Exon repression by polypyrimidine tract binding protein

Abstract

Polypyrimidine tract binding protein (PTB) is known to silence the splicing of many alternative exons. However, exons repressed by PTB are affected by other RNA regulatory elements and proteins. This makes it difficult to dissect the structure of the pre-mRNP complexes that silence splicing, and to understand the role of PTB in this process. We determined the minimal requirements for PTB-mediated splicing repression. We find that the minimal sequence for high affinity binding by PTB is relatively large, containing multiple polypyrimidine elements. Analytical ultracentrifugation and proteolysis mapping of RNA cross-links on the PTB protein indicate that most PTB exists as a monomer, and that a polypyrimidine element extends across multiple PTB domains. The high affinity site is bound initially by a PTB monomer and at higher concentrations by additional PTB molecules. Significantly, this site is not sufficient for splicing repression when placed in the 3' splice site of a strong test exon. Efficient repression requires a second binding site within the exon itself or downstream from it. This second site enhances formation of a multimeric PTB complex, even if it does not bind well to PTB on its own. These experiments show that PTB can be sufficient to repress splicing of an otherwise constitutive exon, without binding sites for additional regulatory proteins and without competing with U2AF binding. The minimal complex mediating splicing repression by PTB requires two binding sites bound by an oligomeric PTB complex.

FIGURE 1.

Characterization of a minimal RNA binding site for PTB. (A) Diagrams of the RNA probes derived from N1 3′ splice site that were used in an electrophoretic mobility shift assay (EMSA). (B) EMSA of recombinant PTB binding to a 38-nt c-src N1 exon 3′ splice site (Probe A). (Lane 1) free RNA (50,000 cpm ~10 fmol). (Lanes 2–6) RNA incubated with increasing amounts of PTB in the amounts shown. The PTB/RNA Complexes 1 and 2 are indicated. (C) Boundary determination of Complexes 1 and 2. An RNA probe comprising the c-src N1 exon and part of the upstream intron was labeled at its 3′ or 5′ end, subjected to partial alkaline hydrolysis (lanes 1,2), and incubated with PTB. The PTB/RNA complexes were separated on a native gel as in B, and the RNA was eluted and analyzed on the denaturing gel shown. The 3′ boundaries of the 5 ′-end-labeled RNA (lanes 3,4) and 5′ boundaries of the 3′-end-labeled RNA (lanes 5,6) for the interaction with PTB are indicated by arrows. A partial RNAse T1 digestion of the 5′ end-labeled RNA (lane 7) and small RNA size markers (lane 8) allowed identification of the fragments in each ladder. The fragments are aligned with the sequence at the top. A diagram of the two possible protein interactions for each complex is shown to the right. (D) EMSA of PTB binding to the wild-type 3′ splice site (Probe B) and to this sequence with one (Probe C) or both (Probe D) CUUCUCUCU elements mutated. Note that Complex 2 forms only on Probe B. (E) EMSA of PTB binding to the wild-type N1 3′ splice site (Probe B) and to probes with the upstream (Probe E) or downstream (Probe F) pyrimidine elements mutated or both (Probe G). Note that loss of the two pyrimidine elements nearly eliminates Complex 2 formation.

FIGURE 2.

PTB is monomeric in solution. (A) The equilibrium absorbance gradient of PTB at 20°C and 12,000 rpm is shown as circles along with a nonlinear least-squares single exponential fit to the data. The average molecular weight from the fit was 61 kDa, and the upper panel shows the deviations of the data from the fit. (B) The sedimentation coefficient distribution (uncorrected for diffusion) of PTB determined from sedimentation velocity at 55,000 rpm and 20°C. The g (s) parameter indicates the fraction of material with the sedimentation coefficient S.

FIGURE 3.

Mapping positions of RNA contact on the PTB protein. (A) Flow chart for mapping the PTB/BS7 RNA cross-link. Partial Glu-C proteolysis maps the cross-link to a position C-terminal of D290. Partial and exhaustive CNBr cleavage (C) maps the cross-link to a 102-residue peptide between M 391 and M 493. Complete digestion with pairs of proteases further delineates the cross-link to the peptide NFQNIFPPSATLHLSNIPPSVSE, encompassing β strand 1 of RRM 4. (B) Partial Glu-C proteolysis of PTB cross-linked to α-³²PUTP-labeled BS7 RNA. After 254 nm UV cross-linking and RNAse A + T1 treatment, PTB was subject to Glu-C digestion. Equivalent digestion reactions were visualized by Coomassie staining (left panel), autoradiography (second panel from the left), and Western blot using antibodies against the C terminus (second panel from the right) or the N terminus (right panel) of PTB. The asterisk indicates the primary labeled (28 kDa) cleavage product from Glu-C digestion. (C) Comparison of the CNBr fragments of PTB after cross-linking to different RNA templates and by different methods. (Left lane) PTB cross-linked to a 5′ end-labeled CUCUCU oligo at 254 nm. (Middle lane) PTB cross-linked to uniformly α-³²PUTP-labeled N1–3′ splice site RNA at 254 nm. (Right lane) PTB cross-linked to 4-thio UTP substituted 3′ splice site RNA at 366 nm UV. Both of the N1–3′ SS RNA samples were treated with RNAse prior to the CNBr digestion. The asterisk indicates the major cross-linking product seen with all methods. (D) PTB cross-linking to a single CUUCUCUCUCG repeat RNA containing single 4-thioU substitutions. (Top left) The efficiency of PTB cross-linking to the CUUCUCUCUCG oligo at 366 nm is similar regardless of the position of the 4-thioU substitution. (Bottom left) Exhaustive CNBr cleavage of the cross-linked proteins shown above. Asterisks indicate the major cross-linked products. Note that these CNBr fragments exhibit slower gel mobility than those in C because they were not digested with RNAse prior to separation. (Right panel) Immunoprecipitation of the PTB CNBr fragments using N- or C-terminal antibodies. The 4-thioU substitution site is indicated at the top. CNBr cleaved protein from each reaction (Load) was subjected to immunoprecipitation with preimmune serum or with antibody raised to the N- or C-terminal 15-residue peptide of PTB. Note that only the smallest peptide is bound by the C-terminal antibody. (E) Proteolysis sites in PTB. Residues subject to specific cleavage are shown in color. Predicted products are shown below.

FIGURE 4.

Two PTB-binding elements are needed to repress splicing of a reporter exon. (A) The structure of DUP-175 minigenes containing various sequences from the src N1 3′ splice site (175 indicates the number of the nucleotides in the reporter exon). The solid thin line represents β-globin intron sequences and the solid thick bar indicates src N1 3′ splice site sequences. The different src sequences added to each construct are indicated to the right side. The 3′ AG and the 5′ GU dinucleotides at the reporter exon splice sites are indicated. Bp represents the β-globin branch point sequence. Pyrimidine deletions in the N1 3′ splice site are indicated by inverted triangles. (B) Primer extension analysis of the RNA isolated from HEK-293 cells after transfection of each minigene. The position of the primer extension products for exon inclusion and exclusion are shown on the left. The name of each construct is indicated at the top. A truncated version of DUP4-1 (DUP4-1S; see Materials and Methods) was used as a control for transfection efficiency. (C) Each primer extension product was quantified, using a Molecular Dynamics Typhoon PhosphorImager and ImageQuant software, and the percent exon exclusion was calculated (exon excluded product/exon excluded product plus exon included product). These results for the gel shown were plotted in the histogram. (D) DUP-51 minigene constructs (with a 51-nt reporter exon) were used to examine PTB-mediated splicing repression on an inefficiently spliced exon. The maps of the constructs are shown as in A. The transfection, primer extension, and quantification were performed as in B.

FIGURE 5.

Splicing repression by CU elements is highly dependent on PTB expression. The DUP 175 and DS9 reporter constructs were transiently expressed in HeLa cells where PTB expression was knocked down with an siRNA and in similar cells expressing a complementing FLAG-PTB. Reporters were transfected with buffer (lanes 1,2), PTB siRNA plus empty pcDNA vector (lanes 3,4), and PTB siRNA plus Flag-PTB expression vector (lanes 5,6). The top panel shows RT-PCR analysis of splicing after transient expression of the constructs indicated with the products diagrammed on the left. Western blot analyses using BB7 antibody (anti PTB), anti-Flag tag antibody (anti-FLAG-PTB), and anti-GAPDH confirmed the PTB expression level. These are shown below, aligned with the RT-PCR panel. Quantification of the splicing is shown in the histogram plotted as percent exon exclusion (n = 3). Calculation of exon exclusion is as in Figure 4C ▶.

FIGURE 6.

PTB binding elements are sufficient to repress splicing in vitro. (A) β-Globin transcripts with or without PTB binding elements are diagrammed at the lower left. These are equivalent to those in Figure 4D ▶ except that they have a T7 promoter in place of the CMV promoter and are truncated within the downstream intron. This allows the assay of the upstream intron A or A1. RNA from these clones was incubated in an in vitro splicing reaction containing HeLa or WERI nuclear extract and resolved on the gel. The positions of two lariats, the unspliced precursor, the spliced product, and the 5′ exon intermediate for each transcript are diagrammed to the left of the gel. The positions of intron A, which is the product of DUP-51 and DS4-51, and of intron A1, which is the product of the DS1–51 and DS5–51, differ due to the different 3′ splice sites in these clones. Similarly the mRNAs for DUP-51 and DS1–51 (235 nt) are shorter than the mRNAs for DS4–51 and DS5–51 (271 nt) due to the downstream PTB sites in the latter RNAs. The double arrow head indicates the position of the intron A1 lariat and the DS5–51 mRNA just below it. Quantification of the spliced mRNA product relative to the DUP51 control is shown in the histogram below. (B) Added PTB eliminates splicing of CU element carrying RNA. The DUP-51 and DS5–51 transcripts were added to splicing reactions containing HeLa (lanes 1,3,5,7) or WERI nuclear extract (lanes 2,4,6,8). In some reactions recombinant PTB (80 ng, 50 nM final concentration) was also added (lanes 3,4,7, 8). Note that PTB strongly inhibits the DS5–51 splicing, but not DUP51.

FIGURE 7.

Weak downstream PTB sites stimulate Complex 2 formation and splicing repression. (A) Probe A contains the N1 exon 3′ splice site. Probe H contains this same sequence, as well as the N1 exon and the weak downstream PTB binding sites. Probe I is the same as Probe H except that two of the CU elements in the downstream intron are mutated as shown below. EMSA was performed on each probe with increasing amounts of PTB protein (indicated at the top). Note that Complex 2 (labeled at the right) forms on Probe H at a much lower PTB concentration than on the other two probes. (B) Primer extension analysis of RNA from HEK-293 cells after transient expression of the minigenes indicated. The position of the spliced products is shown on the left. DS14 and DS15 contain one and two downstream CUUCUCUCU elements, respectively (diagrammed at the top). Each primer extension product was quantified as in Figure 4C ▶ and plotted on the histogram shown.

FIGURE 8.

Models for the multimeric PTB complexes that repress splicing. In Complex 1 (top) one PTB molecule contacts a single CUU CUCUCU RNA element with both RRM 1 and 4. RRM 3 may contact additional RNA in that vicinity or not. This complex does not inhibit splicing. Complex 2 contains two PTB monomers. One is bound at the upstream high affinity site as in Complex 1, using RRMs 1 and 4. The second PTB engages the downstream CU elements, presumably also using RRMs 1 and 4. This PTB also makes a lower affinity contact with the upstream site, perhaps with RRMs 2 or 3. Different arrangements of CU elements may lead to different orientations of the PTB protein such as in Complex 2b. There are other feasible models than those shown. Note that there are likely dimerization interactions between the two PTB monomers that are not shown. These may occur between the RRM 2 domains as proposed (Pérez et al. 1997b).