Depression and anxiety disorders are common psychiatric conditions with high comorbidity and massive medical, social and financial burdens. The cellular and molecular substrate of these disorders, and the search for alleviating treatments, have been a matter of debate since decades. The “monoamine hypothesis”, which identifies a deficiency or imbalances in monoamine neurotransmitters, in particular serotonin, as the cause of depression constitutes the most accepted explanation for the origin of these diseases. The monoamine hypothesis is supported by the properties of many antidepressants that target the monoamine systems. In particular, “selective serotonin reuptake inhibitors” (SSRIs), the most prescribed class of antidepressants, block the membranous serotonin transporter (SERT, SLC6A4, 5-HTT), resulting in increased synaptic serotonin levels and in adaptive responses from the serotonergic system1. However, the aetiology of depression is not fully established, and considering the heterogeneity of symptoms, underlying genetics, and treatment responses, it is generally assumed that the cause of MDD is likely multifactorial. A major limitation of the monoamine hypothesis is that it does not fully explain the kinetics of SSRIs action. In particular, the behavioural benefits of Fluoxetine (Flx), one of the commonly prescribed SSRI, can be typically observed only after 3–4 weeks of chronic treatment, suggesting the involvement of a slow adaptive response, such as increased hippocampal adult neurogenesis2, 3.
The leading monoamine hypothesis of depression coexists with other theories4. Among them, the “GABAergic deficit hypothesis” for depression implicates GABAergic system defects in common phenotypes of MDD, and suggests that antidepressant therapies act on GABAergic neurons5, 6.
This theory is supported by results from clinical studies that showed that depression implicates functional defects in cortical GABAergic inhibition7 and can be associated with reduced GABA brain levels8, 9, reduced expression of glutamic acid decarboxylase, the limiting enzymes for GABA synthesis10, 11 or reduced density or function of GABAergic interneurons subclasses11, 12.
Beside its action on the serotonin transporter, Fluoxetine treatment induces several other changes in the brain13. For instance, Flx administration indeed affects the major class of GABAergic cortical neurons : Parvalbumin (PV)-positive interneurons, which are critical regulators of brain functions, including information processing, fear and stress related behaviours14–16.
Chronic Flx administration induces a strong reduction of Parvalbumin expression and a shrinkage of their extracellular matrix, the perineuronal net (PNN) a hallmark of these neurons’ maturity17, 18. By these actions, fluoxetine treatment induces an adolescent-like period of brain plasticity that allows a strong response to environmental cues that might participate to the antidepressant effect of Flx by allowing a partial rewiring of the brain17. Interestingly, this “dematuration”, or “iPlasticity” requires the direct binding of Flx to a dimerized TrkB receptor (neurotrophic receptor tyrosine kinase 2), independently from the serotonin transporter SERT19, 20. Interestingly, the activation of TrkB signalling selectively in PV neurons is sufficient to promote iPlasticity, by inducing cellular and neuronal network alterations reminiscent of fluoxetine administration, facilitating the influence of environmental factors on neuronal networks21.
Recent findings have shown that both typical (e.g. Flx) and fast-acting antidepressants such as ketamine directly bind to TrkB, suggesting that TrkB is a direct, acute, common target for several classes of antidepressants. It appears, therefore that direct activation of TrkB is a critical step to permit for the rapid biological action of antidepressants22. Other studies have shown that antidepressant also promotes the synthesis and release of the TrkB endogenous ligand BDNF (brain-derived neurotrophic factor)23–25. Together, these experiments suggest an important role of TrkB signalling modulation for the cellular and behavioural effects of antidepressants24, 26. This role is independent from the serotonin transporter SERT19 and, therefore, from variations in serotonin brain concentrations.
Among the downstream effectors of TrkB, the CREB pathway (cAMP Response Element-binding protein) appears to be involved in antidepression27, 28. CREB expression and transcription functions has been shown to be increased after fluoxetine treatment, independently from SERT19, 23. Another intracellular effector of TrkB, the mTOR pathway (mechanistic Target Of Rapamycin), has also been suggested as a possible mediator of rapid anti-depressive effects of ketamine and of long-term Flx treatment, with contrasting results29, 30.
Both adaptations to altered neurotransmission and remodelling of neuronal networks suggested as substrate for the long-term effects of antidepressants involve the progressive adaptation of genetic regulation pathways. Multi-omics analysis of several brain regions of mice treated with Flx for six weeks have revealed a complex, region-specific genetic signature, which includes increased energy metabolism via oxidative phosphorylation and mitochondrial changes, chromatin remodelling and transcription factors regulations31. Large epidemiological studies demonstrate that MDD has a strong familial component, associated to heritable genetic influences that are modulated by patient-specific environmental exposures32–34. Similarly, the efficiency of SSRI treatment on depression is highly variable depending on the patient, suggesting an important role of the genetic background in both the susceptibility to depression, and the response to the treatments35.
In order to better understand the genetic mechanisms implicated in depression and its treatment, the effects of targeted genetic mutations on the responses to antidepressants were studied in the mouse. Some of these mutant mice, even in the absence of pharmacological treatment, showed an “anti-depressant-like” phenotype, with behavioural and histological signs reminiscent of those induced by the chronic administration of antidepressants, highlighting the role the targeted genes for the mechanism of action of the drugs19, 36, 37.
Dlx genes encode for a family of homeodomain transcription factors involved in many developmental processes38, 39. In mammals, six Dlx genes are arranged in three bigenic clusters40. In particular, Dlx5 and Dlx6 are expressed by developing and mature telencephalic and by a pool of diencephalic GABAergic neurons, but not in other brain regions38, 41, 42. Dlx5 and Dlx6 are co-regulated and present redundant functions43.
In the cerebral cortex, Dlx5/6 are particularly important for the differentiation of PV-positive GABAergic interneurons44. Heterozygous Dlx5/6 inactivation in the mouse results in abnormal prefrontal cortex gamma (γ; ∼30–120 Hz) oscillations, which depend on PV-interneurons activity and in working memory deficits45. Targeted inactivation of Dlx5/6 in mouse GABAergic neurons alters behaviour, vocal socialization and metabolism with a reduction of anxiety-like and obsessive-compulsive-like behaviours41, 46. Recently, we have shown that the density of PV-positive neurons in the adult prefrontal cortex and in the hippocampus is directly correlated to Dlx5/6 allelic dosage47. In particular, mice with reduced or absent Dlx5/6 expression in GABAergic neurons present a reduced number of mature PV neurons in the frontal cortex and show reduced anxiety, a peculiar response in the “marble burying test” with some mice not burying any marble which can be interpreted as a strong reduction in anxiety-like behaviours. Remarkably, the phenotypes obtained after Dlx5/6 GABAergic inactivation are reminiscent of phenotypes of mice treated with fluoxetine or other SSRIs18, 48. On the opposite, experimental overexpression of Dlx5 in GABAergic neurons results in increased PV-neurons density in the frontal cortex associated with increased anxiety47.
DLX5 is imprinted in the human brain49, and partially imprinted in the mouse with preferential transcription of the maternal allele. The genomic methylation regulator, MECP2, binds to the genomic region including DLX5/6; it is deregulated in Rett-syndrome, an X-linked neurodevelopmental disorder which provokes a DLX5/6 overexpression50.
DLX5/6 genomic region also contains a long non-coding RNAs (LncRNA), DLX6-AS1 and DLX6-AS2 (Evf2 and Evf1) which form ribonucleoproteic complex to regulate gene expression, including DXL5/651, 52. DLX6-ASs also bind to other non-coding RNAs to regulate their actions, for instance DLX6-AS1 indirectly regulates BDNF pathway through its interaction with miR-107 in neuroblastomas53.
A large-scale transcriptomic analysis of post-mortem brains has shown that the DLX5/6 locus participates to genetic modules altered in GABAergic neuronal function in Autism Spectrum Disorders (ASDs) and schizophrenia54. Namely, DLX6-AS1 has been identified as the most central hub gene in the interneuron module downregulated in schizophrenia and ASD. Patients carrying mutations in the DLX5/6 brain-specific enhancers that includes the I56i, I56ii and MEF255, 56, present a higher incidence of ASD and speech delay11, 57–59.
Mutations involving lifelong brain Dlx5/6 altered expression are, therefore, associated with cognitive modifications both in humans and mice.
In this study, we show that Dlx5/6 expression in the adult brain is can be modulated by Flx administration: Flx treatment induces a rapid and persistent reduction in Dlx5 and Dlx6 expression in the brain, and provokes phenotypic alterations reminiscent of Dlx5/6 invalidation. Furthermore, we show that the rapid effects of Flx administration on Dlx5 expression is mediated through the TrkB-CREB signalling cascade providing a new genetic target to decipher the action of antidepressants.