Host Foxp3⁺CD4⁺ Regulatory T Cells Act as a Negative Regulator of Dendritic Cells in the Peritransplantation Period Free

Allogeneic hematopoietic stem cell transplantation (SCT) has improved long-term survival in patients with hematologic malignancies. However, graft-versus-host disease (GVHD) is a major obstacle to successful allogeneic bone marrow transplantation (BMT) (1), and greatly limits the applications and efficacy of allogeneic SCT. As one of the therapeutic methods for controlling GVHD, donor Foxp3⁺CD4⁺ regulatory T cells (Tregs) were studied extensively (2–8), and these cells have begun to be clinically applied to HLA-haploidentical SCT (9, 10). On the other hand, less attention has been focused on host Foxp3⁺CD4⁺ Tregs.

The fact that host Tregs regulate GVHD was shown from two studies on myeloablative murine BMT models (2, 11). Taylor et al. (2) described that in vivo CD25 depletion of the recipient before transplantation resulted in the acceleration of acute GVHD responses, in which anti-CD25 mAb-treated thymectomized B6 mice were lethally irradiated and received transplants of BALB/c BM and spleen cells. Furthermore, Anderson et al. (11) described, in B10.D2 (H-2^d) → BALB/c (H-2^d), an MHC-compatible, multiple minor histocompatibility Ag-incompatible murine BMT model of chronic GVHD, that host CD4⁺CD25⁺ cells surviving lethal radiation ameliorated chronic GVHD, using RAG^−/− mice as recipients. These studies described alleviation of the clinical symptoms of GVHD and longer survival times because of the survival of host Tregs, but did not provide cell-based data on host Tregs. Furthermore, Bayer et al. showed, in an analysis of cell kinetics 2 wk after BMT, that residual host CD4⁺CD25⁺Foxp3⁺ Tregs survived and even expanded in syngeneic or T cell–depleted MHC-matched allogeneic BMT. In contrast, in T cell–replete allogeneic BMT, no host Tregs were observed in spleen or lymph nodes at 4–5 wk after transplantation (12). Because experimental BMT studies suggested the role of host Tregs in regulating GVHD, we focused on the cell kinetics of host Tregs in the earlier phase of allogeneic SCT.

To clarify the in-depth mechanism of host Tregs for suppressing GVHD, we established a murine MHC-haploidentical BMT model (BDF1 (H-2^b/d) → B6C3F1 (H-2^b/k)), in which transplantation following conditioning with high-dose (13 Gy) or low-dose (5 Gy) total body irradiation (TBI) corresponds to myeloablative SCT (MAST) or reduced-intensity SCT (RIST). All MAST recipients died of GVHD within 70 d, whereas RIST recipients developed almost no GVHD and survived for at least 3 mo. Using this model, comparing with MAST and RIST recipients with or without anti-CD25 mAb treatment before transplantation, we focused on the kinetics of donor conventional T cells, host Tregs, and host dendritic cells (DCs) in the mesenteric lymph nodes (MLNs), as this transplant model showed prominent gut GVHD. Consequently, we found that host Tregs restrained the maturation of host DCs in lymph node organs, thereby dampening the ability of DCs to activate donor conventional T cells and to reduce the magnitude of graft-versus-host reaction.

Female B6C3F1 (C57BL/6 × C3H, H-2^b/k) and BDF1 (C57BL/6 × DBA/2, H-2^b/d) mice were purchased from Japan SLC (Hamamatsu, Japan). CD45.1⁺ C57BL/6 congenic mice were obtained from the Jackson Laboratory (Bar Harbor, ME). To discriminate donor and host cells, we established congenic CD45.1⁺ recipient mice by crossing CD45.1⁺ C57BL/6 congenic mice with C3H (H-2^k) mice. The luciferase-expressing β-actin-luc FVB/N mice (H-2^q) purchased from Taconic (Hudson, NY) were backcrossed onto the C57BL/6J-Tyrc-2J (H-2^b). After the line was crossed with DBA/2 (H-2^d), female heterozygous luc offspring (H-2^b/d) were used for the donor graft in bioluminescence imaging experiments. Recipient mice were 8–12 wk old. All mice were maintained in our specific pathogen-free facility and treated in accordance with the guidelines for animal care approved by Hyogo College of Medicine.

Irradiation was performed with an X-ray unit at a rate of 0.86 Gy/min with a 1-mm Cu filter. Recipient B6C3F1 mice received TBI 13 Gy in two separate doses or 5 Gy as a single dose. On the following day, recipients received T cell–depleted (TCD) bone marrow (BM) cells (5 × 10⁶) and splenocytes (2 × 10⁷) from BDF1 mice via the tail vein. Animals were given acidified water supplemented with antibiotics for 2 wk following transplantation.

bone marrow cells were collected by flushing femurs and tibias from donor mice. For T cell depletion, after red blood lysis, cells were incubated with CD90.2 magnetic beads and were depleted of CD90.2⁺ T cells using the autoMACS system (Miltenyi Biotec, Bergisch Gladbach, Germany).

CD4⁺CD25⁺ T cells were isolated from spleens (>95% purity) using the regulatory T cell isolation kit (Miltenyi Biotec) and autoMACS system, with some modifications. In brief, non-CD4⁺ T cells were labeled with a mixture of biotin-conjugated Abs and anti-biotin magnetic beads. In parallel, cells were labeled with PE-conjugated CD25 mAb. CD4⁺ T cells were enriched by negative selection, followed by positive selection of CD4⁺CD25⁺ T cells using anti-PE–coated magnetic beads.

Splenic DCs were isolated from predigested spleen cells with collagenase D (2 mg/ml) and DNase1 (100 μg/ml; Roche Diagnostics, Indianapolis, IN). After stirring in digestion buffer at 37°C for 45 min, cell suspension was incubated with anti-CD11c–coated magnetic beads and then CD11c⁺ splenic DCs were isolated using the auto MACS system.

Survival was monitored daily, and clinical GVHD scores were assessed weekly by a scoring system, previously reported by Cooke et al. (13). In brief, changes in five clinical parameters (weight loss, posture, fur texture, skin integrity, and activity) were monitored and summed as the GVHD score (maximum score = 10). Samples of skin, liver, and intestine were fixed in 10% formalin, embedded in paraffin, sectioned, mounted on microscope slides, and stained with H&E. Histologic images were captured using a Nikon E600 microscope with a Ds-Fi1-U2 digital camera. Pathologic GVHD scores of the samples were measured with a scoring system, reported by Chen et al. (14).

Tissues were fixed in 3% buffered paraformaldehyde for 30 min, embedded in OCT compound (Sakura Finetek, Tokyo, Japan), cut into 5-μm sections, and mounted on microscope slides. Slides were treated with a biotin blocking system (Dako, Glostrup, Denmark), followed by preincubation with Block-ace (Dainippon Sumitomo Pharma, Osaka, Japan) containing purified rat anti-mouse CD16/32 mAb for 1 h at room temperature to block nonspecific binding. The sections were then stained with FITC-conjugated anti-CD4 and Cy5-conjugated anti-CD8α mAbs (53-6.7), or FITC-conjugated anti-CD11c and primary rat anti-Foxp3 mAbs (FJK-16s) followed by TRITC-conjugated anti-rat IgG Ab. Some sections were stained with biotinylated anti-CD45.1 mAb (A20), primary rat anti-Foxp3, and rabbit polyclonal anti-Ki-67 (ab15580) Abs followed by FITC-conjugated streptavidin, TRITC-conjugated secondary anti-rat IgG, and Cy5-conjugated secondary anti-rabbit IgG Abs. Slides were mounted with Prolong gold antifade reagent containing DAPI (Life Technologies). Triple-colored images were taken using a Zeiss LSM 510 Meta Confocal Microscope (Carl Zeiss, Jena, Germany) and Zeiss LSM 510 software, and images were composed using Adobe Photoshop CS5.

Cells were incubated with anti-CD16/32 (2.4G2) before Ab staining to block nonspecific binding. The following reagents or anti-mouse mAbs were purchased from BD Biosciences (San Jose, CA) or eBioscience (San Diego, CA): FITC-, PE-, PerCP-Cy5.5-, APC-conjugated anti-mouse H-2Kd (SF1-1.1), H-2Kk (36-7-5), I-A^k (10-3.6), CD45.1 (A20), CD62L (MEL-14), CD25 (7D4), CXCR3 (CXCR3-173), CD80 (16-10A1), CD86 (PO3.1), CD4 (RM4-5), CD8b (H35-17.2), CD11c (N418), CD44 (IM7), Foxp3 (FJK-16s), BrdU, Annexin V, and isotype control IgG. 7-Aminoactinomycin D (7-AAD) was used to exclude dead cells. Four-color flow cytometric analysis was performed using FACSCalibur (BD Biosciences). FlowJo software (Tree Star, Ashland, OR) was used for data analysis.

Intracellular IFN-γ staining was performed using the BD Cytofix/Cytoperm Fixation/Permeabilization kit (BD Biosciences), with some modification. In brief, cells isolated from MLNs were stimulated with PMA (20 ng/ml) and ionomycin (2 μg/ml) for 4 h at 37°C in complete RPMI 1640 medium containing Brefeldin A (3 μg/ml). Cells were washed, stained with Abs against surface markers, fixed and permeabilized in permeabilization buffer, and stained with PE-conjugated anti–IFN-γ mAb (XMG1.2). In some experiments, cells were permeabilized and stained for intracellular Foxp3 using APC-conjugated anti-Foxp3 mAb (FJK-16s) and an intracellular staining kit according to the instructions provided by the manufacturer (eBioscience).

To track the division of adoptively transferred cells, isolated BDF1 splenocytes were stained in vitro with 5 μM CFSE (Life Technologies) followed by incubation for 8 min at room temperature. The labeling reaction was quenched by washing in PBS supplemented with 5% FCS. CFSE-labeled BDF1 splenocytes (2 × 10⁷) were injected i.v. with 5 × 10⁶ BDF1 TCD-BM cells into recipient mice. Four days later, T cells were isolated from MLNs, and cell division was tracked based on the decrease in CFSE intensity that was analyzed with flow cytometry and FlowJo software (Tree Star). To detect the proliferation of host Foxp3⁺CD4⁺ Tregs, mice were administered 1 mg of the thymidine analog BrdU i.p., and cells were harvested from MLNs 20 h later. Cells were stained for cell surface markers, followed by incubation in fixation and permeabilization buffer (BD Biosciences), and costained for intracellular BrdU and Foxp3 using a BrdU flow kit (BD Biosciences) according to the instructions to allow Foxp3 costaining.

CD4⁺CD25⁺ T cells freshly isolated from spleens of B6C3F1 mice were incubated for 3 d in the presence of IL-2 (10 ng/ml) and immobilized anti-CD3/CD28 mAb (2.5 μg/ml each). B6C3F1 mice underwent transplantation using TCD-BM cells (5 × 10⁶) and splenocytes (2 × 10⁷) from BDF1 mice with 5 × 10⁵ cultured recipient CD4⁺CD25⁺ T cells after 13 Gy irradiation.

Total RNA was isolated from the intestine using TRIzol reagent (Life Technologies) and was reverse-transcribed using the SuperScript III First-Strand Synthesis System (Life Technologies) according to the manufacturer’s instructions. The methods and primers for quantitative PCR of TLR4 or TLR9 cDNAs (15) and of TNF-α or IL-1β cDNAs (16) were described previously. Real-time PCR was performed with the ABI 7500 Real Time PCR System using SYBR Green ER qPCR SuperMix Universal (Life Technologies). Gene expression levels were normalized relative to the expression levels of the GAPDH gene and were expressed as values relative to the expression levels in normal mice using the ∆∆Ct method.

CD11c⁺ DCs (4 × 10⁴) isolated from B6C3F1 spleens were cultured in 96-well round-bottom plates in the presence of TNF-α (20 ng/ml) or LPS (1 μg/ml), with or without host (H-2K^d−) CD4⁺CD25⁺ T cells (8 × 10⁴) that were sorted from the spleens of RIST recipients on day 4. After 2-d culture, cells were collected and treated with 2 mM EDTA. Cells were stained with FITC-conjugated anti–I-A^k, PE-conjugated anti-CD80, or anti-CD86, and APC-conjugated anti-CD11c and 7-AAD after FcR blocking. The level of CD80/CD86 expression on cultured DCs was analyzed using four-color flow cytometry.

Host Tregs (H-2Kd⁻CD4⁺CD25⁺) were purified from day 4 spleens of RIST recipients using the autoMACS system. Responder (CD4⁺CD25⁻) cells (8 × 10⁴) that were freshly isolated from BDF1 (donor) spleens were cocultured, in triplicate, with stimulators (irradiated B6C3F1 spleen cells: 4 × 10⁵) together with soluble anti-CD3 mAb (0.5 μg/ml). In this MLR system, host Tregs were added to the culture at various ratios to the responder cells. Cultures were pulsed with ³H-thymidine (1 μCi/well) for the final 6 h of incubation and harvested after culture for 72 h in total. In some experiments, CD4⁺CD25⁻ T cells (8 × 10⁴) isolated from BDF1 spleens were labeled with 1 μM CFSE and cultured with B6C3F1 splenic CD11c⁺ DCs (8 × 10³) and soluble anti-CD3 mAb (2 μg/ml) for 3 d, with or without host-type CD4⁺CD25⁺ cells (8 × 10⁴) that were purified from day 4 spleens of the RIST recipients in the presence of TNF-α (20 ng/ml) or LPS (1 μg/ml). Proliferation was measured by the dilution of CFSE intensity using FlowJo software. In addition, freshly purified splenic DCs from B6C3F1 mice were activated with LPS or in the presence or absence of host-type Tregs that were preactivated with IL-2 and immobilized anti-CD3/28 mAbs for 48 h. DCs were purified after the coculture for 24 h by negative selection of CD3⁺ Tregs using PE-conjugated anti-CD3 mAb, followed by anti-PE coated magnetic beads. These purified DCs (2 × 10⁴) were replaced at the designated DC/responder ratio with freshly isolated CD4⁺CD25⁻ responder T cells (2 × 10⁵) from BDF1 spleens. Cultures were pulsed with ³H-thymidine for the final 18 h of incubation and harvested after 96 h in total. For in vitro MLR assay using transwell (17) to determine whether host Tregs suppress alloresponse in a contact-dependent manner or by soluble mediators, freshly isolated CD4⁺CD25⁻ responder conventional T cells (Tcons) (8 × 10⁴) from BDF1 spleens were cocultured in transwell plates (HTS transwell-96 with a 0.4-μm pore; Corning) with irradiated (30 Gy) B6C3F1 spleen cells (4 × 10⁵) in the presence of soluble anti-CD3 mAb (0.5 μg/ml) together with or without host-type H-2K^d− CD4⁺CD25⁺ Tregs (8 × 10⁴) purified from spleens of the RIST recipients on day 4 after transplantation. Cultures were pulsed with ³H-thymidine (1 μCi/well) for the final 18 h of incubation and harvested after 96 h in total. In addition, anti–IL-10 (clone: JES5-2A5, 20 μg/ml) or anti–TGF-β (clone: 1D11.16.8, 20 μg/ml) neutralizing Abs were added in the indicated cultures.

Recipient B6C3F1 mice were thymectomized and injected i.p. with 0.4 mg anti-CD25 mAb (clone: PC61) four times on days −27, −24, −21, and −14.

B6C3F1 recipient mice were coinjected i.v. with splenocytes from F1 offspring of β-actin-luc B6; FVB/N × DBA/2 in addition to TCD-BM cells from wild-type B6D2F1. Recipient mice were anesthetized with isoflurane, and an aqueous solution of d-luciferin (150 mg/kg) was injected i.p. 10 min prior to imaging. In vivo imaging was performed using an IVIS Lumina II (Caliper Life Sciences, Hopkinton, MA). Mice were placed into the light-tight chamber of the CCD camera system. Photons emitted from luciferase-expressing cells within the body were quantified over a period of 1 min using Living Image Software (Caliper Life Sciences). In select experiments, animals were killed humanely after in vivo imaging. Selected tissues were placed in the culture medium containing d-luciferin for 10 min and imaged for 1 min.

Freshly isolated whole B6C3F1 spleen cells were digested with collagenase D and DNase 1 (Roche), and were cultured for 48 h in the presence of IL-2 (10 μg/ml) and GM-CSF (10 μg/ml) after exposure with titrated doses of radiation (0, 5, 9, and 13 Gy). Live (7-AAD^–) cells in gated CD11c⁺ DC or CD4⁺CD25⁺ Tregs population from cultured spleen cells (2 × 10⁵) in each radiation group were analyzed by FACS using the Annexin V Apoptosis Detection Kit 1 (BD Biosciences). The percentage of radiation-related apoptosis was determined by subtraction of percentage of apoptotic cells after 48 h culture without irradiation.

Statistical analysis was performed on Prism version 5.0 software (Graphpad Software). Mean values were compared with two-tailed Student t test. Differences in the CD86- or CD80-positive rate in host DCs between each transplantation groups were analyzed with Mann–Whitney U test. Differences in the radiation-related apoptosis between DCs and Tregs were compared with Fisher exact test using Annexin V–positive or –negative cell number in each irradiation group. Survival data were plotted using Kaplan–Meier estimates, and the Mantel–Cox log-rank test was used to compare survival curves; p < 0.05 was considered significant.

In a murine MHC-haploidentical BMT model (BDF1 (H-2^b/d) → B6C3F1 (H-2^b/k)), recipient mice received TCD-BM cells (5 × 10⁶) and splenocytes (2 × 10⁷) from donor mice after TBI of 13 Gy or 5 Gy. Because TBI of 13 Gy was lethal to the recipient mice but TBI of 5 Gy was not, the two BMT models using TBI 13 Gy and 5 Gy correspond to MAST and RIST, respectively. Both groups achieved full donor chimerism within 14 d, in both the MLNs and spleen (Supplemental Fig. 1); however, the RIST recipients showed significantly fewer clinical GVHD signs (Fig. 1A) and longer survival (Fig. 1B) than the MAST recipients did. All RIST recipients survived for at least 3 mo, whereas MAST recipients died of GVHD by day 60 after transplantation. Because all recipients receiving 5 Gy TBI without transplantation had autologous hematopoietic recovery, this conditioning is categorized as nonmyeloablative conditioning, according to the definition by Champlin et al. (18). However, in the current study, we used RIST as a contrast of MAST.

Histopathologic examination of the MAST recipients revealed various pathologic changes compatible with acute GVHD in the skin, liver, and intestines. In particular, samples from the large intestine on day 14 demonstrated severe histologic damage with massive infiltration of donor CD4⁺/CD8⁺ T cells (Fig. 1C, 1D, left panels). In contrast, those from RIST recipients showed fewer pathologic changes and less infiltration of T cells (Fig. 1C, 1D, right panels). In the evaluation using the pathologic scoring system, the MAST recipients showed significantly higher GVHD scores in the small and large intestines than the RIST recipients did (Fig. 1E). Regarding liver GVHD, the MAST recipients tended to have higher GVHD scores than the RIST recipients did. In vivo imaging analysis clearly indicates the difference in the expansion of donor T cells in GVH organs between the MAST and RIST recipients (Fig. 1F). Ex vivo imaging revealed massive gut infiltration of donor splenocytes in the MAST recipients compared with that in the RIST recipients (Fig. 1G). Quantification of emitted photons over the abdominal region of interest (ROI) showed significantly more emitted photons over the abdominal ROI in the MAST recipients than in the RIST recipients (Fig. 1H). These results clearly showed that dose intensity of TBI had a strong influence on the GVHD severity and survival of the recipients in these BMT models.

To examine the difference in the activation and proliferation of donor T cells between MAST and RIST recipients, we analyzed the kinetics of donor T cells in the MLNs because the pathologic changes of GVHD were most marked in the intestine in our murine model. In both groups, donor CD4⁺ and CD8⁺ T cells increased in number after homing to MLNs, peaked on day 7, and thereafter decreased (Fig. 2A, 2B). There was no significant difference in the numbers of donor T cells retrieved from MLNs on day 1, corresponding to the number of cells homing to MLNs, between MAST and RIST (data not shown); however, donor T cells in MAST expanded more vigorously than in RIST. In fact, the number of donor T cells on day 4 was significantly higher in MAST than in RIST (5 times for CD4⁺ and 7 times for CD8⁺ T cells; Fig. 2A, 2B; Supplemental Fig. 2). We performed the FACS analysis using CFSE for investigating the proliferation status of donor T cells. Representative data are shown in Fig. 2C. The percentage of donor CD4⁺ or CD8⁺ T cells that had divided at least once by day 4 was significantly higher in the MAST than in the RIST (Fig. 2D). The mean numbers of divisions of CD4⁺ and CD8⁺ T cells in the period from days 1 to 4 were 8.5 and 8 in the MAST versus 7 and 5.5 in the RIST, respectively. These results show that a significantly greater percentage of CD4⁺ and CD8⁺ donor T cells in the MAST recipients entered the cell cycle and divided more vigorously than those in the RIST recipients.

We next performed qualitative analysis of proliferating donor T cells. There were no differences in the percentage and expression levels of apoptosis-related markers, including Annexin-V, Fas, or PD-1, on donor T cells that were recovered from the MLNs on days 4 and 7 (data not shown). Next, we investigated the activation status of donor T cells that were recovered from the MLNs on days 4 and 7. T cells mature dependently on cell division (19, 20). In fact, the MAST recipients showed significantly higher values than the RIST recipients did in terms of the percentages of CD44⁺CD62L⁻ (effector/memory phenotype; Fig. 2E), the numbers of IFN-γ–producing cells (Fig. 2F) on day 4, and the numbers of CXCR3-positive cells (Fig. 2G) on day 7. These findings indicate that donor T cells in the MAST were more activated and more differentiated than those in the RIST.

In our BMT model, we next investigated the kinetics of host immune cells in the MLNs. Anti–H-2Kk Ab was used to identify host-derived cells, and CD4⁺Foxp3⁺ T cells were defined as Tregs (Fig. 3A). As shown in Fig. 3B–D, regardless of the intensity of conditioning, host immune cells, including CD4⁺ and CD8⁺ T cells, as well as Tregs, increased in number after transplantation, peaked on day 4, thereafter decreased, and had disappeared by day 14. A significantly greater number of host immune cells survived in the RIST than in the MAST on day 1. A greater number of cells remained on day 1 in the order of CD4⁺ T cells > CD8⁺ T cells > Tregs in both the MAST and the RIST recipients. The proliferation rate in the period from days 1 to 4 was significantly higher in the MAST than in the RIST. The proliferation rate in the period from days 1 to 4 was higher in the order of Tregs > CD4⁺ T cells > CD8⁺ T cells in both the MAST and the RIST recipients. This transient proliferation of host immune cells was considered to occur by the mechanism of homeostatic proliferation based on lymphopenic status. However, when the proliferation of host Tregs was compared between allogeneic and syngeneic RIST settings, a significantly greater number of host Tregs were observed in allogeneic BMT than in syngeneic BMT (Fig. 3E), suggesting that this proliferation was partially related to alloresponse. The increased number of host Tcons (CD4⁺ or CD8⁺ T cells) in the RIST recipients may have exerted host-versus-graft reaction, thereby leading to mitigation of GVHD. However, this may be not the case as the percentage of host effector/memory phenotype (CD44⁺CD62L⁻) T cells was significantly lower in the RIST recipients than in the MAST recipients (Supplemental Fig. 3A, 3B).

We next investigated whether host Foxp3⁺CD4⁺ Tregs could suppress GVHD in in vivo experiments. First, host Tregs were additionally administered to the MAST recipients. Consistent with previous reports (3), freshly purified CD4⁺CD25⁺ Tregs showed only a minimal protective effect on GVHD in the MAST model (data not shown). Next, the MAST recipients additionally received preactivated host-type CD4⁺CD25⁺ Tregs (5 × 10⁵) that had been cultured for 3 d in the presence of IL-2 and immobilized anti-CD3/CD28 mAbs (21). Intriguingly, the addition of preactivated host Tregs significantly reduced clinical GVHD signs (Fig. 4A) and inhibited GVHD mortality (Fig. 4B) in the MAST recipients compared with that of untreated controls. Most recipients that received host Tregs eventually survived for at least 3 mo. We next investigated whether the aggravation of GVHD occurs in the RIST recipients by mitigating host Tregs through the administration of anti-CD25 mAb to the recipients before transplantation. In this experiment, the number of host Foxp3⁺CD4⁺ T cells that were recovered from MLNs on day 1 in mice receiving anti-CD25 mAb significantly decreased (to ∼50%) compared with that in control mice (Supplemental Fig. 4C). Consequently, anti-CD25 mAb-treated RIST recipients had a significantly lower median survival rate along with the exacerbation of GVHD compared with the controls (Fig. 4C, 4D). Some anti-CD25 mAb-treated mice succumbed to GVHD within 70 d after transplantation. These results indicate that host Tregs have the ability to suppress GVHD in vivo.

In vivo imaging analysis using luciferase showed that a greater number of splenocytes in the anti-CD25 mAb-treated mice accumulated in the MLNs and intestines than in the untreated mice (Fig. 4E). Quantification of emitted photons over the abdominal ROI showed significantly more emitted photons in the anti-CD25 mAb-treated mice than in the untreated mice (Fig. 4F).

We next analyzed the localization and proliferative status of host Tregs in the immunohistochemical analysis. As shown in Fig. 5A, host Foxp3⁺CD4⁺ Tregs (yellow-colored) were widely distributed in the MLNs and intestines of the RIST recipients. On the other hand, few host Foxp3⁺CD4⁺ Tregs were observed in the MAST recipients (<10% of the RIST recipients). We next performed Ki-67 staining to reconfirm immunohistochemically the proliferation of host Tregs. The immunohistochemical analysis revealed that ∼40% of host CD45.1⁺Foxp3⁺ Tregs simultaneously exhibited Ki-67-positive staining in lymphoid tissues of the colon and in MLNs of the RIST recipients on day 4 (Fig. 5B). On the other hand, <30% of host CD45.1⁺Foxp3⁺ Tregs exhibited Ki-67–positive staining in the MAST recipients. Furthermore, in vivo BrdU incorporation assay (22) revealed that a significantly greater percentage of host Foxp3⁺ Tregs incorporated BrdU in the RIST recipients than in the MAST recipients. In addition, the ratios of BrdU incorporation among host Tregs were superior to those of host conventional Foxp3⁻ CD4⁺ T cells (Tcons; Fig. 5C), suggesting that host Tregs are not only relatively resistant to TBI, but also have greater proliferative ability even after TBI.

We next investigated the functionality of host Tregs that transiently survive after transplantation in in vitro study. To this end, a suppressive effect of host Tregs against the alloresponse of donor T cells was checked using the MLR assays. In the experiments using purified host CD4⁺CD25⁺ Tregs from the RIST spleens on day 4, we found that host Tregs suppressed the donor T cell proliferation in a dose-dependent manner (Fig. 5D). The inhibition rate of thymidine incorporation of the recovered host Tregs was comparable to that of host Tregs isolated from untreated mice. In addition, we also observed that host Tregs isolated from the RIST spleens on day 4 strongly inhibited the proliferation of CFSE-labeled donor T cells after coculture with host DCs in the presence of LPS and TNF-α (Fig. 5E). These data show that host Tregs were fully functional after transplantation.

Next, to investigate whether the suppression of alloresponse of donor T cells by host Tregs was caused by cell-to-cell contact inhibition or by some soluble mediators, we performed the MLR assay using transwell. Under the use of transwell, Tregs significantly decreased the value of thymidine incorporation (by ∼24%), compared with the control (without Tregs; Fig. 5F). The addition of anti–TGF-β neutralizing Abs into the transwell plates tended to recover the value of thymidine incorporation (by 60%), but that effect was not observed in the experiment using anti–IL-10 Abs (Fig. 5G). On the basis of these data, we concluded that the suppression of alloresponse by host Tregs was caused mainly by cell-to-cell contact inhibition, but partially by soluble mediators, including TGF-β.

We finally addressed the mechanism by which host Tregs that transiently survived after transplantation contributed to the suppression of GVHD. Among several suppressive mechanisms of Tregs that have been reported, we have focused on the residual host DCs because host DCs are considered to be the most potent inducer of GVHD among APCs that present recipient Ags to donor alloreactive T cells during the priming of GVHD (23–25). In fact, immunohistologic analysis revealed that residual host Foxp3⁺ Tregs were widely distributed in MLNs of the RIST recipients, and that 10–20% of those cells were closely aggregating with CD11c⁺ conventional DCs (Fig. 6A, upper panel). Also in MLNs of the MAST, some host Tregs closely contacted with DCs (Fig. 6A, lower panel), despite fewer of those cells. These results suggest that host Tregs directly affect residual host DCs through cell-to-cell contacts.

We next performed in vitro study to determine whether host Tregs were able to suppress directly the activation of host DCs. To this end, host-type DCs that were freshly isolated from B6C3F1 spleens were cultured in the presence of LPS or TNF-α, with or without purified host-type Tregs. Consequently, the expression levels of CD86 and CD80 were significantly suppressed by coculture with host Tregs (Fig. 6B, 6C).

In addition, we analyzed how host Treg influences host DC expression of MHC I and II expressions. In the same experiments as in Fig. 6B and 6C, we analyzed the expression levels of H-2K^k (class I; Fig. 6D) and I-A^k (class II; Fig. 6E) on DCs. Consequently, host Tregs significantly reduced the expression of these MHC Ags. We next investigated using the MLR assay whether host Treg-mediated inhibition of GVH responses could be induced through the modulation of DC activation. Freshly isolated splenic CD11c⁺ DCs from B6C3F1 recipient mice were cultured in the presence of LPS with or without preactivated host Tregs. After 1-d culture, DCs were harvested and purified using magnetic beads. The isolated DCs from coculture with host Tregs were found to be approximately 20-fold less efficient than DCs conditioned with LPS alone in terms of induction of the proliferation of donor Tcons (Fig. 6F).

On the basis of these results, we hypothesized that the difference in the proliferation of donor T cells between the MAST and RIST recipients was brought about by the difference in the maturation status of host DCs. To address this issue, we investigated host DCs in the MLNs in the early transplant period. The FACS analysis, using cells isolated from the recipient MLNs on day 1, which corresponds to the timing of donor T cells’ homing into MLNs, revealed that the RIST recipients showed significantly reduced expression levels of CD86 and CD80 costimulatory molecules compared with the MAST recipients (Fig. 7A).

Regarding the number of host DCs in MLNs, the median DC number on day 1 in the RIST recipients was 9.5-fold greater than that in the MAST recipients (Fig. 7B). From days 1 to 4, the number of host DCs increased 9.1-fold in the MAST recipients and 6.1-fold in the RIST recipients. The median ratio of host DCs to host Tregs in MLNs on day 1 was higher in MAST (9.8:1) than in RIST (4.5:1) (Table I), suggesting that DCs were more resistant to TBI than Tregs. Because of the high ratio of host DCs to host Tregs in MLNs on day 1 in the MAST, host DCs in MAST should be less influenced by the suppressive effects of host Tregs than those in RIST. In fact, the percentage of CD86⁺ host DCs on day 1 tended to be higher in the MAST than in the RIST (a median of 90.7% in the MAST and 68.8% in the RIST; p = 0.067). The median percentages of CD86⁺ host DCs in the MAST and RIST significantly decreased on day 4 (56.9% for the MAST and 53.8% for the RIST; Fig. 7B, 7C), which may have been related to the decrease in the median ratio of DCs to Tregs from days 1 to 4 (9.8:1 to 4.4:1 for the MAST, and 4.5:1 to 2.0:1 for the RIST; Table I). On the other hand, the higher percentage of CD86⁺ DCs in the MAST may have been caused by intensified conditioning as the maturation of DCs is accelerated through the high production of inflammatory cytokines. To address this issue, we analyzed anti-CD25 mAb-treated RIST recipients (Fig. 4C, 4D), in which the extent of inflammation was similar to that in the control RIST recipients (Supplemental Fig. 4A, 4B), but they had significantly lower numbers of host Tregs (Supplemental Fig. 4C), and significantly higher numbers of host DCs (p = 0.016) than the control RIST did, which resulted in a median higher ratio (11.1:1) of DCs to host Tregs on day 1. As expected, in the anti-CD25 mAb-treated RIST recipients, the median percentage of CD86-positive host DCs in the MLNs on day 1 was as high as 98.2% (Fig. 7C, Table I), which was higher than the value of the MAST recipients. As shown in Fig. 7D, a significantly positive correlation between the ratio of host DCs to host Tregs and the percentage of CD86⁺ DCs was observed (r = 0.977, p = 0.0041). Similar to the data on CD86 Ag expression, the percentage of CD80⁺ host DCs on days 1 and 4 was significantly reduced in the RIST recipients compared with that in the MAST recipients, respectively, and that in the anti-CD25 mAb-treated RIST recipients significantly increased compared with that of the RIST recipients on day 1. (Fig. 7E, 7F). Similarly, a significantly positive correlation between the ratio of host DCs to host Tregs and the percentage of CD80⁺ DCs was observed (r = 0.942, p = 0.017; Fig. 7G). The percentage of DCs positive for CD86/CD80 and the ratio of the number of host DCs to host Tregs are summarized in Table I. The fact that the ratio of DCs to Tregs was higher in the MAST recipients than in the RIST recipients strongly suggests that DCs are more resistant to radiation compared with Tregs. To address this possibility, resistance of DCs and Tregs to irradiation was analyzed with an in vitro assay using after irradiation with titrated doses. Consequently, as shown in Fig. 7H, live (Annexin V⁻) Treg counts significantly decreased in a dose-dependent manner (p = 0.0083; in comparison of live cell numbers in 0- and 13-Gy irradiation groups). In contrast, live DC counts were not much influenced by radiation. In each irradiation dose, Tregs had a significantly higher percentage of Annexin V⁺ cells than DCs (Fig. 7I). Thus, we confirmed that DCs are more resistant to irradiation than Tregs.

These results indicate that an increase in the ratio of the number of DCs to host Tregs results in an increase in the percentage of mature DCs. Coupled with the in vitro data, we concluded that host Tregs restrained the maturation of host DCs in a dose-dependent manner in the peritransplantation period.

As experimental murine BMT models, fully MHC-incompatible, minor-incompatible, or parent-to-F1 BMT models have been used. On the other hand, an HLA-haploidentical donor used in clinical BMT settings is haploidentical in the GVH and HVG directions; however, this kind of animal model has never existed. In this point, our haploidentical BMT model is the first one, to our knowledge, that faithfully reflects clinical HLA-haploidentical BMT. The murine bidirectionally haploidentical BMT model that we developed is characterized by mixed chimerism lasting for 2 wk after transplantation, especially in the RIST model, during which host type-immune cells transiently expand, exert some functions, and affect the magnitude of GVH reaction. In particular, host Tregs act to suppress GVH reaction through the interaction with DCs. However, our findings observed in the current study are a ubiquitous phenomenon that is observed in other types of allogeneic BMT settings to a greater or lesser degree. Furthermore, although we focused in the current study on the kinetics of donor and recipient immune cells in MLNs, our findings are not unique to MLNs but also common to secondary lymphoid organs, because similar findings were also observed in spleen.

Host DCs are the most potent inducers of GVHD among APCs that present recipient Ags to donor conventional T cells in secondary lymphoid organs during the priming of GVHD (23–25). In particular, the extent of maturation of host DCs at the time when donor conventional T cells interact with DCs is important in determining the magnitude of subsequent GVH reaction. Namely, mature DCs activate T cells, but immature DCs render T cells anergic. In the current study, we showed that host Tregs restrained the maturation of DCs in a dose-dependent manner, and consequently reduced the magnitude of GVH reaction. Thus, host Tregs naturally act in secondary lymphoid organs in the peritransplantation period and deeply contribute to the suppression of GVHD through regulating the maturation of DCs, although less attention has been focused on host Tregs.

To investigate the role of host Tregs in GVHD, we established a murine MHC-haploidentical BMT model, BDF1 (H-2^b/d) → B6C3F1 (H-2^b/k). Regarding donor conventional T cells, there was no significant difference in the T cell numbers in the MLNs on day 1, corresponding to the number of cells homing to MLNs, between MAST and RIST. After homing to MLNs, donor T cells increased in number and peaked on day 7 regardless of the intensity of conditioning, but donor T cells in MAST expanded more vigorously than those in RIST did. In fact, the number of donor T cells on day 4 was significantly higher in MAST than in RIST (5-fold for CD4⁺ and 7-fold times for CD8⁺ T cells; Fig. 2A, 2B). In general, T cells mature dependently on cell division (19, 20). In fact, the MAST recipients showed significantly higher values than the RIST recipients did in terms of the percentage of CD44⁺CD62L⁻ (effector/memory phenotype; Fig. 2E), the number of IFN-γ–producing cells (Fig. 2F) on day 4, and the number of CXCR3-positive cells (Fig. 2G) on day 7. Another important point is that the peak of the donor T cell number in MLNs occurs as early as day 7, suggesting the importance of early regulation of GVH reaction. On the other hand, amplified host Tregs peaked on day 4 (Fig. 3D), which is considered the appropriate timing for suppressing GVH reaction.

As one of the therapeutic methods for controlling GVHD, donor Foxp3⁺CD4⁺ Tregs have been studied extensively and have begun to be applied clinically to HLA-haploidentical SCT (9, 10). Tregs have been reported to function by several mechanisms, including a cell-cell contact-dependent (26) or cytokine-dependent manner (27, 28). However, increasing attention has recently been focused on the interaction of Tregs with DCs. In fact, Tregs have been recently reported to function through the induction of host DC death and the downregulation of surface proteins (CD86 and CD54) required for donor T cell activation (29).

Regarding host-derived Tregs, several groups, including ours, using different experimental BMT models, showed that host Tregs suppressed GVHD clinically (2, 11); however, the in-depth mechanisms have remained to be determined. In the current study, we showed, in the MLR assay using transwell, that the suppression of alloresponse by host Tregs was caused mainly by cell-to-cell contact inhibition, but partially by soluble mediators, including TGF-β. In fact, we showed that, in keeping with results observed in studies (29–31) of donor Tregs, host Tregs closely contacted with DCs in vivo (Fig. 6A), reduced the expression of MHC class I and class II on DCs (Fig. 6D, 6E), inhibited the maturation of DCs (decreased the expression of CD86 and CD80 Ags; Fig. 6B, 6C), and suppressed allogeneic responses in MLR upon coculture with DCs treated with LPS or TNF-α (Fig. 5E). Furthermore, we showed that host Tregs could proliferate in in vivo study, including immunohistochemistry using Ki-67 staining and BrdU incorporation experiment (Fig. 5B, 5C). These findings indicate that host Tregs were fully functional despite surviving transiently, and that they seemed to act through mechanisms similar to those reported for Tregs.

The ratio of host DCs to host Tregs in MLNs on day 1 was higher in MAST (9.8:1) than in RIST (4.5:1), indicating that DCs were more resistant to TBI than Tregs were. In fact, we could confirm that DCs are more resistant to irradiation than Tregs are in in vitro study (Fig. 7H, 7I). This higher ratio of DCs to Tregs in MAST indicates that host DCs in MAST are less influenced by the suppressive activity of host Tregs than those in RIST, resulting in a higher percentage of DCs with a mature phenotype. The higher percentage of mature DCs in the MAST recipients may have been caused by LPS or TNF-α produced by higher-dose TBI, because these inflammatory cytokines are known to be powerful activators of DCs. However, in anti-CD25 mAb-treated RIST, in which the extent of inflammation was similar to that in the control RIST recipients (Supplemental Fig. 4A, 4B), the number of host Tregs on day 1 significantly decreased, and the number of host DCs on day 1 significantly increased compared with that of the control RIST (Supplemental Fig. 4C), resulting in a median higher ratio (11.1:1) of DCs to host Tregs. In fact, in anti-CD25 mAb-treated RIST, the median percentage of CD86⁺ host DCs in the MLNs on day 1 was as high as 98.2%, which was rather higher than the value of the MAST recipients. These results indicate that the extent of maturation of host DCs is influenced by the ratio of host DCs to host Tregs. Furthermore, as data that consolidated our hypothesis, a significantly positive correlation between the ratio of DCs to host Tregs and the extent of maturation of DCs was observed (Fig. 7D, 7G). Coupled with our in vitro and in vivo studies showing that host Tregs suppress the maturation of DCs, we concluded that host Tregs restrained the maturation of DCs in a cell–dose-dependent manner.

In addition, the fact that the number of host DCs in MLNs on day 1 in the anti-CD25 mAb-treated RIST was significantly (1.4-fold) greater than in the control RIST suggests that host DCs had decreased in number by interaction with host Tregs by day 1, as DCs were reported to be eliminated through Treg-mediated killing via FasL or perforin (29, 32). During days 1–4 after transplantation, the expansion rate of host Tregs was greater than that of host DCs regardless of the intensity of conditioning, leading to a decrease in the ratio of DCs to Tregs in the RIST (4.5:1 to 2.0:1) and the MAST (9.8:1 to 4.4:1), and probably leading to a decrease in the percentage of mature DCs.

Different from donor Tregs, host Tregs are resident in, or have already migrated from damaged end organs to, secondary lymph nodes (LNs) by the time when donor Tregs home there. Host Tregs interact with host DCs there (Fig. 6A) and can regulate DCs quantitatively and qualitatively, before donor conventional T cells migrate there and start to interact with DCs. Thus, host Tregs are considered as potent negative regulators of DCs in the peritransplantation period. Furthermore, host Tregs were reported to be already activated in LNs (33–35). In contrast, donor Tregs need to home to secondary LNs (36–39) and to be activated to exert suppressive activity (6, 29). In allogeneic BMT settings, donor Tregs and donor Tcons may compete in the interaction with DCs in a sense; therefore, in a clinical protocol for the suppression of GVHD, twice as many donor Tregs as donor conventional T cells were infused (10). The acquisition of this large number of donor Tregs is laborious, time-consuming, and expensive. In contrast, host Tregs always act as natural immunologic negative regulators in LNs, can occupy DC sites by the time when donor T cells migrate to secondary LNs, and thus the ratio of DCs to host Tregs may be rather important. As such, host Tregs may function in small numbers, and their preservation may be possible by contriving conditioning treatments.

Interestingly, regardless of the intensity of conditioning, host-derived hematopoietic cells, including conventional T cells, Tregs, and DCs, increased in number after transplantation, peaked on day 4, and subsequently decreased. Although host immune cells peaked earlier than donor T cells, this transient cell expansion is considered mainly to have been due to homeostatic proliferation (40, 41); however, because the magnitude of the transient increase in the number of host Tregs in RIST on day 4 significantly diminished (to ∼30%) in a syngeneic transplant setting (Fig. 3E), this expansion was related partially to allogeneic response. Whether increased host Tcons (CD4⁺ or CD8⁺ T cells) exerted host-versus-graft reaction remains to be determined because it could not be analyzed; however, this may be not the case as the percentage of host effector/memory phenotype (CD44⁺CD62L⁻) T cells was significantly lower in the RIST recipients than in the MAST recipients (Supplemental Fig. 3A, 3B).

Considering the importance of host Tregs for regulation of the GVH reaction, it may be better to select host Treg-friendly conditioning treatments. In the current study, the use of an RIC regimen such as low-dose TBI must have enabled host Tregs to survive longer and enhanced their function. As another therapeutic agent favorable to Tregs, anti–T lymphocyte globulin, which is often used in clinical RIST settings, has been reported to be useful for causing Tregs preferentially to expand or remain (42, 43). Blockade of IL-6 was reported to augment Tregs (44). In contrast, cyclophosphamide (45), fludarabine (46), and mogamulizumab (humanized anti-CCR4 mAb) (47) have been reported to deplete Tregs selectively.

Inflammation, including the production of chemokines and cytokines, in GVH target organs is another important factor in the development of GVHD (48, 49). In the current study comparing MAST and RIST, MAST should have produced stronger inflammation in GVH target organs than RIST, which may have enhanced clinical GVHD in MAST. Despite this distortion, the amplification and differentiation of donor conventional T cells in secondary lymphoid organs, which are regulated by host Tregs through DCs, are invariably important for the development of GVHD. This is rationalized by the results of RIST that was treated with anti-CD25 mAb before transplantation.

In conclusion, we showed that host Tregs restrained the maturation of, and probably decreased the cell numbers of, host DCs in the peritransplantation period, by the time when donor conventional T cells encounter DCs in secondary lymphoid organs, and that diminished maturation of DCs caused by host Tregs resulted in a reduction of the magnitude of the subsequent GVH reaction. Thus, host Tregs are potent negative regulators of DCs.

We thank Dr. Kouji Matsushima and Dr. Satoshi Ueha for technical advice on immunohistochemistry, Takashi Daimon for statistical advice, and Yoko Yamamoto and Hiromi Takeda for technical assistance.

The authors have no financial conflicts of interest.