Donor Requirements for Regulatory T Cell Suppression of Murine Graft-versus-Host Disease

Allogeneic hematopoietic cell transplantation (HCT) is a curative treatment for patients with hematological malignancies and many congenital and genetic disorders. One of the major complications of HCT is graft-versus-host disease (GVHD), a potentially lethal immune reaction caused by donor cells recognizing and destroying host tissues (1). Several studies have demonstrated that CD4⁺CD25⁺Foxp3⁺ regulatory T cells (Treg) control CD4⁺ and CD8⁺ conventional T cell (Tcon) proliferation, limiting GVHD lethality yet retaining antiviral and graft-versus-tumor activity, thus promoting animal survival (2–6). Recently, these promising results have been translated into the clinic confirming that Treg-based cellular therapy is a powerful approach for GVHD prevention (7–10).

Despite the promising results, there are several factors that are limiting the broader application of this treatment (11, 12). Treg from different sources such as the donor, recipient, or third party have been tested in preclinical and clinical transplantation studies, but no comparison between these different donor sources has been systematically reported, and therefore it is unclear which donor source has a greater impact on Tcon proliferation and prevention of GVHD. Several studies demonstrated that Treg exert their suppressive function through different mechanisms that can be contact or cytokine mediated (13). In these animal models it has been demonstrated that Treg undergo expansion and control Tcon proliferation. Treg have been shown to directly home in primary lymphoid tissues after their adoptive transfer where they prevent Tcon proliferation and further homing in GVHD target tissues. Moreover, timing of Treg adoptive transfer is extremely important, as they require to be injected prior to Tcon for conferring the best GVHD protection (14–16).

It is yet unclear whether MHC disparities between Treg and Tcon impact Treg function. Furthermore, several groups are investigating the clinical utility of ex vivo–expanded Treg to increase their number because Treg are a rare cell population and others are improving culturing strategies to enhance Treg function (6, 17–20). Third-party Treg are particularly suitable for such studies, as they can be prepared in advance and then banked for further use. Accordingly, their application may be particularly relevant in cases where the donor is not immediately available such as transplantation from an unrelated or umbilical cord blood donor. In this study we investigated the impact of MHC disparities between Treg and Tcon on alloreactive T cell proliferation and GVHD prevention, aiming to establish the role of MHC-disparate Treg sources, which is of central importance for their clinical application. Using a model of in vivo Treg depletion, we further studied the timing of Treg function in vivo after adoptive transfer. We found that selective in vivo depletion of injected Treg at different time points has a different impact on GVHD onset and lethality, and therefore our study introduces relevant insights to the complex mechanisms through which Treg protect from GVHD.

Experiments used gender-matched mice between 7 and 12 wk old. FVB/N (H-2^q), BALB/c (H-2^d, CD45.2), and C57BL/6 (H-2^b, CD45.2) mice were purchased from The Jackson Laboratory (Sacramento, CA). Luciferase-expressing (luc⁺) C57BL/6 (CD45.1, Thy1.1) and BALB/c (CD45.1, Thy1.1) mice were generated after backcrossing for >10 generations to the C57BL/6 or BALB/c background with the luc⁺ transgenic FVB/N L2G85 mice that were described previously (21). C57BL/6 albino Foxp3 mutant mice expressing diphtheria toxin (DT) receptor, GFP, and luc (Foxp3^DTR/GFP/luc) were a gift from Dr. Günter J. Hämmerling (Division of Molecular Immunology, Deutsches Krebsforschungszentrum, Heidelberg, Germany) and Dr. Andreas Beilhack (Department of Medicine II, Center for Experimental Molecular Medicine, Würzburg, Germany) and were bred in our animal facility. Animal protocols were approved by the Institutional Animal Care and Use Committee of Stanford University.

CD4/CD8 Tcon were prepared from splenocytes and peripheral lymph node cells and enriched with anti-CD4 and anti-CD8 MACS (Miltenyi Biotec, Auburn, CA). Purity was assessed by FACS using an LSR II (BD Biosciences, San Jose, CA). T cell–depleted bone marrow (TCD BM) was prepared by flushing bones and depleting T cells with anti-CD4 and anti-CD8 MACS beads. Treg were prepared from pooled spleens and lymph nodes, by staining for CD25-allophycocyanin and CD4, enriching with anti-allophycocyanin MACS beads, and sorting for CD4⁺CD25^hi cells or for CD4⁺CD25⁺GFP⁺ cells from C57BL/6 albino Foxp3^DTR/GFP/luc mice in the experiments with in vivo Treg depletion on a FACSAria III (BD Biosciences). Approximately 200,000 purified (>98% Foxp3⁺) Treg were routinely obtained per wild-type mouse, and 400,000 purified Treg were obtained per C57BL/6 albino Foxp3^DTR/GFP/luc mouse.

The following anti-mouse Abs were purchased from eBioscience (San Diego, CA) or BioLegend (San Diego, CA): PerCP/Cy5.5 anti-CD4 (GK1.5), allophycocyanin/Cy7 anti-CD8a (53-6.7), Pacific Blue anti-CD44 (IM7), FITC anti-CD62L (MEL-14), allophycocyanin anti-CD25 (3C7), Pacific Blue anti-Foxp3 (MF-14), FITC anti–H-2K^q (KH114), PE anti–H-2D^d (34-2-12), FITC anti–H-2K^b (AF6-88.5), PE/Cy7 anti-CD45.1 (A20), PE anti-CD45.2 (104), allophycocyanin anti-CD90.1 (OX-7), allophycocyanin anti–IL-10 (JES5-16E3), PE anti–IFN-γ (XMG1.2), Alexa Fluor 700 anti-IA/IE (M5/114.15.2), and PE anti-CTLA4 (UC10-4B9). Isotype controls were purchased from the respective vendors. Foxp3 staining was performed with an anti-mouse/rat Foxp3 staining set (eBioscience). Fixable viability dye eFluor 506 (eBioscence) was used for identification and removal of dead cells. Analysis was performed on an LSR II (BD Biosciences). Data were analyzed with FlowJo 10.0.7 (Tree Star, Ashland, OR).

Tcon (1–2 × 10⁵) with or without Treg (for Treg/Tcon ratio see text) were added per well to a 96-well U-bottom plate containing complete RPMI 1640 and 1 × 10⁶ BALB/c splenocytes, which were previously irradiated with 30 Gy. Tcon and Treg were isolated from different mouse strains as described in the text. Total volume per well was 300 μl, and cells were cultured at 37°C and 5% CO₂. For quantifying T cell proliferation, after 4 d of culture, cells were pulsed with 1 mCi/well [³H]thymidine (GE Healthcare, Piscataway, NJ) for 16 h. Cells were harvested onto filter membranes using a Wallac harvester (PerkinElmer, Shelton, CT), and the amount of incorporated [³H]thymidine was measured with a Wallac Betaplate counter (PerkinElmer).

For analyzing surface marker expression and for in vitro cytokine analysis, 5–10 × 10⁴ Treg were added per well of a 96-well U-bottom plate containing complete RPMI 1640 and were stimulated with either IL-2 (Chiron, Emeryville, CA) alone (5000 IU/ml) or IL-2 plus anti-CD3/CD28 beads (Dynabeads mouse T-activator CD3/CD28, Life-Technologies, Logan, UT; Treg/bead ratio of 1:1) or were coincubated with Tcon and irradiated splenocytes as described above. After 3–4 d of culture, supernatants were collected for cytokine analysis and cells were harvested, stained, and FACS analyzed as described above.

For detecting T cell proliferation through in vivo bioluminescence imaging (BLI), BALB/c recipient mice were irradiated at day −2 with total body irradiation (TBI) 2 doses of 4 Gy, 4 h apart with 200-Kv x-ray source. Then, 5 × 10⁵ host-type (BALB/c, H-2^d), donor-type (C57BL/6, H-2^b), or third-party (FVB/N, H-2^q) Treg were administered i.v. at day −2, and 5 × 10⁶ TCD BM cells from CD45.2⁺ C57BL/6 mice and 1 × 10⁶ Tcon from luc⁺ CD45.1⁺ C57BL/6 mice were injected i.v. at day 0. Transplanted animals were housed in autoclaved cages with antibiotic water or antibiotic food (sulfamethoxazole/trimethropim; Schein Pharmaceutical, Corona, CA). In vivo BLI was performed as described (22) with an IVIS 29 charge-coupled device imaging system (Xenogen, Alameda, CA). Images were analyzed with Living Image software 3.4.1 (Xenogen).

For experiments with Treg, in vivo depletion transplantation was performed as described above. Treg from C57BL/6 albino Foxp3^DTR/GFP/luc mice were administered i.v. at day −2. DT was i.p. injected at the dose of 50 mg/kg/mouse daily for 2 consecutive days with the timing described in the text.

For Treg reisolation after transplant, BALB/c recipient mice were transplanted as described above. Mice were euthanized at days 6 and 12 after transplantation, and peripheral blood, spleen, lymph nodes, and liver were harvested. Sera were collected from peripheral blood for cytokine analysis. Spleen and lymph nodes were disassociated for obtaining homogeneous cell suspensions. Liver was mashed and then was passed through a Percoll gradient (GE Healthcare). The lymphocyte layer was collected. All cell suspensions were filtered, stained, and FACS analyzed as described in the text.

For survival experiments, BALB/c recipient mice were transplanted as described above. Mice were weighed weekly and GVHD score was calculated (23).

MHC minor mismatched model of transplantation was performed as described above with the only difference that BALB/b (H-2^b) mice were used as recipients and 2.5 × 10⁷ Tcon were injected for GVHD induction.

For intracellular cytokine staining, cells were stimulated with 20 ng/ml PMA (Sigma-Aldrich, St. Louis, MO) and 1 μg/ml ionomycin (Sigma-Aldrich) for 6 h at 37°C and 5% CO₂ in RPMI 1640 supplemented with 10% FCS, 2 mM l-glutamine (Mediatech, Manassas, VA), 100 U/ml penicillin (Thermo Fisher Scientific, Asheville, NC), and 100 μg/ml streptomycin (Thermo Fisher Scientific). Monensin (BD Biosciences) was used to block cellular protein transport. Cells were fixed and permeabilized (eBioscience, San Diego, CA) prior to staining of intracellular and intranuclear Ags. For quantitative measurement, supernatants were collected from Treg in vitro cultures and sera were obtained from peripheral blood of transplanted mice at days 6 and 12 after transplantation. All samples were analyzed for cytokine concentration through multiplex assay (Luminex, Life Technologies, Logan, UT)

Spleens and livers from transplanted animals were harvested and fixed with 4% paraformaldehyde in PBS. Fixed organs were included in frozen blocks and cryosections were obtained. Slides were stained with PE anti-CD4 Ab. Data were analyzed though Leica SP2 confocal microscopy (Leica Microsystems, Mannheim, Germany)

Differences in animal survival (Kaplan–Meier survival curves) were analyzed with the log-rank test. Weight variation and GVHD score were analyzed with a two-way ANOVA test. All other comparisons were performed with the two-tailed Student t test, and p < 0.05 was considered statistically significant.

To test the impact of MHC disparities on Treg suppressive function, we evaluated the ability of highly purified Treg derived from donors with different MHC backgrounds to suppress proliferation of C57BL/6 (H-2^b) Tcon following exposure to irradiated stimulator splenocytes in the mixed leukocyte reaction in vitro. Both donor-derived C57BL/6 (H-2^b) or third-party FVB/N (H-2^q) Treg suppressed Tcon proliferation induced by exposure to irradiated BALB/c (H-2^d) splenocytes at different Treg/Tcon ratios (Fig. 1A). Similar results were obtained by incubating different Tcon (BALB/c, H-2^d or FVB/N, H-2^q, third party) with irradiated splenocytes from C57BL/6 (H-2^b) mice and Treg from BALB/c (H-2^d) mice. BALB/c Treg suppressed proliferation of both BALB/c and FVB/N Tcon equally (both p = 0.04; Fig. 1B), demonstrating that MHC disparities do not interfere with Treg suppressive function in vitro.

LAG3 and CTLA4 are two key molecules that are upregulated on Treg upon activation (13, 24–26). Moreover, activated Treg release IL-10 and can reduce Tcon IFN-γ production (5, 27, 28). We investigated the impact of MHC disparities on the activation of Treg in vitro. Treg were purified from either BALB/c (H-2^d, matched with stimulators), C57BL/6 (H-2^b, matched with responders), or FVB/N (H-2^q, third party) and incubated with IL-2 and anti-CD3/CD28 beads or were added to a mixed leukocyte reaction assay where C57BL/6 (H-2^b) Tcon were the responders and BALB/c (H-2^d) irradiated splenocytes were the stimulators for 3–4 d. In all of the conditions Treg maintained expression of CD25 and Foxp3 (Supplemental Fig. 1A, 1B). Third-party Treg had the same activation molecule expression patterns as did MHC-matched Treg. LAG3 and CTLA4 surface expression was enhanced after stimulation with IL-2 and anti-CD3/CD28 beads or after 3–4 d of culture with Tcon and irradiated splenocytes (Supplemental Fig. 1C, 1D). Furthermore, third-party and MHC-matched Treg produced the same levels of IL-10 and were similarly able to reduce Tcon IFN-γ production (Supplemental Fig. 1E, 1F), and thus in vitro Treg activation was not impaired by MHC disparities.

We translated these results to in vivo studies in animal models. In these studies TCD BM from C57BL/6 (H-2^b) mice was injected into lethally irradiated (TBI 8 Gy) BALB/c (H-2^d) recipient mice to establish hematopoiesis. GVHD was induced by injecting luc⁺ donor-derived Tcon (1 × 10⁶/mouse). Using this model, freshly isolated CD4⁺CD25⁺Foxp3⁺ Treg derived from BALB/c (H-2^d, host type), C57BL/6 (H-2^b, donor type), or FVB/N (H-2^q, third party) mice at the Treg/Tcon ratio of 1:2 were injected i.v. 2 d before Tcon injection and immediately after irradiation (Fig. 2A). Tcon proliferation was assessed with BLI. In vivo Tcon proliferation was similar in mice that received Tcon alone and in mice that received Tcon plus host type Treg (p > 0.05, Fig. 2B–D) at all time points analyzed, demonstrating that host Treg lack in vivo suppressive activity. On the contrary, Tcon proliferation was markedly reduced in mice that received either donor or third-party Treg 7 d after transplantation in comparison with mice that received Tcon alone (p = 0.0002 and p = 0.0001, respectively, Fig. 2B). After 15 or 29 d, Tcon proliferation was reduced in the mice that received donor Treg, but not in animals that received third-party Treg (at day 15, p = 0.01 and p = NS and at day 29, p = 0.04 and p = ns, Fig. 2C, 2D), suggesting that third-party Treg can control in vivo Tcon proliferation in the early phases after transplantation but lose their activity over time in comparison with donor Treg. BLI of representative animals is shown in Fig. 2E.

We previously demonstrated that during GVHD onset Treg home to secondary lymphoid tissues after injection where they are primed and induce suppression of Tcon proliferation (15, 29). Survival of Treg during this phase is essential for function. Using congenic markers that allow for discrimination of the infused donor Treg and the same transplantation model described above, we compared the in vivo survival of third-party and donor Treg by isolating spleen, lymph nodes, liver, and peripheral blood of transplanted mice at different time points. Tissues from irradiated (TBI 8 Gy) BALB/c (H-2^d) CD45.2 Thy1.2 mice injected with either donor C57BL/6 (H-2^b) CD45.1, third-party FVB/N (H-2^q) CD45.1 Treg, or host BALB/c (H-2^d) Thy1.1 together with C57BL/6 CD45.2 Tcon and TCD BM were harvested at days 6 and 12 after transplantation and analyzed to calculate the number of live injected Treg. Interestingly, in the mice that received third-party Treg, we found less viable injected Treg at day 6 and extremely few cells at day 12 in all the examined organs in comparison with the mice that received donor Treg (at day 6, lymph nodes p = 0.03, spleen p = 0.03, peripheral blood p = 0.03, liver p = NS; at day 12, lymph nodes p = 0.04, spleen p = 0.01, peripheral blood p = 0.02, liver p = 0.04) or host Treg (host-versus-donor Treg in all examined organs p = NS) where injected Treg were readily detected (Fig. 3). Injected donor Treg were found in secondary lymphoid organs such as spleen and lymph nodes and also in the liver, a GVHD target organ, in both the selected time points, whereas only an extremely limited number of third-party Treg survived and could be detected, suggesting that these cells were rejected in vivo.

CD62L is a key molecule that enables T cell homing into lymph nodes via endothelial venules. TCR engagement induces T cells to lose CD62L expression, and thus effector memory CD62L⁻CD44⁺ activated T cells leave lymphoid tissue and migrate to the targets to drive immune responses (30). After priming in lymphoid tissues such as spleen and lymph nodes, Tcon home to target organs such as liver, skin, and gastrointestinal tract and induce GVHD in transplantation animal models. High numbers of CD44⁺ T cells have been found and proliferate in GVHD affected organs (31). Adoptive transfer of CD62L⁺CD4⁺CD25⁺Foxp3⁺ Treg has been shown to limit Tcon priming in lymphoid tissues, consequently controlling their proliferation (16). Using the GVHD model described before, we detected the presence of CD4⁺CD62L⁺ and CD4⁺CD62L⁻CD44⁺ T cells in the lymph nodes and in the liver of BALB/c recipient mice that received either C57BL/6 donor Tcon alone or Tcon and host-type BALB/c Treg or Tcon and C57BL/6 donor Treg or Tcon and FVB/N third-party Treg at day 6 after transplantation. Fewer effector memory CD4⁺CD62L⁻CD44⁺ activated T cells infiltrated the liver in mice that received donor or third-party Treg in comparison with mice that did not receive Treg (p = 0.04 and p = 0.03, respectively), whereas host type Treg did not reduce CD4⁺CD62L⁻CD44⁺ T cell liver infiltration (p > 0.05, Fig. 4A). At the same time, the CD4⁺CD62L⁺CD44⁻/CD4⁺ T cell ratio was higher in the lymph nodes of mice that received donor and third-party Treg (both p = 0.02), but not in the mice that received host-type Treg (p > 0.05, Fig. 4B). These results demonstrate that in the presence of both donor-type and third-party Treg, a higher proportion of Tcon maintain CD62L and did not express CD44 in secondary lymphoid organs; therefore, donor-type and third-party Treg reduce Tcon activation in lymph nodes, thus limiting their proliferation and infiltration into peripheral tissues. We also analyzed LAG3 expression in Treg isolated from the spleen and lymph nodes at day 6 after transplantation and found that although LAG3 was minimally expressed in residual host-type Treg, expression was enhanced in donor and third-party Treg (in lymph nodes, donor versus host type p = 0.006, donor versus third party p > 0.05, third party versus host type p = 0.02; in spleen, donor versus host type p = 0.0002, donor versus third party p > 0.05, third party versus host type p = 0.04; Fig. 4C), demonstrating that both donor and third-party Treg exhibit a suppressive phenotype after their adoptive transfer. Furthermore, because Treg treatment has been previously demonstrated to reduce IFN-γ production by Tcon and increase IL-10 production in GVHD affected mice, we collected sera of transplanted mice where we found that injection of donor Treg induces lower IFN-γ (p = 0.006) and higher IL-10 (p = 0.05) production in comparison with mice that were treated with third-party Treg (Fig. 4D). Little or no difference was found in IL-4 and IL-5 production (data not shown). Therefore, donor Treg more effectively induce a suppressive cytokine profile.

We tested the ability of Treg from different sources to protect mice from GVHD using the above-described in vivo model. We evaluated GVHD onset and mouse survival following the adoptive transfer of freshly isolated CD4⁺CD25⁺Foxp3⁺ Treg derived from BALB/c (H-2^d, host type), C57BL/6 (H-2^b, donor type), or FVB (H-2^q, third party) at the Treg/Tcon ratio of 1:2 (Fig. 5A). Donor Treg exerted the strongest dose-dependent GVHD protection (p = 0.028), whereas host Treg did not improve mouse survival (p = 0.58). Third-party Treg improved mouse survival (p = 0.028), but these animals had worse GVHD score profiles (p < 0.001) and did not recover their weight as well as did mice treated with donor Treg (p < 0.001, Fig. 5). To exclude a possible strain-related effect, we also transplanted BALB/c recipient with FVB/N-derived TCD BM and Tcon and adoptively transferred either BALB/c (host type), FVB/N (donor type), or C57BL/6 (third party) Treg. Even in this case, donor and third-party Treg, but not host Treg improved mouse survival (data not shown). These data confirm that third-party Treg can be a useful tool for GVHD prevention.

Because the major mismatch model produces the most aggressive GVHD, we compared donor and third-party Treg for GVHD protection in a minor mismatch mouse model of transplantation that more closely resembles clinical conditions in HLA-matched human transplantation. In this model, TCD BM from C57BL/6 (H-2^b) mice was injected into lethally irradiated (TBI 8 Gy) BALB/b (H-2^b, minor mismatched) recipient mice. Because it was previously demonstrated that adoptive transfer of higher numbers of Tcon is needed to induce GVHD in minor mismatched compared with major mismatched mouse models (32), we injected 2.5 × 10⁷ donor-derived Tcon. To prevent GVHD onset and lethality, we adoptively transferred 5 × 10⁵ freshly isolated CD4⁺CD25⁺Foxp3⁺ Treg derived from C57BL/6 (H-2^b, donor type) or FVB/N (H-2^q, third party) mice (Fig. 6A). In these studies, third-party and donor Treg equally prolonged survival (p = 0.02 and p = 0.04, respectively, versus Tcon alone group; p > 0.05 when the donor Treg group was compared with third-party Treg group) of these mice (Fig. 6B). No differences could be detected in weight recovery (Fig. 6C) and GVHD score (Fig. 6D), demonstrating that both donor and third-party Treg are able to protect mice from GVHD equally in this MHC minor mismatched model.

Whereas Treg dynamics in vivo after adoptive transfer have been widely studied, less is known regarding the timing of their in vivo function. We used a mouse model of in vivo Treg depletion to explore the timeframe during which injected Treg control Tcon proliferation and homing in GVHD target tissues after transplantation. We injected TCD BM from C57BL/6 mice into lethally irradiated (TBI 8 Gy) BALB/c (H-2^d) recipient mice and GVHD was induced by injecting donor derived Tcon (1 × 10⁶/mouse). Freshly isolated CD4⁺CD25⁺Foxp3⁺GFP⁺ Treg derived from C57BL/6 albino Foxp3^DTR/GFP/luc mice at the Treg/Tcon ratio of 1:2 were injected i.v. 2 d before Tcon injection. In some animals we depleted the injected Treg pool through i.p. administration of DT for 2 consecutive days at two different time points: immediately after their transfer (day −2 and day −1) or at the day of Tcon injection (day 0 and day +1, Fig. 7A). GFP⁺ injected Treg were detectable only in the spleen and peripheral blood of the animals that did not receive DT treatment, proving the efficacy of Treg depletion with DT in vivo (Fig. 7B). Utilizing BLI we found high intensity signal from luc⁺ Treg only in mice that did not receive DT treatment, further confirming that DT was effective in depleting injected Treg (Fig. 7C). Mice that received DT treatment at day 0 had reduced infiltration of donor-derived Tcon in the liver in comparison with mice that received DT treatment at day −2 and to mice that did not receive Treg adoptive transfer and similarly to mice that received Treg but no DT treatment (Supplemental Fig. 2). These results demonstrate that Treg exert their control on Tcon entry in the liver in the first 2 d after their injection.

To further understand the impact of Treg killing on their in vivo function and to explore the mechanism through which third-party Treg are able to prevent GVHD even when rapidly rejected in vivo, we used the mouse model of Treg in vivo depletion as described above and followed animals for survival, weight, and GVHD score. Mice that received Tcon and Treg adoptive transfer and DT treatment at day 0 had improved survival in comparison with mice that received Tcon and Treg adoptive transfer and DT treatment at day −2 and to mice that received Tcon alone; survival of these animals was comparable to animals that did not receive DT treatment (Fig. 7D). Furthermore, DT treatment at day 0 only mildly impacted mouse weight variation and GVHD score after transplantation (Fig. 7E, 7F). We also treated mice that received Tcon adoptive transfer (1 × 10⁶/mouse) without Treg with DT and observed similar survival, indicating that DT toxicity was limited in vivo and had no impact on GVHD (Supplemental Fig. 3). These data demonstrate that Treg exert their in vivo function in the very early phase of transplantation and that their main activity is employed in the absence of Tcon.

In this study, we demonstrate that freshly isolated natural occurring third-party Treg are a valuable and useful alternative to Treg derived from the same donor as the Tcon, as both effectively suppress alloreactive T cell proliferation and GVHD. The major limitation to the adoptive transfer of third-party Treg is their shorter survival in vivo in comparison with donor-derived Treg, which were present for longer periods of time in lymphoid tissues and GVHD target organs. Interestingly, donor-derived Treg could be found in the peripheral blood as well as in lymphoid tissues. Although donor cell rejection occurs because of host-versus-graft reactions that can be sustained by host T cells and recently described subsets of host NK cells (33), third-party Treg may be also rejected by donor Tcon because they are MHC mismatched to both host and donor.

Despite their inferior survival compared with donor Treg, third-party Treg demonstrated effective protection from GVHD. This observation prompted us to study the timing of Treg function in vivo to investigate the mechanism of third-party Treg function. We found that adoptively transferred Treg exert their protection against GVHD mainly in the very early phase after transplantation, thus explaining why third-party Treg are functional even when rejected early. Treg rapidly home to secondary lymphoid organs after injection and a few days later circulate to peripheral tissues (15). In our experiments, even if in vivo Treg depletion occurred only 2 d after Treg injection (day 0), Treg were able to reduce GVHD lethality, proving that their function is mainly sustained by their early presence in lymphoid tissues. Donor and third-party Treg similarly reduce Tcon activation in lymphoid tissues, limiting their further infiltration of GVHD target organs, confirming that in vivo Treg control over Tcon proliferation mainly happens in lymph nodes and is an early event after Treg adoptive transfer. Early rejection of third-party Treg, even if it induces a worse GVHD score profile and a further reduction of body weight in the animals, is not enough to abrogate Treg function, as third-party Treg are anyway able to limit GVHD lethality in the first days after transplantation.

The fact that third-party Treg were efficacious in controlling GVHD also demonstrates that the main mechanisms through which Treg exert their in vivo function are independent of Treg/Tcon MHC identity. Others showed that posttransplant residual host-type dendritic cells interact with donor Tcon, triggering GVHD (34–36). Treg need to be present during the Tcon priming phase for effective suppression (37), and recently they have been shown to disrupt dendritic cell/Tcon interactions, inducing suppression of in vivo Tcon proliferation, limiting GVHD (38). Our data demonstrate that these early events after transplant happen in the absence of MHC identity whereas Treg/Tcon MHC disparities are only responsible for in vivo Treg survival. Furthermore, in our model, Treg exert their function in the first 2 d after adoptive transfer preceding Tcon injection, and therefore Treg function is mainly sustained by Treg interactions with lymphoid environment even in the absence of donor Tcon, suggesting that Treg modify the ability of host APCs to effectively prime Tcon and induce GVHD. Further studies are required to understand the cellular interactions involved and the mechanisms that underlie these early events after transplantation. Because Treg exert their function before Tcon expansion, it can also explain why Treg adoptive transfer is more effective in preventing GVHD when executed before Tcon injection (15).

Our insights on timing of Treg function suggest that Treg adoptive transfer is mainly effective in the early phase of transplantation and demonstrates that timing of Treg injection is crucial for their function, suggesting that delayed GVHD treatment with Treg may be minimally effective, as Treg would not be able to interfere with the Tcon priming phase. Clinical trials are ongoing with the goal of expanding the Treg pool in vivo to reduce GVHD symptoms and possibly limiting lethality (10, 39).

Others demonstrated that Treg that are specific for a third-party Ag are effective in controlling GVHD when reactivated in vivo by providing the specific Ag, suggesting that specific Treg require previous activation to exploit a broader ability of suppression (40). We transferred fresh polyclonal Treg, proving that Ag specificity may be not required or may be acquired in vivo for effective function that does not depend on MHC matching.

Moreover, Treg/Tcon MHC disparities strongly limit natural occurring Treg function when the bone marrow recipient is selected as the source of Treg. Previous studies showed that differently activated host type Treg are able to control donor Tcon proliferation, thus reducing GVHD lethality (6, 41). We demonstrated that adoptive transfer of freshly isolated unmanipulated host-type Treg do not prevent GVHD and do not have any impact on animal survival even if their in vivo detection is still possible several days after injection. Therefore, host Treg may require activation to exert their effect, whereas third-party Treg are functional even in the absence of any ex vivo or in vivo manipulation but presumably are activated in vivo following GVHD induction.

By analyzing different mouse tissues after transplantation, we were also able to detect previously transferred Treg in the peripheral blood as well as spleen, lymph nodes, and liver, demonstrating that Treg presence in the peripheral blood is reflective of their infiltration of lymphoid and peripheral tissues. Such data allow for peripheral blood monitoring of Treg transfer, providing a relevant tool for clinical translation.

When we induced GVHD in an MHC minor mismatch model of transplantation, there was no impact of Treg/Tcon MHC disparities on mouse survival and on GVHD onset. In this model higher numbers of Tcon are required for inducing lethal GVHD, but even Treg that have been injected at an extremely low Treg/Tcon ratio (1:50 in our study) resulted in protection and prolonged survival. Both donor-type and third-party Treg were effective in GVHD protection. Because donor Tcon are activated in vivo by minor Ags and graft–versus-host reactions are weaker because of their MHC class I identity (42), we might expect that third-party Treg that are MHC major mismatched with the host environment would be more effective in this model. Somewhat surprisingly, we found that third-party Treg protect mice from GVHD as well as did donor Treg, strongly suggesting that the role of MHC disparities in Treg activation and the survival of transferred Treg in vivo have limited impact on GVHD protection when minor Ags work as GVHD triggers. In this context, third-party Treg may be rejected less by the donor T cell pool, survive longer, and possibly provide effective in vivo suppressive function even if limited in number, thus explaining the similar survival that we observed in mice that received donor or third-party Treg. Because this model better resembles HLA conditions in human HLA-matched transplantation, these data promote third-party Treg cellular therapy for GVHD prevention in this clinical setting.

Steiner et al. (43) described third-party Treg as an effective alternative tool when injected for overcoming experimental-induced rejection. Our work confirms their observations highlighting third-party Treg value in suppressing GVHD and introduces relevant insights on Treg mechanism of in vivo tolerance induction. In the Steiner et al. transplantation model, donor and third-party Treg are injected in mice that received host type T cells for rejection induction, and therefore both may be rejected by host-mediated alloreactions. In our model the host immune system is quickly overcome by donor T cells, resulting in rapid engraftment and GVHD; even in these conditions that favor donor cell persistence, third-party Treg are still effective in suppressing GVHD.

One of the major limitations to a broader clinical application of Treg adoptive transfer is the difficulty in obtaining enough cells from a donor due to Treg paucity in the periphery. The use of third-party Treg may overcome this issue, as they can be harvested from several donors, pooled when needed, prepared in advance, and banked, allowing for a readily available cell product for clinical use.

MHC matching is a key factor in transplantation, and the new introduction of cellular therapies requires better understanding of MHC interactions and roles in this setting. We think that our study elucidates critical impact of Treg/Tcon MHC disparities and provides important insights into the selection of the source of Treg for the clinical application of Treg cellular therapy in GVHD prevention. Although HLA-matched sibling donors can often provide fresh natural occurring Treg, our findings demonstrate that the use of third-party Treg is a valuable alternative to donor-derived Treg immunotherapy in other clinical settings, where HLA-matched donors are not readily available.

Our study also brings relevant insights about timing of in vivo Treg function and helps elucidate the mechanisms through which Treg suppress GVHD in preclinical and clinical settings. The very early phase of transplantation is extremely important and regulates transplantation tolerance and GVHD onset and lethality. Treg adoptive transfer plays a key role in this crucial phase of transplantation, and treatment interventions that modify the balance between regulatory and effector interactions during this phase strongly impact transplantation outcomes.

We thank the Stanford Shared FACS Facility and the Stanford Center for Innovation in In-Vivo Imaging (SCI3) for providing facilities for FACS analysis and in vivo imaging.

The authors have no financial conflicts of interest.