MBNL proteins and their target RNAs, interaction and splicing regulation

Abstract

Muscleblind-like (MBNL) proteins are key regulators of precursor and mature mRNA metabolism in mammals. Based on published and novel data, we explore models of tissue-specific MBNL interaction with RNA. We portray MBNL domains critical for RNA binding and splicing regulation, and the structure of MBNL's normal and pathogenic RNA targets, particularly in the context of myotonic dystrophy (DM), in which expanded CUG or CCUG repeat transcripts sequester several nuclear proteins including MBNLs. We also review the properties of MBNL/RNA complex, including recent data obtained from UV cross-linking and immunoprecipitation (CLIP-Seq), and discuss how this interaction shapes normal MBNL-dependent alternative splicing regulation. Finally, we review how this acquired knowledge about the pathogenic RNA structure and nature of MBNL sequestration can be translated into the design of therapeutic strategies against DM.

Figure 1.

Organization of human (Hs) and mouse (Mm) MBNL1, 2 and 3 genes. Schematic gene representations are based on sequences from RefSeq database. Exon numbers refer to protein coding exons and nucleotide length of each exon is indicated in brackets. Alternatively spliced exons and ZnFs are marked red and blue, respectively. Note that according to published data, the number of known alternative splicing isoforms of MBNL1–3 is significantly larger than depicted in this figure. Hs and Mm indicate Homo sapiens and Mus musculus, respectively.

Figure 2.

Expression pattern of MBNL mRNAs in different tissues and consequences on alternative splicing. Representative images of semi-quantitative reverse transcriptase-polymerase chain reaction (RT-PCR) analyses of MBNL1 exon 5, MBNL2 exons 6 and 7, MBNL3 exon 10 and NFIX exon 7 distribution (+e, PCR band representing exon inclusion and –e, representing exon exclusion) (A) and multiplex PCR-based quantification of relative MBNL1, 2 and 3 as well as total MBNL mRNA in various tissues (B). Human fetal (F) and adult (A) cDNA panels from Clontech were used as templates for PCRs. Note the predominant MBNL1 transcript expression in most tissues, especially in the adult skeletal muscle. Relatively high amounts of MBNL2 are detected in the brain, and MBNL3 in the liver and placenta. In (C) inclusion of MBNL1 exon 5, MBNL2 exon 7 and NFIX exon 7 was related to the total amount of MBNL mRNA. Note that with increasing amounts of MBNLs during differentiation, the splicing shifts toward MBNL-dependent exon exclusion in all depicted tissues. In contrast, splicing of MBNL2 exon 6 does not depend on MBNL content, as exon 6 is specifically included only in the brain and pancreas [see images in (A)]. AU indicates arbitrary units (P. Konieczny and K. Sobczak, unpublished data).

Figure 3.

The mode of ZnFs binding to single- and double-stranded RNAs. The model shown in (A) is adapted from the model deposited by Teplova and Patel in RCSB Protein Data Bank [3D2S structure, (52)]. The structure shows ZnF3/4 domain interacting with three 5′-CGCUGU-3′ RNA molecules. G2, C3 and U4 residues of one RNA molecule interact with ZnF4 while C1 and G5 from two separate RNA molecules bind to ZnF3. This model indicates the GC sequence as the minimum MBNL1 RNA binding motif. Panels (B and C) and (D and E) depict hypothetical interactions of MBNL1 with single-stranded RNA and double-stranded CUG repeat hairpins with locally unwound GC motifs, respectively. (B andD) are based on (A) while (C) refers to the data obtained by Cass et al., where two GC motifs are separated by only one nucleotide (51). In (E), in which one MBNL1 particle bridges two CUG^exp hairpins, we depict a hypothetical role of MBNL1 in CUG^exp foci formation.

Figure 4.

Important MBNL1 regions responsible for RNA binding and splicing regulation. Schematic representation of binding affinity of MBNL mutants to target RNAs (A) and their splicing activities (B). Data obtained from truncation (,,–50,53) and point mutation studies (47,48) are depicted on the left and right panels of (A) and (B), respectively. In point mutation studies, key amino acids responsible for interaction with guanine and cytosine bases were substituted to alanines as marked in the figure legend. Purcell et al. (48) substituted two amino acids per each ZnF motif while Edge et al. (47) mutated only one amino acid (marked in red). Decreasing binding affinities and splicing activities of distinct MBNL1 variants are marked with inverted triangles. The top of inverted triangles indicates normal/full-length MBNL1 binding affinity and splicing activity. Protein variants showing similar RNA binding or splicing activities are grouped (black vertical lines).

Figure 5.

MBNL binding sequence motifs relative to the regulated exon. Exons are either positively (A) or negatively (B) regulated by MBNL depending on its position relative to the alternative exon (compare with Figure 6A). Two MBNL targets do not follow this positional pattern (underlined). Motifs in white on black background are potential MBNL interaction sites and their mutation reduce or eliminate the effect of MBNL1-dependent regulation. Note that most of them contain YGCY. Motifs in black on gray background are also potential MBNL binding sites, however, mutation of these sites either does not significantly affect MBNL1-dependent regulation [(MAPT), (61)] or was not tested [(NFIX), (12)]. Asterisks show pre-mRNA targets for which additional MBNL binding sites have been indicated. In TPM1, the upstream sequence is critical for exon skipping (62). In (C), the MBNL1 binding motif in pre-miRNA-1–1 is indicated. Note that this is the only target containing a single MBNL consensus sequence. Hs, Mm, Gg and Rr indicate Homo sapiens, Mus musculus, Gallus gallus and Rattus rattus, respectively.

Figure 6.

MBNL-induced alternative splicing regulation. The model depicted in (A) is based on data obtained from the CLIP analysis performed by Wang et al. (7). MBNL1 binding either upstream (red boxes) or downstream (green boxes) to the alternative exon promotes alternative exon exclusion as exemplified in (B) by cTNT exon 5, CLCN1 exon 7A and MBNL1 exon 5, or exon inclusion as exemplified in (C) by INSR exon 11. The mode of MBNL1 regulation of transcripts in parentheses is putative. Interaction of MBNL1 with pre-miRNA hairpin promotes pre-miRNA processing to mature miRNA (D).