Binding of hnRNP H and U2AF65 to respective G-codes and a poly-uridine tract collaborate in the N50-5'ss selection of the REST N exon in H69 cells

Abstract

The splicing of the N exon in the pre-mRNA coding for the RE1-silencing transcription factor (REST) results in a truncated protein that modifies the expression pattern of some of its target genes. A weak 3'ss, three alternative 5'ss (N4-, N50-, and N62-5'ss) and a variety of putative target sites for splicing regulatory proteins are found around the N exon; two GGGG codes (G2-G3) and a poly-Uridine tract (N-PU) are found in front of the N50-5'ss. In this work we analyzed some of the regulatory factors and elements involved in the preferred selection of the N50-5'ss (N50 activation) in the small cell lung cancer cell line H69. Wild type and mutant N exon/β-globin minigenes recapitulated N50 exon splicing in H69 cells, and showed that the N-PU and the G2-G3 elements are required for N50 exon splicing. Biochemical and knockdown experiments identified these elements as U2AF65 and hnRNP H targets, respectively, and that they are also required for N50 exon activation. Compared to normal MRC5 cells, and in keeping with N50 exon activation, U2AF65, hnRNP H and other splicing factors were highly expressed in H69 cells. CLIP experiments revealed that hnRNP H RNA-binding occurs first and is a prerequisite for U2AF65 RNA binding, and EMSA and CLIP experiments suggest that U2AF65-RNA recognition displaces hnRNP H and helps to recruit other splicing factors (at least U1 70K) to the N50-5'ss. Our results evidenced novel hnRNP H and U2AF65 functions: respectively, U2AF65-recruiting to a 5'ss in humans and the hnRNP H-displacing function from two juxtaposed GGGG codes.

Figure 1. Splicing regulatory elements around the N exon.

(A) Schematic representation of the REST gene (drawing not to scale) showing the transcription start site (right arrow), introns (thin lines), lengths of relevant introns, constitutive exons coding for the UTR (dashed boxes) and for the REST protein (white boxes), including the alternative N exon (black box); the relative positions of the alternative N4-, N50- and N62-5'ss appear underneath. Thick arrows indicate the positions of the primers used to amplify the mRNA variants by RT-PCR. The main possible mRNAs and proteins expressed in a) non-neuronal cells and b) SCLC H69 cells are depicted also. Splicing of the N50 exon introduces a stop codon. Nomenclature of protein domains: RD, repressor domain; Zn-DBD, zinc-finger DNA-binding domain; PM, Proline-rich motif; and Zn-RD, zinc-finger repressor domain. (B) Genomic DNA from H69 and MRC5 cells was used to amplify 163 bp (between virgules) that include the N exon by PCR and the sequence identity between the cell lines was confirmed. The names of the relevant transcript elements are shown under the sequence. The AG-UG dinucleotides, corresponding to the 3′ss, and the N4-, N50- and N62-5'ss appear in italics; the upstream Py, G1-G3, and N-PU elements are boxed in gray. (C) Unscaled drawings of the RNA molecules used in this work. The top one is coded by the minigene constructs (271 nt upstream, 62 nt of the N exon, and 72 nt downstream); the sequences between virgules correspond to those shown in B. The full-length wild type (flWT) and G2-G3∼N-PU probes are also shown. Mutant versions (without both elements or without N-PU) of the G2-G3∼N-PU probe were made also. The putative RNA binding sites for U2AF65 (triangles), hnRNP H (black diamonds), TIA-1/TIAR (open circles), and nSR100 (white diamonds) were predicted using the Splicing Rainbow software; the symbols indicate their relative positions in the RNA molecules.

Figure 2. N-PU and G2-G3 codes are required for N50 exon activation in H69 cells.

Mutant N-PU and G2-G3 elements negatively affect N50 exon inclusion. Cells were transfected with the wild type (lanes 1 and 2) or mutant N50/β-globin minigenes (lanes 3–5). 24 h post transfection, N exon inclusion between β-globin exons 1 and 2 was monitored by RT-PCR (with primers T7 and SP6). No RT (- RT) PCR control reaction appears in lane 1. The plot below shows the quantification of the spliced and partially spliced mRNAs in the indicated lanes. The absence of either element decreased N50 exon activation, albeit the lack of the G2-G3 rendered more exon skipping and partially spliced products.

Figure 3. U2AF65 and hnRNP H bind to probes containing the N-PU and the G2-G3 codes.

(A) flWT N exon synthetic transcripts containing the three GGGG codes and the N-PU, and NEx from H69 (lane 1) and MRC5 (lane 2) cells were used in UV crosslinking experiments. Enriched p64 and p50 proteins in H69 are indicated with arrowheads. (B) UV crosslinking (lanes 1 and 2) and CLIP (lanes 3–6) assays were carried out with G2-G3∼N-PU synthetic RNAs, NEx from MRC5 (lanes 1, 3 and 5) or H69 cells (lanes 2, 4 and 6), and with antibodies against U2AF65 (lanes 3, and 4) or an unrelated antigen (lanes 5 and 6). Immunoprecipitation (lanes 7 and 8), UV crosslinking (lane 9) and CLIP (lanes 10 and 11) assays were carried out with the same synthetic RNAs and NEx from H69 cells, using antibodies against hnRNP H (lanes 7 and 10, respectively) or an unrelated antigen (lanes 8 and 11, respectively). A low molecular mass unidentified protein cross-reacted with the probe (asterisk).

Figure 4. U2AF65 and hnRNP H are required for N50 exon activation in H69 cells.

(A) U2AF65 knockdown downregulates REST-N50 mRNA and tREST expression in H69 cells. H69 cells were treated with universal control (iUC) or U2AF65 siRNAs (lanes 1 and 2, respectively), (B) Overexpression of U2AF65 in MRC5 cells promoted REST-N50 mRNA and tREST expression. MRC5 cells were transfected with the U2AF65 cDNA (lanes 1) or with empty pcDNA3 vector (lanes 2). 24 h or 36 h post-transfection of siRNAs or cDNAs, respectively, the mRNAs coding for U2AF65, N50-REST, and Actin were monitored by RT-PCR. Protein expression of U2AF65, REST, tREST, and Actin was monitored by western blots as well. (C) Steady state expression profiles of REST mRNA variants (lanes 1–3) and protein isoforms (lanes 4–5) in H69 (lanes 1 and 4) and MRC5 (lanes 2 and 5) cells, respectively. No RT (- RT) PCR control reaction appears in lane 3. (D) hnRNP H knockdown leads to accumulation of unspliced N50/β-globin precursor. H69 cells were treated with iUC (lanes 1, 3 and 4) or hnRNP H siRNAs (lanes 2, 5–8). After 24 h, the cells were transfected with the wild type (lanes 3–5) or mutant N50/β-globin minigenes (lanes 6–8). 24 h post- transfection, N50/β-globin exon 2 splicing was monitored by RT-PCR (with primers SP6 and rtN62). No RT (- RT) PCR control reaction appears in lane 3. hnRNP H knockdown increased the unspliced precursors two-fold. Accumulation of the precursor increased up to 3 times when one or both of the N-PU (black flags) or G2-G3 (white flags) were mutated. Notably, G2-G3 mutants rendered no spliced products, and the N62/2 (*) product was obtained in double transfection experiments only. The PCR products and proteins were quantified densitometrically using the ImageJ software. IntDen % represents the expression change of the indicated protein or mRNA compared to their respective control treatment.

Figure 5. tREST expression correlates with the robust expression of U2AF65, hnRNP H, and other splicing factors in H69 cells.

(A) The expression of nuclear U2AF65, hnRNP H and snRNP U1 70K in H69 (black bars) and MRC5 (white bars) cells was monitored by western blots. Proteins were quantified densitometrically using the ImageJ software. IntDen % represents the expression change of the indicated protein in H69 cells compared to MRC5 cells. (B) Whole cell protein extracts from MRC5 and H69 cell were probed with antibodies against nSR100, TIA-1/TIAR and Actin antibodies. A robust expression of nSR100 and TIAR was detected in H69 cells only.

Figure 6. U2AF65/N-PU and hnRNP H/G2-G3 interplay in the N50 exon recognition.

(A) hnRNP H-RNA recognition is a prerequisite for U2AF65-RNA binding. The wild type RNA was incubated with whole (lane 2), hnRNP H-depleted (lane 3), and unrelated antigen-depleted H69 NEx (lane 4) or beads alone (lane 5). Then CLIP assays were carried out using U2AF65 antibodies and the recovered bound RNA was monitored by RT-PCR and quantified densitometrically. No RT (– RT) control PCR appears in lane 1. In the absence of hnRNP H, 60% less RNA was recovered with the U2AF65 antibodies, and no difference was observed in the absence of an unrelated antigen. (B) Kinetics of U2AF65- and hnRNP H-RNA binding. H69 NEx were incubated with wild type or mutant (G2-G3∼N-PU-less) RNA probes during the indicated times at room temperature. Then, CLIP assays were carried out with U2AF65 or hnRNP H antibodies and the recovered RNAs were amplified by RT-PCR (top panel, electrophoresis of the amplified products) and quantified densitometrically (plot at the bottom). At each time point, IntDen values obtained from the CLIP with the mutant RNA were subtracted from those obtained with wild type RNA in order to discriminate indirect protein-RNA interactions (e.g. mediated by additional factors) and unspecific binding to G-like and PU-like motifs. Significant amounts of hnRNP H-bound RNA are recovered as early as 4 min plummeting at 10 min, coinciding with the onset of U2AF65-bound RNA recovery. (C) TIA-1/TIAR is able to bind to multiple poly-U tracts. Different probes (wild type G2-G3∼N-PU, G2-G3∼mutant N-PU, and 3'ss control) were used in CLIP experiments as in B. Immunoprecipitations were carried out with antibodies against TIA-1/TIAR and U2AF65, and the recovered RNAs were amplified by RT-PCR and quantified. IntDen % represents the expression change of the indicated protein or mRNA compared to their respective control treatment. ND, not detected.

Figure 7. U2AF65/N-PU recognition displaces hnRNP H binding to the G2-G3 motifs in a dose dependent manner.

Supershift experiments were carried out with the G2-G3∼N-PU probe (free probes are in lanes 1 and 5), NEx from MRC5 cells (lanes 2–4, and 6–10), anti-hnRNP H antibodies (lanes 3, 7–10), antibodies against an unrelated antigen (lane 4), and increasing amounts of rU2AF65 (2, 4, and 6 µg; lanes 8–10, respectively). The complexes supershifted by the hnRNP H antibody (a-H cx) migrated back to the original position of the NEx complexes (NEx cx) as the amount of rU2AF65 increased. Probe and factors depicted as in Fig. 1C.

Figure 8. The RS domain of U2AF65 promotes U1 snRNP recruitment to the N50-5'ss.

(A) The G2-G3∼N-PU probe (free probe in lane 1) was used in EMSA experiments with increasing amounts of rU2AF65 (1–3 µg) forming the complexes r65 cx (lanes 2–4, respectively). The same probe was incubated with MRC5 NEx forming the complexes NEx cx (lanes 5–7). NEx cx were supershifted with anti-U1 70K antibodies (a-U1 70k cx) in the presence of 1–3 µg of rU2AF65 (lanes 5–7, respectively). The amount of a-U1 70k cx increased with the amount of rU2AF65. (B) The same probe (free probe, lane 1) was used in EMSA experiments with NEx from H69 cells (lane 2), or with extracts incubated with oligonucleotides targeted to the 5' end of the U1 snRNA and treated with RNase H (U1 k.o.; lane 3), or with RNase H-treated extracts incubated with scrambled oligonucleotides (c.o.; lane 4). Probe and factors depicted as in Fig. 1C. (C) Nickel or protein-A Sepharose (S) beads covered with rSRSF3-(S), rU2AF65, and the SR domain of U2AF65 (rU2AF65-RS) were used in pull-down experiments. U1 70K, TIA-1/TIAR, and hnRNP H (top, middle, and bottom panels, respectively) were monitored by western blots. Both snRNP U1 70K and TIA-1/TIAR were able to interact with U2AF65 and its RS domain (lanes 2, and 3) but not with a different splicing protein, rSRSF3 (lane 1). BSA covered beads were used for blank controls (lane 4); in these lanes protein-A copurified with BSA. No hnRNP H was pulled-down with rU2AF65. The cartoon depicts the possible U1 snRNP and TIA-1/TIAR interactions with U2AF65 in the recognition of a 5'ss.