Serotonin-Mediated Tuning of Human Helper T Cell Responsiveness to the Chemokine CXCL12

In addition to its role as neurotransmitter, serotonin (5-HT) is an important modulator of inflammation and immunity. Here, we report novel findings suggesting a 5-HT involvement in T cell migration. In particular, we show that 5-HT tunes the responsiveness of human T lymphocytes to the broadly expressed chemokine CXCL12 in transwell migration assays. By real-time PCR, western blot analysis and electrophysiological patch clamp experiments, we demonstrate that the type 3 5-HT receptor (5-HT3) is functionally expressed in human primary T cells. In addition, specific 5-HT3 receptor agonists selectively decrease T cell migration towards gradients of CXCL12 but not of inflammatory chemokines, such as CCL2 and CCL5. In transmigration experiments, 5-HT3 receptor stimulation reverts the inhibitory effect of endothelial-bound CXCL12 on T cell migration. Our data suggest that the reduced T cell responsiveness to CXCL12 induced by 5-HT may occur to facilitate T cell extravasation and migration into inflamed tissues.


Introduction
The monoamine serotonin (5-HT) is known to mediate a plethora of actions throughout the body. In the central and autonomous nervous system, 5-HT is utilized by neurons to signal both volumic and fast synaptic neurotransmission. In the periphery, 5-HT can both contract and relax smooth muscle (vascular and gastrointestinal), contract cardiac muscle, regulate hepatic microcirculation and contribute to blood clotting [1].
Synthesis of 5-HT in the peripheral and central nervous system is respectively operated by two different isoforms of tryptophan hydroxylase (TPH), TPH-1 and -2 [2]. These enzymes catalyze hydroxylation of tryptophan to 5-hydroxytryptophan, the immediate precursor of 5-HT. Though conventionally considered a neurotransmitter, 5-HT is primarily produced by the enterochromaffin cells in the gut [3] and it is taken up via an active transport mechanism into platelets, which provide a rich reservoir of 5-HT in the circulation. Subsequent release of platelet-stored 5-HT can be rapidly triggered by platelet-activating factor, thrombin, complement fragments C3a and C5a, or immunoglobulin Econtaining immune complexes. Thus, at sites of inflammation and platelet activation, local concentrations of 5-HT can greatly exceed the relatively low amounts found free in the serum [4]. Indeed, 5-HT is a known immunoregulator and it is implicated in several inflammatory diseases [1,[5][6][7]. In the context of the adaptive immune response, 5-HT affects both T cell and dendritic cell functions [6,8]. Interestingly, it has been recently reported that dendritic cells uptake 5-HT at sites of inflammation and shuttle it to naive T cells, thereby modulating T-cell activation [9], proliferation and differentiation [10].
The 5-HT receptors have been divided into seven major classes (5-HT 1 -5-HT 7 ). The 5-HT 3 receptor is the most phylogenetically conserved; it is described as a ligand-gated ion channel, permeable to monovalent and divalent cations [11], belonging to the Cysloop receptor super-family and sharing structural and functional features with other members of this channel/receptor group [12]. The 5-HT 3 receptor is constituted of five subunits organized in a channel. To date, five different 5-HT 3 receptor subunits were identified (5-HT 3A-E ) and, among them, the 5-HT 3A subunit is essential to form a functional receptor, whereas the 5-HT 3B subunit is known to alter the pharmacological properties of the channel [13,14].
In this study, we analyzed the possible role of 5-HT in T cell migration. We found that 5-HT selectively inhibits migration towards CXCL12 through the engagement of 5-HT 3 receptor. We propose that this process likely facilitates T cell infiltration in inflamed tissues.

5-HT 3 receptor activity modulates CXCL12-induced T cell migration
First, we investigated on the possible modulation of human and mouse CD4 + primary T cells chemotaxis by 5-HT. The cells were treated with 5-HT and allowed to migrate in a transwell assay towards a gradient of CXCL12, a chemokine that binds and triggers the chemokine receptor CXCR4. After stimulation with 5-HT, human CD4 + T cells displayed decreased migration, whereas the migration of mouse CD4 + T cells was not affected ( Figure 1A). The inhibition of CXCL12-induced human T cell migration exerted by 5-HT was dose-related in the range from 0.003 to 3 mM ( Figure 1B).
The 5-HT 3 receptor subtype is expressed in human but not mouse T cells [9,15] and thus we hypothesized that different effects of 5-HT on human and mouse T cell migration might depend on its specific expression. We evaluated T cell migration after incubation with specific 5-HT 3 receptor agonists, SR57227A and 2-methyl-5-hydroxytryptamine (2M-5HT). As shown in Figures 2A and 2B, 5-HT 3 agonists reduced, in a dose-dependent manner, CXCL12-induced migration of human resting T cells. We observed that 5-HT 3 receptor agonists and 5-HT had a similar inhibitory effect on T cell chemotaxis, suggesting that the effects of 5-HT on CXCL12-induced cell migration were most likely mediated by 5-HT 3 receptor engagement. Accordingly, the two selective 5-HT 3 receptor antagonists enhanced T cell chemotactic response to CXCL12 in the range from 1 to 100 mM ( Figure 2C-D). Similar results were obtained when human T cells activated by anti-CD3 and anti-CD28 antibodies (Abs) were used ( Figure 2E). The use of different chemicals, including 5-HT, did not alter neither spontaneous chemotaxis nor modified T cell chemokinesis (not shown). In addition, in line with the absence of the receptor, 5-HT 3 receptor agonist and antagonist did not affect the migration of mouse T cells ( Figure 2F).

Human T cells express functional 5-HT 3 receptor and produce 5-HT
These results prompted us to verify and characterize the expression of functional 5-HT 3 receptors in human CD4 + T cells. We employed real-time PCR analysis to evaluate the expression of 5-HT 3A and 5-HT 3B subunits in primary, resting T lymphocytes as well as in T cells activated by anti-CD3 and anti-CD28 Abs, or phytohaemagglutinin (PHA). Activation of CD4 + T cell was assessed by the up-regulation of activation surface markers, such as CD69 and CD25 (not shown). As shown in Figure 3A, both resting and activated human T cells expressed the 5-HT 3A receptor subunit. Moreover, the levels of 5-HT 3A transcript increased upon cell activation ( Figure 3A). In contrast, no expression of the 5-HT 3B subunit was found (not shown). Similar results were obtained by Western blot analysis, which showed that the 5-HT 3A subunit was constitutively expressed by CD4 + T cells and that its expression was augmented upon cell activation ( Figure 3B).
Next, we assessed the functional properties of the 5-HT 3A receptor in human CD4 + T lymphocytes through electrophysiological patch clamp experiments performed in freshly-isolated T cells in the whole-cell configuration ( Figure 3C-G). In naive T cells, Kv1.3, a voltage-gated, depolarization-activated potassium channel, is responsible for maintenance of the resting membrane potential (around 250 mV) [16]. In accordance, in primary T lymphocytes we recorded Kv1.3-mediated potassium outward currents, displaying typical Kv1.3 inactivation kinetics, at depolarizing voltages ( Figure 3C, left panel) but no current was observed at hyperpolarizing voltages. Upon addition of 5-HT to the bath, an inward current appeared at negative voltages, indicating that 5-HT alters ionic currents in T cells ( Figure 3C, right panel). In the experiments shown in Figure 3C, both pipette and bath solutions contained gluconate, which did not permeate chloride channels. Therefore, the outward current was due to the exit of potassium via Kv1.3, while the 5-HT-elicited inward current was due to the entrance of cations into the cell. Although 5-HT 3A homomers also allow calcium permeation, in our experimental setting the concentration of calcium in the extracellular solution was too low (1 mM) to give rise to a significant current [16], suggesting that the major charge carrying ion was sodium. In line with this, replacing sodium with N-methyl-D-glucamine ions, which did not permeate through sodium channels and 5-HT 3 receptors, resulted in no agonist-induced inward current (not shown). Next, to eliminate signals due to Kv1.3 activity, in the following experiments we applied hyperpolarizing potentials (#250 mV) ( Figure 3D-G) and cesium -a wellknown blocker of potassium channels -was the only cation present in the pipette solution. Under these ionic conditions, the addition of 2M-5HT to resting T cells, which did not display any current at negative potentials under control condition ( Figure 3D, left panel), elicited activation of negative, inward currents ( Figure 3D, right panel). Moreover, in another representative experiment, addition of 2M-5HT to cells at 290 mV holding potential in the whole-cell configuration caused a significant inward current within few seconds ( Figure 3E). T cells are known to express amiloridesensitive sodium channels [17]. To exclude the possibility that the current recorded after addition of 2M-5HT was due to sodium influx through amiloride-sensitive channels rather than 5HT 3 receptors, currents were recorded either in the absence ( Figure 3F, upper trace) or in the presence of amiloride ( Figure 3F, lower trace). Amiloride did not reduce 2M-5HT-induced currents, indicating that, upon addition of the 5-HT 3 receptor agonist, sodium did not enter the cells via classical amiloride-sensitive  Currents were recorded at 270 mV holding potential in outside-out patch after addition of 3 mM 2M-5HT (F, upper trace) and after addition of sodium channels. In accordance with this observation, 2M-5HT failed to induce inward currents in cells pre-treated with ondansetron ( Figure 3G).
Human T cells were found to express TPH-1 [18], thus we asked whether T cells were able to synthesize 5-HT. To this end, we performed an Ultra-Sensitive Serotonin enzyme immunoassay (EIA) of supernatants obtained from resting and activated T cell cultured in medium supplemented with fetal calf serum (FCS) that was previously treated with charcoal-coated dextran, in order to absorb the endogenous 5-HT. Consistently with published data [9], complete media supplemented with 10% FCS contained 276631 nM 5-HT, and this concentration was reduced to 1.160.3 nM following treatment with charcol-coated dextran (mean 6 SEM, n = 3). Despite the intra-donor variability, in all experiments performed we measured nanomolar concentrations of 5-HT released by both resting and activated cells ( Figure 4A). However, the 5-HT release was reduced upon T cell activation (P,0.01). These observations were confirmed by intracellular staining and flow cytometric analysis ( Figure 4B).

5-HT 3 receptor stimulation enhances T cell transmigration through CXCL12-coated endothelial cells
We asked whether 5-HT modulates chemotactic responses towards chemokines other than CXCL12. Thus, we evaluated the effects of 5-HT 3 receptor agonist ( Figure 5A) and antagonist ( Figure 5B) on human T cell migration towards CCL2 or CCL5 gradients. Both 2M-5HT and ondansetron had no effect on T cell migration towards the inflammatory chemokines CCL2 or CCL5, indicating that the 5-HT/5-HT 3 system did not alter T cell chemotactic responses to all chemokines and did not act on general mechanisms required for migration.
In order to investigate the mechanism of 5-HT effect on CXCL12/CXCR4 response, we evaluated CXCR4 expression, CXCL12 binding and CXCL12-induced actin polymerization in T cells after stimulation or inhibition of 5-HT 3 receptors. As shown in Figure 6, neither 5-HT nor 5-HT 3 receptor agonists or antagonists had any detectable effect on CXCR4 expression (Panel A), CXCL12 binding to its receptor (Panel B) or CXCR4mediated early signaling (Panel C).
CXCL12 is present on the surface of endothelial cells [19] and thus we evaluated if 5-HT might influence T cell extravasation, by regulating endothelial-bound CXCL12-chemoattraction. To address this question, we compared the ability of human T lymphocytes, treated or not with the specific CXCR4 antagonist AMD3100 [20], to migrate towards an inflammatory stimulus, such as CCL2, through CXCL12-coated endothelial cells. Endothelial-bound CXCL12 had an inhibitory effect on T cell transendothelial migration and, interestingly, 5-HT 3 receptor stimulation by 2M-5HT enhanced transendothelial migration of T cells, whereas 5-HT 3 receptor inhibition through ondansetron strongly reduced it ( Figure 7). All these effects were specifically determined by the presence of endothelial-bound CXCL12, as they were abrogated by AMD3100 (Figure 7). Moreover, they did not depend on the presence of CCL2, which was simply used as a chemoattractive stimulus, given that 2M-5HT or ondansetron exhibited no effect on migration towards this chemokine, as previously demonstrated ( Figure 5).

Discussion
The attraction of leukocytes to the sites of inflammation and infection is an essential component of the host response to diseases. The recruitment of circulating leukocytes from the blood stream into inflamed tissues is driven by chemokines, produced either by infiltrating inflammatory cells or by injured tissue cells. This process depends on three distinct phases: vascular attachment of leukocytes to the luminal side of blood vessels, extravasation (transendothelial migration) and cell migration.
Platelets, at the sites of inflammation, play a crucial role as amplifiers of leukocyte adhesion to the endothelium because of their repertoire of adhesion molecules and soluble mediators [21]. 5-HT, one of the soluble mediators locally released by platelets, is emerging as a key regulator of immunity, tuning in a very complex way both innate and acquired responses [3,6,22]. In agreement, 5-HT seems to be directly involved in the pathogenesis of several inflammatory diseases, including irritable bowel syndrome [23], colitis [24], contact allergy [25], atopic dermatitis [26], psoriasis [27] and asthma [28]. Moreover, recent studies have demonstrated that platelet-derived 5-HT aggravates T-cell induced liver damage in viral hepatitis [29,30].
The data reported in this study indicate that 5-HT inhibits migration of human T cells towards CXCL12 gradients and that this effect depends on the activity of 5-HT 3 receptors, which are functionally expressed in human T cells.
The presence of functional 5-HT 3 receptor has been already described in the Jurkat T cell line using the fluorescent probe SBFI/AM [31]. Here, through the patch clamp technique, we show the presence of 5-HT 3 receptor agonist-induced inward currents, which are abolished by pre-incubation with 5-HT 3 receptor antagonist, in human peripheral blood T cells, demonstrating for the first time that 5-HT 3 receptors are also active in human primary T cells.
Interestingly, our data indicate that T cells produce 5-HT and suggest that they use it as an autocrine factor to reduce responsiveness to CXCL12. Indeed, all our chemotaxis experiments have been performed in serum-free conditions and hence in the absence of external or contaminant sources of 5-HT. In these conditions, the boosting effect of 5-HT 3 receptor antagonists on T cell migration suggests -but does not formally prove -that selfproduced 5-HT controls responsiveness to CXCL12. In addition, T cells are able to sense an increase in the concentration of extracellular 5-HT, and respond with a further reduction of responsiveness to CXCL12 stimulation.
One puzzling aspect of our findings concerns the molecular basis for the selective action of the serotonin receptor 5-HT 3 on CXCL12/CXCR4. A trivial explanation for the selective control of CXCL12-induced migration may be a direct modulation of the expression of CXCR4 -the CXCL12 receptor -by 5-HT.
However, CXCR4 expression and its affinity for the chemokine are not altered by 5-HT or by 5-HT 3 receptor agonists or antagonists. Moreover, 5-HT 3 receptor stimulation or inhibition do not alter actin polymerization in response to CXCL12, indicating that the early events following CXCR4 stimulation are not affected by 5-HT. The mechanism responsible for the specific modulation of CXCL12-induced T cell migration by the 5-HT system is therefore still unclear and may involve pH i regulation. Chemotaxis is indeed modulated by the contribution of several different ion channels and by pH i changes [32][33][34][35], although the precise mechanisms are still not clear.
CXCL12 regulates both homeostatic and pro-inflammatory migration and homing of lymphocytes [36][37][38] and we suggest that the selective control of T cell responses to CXCL12 by 5-HT may be important in these contexts. Notably, it has been previously observed that the neutrophil elastase cleaves CXCL12 expressed on the bone marrow stroma, thus facilitating stem cell mobilization from bone marrow into the peripheral circulation [39,40]. In addition, another study has shown that elastase release by transmigrating neutrophils deactivates endothelial-bound CXCL12 and attenuates subsequent T lymphocyte transendothelial migration [41]. Our data indicate that an additional mechanism has evolved to selectively control responsiveness of lymphocytes to CXCL12. We suggest that, reducing responsiveness to endothelialbound CXCL12 chemoattraction, 5-HT facilitates transendothelial T cell migration towards gradients of inflammatory chemokines released in inflamed tissues. Data obtained with 5-HT 3 receptor antagonists do not exclude the possibility that T cells themselves may control their responsiveness to CXCL12 in an autonomous manner. On the other hand, since released local concentrations of 5-HT consequent to platelet activation at sites of damaged endothelium greatly exceeds the relatively low amounts produced by T cells, our straightforward interpretation is that during T cell transendothelial migration the major source of 5-HT is most likely produced by platelets. In addition, the fact that activated T cells produce lower amounts of 5-HT than resting ones but up-regulate expression of 5-HT 3 receptor may indicate that, upon activation, T cells become more dependent on external sources of 5-HT released, for example, by platelets.
In conclusion, our data unveil a novel connection between 5-HT and chemokines that may play a crucial role during tissue damage and inflammatory responses.

Ethics statement
Human peripheral blood T (PBT) cells were isolated from Buffy coats, a by-product of the isolation of blood coagulation factors, obtained from healthy donors at the Centre for Blood Transfusions of the Desio or Padua Hospital. The Centres for Blood Transfusions collected blood donations after a written informed consent, and did not release to our laboratory any personal data of the blood donors. For all these reasons, our Institution did not require approval from the local ethics committee. Procedures

Cell culture and activation
Human CD4 + PBT cells were isolated from the peripheral blood of healthy donors by negative selection using RosetteSep kit (StemCell Technologies) and cultured in RPMI 1640 medium (Lonza) supplemented with non essential amino acids,1 mM sodium pyruvate, 2 mM L-glutamine, 10 mM Hepes buffer (Lonza) (complete) and 10% FCS (Euroclone). In some experiments, human cells were activated by culture for 72 hours (h) in the presence of 2.5 mg/ml plate-bound functional grade purified anti-human CD3 Ab (OKT3; eBioscience) and 1 mg/ml of purified NA/LE anti-human CD28 Ab (CD28.2; BD Biosciences) in solution, with 2.5 mg/ml of PHA (Biochrom AG) or with a combination of 10 nM of phorbol 12-myristate 13-acetate (PMA; Sigma) and 1 mM of ionomycin (Sigma). T cell activation was determined by labeling with anti-human CD25 Ab (2A3; BD Biosciences) and anti-human CD69 Ab (FN50; BD Biosciences). The expression of CXCR4 was assessed by labeling with antihuman CD184 (CXCR4) Ab (12G5; BD Biosciences). Samples were analyzed using a FACS Canto flow cytometer and FACS Diva software (BD Biosciences).
Data were analyzed using the SDS2.2.2 software (Applied Biosystems). The fold increase corresponded to 2 2DDCt , where the DC t was the difference of the mean value of the C t between the studied and the housekeeping gene (18S ribosomal RNA), and the DDC t was the difference of the DC t between activated and resting cells for each gene. The standard deviation (SD) of the 2 2DDCt was calculated applying the formula: SD of DDC t * 2 2DDCt * ln2.

Western blotting
Human CD4 + T cells were collected on ice-cold 50 mM Tris-HCl (pH = 7.5), 2 mM EDTA containing protease inhibitors (Complete EDTA-free protease inhibitor cocktail tablets; Roche) (hypotonic buffer). Nuclei were eliminated by centrifugation (300 g, 10 min, 4uC) and membranes were isolated by further centrifugation (48.000 rpm, rotor Type 90 Ti, 45 min, 4uC) and solubilized in hypotonic buffer containing 1% NP40 (Calbiochem). Total protein concentration was determined by DC Protein assay (Bio-Rad Laboratories). Specifically, 20 mg of total protein for each sample were diluted in XT sample buffer (Bio-Rad Laboratories) containing 10% b-mercaptoethanol (Bio-Rad Laboratories), boiled for 10 min and stored at 220uC until use. Western blots were immunolabeled using polyclonal rabbit anti-human 5HT 3 receptor Ab (1 mg/ml; Imgenex) and secondary HRP-conjugated goat anti-rabbit IgG Ab (GE Healthcare). Membrane chemiluminescence was acquired using a ChemiDoc XRS system and the Quantity One -4.6.1 software (Bio-Rad Laboratories).

Patch clamp
Freshly isolated human CD4 + T lymphocytes were subjected to electrophysiological patch clamp experiments in the whole-cell configuration (or outside out patch in Figure 3F). Whole-cell currents were recorded with an EPC-7 amplifier (List) as previously described [43] and data were analyzed by using pCLAMP8 program set (Axon Instruments). Leak currents were not subtracted. The bath solution was composed of 156 mM Nagluconate, 1 mM CaCl 2 , 1 mM MgCl 2 , 10 mM HEPES (pH = 7.3) and the intracellular solution contained 134 mM Kgluconate, 1 mM CaCl 2 , 1 mM MgCl 2 , 10 mM EGTA, 10 mM HEPES (pH = 7.3) for experiments where Kv1.3 activity and 5HT 3 activities were recorded. In some control experiments, to block inward cation current, in the bath solution Nagluconate was replaced with N-methyl-D-glucamine chloride (NMDGCl). Where indicated, intracellular Kgluconate was replaced with 134 mM CsCl (pH to 7.3 with CsOH) to block outward potassium current and the bath contained NaCl instead of Nagluconate.

CXCL12 binding assay
T cells were washed once with Hanks' balanced salt solution containing 20 mM Hepes and 0.2% BSA (Sigma) at pH = 7.4 and resuspended in the same buffer. First, they were treated for 30 min at RT with the indicated chemicals and then incubated for 30 min at RT with different concentrations of CXCL12-AF647 (Almac Sciences) in the range from 0 to 1000 ng/ml. Thereafter, cells were washed twice in assay buffer and fixed in 1% paraformaldehyde in PBS. MFI was measured by FACS Canto flow cytometer and analyzed with FACS Diva software.

Actin polymerization assay
After 5 hours of serum starvation in RPMI 1640 medium, T cells were treated for 30 min at RT with the indicated chemicals and then stimulated with CXCL12 (100 nM) at 37uC in a rotary shaker. At indicated time points, stimulation was stopped by adding 4% formaldehyde. Intracellular staining for F-actin was performed by AlexaFluor488-phalloidin (Molecular Probes) using the BD Cytofix/Cytoperm kit (BD Biosciences) according to the manufacturer's instructions. MFI was measured by FACS Canto flow cytometer and analyzed with FACS Diva software.

Migration assays
Cell migration in response to chemokine was tested using Transwell assays (24-well plate with inserts, 3 mm pore/6.5 mm, TC-treated; Costar). Briefly, after 2 h of serum starvation in migration medium (serum-free RPMI 1640 supplemented with 0.1% BSA), 5-HT, 5-HT 3 receptor agonists or antagonists were added for 30 min at RT to cell suspension and then cell were used for migration assays. 5-HT 3 receptor agonists and antagonists were present during the migration in both transwell compartments. Human (1.5 nM) and murine (6.3 nM) CXCL12 were purchased from PeproTech; human CCL2 (11.5 nM) and CCL5 (12.8 nM) were purchased from R&D; chemokines were all diluted in migration medium. The number of living cells spontaneously migrating through the filter was determined after 2 h at 37uC by counting all the samples for the same time (1 min) at FACS Canto flow cytometer.
HUVECs were seeded at 15-20610 3 /well and cultured to 90% confluence on 24-well transwell tissue culture inserts (24-well plate with inserts, 5 mm pore/6.5 mm, TC-treated; Costar) as previously described [42]. HUVECs were activated using IL-1b (Peprotech) at 10 ng/ml final concentration for 4 h at 37uC in HUVEC-culture medium and then incubated with CXCL12 20 ng/ml final concentration in migration medium for 15 min at 37uC. T cells were treated for 30 min at RT with the indicated chemicals, then incubated with AMD3100 10 uM and immediately allowed to migrate towards CCL2 (11.5 nM) for 30 min at 37uC. AMD3100, 5-HT 3 receptor agonists and antagonists were present during the migration in both transwell compartments. Migration experiments displayed an inter-donor variability in the absolute number of migrating T cells, thus for each experiment migration data have been normalized considering the number of counted cells in control conditions as 100%.

Statistical analysis
All experiments were performed at least three times. All results are expressed as mean 6 the standard error of the mean (SEM). The statistical significance was calculated using a Student's t-test or a oneway ANOVA, Bonferroni's multiple comparison test (GraphPad Prism 4). Differences were considered significant if P,0.05.