Type I Interferon Activates PD-1 Expression through Activation of the STAT1-IRF2 Pathway in Myeloid Cells

PD-1 (Programmed cell death protein 1) regulates the metabolic reprogramming of myeloid-derived suppressor cells and myeloid cell differentiation, as well as the type I interferon (IFN-I) signaling pathway in myeloid cells in the tumor microenvironment. PD-1, therefore, is a key inhibitory receptor in myeloid cells. However, the regulation of PD-1 expression in myeloid cells is unknown. We report that the expression level of PDCD1, the gene that encodes the PD-1 protein, is positively correlated with the levels of IFNB1 and IFNAR1 in myeloid cells in human colorectal cancer. Treatment of mouse myeloid cell lines with recombinant IFNβ protein elevated PD-1 expression in myeloid cells in vitro. Knocking out IFNAR1, the gene that encodes the IFN-I-specific receptor, diminished the inductive effect of IFNβ on PD-1 expression in myeloid cells in vitro. Treatment of tumor-bearing mice with a lipid nanoparticle-encapsulated IFNβ-encoding plasmid (IFNBCOL01) increased IFNβ expression, resulting in elevated PD-1 expression in tumor-infiltrating myeloid cells. At the molecular level, we determined that IFNβ activates STAT1 (signal transducer and activator of transcription 1) and IRFs (interferon regulatory factors) in myeloid cells. Analysis of the cd279 promoter identified IRF2-binding consensus sequence elements. ChIP (chromatin immunoprecipitation) analysis determined that the pSTAT1 directly binds to the irf2 promoter and that IRF2 directly binds to the cd279 promoter in myeloid cells in vitro and in vivo. In colon cancer patients, the expression levels of STAT1, IRF2 and PDCD1 are positively correlated in tumor-infiltrating myeloid cells. Our findings determine that IFNβ activates PD-1 expression at least in part by an autocrine mechanism via the stimulation of the pSTAT1-IRF2 axis in myeloid cells.

In T cells, PD-1 transcription is regulated by dynamic interplays between epigenetic modifiers and transcription factors at its promoter region, which varies according to the T cell state and the cellular microenvironment [17][18][19]26,27].Demethylation and histone modifications of the Pdcd1 promoter regulatory region following TCR engagement mediate transcription factor association with the Pdcd1 promoter DNA and subsequent Pdcd1 transcription [19,26,28].Multiple transcription factors, including BLIMP-1, Eomes, FOXO1, NFATc1, NF-κB, STAT3, STAT4 and T-Bet have been shown to directly regulate PD-1 expression in T cells at various cellular contexts in virus-infected and tumor-bearing hosts [15,17,[27][28][29][30][31][32].These transcription regulation mechanisms define PD-1 expression's upregulation and T cell function under physiological and pathological conditions, and provide the basis for the development of immune checkpoint inhibitor immunotherapy for human diseases such as cancer [1,11,33].
Although PD-1 is primarily expressed in activated and exhausted T cells, emerging experimental data indicate that PD-1 is also expressed in myeloid cells and plays a critical role in myeloid cell activation and differentiation [34][35][36].Deleting PD-1 only in myeloid cells leads to metabolic reprogramming and the differentiation of myeloid-derived suppressor cells (MDSCs), resulting in increased effector memory T cells and decreased tumor growth despite preserved PD-1 expression in T cells in tumor-bearing mice [36].Furthermore, PD-1 regulates myeloid differentiation and monocytic dendritic cell (moDC) lineage commitment by repressing the expression of HOXA10 and IRF8 to restrain myelocyte differentiation to suppress anti-tumor immunity [34].In addition, tumor cell PD-L1 can bind myeloid cell PD-1 to activate SHP-2.The activated SHP-2 then represses IFN-I expression in myeloid cells to impair the IFN-I signaling pathway.Decreased IFN-I signaling pathway results in decreased Cxcl9/10 expression, leading to impaired T cell tumor recruitment to promote tumor immune evasion [35].It is therefore likely that myeloid cell PD-1 might be a key mechanism of anti-tumor immunity mediated by ICI (immune checkpoint inhibitor) immunotherapy [36].PD-1 expression in myeloid cells therefore is a critical immune suppression mechanism underlying tumor immune evasion.Nevertheless, the mechanism underlying PD-1 expression regulation in myeloid cells is incompletely understood.
We report here that the PDCD1 transcription level exhibits a positive correlation with the IFNB1 and IFNAR1 level in myeloid cells in human colorectal cancer patients.Furthermore, IFNβ upregulates PD-1 expression in mouse myeloid cells both in vitro and in vivo.Mechanistically, we determine that IFNβ induces pSTAT1 binding to the irf2 promoter and IRF2 binding to the cd279 promoter to activate PD-1 transcription.

Materials and Methods
Patient Dataset Analysis.PDCD1, IFNB1, IFNAR1, STAT1 and IRF2 single cell RNAseq datasets of human colorectal cancer and all correlations were extracted from the GEO database (GSE178341) [37].
Mice.BALB/c mice (female, 7-8 weeks old) were purchased from Huafukang, Beijing, China.All studies involving the use of mice were covered by a protocol approved by the Institutional Animal Care and Use Committee of Tianjin University (Approval #TJUE-2021-016.Approval date: 1 March 2021).
Cell Lines.Mouse monocyte-macrophage leukemia cell line RAW264.7 and murine colon tumor cell line CT26 were obtained from Meilunbio, Dalian, China.Bone marrow cells were isolated from BALB/c mice and treated with 2 ng/mL recombinant mouse GM-CSF (Cat# P00195, Solarbio, Beijing, China) to be induced as bone marrow-derived MDSCs.All cell lines were tested every month for mycoplasma, and all cell lines used in this study were mycoplasma-free at the time of the study.
In Vivo Tumor Mouse Model.CT26 cells (2.5 × 10 5 cells/mouse) were harvested and subcutaneously injected to the right flank of BALB/c mice.Ten days after tumor cell inoculation, tumor-bearing mice were randomly separated into two groups.DOTAP-Cholesterol (4 mM LNP, 100 µL/mouse) or IFNBCOL01 (25 µg DNA in 4 mM LNP, 100 µL/mouse) were i.v.injected into mice in two groups, respectively, once every 3 days, totaling four times.
CD11b + Myeloid Cell Isolation.Total tumor tissues were digested in a tissue digestion buffer (collagenase at 1 mg/mL, BS163, Biosharp, Hefei, China; hyaluronidase at 0.1 mg/mL, BS171, Biosharp, Hefei, China and DNase I at 30 U/mL, D8071, Solabrio, Beijing, China) at room temperature for approximately 40 min with agitation.The digests were then grinded and passed through a 70 µm cell strainer (251200, Sorfa, Huzhou, China).The cells were lysed with red cell lysis buffer for 5 min in room temperature and resuspended in PBS with 1% BSA.The CD11b + myeloid cells were then purified from the tumor cell suspension using a Mouse CD11b Selection Kit (Cat# 480110, BioLegend, San Diego, CA, USA) and separated by a magnetic stand.
Western Blotting Analysis.Cells were lysed in total lysis buffer on ice for one hour and measured for protein concentration using a Bradford Kit (Cat# MA0079, Meilunbio, Dalian, China), and cell lysates were separated in SDS-polyacrylamide gels (12% separation gel, 5% stacking gel) and transferred to PVDF membranes (Millipore, Burlington, MA, USA).The membranes were blotted with antibodies and developed by ECL.The antibodies are listed in Table S1.
Flow Cytometry Analysis.For cell surface proteins, cells were harvested, centrifuged and washed in PBS.The cell pellets were then resuspended in 100 µL PBS and stained with fluorescent dye-conjugated antibodies.The suspension was then washed with PBS and resuspended in 300 µL PBS.Stained cells were analyzed using flow cytometry (BD FACS Verse).FlowJo v10 software (BD Biosciences, San Diego, CA, USA) was used to analyze data files.All antibodies were purchased from Biolegend and listed in Table S1.
Chromatin Immunoprecipitation (ChIP) Assay.ChIP assays were carried out according to the protocol of the Chromatin Immunoprecipitation Assay Kit (Cat# 17-295, Merck, NJ, USA).To sum up, cells (1 × 10 6 cells) were harvested and cross-linked using 1% formaldehyde (Cat# F809702, Macklin, Shanghai, China).After a wash, the cells were resuspended in SDS lysis buffer and sonicated before centrifuge.The supernatant fraction was diluted ten fold by the ChIP dilution buffer (save 5% diluent as input).The immunoprecipitating antibody was added into the supernatant and incubated in 4 • C overnight with rotation.The next day, the protein A agarose was added into cell supernatant and incubated in 4 • C for one hour with rotation.The agarose was collected by centrifuge, washed in a low-salt immune complex wash buffer, high-salt immune complex wash buffer, LiCl immune complex wash buffer and TE buffer sequentially, and then eluted by 500 µL elution buffer.The eluants were recovered using phenol/chloroform/isoamyl alcohol (Cat# P59330, Acmec, Shanghai, China).The irf2 and cd279 promoter DNA was detected by quantitative PCR with promoter-DNA-specific primers as listed in Table S2.
Gene Expression Analysis.qPCR and regular PCR analysis were performed as previously described [40].The sequences of primers are listed in Table S2.
Statistical Analysis.All statistical analyses were performed using a two-sided Student t test or a Dunnett's test using the GraphPad Prism program (GraphPad Software, Inc., San Diego, CA, USA).p < 0.05 is considered statistically significant.

PDCD1 Expression has a Positive Correlation with IFN-I Expression in Myeloid Cells in Human Colorectal Cancer
To illustrate the correlation of PDCD1 and IFN-I in human cancer patients, we analyzed the cellular source of PDCD1 in colorectal cancer patients.We downloaded single cell RNAseq datasets (GSE178341) [37] and analyzed PDCD1 and IFN-I expression profiles in human colorectal tumors.Colorectal tumor resident cells were annotated (Figure 1A,B).Cellular subtype analysis revealed that PDCD1 is primarily expressed in T, NK and ILCs (innate lymphocytes) (Figure 1A; combined cells "T, NK and ILCs" are indicated in Figure 1A as "TNKILC"), while IFNB1 is primarily expressed in myeloid cells (Figure 1B) in human colorectal carcinoma.In contrast to IFNB1, IFNA expression is too weak to be detected in human colon carcinoma.Correlation analysis identified there is no correlations between PDCD1 expression level and the IFNB1 expression level in T, NK or ILCs (Figure 1C), but positive correlations do exist in myeloid cells subtypes (Figure 1C).Furthermore, PDCD1 expression is positively correlated with IFNAR1 expression in myeloid cells (Figure 1D).alcohol (Cat# P59330, Acmec, Shanghai, China).The irf2 and cd279 promoter DNA was detected by quantitative PCR with promoter-DNA-specific primers as listed in Table S2.
Gene Expression Analysis.qPCR and regular PCR analysis were performed as previously described [40].The sequences of primers are listed in Table S2.
Statistical Analysis.All statistical analyses were performed using a two-sided Student t test or a Dunnett's test using the GraphPad Prism program (GraphPad Software, Inc., San Diego, CA, USA).p < 0.05 is considered statistically significant.

PDCD1 Expression has a Positive Correlation with IFN-I Expression in Myeloid Cells in Human Colorectal Cancer
To illustrate the correlation of PDCD1 and IFN-I in human cancer patients, we analyzed the cellular source of PDCD1 in colorectal cancer patients.We downloaded single cell RNA-seq datasets (GSE178341) [37] and analyzed PDCD1 and IFN-I expression profiles in human colorectal tumors.Colorectal tumor resident cells were annotated (Figure 1A,B).Cellular subtype analysis revealed that PDCD1 is primarily expressed in T, NK and ILCs (innate lymphocytes) (Figure 1A; combined cells "T, NK and ILCs" are indicated in Figure 1A as "TNKILC"), while IFNΒ1 is primarily expressed in myeloid cells (Figure 1B) in human colorectal carcinoma.In contrast to IFNB1, IFNA expression is too weak to be detected in human colon carcinoma.Correlation analysis identified there is no correlations between PDCD1 expression level and the IFNΒ1 expression level in T, NK or ILCs (Figure 1C), but positive correlations do exist in myeloid cells subtypes (Figure 1C).Furthermore, PDCD1 expression is positively correlated with IFNAR1 expression in myeloid cells (Figure 1D).

Myeloid Cell Intrinsic IFN-I Controls PD-1 Expression In Vitro
To validate the above findings in mouse cells, we made use of mouse macrophage cell line RAW264.7 and bone marrow-derived MDSCs to model human myeloid cells.We first analyzed the basal level of IFNβ in these two cell lines.Although bone marrow-derived MDSCs produced much lower IFNβ levels compared with RAW264.7 cells (Figure 2A), both RAW264.7 cells and bone marrow-derived MDSCs secreted IFNβ.IFN-Is triggered signals through a dimeric IFNAR1/IFNAR2 receptor [41].Additionally, it has been well established that autocrine IFN-I regulates PD-L1 expression through IFNAR1 in MDSCs [42].To determine whether autocrine IFN-I directly regulates PD-1 expression in myeloid cells, Ifnar1 was knocked out in RAW264.7 cells by the CRISPR technique.Flow analysis verified that Ifnar1 had been efficiently knocked out (Figure 2B).The deletion of Ifnar1 not only dramatically reduced the basal expression level of PD-1, but also significantly abrogated IFNβ-induced PD-1 upregulation in RAW264.7 cells (Figure 2C).These findings indicate that IFNβ regulates PD-1 expression in myeloid cells at least in part through an autocrine manner.IRFs (interferon regulatory factors) are transcription factors which are directly stimulated by JAK-STAT signaling pathways and induce different cellular responses by binding to the promoter regions of ISGs (interferon-stimulated genes) including PD-1 [41,43,44].Our above observation that knocking out Ifnar1 leads to downregulation of PD-1 suggests that IFN-I may modulate PD-1 expression through IRF regulation.To test this hypothesis, we analyzed the expression of IRF1-9.As expected, the deletion of Ifnar1 significantly reversed IFNβ-induced IRF elevation in all IRF subtypes, especially IRF1, 2, 5, 7 and 9 (Figure 2D).

Extrinsic IFN-I Controls STATs-IRFs-PD-1 Expression in Myeloid Cells In Vitro
To strengthen the above findings, we took a complimentary approach by culturing RAW264.7 cells in the presence of recombinant IFNα and IFNβ proteins.Consistent with expectations, both IFNα and IFNβ treatment increased PD-1 expression.However, the PD-1 levels after IFNα treatment were slightly increased, but the increase was not statistically significant (Figure 3A).IFNβ significantly elevated PD-1 expression both at the protein level and RNA level (Figure 3B,C).It is well documented that IFNβ binds to IFNAR1 to stimulate ISG transcription through the STATs-IRFs axis.To determine which STATs are activated after IFNβ binding to IFNAR1, we analyzed six known STATs using different treatment times.IFNβ induced STAT1 and STAT3 phosphorylation in a time-dependent manner (Figure 3D).However, phosphorylated STAT2 and STAT4 increased first after 4 h and then decreased after 24 h (Figure 3D).In addition, both phosphorylated STAT5 and STAT6 protein levels were low and were not notably altered by IFNβ (Figure 3D).IFNβ also increased the protein level of non-phosphorylated STAT1 and STAT2 (Figure 3D).At the RNA level, all STATs were induced by IFNβ except for STAT5 and STAT6 (Figure 3E).Furthermore, we analyzed nine known IRFs using different treatment times by regular PCR and qPCR.IFNβ induced all IRFs except for IRF6 (Figure 3F,G), which may be regulated by other transcription repressors which are induced by IFNβ.Among the induced IRFs, IRF7 was stimulated by IFNβ time-dependently, while the other IRFs were stimulated by IFNβ after 4 h, but the inductive effect decreased after 24 h (Figure 3F,G).

IFNβ Controls STATs-IRFs-PD-1 Expression in Myeloid Cells In Vivo
Our above findings indicated that intrinsic IFNAR1 knocked out abolished IFNβinduced PD-1 and IRF upregulation in myeloid cells.On the other hand, extrinsic IFNβ treatment elevated PD-1 expression through the STATs-IRFs axis in myeloid cells in vitro.To determine whether the above findings can be translated to colon tumor growth inhibition and PD-1 expression regulation in vivo, we subcutaneously injected CT26 tumor cells to syngeneic mice to establish an in vivo colon tumor model.The tumor-bearing mice were then intravenously injected with IFNBCOL01 (a lipid nanoparticle-encapsulated IFNβencoding plasmid) [39].IFNBCOL01 effectively delivered cuomIFNβ plasmid DNA to the tumor microenvironment (Figure S1) and cuomIFNβ was highly expressed in total tumor tissue RNA levels (Figure 4B).IFNBCOL01 therapy dramatically repressed the established tumor growth in mice (Figure 4A) and upregulated PD-1 expression in the tumor microenvironment (Figure 4C).Our previous findings demonstrated that the overexpression of IFNβ in a tumor microenvironment leads to the enhanced expression of Gzmb (granzyme B) and MHC I (major histocompatibility complex I), resulting in effective tumor suppression in vivo [39].This will be further discussed in the Discussion section.The analysis of IRFs and STATs in total tumor tissues showed that IFNBCOL01 therapy led to elevated STAT1, 2 and 6 levels (Figure 4D) and expression of all IRFs (Figure 4E).and then decreased after 24 h (Figure 3D).In addition, both phosphorylated STAT5 and STAT6 protein levels were low and were not notably altered by IFNβ (Figure 3D).IFNβ also increased the protein level of non-phosphorylated STAT1 and STAT2 (Figure 3D).At the RNA level, all STATs were induced by IFNβ except for STAT5 and STAT6 (Figure 3E).Furthermore, we analyzed nine known IRFs using different treatment times by regular PCR and qPCR.IFNβ induced all IRFs except for IRF6 (Figure 3F,G), which may be regulated by other transcription repressors which are induced by IFNβ.Among the induced IRFs, IRF7 was stimulated by IFNβ time-dependently, while the other IRFs were stimulated by IFNβ after 4 h, but the inductive effect decreased after 24 h (Figure 3F,G).To demonstrate the effect of IFNBCOL01 in tumor-infiltrating myeloid cells, we isolated CD11b + cells from total tumor tissues by CD11b selection beads.The isolation efficiency is shown in Figure 4F.Consistent with the results in total tumor tissues, cuomIFNβ was highly expressed in the RNA level in IFNBCOL01-treated tumor-infiltrating CD11b + cells (Figure 4G).Accordingly, IFNBCOL01 augmented PD-1 (Figure 4H) and IRFs expression (except IRF4) (Figure 4I) in CD11b + cells.Immune cell profiles analysis revealed that IFNBCOL01 increased CXCL10 and MHC-I expression in CD11b + cells (Figure S2), which may contribute to tumor growth inhibition.

IFNβ Elevates PD-1 Expression through pSTAT1-IRF2 Axis in Myeloid Cells
To reveal the specific STATs-IRFs axis used by IFNβ to modulate PD-1 expression in myeloid cells, we exploited PROMO software to make transcription factor predictions at the mouse cd279 promoter region.As shown in Figure 5A, among the nine known IRFs, only IRF1 and IRF2 were predicted to bind at the cd279 (the gene encodes the murine PD-1) promoter region (−2000 to +2000 bp).IRF2 is known to antagonize IRF1 by competing for binding to the same promoter elements of IFN-I-and IFN-II-inducible genes [45] and by inhibiting the nuclear translocation of IRF1 [46].Furthermore, IRF2 is reported to be associated with T cell exhaustion and function [47].Therefore, we analyzed the protein level of IRF2 at different times.As expected, IRF2 was induced by IFNβ time-dependently as IFNγ (Figure S3).
To validate whether IRF2binds to the mouse cd279 promoter, we treated RAW264.7 cells with recombinant IFNβ and treated CT26 tumor-bearing mice with IFNBCOL01, then performed ChIP analysis using an IRF2-specific antibody and PCR primers that covered the promoters approximately from −2000 to +2000 relative to the transcription start site of the mouse cd279 (Figure 5B).Analysis of the mouse cd279 promoter regions by ChIP illustrated that after IFNβ and IFNBCOL01 treatment, enrichment of irf2 increased in the promoter regions of the cd279 gene in −2000 to +2000 regions in RAW264.7 cells (Figure 5C) and in −2000 to 0, +1000 to +2000 regions in tumor-infiltrating CD11b + cells (Figure 5D).To further determine which specific STAT binds to the irf2 promoter region, we used PROMO software to perform transcription factor prediction at the mouse irf2 promoter region.As shown in Figure 6A, among six known STATs, only STAT1 was predicted to bind at the irf2 promoter region (−2000 to +2000 bp).Consistent with the prediction, IFNAR1 deficiency dramatically attenuated the induction of pSTAT1 by IFNβ (Figure S4).To validate whether pSTAT1 binds to the mouse irf2 promoter, we treated RAW264.7 cells with recombinant IFNβ and treated CT26 tumor-bearing mice with IFNBCOL01, then performed ChIP analysis using a pSTAT1 specific antibody and PCR primers that covered the promoters approximately from −2000 to +2000 relative to the transcription start site of the mouse irf2 (Figure 6B).Analyses of mouse irf2 promoter regions by ChIP demonstrated that after IFNβ and IFNBCOL01 treatment, enrichment of pSTAT1 increased in the promoter regions of the irf2 gene in −2000 to +2000 regions in RAW264.7 cells (Figure 6C), and in −2000 to 0, +1000 to +2000 regions in tumor-infiltrating CD11b + cells (Figure 6D).The immunoprecipitated chromatin fragments were then analyzed using qPCR, as in (C).For ChIP analysis, shown is one representative result of two independent experiments, and the error bar is the mean of the triplicates for one experiment.ns: no significance.

IFNβ Regulates PD-1 Expression through pSTAT1-IRF2 Axis in Bone Marrow-Derived MDSCs
Our above results showed that IFNβ elevated PD-1 expression through the pSTAT1-IRF2 axis in myeloid cells both in vitro and in vivo.To further validate this finding, we made use of another myeloid cell line: bone marrow-derived MDSCs.After GM-CSF induction, around 90% of the bone marrow cells were CD11b + (Figure 7A).These bone marrow-derived MDSCs expressed IFNAR1 (Figure 7B), which suggested that they have the ability to respond to IFN-I through IFNAR1.Just like with the RAW264.7 cells, we cultured bone marrow-derived MDSCs in the presence of recombinant IFNα and IFNβ protein.Only IFNβ treatment augmented PD-1 expression, while IFNα has no inductive effect (Figure 7C).IFNβ upregulated PD-1 expression both at the protein level and RNA level (Figure 7D,E).IFNβ also induced pSTAT1 and IRF2 expression in bone marrow-derived MDSCs (Figure 7F,G).
To validate the binding of pSTAT1 to the irf2 promoter and the binding of IRF2 to the cd279 promoter in bone marrow-derived MDSCs, we treated bone marrow-derived MDSCs with recombinant IFNβ, then performed ChIP analysis.PCR primers that covered The immunoprecipitated chromatin fragments were then analyzed using qPCR with primers shown in (B).(D) Chromatin was prepared from IFNBCOL01-treated CD11b + tumor-infiltrating myeloid cells shown in Figure 4F and analyzed using ChIP with a pSTAT1-specific antibody, as in (C).The immunoprecipitated chromatin fragments were then analyzed using qPCR, as in (C).For ChIP analysis, shown is one representative result of two independent experiments, and the error bar is the mean of the triplicates for one experiment.ns: no significance.

IFNβ Regulates PD-1 Expression through pSTAT1-IRF2 Axis in Bone Marrow-Derived MDSCs
Our above results showed that IFNβ elevated PD-1 expression through the pSTAT1-IRF2 axis in myeloid cells both in vitro and in vivo.To further validate this finding, we made use of another myeloid cell line: bone marrow-derived MDSCs.After GM-CSF induction, around 90% of the bone marrow cells were CD11b + (Figure 7A).These bone marrow-derived MDSCs expressed IFNAR1 (Figure 7B), which suggested that they have the ability to respond to IFN-I through IFNAR1.Just like with the RAW264.7 cells, we cultured bone marrow-derived MDSCs in the presence of recombinant IFNα and IFNβ protein.
the promoters from −2000 to +2000 relative to the transcription start sites of mouse cd279 and irf2 are shown in Figure 5B and Figure 6B.The analysis of mouse cd279 promoter regions by ChIP demonstrated that after IFNβ treatment, enrichment of IRF2 increased in the promoter region of the cd279 gene in −2000 to −1000 region (Figure 7H), and enrichment of pSTAT1 was elevated in the promoter region of the irf2 gene in the 0 to +1000 region (Figure 7I), respectively, in bone marrow-derived MDSCs.5C, and a pSTAT1-specific antibody, (I) as in Figure 6C.The immunoprecipitated chromatin fragments were then analyzed using qPCR with primers shown in Figures 5B and 6B, respectively.For ChIP analysis, shown is one representative result of two independent experiments, and the error bar is the mean of the triplicates for one experiment.
To validate the binding of pSTAT1 to the irf2 promoter and the binding of IRF2 to the cd279 promoter in bone marrow-derived MDSCs, we treated bone marrow-derived MDSCs with recombinant IFNβ, then performed ChIP analysis.PCR primers that covered the promoters from −2000 to +2000 relative to the transcription start sites of mouse cd279 and irf2 are shown in Figures 5B and 6B.The analysis of mouse cd279 promoter regions by ChIP demonstrated that after IFNβ treatment, enrichment of IRF2 increased in the promoter region of the cd279 gene in −2000 to −1000 region (Figure 7H), and enrichment of pSTAT1 was elevated in the promoter region of the irf2 gene in the 0 to +1000 region (Figure 7I), respectively, in bone marrow-derived MDSCs.

IRF2 Expression Exhibits Positive Correlation with STAT1/PDCD1 Expression in Myeloid Cells in Human Colon Cancer
Our above findings indicated IFNβ-regulated PD-1 expression through the pSTAT1-IRF2 axis in three different types of myeloid cells (RAW264.7 cells, tumor-infiltrating CD11b + cells and bone marrow-derived MDSCs).To illustrate the human cancer relevance of our above findings, we then analyzed the cellular source of IRF2 in colon cancer patients.We extracted scRNA-seq datasets (GSE178341) [37] and analyzed IRF2 expression profiles in human colon tumors.Colon tumor-resident cells were annotated (Figure S5A).Cellular subtype analysis indicated that IRF2 is primarily expressed in T, NK, B, stroma and myeloid cells in human colon carcinoma (Figure S5A).
Correlation analysis revealed positive correlations between IRF2 and STAT1 expression levels, IRF2 and PDCD1 expression levels and PDCD1 and STAT1 expression levels (Figure S5B-D).Moreover, IRF2, STAT1 and PDCD1 expression levels are positively correlated with each other in different subtypes of myeloid cells (Figure S5E).

Discussion
PD-1, associated with the expression of additional repressive receptors such as Tim-3, LAG-3 or TIGIT [48][49][50], is a co-inhibitory receptor of the CD28 family and functions as a negative regulator to provide crucial inhibitory signals that maintain immune homeostasis, control inflammation resolution, regulate host immune defense and mediate tolerance.PD-1 therefore functions as an immune checkpoint to prevent autoimmunity [1,5].Nevertheless, under pathological conditions such as chronic viral infection and cancer, sustained PD-1 expression leads to chronic T cell activation to induce T cells into dysfunction and exhaustion.In cancer patients and tumor-bearing mice, tumor cells and tumor cell-induced myeloid cells express high levels of PD-L1 in response to IFNγ secreted by tumor-activated T cells.PD-L1 binds to PD-1 on T cells to induce T cell dysfunction and exhaustion to promote tumor immune evasion and resultant progression [51].Tumor cells hijack the PD-1/PD-L1 immune checkpoint to dampen the anti-tumor immune response.T cellexpressed PD-1 therefore serves as a crucial target for tumor cell counterattack.The expression and regulation of PD-1 in T cells have been extensively studied and multiple layers of transcriptional regulatory mechanisms have been defined in the regulation of PD-1 expression in T cells, including different epigenetic and genetic programs adapted to either transient [52] or chronic antigenic stimulations [26,53,54].However, it has been recently observed that myeloid cell PD-1 plays an essential role in the suppression of anti-tumor immunity [34][35][36].Knocking out myeloid cell PD-1 leads to increased tumor reactive T cells and enhanced tumor growth control despite preserved PD-1 expression in the T cells of the tumor-bearing mice [36].The revelation of this key PD-1 function in myeloid cells in tumor immune suppression and evasion suggests a critical role of myeloid cell PD-1 expression regulation in tumor immune evasion and the mechanism by which a tumor responds to ICI immunotherapy.In this study, we determined that IFNβ upregulates PD-1 expression and knocks out IFNAR1-suppressed PD-1 expression in myeloid cells, revealing that PD-1 expression is regulated by IFN-I/IFNβ at least in part through an autocrine mechanism in myeloid cells in vitro and in tumor-bearing mice in vivo.We also observed that the PD-1 expression level is also positively correlated with the expression levels of IFNB1 and IFNAR1 in the myeloid cells of human colon cancer patients.Therefore, PD-1 expression regulation by IFNβ is a physio-pathologically relevant mechanism.
PD-1 is an inhibitory receptor that inhibits myeloid cell activation and differentiation to impair host anti-tumor immunity [34][35][36][55][56][57][58].Although IFN-I has been shown to promote tumor stemness under certain cellular contexts [59,60], IFN-I is the main defense line against pathogen infection in the host immune surveillance system [61] and is also the master regulator of anti-tumor immunity [62][63][64][65][66].It is also known that the engagement of myeloid cell PD-1 by tumor PD-L1 suppresses IFN-I expression and signaling [35].The upregulation of PD-1 by IFNβ therefore makes no sense in terms of the opposite functions of PD-1 and IFNβ under physiological conditions.However, this phenomenon is not uncommon as it is also known that the expression of PD-L1, the immune-suppressive ligand for PD-1, is also upregulated by IFNγ [67], an essential cytokine for T cell function in host cancer immune surveillance [68].This phenomenon is believed to be an adaptive immune resistance process where the cancer changes its phenotype in response to activated T cell attack and thereby evading it [69,70].We therefore propose that the upregulation of PD-1 by IFNβ in myeloid cells is an indicator of myeloid cell activation and another adaptive immune resistance process in cancer.Myeloid cells respond to cancer by activating IFN-I/IFNβ expression to regulate an anti-tumor immune response, but at the same time, tumor cells may use PD-L1 to activate PD-1-intrinsic inhibitory signaling in myeloid cells to shut down IFN-I expression and therefore function to evade cancer.IFNBCOL01 therapy dramatically improved the expression of Gzmb and MHC I in the tumor microenvironment [39].This tumor-suppression effect may overcome the tumor promotion effect caused by the shut-down of IFN-I signaling in myeloid cells, leading to overall tumor inhibition.
IFN-I transduces signals through a dimeric IFNAR1/IFNAR2 receptor that activates the kinases Jak1 and Tyk2 to initiate STAT phosphorylation to activate the expression of IFN-I-stimulated genes (ISGs) [41].In this study, we determined that IFNβ activates STAT1, and activated STAT1 directly binds to the irf2 promoter to activate IRF2 expression.Furthermore, IRF2 directly binds to the cd279 promoter to activate PD-1 expression in myeloid cells.A positive correlation between STAT1, IRF2 and PD-1 is observed in the myeloid cells in human colon cancer patients.It has been well documented that IRF2 expression by CD8 + tumor-infiltrating lymphocytes drives T cell exhaustion, and irf2 −/− CD8 + T cells expressed lower levels of inhibitory receptors including PD-1 as compared with WT CD8 + T cells [47,71].However, how the IFNβ/STAT1/IRF2 pathway directly regulates PD-1 expression in myeloid cells remains to be determined.Nevertheless, in this research, we determined that the IFNβ/STAT1/IRF2 axis regulates PD-1 expression in myeloid cells.

Conclusions
PD-1 is a co-repressive receptor of T cells and a key target in human cancer immunotherapy.Emerging experiment data revealed a key role of the intrinsic PD-1 pathway in myeloid cells.The regulation of PD-1 expression in T cells has been extensively studied, but the regulation of PD-1 expression in myeloid cells is incompletely understood.We determined that the expression level of PDCD1 is positively correlated with the levels of IFNB1 and IFNAR1 in myeloid cells in human colorectal cancer.Recombinant IFNβ protein treatment elevated PD-1 expression in mouse myeloid cells in vitro.Furthermore, deletion of IFNAR1 reversed the inductive effect of IFNβ on PD-1 expression in myeloid cells in vitro.The forced expression of IFNβ in tumor-bearing mice led to elevated PD-1 expression in tumor-infiltrating myeloid cells in vivo.At the molecular level, we determined that IFNβ activates STAT1.The activated STAT1 (pSTAT1) binds to the irf2 promoter region to elevate IRF2 expression.IRF2 then binds to the Pdcd1 promoter region to increase PD-1 expression.In human colon cancer patients, the expression levels of STAT1, IRF2 and PDCD1 are positively correlated in tumor-infiltrating myeloid cells.Our findings thus determined that IFNβ activates PD-1 expression via the pSTAT1-IRF2 pathway in myeloid

Figure 1 .
Figure 1.PDCD1 and IFNΒ1 expression profiles in human colorectal cancer patients.(A) Shown are the UMAP of major cell subpopulations (right panel) and the PDCD1 expression level (left panel) in the indicated cell subpopulations in human colorectal cancer.The red arrows show the PDCD1 expression in myeloid cells.(B) Shown are the IFNB1 expression level (left panel) in the

Figure 1 .
Figure 1.PDCD1 and IFNB1 expression profiles in human colorectal cancer patients.(A) Shown are the UMAP of major cell subpopulations (right panel) and the PDCD1 expression level (left panel) in the indicated cell subpopulations in human colorectal cancer.The red arrows show the PDCD1 expression in myeloid cells.(B) Shown are the IFNB1 expression level (left panel) in the indicated cell subpopulations and the UMAP of major cell subpopulations (right panel) in human colorectal cancer.

19 Figure 2 .Figure 2 .
Figure 2. Myeloid cell intrinsic IFN-I controls PD-1 expression in vitro.(A) Total RNA was prepared from RAW264.7 and bone marrow-derived MDSCs, and analyzed for basal IFNβ mRNA expression by RT-PCR with β-actin as an internal control.(B) RAW264.7 scramble and RAW264.7 IF-NAR1 KO cells were cultured in vitro for 24 h.Cells were stained with anti-IFNAR1 mAb and analyzed by flow cytometry with IgG as a negative control.The MFI (mean fluorescence intensity) of IFNAR1 was quantified and presented in the right panel.Column: mean; Bar: SEM.(C) RAW264.7 scramble and RAW264.7 IFNAR1 KO cells were treated with IFNβ (100 ng/mL) for 24 h.Cells were stained with anti-PD-1 mAb and analyzed using flow cytometry.Data were analyzed with Dunnett's test.(D) Total RNA was extracted from RAW264.7 scramble and RAW264.7 IFNAR1 KO cellsFigure 2. Myeloid cell intrinsic IFN-I controls PD-1 expression in vitro.(A) Total RNA was prepared from RAW264.7 and bone marrow-derived MDSCs, and analyzed for basal IFNβ mRNA

Figure 4 .
Figure 4. Overexpression of IFNβ in the tumor microenvironment activates STATs and IRFs to stimulate PD-1 expression in myeloid cells in vivo.(A) CT26 cells (2.5 × 10 5 cells/mouse) were subcutaneously injected in mice to establish in vivo tumor models.The tumor-bearing mice were then intravenously treated with LNP (n = 5) and IFNBCOL01 (n = 5) 10 days after tumor inoculation,

Figure 5 .
Figure 5. IRF2 binds to cd279 promoter region in myeloid cells.(A) The mouse cd279 promoter DNA sequence (−2000 to +2000 relative to the cd279 transcription start site) was exported from the mouse genomic database and analyzed for potential IRF-binding elements using the PROMO database.Shown is the mouse cd279 promoter structure with predicted binding locations of IRF1 and IRF2 to the cd279 promoter region.(B) Four pairs of primers spanning from −2000 to +2000 relative to the mouse cd279 promoter region are shown.The red arrow represents the transcription start site.(C)RAW264.7 cells were treated with IFNβ for 24 h.Chromatin was then prepared and analyzed using ChIP with an IRF2-specific antibody.The immunoprecipitated chromatin fragments were then analyzed by qPCR with primers shown in (B).(D) Chromatin was prepared from IFNΒCOL01treated CD11b + tumor-infiltrating myeloid cells shown in Figure4Fand analyzed using ChIP with a IRF2-specific antibody as in (C).The immunoprecipitated chromatin fragments were then analyzed with qPCR as in (C).For ChIP analysis, shown is one representative result of two independent experiments, and the error bar is the mean of the triplicates for one experiment.ns: no significance.

Figure 5 .
Figure 5. IRF2 binds to cd279 promoter region in myeloid cells.(A) The mouse cd279 promoter DNA sequence (−2000 to +2000 relative to the cd279 transcription start site) was exported from the mouse genomic database and analyzed for potential IRF-binding elements using the PROMO database.Shown is the mouse cd279 promoter structure with predicted binding locations of IRF1 and IRF2 to the cd279 promoter region.(B) Four pairs of primers spanning from −2000 to +2000 relative to the mouse cd279 promoter region are shown.The red arrow represents the transcription start site.(C) RAW264.7 cells were treated with IFNβ for 24 h.Chromatin was then prepared and analyzed using ChIP with an IRF2-specific antibody.The immunoprecipitated chromatin fragments were then analyzed by qPCR with primers shown in (B).(D) Chromatin was prepared from IFNBCOL01-treated CD11b + tumor-infiltrating myeloid cells shown in Figure4Fand analyzed using ChIP with a IRF2-specific antibody as in (C).The immunoprecipitated chromatin fragments were then analyzed with qPCR as in (C).For ChIP analysis, shown is one representative result of two independent experiments, and the error bar is the mean of the triplicates for one experiment.ns: no significance.

Figure 6 .
Figure 6.pSTAT1 binds to the irf2 promoter region in myeloid cells.(A) The mouse irf2 promoter DNA sequence (=−2000 to +2000 relative to irf2 transcription start site) was exported from the mouse genomic database and analyzed for potential STAT-binding elements using the PROMO database.Shown is the irf2 structure with predicted binding locations of STAT1β to the irf2 promoter region.(B) Four pairs of primers spanning from −2000 to +2000 relative to the irf2 promoter region are shown.The red arrow represents the transcription start site.(C) Chromatin was prepared from RAW264.7 cells treated with IFNβ for 24 h and analyzed using ChIP with a pSTAT1-specific antibody.The immunoprecipitated chromatin fragments were then analyzed using qPCR with primers shown in (B).(D) Chromatin was prepared from IFNΒCOL01-treated CD11b + tumor-infiltrating myeloid cells shown in Figure 4F and analyzed using ChIP with a pSTAT1-specific antibody, as in (C).The immunoprecipitated chromatin fragments were then analyzed using qPCR, as in (C).For ChIP analysis, shown is one representative result of two independent experiments, and the error bar is the mean of the triplicates for one experiment.ns: no significance.

Figure 6 .
Figure 6.pSTAT1 binds to the irf2 promoter region in myeloid cells.(A) The mouse irf2 promoter DNA sequence (=−2000 to +2000 relative to irf2 transcription start site) was exported from the mouse genomic database and analyzed for potential STAT-binding elements using the PROMO database.Shown is the irf2 structure with predicted binding locations of STAT1β to the irf2 promoter region.(B) Four pairs of primers spanning from −2000 to +2000 relative to the irf2 promoter region are shown.The red arrow represents the transcription start site.(C) Chromatin was prepared from RAW264.7 cells treated with IFNβ for 24 h and analyzed using ChIP with a pSTAT1-specific antibody.The immunoprecipitated chromatin fragments were then analyzed using qPCR with primers shown in (B).(D) Chromatin was prepared from IFNBCOL01-treated CD11b + tumor-infiltrating myeloid cells shown in Figure4Fand analyzed using ChIP with a pSTAT1-specific antibody, as in (C).The immunoprecipitated chromatin fragments were then analyzed using qPCR, as in (C).For ChIP analysis, shown is one representative result of two independent experiments, and the error bar is the mean of the triplicates for one experiment.ns: no significance.

Figure 7 .Figure 7 .
Figure 7. IFNβ-STAT1-IRF2 axis upregulates PD-1 expression in bone marrow-derived MDSCs.(A) Bone marrow cells were prepared from BALB/c WT mice and treated with GM-CSF (2 ng/mL) for 7 days.Cells were then stained with anti-CD11b mAb and analyzed using flow cytometry.The percentage of CD11b + myeloid cells after GM-CSF treatment (right pane) is shown.(B) Bone marrow-derived MDSCs were cultured for 24 h and then stained with anti-IFNAR1 mAb and analyzed by flow cytometry.(C) Bone marrow-derived MDSCs were treated with recombinant IFNα, IFNβ and IFNα+IFNβ, respectively, for 24 h.LPS was used as the positive control.Cells were then stained with anti-PD-1 mAb and analyzed by flow cytometry.ns: no significance.(D) The bone marrowderived MDSCs were treated with IFNβ alone for 24 h, and then stained with anti-PD-1 mAb and analyzed by flow cytometry.(E) Total RNA was isolated from bone marrow-derived MDSCs treated with IFNβ for 4 h and 24 h, respectively, and analyzed for the mRNA level of PD-1 using qPCR with β-actin as the internal control.(F) Bone marrow-derived MDSCs were treated with IFNβ for 4 h and 24 h, respectively, and lysed for total protein, and then analyzed by Western blotting for STAT1 and pSTAT1.β-actin is used as the normalization control.(G) Total RNA was prepared from bone Figure 7. IFNβ-STAT1-IRF2 axis upregulates PD-1 expression in bone marrow-derived MDSCs.(A) Bone marrow cells were prepared from BALB/c WT mice and treated with GM-CSF (2 ng/mL) for 7 days.Cells were then stained with anti-CD11b mAb and analyzed using flow cytometry.The percentage of CD11b + myeloid cells after GM-CSF treatment (right pane) is shown.(B) Bone marrowderived MDSCs were cultured for 24 h and then stained with anti-IFNAR1 mAb and analyzed by flow cytometry.(C) Bone marrow-derived MDSCs were treated with recombinant IFNα, IFNβ and IFNα+IFNβ, respectively, for 24 h.LPS was used as the positive control.Cells were then stained with anti-PD-1 mAb and analyzed by flow cytometry.ns: no significance.(D) The bone marrow-derived MDSCs were treated with IFNβ alone for 24 h, and then stained with anti-PD-1 mAb and analyzed by flow cytometry.(E) Total RNA was isolated from bone marrow-derived MDSCs treated withIFNβ for 4 h and 24 h, respectively, and analyzed for the mRNA level of PD-1 using qPCR with β-actin as the internal control.(F) Bone marrow-derived MDSCs were treated with IFNβ for 4 h and 24 h, respectively, and lysed for total protein, and then analyzed by Western blotting for STAT1 and pSTAT1.β-actin is used as the normalization control.(G) Total RNA was prepared from bone marrow-derived MDSCs treated with IFNβ for 4 h and 24 h, respectively, and analyzed for the expression of IRF2 using qPCR.(H,I) Chromatin was isolated from bone marrow-derived MDSCs treated with IFNβ for 24 h and analyzed by ChIP with an IRF2-specific antibody, (H) as in Figure5C, and a pSTAT1-specific antibody, (I) as in Figure6C.The immunoprecipitated chromatin fragments were then analyzed using qPCR with primers shown in Figures5B and 6B, respectively.For ChIP analysis, shown is one representative result of two independent experiments, and the error bar is the mean of the triplicates for one experiment.
expression by RT-PCR with β-actin as an internal control.(B) RAW264.7 scramble and RAW264.7 IFNAR1 KO cells were cultured in vitro for 24 h.Cells were stained with anti-IFNAR1 mAb and analyzed by flow cytometry with IgG as a negative control.The MFI (mean fluorescence intensity) of IFNAR1 was quantified and presented in the right panel.Column: mean; Bar: SEM.(C) RAW264.7 scramble and RAW264.7 IFNAR1 KO cells were treated with IFNβ (100 ng/mL) for 24 h.Cells were stained with anti-PD-1 mAb and analyzed using flow cytometry.Data were analyzed with Dunnett's test.(D) Total RNA was extracted from RAW264.7 scramble and RAW264.7 IFNAR1 KO cells treated with IFNβ for 4 h and 24 h, respectively, and analyzed using qPCR for IRF1-9 expression.Shown is one representative result of three independent experiments, and the error bar is the mean of the triplicates for one experiment.Data were analyzed with the Student t test.ns: no significance.