IFN-γ–STAT1–iNOS Induces Myeloid Progenitors to Acquire Immunosuppressive Activity

Autoimmune diseases often induce dysregulated hematopoiesis with altered number and function of hematopoietic stem and progenitor cells (HSPCs). However, there are limited studies on the direct regulation of HSPCs on T cells, which are often detrimental to autoimmunity. Here, we found that in a murine model of Concanavalin A-induced autoimmune hepatitis, LSK (Lineage−Sca-1+c-Kit+)-like cells accumulated in liver, spleen, and bone marrow (BM), which were myeloid progenitors (Lineage−Sca-1−c-Kit+) that upregulated Sca-1 expression upon T cell-derived IFN-γ stimulation. Strikingly, BM LSK-like cells from mice induced by Con A to develop autoimmune hepatitis or alternatively myeloid progenitors from wild-type mice possessed strong in vitro suppressive ability. Their suppressive function depended on T cell-derived IFN-γ in a paracrine fashion, which induced STAT1 phosphorylation, inducible nitric oxide synthase expression, and nitric oxide production. Blocking IFN-γ/IFN-γ receptor interaction, knockout of STAT1, or iNOS inhibition abrogated their suppressive function. In addition, the suppressive function was independent of differentiation; mitomycin C-treated myeloid progenitors maintained T cell suppressive ability in vitro. Our data demonstrate a mechanism of inflammation induced suppressive function of myeloid progenitors, which may participate directly in suppressing T cell-mediated immunopathology.

while ST-HSCs and MPPs have less self-renewal capacity and can differentiate into committed progenitors like common lymphoid progenitors and common myeloid progenitors (CMPs), which replenish hematopoietic populations including lymphoid cells and myeloid cells when they are depleted or reduced by aging or consumed (3,4).
Hematopoietic stem and progenitor cells usually reside in a quiescent state within a niche of microenvironment in BM, which generate signals that regulate HSC self-renewal, quiescence, and differentiation (5). During infection and inflammation, HSPCs can be activated, mobilized, and differentiate to help against pathogens, as well as inflammation resolution (6,7). HSC transplantation has been used for decades in autoimmune disease treatment to replace autoreactive cells (8). HSPCs traffic to liver and differentiate into alternatively activated macrophages in a CCR2-dependent way to protect against acetaminophen-induced liver injury (9). Blood-derived CD34 + progenitor cells can rescue mice from severe acute liver injury and induce liver regeneration (10). However, in a murine model of IL-23-dependent colitis, intestinal inflammation increases hematopoietic stem and progenitor cell proliferation and skews their differentiation toward granulocyte-monocyte progenitors (GMP), leading to aggravation of colitis (11). Thus, HSPCs play a controversial role during acute or chronic inflammation, whether they can interact directly with T cells to modulate inflammation remains unclear.
Autoimmune diseases often induce a dysregulated hematopoiesis (12,13). Hematopoietic progenitor cells from patients with autoimmune hepatitis are significantly increased in number but functionally impaired (14). Hence, we chose the model of Con A-induced AIH to dissect the direct interaction between HSPCs and T cells. We report that LSK-like cells accumulate in BM, liver, and spleen after Con A treatment; LSK-like cells are myeloid progenitor cells, which upregulate Sca-1 expression during activation. Most importantly, these HSPCs possess powerful T cell suppressive ability in vitro and their suppressive function is acquired following T cell-derived IFN-γ stimulation, which induces STAT-1 phosphorylation, iNOS expression, and NO production, but is independent of differentiation. Our data demonstrate that myeloid progenitor cells acquire T cell suppressive activity using an IFN-γ-STAT1-iNOS pathway.
autoimmune hepatitis Model 10-week-old male mice were injected i.v. with Con A (10 mg/kg body weight, Sigma-Aldrich, St. Louis, MO, USA) and sacrificed 24 h later or at indicated time points. Serum alanine aminotransferase (ALT) level was quantitated to evaluate liver damage.
In some experiments, HSPCs were treated with 25 µg/ml Mitomycin C (Sigma) for 30 min at 37°C and washed for at least five times before adding to the coculture system; Mitomycin C-treated B cells were used as control.
For mixed proliferation experiment, 5 × 10 4 /well non-CFSElabeled OT-II T cells were added into the coculture system of WT myeloid progenitors and GKO OT-II T cells, while 5 × 10 4 / well non-CFSE-labeled GKO OT-II T cells were added as control. Proliferation of CFSE +/lo GKO OT-II T cells was evaluated.
Cells were cultured in T cell medium. After culture, dead cells were excluded by DAPI staining and T cell proliferation was assessed by CFSE dilution of B220 − CD4 + cells. Percentage of proliferation was normalized by the control system.

Transwell assay
For transwell assays, 2.5 × 10 5 CFSE-labeled OT-II T cells and 5 × 10 4 B cells with or without 5 × 10 4 WT myeloid progenitors were cultured in the top or bottom chamber of Corning Transwell-96 System (0.4 µm PC membrane, corning, NY, USA) for 3 days in the presence of 1 µg/ml OVA323-339 peptide. Cells were collected respectively and proliferation of DAPI − CD4 + T cells was analyzed by CFSE dilution.

statistical analysis
For analysis of statistical significance, we performed a two-tailed unpaired Student's t-test in GraphPad Prism 5 (GraphPad Software, San Diego, CA, USA). Data in all results were from at least two independent experiments and were shown as mean ± SEM. resUlTs T cell-Derived iFn-γ induces lsK (lin − sca-1 + c-kit + )-like cells from Myeloid Progenitors during con a-induced autoimmune hepatitis In this murine model of Con A-induced autoimmune hepatitis, the frequency (Figures 1A,B) and absolute number ( Figure 1C)  Figure 1D) following the increase of serum IFN-γ and ALT ( Figure 1E) level, suggesting that LSK-like cell accumulation was induced by inflammatory cytokines. BM LSK-like cells from Con A-treated mice had increased MHC-I/II and PD-L1 but decreased CD86 expression ( Figure 1F and data not shown). In addition, they demonstrated a higher proliferation ability after Con A treatment as indicated by Ki67 staining (Figure 1G).
In BM lineage − cells, the percentage of myeloid progenitors (Lin − c-Kit + Sca-1 − ) decreased a lot in contrast of LSK-like cells ( Figure S1C in Supplementary Material). Myeloid progenitors can be divided into three subsets, granulocyte-monocyte progenitor (GMP, Lin − c-Kit + Sca-1 − CD34 + CD16/32 hi ), common myeloid progenitor (CMP, Lin − c-Kit + Sca-1 − CD34 + CD16/32 mid ), and (MKEP, Lin − c-Kit + Sca-1 − CD34 − CD16/32 − ). We found that BM LSK-like cells in Con A-treated mice demonstrated a similar subset distribution as myeloid progenitors in WT mice ( Figure  S1D in Supplementary Material). Giemsa staining also demonstrated the morphological feature of Con A LSK-like cells resembled WT myeloid progenitors rather than WT LSK cells, such as larger size and lower nucleocytoplasmic tation ( Figure S1E in Supplementary Material). Myeloid progenitors have the potential to upregulate Sca-1 expression and develop a LSK-like phenotype under Th1 cytokine stimulation (17). Hence, we sorted myeloid progenitors and cocultured with splenic T cells in the presence of Con A. WT myeloid progenitors could upregulate Sca-1 expression and became LSK-like cells following stimulation of Con A-activated T cells. However, neither IFN-γ −/− T cells co-cultured with WT myeloid progenitors nor WT T cells co-cultured with IFN-γR −/− myeloid progenitors resulted in the induction of LSKlike cells ( Figure S1F in Supplementary Material). In addition, IFN-γR −/− myeloid progenitors failed to upregulate Sca-1 expression after IFN-γ stimulation compared to WT myeloid progenitors ( Figure S1G in Supplementary Material). Moreover, LSK-like cell accumulation was suppressed in IFN-γ −/− , IFN-γR −/− mice, and Rag1 −/− mice ( Figure S1H in Supplementary Material). These data indicate that LSK-like cells accumulating in Con A-treated mice are myeloid progenitors, which upregulate Sca-1 expression following stimulation of T cell-derived IFN-γ.

WT Myeloid Progenitors and lsK-like cells from con a-Treated Mice exhibit strong T cell suppressive ability In Vitro
As we observed altered expression of MHC class II, CD86, and PD-L1 on LSK-like cells after Con A treatment or IFN-γ stimulation, we wondered whether these LSK-like cells could regulate T cell responses directly. Hence, we cocultured LSK-like cells from Con A-treated mice with OT-II T cells in the presence of B cells and OVA323-339 peptide in vitro. Strikingly, BM LSK-like cells from Con A-treated mice almost completely suppressed the proliferation of OT-II T cells, while BM LSK cells from WT mice could not (Figures 2A,B). WT myeloid progenitors exhibited a similar suppressive potential (Figures 2C,D). Interestingly, these LSK-like cells or WT myeloid progenitors were powerful suppressors as there was still a significant inhibition even at the ratio of E:T = 1:50 ( Figure 2D). Furthermore, among three subsets of WT myeloid progenitors, GMP and CMP but not MKEP cells exhibited remarkable inhibitory function ( Figure 2E). However, the accumulated LSK-like cells in liver or spleen of Con A-treated mice showed no inhibitory function (Figure 2F). This is because these LSK-like cells were mainly MKEP cells, which did not have in vitro suppressive activity (Figure 2G). These results demonstrate that WT myeloid progenitors possessed strong in vitro T cell suppressive ability.

Myeloid Progenitors acquire T cell suppressive Function through iFn-γ-sTaT1-inOs Pathway
As LSK-like cells induced from myeloid progenitors upregulated MHC-II and downregulated CD86 expression, we first wanted to know whether their suppressive ability was acquired through inducing T cell anergy. Myeloid progenitors from H2Ab1 −/− mice, which do not express MHC-II molecules had intact suppressive ability (Figure 3A), indicating that they did not induce T cell anergy during coculture. On the other hand, although LSK-like cells from Con A-treated mice had increased PD-L1 expression, PD-L1 blockade did not influence their suppressive function ( Figure 3B). Furtherly, we found the inhibitory function of WT myeloid progenitors and Con A LSK-like cells was iNOS dependent and eNOS independent as they lost their suppressive ability when iNOS was inhibited (Figures 3C,D). Indeed, after coculture with T cells, myeloid progenitors significantly upregulated iNOS expression ( Figure 3E). Interestingly, although the suppressive ability of WT myeloid progenitors depends on NO, they failed to transmit through transwell system ( Figure 3F).
Based on the findings that the accumulation of LSK-like cells after Con A treatment depended on IFN-γ and NOS2 is one of the IFN-γ stimulated genes, we investigated whether IFN-γ produced by activated T cells was indispensable for the suppressive function of WT myeloid progenitors. Interestingly, IFN-γR −/− myeloid progenitors failed to suppress OT-II T cell proliferation ( Figure 4A). Meanwhile, myeloid progenitors could not inhibit the proliferation of IFN-γ −/− OT-II T cells, which could not produce IFN-γ upon OVA peptide stimulation ( Figure 4B). WT myeloid progenitors cocultured with IFN-γ −/− OT-II T cells failed to upregulate iNOS expression ( Figure 4C). These results indicate that the suppressive ability of myeloid progenitors depends on activated T cell-derived IFN-γ. However, the suppressive function of WT myeloid progenitors could not be restored by addition of IFN-γ ( Figure 4D), but by addition of OT-II T cells (Figure 4E), which provided IFN-γ in a paracrine fashion. These results furtherly suggested a cell-cell contact dependent mechanism for the acquirement of suppressive ability. BM LSK-like cells from Con A-treated mice had a higher level of STAT1 phosphorylation than WT myeloid progenitors ( Figure 4F). Meanwhile, STAT1 −/− myeloid progenitors could not upregulate expression of Sca-1 under IFN-γ stimulation ( Figure S1I in Supplementary Material), and they also lost inhibitory function because of defective iNOS induction (Figures 4G,H). These results indicate an IFN-γ-STAT1-iNOS pathway in myeloid progenitor suppressive function responding to T cell activation.

T cell suppressive ability of Myeloid Progenitors is independent of Differentiation
During coculture with T cells, WT myeloid progenitors could be activated and undergo proliferation ( Figure 5A) with upregulation  of CD11b and Gr-1 expression (Figure 5B), hence presenting a myeloid-like phenotype. However, the expression of CD11b and Gr-1 was not associated with activated T cell stimulation because myeloid progenitors acquired similar phenotype when cocultured with non-stimulated T cells (Figure 5B). Mitomycin C treatment, which could block cell proliferation and differentiation, did not impair suppressive function of myeloid progenitors (Figure 5C), which was still dependent of iNOS ( Figure 5D). Besides, we sorted CD11b + Ly6G hi cells and CD11b + Ly6C hi cells from the BM and spleen of WT and Con A-treated mice and found that they could not inhibit the proliferation of T cells as myeloid progenitors (Figures 5E,F and data not shown). These results indicate that the inhibitory ability of myeloid progenitors was independent of differentiation.

DiscUssiOn
Our present work demonstrates the suppressive potential of myeloid progenitors and reveals the mechanism that leads to this ability. Different from a previous report that early myeloid progenitors can be immunosuppressive cells in a NO-dependent way (18), we used a more specific coculture system, which may be implicated in suppression of antigen-specific T cell activation. We also demonstrated a mechanism of inflammation-induced suppressive function of myeloid progenitors. Inflammatory conditions such as found in a tumor microenvironment can affect HSPC development and function (4, 6). In cancer patients, circulating HSPCs are myeloid-biased, and tumor cells can induce early myeloid differentiation into immunosuppressive neutrophils, which resemble granulocytic myeloidderived suppressor cells (MDSCs) (19,20). MDSCs are described as a heterogeneous population of myeloid progenitor cells and immature myeloid cells, which have remarkable ability to suppress T cell response especially in tumor (21,22). It is reported that tumor-derived GM-CSF are sufficient to induce MDSCs from myeloid progenitors, but the suppressive function of these tumorinduced MDSCs was independent of IFN-γ (23,24). Myeloid dendritic cell precursors appear transiently during BM cell culture with GM-CSF and can suppress T cell response in vitro in a cell contact and NO-dependent way (25). Without the signals provided by BM niche, we observed spontaneous differentiation of myeloid progenitors in vitro with the phenotype of CD11b + Gr-1 + , resembling granulocytic MDSCs. However, their differentiation was not affected by coculture with T cells, or by IFN-γ stimulation. Importantly, mitomycin C-treated myeloid progenitors, which lost proliferation and differentiation ability maintained in vitro suppressive function. Our data demonstrate that myeloid progenitors acquire T cell suppressive function under the effect of inflammatory cytokine IFN-γ without dependency of differentiation into mature or immature myeloid cells and were different from MDSCs. Antigen-specific CD4 + T cells could convert immunosuppressive MDSCs into powerful non-specific suppressor cells, which depend on cell-cell contact and crosslinking of MHC class II molecule on MDSCs (26). We also found increased MHC class II molecule expression on LSK-like cells after ConA treatment, as well as after coculture with T cells. Besides, they showed decreased costimulatory molecule CD86 expression. Based on the in vitro suppressive function of these LSK-like cells and myeloid progenitors, our initial hypothesis was that they could induce T cell anergy by providing "signal 1" without "signal 2" during T cell activation. However, myeloid progenitors with MHC class II knockout maintained in vitro suppressive function. Also, addition of extra anti-CD28, which provide "signal 2" did not impair their suppressive ability in vitro (data not shown). IFN-γ can induce the expression of MHC class II, PD-L1, and CD86 (27)(28)(29). However, block of PD-1-PD-L1 interaction did not impair the suppressive function of myeloid progenitors. We found that myeloid progenitors express very low level of CD86. After IFN-γ stimulation, they shifted to LSK-like cells, but did not upregulate CD86 expression, resulting in decreased CD86 expression. These results indicated that increased MHC class II and PD-L1, and decreased CD86 expression of LSK-like cells during AIH were not associated with their interaction with T cells.
IFN-γ can activate HSPCs and induce expansion of LSK cells, as well as a myeloid-biased differentiation of HSPCs (17,30). However, chronic IFN-γ stimulation inhibits the generation of myeloid progenitors and prevents myeloid lineage differentiation (31). In our study, we found that T cell-derived IFN-γ functioned through IFN-γRI on myeloid progenitors, which induced the phosphorylation and translocation of STAT1 to the nucleus. Further, p-STAT1 induced the expression of iNOS and the production of NO, which suppressed T cell activation. On the other hand, myeloid progenitors cocultured with GKO OT-II T cells have a higher proliferation ratio than with OT-II T cells, suggesting that IFN-γ suppressed their proliferation in vitro. However, although extra IFN-γ administration did not impair their suppressive ability on OT-II T cells, it could barely restore their suppressive ability on GKO OT-II cells. Thus, we suggested that the suppressive ability of myeloid progenitors may depend on a close contact with T cells and the pulse stimulation of IFN-γ. But whether they can maintain suppressive ability during chronic inflammation needs further investigation.
Although IFN-γ stimulated myeloid progenitors suppressed T cells in a NO-dependent way, our result showed that their suppressive ability failed to transmit through a transwell system. Low dose of NO is reported to promote type I T cell differentiation, while high dose of NO inhibited the proliferation of T cells (32). On the other hand, NO diffuses quickly from its source (33,34), but the concentration of the active form drops sharply within about 100 mm (35). Therefore, NO can act only in close proximity to the cells producing it. In our antigen-specific coculture system, the number of myeloid progenitors may not be enough to generate sufficient NO to transmit through a transwell system. Besides, NO may be consumed by T cells cocultured with myeloid progenitors.
Bone marrow is thought to be an immune privileged site, in which stromal cells and Treg cells compose a tolerated niche to protect HSPCs from environmental insults (36,37). Interestingly, BM is also a major reservoir and site of recruitment for memory T cells and a preferential homing site for autoreactive T cells in type I diabetes (38,39). In our work herein, we found myeloid progenitors could suppress antigen-specific T cell activation in vitro. T cells in the BM are reported to locate near stromal cells, as well as HSPCs (40). Thus, we raise the thesis of whether myeloid progenitors can interact with BM-resident memory T cells and act as vanguard to maintain immune homeostasis in BM. During antigen challenge, memory T cells could be rapidly activated and produce IFN-γ. These HSPCs activated by IFN-γ could suppress further activation of these memory T cells and maintain homeostasis in the BM. This process reflects an interaction between an autoimmune response and hematopoietic system regulation, which has the potential to allow hematopoietic system to influence immune tolerance.
However, liver inflammation was not suppressed by accumulated LSK-like cells in Con A-induced liver injury. On one hand, these LSK-like cells in the liver were mainly MKEP cells, which did not have suppressive function upon T cell-derived IFN-γ stimulation. On the other hand, IFN-γ is thought to be pro-inflammatory in the autoimmune phase of Con A-induced liver injury (41), while their level decreased quickly after 24 h. Moreover, the suppressive activity requires close contact of myeloid progenitors. Thus, other models of autoimmunity with chronic IFN-γ production and transfer experiments may be needed to fully understand the in vivo suppressive function of myeloid progenitors.
In conclusion, we dissect the direct interaction between HSPCs and T cells and our data demonstrate a mechanism of inflammation induced suppressive function of myeloid progenitors, which may participate directly in suppressing T cell-mediated immunopathology.

eThics sTaTeMenT
This study was carried out in accordance with the recommendations of Guide for the Care and Use of Laboratory Animals, USTC Animal Care and Use Committee. The protocol was approved by the USTC Animal Care and Use Committee.
aUThOr cOnTriBUTiOns S-HY, LL, YY, H-DM, and Z-XL: design of the work. S-HY and LL: experimental work, the acquisition, analysis, and interpretation of data for the work; Y-QX, YY, C-YG, and L-HL: experimental