Antagonistic regulation of alpha-actinin alternative splicing by CELF proteins and polypyrimidine tract binding protein

Abstract

The alpha-actinin gene has a pair of alternatively spliced exons. The smooth muscle (SM) exon is repressed in most cell types by polypyrimidine tract binding protein (PTB). CELF (CUG-BP and ETR3-like factors) family proteins, splicing regulators whose activities are altered in myotonic dystrophy, were found to coordinately regulate selection of the two alpha-actinin exons. CUG-BP and ETR3 activated the SM exon, and along with CELF4 they were also able to repress splicing of the NM (nonmuscle) exon both in vivo and in vitro. Activation of SM exon splicing was associated with displacement of PTB from the polypyrimidine tract by binding of CUG-BP at adjacent sites. Our data provides direct evidence for the activity of CELF proteins as both activators and repressors of splicing within a single-model system of alternative splicing, and suggests a model whereby alpha-actinin alternative splicing is regulated by synergistic and antagonistic interactions between members of the CELF and PTB families.

FIGURE 1.

(A) Organization and alternative splicing of α-actinin exons. The NM (nonmuscle) and SM (smooth muscle) exons of α-actinin gene are not usually spliced together because of the close proximity of the SM exon branch point and pyrimidine tract (open circle and rectangle) to the 5′ splice site of the NM exon. The SM exon is repressed in most NM cells, and this repression involves multiple PTB-binding sites (UCUU motifs indicated by the vertical lines) between the exon and its upstream branch point. In brain, the major isoform contains both NM and SM exons (Kremerskothen et al. 2002), and direct splicing of EF1a to EF2 is also observed in differentiated SM cells (C. Gooding and C.W.J. Smith, unpubl. observations). Back circles indicate clusters of CUG motifs, arbitrarily defined as three or more CUGs within 18 nt. Taking into account the base composition of the intron, there are nearly 3× the number of CUG motifs as expected randomly. Black diamonds represent clusters of UG motifs, which are also potential high affinity sites for CELF proteins (Takahashi et al. 2000). (B) CELF proteins share a domain structure with three RRMs and a divergent linker region of unknown function between RRM2 and RRM3.

FIGURE 2.

Switching of actinin splicing by overexpression of CELF proteins. (A) Western blot analysis for overexpressed CELF proteins in L cells transiently transfected with expression vectors for ETR3, CUG-BP, CELF4, or mock-transfected (MOCK). Proteins were detected using AntiXpress-HRP conjugated antibody (Invitrogen) against the N-terminal AntiXpress tag in the expression plasmid. (B–D) L cells were transiently transfected with 0.2 μg of the pA (B), pNM (C) or pSM (D) α-actinin reporters together with 0.8 μg of pGEM4Z (4Z), ETR3, CUG-BP, CELF4, or β-gal expression constructs. In the right-hand section of D, 0.1 μg of expression plasmid for PTB4 was also cotransfected. RNA was harvested after 48 h, and RT-PCR analysis was performed to determine the extent of NM exon skipping and SM exon inclusion. In MOCK lanes, no DNA was transfected into the cells. Values below each lane give the percentage of EF1a-EF2 or EF1a-SM-EF2 splicing and represent mean ±SD for three experiments, except in the experiment with PTB4 cotransfection (D, right panel) where they represent the values for the experiment shown. In each case, the percentage of each splice product was calculated as a total of all (two or three) possible spliced products in the lane. The data show that ETR3 and CUG-BP promote SM exon inclusion and NM exon skipping, whereas CELF4 causes skipping of both NM and SM exons.

FIGURE 3.

(A) Coomassie-stained 12.5% SDS-PAGE of purified N-terminally His-tagged CELF proteins; the C-terminal intein/chitin-binding protein tag was cleaved away during purification. The sizes of the lane M protein markers (in kD) are indicated on the left. (B) Recombinant CELF proteins have no effect upon splicing of an unregulated control RNA. The indicated amounts of full-length CELF proteins were added to 10-μL splicing reactions containing the GC + DX pre-mRNA (Scadden and Smith 1995). Splicing precursor, intermediates, and products are indicated to the right. None of the CELF proteins had an effect upon GC + DX splicing.

FIGURE 4.

CELF proteins inhibit NM exon splicing in vitro. (A) Schematic representation of EF1a-NM RNA, with the NM branch point and pyrimidine tract indicated by the circle and rectangle. The sequence from the NM 3′ splice site to 296 nt upstream is shown with the branch point (indicated by the Å) 191 nt upstream. CUG and UG repeats are shown in bold, and the single UCUU motif is bold underlined. (B) ³²P-labeled EF1a-NM α-actinin transcript was spliced in HeLa nuclear extracts for 2 h in the presence of 0–6 μM CELF proteins or 0–4 μM of the unrelated RNA-binding protein UNR (Hunt et al. 1999). RNA species were resolved on 6% polyacrylamide gels. The initial transcripts, splicing intermediates, and products are indicated next to the gels. ETR3, CUG-BP, and CELF4 inhibited NM exon splicing.

FIGURE 5.

ETR3 and CUG-BP activate SM exon splicing by displacing PTB. (A) Sequence of the actinin NM-SM in vitro splicing substrate. A spacer element (lowercase) inserted between the NM 5′ splice site and SM branch point relieves the steric interference that usually prevents NM-SM splicing. The major branch point is Å. CUG motifs are shown in bold and UCUU motifs (optimal PTB-binding sites; Perez et al. 1997) are bold underlined. The diamond above the cartoon and sequence indicates the position of the junction in the ligated RNA in Figure 6A ▶, and the star indicates the position at which the NM1–4 transcripts were truncated in Figure 7D ▶. (B) ³²P-labeled full-length α-actinin transcript (NM-SM) was spliced for 3 h in HeLa nuclear extract in the presence of increasing concentrations of CELF proteins and analyzed by denaturing 12% PAGE. Splicing precursor, intermediates, and products are indicated to the right. (C) Identical reactions to those in B were incubated for 30 min before UV cross-linking was carried out. Cross-linked proteins were analyzed by SDS PAGE and autoradiography. Cross-linked PTB is indicated by the arrow and CELF proteins by the asterisks. Lane numbers in C are equivalent to those in B, with the omission of lane 1, which contained RNA alone.

FIGURE 6.

Cross-linking of PTB and CUG-BP to the polypyrimidine tract in patch-labeled NM-SM RNA. (A) Patch-labeled NM-SM RNA was prepared using radiolabeled 5′ end RNA and trace-labeled 3′ end RNA. The junction between the labeled and unlabeled fragments is indicated by the diamond (see also Fig. 5A ▶ for junction sequence). (B) Body labeled and patch-labeled NM-SM RNA was incubated in nuclear extract for 3 h in the presence or absence of 3 μM CUG-BP. Lanes T: unprocessed transcript. (C) Cross-linking of body-labeled (lanes 1,2) and patch-labeled (lanes 3,4) NM-SM RNA after 30 min incubation in HeLa nuclear extract alone (lanes 1,3) or supplemented with 3 μM CUG-BP. (Lanes 5,6) Controls carried out with the trace-labeled 3′ RNA used to synthesize the patch-labeled RNA. The reduction in PTB cross-linking upon CUG-BP addition was 36% with the body-labeled RNA (lanes 1,2) and 55% for the patch labeled RNA (lanes 3,4).

FIGURE 7.

Antagonistic binding of PTB and CUG-BP to sites at the 3′ end of the polypyrimidine tract. (A) The sequence of RNA probes 1–4. Lowercase letters represent the partial overlap between adjacent probes. The branch point is represented by Å, CUG motifs are bold, and UCUU motifs are bold underlined. (B) Probes 1–4 were UV cross-linked to recombinant PTB and CUG-BP. In each case the strongest cross-link was seen to RNA 3, although PTB also cross-linked weakly to RNA 4 and CUG-BP to RNA 1. (C) UV cross-linking of HeLa nuclear extract containing increasing concentrations of CUG-BP to wild-type RNA 3 (WT) and RNA 3 with point mutations in the UCUU motifs (ΔPTB) or adjacent CUG motifs (ΔCUG). Mutated nucleotides are indicated in lowercase, CUG motifs in bold, and UCUU motifs in bold underline. Mutation of the CUG motifs inhibits CUG-BP cross-linking and prevents competition of PTB binding, whereas mutation of the UCUU motifs decreased PTB cross-linking with no effect upon CUG-BP. (D) Wild-type (WT) or ΔCUG NM1–4 RNA was incubated in HeLa nuclear extract for 25 min in the presence of increasing concentrations of CUG-BP (0, 0.75, 1.5, 3 μM, lanes 1–4 and 5–8, respectively). Splicing activation by CUG-BP was approximately twofold greater with the wild-type NM1–4 RNA than with the ΔCUG mutant (based upon phosphorimager analysis of intensity of the 5′ exon and precursor bands).