Diagnosing and Treating Nervous System Disorders by Targeting Novel Classes of Non-coding RNAs

Introduction

Recent scientific discoveries have revolutionized our understanding of RNA biology by establishing that non-protein-coding RNAs (ncRNAs) are ubiquitously expressed and exceedingly biologically relevant, with versatile roles impacting virtually all known normal and pathological cellular processes [1-3]. These ncRNAs are highly enriched within the nervous system, expressed in precise temporal and spatial patterns, and implicated in a broad spectrum of neurobiological functions that range from generating neural cellular diversity to promoting synaptic and neural network connectivity and plasticity [4-11]. Indeed, RNA molecules have unique, environmentally responsive biochemical and biophysical properties, including the ability to fold into secondary and tertiary structures that interface with proteins and to engage in sequence-specific interactions with DNA and other RNA molecules, that make ncRNAs particularly well suited to transact diverse structural and regulatory processes important in the nervous system. In fact, emerging observations have begun to reveal how the myriad actions of ncRNAs are critical for promoting brain development and adult homeostasis and, further, how alterations in ncRNAs are also responsible for mediating brain aging and neurological and psychiatric disease pathogenesis. Thus, the study of ncRNAs represents the next frontier for basic and translational neuroscience research and will yield insights that are likely to dramatically transform existing approaches to diagnosing and treating nervous system disorders. Here, we discuss various novel classes of ncRNAs, including both small and long ncRNAs, some of which are already driving cutting-edge innovation in the private sector.

Large-scale genomic analyses leveraging increasingly sophisticated high-throughput sequencing platforms have revealed that nearly every nucleotide (nt) in the human genome acts as a multifunctional transcriptional unit. The genome is transcribed pervasively in both sense and antisense directions, and each transcriptional unit can give rise to multiple partially or completely overlapping transcripts. These observations are striking because less than 2% of the human genome codes for protein. The vast majority of transcriptional units in the genome, therefore, give rise to RNAs that do not code for protein. The detection of these ncRNAs does not represent a totally novel discovery. For example, transfer and ribosomal RNAs (tRNAs and rRNAs) are two very well-known classes of ncRNAs. It is, however, the predicted numbers of ncRNAs and the diversity of classes and subclasses of ncRNAs that have been unexpected. Conservative estimates suggest that there are more than a million individual ncRNAs from known classes, and novel classes of ncRNAs are being identified at a very rapid pace including, astonishingly, ncRNAs which are derived from RNA, rather than conventional DNA templates [12]. Even a single protein-coding gene or ncRNA can have multiple transcription start sites, alternative promoter and enhancer elements, splicing initiation and donor sites, and independently regulated 3’ untranslated regulatory regions.

Novel Classes of Small Non-coding RNAs

The best-characterized small ncRNAs (and associated biogenesis and effector pathways) include endogenous small interfering RNAs (endo-siRNAs), PIWI-interacting RNAs (piRNAs), and microRNA (miRNAs). These ncRNAs all participate in RNA interference (RNAi), an evolutionarily conserved mechanism for post-transcriptional gene silencing mediated by the RNA-Induced Silencing Complex (RISC) ribonucleoprotein complex. However, each of these three classes of ncRNAs associates with distinct protein partners and operates via distinct pathways. Intriguingly, within the nervous system, these classes of ncRNAs may also be involved in promoting neural cellular diversity and synaptic and neural network plasticity through post-transcriptional regulation of RNA turnover, processing, and transport in concert with RNA binding proteins (RBPs), ribonucleoprotein complexes, and associated structures such as neuronal granules.

Endo-siRNAs are 20-25 nt double-stranded RNA (dsRNA) molecules derived from longer dsRNAs that are cleaved by the Dicer ribonuclease. Endo-siRNAs are implicated not only in post-transcriptional gene silencing of complementary messenger RNAs (mRNAs) through sequence-specific interactions but also in guiding chromatin remodeling events at their parent genomic loci [13]. piRNAs are 26-31 nt Dicer-dependent ssRNA transcripts derived from longer RNAs that are transcribed from transposon-containing genomic clusters. piRNAs are implicated in modulating genomic stability by silencing retrotransposons in germ cells and in certain somatic cells. In the nervous system, these ncRNAs may have additional functions that go beyond their accepted roles. For example, it has been suggested that piRNAs may regulate transposable genetic elements, such as L1, which contribute to the establishment of neuronal cellular identity and activity-dependent plasticity [14-17].

miRNAs are 21-24 nt single-stranded RNA (ssRNA) transcripts derived from longer primary transcripts (pri-miRNA) that may contain one or more mature miRNA transcripts. The processing of pri-miRNA occurs in multiple stages and may include steps catalyzed by the Drosha and Dicer ribonucleases in the nucleus and cytoplasm, respectively. A single miRNA can target hundreds of mRNAs via sequence-specific and more degenerate interactions. Conversely, the regulatory and coding regions of individual mRNAs can be targeted by multiple independent miRNAs, creating complex regulatory networks for post-transcriptional control of protein-coding gene expression and function. miRNAs are abundant in the developing and adult nervous systems and found in regional-, cellular-, and subcellular compartment-specific and activity-dependent profiles; and, they play important roles in neural development, adult function, and disease pathophysiology. In addition, analysis of miRNA sequences also reveals the existence of length and sequence heterogeneity even within individual miRNAs, termed isomiRs, including non-template additions of 3’ uridines or adenosines [18]. These isomiRs seem to be relevant in the context of nervous system health and disease. For example, a recent study showed that the vast majority of miRNAs present in the frontal cortex and the striatum of the normal brain represent isomiRs. Moreover, significant deregulation of miRNA and isomiR expression is found in the brains of patients with Huntington's disease (HD) [19].

Small nucleolar RNAs (snoRNAs) are another class of small ncRNAs that act as guides for promoting the modification of other RNAs, such as transfer RNAs and ribosomal RNAs. There are two families of snoRNA, the C/D box and the H/ACA box, which play roles in methylation and pseudouridylation, respectively. Of note, there are only two known mammalian clusters of snoRNA genes, and both of these are enriched in neurons and involved in the pathogenesis of various neural developmental disorders [20].

An increasing number of additional classes of small ncRNAs have now been discovered, however, the pathways responsible for their biogenesis and for mediating their functions remain largely unknown. Many of these classes of ncRNAs are transcribed from genomic loci flanking transcription start sites including promoter-associated small RNAs (PASRs) [21,22], termini-associated short RNAs (TASRs), antisense termini-associated short RNAs (aTASRs) [12], and transcription initiation RNAs (tiRNAs) [23]. PASRs and TASRs seem to mark the boundaries of protein-coding genes. Functionally, these classes of ncRNAs may have roles in epigenetic and transcriptional regulation. Suprisingly, aTASRs are not encoded in the genome explicitly. They are transcribed from the termini of annotated genes but in an antisense orientation, suggesting that they are generated by RNA-templated RNA synthesis. Notably, genes associated with aTASRs are enriched for those with translation-related functions. tiRNAs are ~ 18 nt transcripts associated with transcription start sites of highly expressed transcripts and sites of RNA polymerase II binding that may have roles in nucleosome positioning. Approximately 10,000 tiRNAs have been identified in the human genome.

Splice-site RNAs (spliRNAs) are 17-18 nt transcripts associated with splice sites of highly expressed and alternatively spliced genes, suggesting that these ncRNAs may be implicated in mediating this process [24]. Approximately 4,000 spliRNAs have been identified in the human genome. These spliRNAs have expression profiles that are developmental stage- and tissue-specific, suggesting putative roles in region- and cell type-specific alternative splicing in the brain.

Other recently identified small ncRNAs include those that are not associated with transcription sites but derived from other classes of small ncRNAs. For example, small RNAs derived from small nucleolar RNAs (sdRNAs) are small ncRNAs that seem to have the potential to act similar to miRNAs [25-27]. Another novel class of small ncRNAs are microRNA-offset RNAs (moRNAs), ~20 nt transcripts derived from genomic regions close to pre-miRNAs but with unknown functions [24]. Similarly, tRNAs can be processed into small ncRNA species by angiogenin (ANG) in response to stress. These tRNA-derived stress-induced RNAs induce translational repression via stress granule assembly [28,29]. Interestingly, ANG is a candidate susceptibility gene for amyotrophic lateral sclerosis (ALS) [30].

There are also classes of small ncRNAs derived from structural components of chromosomes, such as centromeres and telomeres [31,32]. Centrosome-associated RNAs (crasiRNAs) are 34–42 nt transcripts derived from centrosomes that may have roles in mediating local chromatin modifications. Telomere small RNAs (telsRNAs) are 24 nt transcripts derived from telomeric repeats that may be involved in telomere maintenance.

The expression and functions of these novel classes of ncRNAs have yet to be elucidated fully. But, it is likely that most, if not all, of these are present in neural cells and, further, that they play numerous important neurobiological roles. In fact, one recently described class of ncRNAs, enhancer RNAs (eRNAs), is derived from the enhancer domains of neuronal genes. Intriguingly, the levels of eRNA expression directly correlate with those of neuronal genes induced by stimuli such as membrane depolarization, implicating eRNAs in activity-dependent transcriptional and/or epigenetic regulation [4].

Long Non-coding RNAs

Long ncRNAs (lncRNAs) are operationally defined as ncRNA transcripts greater than 200 nt in length. They are even more heterogeneous than small ncRNAs with fewer well-defined subclasses. lncRNAs may be 5’ capped, polyadenylated, and spliced, like mRNAs [33-36]. They may have sequence characteristics indicative of high or low degrees of evolutionary conservation and functional secondary structures [37-39]. They may be cleaved to form smaller ncRNAs, such as snoRNAs and miRNAs [40]. These lncRNAs have a broad and increasing range of functions that include forming functional nuclear microdomains (e.g., splicing-associated paraspeckle domains), regulating transcription by recruiting transcription factors to gene regulatory elements, and modulating chromatin structure by recruiting chromatin-remodeling complexes to specific genomic sites. The genomic context from which a lncRNA is transcribed seems to be critical for mediating these activities. Some lncRNAs are found in intergenic regions while others are organized in antisense, bi-directional, or intronic configurations with protein-coding genes and are thought to modulate the expression of these proximally located protein-coding genes through regulatory processes in cis. In addition, lncRNAs may simultaneously be involved in regulating the expression of more distally located protein-coding genes (and potentially other small and long ncRNAs), including those located hundreds of kilobase pairs away or even on different chromosomes, through regulatory processes in trans.

Increasing evidence shows that lncRNAs are important for mediating neurobiological processes and that they are deregulated in disease states [41]. For example, the Malat1/Neat2 lncRNA promotes synapse formation and maintenance [42]. Further, a lncRNA implicated in cortical development and in higher cognitive functions, HAR1, is abnormally repressed in HD by the REST transcriptional and epigenetic regulatory factor [43]. Intriguingly, the expression and function of small and long ncRNAs are intimately linked with REST and its cofactor, CoREST [44]. In turn, these factors are directly linked to a spectrum of neurological diseases including neurodevelopmental [45-49] and neurodegenerative [50-54] disorders, epilepsy [55,56], stroke [57,58], and brain cancers [59-68].

Large intergenic ncRNAs (lincRNAs) represent a class of lncRNAs derived from intergenic regions. Intriguingly, a recent study highlighted the importance of lincRNAs in critical cellular pathways mediated by p53. It reported that the expression of a number of lincRNAs is regulated by p53 [69]. Furthermore, it demonstrated that a lincRNA, lincRNA-p21, represses p53-dependent transcriptional responses and regulates p53-mediated apoptosis. These observations underscore the functional importance of lncRNAs and support the conclusion that they can act as oncogenes and tumor-suppressor genes [70].

Diagnostic and Therapeutic Strategies

Our emerging understanding of RNA biology suggests that interrogating ncRNAs and targeting them directly (or indirectly through their biogenesis and effector pathways) are potentially important, and as yet underdeveloped, strategies for diagnosing and treating diverse pathological processes, including specifically those that affect the nervous system. Not only do neurological and psychiatric diseases represent significant unmet needs in the marketplace; but our growing appreciation for the existence of novel classes of ncRNAs, their primacy in neurobiological processes, and their seemingly ubiquitous deregulation in nervous system disorders imply that targeting these factors may be more effective mechanistic approaches for identifying and modifying disease pathology compared to existing approaches.

Because many classes of ncRNAs seem to be fairly stable and resistant to nucleases, to exhibit tissue-selective expression profiles, and to be relatively sensitive and/or specific for pathological states, they may have great clinical utility [71-76]. In fact, diagnostic assays measuring the expression of particular classes of ncRNAs (e.g., miRNAs), are already in development for cancer and for a spectrum of other disorders [71-76]. ncRNAs, including but not limited to miRNAs, may similarly be useful as biomarkers for early detection and for monitoring disease progression and responses to therapies in neurological and psychiatric diseases [77, 78]. For example, a number of miRNAs have been implicated in the molecular mechanisms underlying stroke pathogenesis including blood-brain barrier disruption, cell death cascades, and inflammatory signaling [79]. Although stroke represents a central nervous system process, ncRNA expression profiles may even provide peripheral signatures of disease as some miRNAs are expressed during ischemia in the brain and the blood (both in animal models of stroke [80] and in stroke patients [81]). Thus, miRNAs and other classes of ncRNAs may represent the long sought after serum biomarkers for rapid detection of stroke in the acute setting, which is often difficult but critical for proper management.

ncRNAs may also have roles in extracellular signaling relevant to the diagnosis of nervous system disease states [82]. For example, mRNAs and ncRNAs are known to circulate in the blood in lipid or lipoprotein complexes such as apoptotic bodies, microvesicles, or exosomes. Furthermore, distinct RNA-containing vesicles have been found in the blood of glioblastoma patients, suggesting putative roles for multiple classes of ncRNAs in intercellular signaling pathways associated with malignant transformation, dissemination and immune responses [83,84]. Further characterization of RNA intercellular transport in cancer and other nervous system disorders may lead to improved diagnostic assessment and risk stratification and customized therapeutic interventions.

Additional diagnostic strategies may include sequencing genomic DNA to identify polymorphisms in regions not associated with protein-coding genes but with novel classes of ncRNAs, such as the intergenic lincRNAs. Further, the characterization of polymorphisms in regulatory regions of neurological and psychiatric disease-associated genes may help to identify pathogenic variability in ncRNA-mRNA interactions that may result in altered transcriptional and post-transcriptional ncRNA target gene regulation. These sequencing approaches may have value for a range of applications including pharmacogenomics for personalized medicine.

Moreover, emerging classes of ncRNAs and, as they are better characterized, their associated biogenesis and effector pathways all represent novel targets for therapeutic modulation. Benefits of targeting ncRNAs may include the potential to very selectively affect an individual gene or transcript because of the sequence-specific functions of ncRNAs; to efficiently modulate large gene networks that underlie cell viability, proliferation and differentiation; and to regulate key post-transcriptional processing events (i.e., alternative splicing and transport) that, together, underlie activity-dependent neuronal gene transcription, synaptic plasticity, and neural stress responses.

Many potential therapeutic strategies can be envisioned including those focused on designing and delivering oligonucleotide molecules with the ability to disrupt and inhibit, or to replace and promote, endogenous ncRNAs and their associated functions. Notably, these can likely be launched into preclinical and clinical development in a rapid and cost-effective manner because of existing expertise in oligonucleotide, RNAi, and related technology platforms. For example, antisense oligonucleotides already in research and development pipelines have the capacity to repair and reprogram aberrant, disease-associated RNAs. The mechanism of action of these agents includes alteration of pre-mRNA processing (e.g., alternative splicing) via splice-switching, exon-skipping, and trans-splicing repair of mRNA [85,86]. Potential therapeutic strategies targeting novel classes of ncRNAs may include those that are focused on disrupting particular ncRNAs and/or inhibiting their functions. These may be oligonucleotide-based approaches analogous to those currently in development such as antisense oligonucleotides, “sponges”, and “masks”. Antisense oligonucleotides complementary to individual ncRNAs may inhibit their activity. Sponges engineered with multiple binding sequences complementary to ncRNAs may deplete cellular pools of these ncRNAs. Masks designed to compete with endogenous ncRNAs for their binding partners and targets may interfere with the actions of the ncRNAs. By contrast, other potential strategies may include those that are focused on replacing particular ncRNAs and/or promoting their functions. These approaches are not as well-established, but they are being explored for miRNAs [87]. They may be similarly justified for other classes of small ncRNAs because of their small sizes and potential to be delivered to the nervous system using modes that are currently being developed for other therapeutics. Indeed, region- or cell type-specific delivery into the nervous system is the single biggest challenge for any of these potential oligonucleotide therapies, and it has not yet been addressed effectively despite vigorous efforts. Some promising approaches include utilization of viral delivery systems [88]; various chemical modification and conjugation strategies [89]; aptamers [90]; nanotechnologies [91]; and, barrier modulation [92]. There are, of course, risks associated with oligonucleotide agents including the previously described immune activation, off-target effects, and delivery vehicle-associated toxicities.

In addition to these oligonucleotide-based approaches, alternative strategies for modulating novel classes of ncRNAs include the identification of small molecule agents that can influence ncRNA biogenesis, degradation, processing, transport, quality control, and effector pathways (e.g., RISC-based mechanisms) [93].

These oligonucleotide and small molecule therapeutic agents hold great promise for dynamically modulating gene expression and functional gene networks that are disrupted in nervous system injury and disease states. Furthermore, through delivery of these therapeutic agents to systemically accessible vascular plexi and associated stem cell niches, these approaches also have the potential to activate endogenous neural stem and progenitor cells and to promote remodeling of dysfunctional brain regions, thereby restoring higher-order cognitive, behavioral and sensorimotor functions seemingly irrevocably lost.

Biography

Irfan Qureshi, M.D., is an Assistant Professor of Neurology and member of the Institute for Brain Disorders & Neural Regeneration at Albert Einstein College of Medicine. He received his B.S. in Biomedical Engineering from Johns Hopkins and worked at Anthem Capital, a venture firm, before completing medical school and neurology training at Einstein.

Mark F. Mehler, M.D., is Alpern Professor and University Chair of the Department of Neurology; Director, Institute for Brain Disorders; and Neurologist-in-Chief, Albert Einstein College of Medicine. His research has focused on mammalian CNS development, stem cell biology, epigenetics, epigenomic and regenerative neurology and the pathogenesis of neurodegenerative diseases.