The aim of this study was to investigate the effect of a chronic serotonin treatment on astrocytes and neurons derived from healthy controls and patients, focusing on mitochondrial function and electrophysiology. The two case study patients were selected for our previous study exploring the ways in which mitochondria can influence cellular function and contribute to the development of depression [33].
The relative expression of serotonin receptors was significantly higher in neurons compared to astrocytes, especially HTR1A. Notably, while HTR1A and HTR3A expression were low in astrocytes, both neurons and astrocytes robustly expressed HTR2A. This suggest that the effects of serotonin in astrocytes were mainly mediated by 5-HT2A signalling. All three serotonin receptors had a similar expression in neurons.
In astrocytes, serotonin downregulated SLC1A2 and HTR2B, encoding the glutamate transporter EAAT2 and the 5-HT2B receptor, respectively. The downregulation of 5-HT2B suggests it had an additional role in mediating serotonin effects in astrocytes and may have been downregulated as a compensatory mechanism. Notably, Chen et al. reported that stimulation of the 5-HT2B receptor in astrocytes increases glutamate production [43]. It is therefore conceivable that the downregulation of both SLC1A2 and HTR2B in response to serotonin might be functionally interrelated, although further studies are needed to explore this potential connection.
Serotonin treatment led to significant alterations in neuronal gene expression. The downregulation of OPA1, MFN1 and DNM1L suggests reduced mitochondrial fusion and fission. While this could be indicative of a compromised mitochondrial network, it may also represent an adaptation to mitigate the impact of cellular stressors. The downregulation of cytochrome C, VDAC1 and VDAC2 mRNA also suggest an impact of serotonin on the metabolic state of neurons. Noteworthy, our results contrast with findings presented by Fanibunda et al. in rat primary neurons, where serotonin treatment exerted a positive trophic effect accompanied by an upregulation of PPARGC1A and SIRT1 expression [32]. HIF-1α downregulation could alter glucose metabolism in neurons. Notably, Shibata et al. have observed trait-dependent expression levels of HIF-1α mRNA in MDD patients’ white blood cells, with remissive patients showing a decreased HIF-1α expression [44]. It is possible that in the latter study, increased serotonin from SSRI medication provoked the HIF-1α downregulation, and that the same mechanism was responsible for the serotonin-mediated HIF-1α downregulation we observed. However, the precise mechanism and its implications requires further investigation.
A notable effect of serotonin on astrocyte respiration was observed only in Ctl17 where several respiratory parameters decreased. This aligns with findings in the rat brain [45]. Contrastingly, Fanibunda et al. found that serotonin increased respiration in rat primary neurons [32]. Assuming that respiration rates are comparable in neurons and astrocytes [46], the decreased respiration in whole brain and certain of our astrocyte, coupled with the reported increase in neurons, suggests a potential opposing effect of serotonin on mitochondrial respiration in astrocytes and neurons. Technical constraints prevented the direct measurement of neuronal OXPHOS in our study. Exploring this aspect in future research would be valuable. Additionally, it is important to consider potential metabolic differences between rodent and human brain cells. They could partly account for discrepancies between our observations and reports in the literature. To the best of our knowledge, this is the first study investigating the effect of serotonin on the mitochondrial and cellular function of human astrocytes.
Although serotonin decreased or did not alter astrocytic respiration in our human model, it seemed to increase ATP levels, suggesting a glycolysis elevation. Consistently, studies showed that serotonin increases glucose uptake by increasing surface expression of glucose transporters [47], and that it stimulates glycolysis by upregulating and activating the rate limiting enzyme phosphofructokinase [48]. Future studies investigating the rate of glycolysis and glucose uptake in serotonin-treated astrocytes would be insightful.
Serotonin altered cytosolic and mitochondrial Ca2+ levels in astrocytes in a way that may suggest an equalizing effect. Indeed, while both the Non-R and the Mito patient initially exhibited elevated cytosolic Ca2+ levels relative to their matched controls, serotonin treatments significantly reduced Ca2+ in these patients’ astrocytes. Further supporting a normalizing influence of serotonin on altered Ca2+ homeostasis, low mitochondrial Ca2+ levels in the Non-R patient increased with serotonin treatment, while elevated Ca2+ in the Mito patient decreased. Serotonin receptors from the 5-HT2 family are coupled to phospholipase C (PLC) signaling and lead to increased Ca2+ levels both by mobilization of internal stores and opening of Ca2+ channels. These mechanisms could play a role in the observed compensatory effect of serotonin.
Remarkably, in neurons, serotonin consistently decreased Ca2+ levels in all cell lines and slowed Ca2+ transients. Lower cytosolic Ca2+ levels could result from an upregulation of the plasma membrane Ca²⁺-ATPase or the sodium-calcium exchanger, leading to enhanced Ca2+ efflux. Decreased cytosolic Ca2+ and slower Ca2+ transients in neurons can influence neurotransmission and impact Ca2+-mediated short-term plasticity mechanisms. Decreasing Ca2+ may also be a mechanism to prevent excitotoxicity.
In addition to cytosolic Ca2+, serotonin consistently decreased mitochondrial Ca2+ levels. This could result from lower cytosolic Ca2+ levels to be taken up through the mitochondrial Ca2+ uniporter or be related to the decreased VDAC expression in serotonin-treated neurons. Indeed, VDAC is considered as the main entry point of Ca2+ into mitochondria [49]. Moreover, mitochondrial distribution, partly mediated by fusion and fission, influences mitochondrial Ca2+ levels [50]. Therefore, decreased mitochondrial dynamics, as suggested by mRNA expression, could play a role in the lower Ca2+ levels. A decrease in mitochondrial Ca2+ levels can significantly impair OXPHOS, and therefore ATP production in neurons.
Serotonin had a consistent depolarizing effect on resting membrane potential (RMP), especially marked in Ctl18 and Mito patient’s neurons. Notably, the inhibition of that effect by M100907, a specific 5HT2A receptor antagonist, suggested it was mediated by 5HT2A signalling. Neurons with depolarized RMP would require smaller stimuli to reach the threshold for action potential (AP) generation. Therefore, serotonin appears to render cortical neurons more excitable. Aligning with this, slice recordings from prefrontal cortical neurons indicate depolarizing effects following 5-HT2A receptor activation [51].
5HT2A receptor stimulates PLC, which hydrolyses phosphatidylinositol (4,5) bisphosphate (PIP2) to produce inositol triphosphate (IP3) and diacylglycerol (DAG). DAG and the IP3-mediated Ca2+ release activate protein kinase C (PKC) [42]. Two components of the Gαq signalling cascade are known to modulate ionic channels and could influence the RMP.
First, PKC regulates ionic channels through mechanisms including altered surface expression and probability of opening [52]. Fryckstedt et al. demonstrated that serotonin decreased the activity of the Na+/K+ ATPase in a PKC-dependent manner [53]. This pump is particularly significant for maintaining a hyperpolarized RMP. Thus, a decrease in its activity could contribute to a depolarized RMP.
Second, PIP2 in the plasma membrane appears to control channel gating. For instance, PIP2 activates the two-pore domain potassium (K2P) channels, which mediate K+ background currents that maintain a hyperpolarized RMP [52]. Therefore, a plausible mechanism underlying the depolarizing effect of serotonin involves the activation of PLC via the 5-HT2A receptor, resulting in PIP2 hydrolysis, decreasing the activation of K2P channels, and consequently depolarizing the RMP.
Serotonin had a remarkable effect on current densities in neurons: both Na+ and K+ current densities were significantly increased in all cell lines. These results suggest that serotonin-treated neurons undergo faster depolarization and repolarization during APs. This could have profound implications on synaptic transmission.
This effect was not mediated by the 5HT2A receptor, as it was exacerbated byM100907. This may suggest that a constitutive activity of 5HT2A receptor partly mitigated the serotonin-mediated increase in Na+ and K+ current density, and that the antagonist lifted that restriction. Constitutive activity of 5HT2A receptors has been reported before and is interestingly believed to play a role in the etiology of depression and in the effect of anti-depressive therapies [54].
Serotonin also prolonged the rise time and decay time of postsynaptic currents (PSCs) in most cell lines. This could affect the integration of events at the soma and lead to an increased summation if the broader excitatory PSCs overlap. Overall, this suggests a stronger effect of the PSCs and therefore an increased excitability.
AP kinetics were also altered by serotonin. The lower threshold observed in most cell lines is consistent with increased Na+ current density and could also indicate a shift in the activation properties of voltage-gated Na+ channels. It suggests weaker stimuli can induce APs. Taken together with a depolarized RMP, serotonin seems to generally increase neurons’ excitability. Higher Na+ current density could also underlie the serotonin-induced increased AP amplitude in Ctl17 and Non-R neurons.
Limitations
Interindividual differences could be exacerbated by the differences in the age of the patient/control pairs [33].
We recognize that our results reflect the observations of independent individuals and cannot be interpreted as generally applicable interpretations of MD and TRD.
Reprogramming fibroblasts may affect the expression of disease-associated epigenetic memories. However, we have already shown that functional mitochondrial phenotypes are transmitted (at least partially) to the iPS-derived lineages [18, 33].
While our cellular models have provided valuable insights into mitochondrial dysfunction in MDD pathophysiology, we recognize the inherent complexity of in vivo systems. Extrapolating findings from isolated cellular contexts to the intricacies of whole organisms involves inherent limitations.