Pancreatic cell lines MiaPaca2, BxPC-3, and PANC-1 were cultured in DMEM (Gibco) with 10% FBS, 10 mM L-glutamine, and antibiotics. Murine MT5 (Kras^LSL−G12D, Trp53^LSL−R270H, and Pdx1-cre) pancreatic cells were a kind gift from Dr Tuveson (Cold Spring Harbor Laboratory, Cold Spring Harbor, NY) and cultured in RPMI (Gibco) with 10% FBS, 10 mM L-glutamine, and antibiotics. Pancreatic cancer patient tissue microarray was purchased from US Biomax (Derwood, MD). Murine antibodies to CD200 (Clone OX-90), PD-1 (Clone RMP1-14), or isotype control (Rat IgG2a, κ, Clones 2A3) were purchased from BioXcell (West Lebanon, NH) for in vivo studies.

Pancreatic tumors from humans undergoing surgical resection at the James Cancer Hospital and Solove Research Institute (Columbus, OH) were obtained under an institutional review board-approved protocol following informed consent. Tissue was dissected with a scalpel into 0.5–1 mm³ pieces, then plated in six-well 10 cm² uncoated culture wells in DMEM with 10% FBS and antibiotics and incubated at 37°C. Stellate cells typically grew out of the tissue in 2–3 weeks, and were characterized by morphology and histological analysis of alpha-smooth muscle actin (α-SMA+) staining. Stellate cells were maintained in culture with fresh media added twice weekly for three passages and then RNA was collected by Trizol extraction. RNA was analyzed using the nCounter PanCancer Immune Profiling Panel (Nanostring Technologies, Seattle WA).

Paraffin embedded pancreatic tumor tissue from surgical resection patients and KPC-Brca2 mice were cut at 5 µM sections on mounted Superfrost Plus slides. Slides were dewaxed and rehydrated, antigen retrieval using a 10 mM sodium citrate buffer and permeabilized using a 0.4% Triton X-100 solution. Non-specific binding was blocked using 5% serum. Tissue was stained with primary antibodies: α-SMA (Clone 1A4; Sigma) and CD200 (Abcam; Cambridge, UK); and secondary antibody Alexa-Fluor 647 donkey antirabbit (Invitrogen). Slides were stained with DAPI to visualize nuclei and mounted with a coverslip. Fluorescent images were analyzed at 20× on a Zeiss Axiovert A1 Observer microscope and analyzed using Image J software.

Immunophenotypic analyses of PMBCs from patients, mouse splenoctyes, and single cell suspensions from mouse tumors were assessed by flow cytometry. Cells were incubated on ice for 30 min with fluorochrome-conjugated antibodies, washed, and fixed in PBS containing 1% formalin for flow cytometric analysis on an Attune (Life Technologies) or Fortessa (BD Biosciences) flow cytometer. Antibodies used to stain for human antigens: CD33 (Clone: P67.6; Biolegend), HLA-DR (Clone: L243; Biolegend), CD11b (Clone: ICRF44; Biolegend), CD15 (Clone: HI98; Biolegend), CD14 (Clone: HCD14; Biolegend), CD80 (Clone: 2D10; Biolegend), CD163 (Clone: RM3/1; Biolegend), CD206 (Clone: I5-2; Biolegend), CD56 (Clone: HCD56; Biolegend), CD3 (Clone: OKT3; Biolegend), CD4 (Clone: OKT4; Biolegend), CD8 (SK1; Biolegend), CD45RA (HI100; Biolegend), CCR7 (Clone: 3D12; BB Biosciences), CD25 (Clone: REA570; Miltenyi Biotec), GITR (Clone: 108–17; Biolegend), CD200R (Clone: OX-108; Biolegend), and IgG1 control (Clone: MOPC-21; Biolegend). Antibodies used to stain for mouse antigens: CD45 (Clone: 30-F11; Biolegend); Ly6-g (Clone: 1A8; Biolegend), Ly6-C (Clone: AL-21; BD Biosciences), CD11b (Clone: M1/70; BD Biosciences), CD4 (Clone: GK1.5; Biolegend), CD8 (Clone: 53–6.7; Biolegend), NK1.1 (Clone: PK136; Biolegend), and F4/80 (Clone: BM8; Biolegend).

KPC-Brca2 mice were generated by interbreeding Brca2^flox2/flox2 ; Kras^LSL−G12D/+ with Brca2^flox2/flox2 ; Trp53^{LSL−R270H/+} ; Pdx1-cre animals.29 The mouse strains p53^LSL−R270H (strain number 01XM3), Kras^LSL−G12D (strain number 01XJ6), and Pdx1-cre (strain number 01XL5) were acquired from the National Cancer Institute (NCI) Frederick Mouse Repository. All transgenic mice generated in this study were maintained on a mixed 129/B6 genetic background.

In vivo treatments were completed as previously described.21 Briefly, KPC-Brca2 mice (5 weeks of age) were treated with isotype control or anti-CD200 Ab at a dose of 200 µg/mouse, three times each week (Monday, Wednesday, and Friday). Following 2 weeks of treatment, animals were euthanized via CO₂ asphyxiation, followed by cardiac puncture. Splenocytes and tumor tissue were collected for further analysis. Pathology was assessed in H&E stained slides to determine the differentiation state of tissue as pancreatic intraepithelial neoplasia (PanIN)−1, PanIN-2, PanIN-3, or PDAC. For studies using MT5 tumor cells, 1×10⁶ cells were injected subcutaneously in the flank of C57BL/6 mice and injected intraperitoneally three times each week with 200 µg/mouse of isotype, anti-CD200 and/or anti-PD-1 Ab (BioXCell) treatment starting once tumors reached 50–100 mm³ volume.

Cryopreserved whole PBMC from PDAC patients (n=4) were thawed, washed, and counted. Cell viability was between 83% and 92%. Single cells were isolated using the Chromium Next GEM 5′ gene expression kit, targeting recovery of 4000 cells per patient. Libraries were constructed and sequenced according to the manufacturer’s instructions (Illumina NovaSeq, Nationwide Children’s Hospital Institute for Genomic Medicine/Genomic Services Laboratory). Sequence data were processed using Cell Ranger V.3.1.0. Cell recovery was 4132±1486 cells per sample. After aggregation, one sample showed significant batch effect and was removed from the analysis. Single-cell gene expression analysis was performed using Monocle V.3.30 Dimensionality reduction was performed using Uniform Manifold Approximation and Projection (UMAP) which is better at preserving local and global structural differences in high-dimensional data compared with tSNE.31 Cell clusters were defined using the Leiden method and cluster top markers were identified by logistic regression.32 Maximum expression of CD200R, DOK1, and DOK2 was plotted by taking the maximum of the scaled, size-factor normalized expression values for these genes in each cell. The genes that were overexpressed by MDSC were analyzed by the Reactome Pathway Profile software to determine potential pathways that may be active in CD200R expressing cells.

PBMC were isolated from source leukocytes of healthy donors (Versiti, Milwaukee, WI) and patients with pancreatic cancer and chronic pancreatitis (CP, from a prospective Institutional Review Board-approved study) via density gradient centrifugation using Ficoll-Paque (Amersham, Pharmacia Biotech, Bjorkgatan, Sweden) as described.33–35 PBMC from healthy donors were cultured in 10% FBS, 10 mM L-glutamine, and 100 µg/mL penicillin/streptomycin in RPMI 1640 (Gibco). To generate functional MDSC, PBMC were cultured with 10 ng/mL of IL-6 and GM-CSF (Peprotech, Rocky Hill, NJ) for 7 days as previously described by our group and others.33 34 PBMC were cocultured with increasing concentration of human recombinant CD200 protein (Sino Biologicals). PBMC were stained for surface markers (HLA-DR, CD11b, and CD33) to confirm percentage of MDSC. To test MDSC suppressive function, MDSC were generated by differentiating PBMC with 10 ng/mL of IL-6 and GM-CSF for 5 days and were then stimulated with vehicle or rhC200 protein for 48 hours. MDSC were cocultured with Carboxyfluorescein succinimidyl ester (CFSE) labeled (Thermo Fisher) autologous negatively selected T cells via RosetteSep (STEMCELL) and activated with anti-CD3/CD28 beads (Gibco) for 4 days. T-cell proliferation was measured on an LSRII flow cytometer.

Cell lysates were analyzed for protein expression by immunoblot analysis with antibodies against CD200 (Abcam), STAT3 (Catalog 4904; Cell Signaling), pSTAT3 (Catalog 9145; Tyr705; Cell Signaling), and β-actin (Clone: BA3R; Thermo Fisher). Following incubation with appropriate conjugated secondary antibodies, immune complexes were detected using an LI-COR CLx imager (LI-COR, Lincoln, NE).

TRIzol reagent (Life Technologies) was used to extract total RNA. Reverse transcription reactions were performed using 400 ng RNA in a reaction with the high-capacity reverse transcription kit (Life Technologies). cDNA was used as a template to measure expression of human IRF-8 (Life Technologies) by quantitative real-time PCR. Human GAPDH (Life Technologies) served as an internal control for each reaction. Real-time PCR reactions were performed using the StepOnePlus Real-time PCR system (Applied Biosystems) with TaqMan (ThermoFisher Scientific) chemistry.

A one-way analysis of variance (ANOVA) was used to compare CD200R percentage and mean fluorescent intensity (MFI) between the three groups on the log-transformed values. A two-sample t-test was used to compare percentage and MFI CD200R expression between CD15+ and CD14+ MDSC. Similarly, two-sample t-tests compared tumor volume, intratumoral MDSC, CD4+ T cells, intratumoral CD8+ T-cell percentages, and lesions between mice receiving anti-CD200 and those receiving isotype controls. Further, differences in tumor growth between these two groups of mice, those receiving anti-CD200 antibodies and those receiving isotype controls, were evaluated with a mixed effects regression model containing random effects for intercept and slope by mouse with an unstructured covariance. Differences in tumor growth rate between mice treated with antibodies to CD200 and/or PD-1 were analyzed similarly. To determine the effects of hrCD200 and IL-6/GM-CSF versus PBS on MDSC expansion, and separately IRF-8 expression, linear regression models were fit with fixed effects for each as well as the interaction between the two. Dunnett’s post-hoc test was used to compare each hrCD200 level to 0. One-way ANOVA models with Tukey-Kramer post-hoc tests assessed each of the remaining outcomes between three groups: healthy donors, patients with PDAC, and patients with CP. For each of the above, log transformations were taken when necessary to satisfy assumptions of constant variance and normality. P-values less than 0.05 were considered significant and all analyses were conducted in SAS V.9.4.

Human pancreatic cancer cell lines (BxPC-3, MiaPaca2, and Panc-1) were confirmed to express CD200 by immunoblot analysis (figure 1A) and a moderate level of surface expression via FACS analysis (figure 1B,C). Immunofluorescence staining of human PDAC surgical specimens provides evidence that CD200 is expressed in the human pancreatic TME (figure 1D) compared with isotype control stained tissue (online supplementary figure 1). Further analysis of CD200 expression in the PDAC tissue led to the observation that CD200 was expressed by both the epithelial tumor cells and a population of α-SMA+ stromal cells. Additionally, investigation of RNA isolated from patient-derived stromal cell lines (n=10 patients) exhibited significantly increased expression of CD200 as compared with RNA isolated from normal human pancreatic fibroblasts (online supplementary figure 2; p<0.01). A commercially available tissue microarray with different levels of pancreatic cancer disease progression (PanIN-1, PanIN-2, and PanIN-3/PDAC) was stained via immunohistochemistry for CD200 expression (online supplementary figure 3A). Quantification of the CD200 expression by histology (H-score) confirmed no difference in expression levels across the spectrum of microscopic progression (ie, PanIN-1/2 compared with PanIN-3/PDAC) (online supplementary figure 3B).

Next, we investigated whether the expression of the CD200 receptor (CD200R) was expressed by MDSC populations from healthy donors, patients with CP, and PDAC. PBMC were stained for granulocytic (CD11b+CD33+HL-DR^−/lowCD15+) and monocytic (CD11b+CD33+HL-DR^−/lowCD14+) MDSC (figure 2A) and analyzed for expression of CD200R (figure 2B). As expected, patients with PDAC had elevated numbers of circulating MDSC compared with healthy donors and patients with CP (figure 2C; p=0.0060 and p=0.0042, respectively). Interestingly, MDSC in patients with PDAC had a significantly higher percentage of cells expressing CD200R (figure 2D; p=0.0433 and p=0.0132) and a higher level expression (MFI) of CD200R than healthy or CP patients (figure 2E; p<0.0001 and p=0.0001). Additionally, we observed a greater percentage (figure 2F; p=0.0007) and greater MFI (figure 2G; p<0.0001) of CD200R in CD15+ MDSC and compared with CD14+ MDSC.

We expanded our analysis of CD200R on other immune populations including macrophages, which have previously been reported to express the receptor.36 PBMC were stained for macrophage phenotypes previously described as M1 (CD14+CD163+) or M2-like (CD14+CD206+) (figure 3A). Patients with PDAC had a significantly higher number of circulating M2 macrophages (figure 3C; p<0.0001) compared with healthy or CP patients (p=0.0007) with no significant difference in M1 macrophages (figure 3B). Both M1 and M2-like macrophages expressed the CD200R (figure 3D,E). There was no significant difference in CD200R percentage expression or MFI observed in M1-like macrophages (figure 3F,G) and in M2-like macrophages (figure 3H,I) in PBMC between groups.

We investigated other relevant immune populations in patients with PDAC for differences in CD200R expression. We analyzed by flow cytometry CD4/8 T-cell populations (online supplementary figure 4A) and subpopulations (online supplementary figure 4B). We observed a higher percentage of terminally differentiated effector (CD45RA+CCR7⁻) CD4+ (online supplementary figure 4C; p=0.0261) and CD8+ (online supplementary figure 4D; p=0.0091) T cells in PDAC compared with healthy controls. However, there was no difference in the expression or number of CD200R+ T cells between the groups within any of the subpopulations (online supplementary figure 4E and F). Additionally, expressions of CD200R in NK (CD3+CD56+; online supplementary figure 5A–C), NKT (CD3^lowCD56+; online supplementary figure 5D and E), and T-regulatory cell (CD4+CD25+GITR+; online supplementary figure 5G and H) populations were found to be very low with no significant difference compared with healthy controls.

We postulated that CD200 expression in the PDAC TME is driving the expansion of MDSC, thus inhibiting the antitumor immune-mediated responses. To test this hypothesis, we utilized murine MT5 tumor cells subcutaneously injected into C57BL/6 mice. As previously described, this cell line was derived from KPC tumors and has both G12D mutated Kras and R172H Trp53.21 37 We confirmed that CD11b+GR1+MDSC from MT5 tumor-bearing mice expressed the CD200R (online supplementary figure 6A and B) similar to what was observed in patient MDSC. Once tumors were palpable, mice were treated with anti-CD200 or isotype control antibodies until study endpoint. A significant inhibition of tumor growth was observed in response to anti-CD200 antibody blockade compared with isotype control treated mice (figure 4A; p=0.0208). Tumors were processed at the study endpoint and underwent phenotypic analysis by flow cytometry. These data confirmed a significantly decreased percentage and number of infiltrating MDSC in the tumor tissue of mice receiving anti-CD200 antibodies compared with mice receiving isotype controls (figure 4B,C; p=0.0035). Additionally, we also observed a trend in an increase in intratumoral CD4+ T cells (figure 4D,E; p=0.07 and p=0.288) and no change in intratumoral CD8+ T-cell percentages and numbers (figure 4F,G).

We investigated whether blockade of CD200 could enhance the efficacy of PD-1 checkpoint antibody blockade in a murine model of PDAC. For these studies, MT5 tumor-bearing mice were treated with antibodies to CD200 and/or PD-1 until study endpoint. Mice receiving single antibody CD200 or PD-1 treatment had a significantly reduced tumor volume compared with isotype control treated mice (figure 4H; p=0.0368 and p=0.0037). Additionally, mice administered the combination of CD200 and PD-1 antibodies demonstrated significantly limited tumor growth compared with single antibody and isotype control treated mice (figure 4H; p<0.05).

To further confirm our data in a genetic model that better recapitulates many aspects of patient tumors, we used a highly aggressive, autochthonous model of spontaneously arising PDAC driven by mutant Kras, Trp53, and Brca2 which expresses CD200 similar to what we observed in human tissue (figure 5A). As previously described, there is a 100% penetrance of PDAC in these KPC-Brca2 mice at age 5–6 weeks of age.21 Additionally, KPC-Brca2 mice have identical histopathology as the KPC model, but have a more aggressive phenotype that decreases the overall survival of these mice. We have reported that KPC-Brca2 mice have similar numbers of CD3+ and F4/80+ macrophage infiltration compared with KPC mice and are typically unresponsive to single-agent immune-based therapies.21 38 Further, the increased rate of disease progression in KPC-Brca2 along with histopathology, stromal involvement, and immune infiltration make this model ideal for therapeutic studies compared with using the classical KPC genetically engineered mouse (GEM) model. KPC-Brca2 began antibody treatment at 6 weeks of age and were treated for 2 weeks with anti-CD200 or isotype control antibodies. Following 2 weeks of treatment, mice were euthanized and pancreata were isolated for histologic analysis (figure 5B). There was a significant increase in the percentage normal tissue, with fewer PanIN-3 and foci of adenocarcinoma in mice treated with antibodies targeting CD200, as compared with animals treated with isotype control Ab (figure 5C; p<0.05).

To better understand the mechanism by which CD200 affects MDSC in vivo, we further investigated PBMC from patients with PDAC and performed single-cell RNA sequencing to identify downstream pathways. Immune populations were defined using graph-based clustering and identified by expression of canonical genes (online supplementary figure 7A–C). These patients were aggregated into a single UMAP plot (figure 6A) and in figure 6B analyzed for over-expression of CD200R and canonical downstream targets Dok1 and Dok2. We identified a list of genes that were overexpressed in the cluster of MDSC that expressed CD200R/Dok1/Dok2. We investigated the differential expression of these genes (online supplementary table 1) and analyzed the signaling pathways in which these genes are active.39 40 We observed significant interactions in signaling by interferons, and other cytokines (figure 6C; p<0.05), indicating that CD200R signaling may potentially activate pathways involved in MDSC expansion.

To test whether CD200 affects MDSC expansion, we performed a series of in vitro assays on healthy donor PBMC. As previously described, healthy donor PBMC stimulated with IL-6 and GM-CSF for 7 days leads to the expansion of functional MDSC in vitro.12 For these experiments, PBMC were stimulated with increasing concentrations of recombinant human CD200 protein (hrCD200) in the absence or presence of cytokines (IL-6 and GM-CSF) for 7 days and analyzed for percentage of MDSC via flow cytometry (figure 7A). Analysis revealed a significant increase in the percentage of cytokine-driven expansion of MDSC when cells were cultured with hrCD200 (figure 7B, p<0.05). When examining downstream signaling pathways involved in MDSC expansion, we observed an increase in phosphorylation of STAT3, when cells were stimulated with IL-6/GM-CSF and rhCD200 (figure 7C). We also observed a significant decrease in the expression of IRF-8 (figure 7D; p<0.05), which can negatively regulate MDSC expansion. To test whether CD200 affects MDSC activity, we generated MDSC in vitro from healthy donor PBMC stimulated with IL-6 and GM-CSF. After 5 days, we removed cytokines and stimulated cells with 200 ng/mL of rhCD200 for 48 hours and tested their ability to inhibit autologous CD3/CD28 mediated T-cell proliferation (figure 7E). MDSC stimulated with rhCD200 resulted in a trend of greater inhibition of CD3/CD28 mediated-T-cell proliferation compared with vehicle stimulated MDSC (figure 7F,G).