Serotonin transporter inhibits antitumor immunity through regulating the intratumoral serotonin axis

Abstract

Identifying additional immune checkpoints hindering antitumor T cell responses is key to the development of next-generation cancer immunotherapies. Here, we report the induction of serotonin transporter (SERT), a regulator of serotonin levels and physiological functions in the brain and peripheral tissues, in tumor-infiltrating CD8 T cells. Inhibition of SERT using selective serotonin reuptake inhibitors (SSRIs), the most widely prescribed antidepressants, significantly suppressed tumor growth and enhanced T cell antitumor immunity in various mouse syngeneic and human xenograft tumor models. Importantly, SSRI treatment exhibited significant therapeutic synergy with programmed cell death protein 1 (PD-1) blockade, and clinical data correlation studies negatively associated intratumoral SERT expression with patient survival in a range of cancers. Mechanistically, SERT functions as a negative-feedback regulator inhibiting CD8 T cell reactivities by depleting intratumoral T cell-autocrine serotonin. These findings highlight the significance of the intratumoral serotonin axis and identify SERT as an immune checkpoint, positioning SSRIs as promising candidates for cancer immunotherapy.

Keywords: PD-1 blockade; SSRIs; T cell antitumor immunity; antidepressants; cancer immunotherapy; immune checkpoint; intratumoral serotonin axis; serotonin; serotonin transporter; tumor microenvironment.

Figure 1.. SERT blockade suppresses tumor growth and enhances cytotoxic CD8 T cell antitumor responses in multiple syngeneic mouse tumor models

(A) Quantitative reverse-transcription PCR (RT-qPCR) analyses of Sert mRNA expression in tumor-infiltrating CD8 T cell subsets (gated as CD45.2⁺TCRβ⁺CD8⁺PD-1^lo or CD45.2⁺TCRβ⁺CD8⁺PD-1^hi) isolated from day 14 B16-OVA tumors grown in WT B6 mice (n = 3). Naive CD8 T cells (gated as TCRβ⁺CD8⁺CD44^loCD62L^hi) sorted from the spleen of tumor-free B6 mice were included as a control (n = 3). (B) Schematics showing the intratumoral serotonin axis in analogy to the serotonin axis in the neuron. In an antitumor CD8 T cell, TPH1 converts Trp into 5-HT, which is secreted into the tumor microenvironment (TME) and in turn stimulates the T cell by binding to its surface 5-HTR. SERT depletes 5-HT from the TME by transporting it back into the T cell, where it is degraded by MAO-A into 5-HIAL. SSRIs can inhibit SERT activity, increasing extracellular 5-HT and preventing degradation. TPH1/2, tryptophan hydroxylase 1/2; Trp, tryptophan; 5-HT, 5-hydroxytryptamine or serotonin; 5-HTR, 5-HT receptor; SERT, serotonin transporter; MAO-A, monoamine oxidase A; 5-HIAL, 5-hydroxyindolealdehyde; SSRIs, selective serotonin reuptake inhibitors; TCR, T cell receptor; MHC, major histocompatibility complex; DC, dendritic cell; TAM, tumor-associated macrophage. (C and D) SSRI treatment in tumor prevention experiments. (C) Experimental design. Six syngeneic mouse tumor models and two SSRIs, fluoxetine (FLX; trade name Prozac) and citalopram (CIT; trade name Celexa), were used. s.c., subcutaneous. (D) Tumor growth and animal survival (n = 5). NT, non-treated. (E–H) SSRI treatment in tumor therapy experiments. (E) Experimental design. Two syngeneic mouse tumor models (B16-OVA and MC38) and two SSRIs (FLX and CIT) were used. (F) Tumor growth and animal survival (n = 5). (G and H) Fluorescence-activated cell-sorting (FACS) analysis of intracellular IFN-γ (G) and Granzyme B (H) production in tumor-infiltrating CD8 T cells isolated from B16-OVA tumors at day 14 (n = 3). MFI, median fluorescence intensity. (I and J) SSRI and anti-PD-1 combination treatment in tumor therapy experiments. (I) Experimental design. Two syngeneic mouse tumor models (B16-OVA and MC38) and one SSRI (FLX) were used. (J) Tumor growth and animal survival (n = 5). Iso, isotype control; αPD-1, anti-PD-1. Representative of two (A, D, G, and H) and three (F and J) experiments. Data are presented as the mean ± SEM. ns, not significant, *p < 0.05, **p < 0.01, and ***p < 0.001 by one-way ANOVA (A, G, and H) or log rank (Mantel-Cox) test adjusted for multiple comparisons (D, F, and J). See also Figures S1 and S7.

Figure 2.. SERT blockade enhances antitumor CD8 T cell effector and proliferating gene profiles

(A) Schematics showing the experimental design to study the in vivo gene profiling of antitumor CD8 T cells using scRNA-seq. CD45⁺ tumor-infiltrating immune cells were sorted from day 10 B16-OVA tumors and then subjected to scRNA-seq analysis. Three experimental groups were included: non-treated (NT), FLX-treated (FLX), and anti-PD-1-treated (αPD-1). (B) scRNA-seq analysis of the total CD45⁺ tumor-infiltrating immune cells combined from all samples. Combined uniform manifold approximation and projection (UMAP) plot is presented, showing the formation of nine major cell clusters. Each dot represents a single cell and is colored according to its cell cluster assignment. Mono, monocytes; NK, natural killer; Th, T helper; Treg, CD4 regulatory T. (C–J) scRNA-seq analysis of the antigen-experienced (CD44⁺) tumor-infiltrating CD8 T cells identified from (B). (C) Combined UMAP plot showing the formation of three major cell clusters. Each dot represents a single cell and is colored according to its cell cluster assignment. Gene signature profiling analysis identified cluster 1 to be the effector/proliferating CD8 T cells, cluster 2 to be the progenitor exhausted CD8 T cells, and cluster 3 to be the terminally exhausted CD8 T cells. (D) Individual UMAP plots showing the three-cell cluster composition of the indicated treatment groups. (E) Bar graphs showing the cell cluster proportions from (D). (F) Violin plots showing the expression distribution of the indicated gene signatures in each treatment group. Box and whisker plots exhibit the minimum, lower quartile, median, upper quartile, and maximum expression levels of each group. (G) Dot plots showing the expression of representative signature genes in each treatment group. Color saturation indicates the strength of averaged gene expression. The dot size indicates the percentage of cells expressing the indicated genes. (H) RNA velocity projected on UMAP plots. Arrows represent the estimates of local average velocity, showing the path (indicated by arrow orientation) and pace (indicated by arrow length) of cell transition. (I) Venn diagram showing numbers of genes upregulated by FLX or anti-PD-1 treatment. (J) Bar plots showing the fold enrichment of indicated pathways upregulated by FLX or anti-PD-1 treatment. The experiment was performed once, and cells isolated from 10 mice of each experimental group were combined for analysis. The p values of violin plots were determined by the Kruskal-Wallis test for the overall comparison and Dunn’s test for post hoc pairwise comparisons between groups (F). p < 0.05 was considered significant. ns, not significant. See also Figures S2 and S7.

Figure 3.. SERT functions as a T cell-intrinsic factor negatively regulating CD8 T cell-mediated antitumor responses

(A–D) Sert-WT and Sert-KO mice tumor challenge experiments. (A) Experimental design. (B) Tumor growth (n = 4–6). (C and D) FACS analyses of the numbers(C) and intracellular Granzyme B production (D) of tumor-infiltrating CD8 T cells at day 14 (n = 4). (E–I) Bone marrow (BM) transfer experiments. (E) Experimental design. (F) Tumor growth (n = 8–13). (G and H) FACS analyses of intracellular IFN-γ (G) and Granzyme B (H) production in tumor-infiltrating CD8 T cells isolated from day 21 tumors (n = 5). (I) FACS analyses of PD-1 expression on tumor-infiltrating CD8 T cells isolated from day 21 tumors (n = 7–8). (J and K) CD8 T cell depletion experiments. (J) Experimental design. (K) Tumor growth (n = 4). αCD8, anti-CD8. (L–N) OT1 T cell adoptive transfer experiment. (L) Experimental design. i.v., intravenous. (M) Tumor growth (n = 6). (N) FACS analyses of surface PD-1 expression on tumor-infiltrating CD8 T cells isolated from day 17 B16-OVA tumors (n = 6). Representative of one (M and N), two (B–D and F–I), and three (K) experiments. Data are presented as the mean ± SEM. ns, not significant, *p < 0.05, **p < 0.01, and ***p < 0.001 by Student’s t test (B–D, F–I, M, and N) or one-way ANOVA (K). See also Figure S3.

Figure 4.. SERT acts as an autonomous factor negatively regulating CD8 T cell antigen responses

(A–G) CD8 T cell antigen response in the absence of SERT. (A) Experimental design. Naive CD8 T cells were purified from Sert-WT and Sert-KO B6 mice and stimulated in vitro with anti-CD3 over 4 days. (B) Cell counts at day 4 (n = 4). (C–E) RT-qPCR analyses of Il2 (C), Ifng (D), and Gzmb (E) expression over time (n = 4). (F and G) FACS analyses of intracellular IFN-γ (F) and Granzyme B (G) production at day 3 (n = 3). αCD3, anti-CD3. (H–L) CD8 T cell antigen response under SSRI treatment. (H) Experimental design. Naive CD8 T cells were purified from Sert-WT B6 mice and stimulated in vitro with anti-CD3 over 4 days in the presence or absence of SSRI (CIT or FLX) treatment. NT, non-treated. (I) Cell counts at day 3 (n = 4). (J) RT-qPCR analyses of effector gene (i.e., Il2, Ifng, Tnf, Gzmb, Prf1) expression at day 2 (n = 4). (K and L) FACS analyses of intracellular IFN-γ (K) and Granzyme B (L) production at day 3 (n = 3). Representative of two (B–G and I) and three (J–L) experiments. Data are presented as the mean ± SEM. *p < 0.05, **p < 0.01, and ***p < 0.001 by Student’s t test (B–G) or one-way ANOVA (I–L). See also Figure S4.

Figure 5.. SERT restrains CD8 T cell antigen responses by directly regulating the autocrine serotonin signaling pathway

(A) Schematics showing the proposed autocrine serotonin signaling pathway in a CD8 T cell. Possible pharmacological interventions are indicated. (B and C) Serotonergic gene expression in Sert-WT CD8 T cells in response to antigen stimulation. Naive CD8 T cells were purified from Sert-WT B6 mice and stimulated with anti-CD3 for 3 days. (B) RT-qPCR analyses of 5-HTR family member gene expression at day 1 (n = 4). Unstimulated naive CD8 T cells were included as a control. (C) RT-qPCR analyses of Sert, Tph1, and Maoa gene expression over time (n = 4). (D–H) Serotonergic gene expression in activated Sert-WT and Sert-KO CD8 T cells. (D) Experimental design. Naive CD8 T cells were purified from Sert-WT and Sert-KO B6 mice and stimulated in vitro with anti-CD3 for 1 day. (E–H) RT-qPCR analyses of Tph1 (E), Maoa (F), Htr2b (G), and Htr7 (H) expression (n = 4). (I–M) Autocrine serotonin signaling in Sert-WT CD8 T cells. (I) Experimental design. Sert-WT CD8 T cells were stimulated with anti-CD3 for 3 days in serotonin-depleted medium in the presence or absence of SSRI (FLX or CIT) and/or 5-HTR antagonist (ASE or RS-127445) treatment. ASE, asenapine (a general antagonist of most 5-HTR subtypes); RS-127445, a 5-HTR2B selective antagonist. (J and K) RT-qPCR analyses of Il2 (J) and Ifng (K) expression in FLX-treated or non-treated (NT) CD8 T cells at day 2, with or without ASE treatment (n = 3). (L) RT-qPCR analyses of Il2 expression in FLX-treated or non-treated (NT) CD8 T cells at day 2, with or without RS-127445 treatment (n = 3). (M) ELISA analyses of serotonin levels over time in culture supernatants of FLX-treated, CIT-treated, or non-treated (NT) CD8 T cells (n = 4). (N and O) Western blot analyses of key signaling molecules involved in the 5-HTR-MAPK (N) and TCR (O) signaling pathways. MAPK, mitogen-activated protein kinase. (P–R) Serotonin levels in Sert-WT and Sert-KO BMT mice bearing B16-OVA tumors (denoted as WT and KO, respectively). (P) Experimental design. (Q and R) HPLC analyses of serotonin levels in tumor (Q, n = 4) and serum (R, n = 5–6) at day 21. (S–U) Serotonin levels in Sert-WT mice bearing B16-OVA tumors, with or without SSRI (i.e., FLX) and/or anti-CD8 depletion antibody treatment. (S) Experimental design. (T and U) HPLC analyses of serotonin levels in tumor (T, n = 7–8) and serum (U, n = 4–5) at day 14. Representative of two (B, E–H, J–M, N, and O) and three (C, Q, R, T, and U) experiments. Data are presented as the mean ± SEM. ns, not significant, *p < 0.05, **p < 0.01, and ***p < 0.001 by Student’s t test (B, E–H, Q, and R), one-way ANOVA (C and M), or two-way ANOVA with Turkey’s multiple comparisons test (J–L, T, and U). See also Figure S5.

Figure 6.. SERT blockade for cancer immunotherapy: Human T cell and clinical data correlation studies

(A–D) Studying human CD8 T cell antigen responses under SSRI treatment. (A) Experimental design. Human naive CD8 T cells were sorted from healthy donor PBMCs and stimulated with anti-CD3/anti-CD28/IL-2 in vitro for 5 days in the absence (non-treated, NT) or presence of SSRI (FLX or CIT) treatment. PBMCs, peripheral blood mononuclear cells; αCD28, anti-CD28. (B) RT-qPCR analyses of the serotonergic gene expression in the indicated CD8 T cells at day 1 (n = 3). Unstimulated naive CD8 T cells were included as a control. (C) Cell counts at day 3 (n = 3). (D) RT-qPCR analyses of effector genes (i.e., IL2, IFNG, and GZMB) at day 3 (n = 4). (E–I) Studying the gene profile of human tumor-infiltrating CD8 T cells. (E) Experimental design. Eight scRNA-seq datasets across seven cancer types (SRA: PRJNA705464; EGA: EGAS00001004809; GEO: GSE123813, GSE164522, GSE179994, GSE181061, GSE200996, and GSE212217) were retrieved from the “uTILity” human TIL scRNA-seq database and combined for the analysis. (F) Combined UMAP plot showing the formation of six major cell clusters of human tumor-infiltrating CD8 T cells. Each dot represents a single cell and is colored according to its cell cluster assignment. TCM, central memory T; TEM, effector memory T; TEMRA, terminally differentiated effector memory CD45RA re-expressing T; TPEX, progenitor exhausted T; TEX, exhausted T. (G) Heatmap showing gene expression in the indicated CD8 T cell clusters. (H) Violin plots showing the expression distribution of the serotonergic gene signature in each CD8 T cell cluster. (I) Heatmap displaying the expression of representative serotonin pathway genes selected from the serotonergic gene signature shown in (H). (J) RT-qPCR analyses of the representative serotonin pathway genes in human CD8 T cells sorted from healthy donor PBMCs and stimulated in vitro without (non-treated, NT) or with SSRI (FLX or CIT) for 3 days (n = 3–4). (K–P) Studying SERT blockade therapy in an A375 human melanoma xenograft model. (K) Schematics showing a human tumor-T cell pair designated for this study. A375-A2-ESO-FG, human A375 melanoma cell line engineered to express the tumor antigen NY-ESO-1, its matching MHC molecule (HLA-A2), and a dual-reporter comprising a firefly luciferase and an enhanced green fluorescence protein (FG). ESO-T, human CD8 T cell engineered to express an NY-ESO-1 antigen-specific TCR. ESOp, NY-ESO-1 peptide. (L) Experimental design. (M) Tumor growth (n = 4). (N–P) FACS analyses of tumor-infiltrating human CD8 T cell numbers(N) and intracellular production of IFN-γ (O) and Granzyme B (P) (n = 3–4). (Q) Clinical data correlation studies. Kaplan-Meier plots are presented, showing the association between SERT expression in tumor and survival of cancer patients in a melanoma cohort (Prediction of Clinical Outcomes from Genomic Profiles [PRECOG]: GSE8401, n = 67), a breast cancer cohort (Molecular Taxonomy of Breast Cancer International Consortium [METABRIC], n = 233), a lung cancer cohort (The Cancer Genome Atlas [TCGA], n = 484), a kidney cancer cohort (TCGA, n = 256), and a sarcoma cohort (TCGA, n = 258). Representative of two (B–D, J, and M–P) experiments. Data are presented as the mean ± SEM. ns, not significant, *p < 0.05, **p < 0.01, and ***p < 0.001 by Student’s t test (B and M–P), one-way ANOVA (C, D, and J), or two-sided Wald test in a Cox-PH regression (Q). p values of violin plots were determined by the Kruskal-Wallis test for the overall comparison and Dunn’s test for post hoc pairwise comparisons between groups (H). See also Figure S6.

Figure 7.. Working model of intratumoral serotonin axis regulation of CD8 T cell anti-tumor immunity

Schematics are presented, showing an intratumoral serotonin axis regulation of CD8 T cell antitumor immunity. In this model, SERT restrains CD8 T cell antitumor responses by inhibiting the CD8 T cell-autocrine 5-HT signaling pathway in a solid tumor. CD8 T cells are major producers of 5-HT (or serotonin) in the tumor microenvironment (TME). Upon recognition of tumor antigen, tumor-infiltrating CD8 T cells upregulate TPH1, which synthesizes 5-HT followed by releasing it into the TME to enhance T cell activation via 5-HT signaling. Meanwhile, tumor-infiltrating CD8 T cells also upregulate SERT, which acts in a negative-feedback loop to downregulate T cell activation by terminating 5-HT signaling via the reuptake of extracellular 5-HT from the TME. Blocking SERT activity using established SSRI antidepressants accumulates 5-HT in the TME, leading to the activation of the 5-HTR-MAPK-TCR signaling pathway and enhancement of CD8 T cell antitumor reactivities. This local serotonin accumulation induced by SSRIs resembles the effect observed in neuronal tissues, where SSRIs block the reuptake of serotonin by the presynaptic neuron, thereby accumulating serotonin secreted by the presynaptic neuron and stimulating neuronal activity. Note SSRIs exhibit different effects on systemic serotonin, where SSRIs block platelet uptake of gut enterochromaffin cell (EC)-produced serotonin and deplete serum serotonin.