UCLA

MicroRNAs (miRNAs) are a class of non-coding RNAs that tune gene expression by negatively regulating at least 60% of protein-coding genes. In animals, they are required for development and cell physiology. Due to their role in controlling gene expression, dysregulation of miRNA biogenesis can cause genetic disorders, immunity deficits, neurological problems, and cancers. miRNA genes are transcribed into primary miRNA (pri-miRNAs), which undergo a multi-step maturation process. The first step occurs in the nucleus, where the pri-miRNA is cleaved by the Microprocessor Complex (MC). The MC contains the ribonuclease Drosha and an RNA-binding hemoprotein DGCR8. The MC specifically recognizes the characteristic pri-miRNA hairpin to initiate miRNA maturation. This thesis centers on the nuclear step of miRNA biogenesis. My research has two main areas of focus; 1) dissecting the molecular and structural determinants that govern pri-miRNA processing, and 2) investigating the regulation of pri-miRNA processing in normal physiology and diseases.

The MC identifies pri-miRNA substrates from a myriad of other RNAs. DGCR8 plays an important role in pri-miRNA recognition, but the mechanism was unknown. DGRC8 contains two double-stranded RNA-binding domains (dsRBDs), but these domains do not provide specificity. It has been shown that “junctions” between single-stranded and double-stranded regions in pri-miRNAs are important features that define MC substrates. However, the protein moiety that recognizes pri-miRNA junctions had not been identified. We discovered that DGCR8 contains an RNA-binding heme domain (Rhed) that directly and specifically binds pri-miRNA junctions. Further, we showed that the Rhed and its RNA-binding surface are required for efficient pri-miRNA processing.

Previous studies of our lab showed that the pri-miRNA processing activity of DGCR8 specifically requires heme in its Fe(III) redox state. However, it is unknown how much Fe(III) heme is available in cells and how much is required to support pri-miRNA processing. During our investigation of the Rhed-pri-miRNA interface, we performed mutagenesis on DGCR8 that led to identification of single mutations with reduced heme affinity but nearly full processing activity in cells. This meant that the Fe(III) heme affinities of these mutants are sufficient for them to acquire heme and to process pri-miRNAs. In contrast, all heme-binding-deficient mutants of DGCR8 we previously characterized are inactive in cells and have lower affinities for Fe(III) heme. We discovered that the Fe(III) heme affinity threshold for activating DGCR8 is shifted by overall heme availability changes in cells. These results indicate the presence of an available ferric heme pool that distinctly determines pri-miRNA processing efficiency in cells. Our study suggests cellular redox state and currently unknown Fe(III) heme-specific transporter proteins may be important for regulating miRNA maturation.

Secondary structure is a defining characteristic of pri-miRNA substrates. Substantial effort has been made to identify the important features of processing-competent pri-miRNA hairpins. Little attention is paid to structures outside the canonical hairpin. We identified a helix (f-helix) flanking the pri-miR-30a hairpin. Using an in vitro processing assay, we found that disrupting the pairing in this structure caused processing defects that could be rescued by repairing the helix. Although f-helices are not found in all pri-miRNAs, our finding has important implications on how to improve DNA vector-based RNA interference technology. The second generation short-hairpin RNAs (shRNAmir) are designed to mimic pri-miRNA and thus produce small RNAs through the natural miRNA maturation pathway. Their structure is most often based on pri-miR-30a. Indeed, we showed that the f-helix was functionally important for shRNAmir processing and knockdown efficiency. This work indicates that structural elements surrounding the pri-miRNA hairpin may serve as regulatory elements for miRNA maturation.

miRNA expression is globally repressed in many tumors, and this dysregulation can drive tumorigenesis. High numbers of somatic mutations are a hallmark of many tumors, but distinguishing between disease-driving and spurious mutations remains a challenge. We analyzed seven missense DGCR8 mutations and found that four are severely defective. We determined that the E518K mutation identified in about 3% of Wilms Tumors, most likely disrupts the folding of DGCR8. The F448L mutation found in colon cancers rendered DGCR8 unable to bind Fe (III) heme and the Drosha protein. This result suggests that heme-binding to DGCR8 may be required for strong association with Drosha. Biochemical characterization of K289E and G336E show they have reduced affinities both pri-miRNA and heme. Altogether, these results suggest that a substantial fraction of tumor-derived DGCR8 mutations results in functional deficits and thereby are likely to make important contribution to a cancer-promoting phenotype.

Overall, my thesis investigates the fundamental mechanism of pri-miRNA processing, and its applications to biotechnology and diseases. My main scientific contribution is the identification and characterization of the unique RNA-binding heme domain in DGCR8. My work demonstrates the importance of heme in pri-miRNA processing, and defines an Fe(III) heme pool that regulates MC activity, thereby linking miRNA biogenesis with heme biology and cellular redox environments. My research extends our understanding of the functional/regulatory elements in pri-miRNA hairpins to include regions flanking the conserved pri-miRNA hairpin. Finally, I identify probable driver mutations for tumor formation by connecting several cancer-associated somatic point-mutants of DGCR8 to functional defects in processing.