recipients, which could be CCR6 dependent, may explain the disease outcome observed in recipient mice. Alloreactive T-cell pool size may vary when there is impaired donor cell recruitment to host secondary lymphoid organs, an altered response or proliferation to alloantigenic activation and/or impaired recruitment of alloreactive cells to target tissues. Analysis of these possibilities showed that CD4⁺ T-cell homing to spleen and lymph nodes is CCR6 independent, as is also the case for allogeneic-driven short-term proliferation of CD4⁺ T cells in secondary lymphoid organs. CCR6-deficient T-cell proliferation is unimpaired in response to TCR stimulation in vitro or to alloantigens in an MLR assay.¹⁰  Lymphopenia-driven homeostatic proliferation is also CCR6 independent, as confirmed by our results from C57BL/6 nu/nu reconstitution experiments (not shown). Although all proliferation-related parameters analyzed were not affected by CCR6 deficiency, the size of the donor alloreactive T-cell pool in spleen was significantly reduced from day 7 ACT in T-cell recipients from CCR6-deficient donors compared with WT donors. This suggests that long-term expansion of CCR6-deficient CD4⁺ T cells is reduced in host secondary lymphoid organs. Analysis of alloreactive CD4⁺ T-cell homing to gut and skin also showed a clear recruitment delay in T-cell recipients from CCR6-deficient mice.

Our results suggest altered recruitment of alloreactive CD4⁺ T cells from CCR6-deficient donors to target organs. In the early inflammation phases, tissue sites and lymph nodes recruit CD4⁺ memory T cells, and recruitment to lymph node may be particularly important for a recall response. Whereas lymphocyte receptors for proinflammatory chemokines such as CCR2, CCR5, and CXCR3 are functional only after cell activation, CCR6 mediates chemotaxis and calcium release in freshly-isolated effector/memory T cells.⁹  This characteristic of CCR6 may provide the mechanism for recruitment of resting cells to activation sites. The data are compatible with a role for CCL20/CCR6 in the first stage of GvHD inflammation, when this chemokine axis contributes to T-cell recruitment to target tissues. This leads to cell activation, IFN-γ and IL-10 production, and eventually enhances expression of chemokines that attract leukocytes to target tissue as the immune response expands.

Considerable evidence supports this hypothesis. In WT naive T-cell recipients, the skin chemokine expression pattern in early GvHD showed that whereas CCL27 mRNA increased only moderately, there was a strong CCL20 mRNA up-regulatory response in the first 2 to 3 weeks ACT, which disappeared concomitantly with the onset of GvHD symptoms. This concurs with reports on CCL27 and CCL20 function in homeostatic lymphocyte trafficking to skin and inflammatory pathologies.^21,31-33  Although there may be some overlap in chemokine activity, individual roles are also clear, since keratinocyte CCL20 production is greatly enhanced by proinflammatory cytokines and T-cell–derived factors.⁸  The skin GvHD data reported here concur with the altered contact and delayed-type hypersensitivity responses in CCR6-deficient mice,¹⁰  suggesting a critical role for CCR6 in cutaneous inflammation.

CCL20/CCR6 also has a clear role in the gut. In both man and mouse, CCL20 is constitutively expressed in mucosa-associated lymphoid tissues.^34-36  It is up-regulated after lipopolysaccharide, IL-1, or tumor necrosis factor α injection^37,38  and in chronically inflamed colon in murine inflammatory bowel disease models.³⁹  Lymphocyte homeostasis is impaired in intestinal mucosa of CCR6-deficient mice,^10,12  whose response to induced intestinal inflammation differs from that of WT mice.¹³ 

In WT naive T-cell recipients, skin and intestine showed marked up-regulation of IL-10 and IFN-γ transcript levels; these early IFN-γ increases are critical for GvHD development.^4,20,40,41  Consistent with this, the IFN-γ–induced chemokines CXCL9, 10, and 11 showed strong, prolonged mRNA up-regulation; some of these potent T-cell chemoattractants may also stimulate T-cell proliferation and effector cytokine production.⁴²  During GvHD, transcript levels of CCL3 and CCL5, 2 chemoattractants of activated T cells and monocytes, also increased significantly in target tissues. Conversely, chemokine transcript up-regulation in intestine and skin of T-cell recipients from CCR6-deficient donors was lower and delayed compared with WT T-cell recipients. Although skin IL-10 and IFN-γ mRNA levels were similar in both recipient groups at approximately 1 month ACT, the kinetics of transcript increase was significantly delayed in T-cell recipients from CCR6-deficient donors. Nonetheless, gut IL-10 and IFN-γ mRNA levels in these mice had not reached levels similar to those in WT T-cell recipients by 1 month ACT. During this first month ACT, we detected no differences in either recipient mouse group over control IL-4 transcript levels in the GvHD target organs or spleen. These data concur with the significantly delayed infiltration of CD45⁺ and CD4⁺ cell subsets to skin and small intestine of T-cell recipients from CCR6-deficient donors.

Further work is needed to define the precise mechanism by which CCR6 exerts its critical role in GvHD development. Cell surface CCR6 expression increases in CD4⁺ T lymphocytes during MLR, an in vitro culture system that activates naive and resting memory T cells, mimicking physiologic conditions.¹¹  Unlike other chemokine receptors whose expression is also modulated in MLR, CCR6 up-regulation is uncoupled from initiation of cell division and does not increase CCL20-mediated cell responses. CCL20 binding to CCR6 may thus be needed to trigger cellular responses other than chemotaxis, such as integrin activation.¹¹  CCL20 induces rapid adhesion of freshly isolated T-cell subsets to intercellular adhesion molecule-1,⁴³  and CCL20/CCR6 signaling is critical for memory T-cell adhesion to endothelial cells prior to extravasation.⁴⁴ 

Our data suggest that the delayed, milder GvHD symptoms in T-cell recipients from CCR6-deficient donors are not due to lack of donor cell engraftment; rather, CCR6 seems to have a major role in GvHD development by facilitating recruitment of alloreactive effector/memory CD4⁺ T cells to target tissues. GvHD is a complex pathologic process that is unlikely to be controlled by a single agent; our results indicate that CCR6 may be a targetable gene for GvHD treatment using molecules able to block the activity of this chemokine receptor.

We thank T. Cuevas and D. Esteban for help with animal work, M. C. Ortiz for advice in cell sorting, and C. Mark for editorial assistance.