Activation of the glucocorticoid receptor rapidly triggers calcium‐dependent serotonin release in vitro

Abstract Aims Glucocorticoids rapidly provoke serotonin (5‐HT) release in vivo. We aimed to investigate molecular mechanisms of glucocorticoid receptor (GR)‐triggered 5‐HT release. Methods Employing 1C11 cells to model 5‐HT neurotransmission, immunofluorescence and Pearson's Correlation Coefficient were used to analyze colocalization of GR, 5‐HT, vesicle membrane protein synaptotagmin 1 and vesicle dye FM4‐64FX. FFN511 and FM4‐64FX dyes as well as calcium imaging were used to visualize vesicular 5‐HT release upon application of GR agonist dexamethasone, GR antagonist mifepristone and voltage‐gated calcium channel (VGCC) inhibitors. Results GR, 5‐HT, synaptotagmin 1 and FM4‐64FX showed overlapping staining patterns, with Pearson's Correlation Coefficient indicating colocalization. Similarly to potassium chloride, dexamethasone caused a release of FFN511 and uptake of FM4‐64FX, indicating vesicular 5‐HT release. Mifepristone, calcium depletion and inhibition of L‐type VGCC significantly diminished dexamethasone‐induced vesicular 5‐HT release. Conclusions In close proximity to 5‐HT releasing sites, activated GR rapidly triggers L‐type VGCC‐dependent vesicular 5‐HT release. These findings provide a better understanding of the interrelationship between glucocorticoids and 5‐HT release.


| INTRODUC TI ON
The serotonin (5-HT) system affects various physiological functions such as circadian rhythm, appetite, behavior and learning.
Alterations of the 5-HT system are believed to play a pivotal role in the pathogenesis of many psychiatric diseases such as depression. 1 In serotonergic neurons, 5-HT is synthesized from L-tryptophan and packed into vesicles. 2 Excitation of 5-HT neurons triggers calcium influx via voltage-gated calcium channels (VGCCs). 3 At the center of the neurotransmitter release process, calcium ions then bind to the vesicle membrane-bound calcium sensor synaptotagmin 1. Upon calcium binding, synaptotagmin 1 forms a complex with soluble NSF attachment protein receptors (SNAREs) and the plasma membrane, which causes vesicle fusion pore opening and subsequent neurotransmitter liberation. 4-7 5-HT neurons also release neurotransmitters from extrasynaptic sites, including the soma. This release mode is known for other monoamine neurotransmitters as well and is termed volume transmission. 8,9 Administration of glucocorticoids triggers a rapid increase in extracellular 5-HT concentration in vivo, which is blocked by the glucocorticoid receptor (GR) antagonist mifepristone. 10 This 5-HT release occurs too quickly to be mediated by the slow, genomic pathway of glucocorticoid action, 11 and is more likely caused by rapid, non-genomic GR actions on vesicle release. 12 Glucocorticoids were shown to have various rapid effects on 5-HT neurons, for example, impacting 5-HT synthesis by affecting tryptophan hydroxylase 2 mRNA levels or causing a rapid upregulation of the cell surface expression of the 5-HT transporter. 13,14 However, it remains unknown how glucocorticoids rapidly trigger 5-HT release.
We employed murine stem cell-derived 5-HT neurons 15,16 to demonstrate that GR resides in close proximity to 5-HT and vesicle release sites, which is a prerequisite for rapid action of GR on vesicular 5-HT liberation. We then show that GR activation initiates rapid F I G U R E 2 Depolarization via potassium chloride triggers FM4-64FX dye uptake in synaptic vesicles in 1C11 5-HT . 1C11 5-HT were maintained in imaging buffer before application of either imaging buffer supplemented with 2 µM FM4-64FX (FM; top row in A and B) or reaction buffer, which contained potassium chloride (KCl) and was also supplemented with FM (bottom row in A and B  15,16 Briefly, 1C11 cells were kept on 100 mm plates (Sarstedt) in DMEM Glutamax with 10% fetal bovine serum, 1% non-essential amino acids, 1% penicillin/streptomycin, and 1% L-glutamine (all Life Technologies) at 37°C and 5% CO 2 . For differentiation to serotonergic 1C11 cells (1C11 5-HT ), 40,000 cells were transferred to a 3.5 cm plate (Sarstedt) containing 3 coverslips or slide chambers (Ibidi) for live cell imaging. Then, culture medium was supplemented with 1 mM dibutyryl cyclic adenosine monophosphate (cAMP) and 0.05% cyclohexanecarboxylic acid for 4 days. On day 4, 1C11 5-HT were shown to display a complete 5-HT metabolism. 16 Fetal bovine serum in the medium, which contains 5-HT, does not impact differentiation to a complete 5-HT phenotype, but reduces 5-HT synthesis, content and uptake by activation of 5-HT 2B and 5-HT 1B/D autoreceptors. 16 As we did not quantify 5-HT synthesis, content

| Colocalization analysis
For immunostainings, 1C11 5-HT were fixed in 1% paraformaldehyde and subsequently permeabilized using 0.1% saponin in blocking so-  staining of 1C11 5-HT showed a similar distribution pattern as observed for synaptotagmin 1. Cells treated with either N-type, R-type or P/Q-type voltage-gated calcium channel (VGCC) inhibitors (inhib.) showed similar FFN511 release upon depolarization compared to cells unexposed to VGCC inhibitors (each cell was exposed to one VGCC blocker only). Inhibition of L-type VGCCs, however, prevented potassium chloride (KCl)-provoked FFN511 release. Extracellular calcium concentration was 5 mM. Within panels of each VGCC inhibitor, arrows and asterisk depict the same neurons; for the P/Q-type VGCC inhibitor, the top neuron dislocated during buffer exchange. Scale bars: 25 µm. (B) Kruskal-Wallis test revealed significant differences between groups (χ 2 (5, N = 114) =90.88, p < 0.001). Post-hoc analysis using Dunn's multiple comparison test showed no significant difference in FFN511 signal intensity between resting cells and L-type VGCC inhibitorexposed, stimulated cells (p = 0.752), whereas stimulated cells exposed to other VGCC blockers significantly lost FFN511 signal intensity (0 mM KCl vs. 60 mM KCl and N-type VGCC blocker (p = 0.006); 0 mM KCl vs. 60 mM KCl and R-type VGCC blocker (p = 0.03); 0 mM KCl vs. 60 mM KCl and P/Q-type VGCC blocker (p = 0.011)). Results of 3 independent live cell imaging experiments are shown, in which a total of 114 cells were analyzed (n 1 = 35; n 2 = 40; n 3 = 39). Data are displayed as median and interquartile range. (C) Fluo4-AM was employed to visualize high KCl-triggered calcium influx. While treatment with neither N-type, R-type or P/Q-type VGCC inhibitors significantly diminished calcium influx upon depolarization, treatment with L-type VGCC inhibitors strongly reduced calcium ion influx (each cell was treated with one VGCC blocker only). Extracellular calcium concentration was 5 mM. Scale bars: 25 µm. (D) Kruskal-Wallis test revealed significant differences between groups (χ 2 (4, N = 27) =21.57, p < 0.001). Post-hoc analysis using Dunn's multiple comparison test showed no significant difference in Fluo4-AM intensity between stimulated 1C11 5-HT unexposed to VGCC inhibitors and stimulated 1C11 5-HT exposed to either N-type (p = 0.788) or R-type (p > 0.999) VGCC inhibitors (for the P/Q-type VGCC inhibitor, only descriptive data is presented).

| GR colocalizes with 5-HT and synaptotagmin 1
To investigate a possible contribution of GR to 5-HT release, we first analyzed the localization of GR in comparison to 5-HT itself and to synaptotagmin 1, a protein involved in vesicular 5-HT release, with immunofluorescence analysis performed in 1C11 5-HT (Figure 1). In cells double-stained for GR and 5-HT, we found that GR localized to cell bodies, including the nucleus, but also to neurites ( Figure 1A

| Depolarization with potassium chloride triggers L-type voltage-gated calcium channeldependent vesicular 5-HT release
Up to date, the molecular framework how GR activation causes a rapid release of 5-HT on the cellular level is not completely understood. To provide more insight into this process, we employed 5-HT neuron-mimicking 1C11 5-HT . As reported previously, high potassium-induced vesicular 5-HT release from 1C11 5-HT can be concentrations also induced calcium influx required for neurotransmitter release, which was visualized using Fluo4-AM (example images in Figure 3C and data analysis in Figure 3D;

| GR activation triggers L-type voltage-gated calcium channel-dependent vesicular 5-HT release
To simulate GR activation in vitro, we applied the selective agonist  Figure 4A and B, data analysis in Figure 4C). When we applied 20 nM dexamethasone in cells loaded with FFN511, a significant decrease of FFN511 fluorescence intensity was observed, indicating a vesicular, GR-dependent dye release ( Figure 5C; Kruskal-Wallis test (χ 2 (3, N = 78) = 40.29, p < 0.001; post-hoc Dunn's multiple comparison test compared to resting cells: p < 0.001). Moreover, dexamethasone-induced FFN511 release occurred within seconds after application, likely indicating a rapid, non-genomic action of GR (exemplary images in Figure 5A, data analysis in Figure 5B). High potassium-induced FFN511 release was not affected by prior incubation of 1C11 5-HT with the GR antagonist mifepristone, whereas dexamethasone-induced FFN511 release was blocked by mifepristone ( Figure 5B). Our previous experiments showed that calcium ion influx via L-type VGCCs contributes to 5-HT release from 1C11 5-HT . Indeed, the absence of calcium ions in the reaction buffer blocked dexamethasone-induced FFN511 release ( Figure 5C; post-hoc Dunn's multiple comparison test: Quantification of FFN511 release from 1C11 5-HT . Kruskal-Wallis test revealed significant differences between groups (χ 2 (3, N = 78) =40.29, p < 0.001). Post-hoc analysis using Dunn's multiple comparison test showed that compared to resting cells, FFN511 signal intensities were significantly diminished in the presence of Dex and calcium ions (p < 0.001), whereas no significant loss of FFN511 intensity was observed when Dex was applied in the absence of calcium ions (p > 0.999). FFN511 intensities were significantly higher in cells exposed to Dex without calcium ions than in cells exposed to Dex with calcium ions (p = 0.003). Results of 3 independent live cell imaging experiments are shown, in which a total of 78 cells were analyzed (n 1 = 25; n 2 = 25; n 3 = 28). Data are displayed as median and interquartile range. (D) FFN511 intensity was quantified after the application of Dex and the L-type VGCC inhibitor nifedipine (Nif). Kruskal-Wallis test showed significant differences between groups (χ 2 (3, N = 150 We employed FM4-64FX as well as FFN511 to visualize GRinduced vesicular release in 1C11 5-HT , as both dyes are wellestablished agents to visualize synaptic, exocytotic events. 3,20,25,26 Dexamethasone resulted in a rapid increase of FM4-64FX and decrease of FFN551 signal intensity, similar to depolarization with high potassium. This change in signal intensity was abolished by prior application of the GR antagonist mifepristone, showing that GR activation by a corticosteroid is required to induce vesicular 5-HT release. In line with the postulated serotonergic volume transmission, 8,9,31 we were able to detect vesicular release not only at neurite terminals but also on the soma and along neurites of 1C11 5-HT .
Our results show that calcium influx is at the center of GR-induced vesicular 5-HT release, similar to depolarization-induced 5-HT release. 32 Interestingly, a study on dopaminergic neurons, which rely on volume transmission just like 5-HT neurons, 8,9 showed that blocking L-type VGCC with nimodipine abolished depolarization-induced dopamine release. 36 Further, GR-induced 5-HT release shows similarities to melatonin-mediated 5-HT liberation. In more detail, excitation induces 5-HT release, but the release response increases when neurons are simultaneously exposed to melatonin. As this melatonin-mediated enhancement of 5-HT release is abolished by incubation with the Ltype VGCC blocker nifedipine, it was hypothesized that the melatonin receptor is coupled with L-type VGCCs, thereby facilitating 5-HT release. 37 As a limitation, however, this study cannot exclude that VGCCs other than the L-type VGCC contribute to calcium-induced vesicular release in 1C11 5-HT . While inhibition of the L-type VGCC significantly reduced vesicular release and calcium influx, it did not completely abolish it. Thus, future research needs to compare the contribution of different VGCCs to vesicular 5-HT release.
The colocalization between GR, 5-HT, synaptotagmin 1, which resides in the membrane of neurotransmitter vesicles, and the styryl vesicle dye FM4-64FX suggests that GR might have a relatively direct, non-genomic effect on the opening of L-type VGCC in proximity to vesicle release sites, as it is hypothesized for the melatonin receptor. 37 Considering the extranuclear GR localization, especially on neurites, this subcellular population of GR molecules may not contribute to genomic events at all. A study by Di et al. 38 suggests that this GR population is bound to the cell membrane, as dexamethasone bound to bovine serum albumin, which cannot cross cell membranes, still caused a rapid release of endocannabinoids.
Further evidence for membrane-bound GR is provided by a study using electron microscopy, which detected the presence of GR in postsynaptic membrane densities. 39 Alternatively, it has been proposed that rapid glucocorticoid effects are mediated by cytosolic GR, which associates with the cell membrane similar to the intracellular estrogen receptor. 40

| CON CLUS ION
This work extends our understanding how glucocorticoids interact with 5-HT neurons. In murine stem cell-derived 5-HT neurons, the proximity of extranuclear GR, 5-HT and synaptotagmin 1 suggests that GR might be directly involved in rapid 5-HT release.
Glucocorticoid binding to GR initiates a rapid vesicular 5-HT release, which depends on extracellular calcium influx via L-type VGCCs.
Extending the knowledge of the acute and chronic effects of glucocorticoids on 5-HT neurons might ultimately contribute to the understanding of the underlying mechanisms of stress-related psychiatric disorders.

ACK N OWLED G M ENTS
We thank Nasser Haddjeri, Stem Cell and Brain Research Institute, INSERM 1208, and Guillaume Lucas, Institut François Magendie, INSERM U1215, for their discussions and comments on the manuscript. We kindly thank Patrick Schloss for his initial support and scientific input to this project. The work of Thorsten Lau was supported by the Hector Stiftung II. The graphical abstract and Data S1 were created with BioRender.com.

CO N FLI C T S O F I NTE R E S T
The authors declare that the research was conducted in the absence of any financial support or relationships that could be construed as a potential conflict of interest.

AUTH O R CO NTR I B UTI O N S
TL, NP, TW and DB contributed to the conception of the study.
TL, NP, JR and SL conducted the experiments and analyzed the data. TL and NP prepared the manuscript. All authors contributed to critical revision of the manuscript and approved the submitted version.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are available from the corresponding author upon reasonable request.