Meta-analysis was performed to examine the HTR mRNA expression levels on 1036 cancer cell lines housed in the Cancer Cell Line Encyclopedia (CCLE) (www.broadinstitute.org/ccle/home) (27). Normalized heatmap data was generated in Cluster 3.0 software (http://bonsai.hgc.jp/~mdehoon/software/cluster/software.htm) (using unsupervised hierarchical clustering analysis (uncentered correlation similarity metric, centroid linkage). Heatmaps were visualized using Java Treeview software (http://jtreeview.sourceforge.net/). Meta-analysis of 5-HT protein expression was performed using existing annotated databases from the Human Protein Atlas (HPA; www.protein atlas.org). Antigen staining was evaluated as positive or negative, and represented as a percentage of the total number of cancer tissues tested for each cancer type. Meta-analysis of HTR and TPH genes in normal mouse breast tissue (N=5), non-metastatic breast tumors from 67NR xenografts (N=5), and metastatic breast tumors from 4T1 ×enografts (N=4) was performed on Affymetrix mouse Genome 430 2.0 Array data housed in Gene Expression Omnibus (GEO; cat. no. GSE62817).

Sarcoma and breast cancer cell lines were cultured in incubators maintained at 37°C in the presence of 5% CO₂. AU565 (cat. no. CRL-2351), HCC70 (cat. no. CRL-2115), and BT-549 (cat. no. HTB-122) human breast cancer lines, and SW872 human liposarcoma cells (cat. no. HTB-92) were grown in RPMI-1640 medium supplemented with 10% fetal bovine serum and penicillin/streptomycin antibiotics (all from ATCC, Manassas, VA, USA). The SK-BR-3 human breast cancer line (cat. no. HTB-30; ATCC) was grown in McCoy's medium supplemented with 10% fetal bovine serum and penicillin/streptomycin antibiotics. The A673 human Ewing's sarcoma cell line (cat. no. CRL-1598) and the HOS human osteosarcoma cell line (cat. no. CRL-1543) (both from ATCC) were grown in Dulbecco's modified eagle medium supplemented with 10% fetal bovine serum and penicillin/streptomycin. The COSB canine hemangiosarcoma cell line (28) was grown in EGM-2 basal medium with the EGM-2 Bullet kit (cat. no. CC-3162; Lonza, Basel, Switzerland). The epidemiology and cancer biology of the canine cell line has been shown to extrapolate directly to humans (28).

Cell lysates were collected as indicated for each experiment, subjected to SDS-PAGE, and transferred to polyvinylidene fluoride membranes using the Trans-Blot Turbo Transfer System (Bio-Rad, Hercules, CA, USA). Membranes were blocked in tris buffered saline plus 3% bovine serum albumin and 0.05% Tween-20, and incubated with the following antibodies as indicated for each experiment: Ki-67 (cat. no. ab16667; 1:1,000 dilution, 1 h incubation at 25°C; Abcam, Cambridge, UK), Kinome View Profiling kit (cat. no. 9812; 1:2,000 dilution for all antibodies, 1 h incubation at 25°C; Cell Signaling, Danvers, MA, USA) or anti-actin (cat. no. sc8432; 1:1,000 dilution, 1 h incubation at 25°C; Santa Cruz Biotech, Dallas, TX, USA). Each primary antibody was detected with 1:1,000 HRP-conjugated secondary antibody [Santa Cruz Biotechnology cat. no. sc-2357 (anti-rabbit) or cat. no. sc-2005 (anti-mouse); 1:1,000 dilution, 1 h incubation at 25°C], subjected to Supersignal West Dura Extended Duration Substrate (Thermo Scientific, Waltham, MA, USA), and digitally captured using a GE Image Quant Las4000 imaging system.

The Phospho-Mitogen-activated protein kinase (MAPK) Antibody Array (cat. no. ARY002B; R&D Systems, Minneapolis, MN, USA) was performed on HOS cells as indicated according to the manufacturer's instructions. Normalized heatmap data was generated in Cluster 3.0 software (http://bonsai.hgc.jp/~mdehoon/software/cluster/software.htm) (using unsupervised hierarchical clustering analysis (uncentered correlation similarity metric, centroid linkage). Heatmaps were visualized using Java Treeview software (http://jtreeview.sourceforge.net/).

To measure the effects of serotonin or 5-HT pharmacological inhibitors on tumor cell viability, sarcoma and breast cancer cells were plated in 96-well plates at approximately ~75% confluence, treated as indicated, and cell viability was measured after 24 h via fluorescent excitation at 530 nm with the Alamar Blue cell viability assay (ThermoFisher Scientific).

HOS cells were grown in hanging drops (3,000 cells/drop) for 48 h as previously described for other cell lines (29), and transferred to non-adherent well plates. Spheroids were treated as indicated for 48 h and photos were taking using bright field microscopy.

All chicken embyo experiments were performed prior to hatching, thus these experiments were considered exempt from ethics approval based on PHS policy. Chorioallantoic membrane (CAM) tumor assays were performed as previously described using rainbow hen eggs (Gallus gallus domesticus) (30). A false air-sac was generated directly over the CAM of fertilized chicken eggs (7 days post-fertilization). 20,000 dissociated CosB tumor cells were soaked onto a 5 mm² gelatin sponge and then placed onto the CAM. A sham solution of isotonic saline solution (N=3) or 100 nM SB-269970 (N=3) was added daily directly onto the CAM tumor. After 72 h of treatment (13 days post-fertilization), the tumors were collected, weighed, and photographed on a lightbox.

Retrospective analysis of clinical data was carried out with the approval of the Texas Tech University Health Sciences Center Institutional Review Board. Analysis of 419 female patients diagnosed with invasive ductal carcinoma at the Texas Tech Breast Care Center between the years of 2010 to 2014 was performed. The demographic characteristics of the study population are described in Table I. IHC for estrogen receptor (ER), progesterone receptor (PR), Her-2/neu, and Ki-67 tumor proliferative index were available for all patients. Breast cancer clinico-pathological features (age at diagnosis, tumor size, tumor grade, lymph node status, hormonal receptor status) were extracted from each case report. Patients were considered positive for SSRI usage if they had been prescribed SSRIs at any point in the year prior to or at the time of diagnosis. Of a total of 419 patients included in this study, we identified 28 patients taking SSRIs and 391 patients who were not taking SSRIs during the defined time frame.

All in vitro experiments were performed at least three independent times, with at least four technical replicates per assay. Unpaired t-tests or one-way analysis of variance (ANOVA) followed by Dunnett's multiple comparison post hoc test were used to determine the statistical significance for all in vitro experiments. Differences were considered statistically significant if the P-value was less than 0.05. The retrospective relationship between SSRI usage and the Ki-67-based proliferative index of the breast tumors was determined with the Mann-Whitney rank sum test. Comparisons of SSRI usage to tumor hormonal receptor status and tumor staging were calculated with the Fisher's exact test. Statistical analyses were carried out using Graphpad Prism. Differences were considered statistically significant if the P-value was less than 0.05.

To evaluate HTR gene (encoding 5-HT receptors) expression in human cancers, we performed meta-analysis on the global gene expression profiles obtained from microarray analysis housed in the Cancer Cell Line Encyclopedia (CCLE; http://portals.broadinstitute.org/ccle). This database contains quantitative data on over 1,000 different cell lines representing a diverse array of cancer types including carcinomas, sarcomas, and hematopoietic cancers. Unique expression patterns emerged for many of the HTR mRNAs (Fig. 1). Analysis of the CCLE data revealed clear clustering of gene over-expression dependent on tumor origin was observed for several HTR mRNAs including HTR3A (largely lung and hematopoietic/lymphoid tumor cell lines), HTR1F (largely hematopoietic/lymphoid, bone sarcoma, and soft tissue sarcoma cell lines), HTR1D (largely mixed digestive track cancer cell lines), HTR2A (largely central nervous system, breast, and bone sarcoma cell lines), and HTR2B (largely skin cancer cell lines). Though clustering of expression patterns were observed other HTR mRNAs, the cell lines composing these clusters were varied among cancer origins. As mRNA expression is not always reflective of protein levels, we performed meta-analysis of tissue pathology data housed in the Human Protein Atlas to analyze 5-HT protein expression across cancers. This repository contains stained tissue samples representing the most common forms of cancer, totaling 216 different cancer samples on which immunohistochemistry data is reported for many proteins. Our meta-analysis of the pathology-based annotation of 5-HT1A, 1D, 1E, 1F, 2A, 2B, 3B, 4, 5A, and 7 protein expression levels is illustrated in Table II. A handful of the 5-HT receptors were not reported due to their absence from the Human Protein Atlas, therefore IHC analysis for these proteins is not included in our results. Our meta-analysis revealed, with the exception of 5-HT1A and 2A, that the majority of 5-HT receptors analyzed were expressed across many cancers.

To determine if serotonin is capable of enhancing the proliferation of rate of cancer cells, we serum starved a panel of diverse tumor cell lines representing 4 breast cancers (AU565, BT549, HCC70, and SK-BR-3), 2 soft tissue sarcomas (CosB hemangiosarcoma cells and SW872 liposarcomas cells), and 2 bone cancer (A673 Ewing's sarcoma cells and HOS osteosarcoma cells) for 24 h. We then treated each cell line with increasing concentrations of serotonin (1 to 1×10⁵ nM), and cell viability was accessed after 48 h. No major changes in the proliferation rate were observed for any of the cell lines following serotonin treatment relative to the untreated control (Fig. 2A). These data were confirmed in HOS osteosarcoma cells via immunoblotting by using the proliferative marker Ki-67 (Fig. 2B).

We next sought to determine if the presence of serotonin alters intracellular signaling of the cancer cell lines. To accomplish this, HOS osteosarcoma cells were serum starved overnight and stimulated with 10 nM serotonin or a sham control for 10 min. Lysates were collected and the phosphorylation status of 24 kinases implicated in mitogenic and survival processes were examined using antibody arrays. Serotonin stimulation resulted in a marked increase in the activation of Akt2, CREB, GSK3, HSP27, and multiple MAPK signaling mediators (Fig. 2C).

Because serotonin modulated the levels of kinases involved in cancer cell signaling pathways, we determined whether inhibition of 5-HT receptor activity was capable of affecting cancer cell viability. We subjected the cell line panel consisting of 4 breast cancers (AU565, BT549, HCC70, and SK-BR-3), 2 soft tissue sarcomas (CosB hemangiosarcoma cells and SW872 liposarcomas cells), and 2 bone cancer (A673 Ewing's sarcoma cells and HOS osteosarcoma cells) to highly selective 5-HT receptor antagonists (Table III), and cell viability was assayed 24 h post-treatment. Dose-dependent reductions in cell viability were observed for most of the 5-HT receptor antagonists across the panel of cancer cell lines (Fig. 3A-G). Notable exceptions were pirenperone (5-HT2 antagonist) and dolasetron mesylate (5-HT3 antagonist), which did not reduce cell viability across any of the cancer cell lines. Moreover, A673, HOS, and SK-BR-3 tumor cells were relatively more resistant to 5-HT antagonism compared to the other cells lines, despite expression of selected HRT mRNAs in each of these cell lines (based on CCLE data). A representative image of the SW872 liposarcoma cell line treated for 24 h with SB-269970 (5-HT7 antagonist) is shown in Fig. 4A. We confirmed the efficacy of SB-269970 to abrogate cell viability in the HOS osteosarcoma cell line using a three-dimensional tumor spheroid model, whereby the 5-HT7 receptor antagonist reduced tumor cell viability in a dose dependent manner after 48 h (Fig. 4B). To expand our in vitro results into an in vivo xenograft tumor model, we subjected CosB hemangiosarcoma tumors grown on CAMs to daily treatments of the 5-HT7 antagonist (5×10⁻⁹ grams/day) or a control sham. Tumors treated with the 5-HT7 antagonist exhibited significantly smaller tumor sizes compared to the sham control (Fig. 4C and D).

To detect the phosphorylation status of proliferation and survival regulators following 5-HT antagonism, we treated HOS osteosarcoma cells with the 5-HT7 receptor antagonist for three h, collected cell lysates, and performed antibody arrays to quantify changes in protein phosphorylation. Antagonist treatment resulted in a marked increase in p53 phosphorylation and reductions in the phosphorylation status of both p38 and p42 MAPK (Fig. 5A). To more broadly evaluate the panel of selective 5-HT receptor antagonists, we collected protein lysates from HOS cells treated for three h with each selective 5-HT antagonist (100 nM), and performed immunoblots using a set of phospho-motif antibodies that cover a large portion of the kinome regulated by diverse kinase families. Our data revealed, with the exception of 5-HT3 antagonists, treatment with many of the antagonists resulted in reductions in substrate phosphorylation for signaling pathways including CDK, MAPK, and AKT (Fig. 5B).

We carried out a retrospective clinical analysis of 419 patients diagnosed with breast cancer to assess the association between the use of SSRIs and breast tumor proliferation rates. Patients were stratified based on SSRI use. A description of the various SSRIs taken by these patients is exhibited in Table IV. The Ki-67-based proliferative index was determined from pathology samples taken at each patient's diagnostic biopsy. No difference was found in tumor staging or hormone receptor status between users of SSRIs and non-users; however, in patients with late stage breast cancer, use of SSRIs was significantly associated with increased tumor proliferative index (2.3-fold increase) compared to patients who were non-users of SSRIs (P=0.03) (Table V, Fig. 6A). To determine if expression changes in 5HT receptors or biosynthetic enzymes contribute to SSRI-dependent alterations in proliferation rates between early and late stage breast cancer, meta-analysis of HTR and TPH genes was performed for normal mouse breast tissue, non-metastatic breast tumors from 67NR xenografts, and metastatic breast tumors from 4T1 ×enografts based on data from the Gene Expression Omnibus (GEO cat. no. GSE62817). No statistical difference in expression patterns was observed for HTR or TPH mRNAs between normal, non-metastatic, and metastatic breast tissue (Fig. 6B).