Chemerin (retinoic acid receptor responder 2, RARRES2) is an endogenous leukocyte chemoattractant that recruits innate immune cells through its receptor, ChemR23. RARRES2 is widely expressed in nonhematopoietic tissues and often downregulated across multiple tumor types compared with normal tissue. Recent studies show that augmenting chemerin in the tumor microenvironment significantly suppresses tumor growth, in part, by immune effector cells recruitment. However, as tumor cells express functional chemokine/chemoattractant receptors that impact their phenotype, we hypothesized that chemerin may have additional, tumor-intrinsic effects.
We investigated the effect of exogenous chemerin on human prostate and sarcoma tumor lines. Key signaling pathway components were elucidated using qPCR, Western blotting, siRNA knockdown, and specific inhibitors. Functional consequences of chemerin treatment were evaluated using in vitro and in vivo studies.
We show for the first time that human tumors exposed to exogenous chemerin significantly upregulate PTEN expression/activity, and concomitantly suppress programmed death ligand-1 (PD-L1) expression. CMKLR1 knockdown abrogated chemerin-induced PTEN and PD-L1 modulation, exposing a novel CMKLR1/PTEN/PD-L1 signaling cascade. Targeted inhibitors suggested signaling was occurring through the PI3K/AKT/mTOR pathway. Chemerin treatment significantly reduced tumor migration, while significantly increasing T-cell–mediated cytotoxicity. Chemerin treatment was as effective as both PD-L1 knockdown and the anti–PD-L1 antibody, atezolizumab, in augmenting T-cell–mediated tumor lysis. Forced expression of chemerin in human DU145 tumors significantly suppressed in vivo tumor growth, and significantly increased PTEN and decreased PD-L1 expression.
Collectively, our data show a novel link between chemerin, PTEN, and PD-L1 in human tumor lines, which may have a role in improving T-cell–mediated immunotherapies.
Loss of the tumor suppressor PTEN in human cancers has recently been shown to contribute to resistance to immunotherapy; unfortunately, therapeutic reactivation of PTEN has remained elusive. Chemerin (retinoic acid receptor responder 2, RARRES2) is a leukocyte chemoattractant known to recruit effector immune cells, and is often downregulated in tumors. Recent data links chemerin to PTEN expression, and thus we hypothesized that chemerin may act to augment PTEN and result in improved responses to immunotherapy. Herein, we describe a novel pathway in human tumors whereby chemerin, through its G protein–coupled receptor CMKRL1, induces PTEN expression and activity, while concurrently suppresses programmed death ligand-1 (PD-L1) expression. We show that chemerin treatment significantly inhibits tumor migration/invasion, increases T-cell–mediated cytotoxicity, and suppresses in vivo tumor growth. Taken together, these results identify chemerin as a promising clinical therapeutic able to reactivate PTEN and suppress PD-L1 expression, thus potentially improving responses to immunotherapy.
Chemerin, or RARRES2 (retinoic acid receptor responder 2), is an endogenous leukocyte chemoattractant but has myriad roles in adipogenesis, metabolism, angiogenesis, microbial defense, and cancer. Chemerin is widely expressed in nonhematopoietic tissues, with low/no expression noted in leukocytes (1). Chemerin recruits innate immune cells along its concentration gradient to sites of inflammation via its G protein–coupled receptor (GPCR) chemokine-like receptor-1 (CMKLR1, also known as ChemR23; refs. 2, 3). In humans, CMKLR1 expression on leukocytes has been shown in macrophages, dendritic cells (DCs), and natural killer (NK) cells with comparable expression in the mouse (3–6). While data are limited, CMKLR1 expression has been detected on human tumor cells (7, 8), suggesting that interaction with its endogenous ligand chemerin may modulate tumor cell phenotype, as seen with other chemokine/receptor pairs (9).
Chemerin/RARRES2 is commonly downregulated across several tumor types, including melanoma, breast, prostate, and sarcoma, compared with their normal tissue counterparts (1). Our group was the first to show that forcible reexpression of chemerin in the tumor microenvironment (TME) resulted in recruitment and increased tumor-infiltrating effector leukocytes, leading to a significant reduction in the growth of aggressive B16 melanoma in a mouse model (10). While recruitment of immune effector cells is important, tumor cell–intrinsic oncogenic signaling pathways can also impact directed immune responses, and thus play a key role in determining therapeutic efficacy.
PTEN (PTEN deleted on chromosome 10) is a critical tumor suppressor whose expression is downregulated and/or lost in many tumor types (11). PTEN loss has been correlated with activation of the PI3K–AKT pathway, which is implicated in the pathogenesis of these cancers, and is particularly relevant in prostate cancer (12). Deleterious PTEN alterations are found in up to approximately 20%–30% of primary prostate cancer tissues and in approximately 40%–60% of metastatic tissues, and are among the most common genomic events in prostate cancer (13). While less commonly mutated in sarcoma, PTEN downregulation has also been shown to play an important role in a subset of soft-tissue sarcomas (STS), with one study showing 57% of STSs with decreased PTEN expression (14). Furthermore, aberrations in the downstream PI3K/Akt pathway are almost always implicated in the pathogenesis of sarcomas, with essentially 100% of advanced-stage osteosarcomas showing dysregulation in this pathway (15).
Here, we examine the effects of chemerin on tumor cell–intrinsic phenotype and describe, for the first time, the ability of chemerin to upregulate the expression and function of PTEN in human prostate and sarcoma tumor cell lines. Importantly, we show, also for the first time, that chemerin treatment of tumor cells results in a concomitant downregulation of programmed death ligand-1 (PD-L1) expression, which directly translates into significantly increased T-cell–mediated cytotoxicity. These effects were dependent on CMKLR1, as siRNA knockdown and specific inhibition with the CMKLR1 antagonist, α-NETA, completely abrogated these effects. In vivo studies using the human DU145 prostate tumor line show that expression of chemerin in the TME significantly suppresses tumor growth, increasing tumor PTEN and decreasing tumor PD-L1 expression compared with controls. Collectively, these studies show that in addition to recruitment of effector leukocytes into the TME, chemerin can also upregulate PTEN expression/function and suppress PD-L1 expression, suppressing in vivo tumor growth, and potentially rendering tumor cells more susceptible to T-cell–mediated immunotherapies.
Materials and Methods
Cell culture and reagents
Cell lines were obtained from ATCC between the years of 2015 and 2018. For experiments, each cell line was used between passage 4 and 12. Cell line authentication was verified by ATCC through PCR, karyotyping, and morphology-based techniques to confirm the tumor line status prior to use. DU145 (human prostate cancer, HTB-81) and PC3 (human prostate cancer, CRL-1435) cells were cultured using RPMI1640 complete media. SKES-1 (human Ewing sarcoma, HTB-86) and U2-OS (human osteosarcoma, HTB-96) cells were cultured with McCoy's 5A media. Cell lines were tested for Mycoplasma every 2–4 weeks, depending on the rate of usage, using the MycoProbe Mycoplasma Detection Kit (CUL001B, R&D Systems). Recombinant human chemerin (2325-CM, R&D Systems) was added at specified concentrations for 48 hours. Cell lines were incubated with vehicle control (captisol) versus α-NETA (10 μmol/L) for up to 24 hours with either PBS or 6 nmol/L chemerin, each reagent was replaced in fresh media every 24 hours. α-NETA (10 μmol/L, Selleck Chemical) or CMKLR1 blocking peptide (5 μmol/L, sc-374570 P, Santa Cruz Biotechnology) was used to block CMKLR1. Everolimus (RAD001, Sigma, mTOR) and CCG-1423 (Cayman Chemical, RhoA/SRF) were used as inhibitors. mTOR inhibitor (Everolimus, Selleck Chemical, 200 nmol/L) was used for complete inhibition for 24 hours. PI3k inhibitor (BEZ235, Selleck Chemical, 100 nmol/L) was used for 24 hours pretreatment.
X-tremegene siRNA Transfection Reagent (No. 4476093001, Roche) and each siRNA was added drop-wise to the cell media. siRNAs (CMKLR1, PTEN, and PD-L1) were all 10-μmol/L stock concentration. The optimal ratio of transfection reagent to siRNA (4:10) gave a final concentration of 40 pmol/L, and signal knockdown was evaluated via Western blot analysis (Supplementary Figs. S1 and S8). ChemR23/CMKLR1 (sc-44633, Santa Cruz Biotechnology), PD-L1 (sc-39699, Santa Cruz Biotechnology), PTEN (6251S, Cell Signaling Technology), and Control (sc-37007, Santa Cruz Biotechnology) siRNA were used for signal knockdown. The control siRNA-A is a nonspecific scrambled sequence used as a negative control in the siRNA-targeted knockdown experiments.
Cells were stained with the target-specific antibody (Supplementary Table S1) as labeled in each figure at 1 μL/1 × 105 cells for 30 minutes at 4°C. Cells were analyzed using a FACSCalibur (BD Biosciences).
Sample RNA was isolated using TRizol (Invitrogen) and RNeasy Mini RNA Isolation Kit (Qiagen). RNA concentrations were verified using NanoDrop 2000 (Thermo Fisher Scientific). Bio-Rad iScript Advanced cDNA Synthesis Kit converted RNA to cDNA via the manufacturer's protocol. cDNA was amplified with iTaq Universal SYBR Green Supermix (Bio-Rad) via the manufacturer's protocol. A CFX96 Real-Time PCR System (Bio-Rad) was used to quantify gene expression via the 2ΔΔCt analysis method. Primer sequences were developed using Primer-Blast software (https://www.ncbi.nlm.nih.gov/tools/primer-blast/). Each sample result was normalized to its respective GAPDH loading control. See Supplementary Table S2 for primer sequences.
RIPA Lysis Buffer (protease/phosphatase inhibitor cocktail, Thermo Fisher Scientific) was used to lyse cells after experiment. Protein concentration was calculated using Pierce BCA Protein Assay (Thermo Fisher Scientific) via the manufacturer's protocol. Bolt 4%–12% Bis-Tris SDS-PAGE Gels (Invitrogen) were loaded with equal sample protein amounts (50 μg/sample). Gels were transferred to NitroBind Nitrocellulose Membrane (Thermo Fisher Scientific). Blots were illuminated using Thermo SuperSignal West Dura per the manufacturer's protocol. Imaging was done using Bio-Rad Molecular Imager ChemiDoc XRS+ System and quantified using the Bio-Rad Image Analysis Software.
PTEN phosphatase activity
Protein was processed as described above using immunoprecipitation Lysis Buffer (Thermo Fisher Scientific). Samples were normalized (200 μg/sample) and anti-human PTEN (138G6, Cell Signaling Technology, 1:250) was added for PTEN immunoprecipitation. To initiate activity, 3 pmol/L phosphatidylinositol-3,4,5-triphosphate (PIP3, Echelon Biosciences, DiC8) was added to each PTEN immunoprecipitation protein sample (200 μg/sample) for 2 hours at 37°C. To measure free phosphate, the Malachite Green Phosphate Detection Kit (12776, Cell Signaling Technology) was followed via the manufacturer's protocol.
Tumor migration/invasion assay
A 24-well plate transwell inserts (6.5-mm, Costar, 8-μm pores) were precoated with 35 μL of 1 mg/mL Matrigel (BD Biosciences) at 37°C for 2 hours. A total of 0.5 × 105 cells of each sample in serum-free medium were plated in the top chamber and media (10% FBS) were added to the bottom well. After 24 hours, the inserts were fixed and stained with 0.1% crystal violet for imaging before being lysed with 10% acetic acid. Absorbances were measured correlating to the number of migrated cells per insert (BioTek Instruments).
Human T cells were isolated from donor peripheral blood mononuclear cells (PBMC) using MojoSort Human CD3 Isolation Kit (catalog No. 480022, BioLegend) via the manufacturer's protocol. T cells were left untreated (naïve T cells) or treated with IL2 + ImmunoCult CD3/CD28/CD2 T-cell tetramers (activated T cells, 25 μL/mL, No. 10970, Stemcell Technologies). Trypsinized tumor cells were counted and stained with carboxyfluorescein diacetate succinimidyl ester (CFSE, 1 μL/mL, No. 423801, BioLegend) via the manufacturer's recommendation. CFSE+ target tumor cells were incubated with naïve (untouched) or activated T cells overnight (∼18 hours) at indicated effector to target (E:T) ratios (typically 3:1). Samples were stained with 7-AAD (5 μL/1 × 106 cells, No. 420404, BioLegend) to identify dead cells. Percent lysed was the fraction of cells that stained positive for both CFSE and 7-AAD. As donor T cells and target tumor cells were not HLA-matched (and thus measured alloreactivity), anti-MHCI, anti-human HLA-ABC (311402, clone W6/32, BioLegend) was used for additional control experiments.
RNA ISH and image analysis
Manual chromogenic RNAScope was performed with RNAScope 2.5 HD Reagent Kit–Brown (ACD, No. 322310), using the optimized manufacturer's protocols. Single ISH detection for PTEN (ACD Probe: 408511), PD-L1 (CD274, ACD Probe: 600861), Positive Control Probe (PPIB - ACD Probe: 313901), and Negative Control Probe (Dapb - ACD Probe: 310043) were performed via the manufacturer's protocols. Three comparable regions of interest (ROI) for each respective sample set were analyzed using HALO Software (three ROIs per sample, repeated for n = 3 independent experiments).
In vivo studies
All mice used in experiments were purchased from The Jackson Laboratory. NOD/SCID/IL2R gamma (null) (NSG; No. 005557, NOD.Cg-PrkdcscidIl2rgtm1Wjl/SzJ) male mice were used at approximately 9–10 weeks of age, as indicated. Mice were maintained in the Washington University facilities (St. Louis, MO) under the direction and guidelines of the Division of Comparative Medicine. All animal experiments were conducted in accordance with approved Washington University (St. Louis, MO) and NIH Institutional Animal Care and Use Committee guidelines under an approved protocol (No. 20170174). To evaluate the effect of constitutive chemerin secretion on in vivo tumor growth, vector control or chemerin-expressing DU145 tumor cells (2.5 × 106) were inoculated subcutaneously into 9- to 10-week-old male mice. The pcDNA3.1+ (Thermo Fisher Scientific) vector was used to produce either vector control or human RARRES2–transfected DU145 cells. Transfected cells were selected using geneticin (G418). Prior to inoculation, DU145 lines were grown to approximately 70%–80% confluence to ensure log-growth kinetics, and cell viability was assessed using trypan blue and cells were used only if approximately 95% viable. Tumor growth was measured every 3–4 days by calipers, and size was expressed as the volume product of perpendicular length by width in square millimeters. Mice were euthanized when tumor size reached approximately 400 mm2 or at indicated timepoints for downstream analyses.
Primary prostate tumor processing
Within 1 hour of resection, primary tissue was processed into single-cell suspensions. To digest the tissue, prostate tissue or metastatic biopsy cores were cut into 1 × 1 mm pieces and incubated with 100 μL of Liberase TL Solution (28 U/mL, Roche Applied Science), and DNase I (20 U/mL, Thermo Fisher Scientific) was added and samples were continuously rotated and incubated at 37°C for 1 hour. Digested cell suspensions were then homogenized by using a 1,000-μL-wide bore pipette tip and samples were passed through a 100-μm strainer. Following processing, cells were ready for use in investigative studies and downstream analysis. All human subjects were consented under the approved institutional review board (IRB) protocol (No. 201411135) titled Tissue, Blood, and Urine Acquisition for Genomic Analysis and Collection of Health Information for Patients with Malignancies of the Genitourinary Tract.
All experiments were done independently (n = 3 or more). Each time, sample replicates were prepared and analyzed independently. Means and SEM were calculated. Paired Student t tests were used for comparison between two groups in each experiment. One-way ANOVA was used to compare more than two groups, including a post hoc Tukey test to confirm differences between groups. A P value of less than 0.05 was considered statistically significant via Microsoft Excel and GraphPad Prism v.8 software.
Chemerin exposure can induce PTEN expression in tumor cell lines
We initially questioned whether chemerin, given its myriad roles, would have an impact on tumor-intrinsic cell functions. Previous studies in mouse models have not shown detectable levels of CMKLR1 in mouse tumor cell lines, nor direct effect of recombinant chemerin exposure on tumor cell phenotype measured (10). Given the prominent role of PTEN dysregulation in prostate and sarcoma tumors, we decided to study these tumor types using human cell lines. Analysis of prostate and sarcoma The Cancer Genome Atlas (TCGA) data shows that patients with higher levels of RARRES2 in their tumors have improved overall survival compared with those with lower expression (Fig. 1A and B), in-line with our and others' analyses in other tumor types (16). We looked at human tumor lines that had detectable CMKLR1 protein expression and genetically intact PTEN (DU145, U2OS, and SKES) and used a CMKLR1+, PTEN-null (−/−) cell line (PC3) as a control. Both prostate and sarcoma cell lines were analyzed for expression of CMKLR1, and showed detectable levels of CMKLR1 protein at both the intracellular and cell surface levels (Supplementary Fig. S1). Cell lines had no detectable chemerin expression using anti-human chemerin ELISA assays (Supplementary Fig. S2F and data not shown). We then investigated the effect of exogenous, recombinant chemerin on these cell lines. Chemerin is found systemically in plasma and most nonhematopoietic tissues, and engages CMKLR1 at low nanomolar concentrations (2); thus, we chose to initially focus in this range of concentrations. Cell lines were plated as indicated with complete media containing 6 nmol/L recombinant chemerin protein or PBS (the diluent control) for 48 hours. Following treatment, we found PTEN mRNA expression was significantly upregulated over control-treated cells in PTEN wild-type (WT) cell lines tested (Fig. 1C and D). As expected, there was no detection of PTEN in the PC3 cells, while we saw an approximately twofold increase in PTEN expression in the DU145 cells, and an approximately 2.5-fold increase in the sarcoma SKES and U2OS cells after incubation with chemerin.
As mRNA expression does not perfectly correlate with protein production (17), we then investigated protein expression after chemerin treatment. Western blot analyses showed a noticeable upregulation of PTEN protein expression in all three cell lines after a 48-hour incubation (Fig. 1E and F). Quantification showed that PTEN expression increased with an increasing concentration of chemerin, suggesting a dose response. PTEN protein was increased approximately 1.5-fold in DU145 cells, and approximately 1.6– to 1.7-fold in U2OS and SKES cells compared with the controls (Fig. 1G–I). Neither in vitro cell proliferation nor apoptosis (Supplementary Figs. S2 and S3) was significantly altered after chemerin exposure over a 72-hour period, compared with controls. Collectively, these results show that exogenous chemerin significantly induces PTEN mRNA and protein expression in a dose-dependent manner, without significant impact on their in vitro proliferation or apoptosis.
Chemerin treatment, mediated by CMKLR1, significantly reduces tumor migration and invasion
PTEN has multiple roles in tumor suppression, including inhibition of tumor cell proliferation, invasion, and migration (18, 19). While chemerin did not detectably impact cell proliferation or apoptosis, we hypothesized it might affect other aspects of tumor cell phenotype. Using a tumor migration model, we examined the effects of chemerin treatment for 48 hours, in line with our previous experimental conditions. Control or chemerin-treated cells were allowed to migrate on Matrigel for 24 hours. Imaging showed a noticeable decrease in tumor migration/invasion in all three of the cell lines treated with chemerin compared with control cells (Fig. 1J). Quantification of migrated tumor cells shows that chemerin treatment significantly reduced tumor cell invasion by 29% in DU145 cells, 31% in U2OS cells, and 22% in SKES cells compared with control-treated samples, respectively (Fig. 1K).
Subsequently, we assessed whether the effect of chemerin treatment on reducing tumor migration/invasion was mediated through its binding to CMKLR1, and not “off-target” effects. CMKLR1 siRNA knockdown experiments were performed as described above, using the Matrigel invasion assay. Knockdown of CMKLR1 protein using siRNA was confirmed at both the intracellular and cell surface levels (Fig. 2A; Supplementary Fig. S1). Mock transfection and control siRNA cells continued to display significantly reduced tumor cell invasion after chemerin treatment, compared with control-treated tumor cells. In general, CMKLR1 siRNA knockdown decreased overall cell invasion in both control- and chemerin-treated groups compared with mock transfection or control siRNA groups. However, CMKLR1 knockdown completely abolished the ability of chemerin to inhibit tumor cell migration compared with control-treated cells in all three lines (Supplementary Fig. S4). Collectively, these studies show a significant functional impact of chemerin treatment on tumor cell lines, and may represent a way to reduce tumor metastatic potential in vivo.
CMKLR1 mediates chemerin-induced PTEN expression and function
Given the two other known receptors for chemerin (CCRL2 and GPR1) are either nonsignaling or have limited tissue expression (20), we focused on the role of CMKLR1, chemerin's main chemotactic receptor, in mediating chemerin-induced PTEN upregulation in these cell lines. Compared with controls, significant increases in PTEN expression, by both qPCR and Western blot analysis, were seen with exposure to 6 nmol/L chemerin during mock and control siRNA transfections in all cell lines. However, only with CMKLR1 siRNA was there a complete abrogation of chemerin-induced PTEN expression (Fig. 2B). This establishes the role of CMKLR1 in mediating the chemerin-induced increase of PTEN expression, at both the mRNA and protein levels.
While PTEN increased because of chemerin in all cell lines, we tested whether the augmented PTEN was indeed functional. PTEN phosphatase activity modulates PI3K-induced PIP3, which is a critical factor in mediating subsequent signaling pathways involved in cell survival, proliferation, and migration (21). We studied the ability of PTEN protein to dephosphorylate PIP3 phosphate, following 48 hours PBS or chemerin incubation. Protein lysates were collected following each 48-hour experiment for PTEN immunoprecipitation. PTEN phosphatase activity was significantly increased after 48-hour chemerin exposure compared with control-treated cells (Fig. 2C), suggesting the chemerin-induced PTEN retained its ability to function as a phosphatase. Likewise, specific CMKLR1 knockdown, but not mock transfection or control siRNA, completely abrogated the chemerin-mediated increase in PTEN phosphatase activity (Fig. 2C). Together, these findings support a role for chemerin to induce significant functional PTEN expression via CMKLR1 in human tumor lines.
Chemerin treatment results in an increase in the transcription factors serum response factor and EGR-1, and correlates with PTEN upregulation
To further elucidate this novel pathway, we investigated the underlying mechanisms of the chemerin–PTEN interaction. A recent study showed that chemerin binding to CMKLR1 leads to transcriptional activation of the serum response factor (SRF; ref. 22). SRF expression has been shown to modulate PTEN expression, as well as induce activation of its target gene EGR-1 (early growth response 1), which directly regulates PTEN expression (23, 24). Aberrant PI3K pathway activation leads to a decrease in SRF levels and results in reduced binding to the EGR-1 promoter necessary for EGR-1 transcription (25). Thus, we hypothesized these components may mediate signaling between chemerin and PTEN, via CMKLR1, in DU145 cells. Therefore, we examined both SRF and EGR-1 expression in chemerin-treated DU145 cells, as previously described above. Concomitant with upregulated PTEN expression, our RT-qPCR results showed significant increases for both SRF (1.75-fold) and EGR-1 (1.91-fold) mRNA expression in the chemerin-treated DU145 cells compared with PBS alone (Fig. 3A). Similarly, Western blot analysis showed a significant increase in SRF and EGR-1 protein expression, 1.68-fold and 1.57-fold, respectively, directly correlating with increased PTEN protein expression (1.67-fold; Fig. 3A). Taken together, these results suggest chemerin-binding CMKLR1 induces increased SRF and EGR-1 expression upstream of augmented PTEN expression and activity in DU145 cells.
Chemerin suppresses pAKT and pS6 expression, correlating with decreased PD-L1 expression
Next, we set out to further characterize the relationship between chemerin, PTEN, and PD-L1. We examined the protein levels of p-Akt (ser473) and pS6 (ser235/236) by Western blotting. PTEN negatively regulates the PI3K/Akt pathway and overall PTEN activation inversely correlates with p-Akt expression (12, 15, 26, 27). Furthermore, PTEN loss or PI3K genetic alterations in prostate, breast, or glioma tumors result in significantly augmented PD-L1 expression (28). Importantly, Lastwika and colleagues showed that PD-L1 expression is tightly regulated by the Akt–mTOR pathway, where activation can lead to immune escape for some tumor types (29). Inhibition studies targeting mTORC1 (pAKT) and mTORC2 (pS6) within the PI3k/Akt/mTOR pathway confirmed their role in control of PD-L1 expression (29), thus we studied these key signaling constituents to further investigate chemerin's downstream impact on PD-L1. Figure 3B shows a significant decrease in p-Akt (ser473) protein expression following chemerin incubation, compared with the PBS treatment. Western blot analysis data from four independent sample sets show a 29% decrease in pAkt (ser473) protein expression in chemerin-treated cells compared with the control PBS group. We also show that chemerin treatment leads to a significant decrease in both phospho-S6 (pS6 ser235/236; 43% decrease) and PD-L1 protein expression (32% decrease) compared with the control PBS group (Fig. 3B). Thus, our experimental results show that chemerin exposure increases PTEN expression leading to a subsequent negative regulation of the Akt–mTOR–PD-L1 signaling cascade. These results are consistent with previous studies looking at the effects of augmented PTEN expression on the PI3k/Akt/mTOR pathway and its signaling constituents (12, 15, 26, 27, 30, 31). To evaluate the effects of chemerin treatment on tumor PI3K/AKT/mTOR pathway components, we used two well-studied inhibitors of PI3K/mTOR (BEZ235) and mTOR (RAD001). PI3K inhibition by BEZ235 had no effect on the increase in tumor PTEN expression seen with chemerin treatment (Fig. 3C), but did significantly reduce the expression of downstream pAKT/AKT, pS6/S6, and PD-L1; these reductions seen with this PI3K/mTOR inhibitor (also known as dactolisib) were not significantly different than those seen with chemerin treatment in the DU145 tumor cells (Fig. 3C). Treatment with the mTOR (mTORC1/2) inhibitor RAD001 suppressed PD-L1 and pS6 expression, as expected (refs. 29, 32; Fig. 3D; Supplementary Fig. S5), in both control and chemerin-treated tumor cells. Suppression of tumor PD-L1 was comparable and not statistically different between chemerin and RAD001 treatment. Following treatment, pS6 protein expression was completely knocked out, as expected (Supplementary Fig. S5). RAD001 treatment completely abrogated the decrease in PD-L1 seen in the DU145 cells after chemerin exposure (Fig. 3D), implicating mTOR as a critical factor in this signaling pathway.
We next used the RhoA/SRF pathway inhibitor, CCG-1423, given CMKLR1 has been shown to signal through RhoA/SRF (22), and again found that both chemerin-induced increases in PTEN (top) and decreases in PD-L1 (bottom) expression were completely abrogated with use of the CCG-1423 inhibitor (Fig. 3D). As siRNA can have off target effects, we used a specific CMKLR1 small-molecule antagonist, α-NETA, that recapitulates a CMKLR1-knockout phenotype (33). We again looked at PTEN/PI3K/mTOR pathway components and found that treatment with α-NETA completely abrogated effects seen with chemerin treatment (Fig. 3E), suggesting chemerin was signaling through CMKLR1 and mediating these effects. A CMKLR1-blocking peptide also showed similar abrogation of PTEN and PD-L1 changes induced by chemerin (Supplementary Fig. S5), suggesting our results seen with CMKLR1 siRNA were unlikely due to off-target effects. Taken together, these data suggest that chemerin treatment results in an increase in tumor PTEN expression, with associated changes in the canonical PTEN–PI3K–AKT–mTOR signaling pathway as well as decreases in pS6/S6 and PD-L1 expression, comparable with PI3K and mTOR inhibitors that have been used in clinical trials.
Chemerin upregulates PTEN and concomitantly decreases PD-L1 expression on tumor cells, via CMKLR1
Recent evidence correlates PTEN expression and function to PD-L1 expression in cancer, and this has been shown to be dependent on the PI3K pathway and S6 kinase (S6K) activation (28, 34–36). Thus, we decided to further study PD-L1 expression in the context of chemerin exposure and PTEN expression. We tested a wide range (3–62 nmol/L) of recombinant chemerin concentrations on DU145 tumor cells, and then assessed for both PTEN and PD-L1 expression. Again, 6 nmol/L chemerin treatment produced the most robust increase in PTEN expression, with an obvious dose–response relationship seen. Importantly, we also saw a concomitant, significant decrease in PD-L1 mRNA expression via qPCR (Fig. 4A).
To further elucidate, we looked at RNA ISH staining for PTEN and PD-L1 using RNA-specific probes (ACDBio). Image analysis showed chemerin significantly upregulated PTEN and simultaneously decreased PD-L1 RNA expression (Fig. 4B). Furthermore, we investigated the role of CMKLR1 in the chemerin-mediated suppression of PD-L1 expression. We found that only knockdown of CMKLR1, and neither mock transfection nor control siRNA, completely abrogated both the significant increase in PTEN and decrease in PD-L1 expression seen following chemerin treatment (Figs. 2B and 4C). In addition, we evaluated the impact of chemerin on tumor cell surface expressed PD-L1 protein, as this ultimately mediates its immunosuppressive effects. FACS staining analyses for PD-L1 revealed a significant decrease in surface expression in the chemerin-treated tumor cells compared with controls by both percent positive (based on isotype control) as well as mean fluorescence intensity (MFI; Fig. 4D). Furthermore, chemerin treatment significantly reduced IFNγ-induced PD-L1 expression compared with PBS-treated DU145 cells (Fig. 4E), suggesting chemerin may blunt the induction of PD-L1 expression in the setting of increased IFNγ that can occur with some immunotherapies.
Collectively, these data confirm a key inverse relationship between the tumor suppressor, PTEN, and a key immune checkpoint inhibitor, PD-L1 (28, 34–36). More importantly, our findings show, for the first time, that chemerin can directly modulate this established PTEN/PD-L1 axis via CMKLR1 in human tumor cells.
Chemerin treatment, mediated through CMKLR1, significantly improves T-cell–mediated cytotoxicity of tumor cells
Our findings that chemerin treatment of tumor cells suppresses PD-L1 expression suggest that it could play a role in T-cell–mediated cytotoxicity. PD-L1 is known to inhibit T-cell function via its interaction with programmed cell death-1 (PD-1) on T cells (37). To investigate, we isolated human T cells from donor PBMCs to target DU145 and U2OS cells. We found that unstimulated, naïve T cells were only able to mediate low levels of DU145 cytotoxicity, as published previously (38). This is not surprising, as our effector T cells were donor-derived and not HLA-matched, thus measuring T-cell alloreactivity to tumors. Both activation (CD2/CD3/CD28 tetramers) and increasing E:T ratio improved tumor cell killing (Supplementary Fig. S6), generally in-line with other studies (39–41). Importantly, we found that chemerin treatment of DU145 tumor cells resulted in a significant increase in activated T-cell–mediated cytotoxicity, compared with controls (Fig. 5A). Treatment of tumor cells with an anti-HLA antibody abrogated the effect of chemerin on augmenting T-cell–mediated lysis, suggesting an MHC-dependent mechanism. Treatment of tumor cells did not change MHCI surface expression in the presence or absence of chemerin (Supplementary Fig. S6).
The increase in cytotoxicity seen with chemerin treatment, however, was only seen at lower E:T ratios (i.e., 0.5:1 to 3:1), while the effect seemed to lessen at higher E:T ratios (Supplementary Fig. S6). This suggests that higher numbers of activated T cells per tumor target cell may act to obscure the effect mediated by chemerin treatment. It is important to note that prostate cancers typically are less infiltrated with immune cells (especially T cells) compared with most other tumor types (42), suggesting the lower E:T ratios used in our assays may, in fact, be more physiologically relevant to the human TME. Furthermore, prostate tumor–infiltrating T cells show an exhausted phenotype, as evidenced by high PD-1 expression and decreased IFNγ (43), similar to T cells used in our assays (Supplementary Fig. S7). Thus, PD-L1 expression on prostate tumor cells is likely to modulate prostate-infiltrating T-cell function; indeed, recent clinical data show that blocking the PD-1/PD-L1 pathway in patients with metastatic colorectal cancer led to disease control rates of up to 22% (44).
We next investigated the mechanisms underlying chemerin's ability to augment T-cell–mediated cytotoxicity. Using siRNA knockdown, we again examined the role of CMKLR1 in mediating chemerin's effects. siRNA significantly reduced both mRNA and surface protein levels of CMKLR1 in tumor cells (Supplementary Fig. S1). Neither control nor CMKLR1 siRNA affected the low level of cytotoxicity seen using naïve T cells in either PBS- or chemerin-treated tumor cells. However, CMKLR1 knockdown completely abrogated chemerin's effect on tumor cell cytotoxicity by activated T cells, whereas control siRNA had no effect (Fig. 5B).
Chemerin augmentation of cytotoxicity is mediated, in part, by PTEN and PD-L1
Given chemerin's impact on both PTEN and PD-L1 expression, we then explored their roles in the cytotoxicity assay. Control or PD-L1 siRNA were then used to look at the role of PD-L1 in this setting (Supplementary Fig. S8). Again, no effect of siRNA transfections was seen with naïve T cells. Using activated T cells, control siRNA again had no impact on the chemerin-mediated increase in tumor killing, while PD-L1 knockdown significantly increased cytotoxicity in both control and chemerin-treated tumor cells (Fig. 5C). This is not surprising given the high levels of PD-1 found on the activated T cells (Supplementary Fig. S7), and known effects of blocking PD-L1 in this setting (37). Interestingly, levels of cytotoxicity in the PD-L1–knockdown groups were comparable with, and not statistically different from, the chemerin-treated/control siRNA groups: control siRNA + chemerin-treated cells displayed 29% lysis compared with 31% and 33% lysis in the control PBS- and chemerin-treated PD-L1–knockdown groups, respectively. While there was a small difference in activated T-cell lysis between chemerin-treated control siRNA cells compared with chemerin-treated PD-L1 siRNA DU145 cells, this was not statistically significant. Similarly, there was no significant difference in activated T-cell lysis between the PBS- versus chemerin-treated PD-L1 siRNA DU145 subsets, showing that PD-L1 was necessary for the effect of chemerin on T-cell–mediated cytotoxicity (Fig. 5C). We then examined the effects of PTEN knockdown via siRNA transfection (Supplementary Fig. S8). In control-treated cells, knockdown of PTEN resulted in significantly less killing by activated T cells (Fig. 5D). PTEN siRNA knockdown in chemerin-treated cells resulted in the complete abrogation of the increase in T-cell killing seen with chemerin-treated/control siRNA cells, to the level of PBS-treated/control siRNA tumor cells (Fig. 5D). PTEN knockdown significantly reduced the effect of chemerin treatment, thus the difference in cytotoxicity seen between control- and chemerin-treated cells using control siRNA was significantly greater than the increase seen using PTEN siRNA. This strongly supports a mechanistic role for PTEN in chemerin-augmented T-cell cytotoxicity. Together, these data support roles for both PTEN and PD-L1 in how chemerin augments sensitivity to T-cell–mediated cytotoxicity.
Chemerin treatment is as effective as atezolizumab at augmenting T-cell–mediated cytotoxicity
While statistically significant increases in T-cell cytotoxicity were seen with chemerin, we compared this result with a clinically validated and approved checkpoint inhibitor, anti–PD-L1 antibody atezolizumab, in our cytotoxicity assays. The addition of isotype antibody to the cytotoxicity assay did not impact the established beneficial effect of chemerin treatment. The addition of atezolizumab significantly increased activated T-cell–mediated cytotoxicity in the control-treated DU145 cells (Fig. 5E), consistent with studies showing the effects of blocking PD-L1 in in vitro cytotoxicity assays (45). There was no statistically significant difference between chemerin-treated DU145 cells with the addition of atezolizumab antibody compared with isotype control, while there was a significant difference with the addition of atezolizumab to PBS control–treated tumor cells compared with isotype control. Blockade of PD-L1 with atezolizumab negated the significant difference in lysis between control- and chemerin-treated DU145 cells (Fig. 5E). Together, these suggest that atezolizumab is effective in control-treated cells, with basal PD-L1 expression, but added no additional significant impact in chemerin-treated DU145 cells, where chemerin pretreatment suppresses PD-L1 expression. With effective PD-L1 blockade by atezolizumab, chemerin treatment did not further augment T-cell cytotoxicity, highlighting PD-L1 as a key downstream pathway component in mediating chemerin's effects on tumor cells.
We then set out to compare the effect of chemerin treatment on T-cell cytotoxic directly with both PD-L1 siRNA knockdown as well as atezolizumab blockade. Using experimental conditions as above, we applied these conditions in parallel, independently repeating with comparable results in both DU145 and U2OS tumor cells. No impact was seen in the cytotoxicity using naïve T cells (data not shown). Using activated T cells, we again found that chemerin treatment significantly increased T-cell–mediated cytotoxicity of DU145 cells (Fig. 5F). Similarly, chemerin treatment of U2OS cells led to significantly increased T-cell–mediated cytotoxicity (Fig. 5G). Importantly, chemerin treatment was as effective at augmenting T-cell–mediated tumor cell lysis in comparison with PD-L1 siRNA or atezolizumab blockade, with no significant differences between the three conditions for both DU145 and U2OS cells (Fig. 5F and G). We looked at treatment of the activated effector T cells, which lack CMLKR1, and saw no impact of chemerin treatment on immune cell PTEN expression, or cytolytic ability (Supplementary Fig. S6), suggesting the effects seen were due to tumor-intrinsic changes after chemerin treatment.
Collectively, these data show that in two different tumor cell lines chemerin, via CMKLR1, can induce the upregulation of PTEN and concurrent downregulation of PD-L1 expression in tumor cells. This results in a significant increase in T-cell–mediated cytotoxicity, comparable, and not statistically different, in our assays with siRNA knockdown of PD-L1 or the clinically approved atezolizumab.
Expression of chemerin in the TME leads to decreased in vivo tumor growth
To test our hypothesis that forced overexpression of chemerin by tumor cells would act to suppress tumor growth, in part, by modulation of PTEN and PD-L1, we used plasmid transfection to introduce the human RARRES2 gene into the DU145 tumor cells. Both WT and vector control DU145 cell lines showed no detectable chemerin by ELISA, while the RARRES2-transfected line showed significant production of secreted chemerin, with no differences seen in in vitro proliferation (Supplementary Fig. S2E and S2F). To determine whether the tumor-secreted chemerin was functional and active, we utilized conditioned media from both control and chemerin-expressing tumor lines in a chemotaxis assay and found that only the chemerin-expressing media were able to mediate chemotaxis of CMKLR1-positive cells (Supplementary Fig. S2G). Tumor cells were inoculated subcutaneously in NSG mice and growth was monitored. Chemerin expression in the TME resulted in significantly reduced tumor growth compared with control tumors (Fig. 6A). While chemerin can recruit immune effector cells into the TME (10, 46), the immunodeficiencies in NSG mice (absent NK/T/B and defective macrophage and DCs) suggested the differences in tumor growth seen could be due, in part, to tumor-intrinsic factors. We analyzed ex vivo tumors and found that chemerin-expressing tumors had significantly higher PTEN and significantly lower PD-L1 levels compared with controls (Fig. 6B). This is consistent with our in vitro data, and suggests chemerin may play a role at modulating tumor PTEN and PD-L1, favorably, in vivo as well.
While tumor line studies are informative, they have limited applicability to the clinical setting. To investigate the effects of chemerin on human primary prostate tumors, we collected primary tumors from 4 patients (two local and two metastatic tumors) under an IRB-approved protocol. One patient had enough tumor collected that allowed us to perform several experiments, while the other three had limited tumor cell content. Primary tumor cells were cultured in the presence or absence of recombinant human chemerin and assessed for changes in PTEN and PD-L1 expression by qPCR. Compared with controls, there was a significant increase in PTEN and decrease in PD-L1 (Fig. 6C and D) in primary human prostate tumor cells treated with chemerin. While limited by the amount of tumor collected from patients, we were able to analyze tumor cells from one patient and found detectable surface expression of CMKLR1 on these tumor cells (not shown), suggesting, as in our tumor cell lines, that chemerin may be acting through CMKLR1 on tumor cells to modulate PTEN and PD-L1.
Finally, we looked at human clinical trial data from patients with metastatic prostate cancer treated with ipilimumab (anti–CTLA-4) on a single institution clinical trial (NCT02113657; ref. 47). Published RNA expression data were used to look at RARRES2, PTEN, and PD-L1 (CD274) in these patients, and evaluate clinical outcomes. Comparison of patients with the highest and lowest quartile RARRES2 expression showed almost threefold increase in PTEN expression in those patients whose tumors had the highest RARRES2 expression (Fig. 6E). Tumor PD-L1 RNA was low and not different between groups (not shown) and mostly undetectable by IHC. However, evaluable TME immune cell PD-L1 and CD8 by IHC was available. Greater than 50% reduction in PD-L1 and approximately fourfold increase in CD8 expression was seen in the highest RARRES2 quartile compared with the lowest (Fig. 6E). Clinical outcomes for patients above and below the median RARRES2 expression were analyzed; the RARRES2-high group had a median overall survival (OS) of 40.3 months compared with 5.8 months for low RARRES2 [HR, 0.83; 95% confidence interval (CI), 0.27–2.6; P = 0.39]. Median PSA progression-free survival (PFS) was also increased in the RARRES2-high compared with the -low group, 11.2 versus 0.7 months (HR, 0.49; 95% CI, 0.16–1.5; P = 0.12; Fig. 6F). Relative abundancies of indicated immune populations [based on RNA-sequencing (RNA-seq) signatures] in both high and low RARRES2 expression groups showed significant increases in immune effector populations in the RARRES2-high group compared with the -low group (Fig. 6G). While limited in sample size, these data suggest that high RARRES2 in the TME is correlated with increased PTEN, decreased PD-L1, and increased immune effector populations. Thus, a strategy for increasing expression of chemerin within the TME in humans may be beneficial in the clinical setting.
Tumors have developed various suppressive mechanisms to evade antitumor immune responses and regulatory signaling that may limit their growth. As both are altered in the TME, further study of the interplay between tumor cell–intrinsic oncogenic signaling and extrinsic antitumor immunosurveillance is necessary to improve current immunotherapies.
The link between PD-L1 (cell-extrinsic immune responses) and PTEN (cell-intrinsic responses) expression has been described, with several examples of PTEN loss or suppression resulting in increased PD-L1 expression in tumors (28, 34, 35). Other studies suggest PD-L1 expression in prostate, breast, and lung carcinoma may be dependent on PI3K, commonly regulated by PTEN (48). However, this association is likely context dependent, as the regulation of PD-L1 expression is controlled by many factors and pathways (reviewed in ref. 49). PD-L1 expression correlates with tumor aggressiveness and poor clinical outcomes (50–53), as does loss of PTEN (52, 54–56), in several datasets, further supporting the clinical impact of alterations in these two key proteins. In prostate, PD-L1 expression has been reported in up to approximately 47% of de novo metastatic prostate cancers (57), and has been found to correlate with poorer prognosis and risk of disease recurrence (52, 53), while PTEN loss has been correlated with both risk of recurrence in localized disease and lethal progression (55, 56), suggesting a therapeutic strategy to augment PTEN expression may reduce prostate cancer lethality.
Recent studies describe functional consequences of modulating PTEN signaling and its impact on immunoresistance. Toso and colleagues used a conditional PTEN-null mouse model to study the impact of PTEN loss within prostate tumors. They found loss of PTEN resulted in a significant increase in several immunosuppressive cytokines as well as infiltration by granulocytic myeloid-derived suppressor cells (58). Furthermore, Peng and colleagues showed that the PTEN loss led to inhibited T-cell–mediated tumor killing and decreased T-cell trafficking into the TME. Importantly, they showed that patients with metastatic melanoma with PTEN-positive tumors treated with anti-PD-1 antibodies had significantly better responses than otherwise matched patients with PTEN-negative tumors. They showed that PI3Kβ inhibition, part of the PI3K/Akt pathway activated with PTEN loss, enhanced the activity of T-cell–mediated immunotherapy in mice bearing PTEN-deficient tumors (31). Additional evidence recently elucidated the importance of PTEN loss in developed resistance to anti-PD-1 immunotherapy in human sarcoma (59), supporting the clinical relevance of this mechanism. Thus, PTEN alterations that impact immunotherapy efficacy are key mechanisms to consider in optimization of these therapies.
It is important to recognize, however, the variety of PTEN alterations that exist across cancers. In addition to deletion, expression can be altered by DNA methylation, transcriptional repression, and translational disorder, reducing PTEN expression in numerous tumor types (60). Deletion can be bi- or monoallelic, with approximately 42% of patients with prostate cancer having monoallelic PTEN loss (61). The various modes of PTEN loss in malignancy can lead to distinctive signaling modulations, which are not always equivalently regulated. Thus, the exact type of PTEN loss in tumors would potentially dictate the relevance of chemerin modulation in humans. For example, complete allelic loss of PTEN (as in our PC3 cells) in tumors might suggest that modulating tumor chemerin levels in these patients would not result in changes in tumor PD-L1; however, the ability of chemerin to recruit immune effector cells into the TME may still have beneficial outcomes. In those tumors with intact, but decreased, PTEN expression, chemerin modulation may then act to increase its expression and potentially decrease PD-L1, suppressing tumor growth and improving responses to immunotherapies.
Our study is the first to show that chemerin, an innate immunocyte chemoattractant, can reactivate PTEN in human prostate and sarcoma tumor lines, while concomitantly suppressing PD-L1 expression. We describe a novel mechanistic link between chemerin/CMKLR1, PTEN, and PD-L1 in tumor cells, and identify key signaling pathway components. We show a beneficial, functional impact of chemerin treatment, with reduced tumor cell migration/invasion and increased T-cell–mediated cytotoxicity, on par with the clinically approved anti–PD-L1 antibody, atezolizumab. In vivo tumor studies showed chemerin expression in the TME significantly reduced tumor growth, with an increase seen in tumor PTEN and decrease in tumor PD-L1. Primary human prostate tumor cultures recapitulated our cell line studies, again showing chemerin treatment resulting in favorable modulation of PTEN and PD-L1.
A recent study showed that chemerin could suppress hepatocellular carcinoma growth and metastases via the PTEN–Akt signaling axis in a mouse model (30). Using human hepatocellular carcinoma cell lines, Li and colleagues showed that chemerin overexpression resulted in PTEN upregulation and suppression of the PI3K/Akt pathway. As in our studies, they also saw a significant decrease in tumor cell migration/invasion with exposure to chemerin. Their data are supportive of our initial findings with PTEN, but our study extends this mechanistically and elucidates a novel signaling cascade in tumors linking chemerin/CMKLR1 to PD-L1. Independent validation of findings across laboratories and tumor types suggests this axis may be biologically and clinically relevant.
In conclusion, we report a previously unidentified signaling cascade linking chemerin/CMKLR1, PTEN, and PD-L1 in human tumor cell lines, resulting in a significant decrease in tumor migration/invasion and increase in T-cell–mediated killing, with significant suppression of in vivo tumor growth. In addition to its already described role of favorably modulating antitumor immune responses by recruitment of immune effector cells into the TME, this data now show a new tumor cell–intrinsic mechanism of chemerin treatment. Ongoing and future studies will further investigate biologic consequences of modulating this axis, with the goal of clinical translation.
Disclosure of Potential Conflicts of Interest
K. Rennier reports grants from Ferring Pharmaceuticals (partially supported the research) during the conduct of the study. R.K. Pachynski reports grants from Prostate Cancer Foundation and Sidney Kimmel Foundation during the conduct of the study; personal fees from AstraZeneca (speaker, advisory), BMS (advisory), EMD Serono (advisory), Pfizer (advisory), Genomic Health (speaker, advisory), Merck (speaker), Sanofi (speaker, advisory), Dendreon (advisory), Jounce Therapeutics (advisory), and Bayer (advisory) outside the submitted work; personal fees and non-financial support from Genentech/Roche (speaker, travel); and is a co-inventor on a U.S. utility application on the use of chemerin fusion proteins to treat cancer that is owned by Washington University and is currently unlicensed, and is also a co-inventor on a provisional patent application on a method to identify high-risk metastatic prostate cancer patients that is also unlicensed. No potential conflicts of interest were disclosed by the other authors.
K. Rennier: Conceptualization, data curation, formal analysis, validation, investigation, visualization, methodology, writing-original draft, writing-review and editing. W.J. Shin: Formal analysis, validation, investigation. E. Krug: Formal analysis, validation, investigation. G. Virdi: Software, formal analysis, validation, investigation. R.K. Pachynski: Conceptualization, resources, supervision, funding acquisition, validation, visualization, methodology, writing-original draft, project administration, writing-review and editing.
We would like to thank Dr. Brian Van Tine for the generous donation of the sarcoma cell lines used in this article. This work was supported, in part, by American Cancer Society MSRG 125078-MRSG-13-244-01-LIB, the Prostate Cancer Foundation, The Kimmel Foundation, and a generous gift from Kerry Preete. K. Rennier was supported, in part, by a fellowship provided by Ferring Pharmaceuticals.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.