Zinc-induced modulation of SRSF6 activity alters Bim splicing to promote generation of the most potent apoptotic isoform BimS

Authors


Abstract

Bim is a member of the pro-apoptotic BH3-only Bcl-2 family of proteins. Bim gene undergoes alternative splicing to produce three predominant splicing variants (BimEL, BimL and BimS). The smallest variant BimS is the most potent inducer of apoptosis. Zinc (Zn2+) has been reported to stimulate apoptosis in various cell types. In this study, we examined whether Zn2+ affects the expression of Bim in human neuroblastoma SH-SY5Y cells. Zn2+ triggered alterations in Bim splicing and induced preferential generation of BimS, but not BimEL and BimL, in a dose- and time-dependent manner. Other metals (cadmium, cobalt and copper) and stresses (oxidative, endoplasmic reticulum and genotoxic stresses) had little or no effect on the expression of BimS. To address the mechanism of Zn2+-induced preferential generation of BimS, which lacks exon 4, we developed a Bim mini-gene construct. Deletion analysis using the Bim mini-gene revealed that predicted binding sites of the SR protein SRSF6, also known as SRp55, are located in the intronic region adjacent to exon 4. We also found that mutations in the predicted SRSF6-binding sites abolished generation of BimS mRNA from the mutated Bim mini-gene. In addition, a UV cross-linking assay followed by Western blotting showed that SRSF6 directly bound to the predicted binding site and Zn2+ suppressed this binding. Moreover, Zn2+ stimulated SRSF6 hyper-phosphorylation. TG003, a cdc2-like kinase inhibitor, partially prevented Zn2+-induced generation of BimS and SRSF6 hyper-phosphorylation. Taken together, our findings suggest that Zn2+ inhibits the activity of SRSF6 and promotes elimination of exon 4, leading to preferential generation of BimS.

Abbreviations
ActD

actinomycin D

Clk1

cdc2-like kinase 1

Dyrk1A

dual-specificity tyrosine-phosphorylation regulated kinase 1A

EGFP

enhanced green fluorescent protein

ESE

exonic splicing enhancer

ISE

intronic splicing enhancer

PUMA, p53 upregulated modulator of apoptosis; RS domain

arginine/serine-rich domain

SR protein

serine/arginine-rich protein

SRSF6

serine/arginine-rich splicing factor 6

Introduction

Zinc (Zn2+) is an essential trace metal that is involved in many biological functions. In the central nervous system, Zn2+ is stored in pre-synaptic vesicles together with glutamate, and released into the extracellular space in an activity-dependent manner [1]. Released Zn2+ functions as a neurotransmitter or a neuromodulator [2, 3]. In contrast, under pathological conditions such as cerebral ischemia, excessive Zn2+ released from pre-synaptic terminals triggers neuronal cell death [4-6]. At present, the mechanism by which high concentrations of Zn2+ cause neurotoxicity is not fully understood. However, it has been reported that Zn2+ induces caspase activation and DNA fragmentation, and that caspase inhibitors attenuate Zn2+-induced toxicity [7-9]. Therefore, Zn2+-induced neurotoxicity is thought to proceed through an apoptotic process.

Members of the Bcl-2 family are major regulators of mitochondrial-mediated apoptosis. The BH3-only proteins such as Bim and PUMA are a sub-group of the pro-apoptotic Bcl-2 family. Bim, which was initially isolated as a Bcl-2 binding protein, inhibits the anti-apoptotic activity of Bcl-2 and Bcl-xL by associating with these proteins [10, 11]. The Bim gene consists of six exons and undergoes alternative splicing to produce diverse splicing variants [12, 13]. BimEL, BimL and BimS are three predominant alternative splicing variants. Although these three Bim isoforms all induce apoptosis, the smallest variant BimS is the most potent inducer of apoptosis [11, 14]. As BimEL and BimL are sequestered to the microtubular cytoskeleton through binding to the dynein light chain LC8, their activity is restrained [15]. In contrast, BimS, which lacks the binding site, is localized in the mitochondria. Recently, it has been demonstrated that BH3-only proteins are involved in the process of neuronal death in stroke and several neurodegenerative disorders [16-18]. Additionally, exposure to Zn2+ has been reported to stimulate expression of PUMA and Bim in neuronal and non-neuronal cell types [19, 20].

Alternative splicing is recognized as an essential mechanism by which to acquire functionally diverse isoforms from a single gene in eukaryotes. Both constitutive and alternative splicing processes are regulated by trans-acting splicing factors such as serine/arginine-rich (SR) proteins [21]. Members of the SR protein family, including SRSF5 (also known as SRp40) and SRSF6 (also known as SRp55), have one or two RNA-recognition motifs and a C-terminal arginine/serine-rich (RS) domain. In addition to these trans-acting splicing factors, cis-elements present in precursor mRNAs (pre-mRNAs) are responsible for exon recognition [22]. These elements are conventionally classified as exonic and intronic splicing enhancers/silencers. SR proteins bind to exonic splicing enhancers (ESEs) and intronic splicing enhancers (ISEs) through the RNA-recognition motifs, and the RS domain participates in protein–protein interactions that facilitate recruitment to the spliceosome. The RS domain is extensively phosphorylated under normal conditions. Alterations of the phosphorylation status have been reported to influence SR protein activity [23, 24].

Recently, evidence has accumulated indicating that a number of apoptotic factors are regulated by alternative splicing [25]. Apoptotic stimuli promote changes in the phosphorylation status of SR proteins and affect the splicing process of apoptotic factors, leading to generation of isoforms with different functions from a common pre-mRNA. For example, Bcl-xL has been shown to be alternatively spliced to produce the pro-apoptotic isoform of Bcl-xL, Bcl-xS. Ceramide, which is known to be an apoptotic mediator, changes the phosphorylation status of SR proteins and promotes the production of Bcl-xS, leading to apoptosis [26-28]. More recently, it was reported that apoptosis of melanoma induced by the specific B-RAFV600E inhibitor PLX4720 causes preferential expression of BimS [29].

Previously, we reported that exposure to excess Zn2+ results in apoptosis in human neuroblastoma SH-SY5Y cells [19]. In this study, we examined the effect of Zn2+ on the expression of Bim in SH-SY5Y cells. Interestingly, Zn2+ preferentially promoted generation of the most potent apoptotic isoform BimS, but not BimEL and BimL. Furthermore, we also demonstrated that the SR protein SRSF6 is involved in this phenomenon.

Results

Zn2+ changes the splicing patterns of Bim pre-mRNA

Three predominant splicing variants generated from Bim pre-mRNA were expressed in SH-SY5Y cells (Fig. 1A). The exon cassette, which comprises exons 2, 3 and 4 without additional intervening sequences, sometimes acts as a single exon for production of BimEL. Exon 3 is removed in BimL, thus acting like an intron. We confirmed that the nuclear sequences of the three predominant PCR products were identical to those of BimEL, BimL and BimS. First, to examine whether Zn2+ affects expression of these Bim splicing variants, SH-SY5Y cells were exposed to various concentrations of ZnSO4 plus pyrithione for 4 h. We used pyrithione, a Zn2+ ionophore, to achieve an increase in the intracellular Zn2+ level. In general, BimEL mRNA was predominantly expressed in SH-SY5Y cells (Fig. 1A). Exposure to Zn2+ preferentially induced expression of BimS mRNA in a dose-dependent manner but had little effect on expression of BimEL and BimL mRNAs (Fig. 1B). Preferential expression of BimS mRNA induced by Zn2+ exposure increased in a time-dependent manner (Fig. 1C). Pyrithione itself failed to induce BimS mRNA expression (Fig. S1). Additionally, to examine whether Zn2+ also induces expression of BimS protein, we performed Western blot analysis. As shown in Fig. 1D, Zn2+ preferentially increased the production of BimS protein. These results indicate that Zn2+ provokes alterations in the expression of alternatively spliced isoforms of Bim mRNA.

Figure 1.

Zn2+ promotes preferential production of BimS. (A) Genomic structure of the Bim gene, and schematic diagram of the three predominant splice variants of Bim. (B) Dose-dependent study. SH-SY5Y cells were exposed to various concentrations of ZnSO4 plus 1 μm pyrithione for 4 h. After this exposure, RT-PCR was performed. mRNA levels were normalized against the GAPDH mRNA level for each sample. The graph represents the fold change in the BimS/BimEL (S/EL) or BimL/BimEL (L/EL) ratio compared with cells that were not exposed to Zn2+. Asterisks indicate statistically significant differences compared with cells not exposed to Zn2+ (S/EL only) (**< 0.01). (C) Time-course study. SH-SY5Y cells were exposed to 30 μm ZnSO4 plus 1 μm pyrithione for the indicated time. After this exposure, RT-PCR was performed. mRNA levels were normalized against the GAPDH mRNA level for each sample. The graph represents the fold change in the BimS/BimEL (S/EL) or BimL/BimEL (L/EL) ratio relative to time 0. Asterisks and hash symbols indicate statistically significant differences compared with time 0 (**< 0.01, S/EL; ##< 0.01, L/EL). (D) Western blot analysis. SH-SY5Y cells were exposed to 20 or 30 μm ZnSO4 plus 1 μm pyrithione for 10 h. After this exposure, the cells were lysed and Western blot analysis was performed using antibody against Bim (E) Effects of Zn2+ on BimS production in other cell types. HepG2 and MCF7 cells were exposed to 30 μm ZnSO4 plus 1 μm pyrithione for 4 h. After this exposure, RT-PCR was performed.

It has been reported that Bim is expressed in various kinds of cell [30-32]. To test whether Zn2+-induced alterations in the splicing patterns of Bim pre-mRNA are observed in other cell types, hepatoma HepG2 cells and human breast cancer MCF7 cells were exposed to ZnSO4 (30 μm). As shown in Fig. 1E, Zn2+ exposure elicited similar effects in both cell types. This result indicates that preferential production of BimS induced by Zn2+ is not a neuron-specific phenomenon.

Other metals and stresses fail to induce changes in Bim splicing patterns

To examine whether metals other than Zn2+ change splicing patterns of Bim pre-mRNA, SH-SY5Y cells were exposed to cadmium chloride (50 μm CdCl2), cobalt chloride (300 μm CoCl2) or copper chloride (300 μm CuCl2). As shown in Fig. 2A, Zn2+ markedly induced the production of BimS mRNA. However, the other metals had little or no effect on the splicing patterns of Bim mRNA. It has been demonstrated that genotoxins such as etoposide induce alterations in splicing [33, 34]. Therefore, to examine the effects of various stresses on this phenomenon, SH-SY5Y cells were exposed to oxidative stress (300 μm hydrogen peroxide), genotoxic stress (50 μm etoposide or 50 μm cisplatin) and endoplasmic reticulum stress (1 μm thapsigargin). However, as shown in Fig. 2B, these stresses hardly affected the alternative splicing of Bim pre-mRNA.

Figure 2.

Effects of various stresses on BimS production. (A) SH-SY5Y cells were exposed to 30 μm ZnSO4 plus 1 μm pyrithione for 6 h, 50 μm CdCl2 for 10 h, 300 μm CoCl2 for 10 h, or 300 μm CuSO4 for 10 h. After this exposure, RT-PCR was performed. The graph represents the fold change in the BimS/BimEL (S/EL) or BimL/BimEL (L/EL) ratio relative to control. Asterisks and hash symbols indicate statistically significant differences compared with control (**< 0.01, S/EL; #< 0.05, L/EL). C, control. (B) SH-SY5Y cells were exposed to 300 μm H2O2, 50 μm etoposide (etop), 50 μm cisplatin (cisp) or 1 μm thapsigargin (Tg) for 10 h. After this exposure, RT-PCR was performed. The graph represents the fold change in the BimS/BimEL (S/EL) or BimL/BimEL (L/EL) ratio relative to control. Asterisks and hash symbols indicate statistically significant differences compared with control (**< 0.01, S/EL; #< 0.05, L/EL). C, control.

The Bim mini-gene recaptures Zn2+-induced generation of BimS

It has been shown that RNA splicing occurs simultaneously with transcription [35]. To clarify how Zn2+ induces preferential BimS production, we used actinomycin D (ActD), a transcription inhibitor. SH-SY5Y cells were exposed to ZnSO4 (30 μm) in the presence or absence of ActD. As shown in Fig. 3A, ActD completely suppressed the production of BimS mRNA caused by Zn2+. This result suggests that the BimS mRNA induced by Zn2+ is generated from newly synthesized Bim pre-mRNA.

Figure 3.

The Bim mini-gene recapitulates endogenous Bim splicing. (A) Effect of ActD on Zn2+-induced BimS generation. SH-SH5Y cells were treated with 1 μg·mL−1 ActD for 30 min, followed by exposure to 30 μm ZnSO4 plus 1 μm pyrithione for 4 h. After this exposure, RT-PCR was performed. The graph represents the fold change in the BimS/BimEL (S/EL) or BimL/BimEL (L/EL) ratio relative to control. C, control; ActD, actinomycin D. Asterisks indicate statistically significant differences compared with control (**< 0.01, S/EL). Hash symbols indicate statistically significant differences compared with Zn2+ alone (##< 0.01, S/EL). (B) Schematic diagram of the Bim mini-gene and the products derived from the mini-gene. Arrows indicate the locations of primers used in RT-PCR analysis for the mini-gene. (C) Detection of BimL–EGFP and BimS–EGFP transcripts derived from the Bim mini-gene construct. SH-SY5Y cells were transfected with the Bim mini-gene. The next day, total RNA was extracted and RT-PCR was performed. RT, reverse transcription. (D) Zn2+ promotes generation of BimS–EGFP transcript. SH-SY5Y cells were transfected with the Bim mini-gene. The next day, the cells were exposed to 20 μm ZnSO4 plus 1 μm pyrithione for 7 h. After this exposure, total RNA was extracted and RT-PCR was performed. (E) Zn2+ promotes generation of the BimS–EGFP protein. SH-SY5Y cells were transfected with the Bim mini-gene. Nine hours later, the cells were exposed to 20 μm ZnSO4 plus 1 μm pyrithione for 15 h. After this exposure, the cells were lysed, and Western blot analysis was performed antibody against Bim.

To address the mechanism underlying Zn2+-induced alterations in splicing patterns of Bim pre-mRNA, we constructed a Bim mini-gene containing exons 2–4, an intronic region adjacent to exon 4, an intronic region adjacent to exon 5, and exon 5 (Fig. 3B). The Bim mini-gene was designed to produce fusion proteins of Bim with EGFP (enhanced green fluorescent protein). First, we examined whether the mini-gene works correctly. SH-SY5Y cells were transfected with the mini-gene, and then RT-PCR analysis was performed to detect BimL–EGFP and BimS–EGFP transcripts. As shown in Fig. 3C, both BimL–EGFP and BimS–EGFP transcripts were detected in SH-SY5Y cells transfected with the Bim mini-gene. In addition, to examine the effect of Zn2+ on splicing of Bim–EGFP pre-mRNA synthesized from the mini-gene, SH-SY5Y cells transfected with the Bim mini-gene were exposed to ZnSO4 (20 μm) for 8 h. As shown in Fig. 3D, BimS–EGFP transcript was preferentially generated upon exposure to Zn2+. Moreover, Western blot analysis revealed that protein levels of BimS–EGFP, but not BimEL–EGFP and BimL–EGFP, were increased by Zn2+ exposure (Fig. 3E). These results indicate that the Bim mini-gene accurately recapitulates the splicing process of the endogenous Bim gene.

Deletion analysis of the Bim mini-gene

As BimS mRNA is a variant that lacks exon 4, we hypothesized that cis-acting elements responsible for the elimination of exon 4 caused by Zn2+ exposure may be located in exon 4. First, to determine whether the elements are within exon 4, we constructed five exon 4 deletion mutants of the Bim mini-gene (Ex del 1–5; Fig. 4A). SH-SY5Y cells transfected with a wild-type (WT) construct or the exonic deletion constructs (Ex del 1–5) were exposed to ZnSO4 (20 μm) for 8 h, and then RT-PCR was performed to detect BimL–EGFP and BimS–EGFP transcripts. As shown in Fig. 4B, Zn2+ increased the ratio of BimS–EGFP transcript to BimL–EGFP transcript in all cells transfected with WT or exonic deletion constructs, although the ratio in cells transfected with the deletion constructs tended to be lower than that in cells transfected with the WT construct. These results suggest that elements involved in Zn2+-induced preferential BimS expression are not present in exon 4.

Figure 4.

Deletion analysis of the Bim mini-gene. (A) Construction of exon deletion mutants of the Bim mini-gene. WT, wild type. (B) Effects of exon 4 deletion mutants on Zn2+-induced elimination of exon 4. SH-SY5Y cells were transfected with WT or exon-deleted constructs of the Bim mini-gene. The next day, the cells were exposed to 20 μm ZnSO4 plus 1 μm pyrithione for 8 h. After this exposure, total RNA was extracted and RT-PCR was performed. The graph represents the fold change in the BimS–EGFP/BimL–EGFP ratio relative to each control. Z, Zn2+. (C) Construction of intron deletion mutants of the Bim mini-gene. WT, wild type. (D) Effects of intron deletions on Zn2+-induced elimination of exon 4. SH-SY5Y cells were transfected with WT or intron-deleted constructs of the Bim mini-gene. The next day, the cells were exposed to 20 μm ZnSO4 plus 1 μm pyrithione for 8 h. After this exposure, total RNA was extracted and RT-PCR was performed. The graph represents the fold change in the BimS–EGFP/BimL–EGFP ratio relative to each control. Z, Zn2+. Asterisks indicate statistically significant differences (**< 0.01).

As no cis-element responsible for Zn2+-induced splicing alterations was found in exon 4, we constructed intronic deletion mutants of the Bim mini-gene (Int del 1 and 2; Fig. 4C), and examined the involvement of the intronic region adjacent to exon 4 in Zn2+-induced generation of BimS. As shown in Fig. 4D, Int del 1 failed to induce BimS–EGFP mRNA expression upon Zn2+ exposure, but no difference in induction of BimS–EGFP transcript was observed between Int del 2 and WT. These results indicate that elements responsible for preferential production of BimS induced by Zn2+ are present in the intronic region adjacent to exon 4.

Involvement of SRSF6 in generation of BimS mRNA

A recent report demonstrated that SRSF6, one of the SR proteins, plays an important role in the splicing process of Bim pre-mRNA [29, 36]. Therefore, we used ESEfinder [37] to search for binding sites for SRSF6 in the intronic region adjacent to exon 4 corresponding to the deleted region of Int del 1. As shown in Fig. 5A, there are several predicted SRSF6-binding sites within the region. In contrast, no predicted SRSF6-binding site was present in exon 4 (data not shown).

Figure 5.

Involvement of SRSF6 in Bim splicing. (A) Analysis of SR protein-binding sites. (B) Upper panel: detection of recombinant SRSF6 protein. HEK293 cells were transfected with the SRSF6 expression vector or the empty vector. Forty-eight hours later, whole-cell lysates were prepared from the transfected cells and subjected to Western blot analysis antibody against FLAG. Lower panel: over-expression of SRSF6 suppresses BimS production. SH-SY5Y cells were co-transfected with the WT or Int del 1 construct together with the SRSF6 expression vector or empty vector. Forty-eight hours later, total RNA was extracted and RT-PCR was performed. The graph represents the fold change in the BimS–EGFP/BimL–EGFP ratio relative to cells co-transfected with WT construct and the empty vector. The asterisk indicates a statistically significant difference (*< 0.05). NS, not significant. (C) Knockdown of SRSF6 induces BimS production. SH-SY5Y cells were co-transfected with the WT or Int del 1 construct together with SRSF6 or control siRNA. Seventy-two hours later, total RNA was extracted and RT-PCR was performed. The graph represents the fold change in the BimS–EGFP/BimL–EGFP ratio relative to cells co-transfected with WT construct and control siRNA. Asterisks indicate a statistically significant difference (**< 0.01). NS, not significant.

To examine whether SRSF6 is involved in splicing of Bim pre-mRNA in SH-SY5Y cells, we constructed an expression vector for SRSF6. Over-expression of SRSF6 protein was observed in HEK293 cells transfected with the SRSF6 expression vector (Fig. 5B). SH-SY5Y cells were co-transfected with the SRSF6 expression vector and the WT or Int del 1 construct. Forty-eight hours later, the expression levels of BimL–EGFP and BimS–EGFP transcripts were assayed using RT-PCR. As shown in Fig. 5B, over-expression of SRSF6 decreased the ratio of BimS–EGFP transcript to BimL–EGFP transcript in cells transfected with the WT construct, but not the Ind del 1 construct. Moreover, to test the effect of SRSF6 knockdown on Bim splicing, we used the RNA interference technique. SH-SY5Y cells were co-transfected with SRSF6 or control siRNA and the WT or Int del 1 construct. Seventy-two hours later, the expression levels of BimL–EGFP and BimS–EGFP transcripts were assayed using RT-PCR. As shown in Fig. 5C, the expression levels of SRSF6 mRNA were suppressed in cells transfected with SRSF6 siRNA. Down-regulation of SRSF6 increased the ratio of BimS–EGFP transcript to BimL–EGFP transcript in cells transfected with the WT construct, but not the Ind del 1 construct (Fig. 5C). These results indicate that SRSF6 plays an important role in the inclusion of exon 4, and ISEs may be located in the intronic region adjacent to exon 4.

Definition of SRSF6-binding site(s) within the intron adjacent to exon 4

As analysis using ESEfinder showed that site 1 and site 2 are predicted SRSF6-binding sites (Figs 5A and 6A), we hypothesized that site 1 and/or site 2 act as ISEs and contribute to Zn2+-induced preferential BimS production. First, to determine whether mutations in site 1 and site 2 cause alterations in Bim splicing, we constructed a mutated Bim mini-gene (Int mut 1 × 2; Fig. 6A). Using ESEfinder, we confirmed that the mutations caused SRSF6-binding sites to disappear. SH-SY5Y cells were transfected with the WT or Int mut 1 × 2 construct, and then RT-PCR was performed to detect BimL–EGFP and BimS–EGFP transcripts. As shown in Fig. 6B, the ratio of BimS–EGFP transcript to BimL–EGFP transcript in cells transfected with the Int mut 1 × 2 construct were higher than that in cells transfected with the WT construct. This result indicated that the mutations promote elimination of exon 4, suggesting that the sites function as ISEs. Next, we examined the effects of the mutations on Zn2+-induced changes in Bim splicing. SH-SY5Y cells transfected with the WT or Int mut 1 × 2 construct were exposed to ZnSO4 (20 μm) for 8 h, and then RT-PCR was performed to detect BimL–EGFP and BimS–EGFP transcripts. As shown in Fig. 6C, the Int mut 1 × 2 construct did not increase generation of BimS–EGFP transcript upon Zn2+ exposure. This result strongly indicates that the sites contribute to Zn2+-induced preferential BimS production.

Figure 6.

Definition of cis-elements responsible for Zn2+-induced elimination of exon 4. (A) Construction of a Bim mini-gene with mutations within predicted SRSF6-binding sites. Boxes show predicted SRSF6-binding sites. The substituted nucleotides are indicated by lower-case letters. WT, wild type. (B) Effects of the Int mut 1 × 2 construct on constitutive splicing of BimS. SH-SY5Y cells were transfected with the WT or Int mut 1 × 2 construct. The next day, total RNA was extracted and RT-PCR was performed. The graph represents the ratio of BimS–EGFP to BimL–EGFP. Asterisks indicate statistically significant differences compared with WT (**< 0.01). (C) Effects of the Int mut 1 × 2 construct on Zn2+-induced elimination of exon 4. SH-SY5Y cells were transfected with the WT or Int mut 1 × 2 construct. The next day, the cells were exposed to 20 μm ZnSO4 plus 1 μm pyrithione for 8 h. After this exposure, total RNA was extracted and RT-PCR was performed. The graph represents the fold change in the BimS–EGFP/BimL–EGFP ratio relative to each control. Asterisks indicate a statistically significant difference (**< 0.01). NS, not significant. (D) Localization and sequences of RNA probes. (E) UV cross-linking study. Nuclear extracts (15 μg) from SH-SY5Y cells were incubated with biotinylated Int 1 or Ex 4 RNA probes (100 pmol). After UV cross-linking, the reaction mixtures were separated by 12% SDS/PAGE, and then transferred onto poly(vinylidene difluoride) membrane. The detection of RNA–protein complexes was performed using the streptavidin-biotin-peroxidase complex method (see Experimental procedures). After removal of the streptavidin-biotin-peroxidase complex, SRSF6 was detected by Western blotting using antibody against SRSF6. (F) Competition assay. Nuclear extracts (15 μg) from SH-SY5Y cells were incubated with biotinylated Int 1 in the presence of a fourfold excess of non-labeled Int 1 (WT) probe or mutant Int 1 (Mut 1) probe. After UV cross-linking, Western blotting with antibody against SRSF6 was performed. The substituted nucleotides are indicated by lower-case letters. (G) Effect of Zn2+ on the binding ability of SRSF6. SH-SY5Y cells were exposed to ZnSO4 (40 μm) plus 1 μm pyrithione for 3 h. After this exposure, nuclear extracts prepared from the cells were subjected to a UV cross-linking assay followed by Western blotting with antibody against SRSF6.

To ascertain whether SRSF6 directly binds to site 1, which showed the highest score for SRSF6 binding (Fig. 5A), we prepared nuclear extracts from SH-SY5Y cells, and performed a UV cross-linking assay followed by Western blotting using antibody against SRSF6. We used a biotinylated Int 1 RNA probe corresponding to site 1 and a biotinylated Ex 4 RNA probe corresponding to the region deleted in Ex del 4 (Fig. 6D). As shown in Fig. 6E, a RNA–protein complex (approximately 50 kDa) was detected using the Int 1 RNA probe, but the band was not observed using the Ex 4 RNA probe. The protein that bound to Int 1 RNA was identified as SRSF6 using Western blotting (Fig. 6E). In addition, we performed competition experiments using non-biotinylated Int 1 (WT) RNA and mutant Int 1 RNA (Mut 1). As shown in Fig. 6F, binding of SRSF6 to the biotinylated Int 1 RNA probe was inhibited in the presence of excess WT RNA, but not Mut 1 RNA. Moreover, we examined whether Zn2+ affects the RNA-binding ability of SRSF6. SH-SY5Y cells were exposed to ZnSO4 (40 μm) for 3 h, and the nuclear extracts were subjected to UV cross-linking assay followed by Western blotting with antibody against SRSF6. As shown in Fig. 6G, Zn2+ exposure suppressed binding of SRSF6 to site 1. These results suggest that SRSF6 directly binds to site 1, and Zn2+-induced changes in the binding ability of SRSF6 promote elimination of exon 4.

Effects of phosphorylation of SRSF6 on Zn2+-induced BimS generation

The phosphorylation status of SR proteins has been shown to influence RNA splicing [23, 24]. Therefore, we investigated the effects of Zn2+ exposure on the phosphorylation status of SR proteins. SH-SY5Y cells were exposed to ZnSO4 (30 μm), and the whole cell lysates were subjected to Western blot analysis using antibodies against phospho-SR proteins (1H4) and SRSF6. As shown in Fig. 7A, Zn2+ markedly phosphorylated SRSF6 and retarded its electrophoretic mobility, but had little effect on the phosphorylation status of other SR proteins. Interestingly, the hyper-phosphorylated form of SRSF6 is unable to bind to site 1 (Fig. 6G). These results indicate that Zn2+ promotes SRSF6 hyper-phosphorylation and inhibits the RNA-binding ability of SRSF6.

Figure 7.

Zn2+-induced changes in the phosphorylation status of SRSF6. (A) Time-course study. SH-SY5Y cells were exposed to 30 μm ZnSO4 plus 1 μm pyrithione for the indicated time. Whole-cell lysates were prepared from cells and subjected to Western blot analysis using antibodies against phospho-SR proteins (1H4) and SRSF6. The graph represents the fold change in the hyper-phosphorylation/hypophosphorylation ratio relative to time 0. Asterisks indicate statistically significant differences compared with time 0 (**< 0.01). hyper and hypo, hyper-phosphorylated and hypo-phosphorylated forms, respectively. (B) Effect of TG003 on Zn2+-induced BimS generation. SH-SY5Y cells were exposed to 30 μm ZnSO4 plus 1 μm pyrithione for 4 h in the presence or absence of TG003 (40 μm). After this exposure, RT-PCR was performed. mRNA levels were normalized against the GAPDH mRNA level for each sample. The graph represents the fold change in the BimS/BimEL (S/EL) or BimL/BimEL (L/EL) ratio relative to control. Asterisks and hash symbols indicate statistically significant differences compared with Zn2+ alone (**< 0.01, S/EL; ##< 0.01, L/EL). C, control. TG, TG003. (C) Effect of harmine on Zn2+-induced BimS generation. SH-SY5Y cells were exposed to 30 μm ZnSO4 plus 1 μm pyrithione for 4 h in the presence or absence of harmine (20 μm). After this exposure, RT-PCR was performed. mRNA levels were normalized against the GAPDH mRNA level for each sample. The graph represents the fold change in the BimS/BimEL (S/EL) or BimL/BimEL (L/EL) ratio relative to control. Asterisks indicate statistically significant differences compared with Zn2+ alone (**< 0.01, S/EL). C, control. Har, harmine. (D) Effect of protein kinase inhibitors on the phosphorylation status of SRSF6. SH-SY5Y cells were exposed to 40 μm ZnSO4 plus 1 μm pyrithione for 3 h in the presence or absence of TG003 (40 μm) or harmine (20 μm). Whole-cell lysates were prepared from cells and subjected to Western blot analysis using antibodies against phospho-SR proteins (1H4) and SRSF6. TG, TG003; Har, harmine.

It has been demonstrated that phosphorylation of SR proteins is regulated by protein kinases, including cdc2-like kinase 1 (Clk1) and dual-specificity tyrosine phosphorylation-regulated kinase 1A (Dyrk1A) [38-41]. To examine the involvement of these protein kinases in Zn2+-induced preferential generation of BimS, we used TG003, a Clk1 inhibitor [38], and harmine, a Dyrk1A inhibitor [42]. SH-SY5Y cells were exposed to ZnSO4 (40 μm) in the presence or absence of these protein kinase inhibitors (40 μm TG003 or 20 μm harmine). Both protein kinase inhibitors partially suppressed the generation of BimS mRNA (Fig. 7B,C). Additionally, we examined the effect of these inhibitors on Zn2+-induced hyper-phosphorylation of SRSF6. As shown in Fig. 7D, the hyper-phosphorylated SRSF6 migrated slightly faster in the presence of TG003 compared with Zn2+ alone, but harmine did not influence the migration of hyper-phosphorylated SRSF6. These results indicate that Zn2+ exposure promotes the hyper-phosphorylation of SRSF6, at least in part, through Clk1.

Discussion

Apoptotic stimuli have been shown to stimulate transcription of the Bim gene [17, 20, 31]. In most cases, the splicing patterns of Bim gene are usually unchanged. However, we showed here that Zn2+ alters splicing of Bim pre-mRNA and preferentially generates the most potent apoptotic isoform BimS, but not BimEL and BimL. In general, splicing of pre-mRNA is thought to occur simultaneously with transcription. However, it has been reported that mature mRNAs are produced from intron-retaining RNAs under stress conditions [43]. In this study, we showed that Zn2+-induced preferential production of BimS was completely suppressed in the presence of ActD (Fig. 3A). This result indicates that Zn2+ promotes elimination of exons 3 and 4 from newly synthesized Bim pre-mRNA to produce BimS preferentially.

There is growing evidence that apoptosis is regulated via alternative splicing of apoptotic factors. Ceramide, which is an important apoptotic mediator, induces expression of Bcl-xS, a pro-apoptotic isoform of Bcl-xL, leading to apoptosis. In addition, various stresses (e.g. heavy metals, genotoxin) have been shown to affect the splicing process of pre-mRNA [33, 34, 44]. However, divalent cations other than Zn2+ and genotoxic, oxidative and endoplasmic reticulum stresses had little or no effect on Bim splicing (Fig. 2). Moreover, Zn2+ did not induce Bcl-xS mRNA expression in SH-SY5Y cells (data not shown). These results indicate that the changes in splicing of Bim pre-mRNA and the subsequent preferential BimS generation that we detected are quite specific to Zn2+. Interestingly, Zn2+ has been reported to promote skipping of exon 12 of hypoxia-inducible factor-1α, and to produce inactive hypoxia-inducible factor-1α protein [45]. Thus, it is unlikely that Zn2+-induced preferential production of BimS from Bim pre-mRNA occurs through non-specific inhibition of the splicing process by Zn2+.

BimS has been shown to be the most potent apoptotic isoform among Bim isoforms because of its inability to associate with the dynein light chain. Whether exon 4 is included or not in Bim mRNAs is thought to affect the apoptotic potency of Bim, as exon 4 encodes the binding site of the dynein light chain. Indeed, BimS mRNA lacks exon 4. Using the Bim mini-gene, we demonstrated here that Zn2+ promotes elimination of exon 4 from Bim pre-mRNA and induces preferential BimS expression (Fig. 3). As cis-acting elements such as ESEs and ISEs have been reported to be necessary for exon recognition, we speculated that regulatory elements involved in Zn2+-induced elimination of exon 4 are present within or near exon 4. As expected, experiments using exonic and intronic deletion mutants of the Bim mini-gene showed that the Int del 1 construct did not induce expression of the BimS–EGFP transcript upon exposure to Zn2+ (Fig. 4D). These results indicate that cis-elements responsible for Zn2+-induced generation of BimS are most likely to be located within the intronic region adjacent to exon 4.

Many trans-acting factors including SR proteins, as well as the cis-acting element, play important roles in splicing of pre-mRNA [21]. SR proteins have been shown to bind directly to cis-elements and modulate exon inclusion/exclusion through interaction with other RNA-binding proteins and the spliceosome. Analysis of the deleted region of the Int del 1 construct using ESEfinder revealed that there are multiple predicted SRSF6-binding sites in this region. We demonstrated here that over-expression of SRSF6 decreased generation of the BimS–EGFP transcript, and knockdown of SRSF6 increased generation of the BimS–EGFP transcript (Fig. 5). Moreover, we identified an SRSF6-binding site in the intronic region. Mutation of the SRSF6-binding site promoted generation of the BimS–EGFP transcript under normal conditions and abolished the induction of expression of the BimS–EGFP transcript caused by Zn2+ (Fig. 6). These results suggest that SRSF6 binds to the ISE(s) in the intronic region adjacent to exon 4 and promotes inclusion of exon 4. In contrast, exon 4 deletion mutant constructs did not influence Zn2+-induced expression of BimS–EGFP transcript (Fig. 4B). Analysis of exon 4 using ESEfinder revealed that there are predicted binding sites for SRSF2 (also known as SC35) and SRSF5, but not for SRSF6, within exon 4 (data not shown).

It was reported that the specific B-RAFV600E inhibitor PLX4720 induces SRSF6 expression, and induction of SRSF6 mediates generation of BimS in melanoma cells treated with B-RAFV600E [29]. Almost 24 h treatment with PLX4720 has been shown to be necessary to induce BimS expression. In contrast, Zn2+-induced generation of BimS mRNA occurred within several hours. These findings indicate that the kinetics of generation of BimS differ between Zn2+ and PLX4720. Therefore, the mechanism underlying preferential production of BimS caused by Zn2+ may differ from that of PLX4720. Indeed, as discussed below, our present data suggest the possibility that Zn2+-induced alterations in the phosphorylation status of SRSF6 contribute to BimS production.

Activation of signaling pathways provoked by extracellular stimuli has been reported to regulate the pre-mRNA splicing process [22]. SR proteins are extensively phosphorylated, and their phosphorylation status affects their functions. We and others have demonstrated that Zn2+ exposure stimulates various signaling pathways, including mitogen-activated protein kinases [19, 46] and Akt [47]. Therefore, we speculated that Zn2+ alters the phosphorylation status of SRSF6. As expected, Western blot analysis revealed that Zn2+ exposure retarded the mobility of SRSF6, suggesting that Zn2+ promotes hyper-phosphorylation of SRSF6. Hyper-phosphorylation of the RS domain has been shown to inhibit the functions of SR proteins [23]. Indeed, we found that hyper-phosphorylation of SRSF6 caused by Zn2+ exposure inhibited the binding ability of SRSF6 (Fig. 6G). It has been reported that hyper-phosphorylated SRSF6 is re-localized to the nuclear speckles, which serve as storage sites for SR proteins [41]. Therefore, it is likely that Zn2+ induced-hyper-phosphorylation of SRSF6 decreases the efficiency of recognition of exon 4 and promotes preferential production of BimS. Several protein kinase families such as Clk1 and Dyrk1A have been shown to be involved in the phosphorylation of SR proteins [21]. We demonstrated here that TG003, a Clk1 inhibitor, or harmine, a Dyrk1A inhibitor, reduced Zn2+-induced preferential production of BimS (Fig. 7). TG003 partially recovered retardation of SRSF6 migration caused by Zn2+, whereas harmine had no effect on its migration. These results suggest that Clk1 is involved, at least in part, in Zn2+-induced hyper-phosphorylation of SRSF6. However, TG003 failed to completely inhibit its hyper-phosphorylation. Therefore, it is possible that protein kinases other than Clk1 contribute to Zn2+-induced hyper-phosphorylation of SRSF6. In contrast, harmine did not influence SRSF6 hyper-phosphorylation. Unfortunately, we do not know the reason why harmine prevents Zn2+-induced expression of BimS mRNA. Harmine may regulate the suppression of BimS production independently of SRSF6, because it also inhibits kinases other than Dyrk1A. Further detailed analysis is required to understand how Zn2+ phosphorylates SRSF6.

In conclusion, we demonstrated that Zn2+ exposure induces preferential generation of BimS. As BimS is the most potent apoptotic isoform, Zn2+-induced apoptosis may be regulated, at least in part, through alterations in Bim splicing. Moreover, our findings suggest that SRSF6 plays an important role in Zn2+-induced BimS production. Further experiments on Zn2+ regulation of Bim splicing are required in order to understand the molecular mechanisms of Zn2+ toxicity observed in neurodegenerative diseases and cerebral ischemia.

Experimental procedures

Materials

Zinc sulfate, cadmium chloride, cobalt chloride, copper sulfate, pyrithione, etoposide and cisplatin were purchased from Wako Pure Chemical (Osaka, Japan). Thapsigargin was purchased from Calbiochem (San Diego, CA, USA). Antibody against Bim was purchased from Cell Signaling Technology (Danvers, MA, USA). Antibodies against phospho-SR proteins (1H4), SRSF6 and actin were purchased from Millipore (Billerica, MA, USA). Antibody against FLAG was purchased from Sigma (St Louis, MO, USA).

Plasmid constructions

PCR amplification of the Bim gene was performed using human genomic DNA as a template. The DNA fragment containing exons 2–4 and an intronic region adjacent to exon 4 (E2–4 fragment) was amplified using primers 5′-GGCTCGAGGCCACCATGGCAAAGCAACCTTCTGA-3′ (forward) and 5′-GTGAATTCGTTTGATAGTTTGACACATC-3′ (reverse). The DNA fragment containing an intronic region adjacent to exon 5 and exon 5 (E5 fragment) was amplified using primers 5′-GTGAATTCCTTATTCTTTAACCAGAGT-3′ (forward) and 5′-ACGGATCCCTCCTTGCATAGTAAGCGT-3′ (reverse). The mini-gene was constructed by a two-step process. In the first step, the E5 fragment was sub-cloned into the EcoRI/BamHI sites of the pEGFP-N1 vector (Clontech, Mountain View, CA, USA). As a second step, the E2–4 fragment was sub-cloned in-frame into the XhoI/EcoRI sites of the pEGFP-N1 vector that includes the E5 fragment. This construct is referred to as the Bim mini-gene. Mutants of the Bim mini-gene were generated using a KOD Plus mutagenesis kit (Toyobo, Osaka, Japan). The following primers were used: Ex del 1 (forward 5′-AAATCAACACAAACCCCAAG-3′; reverse 5′-CCTGTCTGTGTCAAAAGAGA-3′), Ex del 2 (forward 5′-CCAAGTCCTCCTTGCCAGGC-3′; reverse 5′-CATGGGTGCTGGGCTCCTGT-3′), Ex del 3 (forward 5′-CAGGCCTTCAACCACTATCT-3′; reverse 5′-TGATTTGTCACAACTCATGG-3′), Ex del 4 (forward 5′-TATCTCAGTGCAATGGGTAA-3′; reverse 5′-ACTTGGGGTTTGTGTTGATT-3′), Ex del 5 (forward 5′-ATGGGTAAGCAATGCCTGGG-3′; reverse 5′-GGCCTGGCAAGGAGGACTTG-3′), Int del 1 (forward 5′-GGGAATGGAGGATGTGTCAA-3′; reverse 5′-CATTGCTTACCCATTGCACT-3′), Int del 2 (forward 5′-GAATTCTCTTATTCTTTAAC-3′; reverse 5′-ATCTAAAGATGGAATAAATC-3′) and Int mut 1 × 2 (forward 5′-AGGCTTGATATCGGAATTCAAATCCCCTTT-3′; reverse 5′-TCAAGCAGGCTCTACCGAGCTCACGTCAA-3′). The human SRSF6 expression plasmid was constructed by amplifying the SRSF6 cDNA from cDNA prepared from human SH-SY5Y cells using primers 5′-GTGGATCCATGCCGCGCGTCTACATAGGA-3′ (forward) and 5′-GACTCGAGTTAATCTCTGGAACTCGACCT-3′ (reverse). The SRSF6 cDNA obtained was sub-cloned in-frame into the BamHI/XhoI sites of the pcDNA FLAG vector (previously constructed in our laboratory). The nucleotide sequences were checked before use.

Cell culture and treatments

Human neuroblastoma SH-SY5Y cells and HEK293 cells were cultured in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum, 100 units·mL−1 penicillin G and 0.1 mg·mL−1 streptomycin in a humidified 5% CO2/95% air incubator at 37 °C. SH-SY5Y cells were seeded in a dish (6 cm diameter) at a cell density of approximately 1.2 × 106 cells per dish. The next day, the cells were exposed to various concentrations of ZnSO4 plus pyrithione (1 μm), a Zn2+ ionophore, or other reagents for the times indicated in the figure legends. Other metals (50 μm CdCl2, 300 μm CoCl2 and 300 μm CuSO4), and stressors (300 μm H2O2, 50 μm etoposide, 50 μm cisplatin and 1 μm thapsigargin) were also used instead of ZnSO4.

RT-PCR

Total RNA was extracted from the treated cells using TRIzol reagent (Invitrogen, Carlsbad, CA, USA). First-strand cDNA was synthesized from 4 μg total RNA. Aliquots of the reverse transcription reaction mixture (1 μL) were amplified using the common Bim forward primer 5′-TGATGTAAGTTCTGAGTGTG-3′ and the reverse primer 5′-ACGTAACAGTCGTAAGATAA-3′ for all Bim splicing variants or the reverse primer 5′-TATTCAAAAATACCCTCCTT-3′ for the three predominant Bim variants (BimEL, BimL and BimS), or 5′-GAAGGTGAAGGTCGGAGTC-3′ (forward) and 5′-CAAAGTTGTCATGGATGACC-3′ (reverse) for glyceraldehyde-3-phosphate dehydrogenase (GAPDH). To detect two splicing variants derived from the Bim mini-gene, the forward primer for BimL–EGFP (5′-CTACAGACAGAGCCACAAGACAG-3′) or the forward primer for BimS–EGFP (5′-CTACAGACAGAGCCACAAGCTTC-3′) and the common reverse primer 5′-AACTTGTGGCCGTTTACGTC-3′ were used. For amplification of endogenous Bim variants, mini-gene-derived Bim–EGFP variants and GAPDH, PCR was performed as follows: 2 min at 94 °C for one cycle, then 40 s at 94 °C, 40 s at 58 °C and 1 min at 72 °C for 28, 26 and 18 cycles, respectively. Aliquots of the PCR mixtures were separated on a 2% agarose gel and stained with ethidium bromide. Densitometric analyses were performed using the Multi Gauge software (Fujifilm, Tokyo, Japan).

Transfection study of Bim mini-gene constructs

SH-SY5Y cells were transfected with the Bim mini-gene constructs using Lipofectamine 2000 (Invitrogen, Carlsbad, CA, USA) according to the manufacturer's instructions. Briefly, after combining diluted mini-gene constructs (8 μg) and diluted Lipofectamine 2000 (10 μL), SH-SY5Y cell suspensions were mixed with the DNA–Lipofectamine 2000 complexes and seeded in dishes (6 cm diameter) at a cell density of approximately 1.5 × 106 cells per dish. The transfected cells were exposed to ZnSO4 (20 μm) plus pyrithione (1 μm) for 8 h (for RT-PCR analysis) or 15 h (for Western blot analysis). Co-transfection of SH-SY5Y cells with siRNA (200 nm) or SRSF6 expression vector (8 μg) and mini-gene constructs (2 μg) was performed using Lipofectamine 2000. The SRSF6 siRNA used was siGENOME SMART pool siRNA M-016067-01 (Dharmacon, Lafayette, CO, USA) and the control siRNA used was AllStars negative control siRNA (Qiagen, Hilden, Germany). After preparation of siRNA and the mini-gene using Lipofectamine 2000, SH-SY5Y cell suspensions were mixed with siRNA/DNA–Lipofectamine 2000 complexes and seeded in dishes (6 cm diameter) at a cell density of approximately 1.5 × 106 cells per dish. Forty-eight (for SRSF6 over-expression experiments) or seventy-two (for SRSF6 knockdown experiments) hours later, total RNA was extracted and RT-PCR was performed.

Preparation of whole-cell and nuclear extracts

After treatment, cells were washed twice with ice-cold NaCl/Pi. For preparation of whole-cell extracts, the cells were collected using 150 μL lysis buffer (20 mm Tris/HCl pH 7.4, containing 1 mm EDTA, 1 mm EGTA, 1% Triton X-100, 10 mm NaF, 1 mm Na3VO4, 20 mm β-glycerophosphate, 10 μg·mL−1 leupeptin, 1 mm phenylmethylsulfonyl fluoride and 1 mm dithiothreitol), and then lysed on ice for 30 min. The lysates were centrifuged at 14 000 g for 10 min at 4 °C to remove cellular debris. For preparation of nuclear extracts, the cells were collected using buffer A (20 mm HEPES/NaOH, pH 7.8, containing 15 mm KCl, 2 mm MgCl2, 5 μg·mL−1 leupeptin, 0.5 mm phenylmethylsulfonyl fluoride and 2 mm dithiothreitol), and centrifuged at 800 g for 30 sec at 4 °C. The cells were lysed in buffer B (buffer A containing 0.4% Nonidet P-40, Nacalai Tesque, Kyoto, Japan) for 5 min on ice, and centrifuged at 9 200 g for 30 sec at 4 °C. Finally, the pellets were suspended in buffer C (20 mm HEPES/NaOH, pH 7.8, containing 0.4 m NaCl, 10% glycerol, 5 μg·mL−1 leupeptin, 0.5 mm phenylmethylsulfonyl fluoride and 2 mm dithiothreitol), and stood for 30 min on ice. The nuclear extracts were centrifuged at 18 000 g for 10 min at 4 °C to remove cellular debris. The protein content of extracts was determined using Bio-Rad protein assay reagent (Hercules, CA, USA).

UV cross-linking study

RNA binding reactions were performed in a reaction mixture (20 μL) containing 10 mm HEPES pH7.8, 20 mm KCl, 1 mm MgCl2, 1 mm dithiothreitol, 40 μg yeast tRNA, 10 units RNase inhibitor, nuclear extracts (15 μg) and 100 pmol biotnylated RNA probe for 30 min on ice. After the binding reaction, the reaction mixture was irradiated using UV (0.12 J·cm−2) on ice for 15 min. The RNA–protein complexes were denatured, separated by 12% SDS/PAGE, and then transferred onto a poly(vinylidene difluoride) membrane. The membrane was incubated with streptavidin–biotin–peroxidase complex. The RNA–protein complexes were visualized using the chemiluminescence reagent ImmunoStar LD (Wako Pure Chemical , Osaka, Japan). We used the following biotinylated RNA probes: Int 1 RNA probe (5′-AGGCUGUGUGUGGCAUUUA-3′) and Ex 4 RNA probe (5′-CUCCUUGCCAGGCCUUCAACCAC-3′).

Western blotting

Cell lysates (30 μg) were separated by 12% SDS/PAGE, and then transferred onto a poly(vinylidene difluoride) membrane. The membrane was incubated with antibodies against Bim (1 : 3000), phospho-SR proteins (1H4) (1 : 3000), SRSF6 (1 : 3000), FLAG (1 : 3000) or actin (1 : 3000), followed by incubation with horseradish peroxidase-conjugated second antibodies. Proteins were visualized using an ImmunoStar LD or the SuperSignal West Pico detection system (Thermo Scientific, Rockford, IL, USA).

Statistical analysis

Data were analyzed using ANOVA followed by post hoc Bonferroni tests or Student's t test. A P value < 0.05 was considered significant.

Acknowledgements

We thank Kaoru Torii for technical assistance. This work was supported by a Grant-in-Aid for Scientific Research (C) from the Japan Society for the Promotion of Science (number 23590644).

Ancillary