Pancreatic tumor microenvironment modulation by EphB4-ephrinB2 inhibition and radiation combination

Abstract

Purpose:

A driving factor in pancreatic ductal adenocarcinoma (PDAC) treatment resistance is the tumor microenvironment, which is highly immunosuppressive. One potent immunological adjuvant is radiation therapy. Radiation, however, has also been shown to induce immunosuppressive factors, which can contribute to tumor progression and formation of fibrotic tumor stroma. To capitalize on the immunogenic effects of radiation and obtain a durable tumor response, radiation must be rationally combined with targeted therapies to mitigate the influx of immunosuppressive cells and fibrosis. One such target is ephrinB2, which is overexpressed in PDAC and correlates negatively with prognosis.

Experimental Design:

Based on previous studies of ephrinB2 ligand-EphB4 receptor signaling, we hypothesized that inhibition of ephrinB2-EphB4 combined with radiation can regulate the microenvironment response post radiation, leading to increased tumor control in PDAC. This hypothesis was explored using both cell lines and in vivo human and mouse tumor models.

Results:

Our data show this treatment regimen significantly reduces regulatory T-cell, macrophage, and neutrophil infiltration and stromal fibrosis, enhances effector T-cell activation and decreases tumor growth. Further, our data show that depletion of regulatory T-cells in combination with radiation reduces tumor growth and fibrosis.

Conclusion/Discussion:

These are the first findings to suggest that in PDAC, ephrinB2-EphB4 interaction has a profibrotic, pro-tumorigenic role, presenting a novel and promising therapeutic target.

Introduction

Pancreatic ductal adenocarcinoma (PDAC) remains a deadly disease, the third leading cause of cancer related deaths in the United States [1]. The 5-year survival rate for patients with PDAC remains at only 8% [1, 2]. A driving factor in PDAC treatment resistance is the tumor microenvironment (TME), which is fibrotic and highly immunosuppressive [3]. In addition to a desmoplastic stroma, it is composed largely of regulatory T-cells (Tregs), tumor-associated macrophages (TAMs), and myeloid-derived suppressor cells (MDSCs), which block the anti-tumoral activity of effector CD4+ and CD8+ T-cells (Teffs) [4–6]. Numerous clinical trials are considering different approaches either targeting the stroma and/or using immune-modulating agents to overcome this resistance [7, 8]. However, monotherapies aimed at blocking PD1/PDL1, CTLA4, or other immune checkpoint receptors have not demonstrated benefit thus far in clinical trials [9–11].

Radiation therapy (RT) is a potent immunological adjuvant, and it has been shown to increase Teff infiltration and activation of interferon I stimulated genes [12–14]. RT, however, has also been shown to induce infiltration of immunosuppressive populations including Tregs, TAMS, and MDSCs [15–19], which can contribute to tumor progression. Another paradox of RT is that, while very effective at killing cancer cells, it can contribute to the formation of pro-tumor fibrotic stroma by triggering an inflammatory response within the TME, recruiting stromal fibroblasts [20–24]. This process promotes tumor growth [20] and is mediated by secretion of cytokines [25]. Fibrosis is an important consideration in PDAC, which has a characteristically fibrotic and desmoplastic stroma [3] that has been shown to act as a barrier for intratumoral Teff immune infiltration [26] and to correlate with worse disease outcomes [27].

These dichotomies of the effect of RT could in part explain why this treatment has not shown improved overall survival outcomes in patients with PDAC [28]. To gain a benefit from the immunogenic effects of RT and obtain a durable tumor response, RT has to be rationally combined with targeted agents aimed at mitigating the influx of immunosuppressive cells and fibrosis. One such target is ephrinB2 (EFNB2), which is overexpressed in PDAC and correlates negatively with prognosis in multiple cancers including PDAC [29, 30]. EFNB2 is the sole ligand for the EphB4 receptor, a member of the largest family of receptor tyrosine kinases [31]. Eph receptors bind to their membrane-bound ligands, the ephrins, resulting in both forward signaling in the Eph receptor-expressing cell and reverse signaling in the ephrin ligand-expressing cell [31]. This interaction regulates multiple oncogenic processes, including angiogenesis, lymphangiogenesis, hematopoietic cell trafficking, and T-cell proliferation and activation [32–39]. More recently, in non-cancer models of cardiac, skin, and lung injury, EFNB2-EphB4 interaction has also been shown to be a key regulator of fibrosis [40, 41].

We hypothesized that inhibition of EFNB2-EphB4 signaling in combination with radiation in preclinical models of PDAC would maximize the benefit of RT by regulating the infiltrating immune population and reducing angiogenesis and fibrotic responses post RT, leading to increased tumor control. Our data show that antibody-mediated disruption of EFNB2-EphB4 signaling in combination with RT significantly reduces Treg, macrophage, and neutrophil infiltration and stromal fibrosis and enhances Teff activation compared to RT alone, leading to decreased tumor growth. Further, our data show that Treg depletion in combination with RT reduces tumor growth and fibrosis, an effect not seen with neutrophil depletion. These are the first findings to suggest that in PDAC, EFNB2-EphB4 interaction has a profibrotic, pro-tumorigenic role, and point to a novel and promising therapeutic target.

Materials and Methods

Antibodies

B11, a human scFv anti-ephrinB2 antibody, has been shown to inhibit EFNB2-EphB4 interaction and signaling [32, Figure 1] and was obtained from Dr. Jorge Martínez-Torrecuadrada (Centro Nacional de Investigaciones Oncologicas, Spain) following a production protocol previously described [32]. Sterile DPBS (Gibco, MA) was used as a control. When used in vivo, B11 or PBS was injected in the tail vein at 20mg/kg split over 5 doses given every other day, as previously described [32].

Figure 1.

EphrinB2 and EphB4 expression in samples taken from patient human and from syngeneic animal model. (A) EFNB2 and EPHB4 mRNA is overexpressed in PDAC compared to normal pancreas. Data was obtained from the TCGA and GTEX databases. *p<0.001. (B-C) IHC on tumor sections of patient human tissue with PDAC (B) and normal pancreas (C). Three different patients are shown for ephrinB2 and two for EphB4. EphrinB2 tends to be expressed on stroma while EphB4 is expressed on subsets of cancer cells and on the stroma. (D) IHC expression for ephrinB2 and EphB4 in FC1242 syngeneic mouse model. Scale bar = 50um. EphrinB2 and EphB4 were not detectable in normal pancreatic tissue.

Anti-Ly6G antibody clone 1A8 was used for neutrophil depletion. IgG2a isotype clone 2A3 was used as control. When used in vivo, anti-Ly6G and IgG2a were administered intraperitoneally at 300μg in 100μl injections three times a week, as previously described [42], and injections continued throughout the course of the study. Anti-CD25 antibody clone PC-61.5.3 was used for Treg depletion. IgG1 isotype clone TNP6A7 was used as control. When used in vivo, anti-CD25 or IgG1 was administered intraperitoneally at 250μg in 100μl injections once a week throughout the course of the study. All neutralizing antibodies were purchased from Bio X Cell (NH).

Flow cytometry conjugated antibodies used were: LIVE/DEAD Fixable Aqua Stain (Invitrogen, CA), APC-eFluor780-CD8 Clone 53–6.7 (eBioscience, CA), eFluor450-CD4 Clone RM4–5 (eBioscience, CA), AlexaFluor700-CD45 Clone 30-F11 (eBioscience, CA), DyLight350-CD3 Clone 145–2C11 (Novus, CO), PE-Cyanine7-CD11b Clone M1/70 (eBioscience, CA), BV785-Ly6G Clone 1A8 (BioLegend, CA) [42–44], BV605-Ly6C Clone HK1.4 (BioLegend, CA), FITC-FoxP3 Clone FJK-16s (Invitrogen, CA), BV605-Ki67 Clone 16A8 (BioLegend, CA), PECyanine7-IFNγ Clone XMG1.2 (eBioscience, CA), PerCP-Cy5.5-TNFα Clone MP6-XT22 (BD Biosciences, CA), and PE-CD25 Clone 3C7 (BioLegend, CA).

Primary antibodies for WB and IP were obtained from Cell Signaling Technology (MA) (anti-β-actin), Sigma-Aldrich (MO) (anti-phospho-ephrinB2), Abcam (MA) (anti-ephrinB2), Millipore (MA) (anti-phospho-Tyr), and Thermo Fisher Scientific (MA) (anti-EphB4). Horseradish peroxidase–conjugated secondary antibodies were obtained from Sigma-Aldrich (MO).

Primary antibodies for IHC were obtained from Cell Signaling Technology (MA) (anti-CD31), BD Biosciences (CA) (anti-PCNA), Biolegend (CA) (anti-Ly6G) and Invitrogen (CA) (anti-EphB4). To stain ephrinB2, two antibodies were used: anti-ephrinB2 MAb, provided by Dr. Jorge Martínez-Torrecuadrada [32], and Recombinant Mouse EphB4 Fc Chimera Protein (R&D Systems, MN) [45]. Biotinylated secondary antibodies were purchased from Vector Laboratories (CA).

Cell Lines

MIA PaCa-2, procured from the ATCC, is derived from the pancreas adenocarcinoma of a 65 year old male [46]. Cells were maintained in DMEM (Gibco, MA) supplemented with 10% FBS (Gibco, MA) and antibiotics and cultured at 37°C and 5% CO2. For shRNA transfection, an ephrinB2 targeting shRNA plasmid or a scrambled control shRNA plasmid (QIAGEN, Germany) were used with FuGENE HD transfection reagent (Promega, WI) according to manufacturer’s instructions. Briefly, cells were transfected in a 6 well plate using a 3:1 ratio of FuGENE HD (20μL) and plasmid DNA (3.3μg) in serum free OptiMEM (Gibco, MA). Cells were incubated with the transfection complex overnight before medium was replaced with fresh serum- and antibiotic-containing growth medium. G418 (InvivoGen, CA) was used as a selection antibody at a concentration of 3mg/mL. Clones were selected and screened for ephrinB2 expression by western blotting before further use.

FC1242 is a mouse cell line derived from the pancreata of KPC mice [18] and was gifted by Dr. David Tuveson (Cold Spring Harbor Laboratory, NY). Cells were maintained in DMEM supplemented with 10% FBS and antibiotics and cultured at 37°C and 5% CO2 prior to injection into mice.

Western Blot and Immunoprecipitation

Snap-frozen tumors were pulverized using sterile mortar and pestle in liquid nitrogen. For WBs, cells or tumor pieces were lysed in RIPA buffer (Millipore, MA) containing protease inhibitor cocktail (Thermo Fisher Scientific, MA) and phosphatase inhibitors (Sigma-Aldrich, MO) on ice for 30 min. For IP, tumor pieces were lysed in lysis buffer (Cell Signaling Technology, MA), containing phosphatase inhibitor on ice for 15 min. Lysates were collected and protein concentration was determined by BCA assay (Thermo Fisher Scientific, MA). For IP, 500μg of protein lysate in 250μL volume was combined with 5μg anti-EphB4 antibody and rotated end-over-end at 4°C overnight. 50μL of beads was washed of storage buffer, added to the protein+antibody mix, and rotated end-over-end at 4°C overnight. Beads were separated from supernatant by magnet, washed, and separated from attached protein by boiling in reducing agent. For WB, protein lysates (20–30μg) were loaded onto 10–12% SDS-PAGE gels and bands separated by electrophoresis. Proteins were transferred to PVDF membrane and blocked with 5% milk in TBS-Tween. Blots were probed overnight at 4°C with respective antibodies.

IncuCyte live cell analysis

To measure proliferation and apoptosis of MIA PaCa-2 cell lines, the Essen BioScience IncuCyte ZOOM system was used. 10³ cells were plated in a 96 well plate and allowed to attach overnight. Each condition was plated with 6 replicates. The plate was then placed in the Incucyte incubator, where images were taken of the wells at regular intervals for up to several days. For proliferation, well confluence was determined by phase contrast. Hours 24–104 post plating were analyzed to generate the curves. For apoptosis, cells were similarly plated and caspase 3/7 reagent (Essen Bioscience, MI) was added to the wells at 1:1000 dilution. The Incucyte system captured images showing caspase 3/7 activity reported as green fluorescence. Hours 20–120 post plating were analyzed to generate the curves. Apoptotic index was calculated as the number of caspase 3/7 expressing cells divided by percent well confluence.

Animals

Female immunocompetent C57BL/6 mice age 5–8 weeks were purchased from Jackson Labs (ME) and immunocompromised athymic nude mice age 5 weeks were purchased from Envigo (IN). According to guidelines dictated by our animal protocol, mice were sacrificed when tumor volumes reached 2000mm³, tumors ulcerated, or weight loss of more than 15% was evident. Up to 40 mice (80 tumors) were implanted in each experiment. Mice were divided into treatment groups such that each group started with as similar as possible average tumor volume. Each treatment group included a sample size of a 4–10 mice (8–20 tumors).

Tumor models

For in vivo studies, PANC193 (F3) and PANC272 (F4) PDXs were obtained from Dr. Todd Pitts’ lab (University of Colorado, Anschutz Medical Campus). Tumor pieces were implanted in mice as previously described [47]. Also used in vivo, FC1242 cells prior to passage 15 were suspended in an equal volume of Matrigel and serum free DMEM and injected subcutaneously into the flanks of immunocompetent mice at a concentration of 10⁶ cells/0.1mL for each of the two injection sites. For studies testing EFNB2-EphB4 inhibition, once tumors reached an average of about 100mm³, mice were divided into four groups: PBS Control, PBS+RT, B11, and B11+RT. RT was delivered post first B11 injection using 12Gy dose for PDX tumors and 5Gy dose for FC1242 tumors. For studies testing neutrophil depletion, treatment injections started on the same day as injection with FC1242 cells and mice were divided into four groups: IgG2a, IgG2a+RT, anti-Ly6G, & anti-Ly6G+RT. For studies testing Treg depletion, treatment injections started one week before injection with FC1242 cells and mice were divided into two groups: IgG1+RT & anti-CD25+RT. RT was delivered 8 days post tumor cell injection using 5Gy dose for both neutrophil and Treg depletion studies. All tumors were monitored two to three times a week by digital caliper, with volume calculated as (longer diameter) * (smaller diameter)^2 / 2.

Radiation

Radiation was delivered in either 12Gy or 5Gy single dose to flank tumors using one of two irradiators. The first was a 160 KVp source RS-2000 X-ray irradiator (Rad Source Technologies Inc., GA) at 25 mAmp, with a dose rate of 1.24 Gy/minute and using a 0.3 mm copper filter. Radiation in this machine was limited to only the flank tumors by using a customized shield. Mice were irradiated while under ketamine/xylazine anesthesia, administered intraperitoneally at 150mg/kg. The second was the X-RAD SmART Image Guided Irradiator (Precision X-Ray Inc., CT) at 225 kVp and 20 mA, with a dose rate of 5.8 Gy/min and using a 0.3mm copper filter. In this machine, mice under isoflurane anesthesia were placed in the prone orientation and positioned by fluoroscopy, so that each tumor would be in the corner of a 40 mm square beam, sparing all but a small amount of surrounding tissue. Irradiation was performed at the Image-Guided Monitoring and Precision Radiotherapy Shared Resource at the University of Colorado Denver.

Flow cytometry

For flow cytometric analysis of tumor tissue, tumors were digested into single-cell suspension as previously reported [14]. Briefly, tumors were finely cut and placed in HBSS (Gibco, MA) solution containing 200U of Collagenase III (Worthington, NJ) for 20 minutes at 37°C with gentle shaking every 5 minutes. After the incubation period, tumor pieces were passed through a 70μm nylon mesh. The resulting cell suspension was centrifuged and re-suspended in red blood cell (RBC) lysis buffer (Invitrogen, CA) for 3 minutes. HBSS was added to inactivate RBC lysis buffer, cell suspensions were centrifuged, re-suspended, and counted using an automated cell counter. A spleen was also collected and processed into single-cell suspensions through mechanical separation. The spleen was stained with single antibodies or fluorescence minus-one (FMOs) to assist in compensation and gating. Blood was collected by cardiac puncture into EDTA blood collection microtubes (Sarstedt, Germany) and spun at 2000rpm for 15min at 4°C to separate cells from plasma. The resulting pellet was re-suspended in RBC lysis buffer for 3 min and HBSS was added to inactivate lysis. Cells were centrifuged, re-suspended, passed through a 40μm nylon mesh, and counted. Trypan blue was used to determine cell viability. In flow panels where stimulation of T-cells was necessary, 10⁶ live cells were plated in 24-well plates and cultured for 4 hours in the presence of monensin to prevent release of cytokines and PMA and ionomycin to stimulate cytokine production. After the incubation period, cells were plated in a 96-well plate and blocked with anti-CD16/32 antibody before staining with conjugated antibodies. For analysis of cytokines IFNγ and TNFα, cells were fixed in 4% paraformaldehyde for 10 minutes at room temperature and methanol for 30 minutes at 4°C prior to staining with antibodies. For analysis of intracellular markers FoxP3 and Ki67, cells were incubated in fixation/permeabilization buffer (Affymetrix Inc., CA) at 4°C overnight before staining with intracellular antibodies. For proper compensation of flow cytometry channels, beads and single-stain spleen samples were used. For gating, FMO controls were applied. Stained cells were run on the Yeti Cell Analyzer at the University of Colorado Denver Cancer Flow Cytometry Core. Data was analyzed using Kaluza Analysis software.

IHC and fibrosis staining

Harvested tumor tissue was formalin-fixed overnight and processed for paraffin embedding. 4–6 μm thick sections were deparaffinized with xylene and rehydrated with decreasing concentrations of ethanol. For IHC, heat-mediated antigen retrieval was performed using citrate buffer. Tissues were blocked with 2% milk in TBS-Triton-X for 30min at room temperature and stained with primary antibodies overnight at room temperature a dark, humid container. Fibrosis was analyzed with Masson’s Trichrome and PicroSirius Red staining. PicroSirius Red staining was performed as described previously [48] and counterstained with Weigert’s hematoxylin. Processing of tissues for PicroSirius Red staining was completed by the University of Colorado Denver Research Histology Shared Resource. All images except for PicroSirius Red were captured on a 4x, 10x, or 20x objective using Nikon microscope. PicroSirius images were captured with a Nikon Eclipse Ni Microscope and DS-Ri2 Camera with Nikon Elements D software at 1636×1088 pixels in RGB 8-bit mode, through a Nikon Plan Fluor 10X/0.30 objective. Quantification was performed using NIH ImageJ and as previously described [48]. Briefly, fibrosis, shown by Masson’s Trichrome staining, and birefringence, shown by PicroSirius Red staining, were calculated as percent positively stained area over total tumor area. 3–5 tumors per treatment group per tumor model were stained and 5–8 fields were chosen for analysis.

Cytokine arrays and ELISAs

Blood samples were collected by retro-orbital bleed from PANC272 tumor-bearing mice 3 days post RT, FC1242 tumor-bearing B11-treated mice 22 days post RT, and anti-Ly6G-treated mice 17 days post RT. Plasma was isolated from blood as above, stored at −80°C until time of analysis, and subjected to Raybiotech (GA) Mouse Cytokine Array Q1, Meso Scale Diagnostics (MD) U-PLEX Biomarker Group 1 (ms) Assays, R&D Systems (MN) Mouse/Rat/Porcine/Canine TGFβ−1 Quantikine ELISA Kit, and/or BioGems (CA) Human VEGF Pre-Coated ELISA as per manufacturer’s instructions.

Mass spectrometry

Briefly, snap frozen tumor samples of approximately 5mg were powderized in liquid nitrogen. High salt extraction was used to make the cell fraction. Guanidine extraction was used to make the sECM fraction. Hydroxylamine extraction was used to make the iECM fraction. Fractions were digested with trypsin. 600ng of protein was analyzed nano-UHPLC-MS/MS (Easy-nLC1200, Orbitrap Fusion™ Lumos™Tribrid™, Thermo Fisher Scientific). Files were loaded into Proteome Discoverer 2.2 and were searched against Swissprot mouse and human database. In Excel, protein abundances from the three fractions were summed together per protein. Data was visualized in excel and MetaboAnalyst 4.0 [49]. See Supplementary Methods for full details.

Statistics

Experiments were performed in triplicate and repeated two to three times. Quantitative analyses were performed in GraphPad Prism using Student’s t-test to compare two groups, ANOVA to compare multiple groups, nonlinear regression curve fit to compare doubling times of cell lines, or log-rank (Mantel-Cox) test to compare survival curves. Any data point determined to be an outlier by Grubb’s test was excluded from analysis. For survival analysis, Kaplan-Meir curves were analyzed based on the Log-rank (Mantel-Cox) test for comparison of all groups. A p-value of < 0.05 was considered significant. Data in figures represents Mean±SEM.

Study approval

All mice were handled and euthanized in accordance with the ethics guidelines and conditions set and overseen by the University of Colorado, Anschutz Medical Campus Animal Care and Use Committee. All protocols for animal studies were reviewed and approved by the Institutional Animal Care and Use Committee at the University of Colorado Anschutz Medical Campus.

Results

Effect of EFNB2 knockdown on cancer cells in vitro

EFNB2 expression in PDAC has been shown to correlate with poor prognosis [29, 30]. In addition, analysis of gene expression data show that mRNA of EFNB2 and EPHB4 is significantly overexpressed in PDAC (TCGA, n=179) compared to normal pancreas (GTEX, n=171) (Figure 1A) [50]. However, it remains unknown which EFNB2-expressing cells in the tumor impact prognosis. We have previously shown that inhibition of the EFNB2-EphB4 signaling axis in head and neck squamous cell carcinoma cell lines enhances apoptosis [51], but this had not been tested in PDAC cells. We tested whether EFNB2 signaling on cancer cells has direct effect on cell survival by creating an EFNB2 knockdown cell line. MIA PaCa-2, a PDAC cell line showing high expression of EFNB2, was transfected with EFNB2 shRNA or a scrambled shRNA control (Supplementary Figure 1A). Proliferation of this knockdown cell line compared to parental control was assessed by plate confluence. No significant difference in proliferation was observed between the control end EFNB2 knockdown cell lines (Supplementary Figure 1B). Apoptosis was assessed via caspase 3/7 expression, and no difference in apoptotic index was observed in the knockdown cell line as compared to the control (Supplementary Figure 1C). Thus, cancer cell survival in vitro does not appear to be directly influenced by EFNB2 signaling. These results are specific to monolayer growth; however, we determined these assays to be sufficient to demonstrate that the effect of EFNB2 signaling in PDAC is not a cancer cell autonomous mechanism. We hypothesized instead that the negative prognosis correlated with EFNB2 expression in PDAC is attributable to an indirect effect on cancer cells by EFNB2 signaling in the TME, an environment that can only be fully studied in vivo.

To test the feasibility of this hypothesis, we characterized where EFNB2 ligand and its receptor EphB4 are expressed in our tumor models as well as normal pancreatic tissue from patients by immunohistochemistry (IHC). PANC272, a patient derived tumor xenograft (PDX) model, and FC1242, a mouse pancreata tumor-derived model, were grown in mouse flanks, harvested, and stained with anti-ephrinB2 and anti-EphB4 antibodies (Figure 1B,C). Biopsied tissue from patients with PDAC and from normal pancreas were stained similarly (Figure 1B,C). Protein expression of EFNB2 was observed on the stromal cells in human patient tissue (Figure 1B) as well as in our mouse model (Figure 1D). EphB4, on the other hand, was expressed on the cancer cells and on the stroma in human and mouse tissue (Figure 1B,D). No expression of either EFNB2 or EphB4 was found in normal pancreatic tissue (Figure 1C).

Inhibition of EFNB2-EphB4 signaling enhances response to radiation therapy (RT) and reduces tumor proliferation in in vivo PDAC tumor models

We tested if administration of the antibody B11, which inhibits EFNB2-EphB4 signaling [32, Figure 1], would enhance response to RT. B11 was validated to block the phosphorylation of both EFNB2 and EphB4 by western blot (WB) and immunoprecipitation (IP), respectively, in PANC272 tumor lysates (Supplementary Figure 2).

We observed a significant increase in tumor control in tumors receiving B11 combined with RT compared to all other groups in PANC272 and compared to PBS and B11 alone groups in PANC193 (Figure 2A-B). In PANC272, tumors treated with B11+RT reached a Mean ± SEM fold change increase from day 0 of 1.765±0.215, whereas PBS was 5.451±1.329, RT 2.982±0.225, and B11 2.864±0.520 (Figure 2A). In PANC193, there was a significant increase in tumor control in B11+RT (Mean ± SEM fold change increase from day 0 of 1.333±0.062) compared to control (5.918±0.764) or inhibitor alone (4.545±0.531). RT alone reached a fold change increase of 1.417±0.117 (Figure 2B). In murine FC1242 tumors, the 12Gy dose of RT used in the PDX models proved to be too high to detect any differences between the irradiated groups. FC1242 cells in vitro have a survival fraction (SF) of 0.1% when dosed with 12Gy, as determined by in vitro clonogenic assay which prohibits long-term analysis of RT effects (Supplementary Figure 3). Therefore, RT was reduced to 5Gy, with an SF of 15.5%. In FC1242 tumors, B11+RT had a Mean ± SEM fold change increase compared to day 0 of 1.846±0.132, which is significantly smaller than PBS (3.013±0.277), RT (2.252±0.134), or B11 alone (2.473±0.247) (Figure 2C).

Figure 2.

EphB4-ephrinB2 inhibitor (B11) adds to the effect of radiation therapy (RT) in PDAC. A-C. Fold change of tumors over time for PANC272 (A), PANC193 (B), and FC1242 (C), with dot plots of single representative days shown below. Day 0 represents the day mice were divided into treatment groups. # compares PBS to B11+RT, * compares RT to B11+RT, $ compares B11 to B11 + RT. D. Representative IHC images of staining for PCNA in PANC272 and FC1242, with quantification of number of PCNA+ cells per HPF to the right. Scale bar = 50um. n=3–5 tumors per group, 5–8 images per tumor. */#/$ p<0.05, **/##/$ $ p<0.01, ***/###/$ $ $ p<0.001. One-way ANOVA test on PANC272 and PANC193 has a p-value of 0.01 and 0.02, respectively.

To explore mechanisms of how B11+RT reduces tumor growth, tumors were stained for proliferating cell nuclear antigen (PCNA). Proliferation, as determined by PCNA expression, was significantly decreased in PANC272 B11-treated groups compared to PBS by 30–37%. Likewise, in FC1242 B11-treated groups, we observed 24–38% fewer PCNA positively stained cells per high powered field (HPF) compared to RT (Figure 2D). Since inhibition of EphB4 and EFNB2 has been previously shown to negatively regulate cell death and apoptosis [32, 52, 53], we analyzed changes in these pathways. We performed mass-spectrometry on PDX tumors and mapped detectable proteins to human and mouse genomes. This allowed us to identify changes to human tumor cells and mouse stromal components. For apoptosis-related markers, we observed a significant increase in Caspase 3, Caspase 7 and Caspase 8 in the RT+B11 group compared to the PBS or RT groups (Supplementary Figure 4). B11 alone and RT alone were able to increase Caspase 3 only, indicating that the combination of RT and B11 is necessary for driving tumor cell apoptosis. These data substantiate the hypothesis that EFNB2 affects PDAC growth in vivo through signaling in the TME, which indirectly affects tumor cell proliferation and apoptosis.

Inhibition of EFNB2-EphB4 increases Teff activation and decreases immunosuppressive factors in irradiated tumors

We analyzed how B11+RT treatment affects the immune landscape as another potential mechanism of tumor control. FC1242 tumors and peripheral blood were harvested for flow cytometry one-week post RT and 8 days post start of B11 treatment. First, there is evidence of immunologic benefit to RT, shown by increased levels of activation markers IFNγ and TNFα on CD8+ and CD4+ T-cells (Figure 3A-D). RT increased CD8+IFNγ+ and CD4+TNFα+ cells by 2-fold and CD8+TNFα+ and CD4+IFNγ+ cells by 4-fold compared to PBS. These effects were sustained or enhanced by the addition of B11 to RT. Total CD8 and CD4 T-cell populations can be seen in Supplementary Figure 5A-B. Second, there is additional benefit to B11 treatment, shown by a significant decrease of 2–3 fold in intratumoral Tregs in B11-treated groups compared to PBS or RT (Figure 3E). In addition to the effect of B11 and RT on intratumoral Tregs, we assessed the impact on myeloid populations by flow cytometry. We observed a significant decrease in the proportion of intratumoral macrophages in the B11 alone and B11+RT groups compared to PBS or RT alone (Figure 3G). Similarly, there was a significant decrease in tumor infiltrating neutrophils in the B11+RT group compared to PBS or RT alone (Figure 3H) and in circulating neutrophils in the B11+RT group compared to PBS (Supplementary Figure 5C). There was a 1.7–1.9 fold decrease in intratumoral neutrophils in the B11+RT-treated group compared to PBS or RT. To evaluate whether B11 treatment can induce neutrophil reduction in another tumor model, we stained tumor sections from the PANC272 long term effect study for Ly6G (Supplementary Figure 5D). We observed a significant decrease in neutrophils in the B11+RT treated group compared to PBS. Taken together, these data show a critical role for EFNB2-EphB4 inhibition in promoting an anti-tumor immune response and suppressing pro-tumorigenic factors including Tregs, macrophages, and neutrophils.

Figure 3.

EphB4-ephrinB2 inhibition in combination with RT enhances cytotoxic CD4 and CD8 T-cell activation and decreases regulatory Treg, macrophage, and neutrophil infiltration. A-H. Percentages of FC1242 tumor-infiltrating IFNg+CD8+ T-cells (A), TNFa+CD8+ T-cells (B), IFNg+CD4+ T-cells (C), TNFa+CD4+ T-cells (D), regulatory T-cells (E), Treg to Teff (CD8+ Tcells and FoxP3-CD4+ T-cells) ratio (F), macrophages (G), and neutrophils (H). n=3–5 per group. *p<0.05, **p<0.01, *** p<0.001. A and B show percentages of live/CD45+/CD3+/CD8+ parent populations, C-E of live/CD45+/CD3+/CD4+ cells, and F of live/CD45+/CD11b+ cells.

Depletion of neutrophils with or without radiation is not sufficient to control tumor growth

Given that B11 decreased intratumoral neutrophils and the clinical data from Schernburg et al, 2017, showing that neutrophil count is predictive for response to RT [54], we tested if neutrophil depletion was sufficient to inhibit tumor growth when combined with RT. Anti-Ly6G has been previously shown to deplete neutrophils in mice [43, 44]. We first confirmed the neutrophil depletion efficacy by flow cytometry on FC1242 tumors and blood samples from tumor-bearing mice. Circulating neutrophils in the blood were reduced by 99% compared to control and intratumoral neutrophils were reduced by 98% compared to control (Figure 4B). Once depletion was confirmed, we examined the long-term effects of neutrophil depletion combined with RT (Figure 4A). With the exception of one time point (day 22), there was no significant difference between IgG and anti-Ly6G single-therapy groups. At no time point was there any significant difference between IgG+RT and anti-Ly6G+RT groups. A difference in ulceration rate of tumors between anti-Ly6G single therapy group and all others was observed, however, with anti-Ly6G having the slowest rate of ulceration (Figure 4C). These data demonstrate that the therapeutic efficacy of EFNB2-EphB4 inhibition is not solely dependent on neutrophils.

Figure 4.

Neutrophil depletion does not alter disease progression in the FC1242 PDAC model, but Treg depletion does significantly delay tumor growth. A. Tumor volumes treated with IgG control or anti-Ly6G antibody, alone or in combination with 5Gy RT. Day 0 represents the day of tumor cell injection. RT administered on day 8 post tumor cell injection. n=10 tumors per group. B. Neutrophil depletion on circulating blood and tumor. n=3 per group. C. Survival curves based on ulceration of 10 tumors in each group. Ulceration of a tumor was input as an event, thus lowering the percent of non-ulcerated tumors for that group, while harvesting of a non-ulcerated tumor due to ulceration of its biflank partner was censored. *p<0.05, **p<0.01, *** p<0.001, as determined by Student’s t-test on neutrophil levels or by log-rank (Mantel-Cox) test on ulceration rate. D. Volume of tumors treated with 5 Gy RT and IgG control or anti-CD25 antibody. Day 0 represents the day of tumor cell injection. RT administered on day 8 post tumor cell injection. n=10 tumors per group. E. Treg depletion on circulating blood. n=5 per group. F. CD8 and CD4 IFNg and TNFa T-cell populations in tumors 3 days post RT. n=3–4 per group. G. TGFB-1 plasma levels decrease with Treg depletion in FC1242 tumors. n=3 per group. *p<0.05, **p<0.01, *** p<0.001. E shows percentages of live/CD45+/CD3+/CD4+/CD25+ parent population, F % of live/CD45+/CD3+/CD8+ or live/CD45+/CD3+/CD4+ cells.

Depletion of Tregs combined with RT controls tumor growth more effectively than RT alone

Given that B11 decreased the recruitment of Tregs to the tumor, and with previous studies showing that Tregs are associated with prognosis and predictive of survival in PDAC and that blocking Treg migration in PDAC inhibits tumor growth [55, 56], we tested if Treg depletion was sufficient to inhibit tumor growth when combined with RT. A cohort of mice with FC1242 tumors was treated with RT and anti-CD25 antibody or the corresponding IgG control antibody and monitored for tumor growth (Figure 4D). Treg depletion efficacy was confirmed in each mouse using flow cytometry on blood samples (Figure 4E). Depletion of Tregs in combination with RT resulted in significantly increased tumor control compared to RT alone (Figure 4D). At day 26, tumors treated with IgG+RT reached a Mean ± SEM tumor volume of 424.4±25.07 mm³, whereas anti-CD25+RT was 211.2±25.17 mm³. This was accompanied by significantly higher levels of Teff activation 3 days post RT (Figure 4F). Treg depletion combined with RT increased CD8+IFNγ+ cells by 1.6-fold, CD8+TNFα+ cells by 1.9-fold, CD4+IFNγ+ cells by 3-fold, and CD4+TNFα+ cells by 1.4-fold compared to RT alone. Interestingly, plasma samples from the Treg depletion study in FC1242 tumors also showed significantly reduced circulating TGFβ−1 between IgG+RT and anti-CD25+RT (Figure 4G). These data show that Tregs are key regulators of tumor growth in PDAC.

Blockade of EphB4-ephrin-B2 signaling with B11 and depletion of Tregs decreases fibrosis

PDAC-associated fibrosis has been viewed as a physical barrier that compromises drug delivery, reduces immune cell accessibility, and promotes disease aggression and therapy resistance [57]. More recently, EFNB2 has been shown to be a critical regulator of fibrosis in other model systems [40, 41]. We therefore tested the effect of EphB4-EFNB2 inhibition with and without RT on fibrosis. RT significantly increases levels of fibrosis in the PANC272 and FC1242 tumors by 1.6–2.6 fold change in the RT groups as compared to all others, both by Masson’s Trichrome and PicroSirius Red staining (Figure 5A,B). This correlates with significantly reduced circulating levels of the pro-fibrotic cytokine IL-17 in PANC272 B11+RT mice compared to RT alone (Figure 5D) [58]. Similarly, Treg depletion combined with RT in FC1242 tumors reduced fibrosis compared to RT alone, a fold change of 2.2 in Masson’s Trichrome staining and 1.4 in PicroSirius Red staining (Figure 5C). Analysis of changes of relative abundance of extracellular matrix (ECM) proteins showed decrease in fibrillar collagen (i.e. Col1a1, Col1a2, Col2a1, Col3a1, and Col5a2) after treatment with B11 or B11+RT relative to PBS or RT treated groups (Figure 5E). Collectively, these data show that EFNB2-B4 inhibition can significantly reduce PDAC fibrosis, which is a key mediator of treatment resistance in PDAC patients.

Figure 5.

RT increases stromal fibrosis, which is significantly reduced by B11 or Treg depletion. A-C. Fibrosis staining in PANC272 (A) and FC1242 (B) tumors treated with B11 and in FC1242 tumors treated with Treg depleting antibody (C). Representative images to the left, quantification to the right. Masson’s stains collagen in blue and keratin and muscle fibers in red. PicroSirius images taken through polarized light to mark collagen. n=3–5 tumors per group, 5–8 images per tumor. Scale bar = 50um in PANC272, 100um in FC1242. D. IL-17 plasma levels in PANC272 tumor-bearing mice. E. Heatmap of changes in the relative abundance of matricellular proteins and extracellular components from PANC272 tumor-bearing mice treated with B11 with or without radiation. Analysis included of n=2–3 per group. *p<0.05, **p<0.01, *** p<0.001.

Inhibition of EFNB2-EphB4 decreases angiogenesis in irradiated tumors

Knowing EFNB2-EphB4 signaling is important in angiogenesis [37, 38] and that B11 has been shown to inhibit functional blood vessel formation in another pancreatic tumor model [32], we assessed the effect of B11+RT treatment on angiogenesis in tumors, as a potential physiologic mechanism of achieving overall tumor control. Blood vessel density, calculated as percent positive CD31 stained area over total tumor area, is significantly reduced in B11-treated groups compared to PBS or RT in PANC272, and compared to PBS in FC1242 (Figure 6A). In PANC272, B11-treated groups had 46–60% of CD31 staining as PBS and 53–68% of CD31 staining as RT. In FC1242, B11-treated groups had 47–54% of CD31 staining as PBS. Our mass-spectrometry analysis showed that RT and B11 treatment significantly downregulated pro-angiogenic proteins Paxillin and PLCG2 relative to PBS control or RT alone groups (Figure 6B,C). Additionally, circulating VEGF was significantly lower in PANC272 B11-treated groups compared to PBS and in B11+RT compared to RT (Student’s two-tailed t-test between B11 and RT p=0.08) (Figure 6D). B11-treated groups had 26–39% less circulating VEGF compared to PBS in PANC272 and 35–47% less compared to RT. Collectively, these data show that RT-induced pro-angiogenic factors can be reversed with concurrent B11 treatment.

Figure 6.

Inhibition of EFNB2-EphB4 interaction decreases blood vessel density and angiogenic protein levels. A. Representative images of staining for CD31 marking blood vessels in PANC272 (top) and FC1242 (bottom), with quantification of %CD31+ area to the right. n=3–5 tumors per group, 5–8 images per tumor. Scale bar = 100um. B-C. Analysis of protein expression levels of Paxillin (PXN) and PLCG2 using mass-spectrometry in Panc272 PDX tumors. Bars represent SEM of 4 independent samples. D. VEGF levels in plasma from PANC272 tumor-bearing mice. n=3–4 samples per group. *p<0.05, **p<0.01, *** p<0.001.

Discussion

Pancreatic ductal adenocarcinoma is an aggressive and deadly disease. Current treatment modalities are largely ineffective, including radiation therapy. The paradoxical effects of RT may be to blame for its apparent lack of survival benefit for patients with PDAC in clinical trials [28]. In this study, we aimed to suppress the potential negative effects of RT—increased infiltration of immunosuppressive populations and fibrosis—while enhancing its positive effects on Teff activation and cancer cell death. Using three models of PDAC, we found that targeting the EFNB2-EphB4 axis renders tumors more responsive to RT. In exploring potential mechanisms of this effect, reduced Treg, macrophage, and neutrophil infiltration, fibrosis, angiogenesis, and tumor proliferation, and increased Teff activation were evident. Further study on Treg depletion revealed this immune subset to be an important component in PDAC radioresistance and fibrosis. Broad applicability was derived from both human and mice translational models highlighting the potential of the proposed therapeutic approach.

RT has had popularity as an immune adjuvant [12, 13], and our studies in FC1242 support this concept by showing an increase in tumor infiltrating activated Teffs post RT. However, especially in the rich immunosuppressive environment of PDAC, a boost to Teffs is not enough. The benefit of RT did not translate to durable tumor growth delay in our models. However, the addition of B11 not only sustained or in some cases augmented the levels of activated Teffs, it also reduced the intratumoral Treg, macrophage, and neutrophil populations. Altogether, this decreased the Treg to Teff ratio, and tumors receiving B11+RT showed enhanced tumor growth delay.

Based on the observation of reduced Tregs after B11 treatment, we initiated Treg depletion studies to further understand radioresistance mechanisms of PDAC. Tan, et al, 2009, showed that inhibiting Treg recruitment to PDAC tumors by disrupting the CCR5/CCL5 axis slowed tumor growth [56]. Our findings are consistent and show a novel method of Treg reduction in the use of an EFNB2-EphB4 inhibitor and a novel concept of enhanced response to RT when combined with Treg depletion. These findings are clinically relevant and should be further evaluated clinically to overcome resistance to therapy in patients with PDAC.

Treatment with B11 also decreased infiltrating neutrophils, which is of clinical interest. A recent subset analysis of the LAP07 trial, which originally negated any survival benefit in adding RT to chemotherapy in PDAC, showed that baseline neutrophilia or increased neutrophil counts after chemotherapy induction predicts for resistance to chemoradiation [54]. Relatively high numbers of tumor-associated neutrophils (TANs) have been reported in PDAC and correlate with poor prognosis [59, 60], but how TANs contribute to treatment resistance is largely unknown. However, the combination of neutrophil depletion and RT had no additional benefit to delay tumor growth in our models. It is possible that a combination of neutrophil and Treg depletion is needed, or that a subset of neutrophils needs to be targeted rather than the entire population. TANs are known to have dichotomous antitumor (N1) and protumor (N2) phenotypes depending on the TME. Although the entire functionality of these subclasses is not known, it is generally thought that N1s are cytotoxic to tumor cells, inhibit metastasis, and promote an antitumor immune response, while N2s are immunosuppressive and promote angiogenesis and tumor cell survival [61, 62]. Perhaps targeted depletion of N2s, if it becomes available in the future, would result in reduced tumor growth, especially if combined with Treg depletion.

Another apparent effect of B11 treatment is the reduction in infiltrating macrophage levels. As mentioned earlier, TAMs make up a large proportion of the PDAC TME, so this population is likely relevant in treatment resistance [5, 63, 64]. It has also been shown that infiltrating macrophages increase 3 days post RT in the FC1242 model [18], another negative effect of RT. Our data is in concordance with this, as we also see a significant increase in the RT-treated group. The reversal of this effect with B11 treatment likely contributes to overall tumor control. We suspect that B11 is having a direct effect on macrophages, as they are known to express EFNB2 and EphB4 [63, 64], but further study on this is warranted. Like neutrophils, macrophages can have antitumor (M1) and protumor (M2) phenotypes [5], and elucidating how each subtype contributes to tumor growth and is effected by B11 treatment is the subject of further study.

The effect of EFNB2-EphB4 blockade cannot, however, be merely related to the adaptive immune system, as a benefit of B11+RT was also seen in our PDX models implanted in immunocompromised mice. This receptor tyrosine kinase pair has a well-established role in angiogenesis, leading us to analyze vascular markers in our models. Although PDAC is characteristic of low microvascular density and leaky vasculature [65], disrupting the fine angiogenic balance established by the tumor could show therapeutic benefit [3]. This becomes especially relevant when considering high doses of RT, which have been shown to be destructive to the vascular system in other tumor models [66, 67]. A method of regulating angiogenesis post RT could become clinically important if the field continues to push toward higher doses and fewer fractions of RT. Our and other’s findings show B11 as a potential option for treatment [32].

PDAC-associated fibrosis has been viewed as a barrier that compromises drug delivery, reduces immune cell accessibility, and promotes disease aggression and therapy resistance, underscoring the critical importance of tissue mechanics to tumor biology [57, 68]. A major benefit of inhibiting EphB4-EFNB2 interaction is the reduction in fibrosis. We observed a decrease in the levels of abundant fibrillary collagens Col12a1, Col14a1, Col3a1, and Col15a2 with B11 treatment and especially in combination with RT. These collagens are reported to associate with the surface of existing collagen fibrils to create bridges between fibrils [69]. High levels of expression of collagens Col1α2, Col2α1, and Col4α1 have been shown to associate with lower overall survival in patients with PDAC [70]. In a cardiac fibrosis model, it was shown that EFNB2 activates Stat3 and TGFβ/Smad3 signaling to induce cardiac fibrogenesis [40], and in lung and skin fibrosis models, it was shown that soluble EFNB2 promotes fibroblast chemotaxis and activation [41]. EFNB2 is known to be a regulator of TGFβ−1 signaling, and it has been shown that mice injected with lentivirus EFNB2 shRNA produce less active TGFβ−1 than control plasmid injected mice [40]. TGFβ−1 is highly relevant in PDAC, where it is overexpressed and promotes fibrosis [71]. Our Treg depletion studies also showed reduced levels of TGFβ−1, indicating a role by Tregs in TGFβ−1 secretion. However, Tregs cannot be the only population contributing to fibrosis, as both our PDX and mouse cell line tumor models showed a significant reduction in fibrosis when treated with B11 or Treg depletion. This reversal of the fibrotic response to RT is accompanied by a significant reduction in circulating IL-17 in the PANC272 model, a proinflammatory cytokine shown to participate in fibrosis formation [58]. Our findings are supported by a cardiac fibrosis model, where it was shown that EFNB2 activates Stat3 and TGFβ/Smad3 signaling to induce cardiac fibrogenesis [40], and lung and skin fibrosis models, where soluble EFNB2 promotes fibroblast chemotaxis and activation [41].

In conclusion, our study provides strong evidence of Tregs contributing to PDAC resistance to RT and a potential method for reduction in Tregs as well as fibrosis, two areas of clinical importance in PDAC, via EFNB2-EphB4 blockade. The effects of targeting EFNB2-EphB4 are not limited to Tregs and fibrosis, but also include reduced angiogenesis and neutrophil infiltration and increased Teff activation. The ability to target multiple tumorigenic pathways is a strength of targeting this signaling pair. Our findings have translational potential to enhance the therapeutic efficacy of RT in patients with PDAC.

Translational Relevance.

The addition of radiation therapy (RT) has not been shown to improve survival for pancreatic ductal adenocarcinoma (PDAC). RT can act as a strong immunological adjuvant that increases effector T-cell infiltration, but it can also induce the infiltration of immunosuppressive populations and can contribute to the formation of fibrosis. In PDAC these factors have been shown to correlate with poor disease outcome. Therefore, inducing a highly immunogenic response with RT must be combined with targets aimed at mitigating the influx of immunosuppressive cells and fibrosis. Here we show that inhibition of EphB4-ephrinB2 signaling in combination with RT reduces tumor growth, fibrosis, angiogenesis, and macrophage, neutrophil, and regulatory T-cell infiltration. The therapeutic implications of this study hold promise for clinical translation in patients with PDAC.

Abbreviations List

RT

Radiation Therapy

PDAC

Pancreatic ductal adenocarcinoma

TME

Tumor microenvironment

Tregs

Regulatory T-cells

TAMs

Tumor-associated macrophages

MDSCs

Myeloid-derived suppressor cells

Teffs

Effector CD4+ and CD8+ T-cells

EFNB2

EphrinB2

IHC

Immunohistochemistry

PDX

Patient derived tumor xenograft

WB

Western blot

IP

Immunoprecipitation

SF

Survival fraction

PCNA

Proliferating cell nuclear antigen

HPF

High powered field

ECM

Extracellular matrix

TANs

Tumor-associated neutrophils