Noncoding RNAs and neurobehavioral mechanisms in psychiatric disease

Abstract

The human genome project has revolutionized our understanding of the underlying mechanisms in psychiatric disease. It is now abundantly clear that neurobehavioral phenotypes are epigenetically controlled by noncoding RNAs (ncRNAs). The microRNA (miRNA) class of ncRNAs are ubiquitously expressed throughout the brain and govern all major neuronal pathways. The attractive therapeutic potential of miRNAs is underscored by their pleiotropic capacities, putatively targeting multiple pathways within a single neuron. Many psychiatric diseases stem from a multi-factorial origin, thus conventional drug targeting of single proteins may not prove most effective. In this exciting post-genome sequencing era, many new epigenetic targets are emerging for therapeutic investigation. Here we review the reported roles of miRNAs, as well as other ncRNA classes, in the pathology of psychiatric disorders; there are both common and unique ncRNA mechanisms that influence the various diagnoses. Collectively, these potent epigenetic regulators may clarify the disrupted signaling networks in psychiatric phenotypes.

Introduction

A groundbreaking paradigm shift, the discovery of noncoding RNAs (ncRNAs), established the conventional “one gene, one protein” model to be an oversimplified view of gene regulation. Numerous ncRNA classes are now implicated in central nervous system (CNS) functions; microRNAs (miRNAs), natural antisense transcripts (NATs) and long intergenic RNAs (LincRNAs) all have reported regulatory activities in the brain ^1-4. While pleiotropic ncRNAs, including miRNAs, can target large numbers of genes and signaling pathways simultaneously ^{5, 6}, there are also ncRNAs, such as NATs ², which hybridize to a limited and precise subset of candidates. Growing evidence indicates that distinct neuronal ncRNA mechanisms, particularly miRNAs, likely influence the development of psychiatric disease. Here we review the role of miRNAs in the neurobehavioral deficits of CNS disorders and also discuss the reported contributions of other ncRNA classes.

Neuronal and synaptic miRNA biology

miRNAs have recently emerged as a global regulator of gene expression and, ultimately, effector of synaptic physiology. These miRNAs function by several mechanisms, including ribosomal RNA modifications, repression of mRNA expression by RNA interference, alternative splicing, and regulatory mechanisms mediated by RNA-RNA interactions. The processing from a primary (pri) transcript to precursor (pre) and mature miRNA in the brain requires the standard miRNA biogenesis machinery, including Drosha, Dicer, DGCR8, and argonaute (Ago) proteins ^{7, 8}. Functional knockdown of Dicer, which processes the pre-miRNA to its mature transcript, leads to reduced neuronal size and branching as well as aberrant axonal pathfinding ^9-11. Correspondingly, mice with genetic knockout of the pri-miRNA processing protein DGCR8 display a loss of synaptic connectivity and reduced number and size of dendritic spines ^{12, 13}. At the behavioral level, these mice display impaired spatial working memory-dependent tasks ¹². Interestingly, it has been shown that miRNAs are formed in part by processing of pre-miRNAs locally within dendritic spines ^{14, 15}. Furthermore, synaptic stimulation leads to local processing of pre-miRNAs in proximity to the synapse ^16-18; a cohort of miRNAs are localized within the synapse, including miR-219-5p, miR-124, miR-134, miR-138, and miR-125b ^{19, 20}. These miRNAs directly impact learning and memory behaviors, neurotransmission, neurogenesis among other functions ^{1, 21-24} and their disruption contribute to psychiatric impairments ²⁵.

A subset of miRNAs mediate neuronal specification, maturation and function. The transition from neural stem cells (NSCs) to neural progenitors and ultimately to fully differentiated neurons is highly regulated by a complex interaction of miRNAs and other factors ²⁴. In general, let-7 ^{26, 27}, miR-124 ^28-30 and miR-9 ³¹ are thought to reduce NSC proliferation and promote neuronal differentiation. It is typically held that miR-134 ³² and miR-25 ³³ induce the proliferation and/or inhibit the differentiation of NSCs and neural progenitors. In parallel, miR-137 both decreases ^{34, 35} and increases ^{36, 37} NSC proliferation, either enhancing or opposing neuronal maturation. There is an intricate overlap and feedback occurring between these key miRNAs, mediated in part by their target genes. Notably, miRNAs may mediate neurogenesis throughout development from embryo to adult. Adult neurogenesis is reportedly decreased in neurodegenerative disease ^{38, 39} and depression ⁴⁰, and is modulated by antidepressant therapeutics ⁴¹. Thus, the following discussions regarding miRNA pathways in psychiatric disorders (outlined in Table 1) may involve neurons derived at all stages of brain maturity.

Table 1. miRNAs implicated in psychiatric phenotypes.

miRNA	Biological finding	References	Diagnosis
miR-137	schizophrenia risk allele	44-47	schizophrenia
	alters neuronal connectivity, and size of brain tissues	48-49
	direct and indirect targets of miR-137 are linked to schizophrenia	50-53
miRs- 137, 548	associated with schizophrenia (along with miR-137) in largest study to date on common genetic variants	46
miRs- 181b, 26b	the premature and mature transcripts were upregulated in human prefrontal cortex	54
miR-181b, miR-30e	associated with schizophrenia through plasma analysis of human patients	55, 56
miR-219-5p	alters schizophrenia behaviors in mice through NMDA-R signaling; targets CaMKIIgamma	57
	most downregulated miR in cortical synaptosomes from schizophrenia patients; upregulated in tissue (total) from cortex	20
	upregulated in cortical tissue (total) from schizophrenia patients	54
miR-132	repressed in mice deficient for the NMDA-R coagonist D-serine	64
miR-134	altered expression in plasma of bipolar patients	79	bipolar disorder
miRs- let-7b, let-7c, 128a, 24a, 30c, 34a, 221, 144	dysregulated in hippocampus of rats by the mood stabilizers valproate and lithium	78
miRs- 219-5p, 132, 279, 142-3p, 185, 138, let-7b, 125a, 206, 182	linked to circadian rhythm behaviors	65-66; 68-76	circadian function, potential links to multiple psychiatric diagnoses
miR-124a	regulates depression-associated behaviors in rats through BDNF	83	depression
miR-16	mediates depression behaviors through regulation of SERT; mechanistic connection to depression through neurogenesis pathways	88-89
miRs- 96, 141, 182, 183, 183*, 198, 200a, 200a*, 200b, 200b*, 200c, 429	different expression profiles of these miRs between rats with learned helpless (LH) versus non-learned helpless behavior (NLH); possible link to depression phenotypes	114
miRs- 142-5p, 142-3p, 494, 376a*, 496, 369-3p, 23b, 27b, 24-1*, 34b*, 34c, 17*, 20a, 424, 20b, 142, 301a, 324-5p, 497	downregulated in MDD patients, possibly due to alterered transcription from the pri-miRNA source	85
miRs- let-7b, mir-132, 181b, 338-3p, 486, 650	networked together in MDD patients	85
rs76481776 polymorphism in the pre- miR-182	associated with Clock genes	74
miR-135	genetic manipulation of miR-135 leads to precipitation of anxiety and depression and antidepressant response in mice	91	depression/anxiety
miRs- 19b, 223	disrupted in human and animal models of PTSD	97, 101	PTSD
miR-34c	overexpression in mice prevents anxiety-associated responses in stressed mice; may control these behaviors through CRFR1 gene	102	stress/anxiety
miR-34a	controls fear-responses in mice through Notch signaling	103
miRs- 608, 330-3p,	genetic variations identified for these miR binding sites in target mRNAs with association to anxiety	104, 107
miR-124	link to anxiety through glucocorticoid signaling	105
	link to stress through corticosteroid signaling	110
miR-16	linked to anxiety in medical students through peripheral blood profiling	109
	binds to SERT mRNA	108	implicated via SERT-mediated pathways
miRs- 22, 138-2, 148a, 488	genetic variations identified in miR binidng sites of gene targets associated with panic disorders	112	panic disorder

miRNA modulation of behavioral phenotypes in psychiatric diseases

Schizophrenia and bipolar disorder

Substantial evidence indicates that miRNAs mediate schizophrenia phenotypes through a variety of mechanistic pathways. The copy number variant (CNV) at 22q11.2, resulting in DiGeorge syndrome, is the most common rare variant identified in schizophrenia, occurring in approximately one percent of patients ⁴². Many children with 22q11.2 deletion syndrome have developmental delays and learning disabilities, with a substantially elevated risk of developing schizophrenia (30%) ⁴³. Though the 22q11.2 region contains 30 to 40 genes, many have not been well characterized; one notable gene is DiGeorge syndrome critical region gene 8 (DGCR8), an essential contributor to microRNA biogenesis ⁸. Microdeletion of 22q11.2 in mice results in the downregulation of a cluster of miRNAs with corresponding changes in cognitive and behavioral functions¹².

Of the common genetic variants enriched in patients with schizophrenia, there is a well-established signal at miR-137 and several of its target genes ^44-46. Schizophrenia patients with a miR-137 risk allele exhibit greater symptom severity ⁴⁷, significantly altered functional connectivity ⁴⁸, reduced white matter integrity, smaller hippocampi and larger lateral ventricles ⁴⁹. Functionally validated targets of miR-137 include such schizophrenia risk genes as CACNA1C ⁵⁰, TCF4 ⁵⁰ and ZNF804A ⁵¹, while additional putative targets include ERBB4, GABRA1, GRIN2A, GRM5, GSK3B, NRG2, and HTR2C ⁵². Overexpression of miR-137 in a human NSC line identified direct and indirect miR-137 targets in neural cells. While direct miR-137 targets were enriched for transcription factors and cell cycle genes, indirect targets included pathways enriched in schizophrenia genome wide association studies, particularly major histocompatibility complex, synapses, FMRP interacting RNAs and calcium channels ⁵³. More recently, the largest study of common variance identified 108 regions significantly associated with schizophrenia, including significant hits at miR-137 and miR-548 ⁴⁶.

Post-mortem profiling of miRNA expression in the prefrontal cortex of schizophrenia patients identified a global increase in miRNA expression compared to control populations ⁵⁴. More specifically, although the mature and pre-miRNA species were increased (particularly of miR-181b and miR-26b), there was no significant difference in transcription of the source pri-miRNA. These results suggest that the changes were due to increased miRNA biogenesis rather than altered miRNA transcription. The authors further observed upregulated expression of Drosha and DGCR8, both of which are involved in pri-miRNA processing, and speculate this is the ultimate cause of aberrant miRNA levels. Additionally, human analyses have started to reveal a profile signature of miRNA dysregulation in peripheral tissues such as plasma and serum of schizophrenia patients. Intriguingly, plasma levels of miR-181b (discussed above) were found to predict response to antipsychotic treatment ⁵⁵; correspondingly, miR-30e has also been postulated as a plasma biomarker of schizophrenia ⁵⁶.

NMDA-R signaling, one of the most consistently implicated pathways in schizophrenia, is strongly controlled by miR-219 ⁵⁷. Notably, miR-219 is the most enriched miRNA in the human synapse, the most downregulated in cortical synaptosomes of schizophrenia patients ²⁰, and was dysregulated in cortical tissue from two patient cohorts ^{20, 54}. Moreover, neuronal inhibition of miR-219-5p in mice significantly altered the precipitation of schizophrenia behavior by NMDA-R antagonists ⁵⁷. Taken together, miR-219-5p actively mediates synaptic functions and the development of psychiatric phenotypes. NMDA-R also regulates miR-132 ^{21, 58-63}, which is implicated in learning and memory functions ^{21, 59}, long-term potentiation ^58-61 and neurotransmission ⁶¹. miR-132 expression is repressed in mice with genetic disruption of the enzyme which produces the NMDA-R co-agonist D-serine ⁶⁴, and these D-serine deficient mice recapitulate some of the neurobehavioral and cognitive impairments presented in patients with schizophrenia. Expression of miR-132 was dysregulated in human cortical tissue from two schizophrenia datasets but those expression changes were in opposite directions ^{20, 21}; it remains possible that miR-132 is disrupted in specific neuronal cell types or distinct subcellular regions of neurons in schizophrenia. Interestingly, the NMDA-R regulated transcripts miR-132 and miR-219-5p (discussed above) ^{65, 66} regulate circadian rhythm, which is frequently disrupted in schizophrenia and other psychiatric disorders ⁶⁷. In addition, a subset of other miRNAs also reportedly regulate circadian functions, including miRs – 279, 142-3p, 185, 138, let-7b, 125a, 206 and 182 ^68-76. Indeed, miRNA regulators of NMDA-R signaling and circadian rhythm could yield new therapeutic targets for treatment of neurobehavioral deficits.

miRNA expression is responsive to current therapeutics administered for bipolar disorder, which is known to share overlapping genetic links with schizophrenia ⁷⁷. In rats treated with either lithium or valproate, there are a cohort of miRNAs altered in the hippocampus, including let-7b, let-7c, miR-128a, miR-24a, miR-30c, miR-34a, miR-221 and miR-144 ⁷⁸. Additionally, miR-134 is altered in the plasma of treated bipolar disorder patients ⁷⁹. Valproate and lithium significantly modulate Brain-Derived Neurotrophic Factor (BDNF) levels, a critical regulator of neuronal homeostasis ^{80, 81}, which is itself regulated by both short and long ncRNAs ^82-84, such as miR-124a. Notably, miR-124a is linked to depression-related behaviors ⁸³, as discussed in the section below.

As more post-mortem datasets become publicly available, we believe that groups of consistently altered transcripts in schizophrenia and bipolar disorder will emerge and potentially converge. Combining multimodal SNP and exome sequencing genotype information with RNA and miRNA expression datasets will facilitate miRNA-mRNA correlations. Furthermore, in vivo studies will allow for comprehensive investigation of identified candidates. Arguably, some of the more persistent miRNA associations in schizophrenia to date implicate miR-137, miR-181b, and miR-219-5p (Table 1).

Depression

There is accumulating evidence for significant contribution of miRNA mechanisms in mood disorders such as depression. In the prefrontal cortex of depressed subjects who had died by suicide ⁸⁵, 21 miRNAs were significantly down-regulated in the major depressive disorder (MDD) group. More miRNAs were down-regulated than upregulated, implying a global down-regulation of miRNA levels in MDD. Furthermore, almost half of the down-regulated miRNAs were transcribed by the same pri-miRNA gene transcripts (mir-142-5p and 142-3p; mir-494, 376a*, 496, and 369-3p; mir-23b, 27b and 24-1*; mir-34b* and 34c; mir-17* and 20a) or found within the same chromosomal region (mir-424 and 20b at Xq26.2-3, 377 kb apart; mir-142 and 301a at 17q22, 820 kb apart; mir-324-5p and 497 at 17p13.1, 205 kb apart), suggesting that the down-regulated miRNA expression may be due to decreased transcription. In addition, a set of 29 miRNAs formed an inter-connected network in the MDD group: let-7b, mir-132, 181b, 338-3p, 486-5p, and 650 were “hubs”. A recent study of genotyping polymorphisms from three miRNA processing genes (DGCR8, AGO1, and GEMIN4) found that DGCR8 rs3757 was associated with increased risk of suicidal tendency and improvement response to antidepressant treatment, whereas AGO1 rs636832 showed decreased risk of suicidal tendency, suicidal behavior, and recurrence ⁸⁶. Thus, polymorphisms in miRNA processing genes may influence depression risk and treatment.

Molecular targets of anti-depressants frequently engage the transporters of serotonin, a monoamine neurotransmitter ⁸⁷. miRNAs regulate the serotonin transporter (SERT) and its response to serotonin reuptake inhibitor (SSRI) therapeutics. Specifically, SSRI antidepressant fluoxetine (Prozac) treatment in mice induces expression of miR-16 while repressing the miR-16 target SERT ⁸⁸. Notably, miR-16 mediates the depression-related behavior through precise control of neurogenesis ⁸⁹. Another SERT associated miRNA, miR-135, was found to control the onset of co-existing depression and anxiety symptoms in mice ^{90, 91} as well as the response to anti-depressant treatment ^{90, 91}.

Because of limited studies in depressed patients, it is difficult to pinpoint specific miRNAs consistently implicated in the pathogenesis of depression; nevertheless, miRNAs that influence BDNF, including miR-132 and miR-34, and neuroinflammation, such as the let-7 family⁹², may be highly relevant (Table 1). Intriguingly, miR-124a repression of BDNF provokes depression-related behaviors ⁸³ and significantly mediates neurogenesis. Future studies could investigate if current anti-depressant therapeutics signal through miR-124a or other disease-associated miRNAs to enhance adult hippocampal neurogenesis ^{40, 41, 93, 94}.

Stress, anxiety, and fear disorders

The link between ncRNAs and stress, anxiety, or fear related responses opens new avenues for therapeutic intervention⁹⁵. While a range of genetic associations have been loosely implicated in Post-Traumatic Stress Disorder (PTSD) ⁹⁶, epigenetic factors likely play a defining role in disease progression. A panel of disrupted miRNAs in the blood of military veterans with active PTSD symptoms have been identified ⁹⁷; these PTSD-associated miRNAs were significantly associated with immunological pathways, suggestive of their pathogenic mechanisms in the disorder. Indeed, abnormal systemic immune responses are routinely reported in stress-related conditions ^98-100, consistent with its regulation by the PTSD-linked miRNAs. Intriguingly, a few of the miRNA transcripts altered in the veterans suffering from PTSD, including miR-19b and miR-223, were also perturbed in the serum and amygdala of a PTSD animal model ¹⁰¹.

Stress and anxiety, however, are not unique behaviors to PTSD and are provoked through many distinct triggers. Overexpression of miR-34c in the central amygdala elicited an anxiolytic-like effect in mice stressed through acute restraint ¹⁰²; furthermore, miR-34c directly targets a key mediator of stress responses, the corticotropin releasing factor receptor type 1 (CRFR1) gene. Notably, miR-34a was recently reported to regulate fear related responses through Notch signaling in mice ¹⁰³. Although miR-34a didn’t alter anxiety-related parameters in the same manner as miR-34c, it is possible that the miR-34 family is central in behavioral manifestations.

In addition to the above mentioned miRNAs, miRs - 608, 124a, 132, 330-3p, and 16 are also implicated in anxiety phenotypes through genetic associations or in vivo analyses ^104-109. Specifically, miR-124 is linked to both anxiety and stress related behaviors through glucocorticoid and corticosteroid signaling, respectively ^{105, 110}. Moreover, ablation of the miRNA biogenesis enzyme Dicer in the central amygdala of mice provoked the onset of anxiety-like symptoms ¹⁰². Mice with Dicer knockout specifically in dopaminoceptive neurons also exhibit behavioral changes, including ataxia as well as front and hind limb clasping ¹¹¹. These Dicer-related phenotypes may also be independent of neurodegenerative mechanisms ¹¹¹.

Human genetic linkage studies and animal models also revealed miRNA pathways in panic or fear responses. At least four miRNAs have single- nucleotide polymorphisms (SNPs) in miRNA sequences located within panic disorder associated genes, including miR-22, miR-138-2, miR-148a, and miR-488 ¹¹². Furthermore, miR-128 expression is increased with the formation of fear-extinction memory in mice, which is the re-conditioning of the memory to overcome established fear behaviors ¹¹³.

An individual’s ability to cope with stress is critical in the development of MDD. There is contrasting miRNA expression between rats who developed learned helpless (LH), a behavior that resembles stress-induced depression, compared to rats who did not develop depression (non-learned helpless [NLH]) in spite of receiving similar inescapable shocks ¹¹⁴. One set of miRNAs showed large, significant, and consistent down-regulation in the frontal cortex of NLH rats compared to a blunted response in LH rats (miR-96, miR-141, miR-182, miR-183, miR-183*, miR-198, miR-200a, miR-200a*, miR-200b, miR-200b*, miR-200c, and miR-429). These synaptically enriched miRNAs¹⁶ are encoded at a few shared polycistronic loci, suggesting coordinated control of their transcription, and they share 5′-seed motifs which indicate similar or overlapping sets of target mRNAs. Interestingly, half of this set are predicted to hit Creb1 as a target, and binding sites for CREB lie upstream of miR-96, miR-182, miR-183, miR-200a, miR-200b, miR-200c, miR-220a*, and miR-200b*. This suggests that a feedback loop arrangement may also exist between Creb and Creb-stimulated miRNAs and target genes ¹¹⁵. Because these miRNAs are down regulated in NLH rats, but not LH rats, this can be interpreted as a homeostatic response intended to minimize repressive effects on Creb1.

Although more studies are needed to identify a list of miRNAs associated with stress, anxiety and fear, some putative candidates are emerging (Table 1). As discussed previously, miR19b and miR-223 were dysregulated in human and animal models of PTSD ^{97, 101} and should be evaluated further for therapeutic targeting in trauma-induced stress. Notably, these two transcripts share over 80 bioinformatic mRNA targets through Targetscan prediction. Additionally, the miR-34 family may exhibit an overlapping role in anxiety and fear responses ^{102, 103}. Overall, replication studies in human and animals will further define persistent miRNA mechanisms in these neurobehavioral deficits.

Long ncRNAs and the interaction between ncRNA transcripts in psychiatric disease

Although miRNAs are arguably the most extensively characterized class of ncRNA in neurons, long ncRNAs (lncRNA) are increasingly implicated in CNS functions ^{4, 116-118}. Long ncRNAs acts as miRNA sponges and can bind proteins and RNAs that regulate transcriptional changes and epigenetic modifications of chromatin. Indeed, these transcripts regulate basic neuronal biology as well as genes with strong disease-association ^{4, 116-122}. BC1, one of the first lncRNA transcripts characterized in the brain, is now known to modulate metabotropic receptor signaling ^{123, 124}. Select members of the NAT class of ncRNAs in the brain are functional and modulate expression of their sense partner ^{2, 82}; NATs are endogenously transcribed and exhibit at least partial sequence complimentarity to protein-coding genes, ranging from short to long in nucleotide length ². An emerging area of investigation indicates that long and short neuronal ncRNAs are co-regulatory ¹²⁵, which is discussed more throughout the following sections. For example, the BACE1 gene linked to psychiatric and cognitive deficits in Alzheimer’s Disease (AD)] is reportedly modulated by a network of competitive ncRNA interactions ¹²⁶. More specifically, it was shown that miR-485-5p binding sites in BACE1 are masked through a long antisense transcript, BACE1-AS ¹²⁶.

Evidence demonstrates that lncRNAs participate in neural plasticity, supported by the expression of a large number of these transcripts in dendrites ¹²⁷. In addition, a genome-wide analysis show that lncRNAs are modulated by neuronal activity in human brain ¹²⁸. LncRNAs also mediate neurogenesis, neurodevelopment and related pathways ^{129, 130}. Reports suggest that fine-tuned regulatory control of the embryonic brain by these transcripts is required for development of mature CNS functions ¹³⁰. Mouse knockout studies involving long-intergenic ncRNAs (lincRNAs) found that loss of function for linc–Brn1b results in a reduction of cerebral cortex progenitor cells and abnormal cortical lamination amongst other pathologies ³. Furthermore, a subset of lincRNAs bind miRNAs and inhibit their functions (i.e. miRNA sponge), which can reportedly disrupt brain development ¹²⁵. Circular ncRNAs (circRNA) were also recently reported to function as miRNA sponges in neurons. For example, a circRNA was found to have multiple miRNA binding sites, including for miR-7, and both were co-expressed in neocortical and hippocampal neurons ¹³¹. These studies further indicate the importance of interactions between ncRNAs in the brain and the need to uncover their epigenetic networks.

Neurotrophins, which play a key role in maintaining homeostatic activity in the CNS, reportedly respond to lncRNA signaling. For example, the neurotrophin BDNF has sequence complementarity with a conserved NAT, termed BDNF-AS. Functionally, BDNF-AS was reported to modulate neuronal growth in vitro and in vivo ⁸². The sequence complementarity of this noncoding transcript appears to extend into the 3′ untranslated region (3′UTR) of BDNF, which contains regulatory miRNA binding sites such as miR-124a. miR-124, as discussed in previous sections, modulates depression and related neurobehavioral deficits; it is possible that the BDNF-AS partly functions by preventing miR-124 from binding to BDNF.

Susceptibility genes that precipitate psychiatric symptoms are also subject to lncRNA modulation. For example, an endogenous noncoding antisense transcript to the HTT gene (HTTAS) functionally regulates HTT expression¹¹⁹, which is the primary genetic aberration in HD and associated cognitive and neurobehavioral pathology. Susceptibility genes for schizophrenia are also epigenetically controlled by lncRNAs. Genetic variations in the Disrupted in Schizophrenia 1 (DISC1) gene are consistently linked with schizophrenia-associated behaviors ^{132, 133}. A recent report suggests the lncRNA Gomafu mediates DISC1 splicing events, resulting in splice variants linked to schizophrenia ¹³⁴. Gomafu, is implicated in neural development ¹²⁹ and is downregulated in cortical tissue of schizophrenia patients and in activated human neuronal cells ¹³⁴. Additionally, DISC1 is regulated by its lncRNA antisense transcript DISC2 ^{120-122, 132}. Several groups have reported genetic association of DISC2 with schizophrenia as well as other psychiatric disorders ^{120-122, 132}; however, DISC2 is not conserved among species ¹³², hampering investigation of its regulatory mechanisms. One intriguing possibility is the existence of regulatory loop between DISC1 and the two ncRNA transcripts DISC2 and Gomafu.

Recent peripheral blood profiling studies also identified significant disruption of lncRNAs in depression, specifically patients with MDD, 17 of which were documented as depression-related gene in previous studies ¹³⁵. In parallel, this study uncovered potential miRNA-mRNA networks that are consistent with genes previously associated with depression. Future studies could investigate co-regulatory mechanisms between the dysregulated miRNAs and lncRNAs in the disease. Additionally, the lncRNA antisense transcript coded by LOC285758 has been implicated in violent suicide completers ¹³⁶, likely triggered through depression or other psychiatric disorders.

Similar to miRNAs, it is possible some of the lncRNAs listed above may mediate the development of the distinct psychiatric symptoms through control of circadian genes. Global transcriptome profiling studies indicate that lncRNAs, including NATs, epigenetically regulate circadian biology ^{137, 138}. Indeed, an antisense RNA to a circadian gene in Neurospora, termed qrf, is regulated by light and represses the expression of its sense partner frq, through chromatin modifications ¹³⁹.

Summary

Regulating neural plasticity, neurogenesis, and numerous behavioral phenotypes, it is clear that ncRNAs contribute to the pathogenesis of many psychiatric disorders. Moreover, direct evidence comes from human postmortem brain and animal studies indicating perturbed ncRNA levels in disparate disease such as Huntington’s, schizophrenia, depression, PTSD among others.

Despite these findings, one needs to find an integrated view of these ncRNA network(s). It is known that differential co-expression of distinct miRNA groups can directly mediate human disease pathogenesis as well as serve as a biomarker profile for disease diagnosis^140-143. The disrupted miRNAs, and their corresponding mRNA target genes in each psychiatric disease, are likely to interact with and regulate each other, both directly as targets and indirectly as part of larger regulatory networks. Furthermore, correlated miRNAs and mRNAs may be coordinately regulated by a (possibly overlapping) set of transcription factors or other epigenetic influences. It remains to be determined whether the changes in the miRNA/mRNA network are i) similar or different across distinct brain regions ii) cell type-specific and iii) reversible mechanisms.

The underlying reasons for altered ncRNA expression remain unresolved and could result from a number of factors, including genetic changes in the promoter region or other locations within the gene. Additionally, defects in RNA editing or epigenetic suppression of the chromosomal region encoding the ncRNAs can also occur. Finally, miRNAs, lncRNAs, and their processing genes are susceptible to regulation by well-established signaling modulators such as BDNF, CREB, calcium, or calcium responsive neurotransmitters. For example, recently it was reported that Dicer is activated by proteolytic cleavage under conditions of elevated calcium levels ^{14, 16}. The networking of miRNAs and other ncRNAs into critical pathways such as neurotransmission, neurogenesis, and neurodevelopment may open up an entirely new understanding of psychiatric disease. Ultimately, we expect that for many of the complex psychiatric disorders we have considered, novel links between ncRNA mechanisms and the disease pathology will continue to emerge.