Absence of STAT1 in donor-derived plasmacytoid dendritic cells results in increased STAT3 and attenuates murine GVHD

Allogeneic hematopoietic stem cell transplant (alloHSCT) is curative for multiple malignant and nonmalignant diseases. T cells in the donor graft play a critical role in engraftment, antiviral immunity, and graft-versus-tumor (GVT) effects, contributing to remissions and cures.¹  However, the therapeutic benefit of increasing T-cell dose in the stem cell graft or as a donor lymphocyte infusion (DLI) is limited by graft-versus-host-disease (GVHD), a major life-threatening complication subsequent to alloHSCT. Despite advances in the matching of donors and recipients at major histocompatibility complex (MHC) alleles and improvements in immunoprophylaxis, GVHD from mismatched minor histocompatibility antigens (miHAs) occurs in up to 65% of recipients, leading to significant morbidity and mortality.²  Although there is a variety of medications available to prevent and treat GVHD, these drugs mainly target global T-cell function, impairing immunity to pathogens and to malignancy, resulting in increased incidence of infection and relapse.³  Thus novel strategies are needed that can prevent GVHD and can preserve immune competence.

After conditioning with cytotoxic chemotherapy and/or total body irradiation for alloHSCT, there is a “cytokine storm” reflected by an increase in inflammatory cytokines. These cytokines activate APCs in the recipient,²  which act to prime donor T cells against host antigens, inducing GVHD.⁴  One family of cytokines that has drawn considerable interest for its role in GVHD/GVT is the interferon family, which serves as a critical interface between innate and adaptive immunity and is a major component of the alloreactive environment.⁵  Interferon signaling can produce differential effects on GVHD depending on whether responding cells are of donor^6-8  or host origin,^7,9,10  and if they are interrupted in transplanted T cells^6-8,11,12  or non-T cells.^6,7 

Because both type I and type II interferons signal through the transcription factor signal transducer and activator of transcription-1 (STAT1),¹³  there is interest in the role of this transcription factor in GVHD.^2,5,14  Furthermore, developing therapies for GVHD that spare T cells are appealing, because they may preserve or even enhance GVT effects. STAT1 gene expression is elevated in GVHD,¹⁵  and data suggest that adenosine triphosphate stimulates APCs through STAT1 to exacerbate GVHD.¹⁶  In addition, correlative evidence in adults with chronic GVHD demonstrates elevated expression of STAT1.¹⁷  Finally, infusion of donor CD4⁺ T cells deficient in STAT1 were shown to abrogate GVHD through the expansion of regulatory T cells (Tregs).¹²  Although effective in preventing GVHD, targeting STAT1 solely in donor T cells may unintentionally impair overall immunity because deficiency of type I⁷  and type II⁶  interferon signaling in donor T cells attenuates their ability to cause GVT. The goal of this study was to assess the effects of STAT1 modulation in donor-derived APCs as an innovative means of preventing GVHD by leaving T-cell function intact and thus potentially preserving GVT.

C3H.SW (H-2^b), C57BL/6 (B6) × C3H.SW (H-2^b), B6 IFNγR1^−/− (H-2^b), B6.CD11c-cre (H-2^b), B6.129P2 CD19-cre (H-2^b), and B6.129P2 Lysozyme (Lys)-cre (H-2^b) mice were purchased from The Jackson Laboratory (Bar Harbor, ME). 129S6 (H-2^b) and 129S6 STAT1^−/− (H-2^b) mice were purchased from Taconic Farms (Hudson, NY). B6 interferon regulatory factor-9 (IRF9)^−/− mice (H-2^b) were generated by Tadatsugu Taniguchi¹⁸  and were purchased from RIKEN BioResource Center (Ibaraki, Japan). B6 (H-2^b) and C3H.HeNCr (H-2^k) were purchased from the National Cancer Institute (NCI) Animal Production Program (Frederick, MD). Strain 129 mice with a floxed region flanking the STAT1 gene (STAT1^flox/flox) were generated as previously described.¹⁹  S100A9^−/− mice (H-2^b) were a gift from Dr Dmitry Gabrilovich (H. Lee Moffitt Cancer Center, Tampa, FL). Wild-type (WT)-1 T-cell receptor transgenic mice (H-2^b) were generated as described in the supplemental Methods (available on the Blood Web site). Mice were age-matched and were used between 4 and 8 weeks of age. All animals were housed in a pathogen-free facility throughout the study. The Animal Care and Use Committee at the National Institutes of Health approved all protocols.

Donor mice were euthanized by CO₂ asphyxiation following National Institutes of Health Animal Care and Use Committee guidelines. Bone marrow (BM) cells were harvested from the tibias and fibulas by crushing with mortar and pestle in 10% complete mouse media (CMM).⁶  BM cells were passed through a 70-μm nylon filter, and were depleted of erythrocytes using ACK Lysing Buffer (Lonza, Walkersville, MD). Single-cell suspensions were then labeled with CD3 microbeads (Miltenyi Biotec, Auburn CA), followed by a magnetic separation using the AutoMACS Pro separation system (Miltenyi Biotec). Lethally irradiated (10 Gy) recipient mice were injected intravenously through the tail vein on day +0 with 4 × 10⁶ (10 × 10⁶ was used for C3H.HeNCr recipients) CD3-depleted BM cells in serum-free RPMI (Invitrogen). Mice were weighed individually biweekly, and the mean weight of each treatment group was calculated at each time point and was compared with the day +0 weight. GVHD was monitored using a clinical scoring system.²⁰  Examination for moribund mice was performed by a veterinarian and veterinary technicians who were blinded to the experimental groups, and they assessed the mice daily in accordance with approved institutional protocols.

GVHD was induced using a DLI administered either on day +10 or +14 after transplantation. In some experiments a second DLI was infused on day +28. DLIs were generated from single-cell suspensions of pooled cervical, inguinal, and axillary lymph nodes obtained from WT mice matched to the BM donor. Lymph node cells were resuspended in serum-free RPMI and were passed through a 70-μm nylon filter before injection, and then they were administered intravenously through the tail vein when indicated.

Off-target inhibition of STAT1 was achieved by injecting donor mice intraperitoneally with (24 nM) 2.5 ug of Exenatide (Sigma-Aldrich) dissolved in 0.2 mL phosphate-buffered saline (K-D Medical, Columbia, MD) 2 days before BM harvest. Recipient mice continued to receive intraperitoneal injections after TCD alloHSCT once daily for 10 days.

Spleen and BM cells were cultured at 1 × 10⁶ cells/mL in CMM at 37°C with 1 ug/mL CpG oligodeoxynucleotides (gift from Dr Dennis Klinman; NCI, Frederick, MD) for 3 hours in CMM. Supernatants were harvested and analyzed by enzyme-linked immunosorbent assays (ELISA) for murine IFNα (PBL Interferon Source, Piscataway, NJ) and interleukin (IL)-10 (R&D Systems, Minneapolis, MN) according to manufacturer’s instructions. ELISA plates were read on a VersaMAX Microplate Reader at 450 nm and were analyzed using SoftMAX Pro 5 reader (Molecular Devices, Sunnyvale, CA). cDNA was synthesized from the cultured cells, and quantitative RT-PCR was performed using primers specific for IFNα and IL-12 on the StepOne Plus Real-Time PCR System (Applied Biosystems, Foster City, CA). Gene expression was calculated using the ΔΔC_t method.

In brief, 1 × 10⁶ freshly isolated, erythrocyte-depleted splenocytes, lymph node, or BM cells were treated with anti-FcγIII/II receptor monoclonal antibody (moAb) clone 2.4G2 and then were stained at 4°C for 20 minutes with an moAb cocktail containing either CD11c or NK1.1-FITC, B220-PE, CD4 or CD45.2-PerCP Cy5.5, CD8 or CD45.1-APC, CD11b-Pacific Blue (BD Biosciences), CD9-APC, or Siglec H-PerCP Cy5.5 (eBioscience, San Diego, CA), and then they were washed in fluorescence-activated cell-sorting buffer (phosphate-buffered salt solution with 0.2% fetal calf serum and 0.1% sodium azide). To examine for the presence of Tregs, surface staining for CD25-FITC and CD4-PerCP Cy5.5 were performed, followed by Foxp3-PE intracellular staining according to the manufacturer’s instructions (eBioscience).

Depletion of pDCs was achieved using a moAb against murine PDCA-1 (Miltenyi Biotec) resuspended in a phosphate-buffered saline/2 mM EDTA solution (both from K-D Medical). Recipients of a TCD alloHSCT were injected intraperitoneally 2 days before their DLI with 250 ug of PDCA-1 or IgG2b control (Sigma-Aldrich). Mice continued to receive injections every other day for a total of 5 injections. Depletion of pDCs was verified by flow cytometric analysis of representative spleens from each group 24 hours after their final intraperitoneal injection.

The MB49 tumor cell line is derived from a chemically induced urothelial carcinoma in a male B6 mouse and it expresses the male-specific miHA H-Y.⁶  MB49 cells were maintained in culture at 37°C in 5% CO₂ in CMM. Exponentially growing tumor cells were prepared as a single-cell suspension in serum-free media and were injected into the subcutaneous fat of the flank at a dose of 2 × 10⁶ tumor cells on day 42 after BMT. Tumors were measured in 2 dimensions (length × width) 2× per week by digital caliper. Mice were euthanized with CO₂ when tumor diameters reached 2 cm, in accordance with animal protocols. If a mouse was found dead, the previously recorded tumor measurement was carried for the rest of the experiment.

Statistics were performed using GraphPad Prism version 4.0c for the Macintosh OS (GraphPad Software, San Diego, CA). Significant differences when comparing 2 groups were determined by the 2-tailed Mann-Whitney test. The Kruskal-Wallis with Dunn’s multiple-comparison posttest was used to assess statistical differences among 3 or more groups. A P value <.05 was considered statistically significant.

Based on our previous observation that the lack of IFNγ receptor on donor non-T cells prevented GVHD,⁶  we first wanted to determine whether a similar effect occurred by interfering with STAT1, the main component of the canonical signaling pathways of both type I and II interferons, in donor hematopoietic cells. Using a MHC-matched, miHA-mismatched alloHSCT model (129 → C3H.SW) we observed that when STAT1^−/− TCD BM was followed by delayed infusion of donor-derived STAT1^+/+ T cells as a DLI administered on day +14, there were improved clinical GVHD scores (Figure 1A) as a result of the prevention of GVHD-associated weight loss (supplemental Figure 1A), with improved B-cell reconstitution (Figure 1B) and expansion of Tregs (Figure 1C). The GVHD protection from STAT1^−/− BM was T-cell dose-dependent (Figure 1D), as doubling the DLI dose from 20 to 40 × 10⁶ suppressed Treg expansion (Figure 1E) and caused weight loss (Figure 1D and supplemental Figure 1B). GVHD-associated, DLI-induced weight loss was also prevented in a different miHA-mismatched alloHSCT model (129 → B6 × C3H.SW, supplemental Figure 1C). Interestingly, in the setting of T-cell replete (TCR) BM grafts, use of STAT1^−/− BM did not impact GVHD (supplemental Figure 1D). We hypothesized that T cells in the BM graft would encounter persisting recipient STAT1^+/+ APCs following TCR transplant whereas delayed DLI would occur in the context of STAT1^−/− APCs. Indeed, we found that total spleen (supplemental Figure 1E) and splenic dendritic cells (DCs) (supplemental Figure 1F) were mainly donor derived by 14 days post-alloHSCT irrespective of the BM source.

To determine if pharmacologic inhibition would replicate the results with BM genetically deficient in STAT1, we treated mice in the peritransplant period with Exenatide, a glucagon-like peptide approved for treatment of type II diabetes and shown to inhibit of STAT1 in vitro and in vivo (supplemental Figure 1G).^22,23  WT B6 mice were treated with Exenatide and were used as BM donors for transplantation into C3H.SW recipients, which also received Exenatide after transplant but before DLI, to mimic STAT1 deficiency during engraftment. Indeed, similar to transplantation of STAT1-deficient BM, we found that Exenatide decreased GVHD scores (Figure 1F) and attenuated weight loss (Figure 1G).

We next tested whether recipients of female STAT1^−/− BM and female T cells could respond to a male vaccine at the time of DLI as measured by resistance to HY-expressing tumor on day +42 (Figure 2A). T cells were required for tumor control, as recipients of no DLI experienced faster onset of large tumors. Although there was a slight delay in tumor growth after DLI and vaccination against HY in recipients of STAT1^+/+ BM, ultimately recipients succumbed to large tumors associated with the development of GVHD. In contrast, recipients of STAT1^−/− BM, DLI, and HY vaccination controlled tumor growth (Figure 2B), resulting in improved overall survival (Figure 2C) likely from improved GVHD.

Because TCD STAT1^−/− BM was sufficient to protect against GVHD, we hypothesized that a donor-derived APC population was responsible. Using STAT1^flox/flox mice expressing Cre recombinase under the control of promoters active in APCs (CD11c [DCs], Lys [macrophages and neutrophils], or CD19 [B cells]), we examined the role of specific STAT1-deficient APC subsets in modulating GVHD. However, selective STAT1 deficiency in donor-derived CD19⁺ cells, CD11c⁺ cells or Lys-expressing cells did not protect against GVHD after MHC-matched BMT (CD11c Cre⁺ BM [B6 → C3H.SW, supplemental Figure 2A], Lys Cre⁺ BM [B6.129 → C3H.SW, supplemental Figure 2B], CD19 Cre⁺ BM [B6.129 → C3H.SW, supplemental Figure 2C]). Similar results were observed after lethal MHC-mismatched BMT (H-2^b → H-2^k) for TCD CD11c-Cre⁺ (B6 → C3H.HeNCr, Figure 3A) and TCD CD19-Cre⁺ BM (B6.129 → C3H.HeNCr, Figure 3C). Interestingly, transplantation of TCD Lys Cre⁺ BM exacerbated DLI-induced GVHD lethality (B6.129 → C3H.HeNCr, Figure 3B), potentially reflecting modulation of suppressive BM-derived myeloid cells.²⁴ 

We next looked at recipients of TCD whole STAT^−/− BM to determine if immune reconstitution of a particular cell subset could indicate the specific STAT1-deficient donor cell mediating GVHD protection. At the time of DLI on day +14, there were no quantitative differences in total splenocytes, BM cells, DCs, NK cells, Gr-1⁺ cells, CD4⁺ cells, or CD8⁺ cells (supplemental Figure 3A-D). However, the percentage (Figure 3D) and absolute number (Figure 3E) of plasmacytoid DCs (pDCs) in the spleen and BM was significantly increased in recipients of TCD STAT1^−/− BM. Chimerism of pDCs from either WT or STAT1^−/− BM was equivalent and showed donor-derivation by Day +14, at the time of DLI (Figure 3F). Furthermore, isolated pDCs from all 3 of the cre recombinase STAT1^flox/flox mice used as BM donors in Figure 3A-C expressed STAT1 (Figure 3G), suggesting that intermediate expression of CD11c in pDCs was not sufficient to delete STAT1 in STAT1^flox/flox mice.

We hypothesized that pDCs reconstituted from STAT1^−/− BM and unable to respond to interferons would be resistant to activation following alloHSCT. Indeed, the proportion of STAT1^−/− pDCs producing IFNα, the predominant cytokine produced by activated pDCs, was reduced in the spleen on day +14 post-alloHSCT in recipients of STAT1^−/− BM (Figure 4A-B). This was confirmed by reduced IFNα gene expression (Figure 4C) and by reduced ex-vivo IFNα production in sorted pDCs (Figure 4D) from BM following alloHSCT consistent with compartment-specific effects. Interestingly, there was no difference in gene expression or IFNα production by splenic pDCs. Importantly, there was no reduction in IFNα production by pDCs isolated from the BM of naïve STAT1^−/− mice, suggesting that the difference in pDCs reconstituted from STAT1^−/− donors reflected altered response to interferons in the post-alloHSCT environment. Furthermore, we observed decreased IL-12 (Figure 4E) and increased IL-10 (Figure 4F) by STAT1^−/− pDCs.

Data suggest that pDCs can be subclassified into stimulatory or tolerogenic populations based on CD9 and Siglec H expression.²⁵  Indeed, we found an increased ratio of tolerogenic (CD9⁻Siglec H^hi) to stimulatory (CD9⁺Siglec H^lo) pDCs in the BM of STAT1^−/− BM recipients (Figure 4G and supplemental Figure 4A). Taken together, these results demonstrate that increased pDCs in the BM STAT1^−/− BM recipients reflect a specific expansion of a CD9⁻SiglecH^hi pDC subset with reduced IFNα and IL-12-production and increased IL-10 production.

Because the reduced IFNα production and tolerogenic phenotype could be a secondary phenomenon related to reduced inflammatory cytokines from decreased GVHD, we next assessed the functional capacity of pDCs recovered from untransplanted STAT1^+/+ vs STAT1^−/− mice in vitro. Bacterial-derived CpG-rich oligonucleotides have been shown to activate potently pDCs via TLR9. As shown in Figure 5A, the amount of IFNα produced by CpG ODN-stimulated pDCs recovered from STAT1^−/− mice was markedly reduced compared with pDCs from STAT1^+/+ mice. GVHD causes substantial inflammation that activates DCs, leading to free-radical production that potentially contributes to DC function specifically downstream of STAT1.²¹  Thus, we next measured peroxynitrite, as measured by levels of 3-nitrotyrosine, in cell supernatants from expanded pDCs generated from TCD WT STAT1^+/+, IRF9^−/−, IFNγR1^−/−, and STAT1^−/− BM. Although pDCs expanded from IRF9^−/− BM or IFNγR1^−/− BM produced decreased levels of nitrotyrosine compared with STAT1^+/+ BM, pDCs expanded from STAT1^−/− BM produced the least amount of nitrotyrosine, suggesting some interference with free-radical generation and, potentially, GVHD-induced tissue injury (Figure 5B).²⁶  We next assessed the capacity of STAT1^−/− pDCs recovered post-alloHSCT to stimulate T cells in vitro. Indeed, pulsed pDCs isolated from STAT1^−/− BM recipients pulsed with Db126, the immunodominant peptide derived from WT-1, induced less IFNγ production by WT-1–specific T cells when compared with those stimulated by pDCs from STAT1^+/+ BM recipients (Figure 5C). Interestingly, there were no differences in proliferation by TCR Tg T cells (supplemental Figure 4B).

We next established whether the expanded pDC population we identified in recipients of STAT1^−/− BM was necessary for GVHD prevention. After MHC-matched, miHA-mismatched TCD alloHSCT (129 → C3H.SW), recipients were treated with a moAb against PDCA-1 to deplete pDCs during the period around DLI (Figure 6A) resulting in a decrease in splenic pDCs (Figure 6B). Clinical GVHD scores were significantly increased in recipients of STAT1^−/− BM after pDC depletion (Figure 6C), with a decrease in B220⁺ cells in the spleen (Figure 6D) demonstrating that pDCs are critical for full GVHD protection mediated by STAT1^−/− BM grafts. In fact, we observed that adoptive transfer of isolated STAT1^−/− pDCs to alloHSCT recipients of STAT1^+/+ BM prevents GVHD. In contrast, adoptive transfer of STAT1^+/+ pDCs to alloHSCT recipients of STAT1^−/− BM does not exacerbate GVHD, implying the GVHD-mediated protection by STAT1^−/− pDCs is a dominant effect (Figure 6E).

The number of interferon-stimulated genes activated by STAT1 is estimated to be between 600 and 2000.²⁷  Thus, to begin to identify relevant downstream targets of STAT1 involved in GVHD protection, we isolated pDCs generated from TCD STAT1^+/+ and STAT1^−/− BM alloHSCT recipients and performed microarray analysis on cDNA prepared from these cells. The data discussed in this publication have been deposited in the National Center for Biotechnology Information’s Gene Expression Omnibus and are accessible through Gene Expression Omnibus Series accession number: GSE59731. Ingenuity Pathway Analysis identified a twofold increase in the expression of S100A8 and S100A9 (Figure 7A), which are molecules previously associated with myeloid-derived suppressor cells.²⁸  Using quantitative RT-PCR, we verified a 2- to 2.5-fold increased gene expression of S100A8/A9 in pDCs generated from recipients of TCD STAT1^−/−BM as compared with recipients of TCD STAT1^+/+ BM (Figure 7B). Correspondingly, TCD S100A9^−/− BM exacerbated GVHD-free survival after DLI (Figure 7C), with worse clinical scores at the time of takedown (supplemental Figure 4C). Finally, because induction of these molecules has been associated with STAT3,²⁸  a transcription factor important for immune tolerance,²⁹  we wanted to determine whether the absence of STAT1 led to unopposed STAT3 levels in pDCs. Indeed, isolated pDCs from the BM, spleen, and lymph nodes of chimeras generated from TCD STAT1^−/− BM demonstrated increased total STAT3 gene expression (Figure 7D) and STAT3 protein from pDC lysates (Figure 7E).

Preventative GVHD strategies that preserve T-cell function and immunocompetence are appealing for protection against relapse and infections, but they remain elusive. Using MHC-matched, miHA-mismatched alloHSCT models, we found that non-T cells reconstituted from a STAT1-deficient allograft prevent GVHD induced by delayed add-back of high doses of fully competent T cells as a DLI that can respond to DC vaccination resulting in enhanced tumor protection. Furthermore, we demonstrate that pharmacologic off-target inhibition of STAT1 using Exenatide^22,23  can also protect against GVHD. This is consistent with previous observations that inhibition of STAT1 can abrogate inflammation in models of solid organ transplantation.^30,31  Although targeting of STAT1 has been tested in models of GVHD, these studies relied on the manipulation of STAT1 in donor lymphocytes.¹²  Although effective in preventing GVHD, targeting STAT1 solely in donor T cells may unintentionally impair overall immunity because deficiency of type I⁷  and type II⁶  interferon signaling in donor T cells attenuates their ability to cause GVT. Based on our earlier observation that TCD donor BM deficient in IFNγR1 could prevent GVHD, we took a novel approach by targeting STAT1 in donor non-T cells. Using this model, we identified an expanded STAT1-deficient pDC population that was activation deficient and was required for GVHD protection.

Previous studies have shown that treatment with suberonylanilide hydroxamic acid, a histone deacetylation inhibitor, after MHC-mismatched alloHSCT led to decreased GVHD mediated by inhibition of STAT1.³²  One possible strategy to translate our findings is based on the effectiveness of the restricted timing of pharmacologic STAT1 inhibition as demonstrated in the Exenatide experiments. Using our approach, the patient would only need to be treated transiently until DLI, and it is hoped that this approach would preclude the long-term risks immunosuppression. In addition to suberonylanilide hydroxamic acid and Exenatide, the use of JAK1 inhibitors^8,33  and/or the development of rationale STAT1 inhibitors such as H2-2³⁴  may also be promising in GVHD.

pDCs have been shown to mediate tolerance to cardiac allografts,³⁵  and they contribute to oral tolerance.³⁶  Whether pDCs exacerbate or diminish GVHD after alloHSCT has been controversial in preclinical models. Alloantigen expression on pDCs alone is sufficient to prime alloreactive T cells and to cause GVHD when activated by total body irradiation.³⁷  However, depletion of pDCs from BM grafts accelerates GVHD mortality, and GVHD prevents maturation of pDCs leading to generation of a suppressive CD11c^lo120G8⁺ precursor pDC.³⁸  Transplanting donor pre-pDCs, or adding pDCs to donor BM grafts, can also attenuate GVHD and augment GVT.^39,40  Furthermore, CCR9⁺ pDCs can induce Tregs and can inhibit antigen-specific T-cell responses, suppressing GVHD.⁴¹  However, depletion of host-derived pDCs is not sufficient to prevent GVHD.⁴²  Interestingly in humans, decreased levels of pDCs in the circulation have been associated with an increased incidence of acute and chronic GVHD^43,44 ; children without GVHD have high absolute numbers of pDCs,⁴⁵  and absence of GVHD has been associated with improved pDC recovery and improved overall survival.⁴⁶  Collectively, these studies suggest that pDCs, depending on whether they are donor or host-derived, may be able to modulate GVHD, but they can also be activated in the posttransplant setting to support alloreactivity.

Our studies provide a novel strategy potentially to generate a tolerogenic pDC by targeted manipulation of STAT1 in the context of alloHSCT. We demonstrate that this results in expansion of immature CD9⁻SiglecH^hi STAT1^−/− pDCs in the BM and supports the fact that STAT1 is required for DC maturation.⁴⁷  It is possible that pDC expansion is also explained, in part, by the fact that the background strain was 129, which has a higher frequency of pDCs.⁴⁸  However, given that STAT1^−/− mice have normal homeostatic levels of pDCs, but an inability to expand pDC numbers above baseline via IFNα,⁴⁸  our observations that pDCs are expanded after alloHSCT suggests that other factors present in the post-alloHSCT are responsible. CD9⁻SiglecH^hi pDCs have been previously described to be tolerogenic.²⁵  Indeed, we found that reconstituting STAT1^−/− pDCs were poor producers of peroxynitrite, IFNα, and IL-12, but were increased producers of IL-10. Interestingly, STAT1^−/− macrophages are also poor producers of superoxide-free radicals.⁴⁹ 

We also found that pDCs reconstituted from STAT1-deficient BM also expressed increased levels of S100A8/A9 as well as STAT3. Although expression of S100A8/A9 has been associated with the suppressive activity of myeloid cells,²⁸  our observations suggest that S100A8/A9 may also be important in tolerogenic pDCs. Indeed, we found that absence of these molecules from donor BM exacerbates GVHD. STAT1 and STAT3 have been described to oppose one another in other models and through a number of mechanisms such as suppressors of cytokine signaling.^50,51  Furthermore, increased expression of STAT3 has been associated with immune tolerance.²⁹  Thus, it is interesting that we observed increased STAT3 in pDCs derived from a STAT1-deficient allograft. Additional studies will be necessary to confirm whether the increased STAT3 is required for GVHD protection. Finally, we observed that recipients of STAT1^−/− BM showed an expansion of Tregs, and we hypothesized that the pDCs are driving Treg generation. Expansion of activated human pDCs has been previously associated with Treg generation in vitro,⁵²  and further studies should demonstrate whether the Tregs are required for the prevention of GVHD in this model.

In conclusion, inhibition of STAT1 in donor allografts may be a viable approach for protecting patients from GVHD and may provide a platform for infusing high doses of T cells as a delayed DLI. Importantly, we show that STAT1 inhibition is only needed for a short period after engraftment until T-cell infusion, potentially providing opportunities for restricted timing of STAT1 inhibitors and preventing long-term immunosuppression. Indeed, intact vaccine-mediated tumor protection observed in our studies suggests that STAT1 inhibition may allow for the incorporation of emerging immunotherapeutic approaches that are being developed outside of the allogeneic setting.

The authors thank Crystal Mackall for her valuable input and John Buckley for his technical expertise.

This work was supported by grants from the National Institutes of Health/NCI K08 CA174750, Hyundai Hope on Wheels, the Midwest Athletes Against Childhood Cancer Fund, Lisa’s Heart Kids Cancer Research Fund (C.M.C.), the Children’s Cancer Foundation (T.J.F.), and the Intramural Research Program at the National Institutes of Health (J.K., L.H., T.J.F.).

The content of this publication does not necessarily reflect the views or policies of the Department of Health and Human Services, nor does mention of trade names, commercial products, or organizations imply endorsement by the US Government.

Contribution: C.M.C. and N.M.N. designed and performed research; collected, analyzed, and interpreted data; performed statistical analysis; and drafted the manuscript; C.D.C. performed research and collected, analyzed, and interpreted data; S.M.L. and H.Q. performed research; collected, analyzed, and interpreted data; and performed statistical analysis; P.J.K. and L.H. provided STAT1^flox/flox mice, assisted with breeding and genotyping of mice, and revised the manuscript; Y.K.S. and J.K. performed microarray, analyzed and interpreted data, and revised the manuscript; T.J.F. designed and supervised research, analyzed and interpreted data, performed statistical analysis, and revised the manuscript.

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

Correspondence: Christian M. Capitini, University of Wisconsin, 1111 Highland Ave, WIMR 4137, Madison, WI 53705; e-mail: ccapitini@pediatrics.wisc.edu.