Bipolar‐associated miR‐499‐5p controls neuroplasticity by downregulating the Cav1.2 subunit CACNB2

Recently, variants in the MIR499 gene have been associated with BD (Forstner et al, 2015). The main mature miRNA expressed from the MIR499 locus is miR‐499‐5p, whose function has been extensively studied in the cardiovascular system but very little in the brain. To investigate a possible involvement of miR‐499‐5p in the molecular pathophysiology of BD, we decided to study its function in the rat model which is easily amenable for experimental manipulation. We first measured miR‐499‐5p levels in different brain regions of the adult rat brain using qPCR. The expression of miR‐499‐5p was stably detected in all the tested brain regions (Fig 1A), including areas involved in the regulation of affect regulation and cognitive functioning, such as the hippocampus and frontal cortex (Millan et al, 2016; Toda et al, 2019). Juvenile social isolation (JSI) in rats is a common environmental model to study different facets of mental disorders, such as anxiety, drug addiction, and cognitive performance (Fig EV1A; Fone & Porkess, 2008; Valluy et al, 2015; Braun et al, 2019). As expected, JSI rats showed significantly lower hippocampal c‐fos and Arc expression (Fig EV1B and C), neural activity‐induced genes which are known to be down‐regulated upon social isolation in the rat hippocampus (Pisu et al, 2011; Begni et al, 2020). Importantly, JSI induced a highly significant up‐regulation (≈ 3.5 fold) in the expression of miR‐499‐5p compared to group‐housed animals (Fig 1B). The JSI‐mediated miRNA upregulation was specific for miR‐499‐5p, since hippocampal levels of two miRNAs previously implicated in psychiatric disorders (Xu et al, 2010; Lopez et al, 2017) were either reduced (miR‐146a) or unaltered (miR‐30e; Fig EV1D and E) upon JSI. Thus, upregulation of miR‐499‐5p in BD patients is paralleled by elevated miR‐499 levels in the JSI rat model of mental disorders.

We next decided to study the function of miR‐499‐5p at the level of individual brain cells. Using smFISH in dissociated rat primary hippocampal neuron cultures, we detected miR‐499‐5p positive puncta (red, Fig 1C) in the soma and, to a lesser extent, dendrites of excitatory pyramidal neurons which co‐expressed the marker Camk2a mRNA (green, Fig 1C). miR‐499‐5p expression increases during a time‐course of hippocampal neuronal development from DIV 4–20 as assessed by qPCR (Fig EV1F), consistent with a function of miR‐499‐5p in dendrite and/or synapse development. This increase is also observed in glia‐depleted cultures (Fig EV1F), demonstrating the neuronal source of miR‐499‐5p expression. Given our results obtained from JSI in rats (Fig 1B), we studied a potential regulation of miR‐499‐5p by stress signaling in rat hippocampal neuron cultures. Glucocorticoids (GCs) are released by the hypothalamic–pituitary–adrenocortical (HPA) axis in response to stressful events and act on neurons via mineralocorticoid (MRs) and glucocorticoid receptors (GRs) to modulate adaptive responses to stress (Finsterwald & Alberini, 2014; Herman et al, 2016). Moreover, high chronic exposure to GCs results in neurotoxicity in models of chronic stress and major depressive disorder (MDD; Dienes et al, 2013). We observed that a 5‐day treatment of developing hippocampal neurons with the GR agonist Dexamethasone (DEX) significantly induced the expression of miR‐499‐5p (Fig EV1G), which suggests that stress signaling triggers miR‐499‐5p expression in neurons. Next, we explored potential effects of excessive expression of miR‐499‐5p, that is, observed by early life adversity or the activation of stress signaling on neuroplasticity. Therefore, we studied dendritogenesis as a paradigm since defects in dendritic arborization in the hippocampus are frequently observed in BD patients. To model excessive miR‐499‐5p activity, we transfected hippocampal neurons with miR‐499‐5p duplex oligonucleotides (“mimics”), which leads to a strong miR‐499‐5p overexpression (Fig EV1H). Sholl analysis revealed that hippocampal neurons transfected with a miR‐499‐5p mimic displayed a reduced number of intersections compared to control conditions (Figs 1D and E, and EV1I), indicating reduced dendritic arborization upon miR‐499‐5p overexpression. Calcium influx through LVGCCs is required for activity‐dependent dendritogenesis (Redmond & Ghosh, 2005). In addition, both the Ca_v1.2 pore‐forming Cacna1c and the regulatory beta‐subunit Cacnb2 are BD risk genes, and impaired cellular calcium homeostasis is a hallmark of BD (Harrison et al, 2019).This led us to hypothesize that Ca_v1.2 activity could be downstream of miR‐499‐5p in the control of dendritogenesis. In agreement with this hypothesis, hippocampal pyramidal neurons prepared from constitutive heterozygous Cacna1c^+/− rats, a well‐established mental disease animal model (Braun et al, 2018), displayed less complex dendritic branches than WT controls (Fig 1F and G), thereby mimicking the effect of miR‐499‐5p overexpression.

We decided to further explore a potential direct regulation of Ca_v1.2 activity by miR‐499‐5p. By applying miRNA‐binding site prediction tools to genes encoding Ca_v1.2 subunits, we detected a highly conserved miR‐499‐5p‐binding site in the 3′UTR of Cacnb2, but not Cacna1c (Fig 2A). CACNB2 is a validated target of miR‐499‐5p in cardiomyocytes (Ling et al, 2017) and in a very recent GWAS with more than 40,000 BD patients, the CACNB2 locus was significantly associated with BD (Mullins et al, 2021). Therefore, we decided to investigate whether miR‐499‐5p and CACNB2 functionally interact in rat hippocampal neurons to control dendritogenesis.

We first studied the effect of miR‐499‐5p overexpression on Cacnb2 expression. Hippocampal neurons transfected with a miR‐499‐5p mimic exhibited a significant reduction in the levels of Cacnb2 mRNA compared to neurons expressing the control mimic as judged by qPCR (Fig 2B). The same treatment led to reduced Cacnb2 protein levels as assessed by Western blot, although the difference between miR‐499‐5p mimic and control transfected neurons did not reach statistical significance (Fig EV2A and B). Furthermore, Cacnb2 mRNA levels were significantly lower in the hippocampus of JSI compared to group‐housed rats (Fig 2C). The direction of change in socially isolated rats is opposite to the one observed for miR‐499‐5p (Fig 1B), consistent with a negative regulatory role for miR‐499‐5p in Cacnb2 expression. To test a possible direct interaction of miR‐499‐5p with the Cacnb2 3′UTR, we performed reporter gene assays in primary hippocampal neurons using luciferase genes fused to either the wild‐type (WT) Cacnb2 3′UTR or Cacnb2 3′UTR containing mutations in the predicted miR‐499‐5p seed match (MUT; Fig 3C). Overexpression of miR‐499‐5p resulted in a significant repression of luciferase reporter gene expression for the Cacnb2 WT plasmid (Fig 2D), consistent with a posttranscriptional, 3′UTR‐dependent repressive effect of miR‐499‐5p on Cacnb2. This effect was not observed when the Cacnb2 MUT plasmid was transfected (Fig 2D), demonstrating that miR‐499‐5p mediated its repressive effect via direct interaction with the seed match site in the Cacnb2 3′UTR. On the other hand, the inhibition of endogenous miR‐499‐5p in neurons by transfection of an LNA‐modified antisense oligonucleotide (miR‐499‐5p pLNA) led to a significant increase in the expression of the Cacnb2 WT, but not Cacnb2 MUT, luciferase reporter gene (Fig 2E). Together, these experiments establish Cacnb2 as a bona‐fide miR‐499‐5p target gene in rat hippocampal neurons. We next investigated whether miR‐499‐5p‐mediated regulation of Cacnb2 is functionally involved in dendritogenesis. Therefore, we transfected hippocampal neurons with a miR‐499‐5p mimic and examined whether restoring CACNB2 expression by co‐transfection of a Cacnb2 expression plasmid which lacks the 3′UTR (pMT2‐CACNB2; Fig EV2C) was able to rescue impaired dendritogenesis observed upon miR‐499‐5p overexpression. Notably, whereas overexpression of miR‐499‐5p reduced dendritic complexity as expected, dendritic complexity of neurons co‐transfected with miR‐499‐5p and pMT2‐CACNB2 was not significantly different compared to control conditions (Fig 2F and G). This result provides strong evidence that Cacnb2 is an important downstream component of the miR‐499‐5p‐mediated regulation of dendritic development.

The auxiliary β₂ subunit of LVGCCs encoded by the Cacnb2 gene is an essential regulator of Ca_v1.2 cell surface expression (Bichet et al, 2000). To assess whether the regulation of Cacnb2 by miR‐499‐5p had an effect on the surface expression of Ca_v1.2 channels, we performed anti‐HA immunostaining on nonpermeabilized rat hippocampal neurons (“live staining”) which had been transfected with an HA‐tagged α₁ pore‐forming subunit (Cacna1c‐HA; Altier et al, 2002) together with miR‐499‐5p mimic. Since the Cacna1c‐HA‐epitope is only recognized once exposed on the cell surface, this procedure allows to measure Ca_v1.2 cell surface expression. In accordance with previous work showing a localization of Ca_v1.2 in proximal dendrites and cell bodies (Hell et al, 1993), we detected Ca_v1.2‐HA puncta in both of these compartments (Fig 3A). However, surface expression of Ca_v1.2‐HA was reduced upon overexpression of miR‐499‐5p as quantified by a significant decrease in the cell area covered by cell surface clusters of Cav1.2‐HA channels (Fig 3B), integrated density (Fig EV3A), and Ca_v1.2‐HA area (Fig EV3B). We did not observe any changes in cell area covered by total Ca_v1.2‐HA, integrated density, or total Cav1.2 area (Fig EV3C–F) when we performed the HA staining under permeabilized conditions, suggesting that miR‐499‐5p overexpression did not affect overall levels of recombinant HA‐tagged Cacna1c. Since Ca_v1.2 is the predominant LVGCC isoform in hippocampal neurons (54), we hypothesized that the observed reduction in cell surface expression of Ca_v1.2 channels translated into a corresponding loss of LVGCC currents (I_Ca,L). To test this hypothesis, whole‐cell patch‐clamp recordings were performed on hippocampal neurons overexpressing miR‐499‐5p. Calcium currents were evoked in conditions that allowed isolation of I_Ca,L as indicated by a ≈ 50% reduction in current density by nifedipine treatment (Fig EV3G). Consistent with our results from immunostaining (Fig 3A and B), hippocampal neurons overexpressing miR‐499‐5p showed significantly smaller LVGCC current density than control neurons (Fig 3C and D). Neither channel activation (Fig EV3H) nor inactivation (Fig EV3I) was significantly altered by miR‐499‐5p overexpression, suggesting that miR‐499‐5p affected LVGCC activity primarily by regulating Ca_v1.2 membrane incorporation. Furthermore, transfection of the miR‐499‐5p mimic led to a decrease in membrane capacitance, which however did not reach statistical significance (Fig EV3J). This finding correlates well with reduced dendrite arborization observed upon miR‐499‐5p overexpression (Fig 1F and G), which is supposed to lead to an overall decrease in neuronal membrane area. We next asked whether re‐expression of Cacnb2 in the context of miR‐499 mimic was able to rescue impaired LVGCC currents, similar to our results obtained for dendritogenesis. In fact, Cacnb2 expression effectively prevented miR‐499‐5p‐mediated reduction of LVGCC currents (Fig 3E and F), supporting the idea that Cacnb2 is an important functional target of miR‐499‐5p. Moreover, LVGCC currents were strongly elevated upon Cacnb2 overexpression in both control‐ and miR‐499‐5p mimic transfected neurons compared to controls, suggesting a gain‐of‐function effect. We conclude that miR‐499‐5p negatively regulates the function of Ca_v1.2 calcium channels most likely by reducing their surface expression due to an inhibition of Cacnb2 expression.

To further explore the impact of miR‐499‐5p dysregulation and Ca_v1.2 calcium channel dysfunction in vivo, we decided to overexpress miR‐499‐5p in the rat hippocampus. Therefore, a recombinant adeno‐associated virus (rAAV) expressing a miR‐499‐5p precursor RNA together with eGFP under the control of the human synapsin promoter (AAV‐miR‐499) was bilaterally injected into the rat dorsal and ventral hippocampus of adult WT or Cacna1c^+/− rats (Fig 4A; Appendix Fig S1). rAAV‐mediated miR‐499‐5p delivery led to significant overexpression of miR‐499‐5p in both genotypes (Fig 4B and C; Appendix Fig S2A–D). CACNB2 protein levels, as assessed by Western blot, were significantly reduced in the hippocampus of both WT and Cacna1c^+/− rats overexpressing miR‐499‐5p and negatively correlated with miR‐499‐5p levels (Figs 4D–F and EV4A and B). These findings confirm the negative regulation of CACNB2 by miR‐499‐5p in the rat brain in vivo.

We went on to investigate a potential role of elevated miR‐499‐5p for behavioral phenotypes associated with BD. For these experiments, we initially focused on the assessment of memory and anxiety, due to the known involvement of the hippocampus and LVGCC activity in these domains (Moosmang, 2005; Kabir et al, 2017; Dedic et al, 2018; Smedler et al, 2022). In the novel object recognition (NOR) test, which is affected by hippocampal lesions (Broadbent et al, 2010), both WT and Cacna1c^+/− rats typically spent significantly more time exploring the novel object than the familiar one as measured by the increase in the time spent sniffing (percentage of total exploration) the novel object compared to the familiar object (Fig 4G). Overexpression of miR‐499‐5p in the hippocampus of WT animals did not alter the animal's preference for the novel object as the animals still spent significantly more time exploring the novel object. However, in Cacna1c^+/− rats overexpressing miR‐499‐5p, the preference for the novel object was almost completely lost (Fig 4G). No differences were observed in the object acquisition phase of the test (Fig EV4C). To assess anxiety, we used the elevated plus maze (EPM) test, which is considered a gold standard in this domain. Neither the Cacna1c genotype nor miR‐499‐5p had an impact on the time rats spent in the open arm of the EPM (Fig 4H) or on their total entries into the open arms (Fig EV4D), indicating that anxiety‐related behavior was not affected by miR‐499‐5p overexpression. Finally, we assessed motor activity as a potential confounder of the performance of rats in the NOR. No significant genotype‐ or miR‐499‐5p overexpression effects on the total distance traveled in the open field were observed (Fig EV4E), ruling out that impaired performance in the NOR test was due to motor dysfunction. When comparing male and female data (symbols), we did not obtain any indication for sex‐dependent effects on any of the studied behaviors, although the sample size does not provide sufficient power for a conclusive statistical assessment. Taken together, miR‐499‐5p overexpression selectively impairs short‐term recognition memory in the context of reduced Ca_v1.2 expression.

If altered miR‐499‐5p function is indeed involved in human BD etiology, one might expect alterations in miR‐499‐5p expression in BD patients compared to healthy controls. To address this hypothesis, we first assessed the levels of miR‐499‐5p in PBMCs obtained from BD patients, and as a comparison from MDD and SZ patients (Figs 5A and B, and EV5A; Appendix Tables S1 and S2). Peripheral levels of miR‐499‐5p were significantly increased in BD (Fig 5A), but not MDD and SZ patients (Fig 5B) compared to healthy controls (HC), after correction for age and antidepressant treatment (Fig EV5B and C). miR‐499‐5p was upregulated in both main BD subtypes, BDI and BDII (Fig 5C), and at different phases of BD (i.e., depressive, euthymic, hypomanic; Fig 5D). miR‐499 upregulation in BD patients was observed independent of early life history, for example, childhood maltreatment (CMT; Fig EV5D), which by itself was however sufficient to trigger miR‐499‐5p expression in healthy subjects (Fig 5E, Appendix Table S3, Appendix Fig S3A–H). miR‐499‐5p levels in BD patients were not significantly correlated with the degree of manic symptomatology (as assessed by the Young Mania Rating Scale (YMRS; Fig EV5E)) and the severity of depressive symptoms (as assessed by the Beckś depression inventory (BDI; Fig EV5F)). Overall, no major differences in miR‐499‐5p expression between female and male BD patients were observed (Appendix Fig S4A–H).

In conclusion, peripheral miR‐499‐5p expression is specifically upregulated in BD patients independent of disease state or severity, supporting a role for miR‐499‐5p in human BD pathogenesis and identifying miR‐499‐5p as a novel trait biomarker candidate for BD.

Given the negative regulatory role of miR‐499‐5p in dendritogenesis and cognition in rats, we explored a potential association between miR‐499‐5p levels and GMV in HC and BD subjects (Appendix Tables S1–S3), by performing multiple regression analyses with PBMC qPCR and structural MRI data (see Materials and Methods). In HC subjects, miR‐499‐5p levels were negatively associated with a GMV cluster comprising parts of the left Wernicke language area (i.e., temporal transverse gyrus (42%), the left parietal operculum (28%), and the left superior temporal gyrus (6%); k = 1,090, x/y/z = −42/−30/15, t = 4.7, P = 0.045 FWE cluster‐level corrected; Fig 6A and B). No association was present for BD patients only and within the BD + HC analyses at the suggested statistical threshold. Our results are consistent with a role for miR‐499‐5p in the regulation of neuroplasticity in human brain areas associated with cognitive processing.

Bipolar disorder is one of the most heritable neuropsychiatric disorders, and recent GWAS identified numerous genes whose products might be involved in the regulation of cellular functions related to BD, that is, calcium homeostasis and neuroplasticity. Nevertheless, we still know relatively little about specific gene regulatory pathways which link the expression of BD risk genes to cellular function and ultimately BD pathology. Here, we show that excessively high levels of the BD‐associated miR‐499‐5p reduce expression of the auxiliary LVGCC subunit Cacnb2, which in turn leads to diminished surface expression of functional LVGCCs and reduced calcium currents. This likely results in a decrease in intracellular calcium, which in turn might impair downstream signaling events (e.g., CREB‐dependent transcription) required for important aspects of neural circuit development, for example, activity‐dependent dendritogenesis. The resulting defects in neuroplasticity might underlie cognitive and behavioral phenotypes associated with BD (Fig 6C).

Our result that the negative effects of miR‐499‐5p overexpression on dendritogenesis and Cav1.2‐mediated calcium currents were rescued by co‐expression of CACNB2 strongly implicates LVGCC activity as an important downstream component of miR‐499‐dependent regulation of neuroplasticity. Previous studies already demonstrated an important role of LVGCC signaling in dendritogenesis in the context of neuropsychiatric disease. For example, inhibition of LVGCCs with Nifedipine significantly reduced dendritic growth (Redmond et al, 2002), and activation of the Ca_v1.2 downstream effectors CaMK, MAPK, and CREB are critical for dendritic growth (Redmond & Ghosh, 2005). Interestingly, iPSC‐derived neurons from Timothy syndrome (TS) patients, which carry a gain‐of‐function mutation in the CACNA1C gene, are prone to activity‐dependent dendrite retraction (Tian et al, 2014). Intriguingly, this phenotype does not depend on calcium influx through Ca_v1.2, but rather on the recruitment of the small Rho‐family GTPAse Gem to Ca_v1.2 (Krey et al, 2013). By analogy, the reduction in CACNB2 may impair dendritic growth by destabilizing the interaction between Gem and Ca_v1.2 calcium channels. In addition, the β₂ subunit is necessary for the trafficking of fully matured channel complexes from their site of folding to the cell membrane. Consistent with these observations, overexpression of miR‐499‐5p in hippocampal neurons led to reduced Ca_v1.2 cell surface expression and current density. In HEK293 cells, Cacnb2 increased the surface levels of Cav1.2 channels by preventing channel ubiquitination by the E3 ubiquitin ligase Ret Finger Protein 2 (RFP2) and entry into the ER‐associated protein degradation system (Altier et al, 2011). Therefore, in addition to effects on forward trafficking, miR‐499‐5p overexpression might lead to increased Cacnb2 endocytosis followed by proteasome‐mediated degradation.

Although an imbalance in cellular calcium homeostasis is widely accepted in the BD field, it is still controversial in which direction calcium concentrations are changed. Intracellular calcium levels are overall increased in the plasma and lymphocytes of BD patients but decreased in olfactory neurons (Harrison et al, 2019). In human iPSC‐derived neurons from individuals carrying the CACNA1C risk allele rs1006737, higher mRNA levels of CACNA1C and higher L‐type calcium currents compared to neurons carrying the nonrisk variant are observed (Yoshimizu et al, 2015). However, the CACNA1C risk allele rs1006737 has been shown to result in both gain‐ and loss‐of‐function of CACNA1C gene expression (Bigos et al, 2010; Gershon et al, 2014), depending on the brain region examined. Importantly, the effect of Cacna1c haploinsufficiency seems to depend on the developmental stage. Embryonic deletion of Cacna1c in forebrain glutamatergic neurons resulted in increased chronic stress susceptibility while during adulthood produced stress resilience (Dedic et al, 2018). Taken together, these observations suggest that calcium signaling alterations in both directions are possibly detrimental and can lead to defects in neuroplasticity, for example, dendritic impairments.

Our observations that JSI induced the expression of miR‐499‐5p in rat hippocampus and that miR‐499‐5p is upregulated upon dexamethasone treatment in primary rat neurons is consistent with a cell‐autonomous regulation of miR‐499‐5p by stress‐associated signaling in neurons. Interestingly, CMT similarly triggers miR‐499‐5p in PBMCs, suggesting that the gene regulatory pathways leading to miR‐499‐5p upregulation could be conserved between different cell types. However, miR‐499‐5p expression is increased in PBMCs of BD patients irrespective of a previous exposure to CMT (Fig 5D), suggesting that CMT is not the main driver of miR‐499‐5p expression in BD. Clearly, additional genetic and environmental factors involved in the control of miR‐499‐5p expression need to be revealed in future experiments. In this context, it is worth mentioning that re‐sequencing of the miR‐499‐5p locus revealed several rare variants which are overrepresented in BD patients and might affect miR‐499‐5p biogenesis (Tielke A, Martins HC, Pelzl, MA, Maaser‐Hecker A, David FS, Reinbold CS, Streit F, Sirignano L, Schwarz M, Vedder H, Kammerer‐Ciernioch J, Albus M, Borrmann‐Hassenbach M, Hautzinger M, Hünten K, Degenhardt F, Fischer SB, Beins EC, Herms S, Hoffmann P, Schulze TG, Witt SH, Rietschel M, Cichon S, Nöthen MN, Schratt G, Forstner AJ, unpublished data).

Activation of the MYH7B/MIR499 gene by environmental factors, for example, stress, could involve epigenetic modifications, since Tsumagari et al (2013) showed that exons 10, 17, and 18 of the MYH7B/MIR499 gene were hypomethylated in heart tissue compared to muscle progenitor cells, which correlated with higher miR‐499‐5p levels. On the other hand, miR‐499‐5p displays particularly high expression in the cardiovascular system (van Rooij et al, 2009). Thus, chronic stress might alternatively lead to aberrant miRNA‐499‐5p expression in the heart and secretion of miR‐499‐5p into the bloodstream inside exosomes that then reach the brain. In agreement with this hypothesis, dysregulated expression of miR‐499‐5p in prefrontal cortex (PFC) exosomes was previously reported in SZ and BD patients (Banigan et al, 2013). Exosomal transfer of miR‐499‐5p between the cardiovascular system and the brain would be especially relevant in light of the high co‐morbidity between heart disease and BD (McIntyre et al, 2020). Indeed, higher levels of circulating miR‐499‐5p were reported in patients of acute myocardial infarction (Chen et al, 2015), cardiomyopathy (Matkovich et al, 2012), and atrial fibrillation (Ling et al, 2017).

A few previous studies already indicated that alterations in miR‐499‐5p expression could contribute to the development of psychiatric disorders. For example, miR‐499 is downregulated In exosomes derived from the PFC of SZ and BD patients (Banigan et al, 2013) and leukocytes from BD patients in depressive episodes (Banach et al, 2017). On the other hand, post mortem PFC of depressed subjects showed higher levels of miR‐499‐5p (Smalheiser et al, 2014). Here, we report upregulation of miR‐499‐5p in PBMCs of BD patients, but no significant effects in SZ or MDD patients, arguing for a rather specific involvement of miR‐499‐5p upregulation in BD. These inconsistencies between the different studies might be explained by the different organs (brain, blood) and miRNA sources (whole tissue, cells, exosomes) analyzed, as well as by the small sample sizes (n < 15) of the previous studies compared to ours. To further establish miR‐499‐5p as a BD trait biomarker, results should be replicated in different and even larger patient cohorts, ideally comparing miR‐499‐5p expression in different blood (PBMCs, serum, plasma, exosomes) and postmortem brain samples (PFC, ACC, amygdala, hippocampus).

Cognitive impairments, especially related to working memory and executive functioning, are among the core symptoms of BD in addition to mood disturbances (Solé et al, 2017). Our finding that overexpression of miR‐499‐5p in the hippocampus of Cacna1c^+/− rats impairs short‐term recognition memory argues in favor of an important contribution of miR‐499 dysregulation to cognitive symptoms associated with BD. In the novel object recognition test, neither the genotype nor miR‐499‐5p overexpression alone affected the preference for the novel object compared to the familiar one. This observation agrees with a previous study which reported a normal preference for the novel object in Cacna1c^+/− and WT rats of both sexes (Braun et al, 2018). Apparently, miR‐499‐5p‐mediated reduction in calcium signaling is not sufficient to elicit LTP and associated memory impairments. In contrast, in Cacna1c^+/− rats, overexpression of miR‐499‐5p might reduce the surface availability and activity of Ca_v1.2 channels to an extent where memory formation is no longer supported. Cacna1c^+/− mice also showed deficits in spatial memory (Moosmang, 2005) and extinction of contextual fear (Temme & Murphy, 2017). Further behavioral tests are therefore warranted to obtain a more comprehensive picture of cognitive abilities of rats overexpressing miR‐499‐5p.

Besides cognition, dysregulation of emotional processing (e.g., mood instability, impulsivity, mania) is one of the hallmarks of BD. In this regard, deletion of Cacna1c in mice induces the expression of anxiety‐like behaviors and has an anti‐depressant effect (Dao et al, 2010), while in rats it was shown to impair social behavior (Kisko et al, 2018). In contrast, our preliminary behavioral testing did not provide indication for a role of miR‐499‐5p in anxiety‐, depressive‐, or manic behavior (Fig 4H, unpublished observation). However, we so far manipulated miR‐499‐5p exclusively in the hippocampus, which might not be the main brain area associated with these behaviors. In the future, miR‐499‐5p manipulation will be extended to cortical regions (PFC, ACC), amygdala and hypothalamus.

Intriguingly, our results from structural MRI suggest that miR‐499‐5p‐mediated regulation of neuroplasticity might be conserved in humans. In agreement with the observed negative effect of miR‐499‐5p on dendritogenesis in rat hippocampal neurons, GMV in parts of the superior temporal gyrus (STG) was negatively correlated with miR‐499‐5p expression levels in healthy human subjects. The STG is involved in auditory processing, including language, but also has been implicated as a critical structure in social cognition. Reduced STG GMV has been observed in several neuropsychiatric disorders, including BD (Wang et al, 2019). In addition, impairments in auditory processing (Zenisek et al, 2015), verbal memory (Bora et al, 2009), and social cognition (Vlad et al, 2018) are frequently observed in BD. Together, these findings raise the possibility that chronically elevated miR‐499‐5p levels could underlie structural and cognitive impairments observed in BD patients.

In conclusion, we describe a novel mechanism for calcium signaling dysfunction in BD involving miRNA‐dependent regulation of the BD risk gene Cacnb2. Stress‐induced miR‐499‐5p represses CACNB2 translation and reduces calcium influx required for dendritic development and cognitive function. Furthermore, circulating miR‐499‐5p could be used as a biomarker for BD diagnosis or to identify healthy individuals that are at elevated risk to develop a psychiatric condition and would therefore benefit from preventative therapies. Finally, miR‐499‐5p inhibition, for example, via stable brain delivery of a chemically modified, LNA‐based miR‐499‐5p antisense oligonucleotide, could represent a new avenue for the treatment of cognitive impairments associated with BD and other neuropsychiatric disorders.

Bipolar disorder patients (n = 26 female, n = 37 male), MDD patients (n = 18 female, n = 24 female), and healthy controls (n = 26 female, n = 31 male), as well as healthy subjects with (n = 17) or without (n = 18) a history of childhood maltreatment, were obtained from the Departments of Psychiatry at the University of Marburg and the University of Münster, Germany, as part of the FOR2107 cohort (Kircher et al, 2019). The diagnosis was ascertained using the German version of the Structured Clinical Interview for DSM‐IV (SCID‐I; Wittchen et al, 1997) and subsequently translated to ICD‐10 diagnoses (WHO, 1993). Subjects were excluded if they had substance‐related disorders, severe neurological, or other medical disorders. Healthy control subjects underwent the same diagnosis procedure as the patients. Healthy subjects were included in the childhood maltreatment study if at least one subscale of the Childhood Trauma Questionnaire (CTQ) reached the threshold for maltreatment (Emotional Abuse ≥ 10, Physical Abuse ≥ 8, Sexual Abuse ≥ 8, Emotional Neglect ≥ 15, and Physical Neglect ≥ 8; Walker et al, 1999). The detailed demographic and clinical data are summarized in Appendix Tables S1 and S2. The procedures involving humans were approved by the ethics committees of the Medical Faculties of the Universities of Münster (2014‐422‐b‐S) and Marburg (AZ: 07/14). Written informed consent was obtained from all participants before examination. Experiments conformed to the principles set out in the WMA Declaration of Helsinki and the Department of Health and Human Services Belmont Report.