BRCA1 exon 11 a model of long exon splicing regulation

Abstract

BRCA1 exon 11 is one of the biggest human exons, spanning 3426 bases. This gene is potentially involved in DNA repair as well as cell growth and cell cycle control. Exon 11 is regulated at the splicing level producing three main different combinations of BRCA1 mature transcripts; one including the whole of exon 11 (full isoform), one skipping the entire exon (D11 isoform), and one including only 117 base pairs of exon 11 (D11q isoform). Using minigene and deletion analyses, we have previously described important splicing regulatory sequences located at the beginning of this exon (5′ end). We have now found additional important sequences located at its 3′ end. In particular, we describe the presence of a strong splicing enhancer adjacent to the downstream 5′ splice site, which minimizes competition from an upstream 5′ splice site and so ensures long exon inclusion. Analyses of the proteins binding these RNA sequences have revealed that Tra2beta and hnRNP L are involved in the regulation of BRCA1 exon 11 by influencing the recognition of donor sites. Interestingly, BRCA1 exon 11 carrying deletion of the regulatory sequences bound by these factors also showed unexpected responses to up- or downregulation of these regulatory proteins, suggesting that they can also bind elsewhere in this large exon and elicit different effects on its recognition.

 

The identification of sequences and proteins relevant for the regulation of BRCA1 exon 11 now provides better knowledge on how this exon is recognized and may represent an important step toward understanding how large exons are regulated.

Introduction

Splicing is a nuclear mechanism of RNA maturation, which determines the excision of selected sequences (introns) from the pre-mRNA.¹ This mechanism is regulated at several levels that include basic biochemical processes such as transcription speed and binding of specific regulatory proteins to selected sequences to external cellular stimuli and epigenetic effects.^2,3 At the basic level, the most important sequences are the acceptor and donor splice sites (localized at the intron/exon and exon/intron junction, respectively) and the branch point (normally positioned 20–80 nucleotides upstream the acceptor site). In addition, enhancer and silencer sequences are usually found in proximity of the splice sites and bind a huge number of auxiliary factors that are important for the fine tuning of this process.¹ In fact, during pre-mRNA processing it is the concerted action of all these basic sequences and factors that is needed for efficient spliceosome recognition. The spliceosome is a very large RNA–protein complex essential for the recognition of the splice sites and the branch point that uses the information contained in these sequences in order to choose whether to include a particular exon or not in the mature mRNA transcript.¹ This alternative exon inclusion/skipping is a process known as alternative splicing. Alternative splicing, therefore, allows the production of different mature RNA isoforms from the same gene. From the functional point of view, inclusion or exclusion of specific exons in these alternative isoforms may result in downregulation of gene function (production of non-functional products), production of a partially functional protein, or production of proteins with different/antagonistic function. Because this mechanism is determined by the level/presence of splicing regulatory proteins, fluctuation in the level of those proteins often determines variation in specific gene isoforms, and therefore, alteration of their functionality.

In the cancer research field, the mRNA splicing profile of BRCA1 has become an increasingly important research topic, as the connections between altered splicing profiles of this gene and cancer progression have become evident.^4-6 In particular, mutations in this gene that alter splicing and lead to the expression of a particular splicing isoform have begun to be identified in selected cases of hereditary breast cancer.^7-9 However, production of splicing isoforms, per se, is not necessarily an indication of disease. In fact, preliminary studies from our lab have identified the expression of BRCA1 splicing isoforms in MCF-7 cells as well as non-transformed human mammary epithelial cells that also lack BRCA1 sequence changes.¹⁰ For this reason, and in order to better define the relationship between BRCA1 alternative splicing isoforms and cancer progression it is important to understand the proteins and sequences involved in this process.

Among all the alternatively spliced exons, BRCA1 exon 11 stands out from the rest because of its unusual length (more than 3 kb) and because there are conflicting studies regarding the effects it has on mRNA localization and pro-oncogenic protein properties.⁶

In this regard, BRCA1 exon 11 is alternatively spliced on a regular basis, resulting in the production of different levels of BRCA1 isoforms in different tissues, most importantly, in breast cancer cells.^11-13 Exon 11 is defined by three known splice sites; a weak acceptor site at the boundary with intron 10 and two weak donor sites located at the boundary with intron 11 (distal donor site) and within exon 11 (internal donor site). The differential recognition of the splice sites determines the expression levels of the various isoforms: exon 11 inclusion (full isoform; distal donor site), exon 11 skipping (D11 isoform), and partial exon 11 inclusion (D11q isoform; internal donor site). Following partial or total exclusion of exon 11, the open reading frame is maintained and functional BRCA1 proteins are still produced; however, the impact on cellular phenotype of those alternative protein isoform is different and not totally understood.⁶ In addition to this, there is little information on the mechanisms that control this differential exon 11 processing.

We have previously found that a splicing regulatory region exists at the beginning of exon 11 and in particular that deletion of a 20 nucleotide sequence located between the exon 11 acceptor site and the internal donor site causes skipping of the exon and increased production of the D11 isoform.¹⁰ This sequence, probably capable of abolishing the recognition of the exon 11 acceptor site, was uncovered thanks to a combination of evolutionary conservation analysis and minigene deletion analysis.

Now, in order to elucidate what sort of regulation modulates the production of the exon 11 full isoform and D11q isoform, we have analyzed the effect of sequences in the proximity of the exon 11 distal donor site (at the junction between exon 11 and intron 11) on BRCA1 exon 11 splicing. We show here that several regulatory sequences exist at the 3′ end of exon 11 where binding of splicing regulatory proteins Tra2beta and hnRNP L determine the choice of the donor sites. Finally, we use this information to provide a model of long exon splicing regulation that involves binding of splicing regulatory elements to multiple distant sequences.

Results

Deletions at the 3′ end of exon 11 increase the D11q isoform

To identify regulatory regions of splicing located at the 3′ end of BRCA1 exon 11, several deletions have been created in a previously described minigene construct (pB1) containing BRCA1 exon 11 and developed by us (Fig. 1A).¹⁰ Briefly, the pB1 minigene contains a region of genomic BRCA1 from exons 8 to 12 (including part of the flanking intronic sequences) under the control of the CMV promoter and can be used to analyze the production of the following BRCA1 splicing isoform: “full” (includes all exon 11), D11q (includes part of exon 11 from nucleotide 1 to nucleotide 117), and D11 (excludes all exon 11).

Figure 1. Minigene splicing assay of BRCA1 exon 11 carrying deletions. (A) The pB1 wild-type (WT) version of the minigene is shown. PCMV, promoter of the pCDNA3 vector. BRCA1 exon 11 size is reported (1 = position of the intron/exon 11 junction; 116, nucleotide position of the internal donor site of exon 11, relative to the beginning of exon 11; 3426, nucleotide position of the distal donor site of exon 11, relative to the beginning of exon 11). Black lines represent introns and gray boxes represent exons. Dotted lines show deletions within the 3′ end of exon 11. (B) 1.5% agarose gel detection of BRCA1 exon 11 splicing isoforms for pB1 WT minigene (WT), pB1 minigene carrying deletion 7 (del7), and pB1 minigene carrying deletion 6 (del6). A schematic representation of the three splicing isoforms full, D11q, and D11 is reported in correspondence of each band. M is the 100 bp DNA ladder. The lower panel shows a histogram reporting the relative percentage (%) of each isoform calculated against the total intensity of the bands representing each isoform, using Image J. The error bars show the standard deviation calculated from at least three independent replicates.

Exon 11 is composed of 3426 nucleotides and with the aim of finding regulatory regions near the donor site we created minigene deletions between nucleotides 3213–3305 (region 6) and 3316–3417 (region 7); resulting in minigene del6 and minigene del7, respectively. Forty-eight hours after transfection of MCF7 breast cancer cells with these constructs (alongside triplicate transfections of pB1 wt minigene as a control), the RNA was extracted, retro-transcribed, and amplified using specific primers to detect the full, D11q, and D11 isoforms originating from minigene-specific splicing. Splicing products were visualized on an agarose gel and are reported in Figure 1B. As shown in this figure, all three isoforms are produced both by WT and the del6–del7 hybrid minigenes. However, compared with the WT, each deletion caused a relative change in the isoform ratio toward an increase of D11q isoform and a decrease of full FL isoform (Fig. 1B).

This switch in D11q, coupled with a drop in full isoform expression, could be the consequence of deleting important sequences that mediate donor site recognition. To investigate this possibility, we used three in silico programs to predict the possible presence of binding sites for splicing regulatory proteins; SpliceAid, Splicemap, and The Human Splicing Finder (data not shown). Several splicing regulatory sequences and corresponding binding proteins were predicted (inside the region deleted in both del7 and del6), including the splicing factor Tra2beta, SRSF1, hnRNP H (region 7 only), and PTB (region 6 only). Three sequences were predicted to have the highest concentration of splicing enhancers (particularly for Tra2beta binding); region 6s, region 7b, and region 7c (Fig. S1). Therefore, in order to narrow down the identification of splicing regulatory elements, shorter single deletions were created in the pB1 minigene within those regions and also within an additional region (region 7a) found to contain a lower number of predicted enhancer sequences. As a result, the following mutant minigenes were engineered: del6s (exon 11 nucleotide deletion 3252–3275), del7a (exon 11 nucleotide deletion 3317–3349), del7b (exon 11 nucleotide deletion 3350–3379), del7c (exon 11 nucleotide deletion 3380–3417). These hybrid minigenes were transfected in MCF7 cells and the splicing products analyzed as before (Fig. 2A). In comparison with the WT, all hybrid minigenes except del7a showed an increase of the D11q isoform and a decrease of the full isoform relative to the total isoform ratio. In particular, del7c showed a degree of change comparable with del7 while the effect of del6s and del7b was less strong (Fig. 2B).

Figure 2. Minigene splicing assay of BRCA1 exon 11 carrying short deletions. The upper diagram shows a 1.5% agarose gel detection of BRCA1 exon 11 splicing isoforms for pB1 WT minigene (WT), pB1 minigene carrying deletion 7 (del7), deletion 7a (del7a), deletion 7b (del7b), deletion 7c (del7c), and pB1 minigene carrying deletion 6 (del6), and deletion 6s (del 6s). FL, band corresponding to the full isoform; D11q, band corresponding to the D11q isoform; D11, band corresponding to the D11 isoform. The lower panel shows a histogram reporting the relative percentage (%) of each isoform calculated against the total intensity of the bands representing each isoform, using Image J. The error bars show the standard deviation calculated from at least three independent replicates.

Splicing regulatory proteins Tra2beta, hnRNP L, hnRNP H, and PTB bind the 3′ end of BRCA1 exon 11 in vitro

To verify binding of the in silico predicted binding proteins to the identified splicing regulatory sequences we performed affinity purification assay of HeLa nuclear proteins binding to WT RNA sequences including region 7 (7WT), region 6 (6WT), or RNA sequences carrying deletion 7 (7del a), 7b (7del b), 7c (7del c), 6s(6del s). In order to discriminate experimental variability in western blot analysis, all RNAs were provided with a short GU repeat tail for binding of the splicing protein TDP-43. RNA sequences are shown in Figure 3.

Figure 3. Identification of splicing regulatory proteins. (A) The upper diagram shows the RNA sequences 6 WT and del6s used in the pull down and western blot analysis. Dotted lines represent the corresponding deleted region compared with the WT RNA. The GU 3′ nucleotide stretch represents the consensus binding site for TDP43 used as a control to normalize pulldown efficiencies. The lower panels represent the western blot probed for Tra2beta, PTB, hnRNP L, and TDP-43. Beads represents the pulldown control experiment without adding RNA. (B) RNA sequences 7 WT, del7a, del7b, and del7c used in the pulldown and western blot analysis are reported. Dotted lines represent the corresponding deleted region compared with the WT RNA. The GU 3′ nucleotide stretch represents the consensus binding site for TDP-43 that is used as a control to normalize pulldown efficiencies. The lower panel represents the western blot probed for Tra2beta, hnRNP H, and TDP-43. The Beads lane represents the pulldown control experiment without adding RNA.

Before performing western blots, however, the pulldown results were first analyzed by straight SDS-PAGE separation of the nuclear proteins specifically bound to each of the RNA-coated beads. Coomassie staining of these gels showed one band migrating at approximately 53 kDa that could be detected using 7WT RNA but not del7b RNA. Protein sequencing by mass spectrometry reported this band to contain the splicing regulatory protein hnRNP H (boxed in Fig. S2A). Additionally, the Coomassie also highlighted the presence of a protein band migrating at approximately 50 kDa that could be detected using 6WT RNA but not del6s RNA. After excision, protein sequencing by mass spectrometry reported this band to contain the splicing regulatory protein hnRNP L (boxed in Fig. S2B).

In order to verify in vitro binding of these hnRNP H, hnRNP L, and of other enhancer proteins predicted in silico, western blot analyses of the SDS-PAGE gels were performed. Western blot showed variation of Tra2beta and SRSF1 binding to the RNAs and confirmed differential binding of hnRNP H and L (Fig. 3).

In particular, as predicted in silico, Tra2beta was shown to bind both 6WT and 7WT, although binding seemed to be more efficient for 7WT than 6WT. In both cases, however, Tra2beta binding capacity was severely reduced by the mutants del7c and del6s (Fig. 3A and B), suggesting that these two sequences were responsible for the binding. The observation that no decrease in Tra2beta occurred in the del7a and del7b mutants, further suggested that these sequences do not participate in the binding to Tra2beta (Fig. 3B). Regarding other in silico predicted SR protein binding sites, a weak binding of SRSF1 was also found to occur for RNA 6 WT. This weak binding seemed to be specific as it was also lost following the introduction of the del6s mutant (Fig. 3A).

Regarding the other hnRNP factors, hnRNP H was also detected binding RNA 7WT, del7a, del7c, but not RNA del 7b, suggesting that hnRNP H exclusively recognizes region 7b (Fig. 3B) and PTB was detected to bind both 6WT RNA and del6s RNA, suggesting that, as predicted in silico, PTB is binding to the two UCUU motifs outside region 6s (Fig. 3A).

Finally, binding of hnRNP L was demonstrated to RNA 6WT and was also reduced to a very low level in RNA del6s (Fig. 3A).

Most of these findings confirmed the in silico analysis, except that hnRNP H was predicted to also bind region 7a (in silico data not shown), Tra2beta was predicted to also bind region 7b (Fig. S1) and SRSF1 was expected to bind several sequences along region 6 and region 7 (in silico data not shown). These differences suggest that in silico binding predictions can represent a good starting point to look for potential RNA protein interactions but that should also always be validated at the experimental level (even when multiple programs are used to try and obtain a more valid analysis).

Based on these findings, therefore, it was decided to further analyze the major interactors of these regions (Tra2beta, hnRNP H, and hnRNP L) to functionally investigate their ability to modulate BRCA1 exon 11 splicing.

Tra2beta binding to multiple sequences is essential for the recognition of the exon 11 distal donor site

In order to verify a role of the splicing regulatory protein Tra2beta in the recognition of exon 11 donor site, we performed RNA interference and overexpression experiments of Tra2beta in MCF7 cells transfected with minigenes WT, del7b, del7c, and del 6a (Fig. 4). As expected, and considering the results obtained with the in vitro binding assay, depletion of Tra2beta following siRNA treatment (confirmed by western blot of protein extracts from siRNA treated cells and reported in Fig. S3A) caused a reduction in full isoform and an increase in D11q isoform in both WT and del7b minigenes, while it had no effect on exon 11 splicing of minigene del7c or minigene del6s (Fig. 4A). This observation further confirmed that important binding sites for Tra2beta are lost when regions 6s or 7c are deleted.

Figure 4. Knockdown and overexpression of Tra2beta. (A) RT-PCR products from transient transfection and Tra2beta-knockdown experiments using minigene WT and minigenes carrying deletions. The relative fold increase in D11q isoform following Tra2beta knockdown (with standard deviation calculated from three independent triplicates) is shown at the bottom. (B) Effect of Tra2beta overexpression. The gel represents RT-PCR products from transient co-transfection experiments of pB1 minigenes with 1.5 μg of pGFP (/) or pGFP-TRA2B (+). The relative fold increase in D11q isoform expression (calculated from duplicate experiments) following co-transfection of 1.5 μg, 1 μg, and 0.5 μg of pGFP-TRA2B is shown at the bottom. A difference is observed in the relative isoform ratio between A and B. As reported previously,¹⁰ this is probably the consequence of minimal variability in cell culture when the experiment is performed on different days, as BRCA1 splicing of exon 11 is affected by the cell cycle.

Nonetheless, the effect of deletion 7c or deletion 6s on exon 11 splicing is much stronger than the effect of Tra2beta depletion with the WT minigene (compare Figs. 2 and 4A); possibly suggesting that additional enhancer proteins are binding region 6s and region 7c since Tra2beta knockdown was extremely efficient (Fig. S3A). Alternatively, concerning deletion 7c, it is possible that in this mutant region 7b (that binds hnRNP H in vitro) gets closer to exon 11 donor site and this may cause a silencing effect that combines loss of Tra2beta binding and interference of hnRNP H binding near the donor splice site.

Finally, Tra2beta overexpression caused a relative decrease of D11q isoform (Fig. 4B). However, as expected from the siRNA results, del7c and del6s minigenes demonstrated a low level of response to Tra2beta overexpression in comparison to WT and del7b constructs (Fig. 4B).

Interestingly, following overexpression of relatively lower Tra2beta concentrations (1 and 0.5 µg of plasmid), the WT minigene splicing was still affected but to a lesser extent and the splicing pattern of the del7c minigene was not particularly affected (Fig. 4B, lower panel).

Depletion of hnRNP H shows no evidence of regulation of the distal donor site

In vitro binding assay showed that the splicing regulatory protein hnRNP H binds region 7b (Fig. 3B; Fig. S2). If the binding of hnRNP H to region 7b is important for the recognition of the exon 11 distal donor site, then it would be expected that depletion of this protein would cause a decrease of the D11q isoform and an increase of the full isoform in the presence of the WT minigene but not in the presence of del7b minigene. Therefore, we transfected both the WT minigene and the del7b minigene in MCF7 cells treated with hnRNP H siRNA (knockdown of this protein was checked by western blot of protein extracts from siRNA treated cells, see Fig. S3B). However, no effect was observed on exon 11 splicing (Fig. 5). Overexpression of hnRNP H also failed to elicit clear responses from the pB1 minigene with the most consistent finding resulting in a reduction of the D11 isoform (data not shown).

Figure 5. Knockdown of hnRNP H. RT-PCR products from transient transfection of minigenes together with hnRNP H (siH)-knockdown experiments using minigene WT and minigene carrying deletion 7b (del7b). The relative fold increase in D11q isoform following hnRNP H knockdown (calculated from duplicate experiments) is shown at the bottom.

Taken together, the results suggest that either this protein-binding event is not functional or that compensatory interactions by other hnRNPs may be taking place to overcome the effect of silencing.

hnRNP L as a regulator of BRCA1 exon 11

In addition to Tra2beta, also hnRNP L was shown to bind region 6, and particularly, region 6s (Fig. 3A). As previously suggested for hnRNP H, if the binding of hnRNP L to region 6s is important for the recognition of exon 11 distal donor site it would be expected that depletion of this protein would cause a decrease of the D11q isoform and increase of the full isoform of the WT minigene, but not in the del6s minigene where the binding site for hnRNP L has been depleted. Therefore, we transfected the WT minigene or the del6s minigene in MCF7 cells treated with hnRNP L and hnRNP LL siRNA. We chose to use both siRNAs because hnRNP LL is a closely related protein to hnRNP L that is normally expressed at very low levels but is also known to be upregulated following hnRNP L knockdown.¹⁴

Although knockdown did not induce total depletion of hnRNP L (Fig. S3C), it was enough to induce an increase in the full splicing isoforms both with the WT minigene and the del6s minigene (Fig. 6). Unexpectedly, in both minigenes, we also observed a decrease of the D11q isoform and this decrease was actually higher with del6s. These observations support the conclusion that hnRNP L plays a role in the regulation of exon 11 splicing even in the absence of the binding site in region 6s. Furthermore, it suggests that its action does not consist in just the activation of exon 11 distal splice site as initially speculated.

Figure 6. Knockdown of hnRNP L and hnRNP LL. (A) RT-PCR products from transient transfection of minigenes coupled with hnRNP L/hnRNP LL-knockdown experiments using minigene WT and minigene carrying deletion 6s (del6s). The relative fold increase in D11q isoform following hnRNP L/LL knockdown (calculated from duplicate experiments) is shown at the bottom.

Since an excess of hnRNP L was previously reported to activate nonsense-mediated decay of its own RNA,¹⁵ we did not carry out additional experiments of hnRNP L overexpression as for Tra2beta.

Discussion

The large, 3426 Kb BRCA1 exon 11 encodes for 1142 aminoacids, which partially contributes to the ability of BRCA1 to localize in the nucleus and to the functionality of BRCA1 in DNA repair (reviewed in ref. 16). It also contains important phosphorylation sites and sites of interaction with other proteins involved in cell cycle regulation (reviewed in ref. 16).

Although important, this exon is alternatively spliced; producing a small amount of natural transcripts that miss part or the entire exon 11 sequence.

The relative abundance of those transcripts changes during the cell cycle and is cell type-specific,¹¹ suggesting that context-dependent regulation exists and controls alternative splicing of BRCA1 exon 11. Using a minigene splicing assay we have previously shown that removal of an important regulatory sequence (called region 2) at the beginning of exon 11 induces exclusion of the entire exon in the mature transcripts producing almost exclusively the D11 isoform, suggesting that a regulatory element of splicing, controlling the recognition of exon 11 acceptor site, was deleted.¹⁰

Here, we show that additional regulatory elements are present at the end of BRCA1 exon 11. These elements control the choice of two alternative donor sites and removal of these elements caused an increase in isoform D11q. As a consequence, this caused a decrease of the use of exon 11 distal donor site, and therefore, a decrease in the expression of the full isoform. In addition to characterizing these sequences in a minigene system, we further demonstrate that the splicing regulatory protein Tra2beta is the principal modulator of this splicing switch. In fact, we have shown that Tra2beta is able to bind two main sequences: the first located in the proximity of the distal exon 11 donor site and a second located more than 100 nucleotides distant (Fig. 7). At normal Tra2beta levels, in MCF7 cells, deletion of only one of those two sequences is sufficient to compromise Tra2beta enhancer effect on the recognition of the distal donor site; in fact, depletion of Tra2beta has no effect on constructs carrying single deletions of these regions. However, at higher Tra2beta concentrations, following transfection of 1.5 ug of the pGFP-TRA2B vector, the presence of one of the two sequences could still partially compensate for the lack of the other; in fact, Tra2beta overexpression was able to cause a decrease in D11q isoform in hybrid minigenes carrying deletion of one of the two binding regions, although not as much as the decrease observed in WT minigene without deletions or in a hybrid minigene carrying deletion of a region not involved in Tra2beta binding. Moreover, transfection of lower concentrations (0.5 or 1 ug) of the pGFP-TRA2B minigene were still able to cause a decrease of the D11q isoform with the WT minigene but not with the minigene del7c.

Figure 7. Proposed model of long BRCA1 exon 11 splicing regulation. Schematic representation of exon 11 (long gray square), its internal donor site (5′ssD11q), distal donor site (FL), and acceptor site (3′ss). The suggested binding sites of proteins Tra2beta and hnRNPL are reported relative to their in vitro detected binding sites on exon 11 RNA (not in scale). + indicates suggested enhancer effect of splice site recognition (relative to the arrow). - indicates suggested hnRNP L silencer of 5′ssFL splice site recognition throughout competition with Tra2beta binding. The additional RNA binding proteins, hnRNP H detected in vitro to bind the 3′ end of BRCA1 exon 11 are reported relatively to their putative binding region (not in scale). At the moment, however, no clear functional property can be ascribed to its presence.

Albeit, the effect of depletion of Tra2beta was not tested on endogenous BRCA1 transcripts, taken together, the results with the minigene splicing assay strongly suggest that physiological changes in the levels of Tra2beta can control the usage of BRCA1 exon 11 alternative donor sites and that the effect of these changes are dependent on the presence of regions 6s and 7c. Interestingly, this situation is similar to one previously observed for the NaspT exon. In this case, mutation of binding sites for Tra2beta induced a much lower splicing sensitivity to Tra2beta than the WT exon.¹⁷ Furthermore, Tra2beta was previously shown to change its phosphorylation state during the cell cycle and to act as a splicing repressor when dephosphorylated in mitotic cells.¹⁸ Thus, fluctuations in Tra2beta functionality during the cell cycle could also explain the change in BRCA1 isoforms ratio, including D11q and full isoform, previously observed during the cell cycle by Orban and Olah.¹¹

Besides the Tra2beta binding regions demonstrated in this study, in silico analysis also predicted additional Tra2beta binding sites at the 3′ end of exon 11 (region b, Fig. S1), for which we did not observe any evidence of binding in vitro. It is however possible that in exceptional circumstances, such as following massive increases in Tra2beta expression levels, this protein would be able to bind these otherwise inactive sites. In this way, it could also compete with binding of putative silencer proteins, and therefore, enhance recognition of the distal donor site.

In keeping with the existence of silencer elements, in this work we have found that hnRNP L can bind in vitro region 6s in proximity of exon 11 distal donor site. If binding to this region, in vivo, this protein may act as a silencer of the distal donor splice site, although additional studies will be needed in order to clarify the role of this factor in exon 11 recognition by the splicing machinery. If binding to this region, in vivo, this protein is probably acting as a silencer of the distal donor splice site, either directly or through competition with Tra2beta for binding to this region. In support of this hypothesis, the relative decrease in D11q isoform and increase in full isoform observed following hnRNP L knockdown was much lower when the WT minigene was transfected compared with del6 minigene. In fact, in the WT minigene, this downregulation ability may have been partially compensated for by the positive effect of Tra2beta on the distal donor site of exon 11. In other words, at lower levels of hnRNP L, the enhancer protein Tra2beta would bind region 6s more efficiently, and thus, be able to activate the distal exon 11 donor site. The fact that partial depletion of hnRNP L was still able to cause a relative decrease in D11q isoform with del6s minigene also means that hnRNP L is binding additional sequences along exon 11, and that this binding has an enhancer effect on the recognition of the internal donor site, or a silencer effect on the recognition of the distal exon 11 donor site. As a general rule, hnRNP L binding upstream of weak donor sites can enhance their recognition.¹⁹ Hence, it is possible that hnRNP L also binds the internal donor site upstream and acts as an enhancer for the production of the D11q isoform. In support of this we have found that the previously identified regulatory region, called region 2, upstream the donor site of the D11q isoform, is also able to bind hnRNP L in vitro following pulldown assay experiments (Marco Silipo personal communication).

In conclusion, the results reported here have shown that competition exists between the two exon 11 donor sites and suggest that it is in part modulated by the level of the splicing regulatory protein Tra2beta and by the splicing regulatory protein hnRNP L. The interplay between Tra2beta and hnRNP L at cellular level may account for a selective mechanism for full and D11q isoform expression. Although this needs to be further demonstrated, one possible model of this mechanism of action is suggested in Figure 7.

Taken together, these results therefore shed light on the complexity of long BRCA1 exon 11 recognition by the splicing machinery, and further confirm the role played by Tra2beta in the regulation of large exon splicing, as suggested from mouse CLIP coupled with deep sequence data.¹⁷ In this respect, our study is particularly useful because few examples exist in which the splicing mechanisms regulating recognition of long exons have been characterized in detail.

Materials and Methods

In silico predictions

Putative splicing regulatory sequences and their candidate binding proteins were predicted using the computational tools SFmap²⁰ and SpliceAid2²¹ as described before.¹⁰

Binding proteins were also predicted using The Human Splicing Finder.²² In this case only predictions indicating the putative binding protein were selected; which include the ESE Finder matrices analysis²³ and the ESE Motifs from HSF analysis (Threshold values are 75.964 for Tra2beta and 59.245 for 9G8).

Hybrid minigene constructs

Hybrid minigene carrying deletions were obtained through a two-step PCR overlap extension²⁴ using the previously described pB1 WT construct as a template.¹⁰ The pGFP and pGFP-TRA2B minigenes are a kind donation from David Elliott and previously described.²⁵

Cell culture and transfection

Human breast cancer cell lines, MCF7 (ATCC number: HTB-22™), were grown in DMEM medium with 4500 mg/L glucose, pyruvate, and L-glutamine supplemented with 1% penicillin/streptomycin and 10% fetal bovine serum. Cells were incubated at 37 °C in 5% CO2 atmosphere. For minigene transfection experiments, 50% confluent cells were transfected in 6-well plates with 0.5 ug of the minigene using 3 ul of FUGENE (Promega) following the manufacturers instruction. Forty-eight hours after transfection, RNA was extracted and the splicing products analyzed as described previously.¹⁰

Gene knockdown

All siRNA used represent validated sequences in several previous publications from our and other labs. As a control, we have used a scrambled sequence (the luciferase siRNA 5′ CGUACGCGGA AUACUUCGAd TdT 3′).

25 nM final concentration Tra2beta siRNA (5′ GCAUGAAGAC UUUCUGAAAd TdT 3′)²⁶ was transfected with INTERFERin (Polyplus) following the manufactures instructions. Seventy-two hours later, the siRNA was transfected again together with the minigene using jetPRIME (Polyplus) according to the manufacturers instruction. Forty-eight hours later, the RNA was extracted using RNAeasy plus kit (Qiagen) and protein extracted using RIPA buffer.

25 nM final concentration siRNA for hnRNP H (5′-UCAGAAGAUG AAGUCAAAUd TdT-3′),²⁷ hnRNP LL, and hnRNPL (5′-AGUGCAACGU AUUGUUAUAd TdT-3′)²⁸ and (5′-GAAUGGAGUU CAGGCGAUGd TdT-3′)²⁸ were transfected separately with INTERFERin (Polyplus) following the manufacturers instructions. Twenty-four hours later, the siRNA was transfected again together with the minigene using jetPRIME (Polyplus) according to the manufacturers instruction. Forty-eight hours later, the RNA was extracted using RNAeasy plus kit (Qiagen) and proteins extracted using RIPA buffer.

Pull-down assay and western blot

Pulldown assays were performed as described previously²⁹ using HeLa nuclear extract (CILBIOTECH s.a.) and a final concentration of Heparin of 5 mg/mL.

Pulldown samples were loaded on a 10% SDS-PAGE gel and Coomassie stained or electroblotted onto a Hybond ECL membrane (GE-healthcare). Antibody recognition was performed using in-house antibodies against PTB and TDP-43 proteins, the antibody against Tra2beta was a kind gift from Dr Elmar Stickeler, and commercial antibodies against SRSF1 (1H4 antibody, Invitrogen) and hnRNP L (Abcam). Protein bands were detected using an ECL chemiluminescence kit (Pierce) according to the manufacturer’s instructions.

10.4161/rna.28458