5-HT orchestrates histone serotonylation and citrullination to drive neutrophil extracellular traps and liver metastasis

Abstract

Serotonin (5-HT) is a neurotransmitter that has been linked to tumorigenesis. Whether and how 5-HT modulates cells in the microenvironment to regulate tumor metastasis is largely unknown. Here, we demonstrate that 5-HT was secreted by neuroendocrine prostate cancer (NEPC) cells to communicate with neutrophils and to induce the formation of neutrophil extracellular traps (NETs) in the liver, which in turn facilitated the recruitment of disseminated cancer cells and promoted liver metastasis. 5-HT induced histone serotonylation (H3Q5ser) and orchestrated histone citrullination (H3cit) in neutrophils to trigger chromatin decondensation and facilitate the formation of NETs. Interestingly, we uncovered in this process a reciprocally reinforcing effect between H3Q5ser and H3cit and a crosstalk between the respective writers enzyme transglutaminase 2 (TGM2) and peptidylarginine deiminase 4 (PAD4). Genetic ablation or pharmacological targeting of TGM2, or inhibition of the 5-HT transporter (SERT) with the FDA-approved antidepressant drug fluoxetine reduced H3Q5ser and H3cit modifications, suppressed NET formation, and effectively inhibited NEPC, small-cell lung cancer, and thyroid medullary cancer liver metastasis. Collectively, the 5-HT-triggered production of NETs highlights a targetable neurotransmitter/immune axis that drives liver metastasis of NE cancers.

Keywords: Cell biology; Oncology; Prostate cancer.

Figure 1. Neutrophil infiltration and NETs are detected in NEPC liver metastasis.

(A) Immune cell profiles of liver metastases (metas) based on the SU2C PCa dataset (n = 26). Neutro, neutrophils; Mac-M1, M1 macrophages; Mac-M2, M2 macrophages; Mono, monocytes. (B) Neutrophils are the most enriched immune cell component in liver metastases based on the SU2C dataset (adrenal metastases, n = 2; bone metastases, n = 82; liver metastases, n = 26; lymph node metastases, n = 79; prostate tumor, n = 5; soft tissue, n = 14). (C and D) Flow cytometric plots (C) and quantification (D) of CD11b⁺Ly6G⁺ neutrophils in liver (n = 4 mice). (E and F) IHC and IF staining (E) and quantification (F) of NETs in liver (n = 5 mice). Scale bars: 50 μm; 25μm (inserts); original magnification, ×9 (inserts). (G and H) IF staining (G) and quantification (H) of NETs (H3cit⁺) in MPO⁺ neutrophils in livers from vehicle- and DNase-I–treated NEPC tumor–bearing mice (n = 6 mice). Scale bar: 50 μm. (I and J) H&E-stained images (I) of livers from vehicle- and DNase-I–treated NEPC tumor–bearing mice (scale bars: 4 mm) and quantification (J) of liver metastases following vehicle or DNase-I treatment of mice inoculated by i.v. injection of Rb1^Δ/Δ Trp53^Δ/Δ organoids (n = 6 mice). (K) Survival curves for vehicle- and DNase-I–treated NEPC tumor–bearing mice (n = 8 mice). **P < 0.01, by log-rank test. (L) Quantification of migrated Rb1^Δ/Δ Trp53^Δ/Δ PCa cells in Boyden chambers recruited by murine primary neutrophils or NETs (n = 4 biological replicates). (M) Quantification of Rb1^Δ/Δ Trp53^Δ/Δ PCa cells adhesion in murine primary neutrophils or NETs (n = 4 biological replicates). *P < 0.05, **P < 0.01, and ***P < 0.001, by 2-tailed Student’s t test (H, J, L, and M), log-rank test (K), 1-way ANOVA followed by Tukey’s test (D and F), and 1-way ANOVA followed by Dunnett’s test (B). Data are shown as the mean ± SEM.

Figure 2. NEPC-derived 5-HT potentiates the formation of NETs and liver metastasis.

(A and B) IF staining images (A) and quantification results (B) showing NET formation in murine BMDNs in response to 5-HT or vehicle control treatment (n = 10 biological replicates). PMA-induced NET formation and served as a positive control. Scale bars: 50 μm. (C and D) IF staining images (C) and quantification results (D) showing NET formation in murine BMDNs upon the treatment of CM collected from scrambled- and shTph1-infected Rb1^Δ/Δ Trp53^Δ/Δ organoids (n = 9 biological replicates). Scale bars: 50 μm. (E and F) IF images (E) and quantification (F) of the NET-forming capacity of human PBDNs upon 5-HT or vehicle control treatment (n = 10 biological replicates). Scale bars: 50 μm. (G) MNase digestion assay showing that addition of 5-HT induced decondensed NET chromatin in HL-60 granulocytes compared with calcium ionophore treatment. (H–J) In vivo experiments demonstrating a reduction of liver metastatic burden (H) in shTph1-infected versus scrambled shRNA–treated mice, and a regained liver metastatic burden in 5-HTP–treated, shTph1–infected Rb1^Δ/Δ Trp53^Δ/Δ organoid–implanted mice via i.v. injection (n = 7 mice, each group). H&E staining (I) and quantification data (J) validated the significant decrease in liver metastatic foci in Tph1-KD Rb1^Δ/Δ Trp53^Δ/Δ organoid–inoculated mice and regained liver metastatic foci numbers in 5-HTP–treated mice. Scale bars: 2 mm. (K and L) IF images of liver sections (K) and quantification results (L) revealing a significant decline in NET formation in Tph1-KD Rb1^Δ/Δ Trp53^Δ/Δ organoid–inoculated mice (n = 7 mice, each group). *P < 0.05, **P < 0.01, and ***P < 0.001, by Kruskal-Wallis test followed by Dunnett’s test (B, D, F, and L) and 1-way ANOVA followed by Tukey’s test (J). Data are presented as the mean ± SEM.

Figure 3. 5-HT induces histone serotonylation in neutrophils during NET formation.

(A) Schematic showing 5-HT–activated 5-HT/HTR signaling, SERT-mediated 5-HT intake, TPH1-catalyzed 5-HT-biosynthesis, and TGM2-catalyzed H3Q5ser. (B and C) IF staining images (B) and quantification results (C) demonstrate that addition of 5-HT promoted H3Q5ser modification in murine BMDNs and that the SERT inhibitor Fluox and the TGM2 inhibitor LDN compromised 5-HT–induced H3Q5ser (n = 10 biological replicates). Scale bars: 50 μm. (D–G) IF staining results (D and F) and quantification data (E and G) revealed that Fluox and LDN abrogated 5-HT–induced NET formation, as reflected by SytoxGreen (D and E) and H3cit (F and G) signals (n = 10 biological replicates per experiment). Scale bars: 50 μm. (H) The MNase digestion assay showed that LDN treatment of HL-60 granulocytes led to delayed chromatin decondensation compared with vehicle control upon stimulation with the calcium ionophore. (I–K) Tgm2^–/– mouse–derived BMDNs showed significantly reduced NET formation upon 5-HT stimulation, as exemplified by reduced SytoxGreen (I and J) and H3cit (I and K) signals compared with WT mouse–derived BMDNs (n = 10 biological replicates per experiment). Scale bars: 50 μm. *P < 0.05, **P < 0.01, and ***P < 0.001, by Kruskal-Wallis test followed by Dunnett’s test (C, G, J, and K) and 1-way ANOVA followed by Tukey’s test (E). Data are presented as the mean ± SEM.

Figure 4. H3Q5ser modification promotes the formation of NETs.

(A) Schematic diagram showing that 5-HT-mediated histone serotonylation promotes a decondensed chromatin state and facilitates the formation of NETs. (B) Immunoblot results show reduced levels of H3Q5ser and H3Cit modifications in H3.3(Q5A) mutant–transfected versus H3.3(WT)-transfected HL-60 granulocytes. (C–F) IF staining images (C) and quantification data (D–F) showing reductions in H3Q5ser (D) and NET formation in H3.3(Q5A) mutant–transfected HL-60 granulocytes, as reflected by decreased H3cit (E) and SytoxGreen (F) signals (n = 10 biological replicates per experiment). Scale bars: 50 μm (C). (G) MNase assay data revealed that chromatin in H3.3(Q5A)-transfected HL-60 granulocytes was less accessible than that in H3.3(WT)-transfected cells in response to calcium ionophore induction. **P < 0.01 and ***P < 0.001, by 2-tailed Student’s t test (D–F). Data are presented as the mean ± SEM.

Figure 5. TGM2 inhibition abrogates NET formation and liver metastasis in NEPC mouse models.

(A) The TGM2 inhibitor LDN suppressed the liver metastatic burden in mice inoculated by i.v. injection with Rb1^Δ/Δ Trp53^Δ/Δ NEPC organoids (n = 9 mice, each group). (B) H&E staining showing reduced liver metastatic foci in response to treatment with LDN. Scale bars: 4 mm. (C and D) Quantification of liver weights (C) and liver metastatic foci numbers (D) in Rb1^Δ/Δ Trp53^Δ/Δ organoid–inoculated mice upon treatment with vehicle or LDN (n = 9 mice, each group). (E and F) IF images (E) and quantification results (F) showing decreased histone serotonylation (H3Q5ser⁺) in hepatic neutrophils (MPO⁺) in Rb1^Δ/Δ Trp53^Δ/Δ–inoculated mice upon treatment with vehicle or LDN (n = 7 biological replicates). Scale bars: 50 μm. (G and H) IF images (G) and quantification (H) of neutrophil-derived (MPO⁺-derived) NETs (H3cit) in Rb1^Δ/Δ Trp53^Δ/Δ–inoculated mice in response to vehicle or LDN treatment (n = 7 biological replicates). Scale bars: 50 μm. *P < 0.05, **P < 0.01, and ***P < 0.001, by Mann-Whitney U test. Data are presented as the mean ± SEM.

Figure 6. TGM2 collaborates with PAD4 to coordinate histone serotonylation and citrullination.

(A) Schematic diagram showing proximal locations of TGM2-catalyzed H3Q5ser and PAD4-mediated H3R2,8,17cit and PAD2-promoted H3R26cit on histone H3. (B) Immunoblot assays revealed increased H3cit and H3Q5ser levels in HL-60–derived granulocytes upon 5-HT stimulation. (C) Immunoblot data revealed reductions in H3cit and H3Q5ser modifications in either shPAD4 or shTGM2-HL-60 cells compared with scrambled shRNA–transfected cells. (D and E) The MNase assay data demonstrated that KD of either TGM2 (D) or PAD4 (E) in HL60 cells attenuated calcium ionophore–induced chromatin decondensation. (F) Dual expression of HA-tagged TGM2 and Flag-tagged PAD4 in HEK-293T cells showing elevated levels of both H3cit and H3Q5ser modifications. (G) Expression of the enzyme-dead mutant of TGM2-C277S and PAD4-C645S in HEK-293T cells abrogates their synergistic effect in enhancing H3cit and H3Q5ser depositions. (H and I) The co-IP assay showed an exogenous interaction between TGM2 and PAD4 by IP Flag-tagged PAD4 (H) and HA-tagged TGM2 (I) in HEK-293T cells. (J and K) Pulldown assay demonstrates the protein-protein association between TGM2 and PAD4 in vitro.

Figure 7. H3Q5ser and H3cit are mutually enhanced and share chromatin occupancy on a genome-wide scale.

(A) Immunoblotting assay demonstrates that the H3.3(Q5A) mutant leads to deficient H3Q5ser modification and a reduced H3cit level. (B) Immunoblotting data show that either H3.3(R2A), H3.3(R8A), H3.3(R17A) mutants alone or in combination H3.3(R2,8,17A) result in deficient H3cit modification and a repressed H3Q5ser level. (C and D) Mutant H3.3(Q5A) (C) or H3.3(R2,8,17A) (D) attenuates the recruitment of each other’s epigenetic writer, as exemplified by deficient binding of either PAD4 (C) or TGM2 (D) to histone H3. (E and F) In vitro catalytic assay showing that H3Q5ser-modified peptides enhanced PAD4-mediated H3cit compared with the unmodified H3 peptide control. (G and H) In vitro catalytic assay revealing that H3cit-modified peptides promoted TGM2-mediated H3Q5ser compared with the unmodified H3 peptide control. (I) Venn diagrams show the overlapped H3cit and H3Q5ser peaks in HL-60 granulocytes from CUT&Tag sequencing data (2 independent experiments). (J) Heatmap showing the genome-wide Spearman’s correlation between the H3Q5ser and H3cit modifications (2 independent experiments). (K and L) Heatmaps of H3cit (K) and H3Q5ser (L) peaks are shown on a genome-wide scale in HL-60 granulocytes (2 independent experiments). (M) CUT&Tag profiles of co-occupied NET-related genes including ITGB2, DNASE1, ITGAM, RIPK1, MPO, and AKT1 of H3cit and H3Q5ser modifications in HL-60 granulocytes (2 independent experiments).

Figure 8. The SERT inhibitor Fluox represses NET formation and liver metastasis in NE cancers.

(A–C) The SERT inhibitor Fluox suppressed liver metastasis (A) in mice i.v. inoculated with Rb1^Δ/Δ Trp53^Δ/Δ, as exemplified by decreased liver weights (B) and number of metastasis foci (C) compared with vehicle-treated mice (n = 5 mice per group). (D and E) IF staining images (D) and quantification results (E) showing decreased histone serotonylation (H3Q5ser⁺) in hepatic neutrophils (MPO⁺) in Rb1^Δ/Δ Trp53^Δ/Δ–inoculated mice upon treatment with Fluox as compared with vehicle treatment (n = 5 mice per group). Scale bars: 1.5 mm. (F and G) IF images (F) and quantification results (G) of neutrophil-derived (MPO⁺-derived) NETs (H3Cit) in Rb1^Δ/Δ Trp53^Δ/Δ–inoculated mice in response to vehicle or Fluox treatment (n = 5 mice per group). Scale bars: 50 μm. (H–J) H&E-stained images (H) and quantification results showing decreased metastasis foci numbers (I) in i.v. TT cell–inoculated mice following treatment with Fluox compared with their vehicle-treated counterparts (n = 8 mice per group). Scale bars: 1 mm. (K and L) IF staining images (K) and quantification results (L) showing decreased H3Q5ser levels in MPO+ neutrophils in the liver sections of TT cell-inoculated mice upon treatment with Fluox versus treatment with vehicle (n = 8 mice per group). Scale bars: 50 μm. (M and N) IF images (M) and quantification (N) of NETs in TT-inoculated mice in response to vehicle or Fluox treatment (n = 8 mice per group). Scale bars: 50 μm. *P < 0.05, **P < 0.01, and ***P < 0.001, by 2-tailed Student’s t test (B, C, E, J, L, and N) and Mann-Whitney U test (G and I). Data are presented as the mean ± SEM.