Kdown of hnRNP H and F. Western blot (left) and corresponding densitometric analysis (middle) demonstrating the actual silencing of hnRNP H and F proteins in RNAi experiments. (right) Relative expression levels of wild-type and pseudoexon-containing transcripts by qRT-PCR. The ratio between the two isoforms in samples silenced for either hnRNP F or H was 25033180 also calculated. (B) Transient overexpression of hnRNP F. (left) GeneMapper windows displaying fluorescence peaks corresponding to RT-PCR products obtained from the cDNA of cells transfected with constructs expressing the M minigene with or without hnRNP F overexpression. The fluorescence peak areas were measured by the GeneMapper v4.0 software. The X-axis represents data points (size 4 IBP standard peaks are also indicated) and the Y-axis represents fluorescence units. (right) Histograms represent the relative amount of transcripts including or skipping the pseudoexon, as assessed by calculating the ratio of the corresponding fluorescence peak areas (setting the 
sum of all peaks as 100 ). Bars represent mean 6 SD of 3 independent experiments, each performed in triplicate. The results were analyzed by unpaired t-test (*P,0.05; **P,0.01; ***P,0.001). doi:10.1371/journal.pone.0059333.gand Western blot assays, showing significantly lower levels of both endogenous hnRNP F mRNA and protein in HepG2 compared to HeLa cells (Figure S4). Contrary to what was observed for delG2 mutants, all single or combined deletions of G-runs outside the 25-bp SC 1 chemical information region resulted in a marked increase of pseudoexon splicing in both cell types, suggesting that G1 and G3 normally act as repressor elements (Figure 5). As the deletion of the G2 element alone does not completely recapitulate the effect of the ablation of the entire 25-bp region, and considering that hnRNP F has three RNA-recognizing motifs arrayed in the same spacing that can bind to extended purine-rich elements [23], we produced an additional deleted construct (pTFGG-M-del8) lacking the very last 8-bp purine-rich sequence of the 25-bp region. This mutant was transfected in both HeLa and HepG2 cells again showing 1081537 a cell-type specific response (Figure 5). In particular, in HepG2 cells the ablation of the 8-bp element ledto a significantly lower inclusion of the pseudoexon (from 77 to 51 of total FGG transcripts). Taken together, these results demonstrate that G-runs with opposite functions contribute in determining the levels of pseudoexon inclusion in FGG transcript, only in the presence of the IVS6-320A.T mutation.DiscussionDetailed knowledge on the structure of most vertebrate genes has highlighted the presence of a large number of pseudoexon sequences that are physiologically silenced by intrinsically defective splice sites [24], by the presence of silencer elements [25,26,27], or by the formation of inhibiting RNA secondary structures [28]. Even though pseudoexons are expected 
to be low in ESEs, the high degeneration of splicing enhancer motifs and their relative abundance also within introns [5] suggest that pseudoexons probably contain also a number of enhancer motifs,G-runs Regulating FGG Pseudoexon InclusionG-runs Regulating FGG Pseudoexon InclusionFigure 3. In-silico prediction of an ESE-enriched 25-bp sequence and identification of the interacting proteins. (A) Schematic representation of the minigene construct containing the 75-bp FGG pseudoexon activated by the IVS6-320A.T mutation (indicated by an arrow). (bottom) ESE elements predicted by the.Kdown of hnRNP H and F. Western blot (left) and corresponding densitometric analysis (middle) demonstrating the actual silencing of hnRNP H and F proteins in RNAi experiments. (right) Relative expression levels of wild-type and pseudoexon-containing transcripts by qRT-PCR. The ratio between the two isoforms in samples silenced for either hnRNP F or H was 25033180 also calculated. (B) Transient overexpression of hnRNP F. (left) GeneMapper windows displaying fluorescence peaks corresponding to RT-PCR products obtained from the cDNA of cells transfected with constructs expressing the M minigene with or without hnRNP F overexpression. The fluorescence peak areas were measured by the GeneMapper v4.0 software. The X-axis represents data points (size standard peaks are also indicated) and the Y-axis represents fluorescence units. (right) Histograms represent the relative amount of transcripts including or skipping the pseudoexon, as assessed by calculating the ratio of the corresponding fluorescence peak areas (setting the sum of all peaks as 100 ). Bars represent mean 6 SD of 3 independent experiments, each performed in triplicate. The results were analyzed by unpaired t-test (*P,0.05; **P,0.01; ***P,0.001). doi:10.1371/journal.pone.0059333.gand Western blot assays, showing significantly lower levels of both endogenous hnRNP F mRNA and protein in HepG2 compared to HeLa cells (Figure S4). Contrary to what was observed for delG2 mutants, all single or combined deletions of G-runs outside the 25-bp region resulted in a marked increase of pseudoexon splicing in both cell types, suggesting that G1 and G3 normally act as repressor elements (Figure 5). As the deletion of the G2 element alone does not completely recapitulate the effect of the ablation of the entire 25-bp region, and considering that hnRNP F has three RNA-recognizing motifs arrayed in the same spacing that can bind to extended purine-rich elements [23], we produced an additional deleted construct (pTFGG-M-del8) lacking the very last 8-bp purine-rich sequence of the 25-bp region. This mutant was transfected in both HeLa and HepG2 cells again showing 1081537 a cell-type specific response (Figure 5). In particular, in HepG2 cells the ablation of the 8-bp element ledto a significantly lower inclusion of the pseudoexon (from 77 to 51 of total FGG transcripts). Taken together, these results demonstrate that G-runs with opposite functions contribute in determining the levels of pseudoexon inclusion in FGG transcript, only in the presence of the IVS6-320A.T mutation.DiscussionDetailed knowledge on the structure of most vertebrate genes has highlighted the presence of a large number of pseudoexon sequences that are physiologically silenced by intrinsically defective splice sites [24], by the presence of silencer elements [25,26,27], or by the formation of inhibiting RNA secondary structures [28]. Even though pseudoexons are expected to be low in ESEs, the high degeneration of splicing enhancer motifs and their relative abundance also within introns [5] suggest that pseudoexons probably contain also a number of enhancer motifs,G-runs Regulating FGG Pseudoexon InclusionG-runs Regulating FGG Pseudoexon InclusionFigure 3. In-silico prediction of an ESE-enriched 25-bp sequence and identification of the interacting proteins. (A) Schematic representation of the minigene construct containing the 75-bp FGG pseudoexon activated by the IVS6-320A.T mutation (indicated by an arrow). (bottom) ESE elements predicted by the.
