Transcriptome-wide regulation of pre-mRNA splicing and mRNA localization by muscleblind proteins

Abstract

The muscleblind-like (Mbnl) family of RNA-binding proteins plays important roles in muscle and eye development and in myotonic dystrophy (DM), in which expanded CUG or CCUG repeats functionally deplete Mbnl proteins. We identified transcriptome-wide functional and biophysical targets of Mbnl proteins in brain, heart, muscle, and myoblasts by using RNA-seq and CLIP-seq approaches. This analysis identified several hundred splicing events whose regulation depended on Mbnl function in a pattern indicating functional interchangeability between Mbnl1 and Mbnl2. A nucleotide resolution RNA map associated repression or activation of exon splicing with Mbnl binding near either 3' splice site or near the downstream 5' splice site, respectively. Transcriptomic analysis of subcellular compartments uncovered a global role for Mbnls in regulating localization of mRNAs in both mouse and Drosophila cells, and Mbnl-dependent translation and protein secretion were observed for a subset of mRNAs with Mbnl-dependent localization. These findings hold several new implications for DM pathogenesis.

Figure 1. Dependence of splicing changes on total MBNL levels by RNA-Seq analysis

(A) RNA-Seq read coverage across Trim55 exon 8, from wild type and Mbnl1 knockout heart tissue: MISO Ψ values and 95% CI shown at right. (B) Global correlation of Ψ values for cassette exons with dendrogram shown below. (C) The fraction of MBNL1, MBNL2, or total MBNL remaining in the indicated knockout tissues and knockdown myoblasts relative to control tissues/cells. (D & E) Mean splicing change (mean |ΔΨ| correlates much better with total MBNL depletion than with MBNL1 depletion. See also Figures S1, S2 and Table S1.

Figure 2. MBNL binding is conserved and associated with context-dependent splicing activity

(A) MBNL1 and Nova CLIP tags derived from various tissues and cells in the 3' UTR of Reticulon 4. (B) Correlation of CLIP tag densities in 5 nucleotide windows across all 3' UTRs expressed in brain and C2C12 myoblasts. (C) Histogram of Z-scores of 4mers occurring in MBNL1 CLIP clusters from C2C12 myoblasts, relative to control regions from the same genes. (D) Alternative exons with MBNL1 CLIP clusters upstream or downstream are more conserved than other alternative exons expressed at similar levels. (E & F) Dependence on Mbnl levels of PSI values of Fibronectin 1 exon EIIIB and SercaI exon 22, respectively (CLIP data is shown above). (G) Dependence of change in PSI value (wild type – Mbnl1, Mbnl2 knockdown) on MBNL1 binding in the upstream intron (last 300 bases), alternative exon, or downstream intron (first 300 bases). See also Figures S3, S4 and Table S2.

Figure 3. Patterns of MBNL binding and a nucleotide-resolution map of splicing activity

(A) Examples of substitutions observed in MBNL1 CLIP tags and calculation of p(CIS). (B) Cytosine exhibits the greatest p(CIS) within MBNL1 CLIP clusters, and is most often sequenced as thymine. (C) Information (relative entropy) of positions adjacent to frequently substituted cytosines, as a function of substitution frequency. (D) Mean CLIP density near exons activated (red), repressed (blue) by MBNL1, using binding sites with p(CIS)_C→T > 0.1. Density around exons unaffected by Mbnl depletion is shown by dotted line, with 90% confidence intervals in gray. Colored dots denote 100 nucleotide-long windows significantly greater density of CLIP sites at regulated than non-regulated exons. (E) Mean MBNL1 CLIP density at positions along 3' UTRs using 3 CLIP datasets. (F) Conservation in 3' UTRs with and without MBNL1 CLIP clusters with similar expression levels and UTR lengths. (G) Significant Gene Ontology categories of genes with 3' UTRs having MBNL1 CLIP. See also Figures S4 and Table S2.

Figure 4. mRNA localization is associated with functional binding by MBNL

(A) Cellular fractionation scheme in C2C12 mouse myoblasts and fly S2R+ cells. Adherent myoblasts were fractionated on tissue culture plates, while semi-adherent S2R+ cells were fractionated using centrifugation, yielding “cytosolic”, “membrane”, and “insoluble” fractions. (B) Western blot of Hsp90, Calnexin, and Tropomyosin I proteins for each compartment in control and MBNL knockdown myoblasts. (C) Simplex representation of mRNA expression across compartments in control myoblasts, with mRNAs belonging to specific functional categories colored. (D) Density map of MBNL1 CLIP targets in control myoblasts. The relative abundance of genes with 0, ≥4, ≥6, or ≥8 CLIP clusters is illustrated within the simplex, and quantified for areas of the simplex on either side of the dashed line. (E) Cumulative distribution functions of relative enrichment in Insoluble/Cytosolic (left) or Membrane/Cytosolic (right) of genes grouped by number of 3' UTR CLIP clusters, with significance determined by rank-sum test. See also Figures S5 and Table S3.

Figure 5. MBNL binding is associated with regulation of mRNA localization in mouse and fly, and may contribute to protein secretion

(A & B) Change in mRNA localization following depletion of MBNLs in mouse myoblasts and fly S2R+ cells. Arrows represent changes in mRNA localization, where the direction change is encoded by color. (C & D) The aggregate behavior following MBNL depletion of the subset of genes biased toward the insoluble compartment in control cells is shown in a polar plot, where the radial amplitude is the number of genes that relocalized in the given direction following MBNL depletion, stratified by density of CLIP clusters (for mouse) or UGCU 4mers (for fly). The aggregate change in localization towards each compartment is shown bar plots. P-values denote significance when comparing change in localization of MBNL targets versus non-targets (e.g., >2 versus 0–0.5 for mouse myoblasts). (E) The mean fold-change in ribosome footprint density following MBNL1, MBNL2, or double MBNL depletion relative to control was quantitated for mRNAs which were initially biased towards the membrane compartment (top) or the insoluble compartment (bottom), and whose localization shifted towards the insoluble compartment following MBNL depletion. Genes were further stratified by density of MBNL CLIP sites in 3' UTR. The P-value denotes the significance of the change in footprint density when comparing MBNL targets versus non-targets (rank-sum test). (F) Change in mRNA localization of Fibronectin 1 (Fn1) and Biglycan (Bgn) following MBNL depletion in mouse myoblasts (top), MBNL1 CLIP tag density in the 3' UTR of each gene (middle), results of secreted luciferase assay using luciferases containing the Bgn 3' UTR or the Fn1 last intron plus 3' UTR (bottom). See also Figure S6 and Table S4.

Figure 6. MBNL regulates Isoform-specific mRNA localization via binding to alternative 3′ UTRs

(A) The localization of long and short alternative 3' UTR isoforms is shown at left (as in Fig. 4B). Relative enrichment in pairs of cellular compartments is shown for long and short isoforms at right (as in Fig. 4E). (B) CLIP data for the 3' UTR of Gtpbp4. Localization of the long and short isoforms in normal myoblasts is displayed in solid blue and black circles, respectively, and in myoblasts depleted of Mbnl in outlined blue and black circles, respectively, with arrows indicating change in localization. Bar plot shows change in RPKM (Mbnl-depleted versus control myoblasts) for the long and short 3' UTR isoforms of Gtpbp4. (C) Genes were separated into those with and without CLIP clusters in the core or extended 3' UTR region, for short and long isoforms, respectively, and assessed for direction of re-localization following Mbnl depletion. Bar plot shows the extent of re-localization towards each compartment, where the magnitude of re-localization for CLIP targets is normalized relative to non-targets. See also Figure S7 and Table S5.

Figure 7. A model for nuclear and cytoplasmic functions of MBNL

MBNLs repress or activate splicing depending on binding location. In the cytoplasm, MBNL binding in 3' UTRs may facilitate targeting of mRNAs with signal sequences to the rough ER. Alternatively, transcripts may be targeted to membrane-rich organelles for localized translation via actin-, microtubule-, or intermediate filament-based molecular motors. These organelles may include synapses, NMJs, or the plasma membrane, depending on cell type. MBNL may mediate isoform-specific mRNA localization, in which MBNL binding sites within distal 3' UTRs are required for targeting to particular compartments. (Model based on data from previous figures. See also Figure S7.)