Serotonin, also known as 5-hydroxytryptamine (5-HT), is a biogenic amine produced by enterochromaffin (EC) cells of the gastrointestinal tract. In the blood, 5-HT is carried by platelets and exported to various sites in the body, where it produces mitogenic effects.[3-6] Recently, 5-HT has been considered a modifier of liver function due to the autonomic nervous system input as well as the blood flow within this organ. This amine is thought to regulate proliferation and function of a number of key resident liver cells, including hepatocytes, stellate cells and sinusoidal endothelial cells.
5-HT1A,1B,1D,1F,2A,2B,2C,3A,3B receptors are known to be expressed on hepatocytes.[10, 11] While the 5-HT1 family inhibits cAMP, 5-HT2 receptors signal through PLC and 5-HT4,6,7 receptors stimulate PKA/cAMP. Neither the complete 5-HT receptor profile nor the intracellular pathways associated with 5-HT-mediated stimulation of hepatocyte function are known. We hypothesized that PKA-dependent as well as PKA-independent cAMP effects and AKT signaling are linked to 5-HT driven hepatocyte stimulation.
Small intestinal (SI) neuroendocrine neoplasms (NEN) represent a group of tumors derived from 5-HT producing EC cells; they have an approximate 2/100 000 incidence and an estimated prevalence of approximately 30 000 patients. All tumors produce 5-HT in excess (approximately 1 μM), which is usually removed by hepatocyte-mediated monoamine oxidase. However, once a tumor metastasizes to the liver, this physiological clearance is surmounted and patients exhibit an expansion of the 5-HT pool size with striking increases in blood and platelet 5-HT, an elevation of 5-HIAA in urine[14, 15] and the production of a number of symptoms (“carcinoid syndrome”) all effected by elevated 5-HT.[14, 15]
Metastatic SI NEN cells are likely trapped within the hepatic trabecular structures and, in the process of tumor proliferation, enter the sinusoidal spaces, making direct contact with hepatocytes. In this hepatic tumor microenvironment, tumor 5-HT reaches concentrations at least in the μM range: 1000-fold higher than normal serum levels.[16, 17] We postulate a direct cross-talk between metastatic tumor cell and hepatocytes under these conditions, with co-regulation of cell secretion and proliferation. IGF-1 and HGF are crucial growth factors supporting progression of various cancers[18-24]; these are produced and secreted in hepatocytes[25, 26] and, therefore, are of importance in the tumor cell:hepatocyte interaction. We postulated that a hepatocyte : tumor axis existed comprising growth factors and amines. Our aim was to examine the 5-HT receptor profile in hepatocytes, determine the effects of 5-HT on hepatocyte function (signal pathway activation, transcription and growth factor secretion) and evaluate evidence for a 5-HT/IGF-1 axis in an in vivo model of SI NEN metastasis.
Material and Methods
Studies were performed on freshly isolated rat hepatocytes, two well-characterized SI NEN cell lines  and a murine hepatic-metastasis model (SCID/splenic H-STS injection). For hepatocytes, the effects of 5-HT at physiological (10−9 M) and elevated (circulating levels detectable in patients with “carcinoid syndrome” and denoted as “pathological”; 10−6 M) levels[16, 17] on cell viability, PKA/cAMP/CREB, AKT and ERK1/2 (MAPK) signaling were evaluated. Growth factor transcription as well as growth factor secretion (IGF-1, HGF) were determined in hepatocytes after 5-HT administration. Antisense strategies were used to confirm 5-HT7 receptor function. SI NEN cell line proliferation (in response to recombinant IGF-1) was evaluated by both MTT uptake and Glomax-based cytotoxicity assays (in the absence or presence of the small molecule IGF-1R inhibitor BMS-754807) as was IGF-1R phosphorylation (western blot). IGF-1, 5-HT and 5-HT7 expression were evaluated in vivo in the livers of sham-operative and tumor-bearing animals.
Chemicals and antibodies
The 5-HT7 receptor antagonist SB269970 was from Tocris Bioscience (Ellisville, MO, USA), Methiothepin (5-HT5A/B,6,7 antagonist) and Ketanserin (5-HT2A/C antagonist) from Sigma Aldrich, PRX-08066 (5-HT2B antagonist) was a gift of EPIX Pharmaceutical (Lexington, MA, USA) and recombinant IGF-1 from R&D Systems (Minneapolis, MN, USA). Primary antibodies, phosphorylated(p)-AKT(Ser473), AKT, pERK1/2(Thr185, Tyr187), ERK1/2, pCREB(Ser133), CREB, pIGF-1R(Tyr1361) and IGFIR, and HRP-secondary antibodies were from Cell Signaling Technology (Danvers, MA, USA). 5-HT7 (AB9405) was from Millipore (Billerica, MA, USA) and BMS-754807 was from Chemietek (Indianapolis, IN, USA).
Normal rat hepatocyte isolation: Hepatocytes were isolated from rat liver by collagenase perfusion, as described previously, and maintained in Williams E medium supplemented with 10 mM Hepes buffer, 2 mM L-glutamine, 1.8 g/L glucose, 1 μM dexamethasone, 4 mg/L insulin, 100 U/mL penicillin, 100 μg/mL streptomycin and 10 mg/L gentamycin. Preparations with >90% viable cells were used for experiments.
Freshly isolated primary rat hepatocytes were maintained in Williams E medium supplemented with HEPES (2.8 g/L), dextrose (1.8 g/L), l-glutamine (20 mM), gentamycin (800 μg/L), dexamethason (400 μg/L), insulin, 10% FBS, penicillin (100 IU/mL), streptomycin (100 μg/mL) and amphotericin B (2.5 μg/mL). Cells were plated into collagen-coated plates at a density of 3–4 × 104 cells/cm2, and after 24 h hepatocytes were coated with a second layer of rat tail collagen I (BD Biosciences, Bedford, MA, USA) in a “sandwich” configuration.[31, 32] For experiments, cells were cultured under serum-free conditions. All experiments were performed within 48 h of plating.
Hepatocytes (4 × 104 cells/cm2) were plated in 96-well plates (sandwich configuration) and stimulated with 5-HT (10−6 M to 10−12 M: n = 6 wells/concentration). After 1 h, cell viability was evaluated using WST-1 (Roche, Indianapolis, NJ, USA). 5-HT receptor antagonists were included (10−6 M).
PKA/cAMP signaling pathway analysis
After 30 min pre-incubation with 5-HT receptor antagonists (10−6 M), cultured hepatocytes (48 h) were stimulated with physiological (10−9 M) and “pathological” (10−6 M) 5-HT for 0.5 h. PKA activity and cAMP levels were quantified using SuperArray ELISA kits (R&D Systems).
pAKT/AKT signaling pathway analysis
After pre-incubation with 5-HT receptor antagonists (10−6 M), hepatocytes were stimulated with 5-HT (10−9 M, 10−6 M) and pAKT/total AKT signal activity quantified using SuperArray CASE ELISA kits (SABiosciences, Frederick, MD, USA; ERK-FE-002).
Protein extraction and western blot
Fresh isolated hepatocytes (4 × 104 cells/cm2) were maintained with a collagen layer in collagen-coated 6-well plates (Falcon, BD Biosciences) and treated with 5-HT (10−9 M, 10−6 M) for 15 min to 8 h. Selective receptor antagonists were used as described. Whole cell lysates were prepared in ice-cold lysis buffer (10× RIPA lysis buffer [Millipore, Billerica, MA, USA]) and, after centrifugation, the supernatant was transferred and quantified using the BCA protein assay kit (Thermo Fisher Scientific, Rockford, IL, USA). After denaturation (SDS sample buffer), total protein lysates (20 μg) were separated on an SDS-PAGE gel (10%) and transferred to a PVDF membrane (Bio-Rad, Hercules, CA, USA; pore size: 0.45 mM). Following blocking, membranes were incubated with primary antibodies (Cell Signaling Technology, Danvers, MA, USA) overnight at 4°C and, after HRP-secondary antibodies, immunodetection was performed using the Supersignal West Pico Luminol/Enhancer solution (Thermo Fisher Scientific, Rockford, IL, USA) and the blots were exposed on X-OMAT-AR films. The optical density of the appropriately sized bands was quantified using ImageJ software (NIH, Bethesda, MD, USA). The ratio between phosphorylated-protein and total protein was calculated; total protein expression was reported relative to that of β-actin (Sigma-Aldrich, St. Louis, MO, USA).
Hepatocyte transfection with serum response element (SRE) vector
To confirm activation of the MAPK/ERK signaling pathway, hepatocytes were transfected using the pGL4.33 [luc2P/SRE/Hygro] Vector (Promega, Madison, WI, USA). Hepatocytes were plated as described and after 24 h, transfection was undertaken using Lipofectamin 2000 (Invitrogen, Carlsbad, CA, USA) according to the manufacturer's instructions. After 18 h, media was changed and cells were stimulated with 5-HT (10−9 M, 10−6 M) and the 5-HT receptor antagonists for 1 and 4 h. Luminescence was measured using the Bright-Glo Luciferase Assay System (GloMax Multi+ Detection System; Promega).
RNA isolation and reverse transcription
RNA was isolated from rat hepatocytes (1 × 106 cells, n = 6) using TRIzol (Invitrogen), cleaned (Qiagen, RNeasy kit; Qiagen, Valencia, CA, USA) and converted to cDNA using the High Capacity cDNA Archive Kit (Applied Biosystems, Carlsbad, CA, USA).[34, 37]
Primer preparation and QT-PCR
Primers for each subtype of 5-HT receptors (1A,1B,1D,1E,1F,2A,2B,2C,3,4,5A,5B,6,7) were used as described. Appropriate bands were excised and prepared for sequencing (Wizard SV Gel and PCR Clean-UP System [Promega]). Purified PCR products were sequenced (W.M. Keck Biotechnology Resource Laboratory, Yale University) using an automated Applied Biosystems 373A Stretch DNA Sequencer (Perkin-Elmer, Norwalk, CT, USA) utilizing both the forward and reverse primers. Sequencing products were analyzed using Bioedit.
5-HT7 receptor knockdown
A 21-mer oligonucleotide antisense corresponding to 1011–31 of the rat 5HT7 receptor (NM_022938.2) was designed to induce a steric obstacle for protein translation (Yale Medical School Keck Oligonucleotide Synthesis Facility). Control nucleotides were prepared with randomized sequence of matching nucleotides per protocol. In these experiments, isolated hepatocytes cells were exposed to oligonucleotides (antisensense: CACACTCTTCCACCTCCTTCT, or control: CTCCTTCTCAACCCCACTTTC, 150 pmol) prior to sandwiching, culture and study.
RT-PCR analyses were performed using Assays-on Demand and the ABI 7900 Sequence Detection System with primer sets from Applied Biosystems; the presence of single bands for each primer sets was confirmed by PCR mix on gels. Data was normalized using the ∆∆CT approach and the rat Act1b was used as a housekeeping gene.
HGF and IGF-1 secretion
Secretion levels of HGF and IGF-1 were analyzed using commercially available ELISA assays (HGF: B-Bridge International; IGF-1: R&D Systems). Cells were seeded into 6-well plates and stimulated with 5-HT (10−9 M, 10−6 M) and the 5-HT antagonists. After 1 h, growth factor levels were quantified in the supernatant according to the manufacturer's instructions.
Small intestinal neuroendocrine neoplasms and in vivo model
Small intestinal neuroendocrine neoplasm culture and proliferation
Two well-characterized NEN cell lines (KRJ-I, primary tumor; and H-STS, hepatic metastasis) were cultured as floating aggregates in Quantum 263 (PAA, Dartmouth, MA, USA) supplemented with 100 IU penicillin/mL and 100 μg streptomycin/mL. All experiments were performed without antibiotics at day 2 following sub-culture. The proliferation of cell lines (5 × 104 cells/mL) was studied using MTT uptake or luminescent-based protocols (live/dead, caspase 3/7 activation: Promega). in response to recombinant IGF-1 (1–1000 pg/mL) with or without BMS-754807 (10−9 to 10−5 M; 24 h).
Cells (1 × 106/mL) were stimulated with recombinant IGF-1 (1 ng/mL) for 1 h, protein isolated and western blot performed using antibodies against pIGFR, IGFR and β-actin.
In vivo metastasis model
Intrasplenic injection of 10 × 106 (200 μL, 27G needle) H-STS cells was undertaken in laparotomized, isoflurane-anesthetized male or female SCID mice (6–8 weeks; approximately 20 g, Jackson Laboratories, Bar Harbor, ME, USA). The wound was closed in one layer with wound clips; the animals were kept until 6 weeks and then killed. Studies were approved by the Yale University Institutional Animal Care and Use Committee. Clinical standard staining was conducted for pathologist examination using H&E, or Chromgranin A or MKI67 ([MIB1], Dako, Glostrup, Denmark), which were detected using DAB (3,3′-Diaminobenzidine). Normal liver (sham-operated) and liver adjacent to the macroscopic tumor (<2 mm) were processed to protein. IGF-1 and 5-HT levels (Rocky Mountain Diagnostics, Colorado Springs, CO, USA). were measured by ELISA. Western blot was performed for IGF-1 and 5-HT7.
All statistical analyses were performed using Microsoft Excel and Prism 5 (GraphPad Software, San Diego, CA, USA). Nonlinear regression analyses were used to evaluate half maximal effective (EC50) concentration. Cell viability assays were evaluated using two-tailed Student's t-tests; all other data were assessed using the two-tailed Mann–Whitney tests.
Serotonin (5-HT) is not only a neurotransmitter but also has a significant physiological role as a hormone in a number of gastrointestinal systems. An intestinal–hepatic axis linking serotonin producing EC cells with the liver remains a largely underexplored subject. Recent evidence supporting the role of 5-HT in liver regeneration has strengthened the hypothesis that this amine plays a role in liver physiology. These physiological effects are generally considered to be mediated via the 5-HT2 receptor family and are considered to reflect platelets as the aminergic source. The role of 5-HT secreted by SI NEN that have metastasized to the liver (i.e. an exogenous cellular source of the amine) has not been investigated. We examined this at a number of levels, including measuring the effects of different 5-HT concentrations on isolated hepatocytes as well as determining expression of the serotonin-IGF-1 axis in an in vivo liver metastasis model to identify potential therapeutic targets. Our data identify significant differences in signal pathway activation and growth factor synthesis and secretion at physiological (10−9 M) and “pathological” (10−6 M) 5-HT levels, which suggests that high amine production by tumor cells (5-HT reaches concentrations at least in the μM range: 1000 fold higher than normal serum levels)[16, 17] may modify hepatocyte behavior (specifically the production of IGF-1) in a fashion conducive to support local SI NEN cell proliferation. In addition, we demonstrate an intact 5-HT7 receptor signaling pathway, which, together with IGF-1R (on tumor cells) provides potential translational therapeutic targets. Our model proposes that 5-HT secretion from SI NEN activates 5-HT7 receptors on hepatocytes, leading to PKA/pCREB/AKT activation and IGF-1 secretion, which, importantly, is only regulated at “pathological” concentrations of 5-HT.
The liver is considered to express at least 5-HT1A,B,D,F, 2A-C and 3A-B receptors, and although crude hepatocyte fractions are 5-HT2 receptor positive, expression of 5-HT5-7 has not, to date, been examined. In our study, we screened rat hepatocyte preparations using primers developed by Chen et al. and confirmed expression of 5-HT1A,D, 2A,B, 5D, 6 and 7 receptors in these cells. As we were specifically interested in 5-HT7, we examined this in more detail. Our studies confirmed expression of this receptor on rat, mouse and human hepatocytes.
An analysis of the effects of 5-HT on rat hepatocyte cell viability (measured by WST-1 uptake and elevated Ki67 transcripts) identified that the 5-HT2 receptors were principally responsible for mediating this phenomenon. The combination of the 5-HT2 receptor antagonists, ketanserin and PRX-08066, could reverse increases in WST-1 and Ki67. In contrast, the 5-HT5–6 receptor antagonist, methiothepin, as well as the 5-HT7 antagonist, SB269970, did not have a significant inhibitory effect. These results confirm that hepatocyte proliferation or viability is principally 5-HT2 driven. However, ERK activation (inhibited by methiothepin) and Ki67 transcription (inhibited by Ketanserin/PRX-08066 and SB269970) suggest that hepatocyte proliferation is complex and may be regulated by 5-HT at a number of levels.
5-HT2 receptors principally signal via PLC while 5HT7 (and receptors 4,6) signal via cAMP/PKA. We measured a dose-dependent increase in cAMP levels as well as PKA activity in cultured hepatocytes stimulated with either physiological or pathological 5-HT concentrations. These effects were completely reversible by SB269970, confirming signaling through the 5-HT7 receptor. Interestingly, the combination of ketanserin and PRX-08066 also decreased PKA activity. As ketanserin is known to show an affinity for 5-HT7 receptors,[45, 46] the measured effect could be due to non-specific antagonism of this receptor subtype.
Given prior observations that activated cAMP may also be associated with AKT as well as the ERK (MAPK) alterations in hepatocytes,[13, 47, 48] we investigated the interaction with the AKT as well as the ERK (MAPK) pathway. A significant increase in pAKT/AKT at pathological and physiological 5-HT concentrations was identified. Of note was the observation that the effect was significantly higher at pathological levels. We interpret this to indicate that AKT signaling in hepatocytes may be preferentially activated by 5-HT secreting SI NEN metastases. In our studies, pAKT/AKT activity was completely reversed by SB269970, suggesting that AKT signaling is predominantly a 5-HT7 receptor-mediated effect via the PKA-independent cAMP-GEF/Rap pathway.[49-51] A partial but not complete response was evident with Ketanserin/PRX-08066, suggesting either antagonism at the 5-HT7 receptor level or an effect via 5-HT2 receptors.
While both protein kinases AKT and ERK are suggested downstream effectors of Rap, it is known that in hepatocytes, cAMP activates AKT and inhibits ERK,[13, 52-58] effects that play an important role in hepatocyte survival.[56, 59-61] The precise role of ERK in hepatocytes, however, is controversial. While some studies suggest that ERK activation has an anti-apoptotic effect[60, 62, 63] others indicate only a mild effect on hepatocyte survival.[13, 53, 56, 64, 65] Cullen and colleagues demonstrate a decrease in pERK levels and an increase in pAKT levels in hepatocyte signaling through PKA independent cAMP activation. Our findings, in contrast, demonstrated a significant increase in pERK/ERK activation at both 5-HT concentrations, an effect that was only reversible by methiothepin, which is known to have high affinity for 5-HT6 and 7 receptors and moderate affinity for 5-HT1, 2 and 5 receptors. Importantly, no effect was determined with the specific 5-HT7 receptor antagonist SB269970, suggesting that ERK activation is not 5-HT7 receptor-linked. Interestingly, ERK activation was increased over time, suggesting that MAKP signaling was not only a direct consequence of 5-HT but could also be a secondary effect, perhaps mediated via growth factor auto-secretion of IGF-1.
We next investigated the time-dependent effects of the two 5-HT concentrations on pCREB (cAMP response element-binding protein) as well as pAKT. Increases in pCREB were detected at both concentrations, but this was more evident in hepatocytes stimulated with pathological 5-HT levels. Use of physiological 5-HT levels resulted in an increase of pCREB after 2 h, with a maximum at 4 h (184 ± 3.3%); this effect was earlier (after 1 h), prolonged (maximum at 8 h) and increased (maximum 271 ± 64%) by pathological 5-HT levels. Time course studies for pAKT identified an increase of protein activation at both concentrations (10−9 and 10−6 M), with a maximum after 30 min. This PKA-independent response was earlier (30 min) compared to the PKA-dependent CREB response (240 min). These findings suggest that 5-HT signals in hepatocytes through both PKA-dependent as well as PKA-independent pathways; these are associated with differences in both time response and extent of effect.
CREB is known to regulate glucose homeostasis as well as growth factor-dependent cell survival. A total of 155 target genes have been identified, including transcripts for cell survival and growth factor secretion. IGF-1 gene transcription, in particular, is upregulated by CREB in osteoblasts and mesenchymal cells.[68-71] Under normal[25, 26, 72-74] and pathological conditions (e.g. tumor development and progression of a variety of different cancers such as pancreatic carcinoma, colon carcinoma, esophageal carcinoma and non-small cell lung cancer),[1, 18-24] hepatocytes produce a variety of agents, including IGF-1 and HGF. In our study, HFG secretion but not transcription was significantly elevated at both physiological and pathological 5-HT concentrations, suggesting HGF release was amine-regulatable. However, no significant difference was evident between 10−9 and 10−6 M 5-HT. The HGF promoter region is characterized by a number of regulatory elements but is not regulated by cAMP. In contrast to HGF, both transcription and secretion of IGF-1 was elevated by 5-HT, an effect only noted at pathological 5-HT concentrations. This was completely reversed by SB269970, suggesting that hepatocytes exposed to pathological 5-HT levels produce and secrete IGF-1 via 5-HT7 receptor mediated pathways. IGF-1 is a well-known proliferative regulator of NEN;[76, 77] our identification of pro-proliferative effects for IGF-1 on the two different SI NEN cell lines as well as activation (and reversal by BMS-754807) of the IGF-1R pathway confirm the link between the growth factor and these tumors.
We postulate that hepatocytes respond to elevated 5-HT levels produced by specific SI NEN metastases (tumors that express TPH1 and synthesize this amine) with increased synthesis and secretion of IGF-1 via the AKT pathway; a 5-HT-mediated paracrine production of IGF-1 most likely supports tumor cell proliferation. Our in vivo model supports such an axis. IFG-1 was increased in hepatocytes adjacent to the H-STS cell tumors (a 5-HT secreting tumor), while 5-HT7 was upregulated in livers compared to sham-operated animals. Hepatocytes also responded with upregulation of Ki67, suggesting a co-proliferative drive.
These observations relating a metastasis to 5HT7 receptors on hepatocytes would be specific for tumors that synthesize and secrete serotonin; other neuroendocrine tumors that are non-serotonergic, for example,rectal NEN, would not directly activate hepatocytes via this mechanism. It is, however, possible that metastases from other NEN or other sites, for example, colorectal adenocarcinoma, may indirectly regulate hepatocyte function through 5-HT receptors. Such a mechanism would require tumor-directed 5-HT release from platelets as a cellular source.
In conclusion, we have confirmed a role for 5-HT2 receptors in hepatocyte viability and demonstrated the existence of a functional 5-HT7 receptor in isolated hepatocyte preparations. Furthermore, we have dissected the differences in pathway activation at physiological (10−9 M) versus pathological (10−6 M) 5-HT levels on these cells. The former, which may be related to platelet-derived circulating serotonin or neural effectors, was related to HGF production and hepatocyte viability. The latter was associated with increased levels of IGF-1, which play a crucial role in regulating NEN proliferation. Overall, elevated 5-HT levels, commensurate with metastatic SI NEN, are associated with increased levels of hepatocyte-derived IGF-1 synthesis and release. 5-HT-mediated perturbations in this growth factor, which plays a crucial supportive role in SI NEN proliferation, identify the significance of the hepatocyte-NEN microenvironment in supporting metastatic neuroendocrine cell proliferation. Based on these observations, there is evidence to propose that 5-HT7, AKT and the IGF-1R may both be viable targets specifically for SI NEN liver metastasis therapy.