Crosstalk between costimulatory and coinhibitory ligands are a prominent node of immune cell regulation. Mounting evidence points toward a critical role for CD155, the poliovirus receptor, in suppressing T cell function, particularly in cancer. However, relative to other known costimulatory/coinhibitory ligands (e.g., CD86, CD80, PD-L1), the physiological functions of CD155 and the mechanisms controlling its expression remain unclear. We discovered that CD155 expression is coregulated with PD-L1 on tumor-associated macrophages, is transcriptionally regulated by persistently active aryl hydrocarbon receptor (AhR), and can be targeted for suppression via AhR inhibition in vivo. Therapeutic inhibition of AhR reversed tumor immunosuppression in an immune competent murine tumor model, and markers of AhR activity were highly correlated with tumor-associated macrophage markers in human glioblastomas. Thus, CD155 functions within a broader, AhR-controlled macrophage activation phenotype that can be targeted to reverse tumor immunosuppression.
Elucidating the biology of immune checkpoints and costimulatory molecules inspired the development of effective therapeutics for cancer, autoimmunity, and organ transplantation. CD155, originally discovered as the poliovirus receptor (PVR) (1), is an immune checkpoint gaining interest as a target for cancer immunotherapy (2). CD155 is virtually universally expressed in solid neoplasia (3) and genetic ablation of the murine CD155 ortholog (gene name PVR) restricts tumor growth and metastases and bolsters the antitumor efficacy of anti-PD1 and anti-CTLA4 (4). This indicated a nonredundant role for CD155 as an immune checkpoint (4).
However, targeting CD155 itself for therapy is hindered by its widespread expression outside of the tumor site [e.g., on vascular endothelial cells (3) or spinal cord anterior horn motor neurons (5)]. Moreover, CD155 binding–activated receptors have both activating (e.g., DNAM-1) (6) and suppressive (e.g., TIGIT) (7) roles in shaping the tumor immune landscape. Thus, unraveling the biological context(s) and mechanisms of CD155 expression control may reveal more tractable routes to target this immune checkpoint. Given the functional importance of CD155 (4) and PD-L1 (8) in myeloid cells, the influence of PD-L1 on cancer immune resistance (9), as well as the pivotal role of APCs in determining T cell function, we investigated the biological role of CD155 expression in tumor-associated macrophages (TAMs)/human macrophages.
In this study, we define a relationship between CD155 and PD-L1 expression on TAMs, induced by signals that have historically been associated with “classical” or “alternative” activation in macrophages, and discover a mechanism of CD155 transcriptional regulation via persistently active aryl hydrocarbon receptor (AhR). Inhibiting AhR mitigated CD155 expression on TAMs and reversed tumor-intrinsic immune suppression in a murine immunocompetent tumor model, revealing a node at which the CD155 checkpoint may be clinically targeted.
Materials and Methods
Analysis of glioblastoma samples
Deidentified surgically resected tumor tissue was collected from consented subjects under an institutional review board–approved protocol. Glioblastoma (GBM) tumor tissue was collected within 1 h of resection by the Duke Preston Robert Tisch Brain Tumor Center BioRepository. Specimens were dissociated in RPMI 1640 containing 100 μg/ml Liberase-TM (Sigma-Aldrich) and 10 μg/ml DNAse I (Roche Diagnostics) for 20 min at 37°C with agitation. Single-cell suspensions were filtered through 70- and 40-μm cell strainers (Olympus Plastics), washed in PBS (Life Technologies), and reconstituted in PBS containing 2% FBS (Sigma-Aldrich) with 1:20 Human Tru-Stain FcX Block (BioLegend). Cell suspensions were stained with Abs against CD45-BUV395 (BD Biosciences), CD14-BV421, CD33-BV510, HLA-DR-BV786, CD31-FITC, CD3/19-BUV737, CD11b-allophycocyanin, CD16-BV711, CD15-allophycocyanin-fire7, and either CD155-PE and PD-L1–BV605 or isotype control-PE and -BV605 Ab (all BioLegend), followed by washing and reconstitution in 7-AAD containing PBS + 2% FBS. Cells were gated for appropriate size on side scatter area (SSC-A) and forward scatter area (FSC-A); single cells by proportionate FSC-height and FSC-A size; live cells by 7-AADNeg; nonendothelial cells by CD31–FITC-ANeg; CD45 by CD45-BUV396-A+; non–NK, B, or T cells by CD56–BUV737-A−, CD19–BUV737-A−, and CD3–BUV737-A−; and for macrophages by CD11b–allophycocyanin-A+, CD16-BV711-A+, and CD14-BV421-A+.
LPS (Invivogen), IL-4, M-CSF, CH223191, SR1 (StemCell Technologies), JW67, PMA, GNF-351, and SB203580 (Sigma-Aldrich) were reconstituted per manufacturer’s instructions; concentrations are noted in the figure legends. Leukopaks (StemCell Technologies) from four different deidentified donors were processed using Leucosep tubes (Greiner Bio-One) and Ficoll-Paque Plus (GE Healthcare) following the manufacturer’s instructions to isolate PBMCs. E0771 cells (a gift from Greg Palmer, Duke University) were grown in high-glucose DMEM (Life Technologies) supplemented with 10% FBS. All cell lines were confirmed mycoplasma free. THP-1 and U937 cells (both American Type Culture Collection) were cultured in suspension in RPMI 1640 supplemented with 10% FBS, antibiotics, and 0.1% β-ME (THP-1; Sigma-Aldrich).
In vitro experiments
Frozen PBMCs were thawed in 10 ml AIM-V media (Invitrogen), and incubated in 2 ml of AIM-V media containing 10 μg/ml DNAse I (Roche Diagnostics) (15 min). Cells were spun down and plated at a density of 1 × 106 PBMCs per well in six-well plates in DMEM (Life Technologies) containing 10% FBS, in the presence of 50 μg/ml M-CSF (StemCell Technologies) for 7 d. THP-1 and U937 cells were plated at a concentration of 1.2 × 106 cells per well (35 mm well) and differentiated using PMA (100 μg/ml for THP-1, 50 μg/ml for U937) for 48 h (THP-1) or 72 h (U937). Media were changed and cells were allowed to rest (24 h) before treatment. Unless otherwise noted, IL-4 or LPS treatment was for 24 h.
Immunoblots were performed as reported earlier (10) with Abs against CD155, PD-L1, p-STAT6(Y641), STAT6, p-p38, p38, α-tubulin, p–NF-κB(S536), NF-κB, p-JNK, JNK, p-STAT1 (Y701), STAT1, p-MK2(T334), MK2, AhR, PARP, GAPDH (Cell Signaling Technology). Unless otherwise noted, immunoblots were performed using whole-cell lysates. Where relevant, densitometric quantifications are shown as a change in protein level normalized to tubulin compared with control. In the case of CD155 and PD-L1 induction, fold change post–IL-4 or -LPS treatment is shown in comparison with unstimulated cells treated with the same inhibitor or small interfering RNA (siRNA). In the case of STAT6 and AhR, fold change is shown in comparison with untreated cells.
siRNA, fractionation, and quantitative RT-PCR
siRNA against STAT6, AhR, or All-Stars negative control siRNA (Qiagen) were complexed with lipofectamine LTX following the manufacturer’s instructions, delivering 100 pmol of siRNA per 35-mm dish for 6 h. Treatment of siRNA-treated monocyte-derived macrophages (MDMs) occurred 36 h after transfection. For cell fractionation, NE-PER Nuclear and Cytoplasmic Extraction Reagents (catalog no. 78835; Thermo Fisher Scientific) were used according to manufacturer’s protocol. For quantitative real-time PCR (qRT-PCR), cells were pelleted and lysed in TRIzol for total RNA extraction as described previously (11). RNA samples were treated with GeneJET RNA Cleanup and Concentration Micro Kit (K0842; Thermo Fisher Scientific) according to the manufacturer’s protocol. Mixtures of 30 ng of sample RNA, 10 μL of 2×, and 1 μL of probe (human PVR [Hs00197846_m1] and AHRR [Hs01005075_m1] and human and mouse 18S RNA [Hs03003631_g1] and CYP1B1 [Hs00164383_m1, Mm00487229_m1]; all Thermo Fisher Scientific) were pipetted into 96-well plates in triplicate for each sample and analyzed as previously described (11). Where 18S RNA was not detected in a sample, data were excluded.
Mice, vaccination, and analysis of spleens
C57BL/6J female mice were purchased from Jackson Laboratory and were used in accordance with Duke Institutional Animal Care and Use Committee–approved protocols. Mice were cohoused with littermates at a maximum of five animals per cage. For immunization, CFA and IFA (Invivogen) were diluted to 50% in PBS and complexed to chicken OVA protein (Sigma-Aldrich) for treatment of 8-wk-old female C57BL/6J mice with 50 μg of OVA protein in 200 μl of adjuvant/PBS solution or PBS. Mice were challenged with 50 μg OVA in 200 μl 1 wk after vaccination or treated with vehicle control. Mice were treated with 40 μg of SR1 diluted in 5% DMSO/95% olive oil or vehicle control 48 and 1 h before vaccination, and every 48 h after vaccination. Mice were euthanized 48 h after challenge, and spleens were harvested and stored in ice-cold PBS, crushed through 70-μm strainers in PBS and pelleted. The splenic cell pellet was reconstituted in RBC lysis buffer (Sigma-Aldrich), processed through 40-μm strainers, washed with PBS + 2% FBS, and spun down for flow staining. Single-cell suspensions were stained with CD45.2-BUV395, NK1.1-BV421, CD11c-BV510, F4/80-BV605, CD11b-BV711, CD19-FITC, CD3-FITC, 7-AAD, and either CD155-PE and PD-L1–allophycocyanin or isotype control-PE and isotype control-allophycocyanin as shown in Fig. 4A–D. Cells were gated on FSC-A and SSC-A for appropriate size, on FSC-A and FSC-height for single cells, live by negative for 7-AAD, CD45.2+ by CD45.2–allophycocyanin–Cy7-A+, non–T and non–B cells by CD19–FITC-ANeg and CD3–FITC-ANeg, non–NK cells by NK1.1-BV421-ANeg, CD11b+ by CD11b–BV711-A+, and for macrophages by SSC-A and F4/80-BV605+. Splenic cells shown in Fig. 4E were stained with Zombie Aqua-BV510 (1 h; BioLegend), blocked with mouse Fc block (30 min; BioLegend), and stained with CD45.2-BUV395, NK1.1-BV421, Ly6G-BV605, CD11b-BV711, CD19-FITC, CD3-FITC, F4/80–PE-Cy5, CD11c-allophycocyanin-Cy7, and either CD155-PE and PD-L1–BV786 or isotype control-PE and isotype control-BV786. Cells were gated for appropriately sized single cells, for live by Zombie Aqua-BV510Neg, CD45.2+ by CD45.2-BUV396-A+, non–T and non–B cells by CD19-FITC-A− and CD3-FITC-A−, non–NK cells by NK1.1-BV421-A−, nonneutrophils by Ly6G–BV605-A−, non–dendritic cells by CD11c–allophycocyanin–Cy7−, and for macrophages by CD11b–BV711-A+ and F4/80–PE–Cy5-A+.
Tumor implantation, treatment, dissociation, and flow cytometry
E0771 cells were injected into the fat pad of 8-wk-old female C57BL/6J mice. When tumors were >10 mm3, mice were treated with 40 μg of SR1 in 200 μl of olive oil i.p. or vehicle control every 48 h until euthanasia at 5 or 14 d after starting treatment. Tumors were harvested and stored in ice-cold PBS until processing. Tumors were minced and dissociated in RPMI 1640 containing 100 μg/ml Liberase-TM (Sigma-Aldrich) and 10 μg/ml DNAse I (Roche Diagnostics) for 20 min at 37°C with agitation. Dissociated cell suspensions were centrifuged (500 G × 3 min) and supernatant was retained for cytokine analysis. The pellet was reconstituted in FACS buffer and filtered through 70- and 40-μm strainers, washed, and reconstituted in PBS + 2% FBS for staining. Prior to staining, ∼10% of volume was separated, centrifuged, and lysed in TRIzol solution for RNA analysis. Single-cell suspension was stained with Zombie Aqua-BV510 (BioLegend) for 1 h, blocked with Fc block (BioLegend) for 1 h, and stained and gated as previously described for Fig. 4E. For T cell analysis, cell suspensions were stained with CD45.2-BUV395 (BD Bioscience), CD4-FITC, CD8-BV421, CTLA4-BV605, OX40-BV711, CD25-PE, TIGIT–PE–Texas Red, Foxp3–PE–Cy5, and CD3-allophycocyanin (all BioLegend). Cells were gated for appropriate size, single cells, live cells, and CD45.2 as described above. Cells were selected for CD3+ by CD3–allophycocyanin-A, for CD8+ by CD8–BV421-A+ and CD4–FITC-A– or T regulatory cells (TRegs) by CD4–FITC-A+ and CD8–BV421-A– followed by FoxP3–PE–Cy5-A+.
LEGENDplex (BioLegend) assays were used to measure cytokines with the Mouse Macrophage/Microglia, or Mouse Th Cytokine Panels on a BD Fortessa X-20 Flow Cytometer. LEGENDplex software was used to determine analyte concentrations per manufacturer’s instructions. Minimum threshold values were shown if cytokine concentrations were below sensitivity of detection (automatically determined by analysis software); in rare cases in which analyte concentration exceeded maximum threshold values, the maximum value was shown. Tumor homogenate cytokine analysis was performed as previously described (12).
Statistical analyses and The Cancer Genome Atlas analysis
Assay-specific statistical tests are indicated in the corresponding figures. GraphPad Prism 8 was used to perform all statistical analyses and plot data. GBM patient data were obtained from The Cancer Genome Atlas Research Network (https://www.cancer.gov/tcga) and were analyzed on cbioportal.org (13, 14) for Spearman correlation and p values modified for multiple comparison (q values).
“Classically” and “alternatively” activated macrophages have elevated levels of CD155 and PD-L1 expression
Reverberating with prior observations (15), analyses of World Health Organization grade IV malignant glioma (GBM) patient ex vivo tumor specimens revealed expression of CD155 and PD-L1 on TAMs (Fig. 1A). Indeed, PVR RNA expression was correlated with PD-L1 (CD274) and with multiple prominent myeloid cell markers in GBM (TCGA; Fig. 1B). Macrophages constitute a large percentage of tumor-associated myeloid cells in GBM (16), exhibiting phenotypes on a broad polarization spectrum (17). Conventionally, “classic” macrophage activation, which historically has been simulated in vitro upon LPS or IFN-γ stimulation, is contrarian to “alternative” activation simulated upon stimulation with IL-4 (18). Based on the historical record with these stimuli, we investigated CD155 and PD-L1 expression in response to LPS and IL-4 stimulation of macrophage-lineage cells. We found that both LPS and IL-4 induced CD155 (and PD-L1) within 24 h after treatment of primary human MDMs, and in phorbol ester–differentiated acute monocytic leukemia (THP-1) cells, indicating THP-1 cells are an appropriate model for our node of interest (Fig. 1C). IL-4 and LPS treatment elicited the expected, canonical signaling response with STAT6(Y641) and p38 MAPK phosphorylation in both cell types, respectively. Phorbol ester–differentiated U937 (histiocytic lymphoma) cells, commonly used for studies in cells of monocytic lineage, neither yielded PD-L1 expression at baseline or upon IL-4 or LPS stimulation, nor responded in a canonical manner to LPS [i.e., induction of p-p38 MAPK (19)]. However, U937 recapitulated IL-4–stimulated CD155 induction seen in MDMs and in THP-1 cells (Fig. 1C). Flow cytometry analyses of MDMs confirmed that increased CD155 expression, monitored by immunoblot in Fig. 1C, occurred at the cell surface (Fig. 1D).
Induction of CD155 expression on macrophages is STAT6 (after IL-4 treatment) and p38 MAPK (after LPS treatment) dependent
To decipher cell signaling cascades leading to CD155 and PD-L1 induction, we first performed detailed time course assays with IL-4 and LPS stimulation in our monocytic lineage panel (LPS stimulation was only tested in MDMs and THP-1 cells) (Fig. 2). IL-4 treatment produced canonical STAT6(Y641) phosphorylation within 30 min, followed by CD155 and PD-L1 induction with a 4–24 h delay (Fig. 2A). Transient, siRNA-mediated depletion of STAT6 in MDMs greatly diminished STAT6/p-STAT6(Y641) levels, and abrogated CD155 and PD-L1 induction after 24 h of stimulation with IL-4 (Fig. 2B). Quantification of CD155 mRNA by qRT-PCR after IL-4 treatment (16 h) showed induction at the transcript level (Fig. 2C). Thus, in aggregate, the time course of CD155/PD-L1 induction, dependency on STAT6, and the IL-4–induced increase of CD155 template, indicate a transcriptional response in line with the canonical signal transduction pathway induced by IL-4R activation (20).
CD155/PD-L1 induction upon LPS treatment mirrored the response to IL-4 (Fig. 2D). Canonical downstream TLR4 signals—p-p38 MAPK, p-JNK and p-NF-κB(S536)—were evident by 30 min, followed by delayed induction of CD155 and PD-L1 at 4–24 h (Fig. 2D). We also observed delayed (relative to p38/Jnk/NF-κB phosphorylation) activation of STAT1, evident as p-STAT1(Y701) accumulation by ∼4 h after LPS stimulation (Fig. 2C). P38 MAPK inhibition with SB203580, evident as blockade of p38-downstream MK2(T334) phosphorylation, demonstrated that LPS-induced CD155/PD-L1 in MDMs and THP-1 cells depends on p38 MAPK activation (Fig. 2E). Thus, distinct signaling networks influence CD155 and PD-L1 induction; these may reflect divergent regulatory pathways, or may involve a shared, convergent signaling nexus downstream of STAT6 and p38.
CD155 induction in response to both IL-4 and LPS is dependent on AhR
Because STAT6(Y641) and p38 were phosphorylated within 0.5 h of IL-4 or LPS stimulation, respectively, but CD155 expression levels only increased >4 h thereafter, PVR induction likely is a transcriptional response to activation of multiple signaling relays. Therefore, we submitted the PVR region upstream of the known PVR transcriptional start site (21) for putative transcription factor binding sites to JASPAR analysis. Fig. 3A depicts the promoter sequence upstream of the transcriptional start site in the PVR gene [GenBank reference standard (RefSeqGene) for PVR: NG_008781 (21)]. JASPAR analysis revealed the presence of a dioxin-responsive element [DRE; defined by the substitution-intolerant core 5′-GCGTG-3′ (22)] ∼310 nt upstream of the PVR transcriptional start site, indicating a possible AhR-responsive site (Fig. 3A). We used a previously established “position weight matrix” examining the putative influence of the DRE sequence context on AhR binding. This matrix is based on analyses of the DRE sequence context within the promoter regions of a set of confirmed, bona fide AhR targets (22). The PVR upstream region DRE’s position weight matrix is >0.85, above the threshold established for DREs in the promoter regions of bona fide AhR target genes (22). Additionally, both STAT6 (23) and p38 (24) signaling have been linked to the regulation of AhR; and AhR signaling has recently been connected to PD-L1 expression (25). AhR is prominently implicated in TAM polarization in GBM (26), in mediating microglial CNS inflammation (27) and in LPS-induced macrophage inflammation (28). Therefore, an involvement of AhR in immunomodulatory regulation of PVR in macrophages was plausible.
AhR, upon activation by xenobiotic (e.g., polycyclic aromatic hydrocarbons) or intrinsic (e.g., tryptophan metabolite) ligands, is freed from chaperone interactions for nuclear translocation and association with the AhR nuclear translocator. Primary products of AhR-transcriptional activation are cytochrome P450-dependent monooxygenases (CYPs), catalyzers of polycyclic aromatic hydrocarbon metabolism, and the negative feedback AhR repressor (AhRR). Once inside nuclei, AhR is prone to degradation; however, AhR binding to β-catenin yields mutual stabilization with sustained promoter activity (29). The AhR/β-catenin form is associated with an altered transcriptional target profile, centered on CYP1B1 (29).
To unravel a possible involvement of AhR in CD155 induction upon macrophage stimulation, we used various small molecule inhibitors of AhR activation (Fig. 3B–D) and of AhR/β-catenin interaction (Fig. 3E). SR1 is a synthetic heterocyclic compound shown to bind AhR and inhibit its transcription factor activity (30). We confirmed this: immunoblot analyses in fractionated cell lysates revealed that SR1 led to cytosolic retention/blocked nuclear accumulation of AhR (Fig. 3B). Accordingly, (AhR-transcriptional target) AhRR and CYP1B1 template expression was virtually abolished in SR1-treated THP-1 cells, either untreated or stimulated with IL-4 or LPS (Fig. 3C). SR1 prevented CD155 induction upon LPS and IL-4 stimulation in THP-1 cells, without disrupting canonical LPS- and IL-4–induced signaling (Fig. 3D). PD-L1 induction was only modestly affected by AhR inhibition with SR1 after IL-4 treatment (Fig. 3D; we did not identify a DRE in the CD274 promoter), possibly indicating redundant and/or independent mechanisms inducing PD-L1 after macrophage activation.
As LPS or IL-4 stimulation did not produce signs of induced AhR activity (e.g., increased AhR nuclear accumulation, increased AHRR or CYP1B1 mRNA levels; Fig. 3B, 3C), we investigated persistent AhR activity resulting from β-catenin binding. JW67, a synthetic compound identified in a screen for WNT signaling inhibitors, accelerates β-catenin degradation (31). Similar to SR1, JW67 prevented CD155 induction in LPS/IL-4 treated THP-1 cells without disrupting canonical LPS/IL-4 signaling (Fig. 3E). In a pattern mirroring the response to AhR inhibition with SR1, PD-L1 induction was only modestly responsive in THP-1 cells stimulated with IL-4, but not LPS (Fig. 3E). Furthermore, siRNA-mediated depletion of AhR prior to IL-4 treatment of THP-1 cells prevented CD155 induction (Fig. 3F). We confirmed AhR-dependent CD155 induction in IL-4/LPS stimulated MDMs by quantitating immunoblots after treatment with SR1 and 2 additional AhR inhibitors (GNF-351 and CH223191) (Fig. 3G). Our investigations, showing that CD155 induction is abrogated by the inhibition of AhR-transcriptional activity and of the AhR/β-catenin interface, indicate that persistently active AhR controls CD155 induction upon macrophage activation.
CD155 and PD-L1 are induced in Th1 and Th2 contexts in mouse splenic macrophages
Immune checkpoints generally exert their function within the context of immunological synapses, where their expression may also be influenced by T cell–secreted cytokines [e.g., IFNγ, IL-4 (32)]. Thus, we investigated if CD155 and PD-L1 induction on macrophages occurs in the context of adaptive immune responses in vivo. To this end, we vaccinated mice with synthetic OVA peptide mixed in CFA or IFA, respectively. By virtue of containing agonists for macrophage inducible Ca++-dependent lectin receptor (MINCLE), and for TLR-2, -4, and -9, CFA elicits Th1-type responses, whereas IFA induces Th2-type responses. Mice were rechallenged with OVA 1 wk after immunization and euthanized 48 h later. Their spleens were removed, and splenic macrophages were harvested for flow cytometry (Fig. 4). Baseline CD155 expression increased broadly after CFA or IFA vaccination and OVA rechallenge (Fig. 4A). PD-L1 expression was also induced by both immunizations but on a smaller proportion of macrophages than CD155 (Fig. 4B). Intriguingly, PD-L1 induction only occurred in combination with CD155 (Fig. 4C, 4D).
Mouse PVR [mPVR, also known as Tage4 (33)] encodes for murine CD155, which is only 42% identical to its human ortholog (34) [RefSeqGene for mPVR: NM_027514.2 (35)]. We performed JASPAR analysis of the mPVR region upstream of the known transcriptional start site, similar to our approach to human PVR (see above). As human PVR, the mPVR upstream region carries a substitution-intolerant core DRE (22) in a similar position (∼430 nt upstream of the transcriptional start site), suggesting AhR involvement in mPVR control. Lastly, we found a cluster of DREs in a similar position (∼620 nt upstream of the transcriptional start site) of the rat PVR gene (RefSeqGene for rat PVR: NM_017076.2) (Fig. 4E). Positional conservation of DREs within comparable distance to the transcriptional start site across homo, rattus and mus has been established as a predictor of AhR-binding activity in putative target genes (22).
Conserved gene regulatory activity of the upstream regions of human and murine PVR is evident in transgenic mice expressing (the human PVR) CD155 under control of either element. These mice displayed similar CD155 distribution patterns and pathogenic profiles upon poliovirus infection (36). To determine if in vivo induction of CD155 is AhR dependent, as suggested by our in vitro data and JASPAR analysis of the mPVR upstream region (Fig. 4E), we treated mice with SR1 48 and 1 h before IFA-OVA immunization, and 48 and 1 h before OVA challenge. Reverberating with our in vitro findings, AhR inhibition prevented CD155 induction. In SR1-treated mice challenged with IFA, the percentage of CD155+ macrophages remained at baseline (Fig. 4F). Also, SR1 treatment had no effect on PD-L1 induction in vivo (Fig. 4F). Thus, AhR signaling controls CD155 induction during immunologic responses in vivo.
AhR activity mediates tumor immunosuppression and is associated with transcriptomic markers of TAMs and survival of patients with GBM
Both CD155 (4) and AhR (37) are implicated in cancer immune suppression and progression. Therefore, we investigated the implications of AhR signaling in a syngeneic, immunocompetent murine breast cancer model (E0771). AhR inhibition with SR1 modestly impaired tumor growth (Fig. 5A), repressed (AhR-transcriptional target) CYP1B1 mRNA in tumors (Fig. 5A), decreased overall TAM levels (Fig. 5B), and diminished CD155 expression on TAMs (Fig. 5B). This was associated with a significantly elevated CD8/TReg ratio, an indication of more aggressive and less suppressive T cell phenotypes (Fig. 5C). CD25 (late) and OX40 (early) markers of CD8 T cell activation were elevated on intratumor CD8 T cells after SR1 treatment at 14 and 5 d, respectively (Fig. 5C). SR1 shifted intratumor cytokine profiles toward proinflammatory and Th1 phenotypes (Fig. 5D). Together, these findings confirm a role for AhR in controlling TAM CD155 expression, TAM density, and tumor-intrinsic inflammation. To test if CD155/PD-L1 expression is linked to AhR in human tumors, we compared markers of AhR expression and of myeloid cells using the TCGA GBM cohort (Fig. 5E). We observed strong correlations between markers of both induced (CYP1A1, AHRR) and persistent (CYP1B1) AhR activity with CD274, CD276, and PVR expression (Fig. 5E). Both CYP1B1 (Fig. 5E) and CD274 (Fig. 1B) were highly correlated with macrophage markers. PVR showed strong correlation to markers of AhR activity (Fig. 5E). To determine which of these factors, if any, dictate patient outcome, we divided samples into low (z-score < 0) and high (z-score > 0) expression for each gene shown (Fig. 1F) and compared cohorts for survival. We only found a significant survival correlation for CYP1B1, indicating that the sustained AhR activity associated with AhR/β-catenin correlates with an unfavorable prognosis in GBM patients (Fig. 5F). In aggregate, we uncovered a role for AhR in controlling CD155 expression on TAMs and defined a role for AhR in shaping the immunosuppressive tumor microenvironment (e.g., in GBM).
Targeting of CD155 [with recombinant poliovirus (38)] and its binding partner TIGIT (39) is being intensely pursued for cancer immunotherapy. Despite evidence for inhibitory and activating influences on immune effector cell subsets, depending on engagement with its various binding partners, recent insight indicates an immune suppressive role for CD155 in cancer (4). Indeed, CD155 is virtually universally expressed on neoplastic cells of solid cancers (2, 3, 40). Our work identifies persistently active AhR signaling as a mechanism controlling CD155 expression on macrophages in various activation contexts, including within the microenvironment of neoplastic lesions. Thus, AhR signaling represents a broader biological program exploited by tumors to institute immune subversion and tumor progression. CD155, AhR, and PD-L1 expression were associated with TAM markers in human gliomas, and AhR inhibition in vivo reduced TAM density in murine tumors. Together, these findings confirm that AhR signaling modulates suppressive TAM phenotypes (26), including via the CD155 immune checkpoint.
The observation that macrophage activation-induced PD-L1 only occurs in combination with induced CD155 indicates a generic role for both immune checkpoints in inflammation. Yet, nuances in the regulation of CD155 versus PD-L1 (e.g., AhR-independent PD-L1 induction by LPS), in addition to their engagement of distinct T cell/NK cell receptors, indicate varied contributions in a broader network of gene expression programs that fine-tune immune responses. Expression of multiple coinhibitory receptors by myeloid cells underscores a weakness of targeting single molecules for cancer immunotherapy.
AhR is best known for its role in recognizing xenobiotic chemicals and initiating transcriptional responses to metabolize them (41). However, AhR also assumes immune suppressive and stimulatory roles in innate and adaptive immunity (42, 43). Thus, fittingly, CD155 with competing immune suppressive versus stimulatory activities, is a transcriptional target of AhR. AhR control of the immune system is multifaceted and stretches well beyond CD155. For example, AhR signaling contributes to IDO expression, dendritic cell immunogenicity (44), TReg induction (45), and Th17 polarization (42). Thus, the observed effects of AhR inhibition on the tumor microenvironment are due to events beyond restricted CD155 expression. We propose that CD155 belongs to a broader AhR-directed gene expression program that, in concert with other effector molecules, sculpts immunological responses.
AhR inhibition only had modest tumor-suppressive effects, despite an increased CD8/TReg ratio, lower TAM density, downregulation of CD155 expression on TAMs, and increased intratumor inflammatory cytokine signatures. Therefore, although AhR inhibition may be therapeutically attractive, our findings suggest it is insufficient to mediate tumor regression on its own. Moreover, antitumor effects of AhR inhibition have previously been documented elsewhere (26). AhR inhibition did not restrict PD-L1 expression on macrophages in vivo, possibly indicating that combination with PD1/PD-L1 blockade may be necessary to enhance antitumor effects of AhR inhibition. Indeed, combining (mouse ortholog) CD155 depletion with immune checkpoint blockade mediated antitumor efficacy in an immunocompetent mouse tumor model (4).
In conclusion, persistently active AhR signaling controls expression of the immune checkpoint CD155 in macrophages. Thus, AhR inhibition may provide a therapeutically tractable approach to target CD155 in cancer. Our work further supports mounting evidence for an AhR-driven immune axis that determines immune homeostasis, immune cell differentiation/activation, and function and places CD155 as an effector molecule within this signaling network.
We thank the Duke Preston Robert Tisch Brain Tumor Center BioRepository for providing explant GBM tumor tissue, Greg Palmer for providing E0771 cells, and Elena Dobrikova, Mikhail Dobrikov, Jonathan Kastan, and Wafa Hassen for insightful comments and technical support.
This work was supported by Public Health Service National Institute of Neurological Disorders and Stroke Grant R01 NS108773 (to M.G.), Kirschstein National Research Service National Cancer Institute Award F32 CA224593 (to M.C.B.), and a National Cancer Center Breast Cancer Fellowship (to M.C.B.).
Abbreviations used in this article:
aryl hydrocarbon receptor
forward scatter area
quantitative real-time PCR
small interfering RNA
side scatter area
T regulatory cell.
M.G. holds equity in, is an advisor and compensated consultant of, and is an inventor of intellectual property licensed to Istari Oncology, Inc. M.C.B. is an inventor of intellectual property licensed to Istari Oncology, Inc. The other author has no financial conflicts of interest.