miRNA degradation in the mammalian brain

Abstract

MicroRNAs (miRNAs) are short, noncoding RNAs that are evolutionarily conserved across many different species. miRNA regulation of gene expression, specifically in the context of the mammalian brain, has been well characterized; however, the regulation of miRNA degradation is still a focus of ongoing research. This review focuses on recent findings concerning the cellular mechanisms that govern miRNA degradation, with an emphasis on target-mediated miRNA degradation and how this phenomenon is uniquely poised to maintain homeostasis in neuronal systems.

INTRODUCTION

MicroRNAs (miRNAs) are small, noncoding RNAs that modulate the translation of more than 60% of all protein-coding mRNAs in the cell, providing an intermediate regulatory step between gene transcription and translation (14). miRNAs are transcribed from the genome into a long primary structure known as primary miRNAs (pri-miRNAs). These pri-miRNAs are subsequently processed in the nucleus by the microprocessor complex consisting of Drosha, a ribonuclease, and DGCR8, an RNA-binding protein. This results in the formation of precursor miRNAs (pre-miRNAs) which are ~70 nucleotides in length. pre-miRNAs are exported to the cytoplasm and processed further by another ribonuclease (Dicer) to form the mature miRNA duplex. From this duplex, a single mature miRNA strand is selected to function within the RNA-induced silencing complex (RISC) to target specific mRNAs on the basis of sequence complementarity (46). There is a high degree of redundancy and flexibility in both the miRNA sequences and their mRNA targets, making them critical gatekeepers for a variety of cellular functions throughout the lifespan of an organism. The major function of most miRNAs in mammalian systems is translational repression, followed by subsequent degradation of complementary mRNA targets. These processes and their physiological consequences have been extensively studied in a variety of model systems since miRNAs were first discovered in the early 1990s (46). However, more recent work has been focused on the cellular mechanisms that control the degradation of mature miRNAs. This mini-review highlights the current understanding of miRNA degradation in the context of the mammalian central nervous system, with particular emphasis on the exciting findings concerning the role of neuronal TDMD, or target-directed miRNA degradation.

RAPID miRNA DEGRADATION KINETICS IN THE BRAIN

In general, mature miRNAs are relatively stable in mammalian systems, where they exhibit half-lives in the hours timescale (19, 27, 30, 39). This stability is thought to be imparted by its association with the multiprotein complex called RISC. Structural analyses revealed that the miRNA 5′ and 3′ ends are embedded into an Argonaute (AGO) protein, an RNA-binding component of RISC, and protects the miRNA from cellular exoribonucleases (RNases) (44, 47). Notably, the central nervous system represents a striking exception to the global stability of miRNAs in other tissues. Various reports have shown that miRNA degradation kinetics are extremely rapid (in the timescale of minutes to hours) in neurons (10, 26, 42). For instance, Krol et al. (29) reported that the miRNA cluster miR-183/96/182 was regulated by light/dark stimuli and that their levels were drastically altered within ~90 min in retinal neurons by the combination of rapid decay and increased transcription. Moreover, neuronal miRNAs that were not regulated by light, such as miR-15, -16, and -29c, still exhibited rapid turnover, suggesting that the rapid kinetics were an intrinsic property of these neuronal miRNAs (29). These findings were later corroborated in other neuronal model systems. For instance, neuronally enriched miR-9-5p and miR-9-3p, which are derived from the 5′ and 3′ arms of the miRNA precursor, respectively, were degraded in the minutes times scale in a rat hypothalamic cell line (IVB) and in multiple brain regions (26). Rapid miRNA turnover was also reported in primary neuronal cultures and post mortem human brain tissues (1~3.5 h) (42). Together, these data suggest that neuronal mechanisms regulating miRNA degradation are unique compared with miRNAs found in peripheral tissues, where mature miRNA expression can be observed for days following inhibition of miRNA transcription or processing in cardiomyocytes and mouse embryonic fibroblasts (15, 45). However, it remains to seen whether rapid degradation is a general property of neuronal miRNAs or whether these observed half-lives are a unique characteristic of a specific miRNA subset.

TARGET DIRECTED miRNA DEGRADATION

This unique regulation can be explained, in part, by the process of TDMD (target-directed miRNA degradation), which seems to be more efficacious in primary neurons (10). TDMD is a phenomenon whereby a target RNA is capable of inducing the degradation of its complementary miRNA counterpart, which then allows the mRNA target to escape the canonical miRNA-silencing pathway. Specifically, extensive complementary binding at the 3′ region of the miRNA induces a conformational change that releases the embedded 3′ end of the miRNA from AGO (38, 44), which then exposes the 3′ end to various enzymatic processes that ultimately lead to its degradation. Notably, the binding requirements for TDMD are distinct from siRNA-mediated silencing, which also requires extensive 3′ complementarity. Specifically, structural modeling of the miRNA-target duplex within AGO2 indicates that target pairing in the central regions beyond the 11th nucleotide position is sterically inhibited. This contrasts starkly with the siRNA-mediated silencing mechanisms that require complementarity at the central regions in addition to the 3′ end, indicating that AGO2 adopts a structurally distinct conformation for TDMD compared with siRNA-mediated slicing (44). The minimal architectural binding requirements of the miRNA:target to elicit TDMD are still the focus of ongoing research. In Drosophila, a mismatch of up to 8 nucleotides at the 3′-end was tolerated before TDMD impairment was observed (2), whereas a 4 nucleotide central bulge flanked by extensive complementary binding was preferred to elicit TDMD in mammalian neurons (10). The amount of mismatched base pairs at the 3′ end that are tolerated to initiate TDMD likely varies between species and cell types; however, near-perfect seed sequence complementarity seems to be a requirement. Notably, artificial mRNA targets that lacked seed complementarity were able to induce TDMD in vitro (37), but it remains to be seen whether such examples can arise in an endogenous cellular environment.

The basic principles of TDMD were initially discovered by the use of artificial mRNA targets, such as antagomiRs and miRNA sponges. These synthetic targets were designed with extensive complementarity and have been shown to accelerate the rate of decay for their respective miRNAs in mammalian cells (3). miRNA sponges are highly complementary RNAs transcribed from stably integrated genetic components and have been shown to elicit TDMD in multiple cell types. For instance, these genetically engineered miRNA sponges effectively reduced endogenous levels of miR-16 and miR-20 in 293T and HeLa cells (13), as well as miR-223 in hematopoietic stem cells (16), although the efficacy of TDMD was much more potent in neuronal cells compared with other cell types (10, 28). Viruses have also evolved a cellular mechanism reminiscent of TDMD to attack host defense systems, specifically through their production of viral RNAs that can destabilize host miRNAs (18, 36). The Herpesvirus saimiri (HVS) virus, for example, expresses a U-rich small noncoding RNA called HSUR1 that destabilizes the miR-27 family via extensive base pairing at the 3′ end. Similarly, the murine cytomegalovirus (mCMV) elicits the rapid TDMD of the miR-27 family through its own viral transcript, m169 (32). The degradation of miR-27, which targets genes necessary to mount an appropriate immune response, allows for the propagation of the viral genome.

While TDMD had been successfully demonstrated using artificial or viral RNA targets, it remained unclear whether this phenomenon could occur in normal physiology. This was partly due to the apparent dearth of highly complementary miRNA targets present in the mammalian genome. However, evidence for endogenous mRNA targets capable of eliciting TDMD are starting to emerge, especially in the context of the central nervous system (Table 1) (4, 10, 11, 28, 49). In the cerebellum of zebrafish, libra, a long noncoding RNA, has been shown to destabilize miR-29b. Similarly, miR-29b was targeted for degradation by nrep (neuronal regeneration-related protein) a libra homolog, in the mouse cerebellum (4). More recently, the Bartel laboratory demonstrated a complex network of TDMD pathways in the mouse brain. Specifically, the long noncoding RNA Cyrano, a key regulator in development, was found to trigger TDMD of the miR-7 family. The reduction of miR-7 subsequently allowed for the accumulation of its downstream circular RNA (circRNA) target cdr1as. This target circRNA, in turn, elicited a modest reduction of miR-671, underscoring the interconnectivity of degradation pathways among varying noncoding RNA species (28). Taken together, these studies provide evidence that TDMD occurs within neuronal systems, with evidence from both endogenous and exogenous RNA targets. Furthermore, the efficacy of TDMD appears to be higher in the central nervous system, as exogenous RNA targets elicited the degradation of their respective miRNA to a greater degree in primary hippocampal neurons and cerebellar granule cells compared with HEK293T and mouse embryonic fibroblasts (10). However, it should be noted that this phenomenon has also been reported to some degree in nonneuronal cells. In mouse fibroblasts, Ghini et al. (17) identified hundreds of potential endogenous RNA targets capable of eliciting TDMD. Among these, the endogenous mRNA Serpine1 was found to destabilize miR-30b-5p and miR-30c-5p to regulate cell cycle reentry processes. It remains to be seen whether other endogenous targets will emerge in the coming years, specifically in the central nervous system. Nevertheless, further investigation is warranted to determine the extent to which TDMD can explain the rapid miRNA degradation observed in neurons.

Table 1.

RNA targets that elicit TDMD in the central nervous system

miR:TDMD Target	Species/Tissue	Sequence Complementarity	Reference
miR-26b: NREP	Zebrafish/cerebellum	UAGCACCAUUUGAAAUCAGUGUU AUCGUGGUAAG_U_AGUCACAGA	(4)
miR-7: Cyrano	Mouse/cerebellum, cortex	UGGAAGAC_UA_GUGAUUUUGUUGUU ACCUUCUGUAACCACUAAAACAACAA	(29)
miR-132: Artificial construct	Rat/primary hippocampal neurons	GCUGGUACCGACAUCUGACAAU CGACCAUGGCNNN_GACUGUUA	(10)
miR-124: Artificial construct	Rat/primary hippocampal neurons	CCGUAAGUGGCGCACGGAAU GGCAUUCANNN_GUGCCUUA	(10)

Boldface sequences represent mismatched nucleotides between the miRNA and the mRNA target. miR, microRNA; TDMD, target-mediated miRNA degradation; NREP, neuronal regeneration-related protein; Cyrano, a long, noncoding RNA.

POTENTIAL ENZYMES IN THE TDMD PATHWAY

The conformational changes induced by the miRNA:target interaction expose the miRNA to further enzymatic processing. Tailing, or nontemplated addition, to the miRNA 3′ end is performed by terminal nucleotidyltransferases (TNTases), and these events have been shown to affect the stability of miRNAs (Table 2). For example, TUT4, a terminal uridyltransferase, was found to oligouridylate pre-let-7 at the 3′ end. This oligo(U) tail subsequently served as a signal for decay, catalyzed by the Perlman syndrome exonuclease DIS3L2 (6). In an alternative pathway, oligoadenylation at the 3′ end mediated by PAPD5 [noncanonical poly(A) RNA polymerase] destabilized miR-21 via degradation by PARN [poly(A)-specific ribonuclease], underscoring the potential diversity of miRNA degradation processes (5). Furthermore, the length of the nucleotidyl additions was critical for determining the stability of the miRNA. For instance, monoadenylation of mature miR-122 catalyzed by GLD2, another poly(A) RNA polymerase, resulted in a stabilizing phenotype whereas oligoadenylation of the same miRNA promoted its degradation through PARN (9, 24, 25). In contrast, monoadenylation by GLD2 had no effect on the stability of miRNAs in the hippocampus, indicating tissue-specific regulation of miRNA degradation (34). Recently, RNA modifications, such as the 3′-terminal 2′-O methylation, was also shown to enhance the stability of miR-21-5p in non-small-cell lung cancer (NSCLC) (31).

Table 2.

Tailing enzymes potentially involved in regulation of TDMD in the brain

Enzyme	Function	Brain Region Specificity (Protein Atlas)	Reference
TUT1 (PAPD2)	Interacts with miR-27 isoforms upon TDMD induction	Expressed in all regions	(20)
TUT2 (GLD2)	Monoadenylate miR-122, stabilizes miRNA	Expressed in all regions	(9, 25)
TUT3 (PAPD5)	Oligoadenylate miR-21, destabilizes miRNA	Expressed in all regions	(5)
TUT4	Oligouridylate pre-let-7, signal for decay	Expressed in all regions	(6)

TDMD, target-mediated miRNA degradation; TUT, terminal uridyltransferase; GLD2, a poly(A) RNA polymerase; PAPD5, a noncanonical poly(A) RNA polymerase.

However, the extent to which these enzymatically edited miRNA isoforms are functionally coupled to TDMD processes remains unclear. For instance, only DIS3L2 has been observed to degrade miRNAs after binding to a highly complementary mRNA target (20). Additionally, after transfection with an artificial RNA target, DIS3L2 was copurified with several components of both the TDMD and RNA-induced silencing complex TUT1, exoribonuclease 2 (XRN2), and AGO2 (20), underscoring the close proximity of the tailing and degradation machinery. Furthermore, tailing processes were initiated with the miRNA still bound to AGO2 (10, 20), implicating the critical role of target binding to the 3′ modification of miRNA. In contrast, it appears that PARN-mediated miRNA degradation is independent of TDMD, as degradation was observed to occur outside of RISC (24). However, the possibility remains that target binding is able to eject the miRNA from RISC where it would be exposed to cytoplasmic exonucleases such as PARN and XRN2. In fact, highly complementary targets have been shown to dissociate guide RNAs from RISC in vitro (12). This pathway would reveal a novel and understudied tailing-independent mechanism of TDMD.

Recent evidence, however, seems to indicate that tailing mechanisms are, in large part, independent of TDMD. For example, double knockout of both TUT4 and TUT7 did not impair HSUR1-mediated TDMD of the miR-27 family (44). Likewise, the viral m169 transcript was able to invoke TDMD of the miR-27 family even after TUT1 knockdown (20). Finally, in the mouse brain, GLD2 knockout had no effect on Cyrano-mediated TDMD of miR-7 (28), suggesting that, although TDMD can expose the miRNA 3′ end to tailing modifications, this enzymatic event does not necessarily initiate miRNA degradation. It is important to note, however, that there are 11 terminal nucleotidyl transferases (TNTases) expressed in mammals with redundant functions (48); therefore, it remains to be seen whether the miRNA tailing mechanisms could be compensated for by other TNTases in these knockout experiments.

Advantages of TDMD in Neurons

Further investigation is required to understand the molecular mechanisms of TDMD, specifically in regard to its potency in the mammalian brain. The evidence indicates that nonneuronal cells also possess the required molecular machinery for TDMD processes, as transfection of artificial targets successfully elicits TDMD. For instance, the Perlman syndrome exoribonuclease (DIS3L2), terminal uridylyl transferase 1 (TUT1), and 5′-3′ exoribonuclease 2 (XRN2) are all constitutively expressed with low tissue specificity. Therefore, one possibility for the observed potency of TDMD in the brain could be due to differential expression of endogenous targets. An exciting possibility for endogenous TDMD targets in the brain could be circRNAs. These circularized transcripts are noncoding RNAs that are formed from back-splicing events and are highly expressed in the brain, where, notably, ~20% of protein coding genes were linked to circRNA formation (50). Furthermore, circRNAs are relatively stable compared with other linear RNAs, as they are immune from degradation through exoribonucleases. Additionally, circRNAs have already been well characterized as miRNA sponges, potentially having over 70 miRNA-binding sites (21). Therefore, it is not difficult to envision that these highly expressed, yet stable, RNAs serve as endogenous targets for TDMD in the mammalian brain, as their stability could allow escape from canonical miRNA-mediated silencing. Although circRNA-mediated degradation of miRNAs remains highly speculative, its functional role in the brain provides exciting avenues for future research.

Nevertheless, the advantages of TDMD in neurons, especially at the synaptic terminals, are abundantly clear. Neurons simultaneously integrate a variety of external stimuli at the synapse, and recent evidence indicates that miRNAs along with the RISC components are spatiotemporally localized to the synaptic terminals, primed to receive external input (22). Indeed, miRNA abundance can be directly regulated by synaptic activity, which in turn alters the landscape of local protein synthesis in dendrites and axons (22). Therefore, the ability to rapidly adjust miRNA to target mRNA stoichiometry is critical in the maintenance of normal neuronal function, but it remains unknown whether TDMD processes can be regulated by synaptic activity. Interestingly, circRNAs have been shown to be enriched in the synapse as well (41), providing further evidence for its potential role as an endogenous neuronal substrate for TDMD. The regulation afforded by TDMD would allow for the rapid fine-tuning of miRNA levels at the synapse, which has been shown to be critical for the maintenance of neuronal homeostasis.

Clinical Implications of Dysregulated miRNA Degradation Pathways

The dysregulation of miRNAs in the mammalian brain has been well characterized in the pathogenesis of various neurodegenerative diseases (Table 3) (see reviews in Refs. 40 and 43) Whereas miRNA dysregulation is likely attributed to deficits in various facets of multiple interconnected molecular pathways, it can be speculated that the dysregulation of miRNA degradation machinery is one mechanism for miRNA overexpression in neuropathology. For instance, miRNA overexpression is apparent in Alzheimer disease (AD), where miR-125b upregulation enhanced tau phosphorylation in primary neurons and in mouse brains, linking miR-125b to a molecular feature of AD pathology (8, 33). Similarly, the upregulation of miR-146a led to an increase in the processing of amyloid precursor protein (APP), which subsequently resulted in amyloid accumulation, both hallmarks of AD (42). Moreover, a group of miRNAs (miR-26b, miR-244, and miR-106b), were upregulated in the substantia nigra and amygdala of patients with Parkinson disease (PD). Increased levels of these miRNAs inhibited genes in the chaperone-mediated autophagy pathway, which ultimately impaired the sufficient degradation of α-synuclein protein, allowing for Lewy body deposit formation (1). The cause of the increased miRNA levels was not determined in these cases; however, it is reasonable to predict that defects in mature miRNA degradation might be a possible explanation. The dysregulation of miRNAs has also been associated with other neurodegenerative diseases such as Huntington disease (HD) and schizophrenia (23), underscoring the necessity of cellular degradation processes to prevent overabundant miRNA expression in the brain. Furthermore, if TDMD is indeed more efficacious in the context of the central nervous system, the exciting therapeutic potential of exogenous RNA molecules provides a theoretical solution to reducing overabundant miRNA expression in the brain. Therefore, further investigation on the causal link between miRNA dysregulation and TDMD impairment is critical to reveal novel molecular pathways contributing to neurodegenerative disease.

Table 3.

MicroRNAs associated with neurodegenerative disease

Pathology	Dysregulated MicroRNAs	Reference
Alzheimer’s disease	miR-9, miR-206, miR-219, miR-132-3p, miR-212, miE-9, miR-125b, miR-128, miR-146a, miR-155, miR-34a, miR-34c, let-7, miR-33, miR-106b, miR-758	(8, 33, 40, 42, 43)
Parkinson’s disease	miR-26b, miR-244, miR-106b, miR-494, miR-7, miR-205, let-7, miR-1, miR-29, miR-133b	(1, 40, 43)
Huntington’s disease	miR-132, miR-9-5p, miR-9-3p, miR-10b-5p, miR-34b	(23, 43)
Schizophrenia	miR-16, miR-20a, miR-128a, miR-338	(23)

TECHNICAL CONSIDERATIONS FOR MEASURING miRNA DEGRADATION AND FUTURE DIRECTIONS

Various experimental approaches have been used to assess mature miRNA stability in cell-free, whole cell, and whole tissue models. Cell-free in vitro degradation assays provide a useful biochemical approach to the investigation of miRNA degradation kinetics. Briefly, synthetic radiolabeled miRNA sequences are incubated with cell or whole tissue lysate across multiple time points (7, 26). The ability to specifically manipulate the miRNA sequence and various components of the lysate allows for a detailed investigation of the critical cis and trans determinants contributing to miRNA degradation across multiple paradigms (26). However, lysis procedures disrupt the structural integrity of the cellular microenvironment and affect the normal distribution and subcellular localization of its molecular constituents, making it difficult to attribute the half-lives obtained from these studies as representative of endogenous systems. In whole cell models, transcriptional inhibition with actinomycin D or DRB (5,6-dichloro-1-β-d-ribofuranosylbenzimidazole) has been the most common method to assay for miRNA abundance. This method inhibits transcription of pri-miRNAs, which then allows for quantification of the remaining pool of mature miRNAs at subsequent time points (26, 42). Although this method is beneficial in determining the degradation kinetics of miRNAs in intact cells, transcriptional inhibition is also cytotoxic, as it initiates apoptotic processes. Therefore, it is difficult to extrapolate the degradation kinetics obtained from these studies to endogenous systems. Recently, metabolic labeling has been adapted to globally assess miRNA turnover kinetics in intact cells (27, 35). Briefly, newly transcribed RNAs are labeled with either 4-thiouridine or 5-ethynyl uridine, and these nascent transcripts are subsequently purified for RT-qPCR or RNA-seq analysis at various time points to model their rate of decay. However, an important technical limitation in this approach is that the relative abundance of U nucleotides available for labeling differs significantly between miRNAs, making it difficult to compare different, or low abundant, miRNAs within the same cell or tissue. Despite this limitation, this approach is the most likely to provide the best approximations for the endogenous half-lives of highly abundant miRNAs with high U content in a physiological context. A critical limitation for these experimental procedures is that kinetic measurements in an in vivo context remains elusive. Specifically, the capability to measure miRNA degradation kinetics in an intact animal brain would significantly further our understanding of miRNA biology apart from simply homogenous cell populations. Furthermore, current approaches reveal relatively little concerning the spatiotemporal localization of endogenous miRNAs and how this relates to its degradation. Improvements to current imaging techniques would allow for the visualization of endogenous miRNAs throughout their life cycle, potentially revealing novel RNA-binding proteins or subcellular compartments of interest. Therefore, the continued development of new experimental strategies to analyze endogenous miRNAs is critical for our understanding of miRNA degradation in the context of an intact and integrated nervous system.

CONCLUSIONS

Recent developments have shed light on the ability of highly complementary endogenous targets to invoke the degradation of miRNAs, particularly in the central nervous system. However, there is a relative lack of knowledge concerning the mechanistic details of the TDMD pathway and how this pathway can be regulated in different cellular environments. Furthermore, future challenges include determining the prevalence of TDMD in the global pathways governing miRNA degradation. Undoubtedly, novel endogenous targets are likely to emerge, particularly in neuronal model systems, and their functional role in the regulation of neuronal physiology and disease is yet to be explored. Moreover, although TDMD appears to be an important process for some miRNAs, there are clearly other important degradation pathways yet to be defined. These alternative pathways represent a major unexplored area, and understanding them in an intact physiological context will likely require improved technological capabilities to advance our current understanding of miRNA degradation.

GRANTS

This work was supported in part by National Institute on Aging Grant AG0336.

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

C.K.K. drafted manuscript; C.K.K. and T.R.P. edited and revised manuscript; C.K.K. and T.R.P. approved final version of manuscript.