Cardiac glycosides correct aberrant splicing of IKBKAP-encoded mRNA in familial dysautonomia derived cells by suppressing expression of SRSF3

Authors

  • Bo Liu,

    1. Laboratory for Familial Dysautonomia Research, Department of Biological Sciences, Fordham University, Bronx, NY, USA
    Search for more papers by this author
  • Sylvia L. Anderson,

    1. Laboratory for Familial Dysautonomia Research, Department of Biological Sciences, Fordham University, Bronx, NY, USA
    Search for more papers by this author
  • Jinsong Qiu,

    1. Laboratory for Familial Dysautonomia Research, Department of Biological Sciences, Fordham University, Bronx, NY, USA
    2. Department of Cellular and Molecular Medicine, University of California (San Diego), La Jolla, CA, USA
    Search for more papers by this author
  • Berish Y. Rubin

    Corresponding author
    1. Laboratory for Familial Dysautonomia Research, Department of Biological Sciences, Fordham University, Bronx, NY, USA
    • Correspondence

      B. Y. Rubin, Department of Biological Sciences, Fordham University, Bronx, NY 10458, USA

      Fax: +718 817 2792

      Tel: +718 817 3637

      E-mail: rubin@fordham.edu

    Search for more papers by this author

Abstract

The ability to modulate the production of the wild-type transcript in cells bearing the splice-altering familial dysautonomia (FD) causing mutation in the IKBKAP gene prompted a study of the impact of a panel of pharmaceuticals on the splicing of this transcript, which revealed the ability of the cardiac glycoside digoxin to increase the production of the wild-type, exon-20-containing, IKBKAP-encoded transcript and the full-length IκB-kinase-complex-associated protein in FD-derived cells. Characterization of the cis elements and trans factors involved in the digoxin-mediated effect on splicing reveals that this response is dependent on an SRSF3 binding site(s) located in the intron 5′ of the alternatively spliced exon and that digoxin mediates its effect by suppressing the level of the SRSF3 protein. Characterization of the digoxin-mediated effect on the RNA splicing process was facilitated by the identification of several RNA splicing events in which digoxin treatment mediates the enhanced inclusion of exonic sequence. Moreover, we demonstrate the ability of digoxin to impact the splicing process in neuronal cells, a cell type profoundly impacted by FD. This study represents the first demonstration that digoxin possesses splice-altering capabilities that are capable of reversing the impact of the FD-causing mutation. These findings support the clinical evaluation of the impact of digoxin on the FD patient population.

Abbreviations
CLK1

CDC-like kinase 1

EGCG

epigallocatechin gallate

ESS

exonic splicing suppressor

FD

familial dysautonomia

hnRNP

heterogeneous nuclear ribonucleoprotein

IKAP

IκB-kinase-complex-associated protein

IKBKAP

inhibitor of κ light polypeptide gene enhancer in B cells, kinase-complex-associated protein

PDK1

pyruvate dehydrogenase kinase, isozyme 1

siRNA

small interfering RNA

SRSF3

serine/arginine-rich protein splicing factor

TP53I3

tumor protein p53 inducible protein 3

Introduction

Familial dysautonomia (FD) is an autosomal recessive disorder that impacts the development and survival of sensory, sympathetic and some parasympathetic neurons [1, 2]. Individuals with FD exhibit cardiovascular instability, recurrent pneumonias, vomiting/dysautonomic crises, gastrointestinal dysfunction, decreased sensitivity to pain and temperature, and defective lacrimation [1-4]. FD is a life-threatening disorder with a high mortality rate. Two FD-causing mutations have been identified in individuals of Ashkenazi Jewish descent [5, 6]. The more common mutation is a T→C transition in the sixth position of the splice donor site of intron 20 (termed IVS20+6T→C) of the IKBKAP gene. This mutation alters the splicing process, resulting in the almost exclusive production of an exon-20-lacking transcript which, when translated, produces a truncated protein.

The IKBKAP-encoded IκB-kinase-complex-associated protein (IKAP), initially named on the basis of its reported ability to bind and serve as a scaffold protein for the IκB kinase complex [7], was later found not to be associated with this complex [8, 9]. IKAP appears to have multiple cellular functions. It has been reported to be a subunit of the Elongator complex [10] and to play roles in c-Jun N-terminal kinase (JNK) signaling [11], neuronal development during embryogenesis [12], tRNA modification [13], exocytosis [14] and actin cytoskeleton regulation [15].

The observed production of a small amount of the exon-20-containing IKBKAP transcript in cell lines and tissues from individuals with FD [16] prompted a search for therapeutic modalities capable of facilitating the production of the exon-20-containing IKBKAP transcript in those affected with this disorder. In 2003, we reported that tocotrienols, members of the vitamin E family [16], and epigallocatechin gallate (EGCG), a flavonoid present in plants, elevate the production of the exon-20-containing transcript in FD-derived cells [17]. Tocotrienols mediate their effect by increasing the rate of transcription of the IKBKAP gene [16] and EGCG alters the splicing process, resulting in the inclusion of exon 20 in the IKBKAP-derived transcript [17]. Others have reported the ability of both kinetin, a plant cytokinin which alters the splicing process [18-21], and phosphatidylserine, through its effect on IKBKAP gene expression [21], to increase the production of the full-length transcript from the IVS20+6T→C-bearing allele. A recent large-scale screening program performed on FD-induced pluripotent stem cells has identified additional compounds that increase IKBKAP expression [22].

Clinical studies of the impact of tocotrienol administration in individuals with FD revealed (a) a tocotrienol-mediated elevated production of the full-length IKBKAP transcript in peripheral blood cells, (b) an improvement of cardiac function, (c) a reduced incidence of dysautonomic crises and (d) increased eye moisture [4, 23-25] (http://emedicine.medscape.com/article/1200921). Experiments to examine the impact of the combined treatment of cells with EGCG and tocotrienols revealed a synergistic production of the exon-20-containing transcript and functional IKAP protein [17].

In addition, we recently reported that the isoflavones genistein and daidzein, which are found in soy, also modulate the splicing of the IKBKAP-derived transcript in FD-derived cells, resulting in the enhanced production of the wild-type, exon-20-containing, transcript from the IKBKAP gene bearing the FD-causing mutation [26]. We further showed that treatment of FD-derived cells with a combination of genistein and EGCG induces cellular levels of IKAP protein that are indistinguishable from those produced in normal cells.

In recent years, there has been a growing interest in re-purposing pharmaceuticals that have already been approved by the US Food and Drug Administration (FDA) [27-29]. The well characterized pharmacology and safety profiles that exist for these drugs allow for their clinical evaluation with little delay. To examine the impact of such compounds on the production of the functional IKBKAP-derived transcript in FD-derived cells, we performed high-throughput screening of a library of 800 products that included FDA-approved compounds and report that digoxin and various related cardiac glycosides alter the splicing process of IKBKAP-encoded mRNA, resulting in the enhanced production of full-length IKBKAP-encoded transcript and protein in FD-derived cells. Characterization of the molecular mechanism involved reveals that digoxin mediates its effect through suppression of the levels of the splicing factor SRSF3.

Results

To study the impact of large numbers of compounds on the splicing of the exon-20-encoded sequence of the IKBKAP transcript produced by the allele bearing the IVS20+6T→C mutation, we generated an HEK293 cell line that constitutively contains and transcribes a minigene construct containing a genomic DNA fragment that spans exons 19–21 of the FD mutation-bearing IKBKAP allele (Fig. 1A). RT-PCR amplification of the vector-derived transcript revealed the production of both an exon-20-containing (normal, i.e. wild-type) and an exon-20-lacking (mutant) transcript (Fig. 1B). The relative amounts of these transcripts produced in the minigene-containing cells are comparable with that observed in cells homozygous for the FD-causing mutation (Fig. 1B). High-throughput screening of library compounds including FDA-approved pharmaceuticals revealed that treatment with digoxin resulted in an enhanced production of the minigene-encoded exon-20-containing transcript (Fig. 1C).

Figure 1.

HEK293 cells constitutively expressing exon-20-containing and -lacking IKBKAP transcripts from the IKBKAP-FD minigene and FD-derived cell lines produce more exon-20-containing transcript in response to digoxin. (A) A pcDNA3.1/v5-His-TOPO vector construct containing a genomic DNA fragment encoding exons 19–21 of IKBKAP that contains the FD-causing IVS20+6T→C mutation was stably transfected into HEK293 cells. The diagram depicts the two IKBKAP-encoded transcripts constitutively expressed from the minigene: the normal, exon-20-containing, and the aberrant, exon-20-lacking, transcripts. (B) RNA purified from two FD-derived cell lines, GM04589 (FD1) and GM04899 (FD2), and the minigene-containing HEK293 cell line was subjected to two-band RT-PCR with primers spanning exons 19–23 (for the FD-derived cell lines) or with a vector-specific primer and a primer in exon 21 (for the minigene-transfected HEK293 cell line) of the IKBKAP transcript. RT-PCR products were separated on agarose gels with a typical experiment depicted. (C) RNA purified from minigene-transfected HEK293 cells incubated in the absence (−) or presence (+) of 100 ng·mL−1 digoxin for 48 h was subjected to RT-PCR as described above. (D) RNA purified from two FD-derived cell lines, GM04589 (FD1) and GM04899 (FD2), incubated in the absence (−) or presence (+) of 100 ng·mL−1 digoxin for 48 h was subjected to RT-PCR as described above.

To examine the effect of digoxin on the IKBKAP transcript in FD-derived cell lines, two FD-derived fibroblast cell lines were incubated in the absence or presence of digoxin and the IKBKAP transcripts generated in these cells were characterized by RT-PCR. Digoxin treatment resulted in the almost exclusive production of the exon-20-containing transcript (Fig. 1D). Incubation of FD-derived cells with eight related cardiac glycosides (digitonin, digitoxin, digoxigenin, digitoxigenin, acetyldigitoxin, bufalin, ouabagenin and ouabain) revealed similar responses (data not shown). As digoxin is a widely used drug with a well characterized safety profile, we continued our study of the splice-altering activity of the cardiac glycosides using this compound.

The goal of the screening program was to identify compounds with therapeutic potential for individuals with FD and, as the therapeutic digoxin concentration range observed in the sera of patients being treated for congestive heart failure and atrial fibrillation is between 0.5 and 2.0 ng·mL−1 [30], we examined the dose responsiveness of the FD-derived cells. The IKBKAP transcripts generated in these digoxin-treated cells were characterized by qRT-PCR using primers that amplify the exon-20-containing transcript, the exon-20-lacking transcript and the exon 34–35 region of the IKBKAP transcript (which is not impacted by the FD-causing mutation) to measure the ‘total’ amount of the IKBKAP transcript present. A statistically significant enhanced production of the exon-20-containing transcript (Fig. 2A) was detected in the presence of as little as 0.5 ng·mL−1 digoxin, and a reduced production of the exon-20-lacking transcript (Fig. 2B) was detected in the presence of as little as 0.2 ng·mL−1 digoxin. As measured using primers recognizing the exon 34–35 region of the IKBKAP transcript, digoxin did not affect the overall cellular level of IKBKAP mRNA (Fig. 2C). To determine whether the changes in the exon-20-containing transcript resulted in an increased presence of the full-length protein, western blot analysis of IKAP levels in protein extracts from FD-derived cell lines incubated in the presence of varying concentrations of digoxin revealed a responsiveness at therapeutic concentrations achieved in patients treated for cardiac failure (Fig. 2D).

Figure 2.

Digoxin upregulates wild-type IKBKAP-encoded transcript and protein at therapeutic concentrations. (A) RNA, isolated from the FD-derived GM04589 cell line incubated either in the absence of digoxin or in the presence of varying concentrations of digoxin for 48 h, was subjected to qRT-PCR analysis to quantitate the relative levels of wild-type exon-20-containing IKBKAP transcript (A), exon-20-lacking IKBKAP transcript (B) and total IKBKAP transcript (C). ‘Relative level’ refers to the relative increase in IKBKAP transcript in the digoxin-treated cells compared with untreated cells. Experiments were done three times and the results presented represent the mean ± SD; *< 0.05. (D) FD-derived cells (GM04589) were incubated in the absence of digoxin or in the presence of varying concentrations of digoxin for 48 h. Cells were harvested and protein extracts were prepared and subjected to western blot analysis using a monoclonal antibody to IKAP. Blots were then scanned and the relative densities of the bands were determined with imagej software. A representative blot is depicted. Experiments were done three times and the results presented represent the mean ± SD; *< 0.05.

As FD is a disorder that primarily impacts neuronal cells, before considering the use of digoxin in the FD patient population, we characterized the impact of digoxin on the splicing process in neuronal cell lines. As IVS20+6T→C -containing, FD-derived neuronal cells are not available, we sought to identify additional digoxin-mediated alternative splicing events that could be examined in neuronally derived cells. To select alternative splicing events to be characterized, we chose 35 genes from the kinase and cancer gene lists on the Exonhit Therapeutics (Gaithersburg, MD, USA) splice array portal (http://portal.splicearray.com/PortalHome/) and designed primers to amplify the RNAs resulting from 134 alternative splicing events previously reported to occur in the processing of these gene transcripts. We first measured the impact of digoxin on these events by performing RT-PCR on RNA isolated from untreated and digoxin-treated FD-derived fibroblast cells. We observed that digoxin modulated three splicing events: the inclusion of exon-4-encoded sequence in the CDC-like kinase 1 (CLK1) encoded transcript, the inclusion of exon-4-encoded sequence in the tumor protein p53 inducible protein 3 (TP53I3) encoded transcript and the inclusion of exon-3a-encoded sequence (unreported 60-nucleotide segment retained from intron 3 of PDK1, corresponds to nucleotides 23,636,887–23,636,946 of accession # NT_005403.17) in the pyruvate dehydrogenase kinase, isozyme 1 (PDK1) transcript (Fig. 3A). Examination of the processing of these transcripts in untreated and digoxin-treated neuronally derived cell lines revealed that the splice-modulating activity of digoxin also occurs in cells of neuronal origin (Fig. 3B).

Figure 3.

CLK1, TP53I3 and PDK1 transcripts are alternatively spliced in FD-derived and neuronal cell lines treated with digoxin. (A) RNA was isolated from two FD-derived cell lines, GM04589 (FD1) and GM04899 (FD2), that had been incubated for 48 h in the absence (−) or presence (+) of 100 ng·mL−1 digoxin and subjected to two-band RT-PCR. RT-PCR products were separated on agarose gels with a typical experiment depicted. The schematic depicts alternatively spliced transcripts visualized on the agarose gels shown. The larger CLK1 transcript contains exon 4, while in the smaller one exon 4 is spliced out. The larger TP53I3 transcript contains exon 4, whereas the smaller one does not. The larger PDK1 transcript contains an alternative exon 3a that is not present in the smaller product. (B) RNA was isolated from three neuronally derived cell lines, BE(2)-M17, LA1-55n and Be(2)-C, that had been incubated for 48 h in the absence (−) or presence (+) of 100 ng·mL−1 digoxin and the CLK1, TP53I3 and PDK1 transcripts were analyzed by two-band RT-PCR.

To elucidate the molecular mechanism by which digoxin mediates its splice-altering effect on the IKBKAP transcript generated from the IVS20+6T→C -bearing allele, deletions were introduced into the exon 20 sequence of the IVS20+6T→C -bearing IKBKAP-FD minigene construct and the transcripts generated in cells transfected with these constructs were characterized. Some of the constructs were observed to produce both the exon-20-lacking and the exon-20-containing transcripts while other constructs failed to generate any detectable exon-20-containing transcript (Fig. 4). Digoxin treatment of the cells transfected with these constructs revealed that all of the constructs bearing mutations that eliminated constitutive production of the exon-20-containing transcript also failed to respond to the digoxin (Fig. 4). As exon 20 of the IKBKAP-derived transcript has been reported to be a ‘weak’ exon [31] and as mutations introduced into this exon by others have also resulted in an elimination of the constitutive production of the exon-20-containing transcript [32], we questioned the validity and usefulness of the data generated using this minigene construct to identify the cis elements involved in the splice-altering effect of digoxin.

Figure 4.

IKBKAP-FD minigene exon 20 deletion mutants vary both in their constitutive and digoxin-regulated expression of exon-20-containing and -lacking IKBKAP transcripts. HEK293 cells transfected with IKBKAP-FD minigene constructs carrying no exon 20 deletion (no del) or the 5-nucleotide-long deletions indicated were incubated either in the absence (−) or presence (+) of 100 ng·mL−1 digoxin for 48 h. The schematic depicts the deleted bases in each minigene construct. RNA was then purified and subjected to two-band RT-PCR with a vector-specific primer and a primer in exon 21 of IKBKAP transcript. RT-PCR products were separated on agarose gels with a typical experiment depicted.

Having identified other transcripts in which digoxin treatment facilitates the inclusion of exonic sequence, we generated a minigene construct containing genomic DNA spanning exons 3–5 of the CLK1 gene in which the effect of digoxin on the inclusion of exon-4-encoded sequence could be evaluated (Fig. 5A). RT-PCR amplification of the minigene-derived transcript expressed in transiently transfected WI-38 cells revealed the predominant production of the CLK1 exon-4-lacking transcript in the untreated cells whereas digoxin treatment resulted in inclusion of the exon-4-encoded sequence (Fig. 5B).

Figure 5.

A CLK1 minigene construct expressing exon-4-containing and exon-4-lacking CLK1 mRNA splice variants expresses more exon-4-containing CLK1 transcript in response to digoxin. (A) A pcDNA3.1/v5-His-TOPO vector construct containing a genomic DNA fragment encoding exons 3–5 of CLK1 was transiently transfected into WI-38 cells. The diagram depicts the alternative splicing of exon 4 of CLK1 transcript expressed from the minigene. (B) RNA purified from minigene-tranfected WI-38 cells incubated either in the absence (−) or presence (+) of 100 ng·mL−1 digoxin was subjected to two-band RT-PCR of minigene-encoded CLK1 transcripts. RT-PCR products were separated on agarose gels with a typical experiment depicted.

To characterize the digoxin-responsive cis element(s) present in CLK1 that may be responsible for the digoxin responsiveness, approximately 20-nucleotide-long deletions were introduced into the exon 4 sequence of this minigene and digoxin responsiveness of these minigenes was evaluated in transiently transfected WI-38 cells (Fig. 6A). All of the deletion-bearing minigene constructs continued to constitutively produce some exon-4-containing transcript, and in all of these mutated constructs digoxin treatment facilitated the inclusion of the exon-4-encoded sequence (Fig. 6A). Transient transfection of WI-38 cells with constructs lacking either the entire CLK1 exon 3 or exon 5 sequence continued to constitutively produce some exon-4-containing transcript and the treatment of these cells with digoxin resulted in the enhanced inclusion of the exon 4 sequence (data not shown). Analysis of the RNA isolated from cells transfected with CLK1 minigene constructs bearing deletions in intron 4, spanning all but the conserved intron donor and acceptor sequences, revealed the continued constitutive production of the exon-4-containing transcript and the digoxin responsiveness of the transcripts generated from all of these minigene constructs (data not shown).

Figure 6.

A CLK1 minigene carrying mutations in a putative SRSF3 binding site fails to respond to digoxin. (A) WI-38 cells transiently transfected with CLK1 minigene constructs carrying no deletion (no del) or deletions in exon 4 as shown were incubated either in the absence (−) or presence (+) of 100 ng·mL−1 digoxin for 48 h. RNA was then purified and subjected to two-band RT-PCR of minigene-expressed CLK1 transcripts. RT-PCR products were separated on agarose gels with a typical experiment depicted. (B) RNA isolated from WI-38 cells that were transiently transfected with CLK1 minigene constructs carrying no deletion (no del) or deletions in intron 3 (the approximate positions of constructs A→O in the 1022-nucleotide-long intron 3 are depicted in the schematic) as shown and incubated in the absence (−) or presence (+) of 100 ng·mL−1 digoxin for 48 h was subjected to two-band RT-PCR. (C) WI-38 cells were transiently transfected with CLK1 minigene constructs carrying no deletion (no del), deletion of nucleotides 829–838 of intron 3 (Δ829–838), deletion of nucleotides 831–838 of intron 3 (Δ831–838) or mutations in nucleotides 833–841 of intron 3 (m833–841) and incubated in the absence of any reagent (−) or in the presence of 100 ng·mL−1 digoxin (D) or 50 μg·mL−1 genistein (G) for 48 h. RNA was then purified and subjected to two-band RT-PCR analysis. The nucleotides in bold font and underlined represent a potential SRSF3 binding site predicted by the online program splicing rainbow. The nucleotides in bold italic font represent the m833–841 construct bearing the mutated bases that replace the normal 833–841 nucleotides in intron 3.

Characterization of the RNA isolated from cells transiently transfected with CLK1 minigene constructs containing deletions of segments of the 1022-nucleotide-long intron 3 revealed that the transcript generated from the minigene bearing a deletion of nucleotides 829–838 (Δ829–838) of intron 3 lost its responsiveness to the splice-altering effect of digoxin (Fig. 6B). Analysis of this region of intron 3 using splicing rainbow software [33], which identifies splice-regulating elements, revealed the presence of a potential SRSF3 binding site spanning nucleotides 833–841 of this intron (Fig. 6C). To further characterize the possible role of this potential SRSF3 binding site on the digoxin-mediated inclusion of exon 4 of the CLK1 minigene derived transcript, additional constructs were developed in which either nucleotides 831–838 of intron 3 were deleted (Δ831–838) or the nucleotide sequence between bases 833 and 841 was mutated (m833–841) (Fig. 6C). Characterization of the RNA isolated from cells transiently transfected with the Δ829–838-, Δ831–838- or m833–841-bearing constructs revealed that the transcripts generated from these minigenes were no longer responsive to digoxin (Fig. 6C). In the light of our recent report demonstrating the ability of genistein to promote the inclusion of exon 4 in the CLK1 transcript [26], we examined the impact of genistein on the transcripts generated from these mutated constructs. A clear genistein-mediated enhancement of exon 4 inclusion was observed, demonstrating that, while the base changes spanning nucleotides 829–838 disrupt the sequence required for digoxin responsiveness, these changes have no effect on the genistein-mediated response (Fig. 6C).

Thus, we have identified a putative SRSF3 binding site as a potential cis element involved in the digoxin-mediated effect on splicing of exon 4 of CLK1. To characterize the effect of digoxin treatment on the SRSF3 encoded transcript and protein, we examined RNA and protein extracts from three cell lines incubated for 48 h in either the absence or presence of digoxin. qRT-PCR and western blot analyses clearly demonstrate a digoxin-mediated suppression of the level of the SRSF3 RNA and protein (Fig. 7A,B, respectively).

Figure 7.

Expression of SRSF3 RNA and protein is reduced in cells treated with digoxin. (A) An FD-derived cell line (GM04589), WI-38 and HEK293 cell lines were incubated either in the absence (−) or presence (+) of 100 ng·mL−1 digoxin for 48 h. RNA was purified and subjected to qRT-PCR analysis of the SRSF3 transcript. Primers used are shown in Table S1. Values obtained for the housekeeping gene, GAPDH, were used to standardize results. Experiments were done three times and the results presented represent the mean ± SD; *< 0.01. (B) Protein extracts were prepared from the cells described above and subjected to western blot analysis using a monoclonal antibody to SRSF3. Blots were scanned and the relative densities of the bands were determined with imagej software. The graph shows the mean ± SD from three independent experiments; *< 0.03. A typical blot is presented.

To determine whether SRSF3 acts as a trans factor in the regulation of the splicing of exon 4 of CLK1, cells transiently transfected with and expressing the CLK1 minigene were transfected with control or SRSF3 small interfering RNA (siRNA) and the effect on the production of the SRSF3 protein and on the splicing of exon 4 of CLK1 was examined. As seen in Fig. 8A, SRSF3 siRNA reduces the cellular presence of the SRSF3 protein to approximately 40% of that observed in the control siRNA-transfected cells. Examination of CLK1 minigene-derived and endogenous CLK1 transcripts in the SRSF3 siRNA-treated cells reveals an enhanced inclusion of exon 4 in both transcripts compared with the control siRNA-treated cells (Fig. 8B). The treatment of WI-38 cells transiently transfected with the m833–841-bearing CLK1 minigene construct, in which the minigene SRSF3 site is disrupted, with SRSF3 siRNA mediated no effect on the level of the mutated minigene-derived exon-4-containing CLK1 transcript, while the SRSF3 siRNA treatment resulted in an enhanced production of endogenous exon-4-containing transcript from the native CLK1 gene (Fig. 8C).

Figure 8.

Reduction of SRSF3 protein by RNA interference increases the production of exon-4-containing CLK1 transcript. (A) Protein extracts were prepared from WI-38 cells transfected with either control siRNA or SRSF3-specific siRNA and subjected to western blot analysis using a monoclonal antibody to SRSF3. The graph shows the mean ± SD from three independent experiments; *= 0.012. A typical blot is presented. (B) WI-38 cells were co-transfected with CLK1 minigene construct (wild-type) and siRNA (control siRNA or SRSF3-specific siRNA). After 24 h, RNA was purified and subjected to two-band RT-PCR of minigene-derived and endogenous CLK1 transcripts. (C) RNA isolated from WI-38 cells co-transfected with a CLK1 minigene construct carrying mutations in nucleotides 833–841 of intron 3 (intron 3 m833–841) and siRNA (control siRNA or SRSF3-specific siRNA) was subjected to two-band RT-PCR analysis.

Examination of the 364-nucleotide intron 19 sequence of the IKBKAP gene using the splicing rainbow software revealed the presence of several SRSF3-like elements (Fig. 9A). We therefore examined the effect of control and SRSF3 siRNAs on the production of the exon-20-containing IKBKAP transcript in the HEK293-FD minigene-bearing cell line. A clear suppression of the level of the SRSF3 protein (Fig. 9B) and a corresponding increase in the production of the exon-20-containing IKBKAP transcript (Fig. 9C) was observed in the cells transfected with SRSF3 siRNA.

Figure 9.

SRSF3 knockdown increases the production of exon-20-containing IKBKAP transcript, exon-3a-containing PDK1 transcript and exon-4-containing TP53I3 transcript. (A) Seven potential SRSF3 binding sites in five regions of intron 19 of IKBKAP predicted by splicing rainbow are shown in a bold, capitalized and underlined font. Three overlapping SRSF3 binding sites are located within nucleotides 140–151 (see * in figure). (B) Protein extracts prepared from HEK293 cells constitutively expressing the IKBKAP-FD minigene and transfected with either control siRNA or SRSF3-specific siRNA were subjected to western blot analysis using a monoclonal antibody to SRSF3. The graph shows the mean density of SRSF3 protein bands ± SD from three independent experiments; *= 0.003. (C)–(E) RNA was purified from HEK-FD cells transfected with siRNA as described in (B) and subjected to two-band RT-PCR of minigene-expressed exon-20-containing and -lacking IKBKAP transcripts (C, top), endogenous exon-3a-containing and -lacking PDK1 (D, top) and endogenous exon-4-containing and -lacking TP53I3 transcripts (E, top). RT-PCR products were analyzed on agarose gels with a typical experiment depicted. The RNA was also subjected to qRT-PCR analysis to detect relative levels of minigene-expressed exon-20-containing IKBKAP RNA (C, graph), endogenous levels of exon-3a-containing PDK1 (D, graph) and endogenous exon-4-containing TP53I3 transcripts (E, graph). The graphs in (C)–(E) show the mean relative levels of the respective transcripts ± SD from three independent experiments; *< 0.001 (C); *= 0.003 (D); *< 0.001 (E).

As digoxin treatment was also observed to facilitate exon inclusion in the PDK1 and TP53I3 transcripts in several cell lines (Fig. 3), the impact of transfection with SRSF3 and control siRNA on the inclusion of these exons was examined in HEK293-FD cells. A clear and statistically significant increase in the inclusion of these exons was observed in the cells transfected with SRSF3 siRNA, while no effect was detected in the cells transfected with control siRNA (Fig. 9D,E).

Discussion

The ability of chemical compounds to modulate heterogeneous nuclear ribonucleoprotein (hnRNP) expression and to facilitate the inclusion of exon 20 of the IKBKAP transcript bearing the FD-causing mutation [17] prompted us to embark on a high-throughput screening program designed to identify additional compounds that facilitate the production of the exon-20-containing IKBKAP transcript in cells bearing the FD-causing mutation. The screening of FDA-approved drugs from a compound library in cells transfected with an IKBKAP minigene construct bearing the IVS20+6T→C mutation revealed that the cardiac glycoside, digoxin, facilitated the production of the exon-20-containing transcript. Digoxin treatment of FD-derived fibroblast cells resulted in an increase in the level of the exon-20-containing IKBKAP transcript and a corresponding increase in the amount of the IKAP protein, at concentrations of digoxin observed in sera of patients receiving this drug for cardiac failure [30]. With the recent development of mouse models for FD [34, 35], the in vivo response to digoxin can be evaluated in these animals.

An expanded analysis of the impact of digoxin on the splicing process revealed that digoxin treatment of fibroblast and neuronally derived cells facilitates the inclusion of exon 4 in the CLK1 transcript and exons 3a and 4 in the PDK1 and TP53I3 transcripts, respectively. The responsiveness of the neuronal cells to the splice-altering activity of digoxin is significant as it reveals digoxin responsiveness in the cell type that is primarily impacted in individuals with FD [4]. In 2008, Stoilov et al. [36] demonstrated a digoxin-mediated inclusion of exon 10 in the microtubule-associated protein tau (MAPT) – encoded transcripts expressed in a neuroblastoma cell line, SHSY-5Y.

Initial efforts designed to characterize the cis elements and trans factors involved in the digoxin-mediated effect on the IVS20+6T→C mutation-bearing IKBKAP transcript were unsuccessful, as several mutations introduced into the IKBKAP minigene bearing the IVS20+6T→C mutation resulted in a loss of the constitutively expressed exon-20-containing transcript and a corresponding loss of digoxin responsiveness. This effect was not particularly surprising, as exon 20 of the IKBKAP transcript has been determined to be a ‘weak’ exon [31] and the loss of the constitutive production of the exon-20-containing transcript following the introduction of mutations into this sequence has been noted by others [32]. To overcome this obstacle, we investigated the cis elements and trans factors involved in the responsiveness of the CLK1 transcript to digoxin and found that mutations introduced into a putative SRSF3 binding site located in intron 3 disrupted the digoxin-mediated production of the exon-4-containing transcript. Further experiments revealed that digoxin treatment of cells resulted in a reduced cellular presence of both the SRSF3 transcript and protein and that transfection with SRSF3 siRNA enhanced inclusion of exon 4 in the minigene-derived and endogenously produced CLK1 transcripts and exon 20 in the minigene-derived IKBKAP transcript. The ability of the SRSF3 siRNA to reduce expression of SRSF3 protein and facilitate inclusion of exon 4 in both the endogenous and minigene-derived CLK1 transcripts and the inability of this siRNA to facilitate exon 4 inclusion in the CLK1 transcript generated from the minigene in which the putative SRSF3 site is mutated demonstrated that SRSF3 is acting as a trans factor that is binding a cis element overlapping or within the sequence encompassing nucleotides 833–841 of intron 3.

Using crosslinking and immunoprecipitation to enable high resolution identification of in vivo binding specificity and endogenous RNA targets of SRSF3, Änkö and coworkers recently identified a pentamer consensus sequence for the in vivo binding of murine SRSF3 and identified transcripts with significant SRSF3 crosslink sites [37]. As the amino acid sequences of the murine and human SRSF3 proteins are identical, it is to be expected that the human SRSF3 protein will bind to the same nucleotide sequences identified in murine cells. An examination of the sequence that includes nucleotides 833–841 of intron 3 of CLK1 (Fig. 6) reveals the presence of three overlapping pentamers (UUAAC, UAACA and CAUAA), which fit within the consensus sequence required for in vivo SRSF3 binding [37]. All three pentamers are disrupted by the base substitutions in the m833–841 CLK1 minigene construct. In addition, Änkö et al. demonstrate that CLK1 transcript is bound by SRSF3 [37]. It is interesting to note that bases 833–841 of intron 3 of human CLK1 are located within a 141-nucleotide stretch that is 100% homologous to the sequence of the analogous intron in the mouse. The presence of this sequence in the murine CLK1 gene may mediate the binding of SRSF3 to the CLK1 transcript. Examination of the putative SRSF3 sites in the 364-nucleotide-long intron 19 of the human IKBKAP gene (Fig. 9) reveals a number of the in vivo identified SRSF3 binding pentamer consensus sequences within these sites. The absence of any significant homology between the human and mouse intron 19 sequences (364 and 334 nucleotides, respectively) of the IKBKAP gene may explain why Änkö and coworkers did not observe the binding of SRSF3 to the murine IKBKAP transcript.

RNA splicing is a critical step in the expression of almost all eukaryotic genes. Nucleotide elements within exonic and intronic sequences of the pre-mRNA direct the splicing machinery. These cis elements include splicing enhancer and splicing suppressor sequences that promote or disrupt exon exclusion, respectively. The splicing-enhancing and splicing-suppressing elements are present in both intronic and exonic regions of the pre-mRNA. hnRNPs typically bind to exonic splicing suppressor (ESS) elements and disrupt spliceosome assembly and exon inclusion while SR proteins typically bind to the exonic splicing enhancer elements and promote spliceosome assembly and exon inclusion [38]. In some cases, however, SR proteins inhibit exon selection and hnRNPs activate splicing. Splicing activation and repression by SR and hnRNP proteins are dependent on the position of the binding sites for these splice-regulating factors [39]. SRSF3 has been reported to promote both exon inclusion and exon skipping. SRSF3-mediated exon inclusion has been observed for exon 11 of the insulin receptor [40], exon 10 of tau [41] and exon v9 of CD44 [42]. SRSF3-mediated exon skipping has been reported for exon 11 of the low-density lipoprotein receptor [43], exon 3b of the small GTPase Rac 1 [44] and the EDA exon of fibronectin [45].

Sterne-Weiler and coworkers recently reported that the siRNA-targeted reduction of SRSF3 results in exon inclusion in a minigene construct bearing the ESS sequence ACUAGG. As this ESS hexamer is created de novo in 83 different inherited disease mutations in 67 different genes, the authors suggest that SRSF3 may be responsible for the exon skipping and hence the disease-causing impact of these mutations [46]. The ability of digoxin to suppress SRSF3 RNA and protein levels supports a systematic investigation of the possible use of digoxin as a therapeutic modality for the diseases caused by these mutations.

In addition to the role aberrant RNA splicing plays in the manifestation of genetic diseases, it may also play a role in tumor progression and metastasis [47-49]. Characterization of the RNAs present in cancerous cells reveals the presence of many alternatively spliced RNAs [50]. Analysis of the alternatively spliced RNAs present in cancer cells indicates that the generation of these alternative spliced molecules is primarily due to changes in the expression of the trans-acting factors present in these cells [51]. Altered expression of the SR proteins has been observed in tumor cells and this changed expression pattern has been demonstrated to result in the generation of alternatively spliced transcripts [52]. Study of the SR proteins in development and growth of cancer cells has led to the observation that SRSF3 acts as a protooncogene that is critical for cancer cell growth [53] and that knockdown of SRSF3 expression results in growth arrest and apoptosis in cancer cells [53, 54]. While SRSF3 has been demonstrated to exhibit many biological effects, it is possible that this SR protein is acting as a protooncogene through its ability to modify the splicing process and cause the production of alternatively spliced transcripts.

Cardiac glycosides (e.g. digoxin, digitoxin, ouabain) are plant-derived compounds that have been used for centuries to treat congestive heart disease and atrial fibrillation. Epidemiological studies reporting improved outcomes in breast cancer patients receiving cardiac glycosides [55] and the reported lower incidence of leukemia, kidney and urinary tract cancer in individuals taking cardiac glycosides [56] prompted studies examining the effects of these compounds on the growth and viability of cancer cells. Numerous pathways and mechanisms have been examined to determine the mechanism by which these compounds exert cytotoxicity and evidence has been generated that implicates the cardiac-glycoside-mediated inhibition of glycolysis [57], an inhibition of N-glycan expression [58], inhibition of HIF-1 synthesis [59], inhibition of topoisomerase II [60], downregulation of c-MYC [61], increased production of reactive oxygen species [62], upregulation of death receptors DR4 and DR5 [63], altered membrane fluidity [64], inhibition of p53 synthesis [65] and induction of caspase-dependent apoptosis [66]. Our findings suggest the possibility that digoxin may in part mediate its anticancer effect through its ability to suppress expression of the protooncogene SRSF3 and thereby reduce the production of alternatively spliced transcripts.

The findings presented in this study suggest that the digoxin-mediated repression of SRSF3 expression plays a role in the digoxin-mediated inclusion of exon 20 in the IKBKAP transcript generated from the IVS20+6T→C bearing allele and support an investigation of the possible use of digoxin as a therapeutic modality for FD and perhaps other genetic diseases that are the result of exon skipping.

Materials and methods

Cell lines and reagents

The GM04589 and GM04899 cell lines, homozygous for the IVS20+6T→C FD-causing mutation, were obtained from the NIGMS Human Genetic Mutant Cell Repository (Coriell Institute for Medical Research, Camden, NJ, USA). WI-38 and HEK293 cells were purchased from the American Type Culture Collection (Manassas, VA, USA). The LA1-55n, BE(2)-C and BE(2)-M17 neuronal cell lines were kindly provided by R. A. Ross. All cell lines were maintained at 37 °C in 5% CO2 in MEM (fibroblast, HEK293 and WI-38 lines) or DMEM (neuronal lines) containing 10% fetal bovine serum. Purified digoxin, digitonin, digitoxin, digoxigenin, digitoxigenin, acetyldigitoxin, bufalin, ouabagenin and ouabain were purchased from Sigma-Aldrich (St Louis, MO, USA). The MicroSource Natural Products Library (Discovery Systems Inc., Gaylordsville, CT, USA) contains digoxin, several other cardiac glycosides and a large number of FDA-approved compounds within its 800-product collection; all compounds were provided at a 10 mm concentration in dimethylsulfoxide. The impact of the compounds on splicing was performed at the following concentrations for 48 h: 100, 30, 10 and 3 μm. Genistein was purchased from EMD Millipore (Billerica, MA, USA).

RNA preparation and RT-PCR

Total RNA was purified using RNeasy Plus Mini Kits (Qiagen, Germantown, MD, USA) according to the manufacturer's directions. 25 ng of total RNA was amplified in 20 μL RT-PCR reactions using OneStep RT-PCR Kits (Qiagen). One-step RT-PCR was carried out as follows: one cycle of 50 °C × 30 min and 95 °C × 15 min, followed by 42 cycles of 94 °C × 20 s, 58–60 °C × 30 s and 72 °C × 30 s, and then a final extension of 72 °C × 2 min. PCR products were analyzed on 2% agarose gels. Primers, used at a concentration of 0.5 μm, are shown in Table S1.

Quantitect SYBR Green RT-PCR Kits (Qiagen) were used for real-time RT-PCR (qRT-PCR) analysis of the relative quantities of the RNAs being studied. For this analysis, primers were designed to amplify the IKBKAP-encoded exon-20-containing (wild-type) and exon-20-lacking (mutant) transcripts as well as an exon 34–35-containing transcript that is unaffected by the FD-causing mutation (total). qRT-PCR of minigene-encoded exon-20-containing IKBKAP transcript (wild-type) was performed with one vector-specific primer and one primer in exon 20 of IKBKAP. qRT-PCR with one vector-specific primer and one primer in exon 19 of IKBKAP was performed to measure the total level of minigene-derived IKBKAP transcripts and used to standardize the results of minigene-encoded exon-20-containing IKBKAP transcript. In addition, primers were designed to quantitate the exon-4-containing CLK1 transcript, the exon-4-containing TP53I3 transcript, the exon-3a-containing PDK1 and the SRSF3 transcripts. Primers used are listed in Table S1. ABI PRISM 7000 and 7500 Sequence Detection Systems (Life Technologies, Grand Island, NY, USA), programmed as follows, were used to perform the qRT-PCR and analysis: 50 °C × 30 min and 95 °C × 15 min for one cycle, followed by 40 cycles of 94 °C × 15 s, 57–60 °C × 30 s and 72 °C × 30–34 s. Results obtained are in the form of threshold cycle, or CT, which refers to the PCR cycle at which the fluorescence of the PCR is increased to a calculated level above background. A change of 1.0 in CT, assuming 100% PCR efficiency, reflects a two-fold change in the starting amount of the RNA template that was amplified. To present relative amounts of PCR product obtained, results are expressed as ‘relative level’ of gene-specific transcript in treated cells compared with that present in untreated cells. To control for the amount of RNA present in the samples, RT-PCR amplification of GAPDH mRNA was performed and used to standardize the IKBKAP transcript results.

Two-band RT-PCR analysis of IKBKAP-, CLK1-, PDK1- and TP53I3-encoded RNAs in fibroblast and neuronal cells

Cells were seeded into 25 cm2 flasks 1 week (fibroblast cell lines) or 24 h (neuronal cell lines) before treatment. Fibroblasts were confluent when treated and neuronal cells were approximately 25% confluent when treated. After 48 h of treatment, cells were washed with NaCl/Pi and total RNA was purified as described above. Primers that spanned regions of alternative splicing in IKBKAP, CLK1, PDK1 or TP53I3 transcripts were used to simultaneously amplify both the wild-type and alternatively spliced transcripts in fibroblast and neuronal cells. The primers used are listed in Table S1. RT-PCR products were analyzed on 2% agarose gels.

Western blot analysis

Cells treated as described were washed twice with NaCl/Pi and lysed in NuPAGE LDS sample buffer (Life Technologies). Western blot analysis was performed essentially as described previously [67]. Equal amounts of protein fractionated on a 7% NuPAGE Tris–acetate gel (Life Technologies) were blotted onto nitrocellulose (Bio-Rad, Hercules, CA, USA) and probed overnight with a monoclonal antibody raised against amino acids 796–1008 of IKAP (Santa Cruz Biotechnology, Dallas, TX, USA) or a monoclonal antibody raised against amino acids 1–86 of SRSF3 (Sigma-Aldrich). The blots were then washed and probed with a goat anti-mouse IgG (H+L) conjugated to alkaline phosphatase (Promega, Madison, WI, USA), followed by detection with Western Blue Substrate solution (Promega). The loading of equal amounts of protein was confirmed by Coomassie Blue staining of a 7% NuPAGE Tris–acetate gel that was run in parallel. Quantification of bands on westerns was accomplished by scanning the blots and then determining the densities of the bands using imagej software (http://imagej.nih.gov/ij/).

Construction of minigenes

DNA derived from a cell line homozygous for the IVS20+6T→C FD-causing mutation was used to amplify DNA fragments containing IKBKAP genomic sequence from exon 19 to exon 21 and CLK1 genomic sequence from exon 3 to exon 5 with PfuUltra II Fusion HS DNA polymerase (Agilent Technologies, Santa Clara, CA, USA). Primers used are listed in Table S1. These DNA fragments were inserted into the pcDNA3.1/V5-His-Topo vector (Life Technologies) according to the manufacturer's directions. The resulting minigenes were purified using a QIAprep Spin Miniprep Kit (Qiagen) and sequenced (Genewiz, South Plainfield, NJ, USA).

Transfection of the CLK1 and IKBKAP-FD minigenes

For transient transfection experiments, WI-38 cells were seeded in 24-well plates 1 day before transfection with appropriate cell numbers to achieve approximately 80% confluence at the time of transfection. The cells were transfected with 400 ng of the minigene constructs in the presence of Lipofectamine 2000 (Life Technologies) in Opti-MEM-reduced serum medium according to the manufacturer's directions. Six hours after transfection, the medium was removed and replaced with fresh culture medium with or without the compound being studied. Following a 48-h incubation period, the cells were harvested and their RNA was purified as described above.

To generate HEK293 cells constitutively expressing the IKBKAP-FD minigene, HEK293 cells seeded for 24 h in a 25 cm2 flask were transfected with 2.4 μg of the IKBKAP-FD minigene construct in the presence of Lipofectamine 2000 (Life Technologies) in Opti-MEM-reduced serum medium according to the manufacturer's directions. Six hours after transfection, the medium was changed to MEM containing 10% fetal bovine serum and 1 mg·mL−1 G418 (geneticin) (Life Technologies). Surviving cells isolated following a 4-week exposure to G418 were permanently maintained in MEM containing 10% fetal bovine serum and 1 mg·mL−1 G418.

Splicing assay for IKBKAP and CLK1 minigenes

HEK293 cells constitutively expressing the IKBKAP-FD minigene were seeded into wells 24 h before exposure to the compounds being studied. For experiments in which cells were transiently transfected with the CLK1 or IKBKAP-FD minigene, cells were treated with digoxin 6 h after transfection. Following a 48-h incubation period, the cells were harvested and their RNA purified as described above. RT-PCR was performed as described above using primers (see Table S1) that specifically amplified either the minigene-derived or the endogenous IKBKAP or CLK1 transcripts.

Site-directed mutagenesis of minigene constructs

Mutations in the minigene constructs were introduced by site-directed mutagenesis using PfuUltra II Fusion HS DNA polymerase (Agilent Technologies) with primers containing the desired mutations (see Table S2). The reactions were carried out as follows: 95 °C × 30 s, followed by 20 cycles of 95 °C × 30 s, 55 °C × 1 min and 68 °C × 4 min. The reactions were then digested with Dpn I (New England Biolabs, Ipswich, MA, USA) and 2 μL from each reaction were used to transform JM109 competent cells (Promega). Plasmids were purified from bacterial colonies using QIAprep Spin Miniprep Kits (Qiagen) and sequenced to confirm mutations (Genewiz).

Reduction of SRSF3 expression by RNA interference

WI-38 cells seeded into 12-well plates 24 h before transfection were co-transfected with 800 ng of CLK1 minigene construct and either (a) 20 pmol control siRNA (non-targeting siRNA) or (b) human SRSF3-targeting siRNA (siGENOME SMART pool for human SRSF3; Thermo Fisher Scientific, Pittsburgh, PA, USA), or were transfected with control or SRSF3-targeting siRNA alone, in the presence of 4 μL Lipofectamine 2000 (Life Technologies). HEK293 cells constitutively expressing the IKBKAP-FD minigene seeded into 12-well plates 24 h before transfection, such that wells were 50% confluent at the time of transfection, were incubated with 20 pmol control siRNA or human SRSF3-targeting siRNA in the presence of 2 μL Lipofectamine 2000. Six hours after transfection of all cells, the medium was removed and replaced with fresh culture medium. Cells were harvested after 24 h and RNA was purified and subjected to two-band RT-PCR analysis of transcripts appropriate to each experiment as described above. Two-band RT-PCR products were analyzed on 2% agarose gels. qRT-PCR analysis of the relative quantities of minigene-encoded or endogenously encoded transcripts appropriate to each experiment was performed as described above.

Acknowledgements

This work was funded by grants from Familial Dysautonomia Now, the Cure FD Foundation and the Eric Alterman Foundation for F.D. Cure. We also acknowledge the generous support of NYC Council Member G. Oliver Koppell and his securing of funds from the New York City Council to support this research effort.

Ancillary