The microprocessor is a complex comprising the RNase III enzyme Drosha and the double-stranded RNA-binding protein DGCR8 (DiGeorge syndrome critical region 8 gene) that catalyses the nuclear step of miRNA (microRNA) biogenesis. DGCR8 recognizes the RNA substrate, whereas Drosha functions as an endonuclease. Recent global analyses of microprocessor and Dicer proteins have suggested novel functions for these components independent of their role in miRNA biogenesis. A HITS-CLIP (high-throughput sequencing of RNA isolated by cross-linking immunoprecipitation) experiment designed to identify novel substrates of the microprocessor revealed that this complex binds and regulates a large variety of cellular RNAs. The microprocessor-mediated cleavage of several classes of RNAs not only regulates transcript levels, but also modulates alternative splicing events, independently of miRNA function. Importantly, DGCR8 can also associate with other nucleases, suggesting the existence of alternative DGCR8 complexes that may regulate the fate of a subset of cellular RNAs. The aim of the present review is to provide an overview of the diverse functional roles of the microprocessor.
miRNAs (microRNAs) are small non-coding RNAs of approximately 22 nucleotides in length that negatively regulate the expression of complementary mRNAs. The mechanism of miRNA-mediated inhibition of gene expression operates at the level of mRNA stability and/or inhibition of translation . miRNAs have diverse unique expression patterns and are involved in developmental and differentiation processes in animals, regulating a wide variety of cellular pathways . It is becoming increasingly clear that miRNA biogenesis and function has to be tightly controlled and that aberrant miRNA expression can lead to the progression of human disease . In the present article, we review the biogenesis pathway of miRNAs, with a particular emphasis on the recent discovery of non-canonical functions for the microprocessor.
Mammalian miRNAs are embedded in long pri-miRNAs (primary miRNAs), which are independent transcription units or can alternatively be located within introns of pre-mRNAs (precursor mRNAs). They are transcribed by RNA polymerase II and processing of pri-miRNAs by subsequent nuclear and cytoplasmic processing events gives rise to mature miRNAs. The nuclear step is catalysed by the microprocessor complex, comprising Drosha and its partner DGCR8 (DiGeorge syndrome critical region 8 gene) and results in the production of pre-miRNAs (precursor miRNAs). Subsequently, pre-miRNAs are exported from the nucleus by Exportin 5 and are further processed in the cytoplasm by the RNase III enzyme Dicer into mature miRNAs [4,5].
Biochemical studies of the microprocessor complex in human cells have shown that Drosha can form two different complexes. A smaller complex, comprising Drosha and its partner DGCR8, constitutes the minimal catalytically active complex in vitro that directs cleavage and production of pre-miRNAs [6,7]. A larger complex contains additional RNA-BPs (RNA-binding proteins), such as RNA helicases, hnRNPs (heterogeneous nuclear ribonucleoproteins) and other associated proteins that can regulate its activity . DGCR8 is the microprocessor component that directly interacts with pri-miRNAs and functions as a molecular anchor that measures the distance from the dsRNA (double-stranded RNA)–ssRNA (single-stranded RNA) junction to direct the cleavage by the endonuclease Drosha 11 bp away from this junction [9,10]. In the cytoplasm, Dicer catalyses an additional processing step (dicing) that results in the production of approximately 22 nt miRNA duplexes. In humans, Dicer interacts with two closely related proteins, TRBP [TAR (transactivating response) RNA-binding protein; also known as TARBP2] and PACT (protein activator of the interferoninduced protein kinase; also known as PRKRA), which are not required for activity, but have been implicated in defining the cleavage site and facilitating the formation of the RISC (RNA-induced silencing complex) (, reviewed in ). One strand of this duplex is preferentially selected to bind to one of the Ago (Argonaute) proteins (Figure 1). Mammals have multiple Ago proteins, with Ago2 being the only one to have retained a slicer domain. The miRNA guides Ago to target mRNAs and can cause degradation or translation inhibition of the transcripts, depending on the extent of complementarity . The canonical miRNA biogenesis pathway driven by the RNase III enzymes Drosha and Dicer has been challenged by the discovery of an unexpected variety of alternative mechanisms that generate functional miRNAs. These non-canonical pathways include, among others, the generation of mature miRNAs by the action of the spliceosome (mirtrons) (reviewed in ).
miRNA biogenesis pathway
The production of mature miRNAs is tightly regulated at the post-transcriptional level at different steps throughout the biogenesis pathway. This can be achieved by proteins present in the large microprocessor complex and/or by auxiliary factors that directly bind to pre-miRNAs [14,15]. The most extensively studied example is the role of the RNA-BP Lin28 in the regulation of let-7 miRNA family biogenesis, which is important for stem cell differentiation and normal development. Lin28 binds to the terminal loop of let-7 family members and blocks its processing at both the Drosha and/or Dicer level . Another interesting example is the activity of TGFβ (transforming growth factor β) and BMP (bone morphogenetic protein) signalling pathways that stimulate the Drosha-mediated processing of an individual miRNA miR-21 . In some cases, the same RNA-BP can display either stimulatory or inhibitory activities with different subsets of miRNAs. For instance hnRNP A1, a protein involved in many aspects of RNA processing, acts as an auxiliary factor for the Drosha-mediated processing of pri-miR-18a. It binds to the CTL (conserved terminal loop) of pri-miR-18a and rearranges its structure, creating a more favourable cleavage site for Drosha processing [18,19]. However, binding of hnRNP A1 to a CTL does not always lead to stimulation of miRNA processing. It was shown that hnRNP A1 is a negative regulator of let-7a in differentiated cells by antagonizing the positive role of KSRP (KH-type splicing regulatory protein) [20,21]. Altogether, these data suggest the existence of positive and/or negative auxiliary factors for the processing and production of specific miRNAs. Previous findings suggest that this type of regulation is extensive and in most cases involves recognition of conserved sequences in the terminal loop of pre-miRNAs (reviewed in ).
Non-canonical functions of the microprocessor
Functional studies of the miRNA processing machinery using mouse models showed that Drosha, DGCR8 and Dicer deficiencies resulted in all cases in early embryonic lethality, confirming a critical role for miRNAs in normal development [22–24]. MEFs (mouse embryonic fibroblasts) derived from these knockout mice showed some phenotypic differences as well as poor correlation between the populations of mRNAs affected by these two endonucleases. Dicer-null ES (embryonic stem) cells have normal morphology, but their proliferation potential is compromised [25,26]. Dgcr8-null ES cells cannot fully down-regulate pluripotency markers, which interferes with their ability to fully differentiate, suggesting that miRNAs have a crucial role in silencing ES cell self-renewal .
Transcriptional analyses of Dicer-null ES cells show that they are devoid of all canonical and most non-canonical miRNAs and siRNAs (small interfering RNAs), whereas Dgcr8-null ES cells are only lacking canonical miRNAs . Thus these differences strongly suggested that Dicer has miRNA-independent roles in ES cell function. Furthermore, the fact that Dgcr8-null cells have less severe phenotypes than Dicer-null cells suggested an important role for microprocessor-independent, Dicer-dependent small RNAs. Indeed, profiling small RNAs for these two populations of cells identified DGCR8-independent, Dicer-dependent noncanonical miRNAs, which include mirtrons, endo-siRNAs and endo-shRNAs (small hairpin RNAs) .
Comparisons between Dgcr8/Drosha and Dicer knockout phenotypes were also pursued to elucidate their roles within differentiated cells and tissues, such as NK (natural killer) cells, regulatory thymocyte cells, epidermal and hair follicle epithelial cells [28–30]. These studies found that inactivation of Drosha/DGCR8 or Dicer gave rise to highly overlapping phenotypes, suggesting a common affected pathway. Nevertheless, in early T-cell progenitors, Drosha and Dicer conditional knockouts did not produce identical phenotypes, indicating non-redundant functions of Drosha and Dicer in miRNA processing during early T-cell development . Transcriptomic changes that were unique to Drosha, but not Dicer, highlighted Drosha-dependent, Dicer-independent RNAs. Some of the mRNAs that were up-regulated in the absence of Drosha also contained a pri-miRNA-like structures which can be cleaved by the microprocessor, independently of miRNA production . Indeed, Drosha had been shown to cleave pri-miRNA-like hairpins harboured within the 5′UTR (untranslated region) of the mRNA encoding the DGCR8 protein [31,32]. This negative regulatory feedback has also been reported in Drosophila melanogaster, indicating that this is a conserved regulatory pathway . Comparison of the up-regulated mRNA populations in the absence of Drosha or Dicer identified another 25 transcripts that were proposed to be direct targets of the microprocessor. In addition, deep sequencing of Drosha cleavage products uncovered more mRNAs that could be regulated by endonucleolytic cleavage . This regulation has been shown to be critical during mouse neurogenesis, where microprocessor-mediated destabilization of Neurog2 mRNA facilitates neural stem cell maintenance . The pre-mRNA encoding the KapB gene of KSHV (Karposi's sarcoma-associated herpesvirus) is also processed by the microprocessor as it contains a pri-miRNA located at the 3′UTR that also causes concomitant transcript destabilization . It was reported recently that small RNAs that are recruited to the sites of DNA damage are Drosha- and Dicer-dependent; however, they are distinct from miRNAs . These small RNAs, termed DDRNAs (DNA damage response RNAs), bind to double-strand breaks and recruit DDR (DNA damage response) proteins, leading to activation of the DNA damage response. Interestingly, defects in DNA damage repair have been seen before in miRNA processing knockouts . Finally, microprocessor-mediated RNA cleavage can cause premature transcription termination of the TAR mRNA, encoded by HIV. In this case, Drosha-mediated endonucleolytic cleavage opens up the transcript for exonucleolytic degradation by Xrn2 and RRP6 . Altogether, the evidence described above has strongly suggested an expanded role for non-canonical functions of the microprocessor and/or Dicer.
Direct RNA targets of the microprocessor complex
Computational scanning of the entire human genome searching for hairpin structures enabled the identification of approximately 11 million hairpins resembling validated pri-miRNA hairpins . However, it is still not completely understood how the microprocessor complex selectively recognizes these structures. DGCR8 is the microprocessor component that binds and recognizes the dsRNA–ssRNA junction at the base of the pri-miRNA hairpin ; however, similar RNA secondary structures are found in many other cellular RNAs, suggesting that these could also be recognized and cleaved. The evidence described above pointed to a more extended role of the microprocessor in the regulation of a plethora of cellular RNAs. Recently, we took an alternative approach to identify direct RNA targets of the microprocessor complex, on the basis of sequencing the transcripts associated with DGCR8 by the HITS-CLIP (high-throughput sequencing of RNA isolated by cross-linking immunoprecipitation) technique . This technique relies on UV light irradiation of living cells to induce covalent cross-links between RNA and protein, allowing the use of very stringent washing conditions during the immunoprecipitation of the protein of interest to identify associated RNAs. It has been widely used to map in vivo RNA–protein interactions for several RNA-BPs involved in different aspects of RNA metabolism [41,42]. The study revealed that DGCR8 binds to many other types of structured RNAs other than miRNAs that harbour a predicted secondary structure resembling that of a pri-miRNA. These cellular targets comprised several hundred mRNAs, as well as snoRNAs (small nucleolar RNAs) and lincRNAs (long intergenic non-coding RNAs) (Figure 2) . Direct binding was observed for more than 2000 mRNAs and depletion of the microprocessor resulted in an up-regulation of selected mRNAs, suggesting that these were direct RNA processing targets. Furthermore, in vitro cleavage by Drosha was confirmed for a handful of mRNA targets. Interestingly, DGCR8 was also shown to bind cassette exons, suggesting a role for this complex in the regulation of alternative splicing. Remarkably, in the absence of DGCR8 more than 300 alternatively spliced events were altered. It was found that the microprocessor cleaves and destabilizes mRNA isoforms harbouring DGCR8 binding sites in cassette exons, resulting in an altered ratio of alternatively spliced isoforms. The discovery that DGCR8 binding to cassette exons may act to influence the relative abundance of alternatively spliced isoforms in a physiological manner is an exciting finding, but needs further investigation. In summary, this analysis revealed that the microprocessor binds and cleaves the mRNA of protein-coding genes, affecting mRNA abundance and also impacting alternative splicing regulation.
Non-canonical functions of the microprocessor
An alternative DGCR8 complex
DGCR8 targets include snoRNAs, short non-coding RNAs that modify rRNA, tRNA and snRNAs (small nuclear RNAs) by direct base pairing and recruitment of modifying enzymes, which guide 2′-O-methylation or pseudourydilation of the target RNA. Interestingly, snoRNAs are also a source of small RNAs, known as sdRNAs (snoRNA-derived RNAs) that associate with Ago proteins and act as functional miRNAs . Whereas sdRNAs have been shown to be Dicer-dependent, but Drosha-independent in some cases , other reports have shown that sdRNAs are dependent on both DGCR8 and Dicer in several organisms, such as human, mouse, chicken, fruitfly, Arabidopsis and fission yeast . Unexpectedly, DGCR8-mediated cleavage of snoRNAs is independent of Drosha, strongly suggesting that DGCR8 can associate with other nucleases to cleave subsets of target RNAs . It may well be that snoRNAs are only one type of cellular RNA target recognized by this putative alternative DGCR8 complex. This opens an exciting avenue of new research aiming to identify additional nuclease(s) that associate with DGCR8 in alternative complexes and to identify cellular RNA targets of this complex(es).
Microprocessor functions and DiGeorge syndrome
The DGCR8 gene is located in the 22q11.2 genomic region, which is deleted in DiGeorge syndrome patients [7,46]. This is the most common human genetic deletion syndrome that affects approximately one in 3000 live births. DiGeorge syndrome patients can present two different types of deletion, a 3.0 Mb or 1.5 Mb deletion, with the former being the most prevalent and spanning approximately 30 genes. In order to gain insight into this condition, an engineered mouse strain carrying a hemizygous chromosomal deficiency on chromosome 16 (that spans a segment syntenic to the human 1.5-Mb 22q11.2 microdeletion) was constructed. This Dgcr8+/− mutant mice [Df(16)A+/−] displayed a minor alteration of miRNA production in the brain, but still exhibited behavioural and cognitive defects and cardiac abnormalities that resemble the human condition [47,48]. Further analysis of the Df(16)A+/− mice revealed abnormalities in the formation of neuronal dendrites and spines, as well as altered brain miRNAs. This phenotype was partially due to a drastic reduction in the levels of a single miRNA, miR-185, which resides within the 22q11.2 locus. Thus phenotypes observed in this mouse model could be attributed to the combined effect of miR-185 and Dgcr8 hemizygosity . These studies highlighted an important role for miRNA biogenesis in this disease; however, it would be of great interest to determine whether altered non-canonical functions of the microprocessor could have any role in the origin and development of DiGeorge syndrome. The development of specific rescue experiments will unravel precisely which miRNAs are affected and whether the alternative functions of the microprocessor are necessary to restore brain function and revert the neuropsychiatric traits related to this deficiency.
Conclusions and future directions
In the present article, we reviewed evidence describing a more general role for the microprocessor that is not limited to the production of pre-miRNAs, but also directs the processing of a large number of cellular RNAs. The identification of non-canonical functions for the microprocessor complex highlights the necessity to reinterpret the phenotypes observed in DGCR8- and/or Drosha-deficient cells and animal models that were exclusively attributed to a defect in miRNA biogenesis. Evidence discussed in the present review also provides new insights into the complex role of the microprocessor in controlling gene expression by affecting the fate of several classes of RNAs.
Biogenesis and Turnover of Small RNAs: A Biochemical Society Focused Meeting held at the Royal Society, Edinburgh, U.K., 15–17 January 2013. Organized and Edited by Richard Bowater (University of East Anglia, U.K.), Amy Buck (Edinburgh, U.K.) and Javier Cáceres (Edinburgh, U.K.).
conserved terminal loop
DiGeorge syndrome critical region 8 gene
high-throughput sequencing of RNA isolated by cross-linking immunoprecipitation
heterogeneous nuclear ribonucleoprotein
long intergenic non-coding RNA
primary miRNA transcript
small interfering RNA
small nucleolar RNA
small nuclear RNA
We thank Elena Miñones-Moyano (Medical Research Council Human Genetics Unit) for comments on this review before submission.
This work was supported by core funding from the Medical Research Council and by the Wellcome Trust [grant number 095518/Z/11/Z].