Neuron-specific repression of alternative splicing by the conserved CELF protein UNC-75 in Caenorhabditis elegans

Abstract

Tissue-regulated alternative exons are dictated by the interplay between cis-elements and trans-regulatory factors such as RNA-binding proteins (RBPs). Despite extensive research on splicing regulation, the full repertoire of these cis and trans features and their evolutionary dynamics across species are yet to be fully characterized. Members of the CUG-binding protein and ETR-like family (CELF) of RBPs are known to play a key role in the regulation of tissue-biased splicing patterns, and when mutated, these proteins have been implicated in a number of neurological and muscular disorders. In this study, we sought to characterize specific mechanisms that drive tissue-specific splicing in vivo of a model switch-like exon regulated by the neuronal-enriched CELF ortholog in Caenorhabditis elegans, UNC-75. Using sequence alignments, we identified deeply conserved intronic UNC-75 binding motifs overlapping the 5' splice site and upstream of the 3' splice site, flanking a strongly neural-repressed alternative exon in the Zonula Occludens gene zoo-1. We confirmed that loss of UNC-75 or mutations in either of these cis-elements lead to substantial de-repression of the alternative exon in neurons. Moreover, mis-expression of UNC-75 in muscle cells is sufficient to induce the neuron-like robust skipping of this alternative exon. Lastly, we demonstrate that overlapping an UNC-75 motif within a heterologous 5' splice site leads to increased skipping of the adjacent alternative exon in an unrelated splicing event. Together, we have demonstrated that a specific configuration and combination of cis elements bound by this important family of RBPs can achieve robust splicing outcomes in vivo.

Keywords: C. elegans; CELF proteins; RNA binding proteins; RNA processing; UNC-75; WormBase; alternative splicing; gene expression.

Fig. 1.

A switch-like alternative exon in the zonula occludens ortholog zoo-1 is strongly skipped in neurons. a) Schematic of two-color splicing reporter. Reporters include the alternative exon (exon 2 in diagram) and 2 flanking introns (black lines) and exons (exons 1 and 3 in diagram). One additional nucleotide is inserted into the alternative exon, thus shifting the reading frame when included to generate mCherry, whereas when skipped the reporters generate GFP in an alternate reading frame. These reporters also contain two 2A peptide linkers and 4 NLS signals, allowing fluorescent proteins to be uncoupled from the rest of the translated peptides and localized to the nucleus. b) Fluorescence images of zoo-1 exon 9 two-color reporters expressed in neurons and muscle cells in the same animal. Representative neuronal nuclei and muscle cell nuclei are labeled with arrowheads in GFP and mCherry panels, respectively. c and d) Representative RT-PCR c) and densitometric measurements d) of zoo-1 exon 9 splicing patterns in neurons. c) The diagram shows where PCR primers anneal on the reporter and exon-included and skipped isoforms are labeled on gel. d) Densitometric measurements of RT-PCRs monitoring zoo1 exon 9 splicing in vivo and PSI values calculated from tissue-specific TRAP-seq data sets from Koterniak et al. (2020). n = 3 replicates per sample, and mean PSI ± 1 SD is plotted. P-values were calculated from a Student’s t-test.

Fig. 2.

UNC-75 is necessary in neurons and sufficient in muscle cells to elicit robust exon skipping. a) Top: schematic of UNC-75 gene architecture highlighting the different isoforms and 3 RRM domains (purple, green, and black). Exon 8 is alternatively spliced, thereby endogenously generating a 3 RRM and 2 RRM isoform. Bottom: schematic of GFP::2A peptide::unc-75 cDNA::tbb-2 3′ UTR transgene used in c and d). b–d) Representative RT-PCR and densitometric measurements of zoo-1 exon 9 splicing reporter (top gels) or RT-PCR of UNC-75 transgenes (bottom gels) in neurons or muscle cells as labeled. b) Reporter splicing patterns are measured in wild-type worms or unc-75 null mutants. c) and d) Reporter splicing patterns are compared between wild-type animals and animals overexpressing the UNC-75 3 RRM-containing isoform c) or 2 RRM-containing isoform d) in neurons or muscle cells. n = 3 replicates for each data point, while n = 2 replicates for data points in c) and d). Mean PSI ± 1 SD is plotted, and P-values are calculated from Student's t-test.

Fig. 3.

Conserved intronic UNC-75 binding sites flank the 3′ and 5′ splice sites surrounding zoo-1 alternative exon 9. zoo-1 ortholog sequence alignments were obtained from 17 Caenorhabditis species in the elegans super group. Sixty nucleotides upstream of exon 9 (including the 3′ splice site) and 12 nucleotides downstream of the 5′ splice site are shown. Canonical splice sites are highlighted in yellow, while sequences matching the UNC-75 consensus motif are highlighted in light blue. TT dinucleotides in bold and underlined are the nucleotides mutated (to CC) for regions 1, 2, 3A, and 3B, as tested in Fig. 4. The UNC-75 consensus binding motif identified by RNACompete is displayed (bottom left) for reference (Norris et al. 2014). Bottom right: schematic of 3 potential cis-regulatory elements (light blue boxes; regions 1, 2, and 3A/3B as referenced in text) identified by multiple species alignments and tested in Fig. 4. Species abbreviations: kama, kamaaina; oiwi, oiwi; trop, tropicalis; bren, brenneri; inop, inopinata; doug, doughertyi; late, latens; rema, remanei; brig, briggsae; nigo—nigoni; trib, tribulationis; zanz, Zanzibari; eleg, elegans.

Fig. 4.

UNC-75 consensus UG-rich elements in intronic regions flanking both sides of zoo-1 exon 9 are required for exon skipping in neurons. a and b) Top: schematics of mutagenesis experiments disrupting 2 UNC-75 consensus sequences immediately adjacent to the 5′ splice site flanking exon 9 and 2 regions upstream of exon 9. Pink labeled boxes denote regions mutated as shown in Fig. 3. Bottom: representative RT-PCR and densitometric measurements assessing zoo-1 exon 9 reporter splicing patterns in neurons and muscle cells, with each cis-element mutation labeled accordingly. n = 3 replicates for each data point. Mean PSI ± 1 SD is plotted, and P-values are calculated from Student's t-test.

Fig. 5.

5′ splice site-adjacent UNC-75 consensus sequences are sufficient to increase exon skipping in neurons. a) Schematic of unc-16 alternative exon 16 mutagenesis experiment, displaying the native reporter, a reporter with the intronic UNC-75-binding motif mutated (UGUUGUG to UGCCGUG; motif mut), and a series of reporters shifting the endogenous UNC-75 consensus sequence closer to the 5′ splice site (from positions +7 to +1) while preserving the overall intron size, endogenous flanking sequences, and maintaining the position of the neighboring EXC-7-binding site. b) Representative RT-PCR, and densitometric measurements of the unc-16 reporters. c) Representative RT-PCR, and densitometric measurements of the native and the +1 motif mutant unc-16 reporters expressed in wild-type or unc-75 null mutant genetic backgrounds. Asterisk denotes likely heteroduplex species. n = 3 replicates for each data point. Mean PSI ± 1 SD is plotted, and P-values are calculated from student's t-test.

Fig. 6.

Model for UNC-75 mediated switch-like regulation of zoo-1 exon 9 in neurons and nonneuronal cells. a) Schematic of proposed mechanisms of zoo-1 exon 9 splicing in neurons. In model 1, binding of UNC-75 to upstream and downstream intronic regions prevents the binding and/or recruitment of the U1 snRNP, U2AF complex (UAF-1/2) and the U2 snRNP. In this model, multiple independent binding events create a robust repression of exon usage in neurons. In model 2, which may not be mutually exclusive, one or more UNC-75 proteins may interact with both upstream and downstream intronic-binding sites to create a looped conformation. b) Schematic of proposed mechanism of zoo-1 exon 9 splicing in muscle cells, where in the absence of UNC-75 the alternative exon is highly included, likely through an exon definition mechanism. In all diagrams, yet to be identified cofactors are included (gray ovals), which likely facilitate the switch-like behavior in each cell type.