hnRNP L is essential for myogenic differentiation and modulates myotonic dystrophy pathologies

Abstract

Introduction: RNA-binding proteins (RBPs) play an important role in skeletal muscle development and disease by regulating RNA splicing. In myotonic dystrophy type 1 (DM1), the RBP MBNL1 (muscleblind-like) is sequestered by toxic CUG repeats, leading to missplicing of MBNL1 targets. Mounting evidence from the literature has implicated other factors in the pathogenesis of DM1. Herein we sought to evaluate the functional role of the splicing factor hnRNP L in normal and DM1 muscle cells.

Methods: Co-immunoprecipitation assays using hnRNPL and MBNL1 expression constructs and splicing profiling in normal and DM1 muscle cell lines were performed. Zebrafish morpholinos targeting hnrpl and hnrnpl2 were injected into one-cell zebrafish for developmental and muscle analysis. In human myoblasts downregulation of hnRNP L was achieved with shRNAi. Ascochlorin administration to DM1 myoblasts was performed and expression of the CUG repeats, DM1 splicing biomarkers, and hnRNP L expression levels were evaluated.

Results: Using DM1 patient myoblast cell lines we observed the formation of abnormal hnRNP L nuclear foci within and outside the expanded CUG repeats, suggesting a role for this factor in DM1 pathology. We showed that the antiviral and antitumorigenic isoprenoid compound ascochlorin increased MBNL1 and hnRNP L expression levels. Drug treatment of DM1 muscle cells with ascochlorin partially rescued missplicing of established early biomarkers of DM1 and improved the defective myotube formation displayed by DM1 muscle cells.

Discussion: Together, these studies revealed that hnRNP L can modulate DM1 pathologies and is a potential therapeutic target.

Keywords: MBNL; ascochlorin; hnRNP L; myotonic dystrophy; smooth.

Figure 1.. shRNAi–mediated knockdown of HNRNPL in human myoblasts impairs myotubes fusion.

A. WB analysis showing hnRNP L expression in proliferating and differentiating normal human myoblasts. D = day. B. D5 myotubes that expressed RNAi-inducing shHNRNPL showed decreased hnRNP L protein levels in whole cell lysate (versus normal levels in shLuc control-expressing myotubes). GAPDH: lysate loading. shLuc, anti-luciferase control virus.; shHNRNPL, anti-hnRNP L lentivirus. C. Downregulation of hnRNP L (bottom row) led to a marked reduction in multi-nucleated myotubes at D4 differentiation (red circle: the culture showed residual myoblasts), compared to healthy control myotubes (top). shHNRNPL D4 residual myoblasts can later differentiate into few myotubes that are disorganized (D14, arrow) compared to normal D14 myotubes (top). N= 3 biological replicates, one representative experiment is shown. Scale bars = 50 μm. D. Quantification of myogenic fusion indices between the shLuc and shHNRNPL muscle cohorts from 30% confluent myoblasts to day 14-differentiated myotubes. ****p-value < 0.005 ***p-value < 0.05. E. Immunostaining of shLuc control virus (left), and shHNRNPL virus (right), infected myotubes showed increased caspase-3 activity in shHNRNPL cells. Arrowheads indicate cleaved caspase-3 positive (red) cells. N=3 wells/-cohort (shLuc-, or shHNRNPL-expressing D4-differentiated normal human myotubes at virus MOI 10). Scale bars = 50 μm. F. A Caspase-Glo assay revealed limited apoptosis in shLuc control- D4-differentiated myotubes). In contrast, shHNRNPL myotubes displayed marked apoptosis at MOI 10 and above.

Figure 2.. Morpholino-mediated hnrpnl2 knockdown in zebrafish results in early death and abnormal muscle birefringence.

A. Position of the two non-overlapping antisense translational-blocking morpholinos (MO1/grey, MO2/blue) on partial hnrnpl2 sequences. B. The MO1, or a control (non-targeting) MO, was microinjected into one-cell AB wild-type fish embryos. Phase microscopy (top panel) revealed abnormal morphology in Hnrnpl2 morphants at 4 dpf. Muscle birefringence (bottom panel) showed disrupted dorsal muscle architecture in hnRNP L morphants. The abnormal bent tail and pericardial edema are indicated by a black arrow, and white arrowhead. Scale bars = 500 μm. C. Hnrnpl2 protein expression levels in hnrnpl2 MO1, control MO, and uninjected fish whole body protein lysates. The low MW Hnrnpl2 protein isoform (29 kDa) is sensitive to translation blocking. β-tubulin: loading control. Quantification of depletion of the Hnrnpl2 protein lower band (at 6ng/μl MO). *p-value < 0.05 ****p-value < 0.005. D. The severity of the muscle phenotype induced by MO1, versus MO2, morpholino (at 6 ng/μl) correlates with the degree of reduction in Hnrnpl2 levels. Scale bars = 500 μm. E. Graph showing percent normal-appearing (blue), abnormal-appearing (red) and dead (grey) fish in each experimental cohort.

Figure 3.. Multiple mRNA targets that contribute to DM1 pathology are found downstream of hnRNP L.

A compendium of known human alternative splicing events was assessed to identify sequences that include the hnRNP L CA-rich binding motif (illustrated by a diagram shown on top), and that are expressed in skeletal muscle and/or the heart (details of the bioinformatics analysis are provided in Supplemental Methods). 47 conserved putative targets of hnRNP L were identified, shown as nodes on a STRING functional protein association network graphic. These include eleven genes/transcripts known to be aberrantly spliced in DM1 patients (black circles and listed in the inset). The pathway comprises multiple components of the contractile apparatus and actin cytoskeleton, as well as SAMD4A, a genetic suppressor or of myotonic dystrophy characterized in human myoblasts (red circle). A filled node indicates a known or predicted 3D structure. The edges represent protein-protein associations (the proteins contribute to a shared function, with or without being physically bound). The legend for the edges’ colors is shown.

Figure 4.. HnRNP L interacts with MBNL1 in HEK293, and aggregates in DM1 patient myoblasts.

A. Top: Co-immunoprecipitation (co-IP) of FLAG-tagged hnRNP L (589 aa, or 456 aa, isoform) with HA-tagged-MBNL1. IP performed with HA-beads (MBNL1) and immunoblot with anti-FLAG antibody. Middle: reverse co-IP using FLAG-beads (HNRNP L). Multiple known MBNL1 isoforms were immunoprecipitated. Bottom: Input lysate control immunoblots. B. Human hnRNP L formed abnormal foci in DM1-diseased myoblasts, and partially co-localized with CUG repeats (white arrowheads). Combined fluorescent in situ hybridization (FISH) and immunofluorescence using FITC-labeled oligonucleotide probe (green), hnRNP L antibody (red) and DAPI (blue) showing hnRNP L protein aggregates in DM1 patient myoblasts (DM1A, DM1B and DM1C.). 30–40 nuclei/genotype were analyzed. Scale bar = 10 μm. C. Quantification of hnRNP L localization in the nuclei of normal and DM1-diseased myoblasts (mean ± SD, n= 30–40 nuclei from 3 independent experiments). Equal regions of individual nuclei of control and DM1 myoblasts were quantified. HnRNP L homogeneous distribution in the nucleoplasm significantly decreased in DM1B and DM1C myoblasts, compared to that in normal (control) myoblasts set at 100%. D. Quantification of hnRNP L-positive foci partially co-localized with CUG RNA foci in DM1 diseased myoblasts (10–60% overlap). * p<0.05, ** p<0.01.

Figure 5.. Ascochlorin (ASC) treatment increased hnRNP L and MBNL1 protein levels, and partially rescued abnormal splicing associated with DM1.

A. Diagram illustrating potential rescue of MBNL1 function by targeting both MBNL1 and its partner, hnRNP L. B. Treatment with ASC increased hnRNP L and MBNL1 protein levels in human DM1-diseased myotubes. At day 3 of differentiation, DM1B diseased myotubes cultured in 12 well-plates were treated with either ASC at the indicated concentration, or DMSO vehicle alone. Fresh media (+/− the drug) was added every 3 days. The cultures were stopped at day 13 of differentiation and Western blot analysis was performed using anti-hnRNP L and anti-MBNL1 antibodies. The blots were re-probed with anti-GAPDH and tubulin antibodies. Pools of three wells of cultured myotubes per treatment condition were analyzed. C. RNA splicing analysis of four validated DM1 biomarkers, NFIX, BIN1, INSR and TTN, in normal and DM1 human primary myotubes, treated either with ASC 10⁻⁶M or vehicle alone (V; DMSO). The experimental layout is shown above. ASC treatment fully restored normal splicing of INSR and TTN in DM1 myotubes (arrows). GAPDH: loading control. -RT: internal experimental control. N=3 experiments, results from one representative example are shown.

Figure 6.. Ascochlorin treatment partially rescued abnormal cellular parameters associated with DM1.

A. Microscope field showing human normal and DM1A-diseased myoblasts, and corresponding day 14-differentiated myotubes (representative examples). The differentiation medium contained either ASC (10⁻⁶ M, +) or vehicle alone (−). In the absence of the drug, DM1 myoblasts differentiated into few multinucleated myotubes (defined by at least 3 nuclei/desmin-positive cell). B. Fusion index box-plots of normal and DM1 myoblasts +/− ASC N= 3 biological replicates(5 distinct microscope fields were assessed/replicate). Statistical significance, ns, not significant; ***, p < 0.001. Scale bar = 50 μm. C. FISH (CUG probe) and hnRNP L immunofluorescence in normal human (control) and DM1 myoblasts treated with ASC (10⁻⁶M). Drug treatment increased hnRNP L diffuse nucleoplasmic expression in both control and DM1 myoblasts; and decreased CUG RNA foci in DM1 myoblasts. 30 myoblasts/condition were analyzed. Scale bar = 10 μm. D. ASC treatment increased pan-nucleoplasmic hnRNP L in normal human myoblasts. E. ASC treatment increases pan-nucleoplasmic hnRNP L in DM1C myoblasts and decreased the number of CUG foci. CUG foci were counted in 30 nuclei in ASC-treated, and untreated, DM1C myoblasts (N= 3 replicates). Numbers were normalized to those obtained in untreated DM1C myoblasts. ** p<0.01.