The hnRNPs F and H2 bind to similar sequences to influence gene expression

Abstract

The hnRNPs (heterogeneous nuclear ribonucleoproteins) F and H2 share a similar protein structure. Both have been implicated as regulating polyadenylation, but hnRNP H2 had a positive effect, whereas hnRNP F acted negatively. We therefore carried out side-by-side comparisons of their RNA-binding and in vivo actions. The binding of the CstF2 (64 kDa cleavage stimulatory factor) to SV40 (simian virus 40) late pre-mRNA substrates containing a downstream GRS (guanine-rich sequence) was reduced by hnRNP F, but not by hnRNP H2, in a UV-cross-linking assay. Point mutations of the 14-nt GRS influenced the binding of purified hnRNP F or H2 in parallel. Co-operative binding of the individual proteins to RNA was lost with mutations of the GRS in the G1-5 or G12-14 regions; both regions seem to be necessary for optimal interactions. Using a reporter green fluorescent protein assay with the GRS inserted downstream of the poly(A) (polyadenine) signal, expression in vivo was diminished by a mutant G1-5 sequence which decreased binding of both hnRNPs (SAA20) and was enhanced by a 12-14-nt mutant that showed enhanced hnRNP F or H2 binding (SAA10). Using small interfering RNA, down-regulation of hnRNP H2 levels diminished reporter expression, confirming that hnRNP H2 confers a positive influence; in contrast, decreasing hnRNP F levels had a negligible influence on reporter expression with the intact GRS. A pronounced diminution in reporter expression was seen with the SAA20 mutant for both. Thus the relative levels of hnRNP F and H2 in cells, as well as the target sequences in the downstream GRS on pre-mRNA, influence gene expression.

Figure 1. SVL pre-mRNA based probes used in the present study

The name and length of each RNA probe is indicated on the left. The relative binding affinities to hnRNP F as determined from data summarized in Figure 3 are indicated in the right column. Probe A is wtSVL; GU, GU-rich binding site for CstF-64; GEM, vector pGem sequence. mt indicates the site of mutation.

Figure 2. hnRNP F inhibits CstF-64 binding

Purified hnRNP proteins and/or the RNA-binding domain of CstF-64 were incubated with ³²P-labelled RNA under UV-cross-linking conditions (see the Experimental section) and subjected to UV-irradiation on ice. Following RNase digestion, proteins were separated by SDS/10% PAGE. Dried gels were exposed to film. In the indicated lanes, 50 nM GST-tagged hnRNP H2 or 260 nM GST–CstF-64 RBD were added. (A) Lane 1, 40 nM His-hnRNP F; lane 2, GST-hnRNP H2; lane 3, GST–CstF-64 RBD; lane 4, 40 nM His–hnRNP F plus GST–hnRNP H2; lane 5, GST–hnRNP H2 plus GST–CstF-64 RBD; lane 6, 40 nM His–hnRNP F plus GST–CstF-64 RBD; lane 7, 20 nM His–hnRNP F plus GST–CstF-64 RBD; lane 8, 10 nM His–hnRNP F plus GST–CstF-64 RBD; lane 9, 5 nM His–hnRNP F plus GST–CstF-64 RBD; lane 10, 40 nM His–hnRNP F, GST–CstF-64 RBD and GST–hnRNP H2. (B) SVL or SVL-GEM, probes A or E incubated with purified proteins; lanes 1 and 4, 40 nM His–hnRNP F; lanes 2 and 5, 50 nM GST–hnRNP H2; lanes 3 and 6, 260 nM GST–CstF-64 RBD. The position of the 75 kDa protein is indicated. F, hnRNP F; H, hnRNP H; 64, CstF-64.

Figure 3. hnRNP F binds to the GRS in SVL pre-mRNA in EMSAs

(A) EMSAs were performed with 50 pM [³²P]GTP-labelled SVL pre-mRNAs in 20 μl reaction volumes. Recombinant His-hnRNP F (0, 55, 110 or 220 nM) was used to shift each RNA probe. Samples were run on a 4% (20:0.25) native acrylamide gel containing 0.5% agarose. Stars indicate the positions of hnRNP F-RNA complexes. Cross-hairs show the positions of unshifted RNA probes. (B) EMSAs were performed as described in (A). Samples were run on 8% (20:0.25) native acrylamide gels containing 0.5% agarose. Probe G, 250 pM, was used in the last four lanes. (C) EMSAs were performed with 50 pM probe B and 0.5, 5 or 50 nM specific (S) or non-specific (NS) competitors in the presence (+) or absence (−) of 165 nM His-hnRNP F.

Figure 4. hnRNP F and H2 show similar binding to mutations in the GRS

(A, B) Relative binding of wild-type and mutant GRS oligoribonucleotides whose sequences are shown in Table 1 to hnRNP F (F) and hnRNP H2 (H2) proteins. EMSAs were performed with 250 pM of each probe and 330 nM His–hnRNP F or 210 nM GST–hnRNP H2. Relative binding of each mutant to wild-type GRS was set as 1.0 as shown in the histogram. Results are means±S.E.M. for at least three determinations.

Figure 5. Filter binding assay with hnRNP H2 and hnRNP F

(A) Binding curves of GST–hnRNP H2 to 250 pM wild-type probe B in Figure 1 or probe B having the SAA20 mutation, as determined by filter binding assay. Each reaction was performed three times, and results are means±S.E.M. (B) Binding curves of His–hnRNP F to the same RNA molecules were obtained as described in (A). Each reaction was performed three times, and results are means±S.E.M. (C) Binding was performed with His–hnRNP F and 250 pM RNA probes (50 nt) containing wild-type (w.t.), SAA10 or SAA21 GRS regions. Results are the means of at least two determinations.

Figure 6. Decreasing the binding of hnRNP F or hnRNP H2 to the downstream region of the GFP transcript reduces the GFP expression in vivo

(A) Mean GFP fluorescence intensities corresponding to pEGFP-nopoly(A) (NopolyA), pEGFP-wtSVL (SVLwt), pEGFP-SAA20 (SAA20) and pEGFP-SAA10 (SAA10) transfected HEK-293T cells were quantified by flow cytometry. The DsRed fluorescence from a separate plasmid served as a transfection control for normalization (results not shown). Each transfection was performed at least three times, and results are means±S.E.M. (B) GFP levels in pEGFP-wtSVL (Wt) or pEGFP-SAA20 (SAA20)-transfected A20 or HEK-293T cells were determined by Western blots with anti-GFP (α-GFP) antibody. Neomycin encoded on the GFP–plasmid was used for the protein blot loading control (α-Neomycin). Relative hnRNP H and hnRNP F levels in A20 and HEK-293T extracts were detected using a Western blot assay with anti-(hnRNP H) (H) and anti-(hnRNP F) (F) antibodies.

Figure 7. Effects of down-regulation of hnRNP F and hnRNP H by siRNAs on GFP expression

(A) Western blots performed with anti-hnRNP H1/H2 (α-hnRNP H), anti-hnRNP F (α-hnRNP F) and anti-GAPDH (glyceraldehyde-3-phosphate dehydrogenase; α-GAPDH) antibodies. The lane marked 0 shows mock-transfected cells, F is hnRNP F siRNA, H2 is hnRNP H2 siRNA, and Scr (scrambled) is a negative-control siRNA that does not hybridize any known mRNA. (B, C) Quantification of GFP levels by flow cytometry. (B) shows the data from HEK-293T cells co-transfected with pEGFP-wtSVL, whereas, in (C), the results were obtained from cells transfected with pEGFP-SAA20 and the indicated siRNAs. Results are means±S.E.M.

Figure 8. Alignment of hnRNP F- and hnRNP H-binding sites

Sequences from the indicated genes implicated as binding to hnRNP F or hnRNP H/H2 were taken from the references indicated in the Discussion. The genes are: V1loopcons, consensus for Exon 6D splicing regulatory region of divergent HIV-1 isolates; Cftr-exon9-155G, natural mutation 155G of cystic fibrosis transmembrane regulator Exon 9; RSV, negative regulator of splicing element from rous sarcoma virus; Throidhorm, TRα2 isoform of thyroid hormone receptor gene c-erbAα; C-SRC, intronic splicing enhancer downstream of the N1 5′ splice site of the c-src N1 exon; Tat, a second exonic splicing silencer of tat exon 2 within HIV type I; BETA-TRP, exonic splicing silencer of the rat β-tropomyosin gene; WA19tat, high-affinity hnRNP A1-binding sequence identified by SELEX (systematic evolution of ligands by exponential enrichment); 844 ins68_CBS, 3′ splice site in 884ins68 polymorphism of the cystathionine β-synthase gene. The sequences between nucleotides 1–5 and 12–14 of the GRS are in the boxes.