Pharmacologic targeting of the dopamine D2 receptor impacts the efficacy of immune checkpoint blockade in melanoma

Abstract

Background

Immune checkpoint inhibitors (ICIs) have been successful in treating advanced melanoma, yet, the 10-year melanoma-specific survival is only 52%. Our prior work using genetic linkage analysis revealed that the murine prolactin (PRL) locus associates with ICI response in C57BL/6 (B6)-syngeneic B16F0 melanoma. This was validated in F1 crosses of B6 with Collaborative Cross mice selected as potential non-responders or responders in the PRL locus and directly by coadministration of PRL with ICIs which slowed B16F0 growth compared with ICIs alone. This study uses Food and Drug Administration (FDA)-approved drugs that act on the dopamine D2 receptor (D2R), which inhibits PRL release from the pituitary, and suggests the potential of this receptor as a target to improve ICI outcomes.

Methods

We used FDA-approved D2R agonist bromocriptine (BRC) and D2R antagonist metoclopramide (MCP) to lower and raise systemic PRL, respectively in vivo. PRL-locus ICI responder and non-responder animal models were employed to assess the effect of D2R modulation with ICIs on tumor growth. Immunohistochemistry, single-cell and bulk RNA sequencing, and flow cytometry were used to explore the mechanism by which pharmacologic D2R targeting impacts antitumor immunity.

Results

BRC accelerated B16F0 growth and diminished intratumoral CD8+ T cell infiltration with ICIs in a PRL-locus ICI responder model. Conversely, MCP with ICIs slowed both B16F0 and MEL11443 growth and increased intratumoral CD8+ T cell infiltration in B6 mice, a PRL-locus ICI non-responder model. Bulk RNA sequencing revealed that MCP enhanced intratumoral immune-mediated processes based on biological sex. Single-cell RNA sequencing uncovered enhanced activity of intratumoral CD8+ T cells in a PRL-locus ICI responder compared with the B6 PRL-locus ICI non-responder model. Strikingly, MCP directly enhanced major histocompatibility complex expression on B16F0 and MEL11443 cells, increased antigen-specific CD8+ T cell activation, and modulated bone marrow-derived macrophage polarization in vitro, suggesting an additional mechanism independent of PRL in promoting antitumor immunity.

Conclusions

These findings demonstrate that pharmacologic modulation of D2R impacts the efficacy of immune checkpoint blockade in murine melanoma and raise the possibility of using FDA-approved D2R targeting therapies as cotherapeutics to overcome ICI resistance in patients with advanced melanoma.

WHAT IS ALREADY KNOWN ON THIS TOPIC

• Immune checkpoint inhibitors (ICIs) have improved advanced melanoma survival but benefit only half of patients. Therefore, investigation is needed to identify strategies to improve ICI response.

WHAT THIS STUDY ADDS

• This study reveals that pharmacologic D2 receptor (D2R) activation or inhibition leads to blunted or improved ICI response, respectively, in mouse melanoma models. D2R inhibition directly increases antigen presentation genes on melanoma cells, enhances CD8+T cell activation, and modulates macrophage polarization toward an antitumor phenotype.

HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY

• Our observation that pharmacologic D2R antagonism enhances ICI efficacy offers a potential avenue for repurposing Food and Drug Administration-approved D2R-targeting medications as co-therapeutics to overcome ICI resistance in patients with advanced melanoma.

Introduction

Immune checkpoint inhibitors (ICIs) have been successful in treating advanced melanoma, though the 10-year melanoma-specific survival with ICIs is only ∼52% which raises a need to improve response rates.1,3 We previously reported that host genetic background influences ICI outcomes in genetically heterogeneous F1 mouse models bearing C57BL/6 (B6)-syngeneic B16F0 melanoma.⁴ Host genetic linkage analysis demonstrated a strong association between prolactin (PRL) and response to ICIs.⁴ Cotherapy of PRL with ICIs reduced B16F0 tumor growth and enhanced intratumoral CD8+ T cell infiltration, suggesting that systemic PRL levels correlate with ICI outcomes.⁴

PRL is predominantly secreted by hypothalamic pituitary lactotrophs. The hypothalamus inhibits pituitary release of PRL through dopamine via the D2 receptor (D2R).^{5 6} Metoclopramide (MCP) is a D2R antagonist used to promote gastrointestinal motility and leads to hyperprolactinemia in humans and when administered daily to mice.^{7 8} Bromocriptine (BRC) is a D2R agonist indicated for treatment of conditions associated with hyperprolactinemia, including Parkinson’s disease, acromegaly, diabetes mellitus, and prolactinomas.⁹ BRC lowers systemic PRL levels in humans and when administered daily to mice.¹⁰

PRL improves macrophage release of proinflammatory cytokines interleukin (IL)-1 and IL-6 and decreases mortality in a mouse model of sepsis.¹¹ MCP treatment enhances the proliferative capacity of splenocytes and improves phagocytosis of macrophages in an ex vivo trauma-hemorrhage model.¹² Conversely, BRC demonstrates immunosuppressive activity, suppressing autoantibody production in murine models of lupus and antiphospholipid antibody syndrome.¹³ BRC has also been reported to improve flares in patients with various autoimmune conditions.¹⁴ Also, BRC directly suppresses immune activity as in vitro administration to mouse and human T cells blunted their proliferation and activation.¹⁵

That MCP and BRC differentially impact immune cell subsets suggests that they could influence anti-tumor immunity. In our study, we find that MCP and BRC play a role during melanoma immunotherapy by altering the tumor microenvironment (TME) and impacting ICI efficacy (figure 1). This investigation provides novel insights into the impact of pharmacologic modulation of D2R during ICI therapy and may translate to combining Food and Drug Administration (FDA)-approved D2R targeting therapies with ICIs to improve advanced melanoma outcomes.

Figure 1. Graphical abstract. Figure created in BioRender. BRC, bromocriptine; D2R, D2 receptor; HAL, haloperidol; ICI, immune checkpoint inhibitor; IL, interleukin; MCP, metoclopramide; MHC, major histocompatibility complex; PRL, prolactin.

Materials and methods

Mice and cell lines

C57BL/6 (strain# 000664, abbreviated B6 here) and C57BL/6-Tg(TcraTcrb)1100Mjb/J OT-I (strain# 003831, abbreviated OT-I here) mice were purchased from the Jackson Laboratory. The Collaborative Cross (CC) lines CC051/TauUnc and CC075/UncJ were purchased from the Systems Genetics Core Facility at the University of North Carolina (abbreviated to CC51 and CC75 here, respectively).¹⁶ (CC51xB6)F1 and (CC75xB6)F1 mice were generated in-house by breeding B6 sires with CC51 and CC75 dams, respectively.

B16F0 (American Type Culture Collection (ATCC), cat. #CRL-6322) and MEL11443 (ATCC, cat. #CRL-3473) cell lines were purchased from ATCC, cultured in a sterile environment at 37°C and 5% CO₂, and maintained for fewer than five passages. B16F0 cells were cultured in D10 medium: Dulbecco’s Modified Eagle’s Medium (DMEM) (Thermo Fisher, cat. #11965092) containing 10% fetal bovine serum (FBS) (Gibco, cat. #A52568-01) and penicillin (100 U/mL) plus streptomycin (100 U/mL) (Gibco, cat. #15140122). MEL11443 cells were cultured in R10: same as D10 except DMEM replaced with RPMI (Thermo Fisher, cat. #72400047). All in vivo experiments were performed on B6 mice that were 8 weeks old and (CC51xB6)F1 and (CC75xB6)F1 mice that were 8–24 weeks old with inoculations performed on the right inguinal region with 2×10⁵ B16F0 or MEL11443 cells suspended in DMEM or RPMI, respectively. Tumor growth was monitored three times per week by caliper measurement. Tumor volume was calculated as (W×W× L)/2. Endpoints for survival studies included the tumor volume reaching >400 mm³ and excluded mice becoming lethargic or losing >10% of body weight. Mice were housed with five or fewer per cage and fed according to specific pathogen free grade requirements. Euthanasia was performed using CO2 asphyxiation followed by cervical dislocation. Male and female mice were equally represented. Animal handling and experiments were conducted in accordance with guidelines and regulations set by the Wayne State University Institutional Animal Care and Use Committee.

T cell culture

OT-I CD8+ T cells were isolated from splenocytes of OT-I mice using the EasyEights EasySep Magnet (STEMCELL Technologies, cat. #18103) and the EasySep Mouse CD8+ T cell Isolation Kit (STEMCELL Technologies, cat. #19853). Isolated OT-I CD8+ T cells were cultured in complete RPMI: RPMI containing 10% FBS, penicillin (100 U/mL) plus streptomycin (100 U/mL), 1 mM HEPES (Gibco, cat. #15630080), 1 mM sodium pyruvate (Gibco, cat. #11360070), 0.1 mM EmbryoMax Non-Essential Amino Acids (Millipore, cat.# TMS-001-C), 55 mM 2-mercaptoethanol (Gibco, cat. #21985023).

Macrophage culture

Bone marrow (BM) cells were collected at the time of sacrifice and isolated as previously described.¹⁷ BM cells were differentiated into BM-derived macrophages (BMDM) on non-tissue culture treated petri dishes using M0 medium (70% R10 and 30% L-929 conditioned R10) for 7–9 days.

Treatment with ICIs

Mice were administered ICIs every 3 or 4 days starting 3 days after tumor inoculation by intraperitoneal injection with 200 µg anti-programmed cell death protein-1 (αPD-1) (Leinco, clone RMP1-14, cat. #P362) and 100 µg anti-cytotoxic T-lymphocyte-associated protein 4 (αCTLA-4) (Leinco, clone 9D9, cat. #C2855) in 200 µL normal saline (NS).

Administration of MCP or BRC

MCP was obtained from Heartland Veterinary Supply and Pharmacy (SKU: 28992_-RX) as a pharmaceutical grade 5 mg/mL solution in NS. B6 mice were subcutaneously administered MCP at 8 mg/kg or volume-matched NS starting 1 day before B16F0 inoculation. BRC was obtained from Cayman Chemical (cat. #14598), dissolved in a 1:1 mixture of Tween 80:Ethanol at 20 mg/mL.¹⁸ Then, the solution was filter-sterilized through a Costar Spin-X Centrifuge Tube Filter, 0.22 µm Pore CA Membrane (cat. #8160). (CC51xB6)F1 mice were intraperitoneally administered BRC at 10 mg/kg or volume-matched 1:1 Tween 80:Ethanol vehicle (Veh) starting 3 days after tumor inoculation.

Blood collection

Blood was drawn from the submandibular vein using a lancet. Serum was collected by allowing blood to clot for 30 min before centrifugation at 400 × g for 5 min.

Serum PRL measurement

Serum PRL was measured using the Thermo Fisher Mouse Prolactin ELISA Kit (cat. #EMPRL) following the manufacturer instructions.

Immunohistochemistry and Weka Segmentation

Intratumoral CD8+ T cell infiltration was determined via immunohistochemistry (IHC) staining and Weka Segmentation as previously described.^{4 19}

RNA isolation, library preparation, and bulk RNA sequencing

Tumor tissue was flash-frozen in liquid nitrogen and stored at −80°C. Following homogenization using the Tissue-Tearor (BioSpec, cat. #985370) Total RNA was purified using the RNeasy Mini kit as described by the manufacturer (Qiagen, cat. # 74106). RNA integrity, contamination, and quantity were monitored using the Agilent TapeStation platform and absorbance ratios for 260/280 nm and 260/230 nm using the Nanodrop 2000. Library construction was performed according to the Illumina Stranded Total RNA Prep with Ribo-Zero Plus protocol followed by sequencing on the NovaSeq 6000 platform at >20 M reads per sample.

Bulk RNA sequencing data analysis

Postprocessing of the run to generate FASTQ files was performed at the Wayne State University Genome Sciences Core. The quality of the sequencing was first assessed using the FastQC tool V.0.11.9. Raw reads were then quality trimmed and filtered by a length of 20 bases using Trimmomatic V.0.39. Trimmed reads were analyzed with the mouse GRCm38 reference genome and resulting quantification files were imported into R to obtain raw transcript counts for all annotated mouse genes. Principal component and differential gene expression analyses were performed using R package DESeq2. Principal component analysis identified five outlier samples, which were removed from further analysis. For comparison between biologic sexes, X and Y chromosome genes were included in differential gene expression analysis, whereas for comparison within biologic sex, these genes were removed. For downstream analysis, differentially expressed genes with adjusted false discovery rate ≤0.05 and |log₂FC| ≥1 were included. Gene ontology functional enrichment of gene expression changes in tumors was performed using Metascape.²⁰

Single-cell RNA sequencing

B16F0 tumors were dissociated (Miltenyi, cat. #130-096-730), run through a gradient of 40% and 80% Percoll PLUS (Cytiva, cat. #17544501), and CD45+ enriched (Miltenyi, cat. #130-110-618). Replicates were tagged using the 10X Genomics 3’ CellPlex Cell Multiplexing Oligo (CMO) system (10X Genomics, cat. #PN-1000261) with library preparation using the Chromium Next GEM Single Cell 3’ Kit (10X Genomics, cat. #PN-1000268) and sequencing was performed on the NovaSeq 6000 (Illumina). Run parameters and quality control metrics recommended by Illumina and 10X Genomics were all met or exceeded. Alignment was performed using the Cellranger pipeline (GSC Alignment refdata-cellranger-mm10-3.0.0). Downstream data analyses were performed using Seurat in R V.4.5.0. Inclusion criteria were defined as follows: >200 genes expressed, <10 k RNA counts, <10% mitochondrial gene content. Over 30,000 cells were included in downstream analyses. Cells were clustered on the top 2,000 most variable genes. Doublet exclusion was achieved by removing cells with >1 CMO tag. Manual cell type identification was performed based on canonical marker genes. Cluster proportions were determined by dividing the number of cells in each cluster by the total number of cells in all clusters.

Treatment with MCP and haloperidol in vitro

Haloperidol (HAL) was obtained at a 1 mg/mL solution in methanol (Sigma Aldrich, cat. #H-030). B16F0 and MEL11443 cells were seeded into a 12-well plate at 100,000 cells/mL and administered 100 µg/mL MCP with or without 0, 2, 20, 200, 2,000, or 20,000 pg/mL interferon gamma (IFN-γ) or treated with 1 µg/mL HAL for 48 hours. For HAL, 0.1% methanol was used as Veh control.

OT-I CD8+ T cells were seeded into a 96-well flat bottom plate at 1,000,000 cells/mL and activated with 50 ng/mL SIINFEKL (N4), SIIQFEKL (Q4), SIITFEKL (T4), or SIIVFEKL (V4) peptide with or without 100 µg/mL MCP or 1 µg/mL HAL for 48 hours. The peptide variants were obtained from GenScript.

BMDM were pretreated with 100 µg/mL MCP for 2 hour then polarized to the M1 phenotype using 50 ng/mL IFN-γ and 10 ng/mL lipopolysaccharide (LPS) or to the M2 phenotype using 20 ng/mL IL-4 for 24 hours.

For all experiments, each condition was performed in triplicate.

Flow cytometry

Cells were stained with 1:10 000 ViaDye Red (Cytek, SKU R7-60008) viability dye for 15 min at 4°C and washed twice with 1× phosphate-buffered saline (PBS, Gibco, cat. #70011044) containing 1% FBS. Then, cells were incubated with 1:200 anti-CD16/32 (Fc Block, BD Pharmingen, cat. #553142) for 15 min at 4°C followed by staining with the following fluorochrome conjugated antibodies: 1:200 major histocompatibility complex (MHC) I (H-2Kb) PE (Invitrogen, cat. #12-5958-82), 1:200 MHC II BV786 (I-A,I-E) (Invitrogen, cat. #417-5321-80), 1:200 CD44 PE (BioLegend, cat. #103008), 1:200 CD25 PE-Cy7 (BioLegend, cat. #102016), 1:200 CD69 PE-Cy5 (BioLegend, cat. #104510), and 1:200 PD-1 BV605 (BD Biosciences, cat. #568868). Following surface staining, cells were washed twice with 1× PBS containing 1% FBS. All data were acquired on Cytek NorthernLights 3L V/B/R (Cytek) and analyzed using FCS Express 7 (De Novo Software).

Quantitative reverse transcription PCR

RNA was extracted via TRIzol (Thermo Fisher, cat. #15596026). Complementary DNA (cDNA) was prepared using LunaScript RT SuperMix Kit (New England Biolabs, cat. #M3010L). Quantitative reverse transcription PCR was performed using iTaq Universal Probes Supermix (Bio-Rad, cat. #1725134) using 10 ng cDNA/well and 500 nM primers. Primers include murine Gapdh (Mm99999915_g1), B2m (Mm00437762_m1), Arg1 (Mm00475988_m1), Nos2 (Mm00440502_m1), Il-12β (Mm01288989_m1), and PRL (Mm00599950_m1). Messenger RNA quantification was reported as a difference in cycle threshold (ΔCT) relative to Gapdh.

Western blot

Flash-frozen tumor was homogenized in 1× RIPA buffer (Cell Signaling Tech, cat. #9806S) with protease/phosphatase inhibitor (Cell Signaling Tech, cat. #5872) using a Tissue-Tearor. Lysates were collected after centrifugation at 15,000 g for 5 min. Protein concentration was measured by bicinchoninic acid (BCA) protein assay kit (Thermo, cat. #23227) and lysates were prepared for loading with Laemmli buffer (Bio-Rad, cat. #1610747) and boiled for 7 min at 95°C. Samples were loaded in 10% precast polyacrylamide gels (Bio-Rad, cat. #4561036), run at 100 V in Tris/Glycine/sodium dodecyl sulfate (SDS) buffer (Bio-Rad 1610732), transferred to polyvinylidene fluoride (PVDF) membrane (Bio-Rad, cat. #1704272), blocked for 1 hour with 5% dry milk in tris buffered saline-Tween (TBS-T), and probed for PRL (R&D, cat. #AF1445), D2R (Proteintech, cat. #55084-1-AP), and β-Actin-HRP (BioLegend, cat. #664803) in blocking buffer. Secondary antibodies for PRL and D2R were αGoat-HRP (Invitrogen, cat. #PA128664) and αRabbit-HRP (Invitrogen, cat. #31460), respectively. Membranes were incubated for 5 min in ECL reagent (Bio-Rad, cat. #1705061) and visualized via ChemiDoc imaging system.

Data and statistical analysis

Tumor volumes were analyzed using a linear regression to compare experimental groups. Regression analysis was performed on tumor growth curves by a log10 transformation followed by a regression against days post-inoculation then extrapolation to 400 mm³ tumor volume. To have 80% power with a two-sided Type I error rate of 0.05 to detect a difference of 1.5 SD units between two groups, eight mice per group were used. Statistical analysis was conducted using GraphPad Prism V.10.6.0. Data are presented as the mean±SD unless otherwise noted. Statistical comparisons between groups were performed as stated in figure legends. A p value <0.05 is considered statistically significant.

Results

BRC attenuates ICI response in a PRL locus-driven ICI responder model

Our prior work identified PRL as a candidate regulator of ICI response.⁴ For validation, CC mouse strains were selected by identifying putative ICI responders that had the NZO founder strain genotype or non-responders that had the B6 founder strain genotype at the PRL locus.⁴ These CC mice were crossed with B6 mice to produce (CCxB6)F1 progeny.⁴ (CC51xB6)F1 mice were the ICI responder model that demonstrated the greatest B16F0 tumor control and infiltration of intratumoral CD8+ T cells during ICI therapy.⁴ To better understand the role PRL might play in the ICI response, we inhibited PRL release with the D2R agonist BRC in B16F0-bearing (CC51xB6)F1 mice then treated with αPD1 and αCTLA-4 dual ICI therapy, using Veh-matched mice as controls (figure 2A). Due to the aggressive growth rate of B16F0 melanoma, ICIs were administered 3, 6, and 10 days after tumor inoculation.^{4 21} We observed a significant delay in tumor growth with ICIs in female (CC51xB6)F1 mice, which was abrogated in the presence of BRC (figure 2B). In contrast, male (CC51xB6)F1 mice failed to respond to ICI therapy, regardless of BRC administration (figure 2C). Because of the fast growth rate of B16F0, survival analysis was performed to determine the days post-inoculation to reach a 400 mm³ tumor volume.^{4 21} The addition of BRC to ICIs substantially reduced survival in female, but not male, mice (figure 2D,E).

Figure 2. BRC attenuates tumor control during immune checkpoint blockade in a PRL locus ICI responder model. (A) Experimental scheme. B16F0-bearing female (F) and male (M) (CC51xB6)F1 mice received 10 mg/kg BRC daily and/or 100 µg αCTLA-4 and 200 µg αPD-1 on days 3, 6, and 10 after tumor inoculation. An equivalent volume of 1:1 Tween 80:Ethanol (Veh) was given to control mice. Individual tumor growth curves of female (B) and male (C) mice. Survival analyses of female (D) and male (E) mice. Intratumoral CD8 IHC performed with Vector Red substrate of B16F0-bearing female (F) and male (G) mice that received Veh, Veh+ICI, BRC, or BRC+ICI. ELISA of serum mouse PRL from B16F0-bearing female (H) and male (I) mice administered BRC or Veh. (D, E) p values determined by log rank test. (F, G, H, I) P values determined by Student’s t-test. *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001. Data are representative of three independent experiments. αCTLA-4, anti-cytotoxic T-lymphocyte-associated protein 4; αPD-1, anti-programmed cell death protein-1; BRC, bromocriptine; ICI, immune checkpoint inhibitor; IHC, immunohistochemistry; PRL, prolactin; Veh, vehicle.

To investigate whether the impact of BRC on ICI response is immune-mediated, we performed IHC staining on tumors from (CC51xB6)F1 mice to determine CD8+ T cell infiltration. We found that ICI-treated male and female (CC51xB6)F1 mice demonstrated enhanced CD8+ T cell infiltration during ICI therapy alone (figure 2F,G, online supplemental figure S1A), consistent with our previous work.⁴ Combination therapy with BRC and ICIs led to a decrease in intratumoral CD8+ T cell infiltration over ICI therapy alone in female, but not male, mice (figure 2F,G, online supplemental figure S1A) in agreement with the deleterious impact of BRC on tumor growth and survival of female mice treated with ICIs (figure 2B,D). Interestingly, ICI therapy led to increased intratumoral infiltration of immunosuppressive Foxp3+ cells in male, but not female mice (online supplemental figure S2). This is consistent with our observation that immune checkpoint blockade was unable to enhance antitumor control despite leading to increased intratumoral CD8+ infiltrates in male (CC51xB6)F1 mice (figure 2C,E,G). We verified by ELISA that daily BRC administration reduced serum PRL levels in female and male mice, with higher baseline PRL levels in female than male mice (figure 2H,I). Thus, BRC blunted the antitumor efficacy of dual ICI therapy in female (CC51xB6)F1 mice that are ICI responders in the PRL locus, suggesting that pituitary-derived PRL may function to enhance the efficacy of ICI, potentially in a sex-dependent manner in this genetic background.

BRC does not impact ICI response in an ICI responder model in a non-PRL locus

To determine if the deleterious effect of BRC on ICI response depends on PRL locus positive driver genetics, we tested a model whose ICI response is driven by a non-PRL locus. We used (CC75xB6)F1 mice which have the CAST founder strain genotype in the PRL locus (online supplemental figure S3A), which is expected to be neutral.⁴ In B16F0-bearing (CC75xB6)F1 mice administered dual ICIs, 4/6 mice eliminated the tumor with protection from contralateral rechallenge (online supplemental figure S3B). Survival analysis revealed a marked improvement in overall survival during immune checkpoint blockade (online supplemental figure S3C). The (CC75xB6)F1 mice are therefore an ICI responder model whose positive driver genetics are outside of the PRL locus.

(CC75xB6)F1 mice bearing B16F0 melanoma tumors were administered BRC or Veh daily with or without dual ICIs (online supplemental figure S4A). Treatment with ICIs led to slower tumor growth regardless of coadministration with BRC and independent of sex (online supplemental figure S4B). Moreover, 3/8 mice treated with ICIs and 7/9 mice treated with BRC and ICIs eliminated the tumor with protection from contralateral rechallenge (online supplemental figure S4B). Survival analysis revealed prolonged survival during ICI therapy that was not impacted by the addition of BRC (online supplemental figure S4C). We did not find any differences in intratumoral CD8+ T cell infiltrates between mice administered ICIs alone and those that additionally received BRC (online supplemental figure S4D, online supplemental figure S1B). Strikingly, similarly high numbers of CD8+ infiltrates were observed in mice administered Veh alone and Veh with ICIs, suggesting high baseline intratumoral immune activity prior to immune checkpoint blockade in this genetic background (online supplemental figure S4D, online supplemental figure S1B). We verified that BRC lowered serum PRL levels in both female and male mice (online supplemental figure S4E,F). That the reduction of serum PRL with BRC did not blunt ICI response in (CC75xB6)F1 mice suggests that the ability of BRC to modulate ICI efficacy depends on the genetic background in the PRL locus.

MCP ameliorates ICI response

We then asked if pharmacologic D2R antagonism with MCP would improve ICI response in inbred B6 mice, which historically exhibit poor response to ICIs.^{4 21 22} B6 mice bearing B16F0 or MEL11443 melanoma tumors were treated with 8 mg/kg MCP or NS daily with or without dual ICIs (figure 3A). MCP significantly delayed B16F0 (figure 3B,C) and MEL11443 (figure 3D,E) tumor growth and prolonged survival with ICIs. We found by ELISA that daily MCP administration raised serum levels of PRL in B16F0-bearing male, but not female mice (figure 3F,G). MCP with ICIs significantly increased intratumoral CD8+ T cell infiltrates fivefold over ICIs alone, comparable to our prior work using a slow-release PRL pellet (figure 3H, online supplemental figure S1C and study by Hackett et al ⁴). Moreover, MCP improved anti-tumor control with immune checkpoint blockade in B16F0 tumors to a greater extent than a slow-release PRL pellet in our prior study.⁴ This suggests that MCP may be enhancing ICI efficacy in part through a PRL-independent mechanism in B16F0 melanoma. Interestingly, MCP enhanced ICI efficacy in B16F0-bearing female, but not male, mice (online supplemental figure S5A–D). In MEL11443 tumors, MCP with ICIs improved tumor control compared with ICIs alone in both female and male mice (online supplemental figure S5E–H). The female sex-specific benefit of MCP during immune checkpoint blockade in B16F0 melanoma may be due to the male genotype of this tumor model with Y chromosome-encoded proteins acting as foreign antigens in female hosts and thereby enhancing anti-tumor response.²³ Conversely, that MCP ameliorated ICI efficacy in MEL11443 melanoma for both female and male mice may be explained by the female genotype of this tumor model.²⁴

Figure 3. MCP ameliorates tumor control during immune checkpoint blockade in an ICI non-responder model. (A) Experimental schema. B16F0-bearing or MEL11443-bearing B6 mice received 8 mg/kg MCP daily and/or 100 µg αCTLA-4 and 200 µg αPD-1 two times per week starting on day 3 after tumor inoculation. An equivalent volume of NS was given to control mice. (B, D) Individual tumor growth curves and (C, E) survival analyses of B16F0-bearing and MEL11443-bearing mice. ELISA of serum mouse prolactin from B16F0-bearing female (F) and male (G) mice administered MCP or NS. (H) Intratumoral CD8 IHC performed with Vector Red substrate of B16F0-bearing B6 mice that received NS, NS+ICI, MCP, or MCP+ICI. (C, E) P values determined by log-rank test. (F, G, H) P values determined by Student’s t-test. *p<0.05. ****p<0.0001. αCTLA-4, anti-cytotoxic T-lymphocyte-associated protein 4; αPD-1, anti-programmed cell death protein-1; ICI, immune checkpoint inhibitor; IHC, immunohistochemistry; MCP, metoclopramide; NS, normal saline; PRL, prolactin.

To further profile the TME, we analyzed B16F0 tumors by bulk RNA sequencing. Interestingly, principal component analysis showed clear separation in male and female mice receiving MCP with or without ICIs (figure 4A,B). To investigate the effects of MCP alone, we compared tumors from male and female mice receiving MCP alone. We found elevated expression of MHC and MHC-associated transcripts in tumors from MCP-treated male mice relative to female mice, suggesting heightened antigen presentation to T cells (figure 4C). The immune checkpoint molecules Lilr4b (also known as Lilrb4) and Selplg were also increased in tumors from male mice, in addition to several IFN-stimulated genes (figure 4C).²⁵ Gene ontology analysis identified multiple intratumoral immune-related pathways that were upregulated in male compared with female mice that received MCP (figure 4D). This may be explained by the systemic PRL-raising effect of MCP in male, but not female mice (figure 3F,G). Correspondingly, we observed that MCP enhanced ICI efficacy in female, but not male B16F0-bearing mice (online supplemental figure S5A–D). This prompted us to compare tumors from female mice treated with ICIs alone to MCP combined with ICIs (figure 4E). Macrophage-related transcripts (Csf1r, Adgre1, Fce1rg) and innate-immunity molecules (C1qb, Ccl8, Il1rl1) were upregulated in tumors of female mice administered MCP with ICIs compared to ICIs alone (figure 4E). These data suggest that during immune checkpoint blockade in B16F0 melanoma in female mice, MCP reprograms the TME toward enhanced macrophage activity.

Figure 4. MCP differentially impacts the tumor immune microenvironment based on sex. (A, B) PCA of tumors from B16F0-bearing female (F) and male (M) B6 mice that received NS, NS+ICI, MCP, or MCP+ICI. (C) Volcano plot highlighting DEGs between B16F0-bearing F and M mice that received MCP. Significant DEG (padj≤0.05, |log2FC | ≥ 1) were determined using DESeq2 and are colored (red=upregulated in MCP F mice; blue=upregulated in MCP M mice). (D) GO biological process enrichment from Metascape of significant DEG from (C). X-axis represents log10(padj). (E) Volcano plot of ICIs alone (orange) versus MCP with ICIs (green) as in (C). DEG, differentially expressed gene; GO, Gene Ontology; ICI, immune checkpoint inhibitor; MCP, metoclopramide; NS, normal saline; PCA, principal component analysis.

We found that female (CC51xB6)F1 mice with positive PRL locus driver genetics⁴ respond to ICIs (figure 2D). Given our observation that MCP raises PRL and enhances ICI efficacy in B6 mice (figure 3C,E), we asked if the ICI response in (CC51xB6)F1 mice correlates with the amount of circulating PRL. Baseline serum PRL measurement in tumor-free (CC51xB6)F1 and B6 mice did not reveal any differences for female or male mice (online supplemental figure S6). Yet, depletion of pituitary-derived PRL via BRC abrogated ICI response in female (CC51xB6)F1 mice (figure 2D) which may in part be explained by a PRL-independent mechanism of BRC, the impact of local PRL concentrations, or immune cell expression of PRL within the TME.

Next, we asked if MCP could further improve ICI response in the (CC51xB6)F1 ICI responder model. B16F0-bearing (CC51xB6)F1 mice were administered 8 mg/kg MCP or NS daily with or without dual ICIs (online supplemental figure S7A). (CC51xB6)F1 mice treated with ICIs demonstrated slower tumor growth and 1/8 mice treated with both MCP and ICIs eliminated the tumor with protection from contralateral rechallenge (online supplemental figure S7B). Survival analysis showed that ICIs prolonged survival (online supplemental figure S7C). MCP with ICIs led to a longer median survival compared to ICIs alone, but this difference was not statistically significant (online supplemental figure S7C). We found enhanced CD8+ T cell abundance in ICI-treated tumors that was not enhanced with MCP (online supplemental figure S7D,E). MCP may thus confer a slight improvement in ICI efficacy in (CC51xB6)F1 mice, suggesting that PRL-dependent signaling may already be maximally activated in this genetic background.

We next compared the immune TME between the (CC51xB6)F1 and B6 models. Mice were inoculated with B16F0 and groups were divided into untreated and treated with dual ICIs (figure 5A). On day 13 after tumor inoculation, tumors were harvested, dissociated and enriched for CD45+ tumor-infiltrating leukocytes (TILs) which were then profiled by single-cell RNA sequencing (scRNA-seq). Unbiased clustering revealed 16 clusters (online supplemental figure S8A) with differences in overlap between the TILs of ICI-treated (CC51xB6)F1 and B6 mice (online supplemental figure S8B–D). Manual cell type annotation using canonical markers revealed various intratumoral immune cell subsets, including macrophages, monocytes, myeloid-derived suppressor cells, CD4+ T cells, CD8+ T cells, natural killer (NK) T cells, B cells, and dendritic cells (DCs) (figure 5B). We found that tumor immune infiltrates from ICI-treated B6 and (CC51xB6)F1 mice dominated the makeup of clusters 2, 4, and 5 which represent CD8+ T cells (online supplemental figure S8D, figure 5B). Macrophages constituted the greatest proportion of TILs regardless of genetic background and ICI treatment status (figure 5C). These tumor-infiltrating macrophages expressed low levels of proinflammatory markers Il-12β, Cd80 and Nos2, low levels of protumor marker Cd163, and high levels of protumor markers Arg1 and Mrc1 (figure 5D). This suggests that the B16F0-infiltrating macrophages are polarized toward the M2-like immunosuppressive phenotype.²⁶ Moreover, in B6 mice, there were similar proportions of tumor-infiltrating macrophages regardless of ICI treatment status (figure 5C). In contrast, the proportion of macrophages decreased with ICI therapy in (CC51xB6)F1 mice (figure 5C), consistent with our findings that ICIs enhanced intratumoral CD8+ T cell infiltration in this ICI responder model (figure 2F,G). Accordingly, ICIs led to a greater proportion of tumor-infiltrating CD8+ T cells in (CC51xB6)F1 compared with B6 mice (figure 5C). Interestingly, we found a small subset of CD8+ T cells expressing B cell-lineage marker CD20 that clustered together with B cells (figure 5B). These cells may represent terminally exhausted CD8+ T cells that have been previously shown to upregulate B cell-associated genes during chronic viral infection.²⁷

Figure 5. Tumor immune microenvironment in prolactin-locus ICI non-responder and responder models. (A) Experimental scheme. B16F0-bearing (CC51xB6)F1 and B6 mice received 100 µg αCTLA-4 and 200 µg αPD-1 on days 3, 6, and 10 after tumor inoculation. On day 13 after tumor inoculation, tumors were enriched for CD45+ TILs then processed for scRNA-seq. (B) UMAP and (C) composition analysis showing the breakdown of B6U, B6T, CC51U, CC51T annotated cell types overlaying the UMAP (B) and proportions of each cell type cluster. For each experimental condition, cell proportions (Y axis) were calculated by dividing the number of cells of a given type by the total number of cells. B6U, B6 untreated; B6T, B6 ICI-treated; CC51U, (CC51xB6)F1 untreated; CC51T, (CC51xB6)F1 ICI-treated. Cell types annotated manually using canonical markers. (D) Expression levels of M1 and M2 markers in the MΦ cluster. (E) Volcano plot highlighting DEGs in the CD20− CD8+ T cluster between B6T and CC51T. Significant DEG (padj≤0.05, |log2FC | ≥ 1) were determined using FindMarkers and are colored (red=upregulated in B6T; purple=upregulated in CC51T). (F) Expression levels of Prlr, Drd2, Htr3a and Htr4 in all clusters. αCTLA-4, anti-cytotoxic T-lymphocyte-associated protein 4; αPD-1, anti-programmed cell death protein-1; B, B cells; B16, B16 F0 cells; CD4+ T, CD4+ T cells; CD20- CD8+ T, CD8+ T cells; CD20+ CD8+ T, CD20+ CD8 T cells; cDC1, conventional dendritic cells, type I; DEGs, differentially expressed genes; Drd2, dopamine D2 receptor; Htr3a, serotonin 5HT3 receptor; Htr4, 5HT4 receptors; ICI, immune checkpoint inhibitor; MΦ, macrophages; MDSC, myeloid-derived suppressor cells; Mono, monocytes; NK T, natural killer T cells; Prlr, prolactin receptor; scRNA-seq, single-cell RNA sequencing; TILs, tumor-infiltrating leukocytes.

We then asked if there were transcriptional differences in tumor-infiltrating CD8+ T cells with ICI therapy between (CC51xB6)F1 and B6 mice. Differential gene expression analysis of the CD20− CD8+ T cell cluster revealed changes in expression of several transcripts between the two genetic backgrounds (figure 5E). CD8+ T cells from (CC51xB6)F1 mice demonstrated high expression of MHC I related genes (H2-D1, H60b) (figure 5E). Compared with B6 mice, the CD8+ T cell cluster in (CC51xB6)F1 had high expression of T cell activation marker Ly6a and IFN-γ signaling genes (Ifng, Ifi202b) (figure 5E).²⁸ On the other hand, B6 CD8+ T cells expressed high levels of prosurvival factor Il7r, homing marker Itgav, immunosuppressive markers Foxp3 and Ikzf2 (encoding Helios), B-cell associated immunoglobulin Igkc1, and T cell receptor (TCR) transcripts (Trav4, Trbv19) (figure 5E).²⁸ These findings suggest that tumor-infiltrating CD8+ T cells in B6 mice undergo TCR engagement during immune checkpoint blockade but also exhibit exhaustion and a non-T cell lineage phenotype. This contrasts with the (CC51xB6)F1 background where effector T cells produce high levels of antigen-associated and tumor-lytic genes that are crucial for effective antitumor immunity. Interestingly, we did not observe transcript-level expression of PRL, D2R, or the serotonin receptors Htr3a, Htr4 that are known targets of MCP and BRC (figure 5F).^{9 29} Low transcript-level expression of PRL receptor (PRLR) was detected in TILs and B16 cells (figure 5F). This suggests that the differential impact on tumor control during immune checkpoint blockade and co-therapy with BRC and MCP may be functioning at the protein level, operating outside of the TME, or targeting a non-canonical receptor.

MCP alters immune function in vitro

We found that MCP enhanced intratumoral CD8+ T cell abundance (figure 3H). Moreover, while MCP did not raise PRL levels in female mice (figure 3F), it ameliorated ICI efficacy (online supplemental figure S5A,C,E,G) and led to increased expression of intratumoral transcripts related to macrophage activity during immune checkpoint blockade (figure 4E). This suggests that D2R antagonism may act directly on cells within the TME through a PRL-independent mode of action to enhance antitumor immunity. To investigate the underlying mechanisms, we tested the direct impact of MCP on tumor cells and immune cells in vitro. We asked if MCP affects tumor cell expression of MHC I and II which have well-established roles in presenting peptides to CD8+ and CD4+ T cells, respectively.³⁰ B16 and MEL11443 cells were treated with MCP with or without a dose escalation of IFN-γ for 48 hours. B16 cells demonstrated enhanced expression of MHC I and II in an IFN-γ-dependent manner consistent with prior studies (figure 6A,B and ³⁰). Addition of MCP enhanced MHC I and II expression on B16 cells at low but not high IFN-γ concentrations (figure 6A,B). We also found that MCP increased MHC I on MEL11443 cells independent of IFN-γ (figure 6C). Interestingly, MCP increased MHC II expression on MEL11443 cells, whereas IFN-γ had no effect at any of the concentrations tested (figure 6D). To extend our findings to another FDA-approved pharmacologic D2R antagonist, we used haloperidol (HAL), an atypical antipsychotic with a roughly 30-fold higher affinity for D2R than MCP.^{31 32} HAL also enhanced MHC I and II expression on B16 and MEL11443 cells (online supplemental figure S9A–D). To establish that MEL11443 and B16F0 cells express the primary receptor targeted by MCP and HAL, we confirmed D2R expression at the protein level in vitro and in vivo (online supplemental figure S10A–D). To assess the presence of a tumor-intrinsic autocrine D2R-PRL signaling axis, we measured transcript-level PRL expression in B16 and MEL11443 cells treated with MCP or HAL. PRL was not endogenously expressed, nor was its expression stimulated by either D2R antagonist at the transcript level for both tumor cell lines (online supplemental figure S9E). This suggests that pharmacologic D2R antagonism does not impact tumor-derived PRL expression in vitro. Similarly, we did not detect protein-level expression of PRL in B16F0 or MEL11443 tumors in vivo (online supplemental figure S10A–C).

Figure 6. MCP enhances MHC I and II expression on tumor cells. (A, C) MHC I expression measured by flow cytometry as MFI on B16 and MEL11443 cells treated with 100 µg/mL MCP with or without 0, 2, 20, 200, 2,000, or 20,000 pg/mL IFN-γ for 48 hours in vitro. (B, D) MHC II expression. P values determined by Student’s t-test. *p<0.05. **p<0.01. ***p<0.001. ****p<0.0001. IFN-γ, interferon gamma; MCP, metoclopramide; MFI, mean fluorescence intensity; MHC, major histocompatibility complex; ns, not significant.

We next determined how MCP affected the activation of CD8+ T cells. OT-I CD8+ T cells were activated with the SIINFEKL ovalbumin_(257–264) peptide (N4) or with the reduced OT-I TCR affinity variants SIIQFEKL (Q4), SIITFEKL (T4), and SIIVFEKL (V4) (∼18-fold, ∼70-fold, ∼200-fold fold affinity reduction, respectively), with or without MCP.³³ Then, we assessed early CD8+ T cell activation by measuring CD44, CD25, CD69, and PD-1 by flow cytometry.²⁸ The reduced affinity OT-I peptide variants were selected to represent tumor antigens with differential affinities for the TCR. 48 hours after peptide stimulation, higher levels of CD44 and CD69 were observed with MCP with all peptides except for V4 (figure 7A,C). MCP led to higher CD25 expression for Q4 and T4 stimulation but did not impact PD-1 expression (figure 7B,D). Repeating this assay with HAL, we observed higher CD44 and CD25 expression during T4 and V4 stimulation and enhanced CD69 expression with all peptides (online supplemental figure S11A–C). Interestingly, in contrast to MCP, HAL increased PD1 expression during T4 and V4 stimulation (online supplemental figure S11D, figure 7D). To establish that T cells express D2R, we confirmed the presence of D2R at the protein level in vitro (online supplemental figure S10D). As observed with B16 and MEL11443 cells, we did not detect transcript-level or protein-level expression of PRL in T cells regardless of N4 peptide stimulation and co-treatment with MCP or HAL, suggesting an absence of T cell-intrinsic autocrine D2R-PRL signaling in this in vitro system (onlinesupplemental figures S10D S11E).

Figure 7. MCP enhances T cell activation and modulates macrophage polarization. (A) CD44 expression measured by flow cytometry as MFI on OT-I CD8+ T cells activated with 50 ng/mL SIINFEKL, SIIQFEKL, SIITFEKL, or SIIVFEKL with or without 100 µg/mL MCP for 48 hours in vitro. (B, C, D) CD25, CD69, PD-1 expression. IL-12β (E) and NOS2 (F) expression measured by quantitative reverse transcription PCR in bone marrow-derived M0 macrophages polarized to the M1 phenotype with or without 100 µg/mL MCP for 24 hours in vitro. (G) Arg1 expression in M0 macrophages polarized to the M2 phenotype with or without MCP for 24 hours in vitro. P values determined by Student’s t-test. *p<0.05; **p<0.01. IL, interleukin; MCP, metoclopramide; MFI, mean fluorescence intensity; PD-1, programmed cell death protein-1.

We also assessed the impact of MCP on BMDM polarization. BMDM were pretreated with MCP for 2 hours then cytokine polarized to the M1 (IFN-γ, LPS) or M2 (IL-4) phenotype overnight. M1-polarized BMDM expressed slightly higher levels of M1 markers Il-12β and Nos2, but this difference was not statistically significant (figure 7E,F).²⁶ MCP decreased expression of M2 marker Arg1²⁶ in M2-polarized BMDM (figure 7G). Together, our results demonstrate that pharmacologic D2R antagonism enhances tumor immunogenicity and effector T cell activation, and suppresses the protumor activity of primary macrophages in vitro which suggests a PRL-independent mechanism in anti-tumor immunity.

Discussion

Here we present the novel observation that pharmacologic targeting of D2R impacts ICI efficacy, the TME, and antitumor immunity in murine melanoma models B16F0 and MEL11443. BRC accelerated B16F0 progression and dampened CD8 infiltrates in ICI-treated (CC51xB6)F1 mice which are ICI responders in the PRL locus.⁴ Conversely, MCP improved tumor control in B16F0 and MEL11443 and enhanced intratumoral CD8 infiltrates when paired with ICIs in B6 mice which are historic ICI non-responders.^{4 21 22} These results suggest that D2R inhibition and ICIs may act together to enhance immune cell function to a greater degree than either intervention alone.

Given the observed impact on immune checkpoint blockade, D2R targeting could directly affect the antitumor immune response. In resting immune cells, D2R is expressed mostly on B cells and NK cells with little expression on eosinophils, neutrophils, T cells, and monocytes.³⁴ In macrophages, D2R signaling increases secretion of proinflammatory cytokines tumor necrosis factor (TNF)-α and IL-6 at low dopamine concentrations with the opposite effect at high concentrations.³⁵ Accordingly, we observed that D2R antagonism modulated the polarization of primary BMDM, shifting them away from a protumor M2 phenotype. Previous studies have shown that stimulation of D2R with dopamine promotes secretion of the immunosuppressive cytokine IL-10 in T cells, a potential mechanism that may impair antitumor immunity and potency of immunotherapy.³⁶ Consistently, we found that antagonism of D2R with MCP enhanced activation of CD8+ T cells in vitro and enhanced intratumoral CD8+ T cell abundance with ICIs in vivo. Further studies investigating how MCP enhances tumor-infiltrating CD8+ T cell activation and abundance during ICI therapy may include TCR sequencing. Enhanced TCR repertoire diversity with MCP treatment would suggest T cell recognition of a broad array of antigens, whereas an unchanged repertoire would be attributed to increased proliferation of existing antitumor CD8 T cells during immune checkpoint blockade.

The effects of MCP and BRC on ICI response may additionally be explained by indirect effects on PRL levels. PRL has been linked to several autoimmune diseases.³⁷ In patients with lupus, ex vivo depletion of PRL with anti-PRL antibodies led to reduction of CD69 and CD40L in peripheral blood mononuclear cells³⁸ indicating that PRL plays a role in antigen presentation and co-stimulation. Additionally, a single-nucleotide polymorphism in the human PRL extrapituitary promoter drives elevated expression of PRL in in vitro stimulated lymphocytes and is found to associate with lupus onset and severity.³⁹ As tumors arise from self-tissue, the mechanisms involved in autoimmunity may also play a role in immunotherapy response which relies on successful immune-mediated tumor rejection.

Beyond its involvement in autoimmune disease, PRL directly impacts the phenotype of various immune cell subsets. PRL enhances T cell priming against antigen presenting cells loaded with tumor-associated carcinoembryonic antigen in vitro in a bimodal manner where a low dose (25 ng/mL) enhanced priming whereas a high dose (200 ng/mL) demonstrated the opposite effect.⁴⁰ This suggests that the benefit to ICI efficacy may exist with mild rather than severe hyperprolactinemia and is consistent with our observation that MCP led to a roughly 50% increase of serum PRL from 10 ng/mL to 15 ng/mL (figure 3G). Additionally, the relationship between PRL and regulatory T cells (Tregs) is complex. In murine 4T1 breast cancer, knockdown of the long-form PRLR reduced intratumoral Treg recruitment and their immunosuppressive activity.⁴¹ Conversely, ex vivo addition of PRL reduced the ability of human Tregs to suppress effector T cell proliferation.⁴² These findings suggest that PRL’s influence on Tregs and whether it promotes immunosuppression or immune activation may be context-dependent. DCs treated with PRL upregulate expression of MHC I, MHC II, and CD40 as well as proinflammatory cytokines IL-12β, TNF-α, and IL-1β.^{43 44} Furthermore, PRL enhances the phagocytic activity of macrophages in vitro.⁴⁵ Thus, our findings are consistent with a model where elevated PRL may improve the function of various immune cell subsets and lead to tumor control.

Our observation that BRC blunted ICI efficacy in female but not male mice supports a growing body of evidence that biological sex affects immune response. A recent study highlighting female sex-based autoimmunity showed that expression of the ribonucleoprotein complex containing the female sex-specific Xist long non-coding RNA in male mice led to formation of autoantibodies which was validated in patients with lupus.⁴⁶ In humans, females have higher CD4+ T cell counts and CD4+/CD8+ T cell ratios whereas males have higher CD8+ T cell frequencies.⁴⁷ Furthermore, ex vivo PMA/ionomycin-stimulated T cells from females have greater cytotoxic activity than males with half of the activated genes in female T cells having estrogen response elements in their promoters.⁴⁸ Given that females have higher PRL levels than males, our findings imply that females and males may have differential sensitivity to changes in serum PRL levels which may in part contribute to sex-based differences in ICI efficacy. In our study, we noted sex-based differences in tumor control with MCP and ICIs in B16F0, but not MEL11443, which may be attributable to Y chromosome proteins acting as neoantigens in B16F0-bearing female mice.

Beyond immune cells, direct effects on the tumor itself may also play a role. Studies in murine xenograft models show that D2R inhibition on its own was sufficient to reduce growth.^{49 50} We found that MCP-mediated D2R blockade enhanced expression of MHC I and II on B16F0 and MEL11443 cells which may serve to enhance the oncolytic capacity of tumor-infiltrating T cells. However, the mechanistic relationship between D2R signaling, tumor growth, and antitumor immunity is not well characterized for any cancer type, including melanoma. Under stress-simulated conditions, B16F10 murine melanoma cells express high levels of D2R which promote tumor progression, suggesting that these cells may be sensitive to D2R blockade.⁵¹ Interestingly, we did not detect transcript-level expression of D2R in TILs or tumor cells in our scRNAseq data, yet we did observe protein-level D2R in both melanoma cell lines and ex vivo tumor tissue. This discrepancy could be due to limitations in detection by scRNA-seq. Further studies with conditional knockout of D2R in melanoma cells and various immune cell subsets are warranted to further dissect its role in tumor progression and immune modulation. Additionally, investigations into how D2R modulation affects classic PRL and D2R signaling proteins such as JAK2, STAT5, ERK, and MAPK will yield deeper mechanistic insight.^{52 53}

A key limitation of our study is the reliance on subcutaneous tumor models, which may not truly recapitulate the native TME of cutaneous melanoma. Because subcutaneous tumors lack the stromal structure, vascular organization, and immune landscape present in orthotopic lesions, these models may influence both therapeutic responses and patterns of immune infiltration.^{21 54 55} Nonetheless, D2R expression has been detected in cutaneous and uveal melanoma which raises the possibility that human melanoma may respond to D2R inhibition.^{56 57} The extent to which D2R activity affects the progression and treatment of human melanoma as well as immunotherapy response remains to be completely elucidated. Moreover, the mechanistic link between D2R and ICI resistance will be investigated in future studies. Currently there are multiple FDA-approved D2R antagonists, which opens the opportunity for clinical exploration of combination therapy to improve ICI response in patients with advanced melanoma.

Data availability statement

All data relevant to the study are included in the article or uploaded as supplementary information.