Stress alters the subcellular distribution of hSlu7 and thus modulates alternative splicing

Most RNA-polymerase-II transcripts of higher eukaryotes are synthesized as precursor mRNA (pre-mRNA) containing exons and introns. During the splicing reaction, introns are removed and exons are ligated to form an mRNA, which is transported to the cytoplasm and translated to form a protein (Burge et al., 1999; Jurica and Moore, 2003; Rappsilber et al., 2002). Four short degenerate sequences are essential for splicing; the 5′ and 3′ splice sites (5′ss and 3′ss) (GU and AG, respectively) located at the 5′ and 3′ ends of the intron, and the polypyrimidine tract and branch-site sequence, which are located upstream of the 3′ss (Brow, 2002). Alternative splicing, in which different subsets of exons or splice sites in a given pre-mRNA are chosen, is a major mechanism that diversifies the genetic information by producing more than one type of mRNA from a single gene (Graveley, 2001). Exon choice is determined by different nuclear protein concentrations that vary among tissues (Hanamura et al., 1998) and cell types (Jensen et al., 2000) or during development (Lallena et al., 2002; Mahe et al., 2000; Smith and Valcarcel, 2000; Wang and Manley, 1995), and these can be altered by various cellular circumstances such as physiological stimuli (van der Houven van Oordt et al., 2000), environmental effects (Meshorer et al., 2002) or phosphorylation state (Xie et al., 2003). Aberrant regulation of splicing has been implicated in an increasing number of human diseases (Philips and Cooper, 2000; Stamm, 2002; Stoilov et al., 2002), including cancer (Hastings and Krainer, 2001; Modrek and Lee, 2002; Stoilov et al., 2002).

The splicing protein Slu7 (Frank and Guthrie, 1992) (synergistic lethal with U5 snRNA) and its human homolog hSlu7 (Chua and Reed, 1999a) have been shown to play important roles in 3′ss selection during step two of splicing. Human nuclear extract immunodepleted of hSlu7 reduces the affinity of the 5′ exon to the spliceosome, leading to different selection of AGs at the 3′ss. This suggests a role for hSlu7 in holding the 5′ exon within the spliceosome for the correct 3′ss attack (Chua and Reed, 1999b). Both Slu7 and hSlu7 act in selection of the distal 3′ss when the distance between the branch point and the 3′ss is greater than 7 or 23 nucleotides, respectively (Chua and Reed, 1999b; Zhang and Schwer, 1997). Although many attempts have been carried out to define the list and function of each protein involved in the splicing reaction, the comprehensive picture is still far from understood.

We have recently identified by bioinformatics and mutational analyses three functional domains of the hSlu7 protein that have distinct roles in its subcellular localization – a nuclear localization signal (NLS), a zinc-knuckle motif and a lysine-rich region (Shomron et al., 2004). The zinc-knuckle motif overlaps the NLS to form a unique functional structure that is required not for hSlu7's entrance into the nucleus but rather to maintain hSlu7 inside it, preventing its shuttle back to the cytoplasm via the chromosomal region maintenance 1 (CRM1) pathway. The selective synchronized translocation of proteins between the nucleus and the cytoplasm by eukaryotic cells provides a powerful mechanism for the regulation of gene expression and for many other cellular processes (Gorlich and Kutay, 1999; Mattaj and Englmeier, 1998; Weis, 1998), including the regulation of alternative splicing (van der Houven van Oordt et al., 2000; Xie et al., 2003). Identifying the signals causing this nuclear-cytoplasmic translocation is a key step in understanding the role of these proteins and the regulation of their substrates. We have shown that the zinc-knuckle motif of hSlu7 determines the cellular localization of the protein through a nucleocytoplasmic-sensitive shuttling balance (Shomron et al., 2004). At this point, we asked whether physiological conditions can shift the endogenous nuclear hSlu7 concentrations towards cytoplasmic accumulation. Therefore, we examined the effect of stress on hSlu7 cellular localization and whether these concentrations are crucial for alternative splicing or 3′ss selection during the splicing reaction. We found that two stress conditions, ultraviolet-C light (UV-C) and heat, alter the cellular distribution of hSlu7, which is dependent mostly on the Jun N-terminal kinase (JNK)-kinase cascade. The differential cellular distribution of hSlu7 affects alternative splicing and 3′ss selection, although the protein is not essential for cell viability. Altogether, these results suggest that the stress-induced nuclear-cytoplasmic shift might be a spatial and temporal regulatory mechanism by which hSlu7 protein levels are regulated within the nucleus. This process affects the alternative splicing inclusion/skipping ratio of target exons.

HeLa, 293T, HepG2 and U2OS cells were grown on cover slides in 24-well plates or 100 mm culture dishes and maintained in Dulbecco's modified Eagle's medium (DMEM) with 10% fetal calf serum (FCS), 0.29 mg ml^–1 L-glutamine, 100 U ml^–1 penicillin, 0.1 mg ml^–1 streptomycin and 1 U ml^–1 nystatin at 37°C in a humidified atmosphere with 5% CO₂. HeLa, HepG2 and U2OS cells were used for immunofluorescence studies and 293T cells were used for transfection experiments owing to their high transfectability (>95%). Transfection with ADAR minigenes (Lev-Maor et al., 2003) and a JNK1-cDNA-containing plasmid (provided by R. Seger, Weizmann Institute of Science, Rehovot, Israel; a mutant JNK1-containing plasmid was provided by T. Kallunki, Institute for Cancer Biology, Copenhagen, Denmark) was performed using FuGENE6 (Roche) according to the manufacturer's protocol. Transcription inhibition was carried out by administering 5 μg ml^–1 actinomycin-D (Sigma) in serum-free DMEM for 3 hours at 37°C. Similarly, puromycin (Sigma) was used at 300 ng ml^–1 to inhibit protein synthesis, and SB203580 or SP600125 (Sigma) to inhibit the p38- or JNK-mediated phosphorylation pathways, respectively. The latter reagents were added for 1 hour before UV-C treatment. For stress analysis, exponentially growing cells were mock-treated or treated with heat (42°C for 1 hour), 0.3 J cm^–2 or 0.9 J cm^–2 UV-C light (Vilbert-Lourmat, Marne La Vallée, France) (except for JNK1-activation experiments, which were irradiated at 0.01 J cm^–2), H₂O₂ (up to 100 mM), sorbitol (up to 600 mM), cisplatin (up to 1 mM), or Neocarzinostatin (NCS; up to 200 ng ml^–1). Cells were left to recover at 37°C for 3 hours and then fixed for immunofluorescence studies. The UV-C irradiation was administered without cell medium, which was subsequently added for the recovery period.

A recombinant fusion between glutathione-S-transferase (GST) and hSlu7 was produced as described (Shomron and Ast, 2003). This protein was used for polyclonal antibody production in rabbits according to standard procedures (Green and Manson, 1998).

Following stress treatment, cells were washed with PBS and fixed for 10 minutes in 4% paraformaldehyde (PFA). Permeabilization was performed with 0.5% Triton X-100 in PBS for 10 minutes, then cells were incubated for 5 minutes with 0.1 mg ml^–1 4,6-diamidino-2-phenylindole (DAPI) to mark the nucleus. To observe hSlu7, Sm or JNK1 proteins, the cells were incubated with anti-hSlu7 (Shomron and Ast, 2003), anti-Sm (provided by J. Steitz, Yale University, New Haven, CT) or anti-hemagglutinin (anti-HA) (fused to JNK1 cDNA; provided by R. Seger), respectively, for 1 hour and then with rhodamine-conjugated anti-IgG or FITC-conjugated anti-IgG (Jackson ImmunoResearch) in blocking solution [5% FCS, 5% bovine serum albumin (BSA) in PBS] for half an hour. Fluorescent images were observed under a 410 Carl Zeiss confocal laser-scanning microscope (CLSM410) at the 365 nm, 488 nm and 568 nm laser lines. Images were obtained using the same parameters.

For determination of hSlu7 localization, multiple fields were examined to count at least 200 cells. Images were acquired under identical conditions, ensuring that the maximal signal was not saturated. Acquisition of images and measurement of speckles intensity was performed using Adobe Photoshop, TINA and analySIS (Soft Imaging System) software. Most of the results represent values obtained from at least three separate experiments, and the results are average values. Small nuclear ribonucleoprotein (snRNP) localization was determined similarly.

Cells were grown in 100 mm culture dish and transfection was performed using 4 μg indicated plasmid DNA. After 48 hours, total cytoplasmic RNA was extracted using TRI reagent (Sigma) followed by treatment with 2 U DNase (RNase-free; Ambion). First-strand oligo(dT)-primed cDNA synthesized with reverse transcriptase (RT) from avian myelobstosis virus (RT-AMV, Roche) from 2 μg total RNA was amplified with High Fidelity Taq polymerase (Roche) and ADAR2, DDO, GAPDH and E1A primers for 30 cycles, consisting of 94°C for 30 seconds, 60-65°C for 45 seconds and 72°C for 1 minute. The products were separated on 2% agarose gels and confirmed by sequencing.

Lysis buffer (50 mM Tris 7.5, 1% NP40, 150 mM NaCl, 0.1% SDS, 0.5% deoxycholic acid, protease inhibitor cocktail and phosphatase inhibitor cocktails I and II; Sigma) was used to dissolve the total proteins extracted from the cells, which were then collected after cold centrifugation for 30 minutes at 17,000 g. Total protein concentrations were measured using BioRad Protein Assay (BioRad). Proteins were separated in 12% SDS-PAGE and then electroblotted onto a Protran membrane (Schleicher and Schuell). The membranes were probed with anti-hSlu7, anti-hPrp18 (provided by D. Horowitz, Uniformed Services University of the Health Sciences, Bethesda, MD), anti-hPrp8 (p220; provided by M. Moore, Brandeis University, Waltham, MA), anti-glyceraldehyde-3-phosphate-dehydrogenase (anti-GAPDH) (provided by I. Yron, Tel Aviv University, Tel Aviv, Israel), anti-phospho-p38-specific (Santa Cruz Biotechnology) or anti-phospho-Jun-specific (US Biological) antibody, followed by the appropriate secondary antibody. Immunoblots were visualized by enhanced chemiluminescence (Lumi-Light Western Blotting Substrate; Roche) and exposure to X-ray film.

RNA interference (RNAi) was performed using the pSUPER plasmid (provided by R. Agami, The Netherlands Cancer Institute, Amsterdam, The Netherlands) as described in (Brummelkamp et al., 2002) with the following modifications. 21-nucleotide double-stranded DNA with a sequence homologous to the zinc-knuckle region of hSlu7 mRNA, containing sticky-end restriction enzymes on either side, was chemically synthesized and purified by Sigma. Transfection of 3 μg cloned pSUPER plasmids was performed using FuGENE6 (Roche) in 100 mm culture as described above.

We produced polyclonal antibodies directed against hSlu7 and used them to detect the cellular localization of hSlu7. The hSlu7 protein is localized to the nucleus in a distinct speckled pattern (Shomron et al., 2004), which is similar to other splicing proteins. However, the absence or presence of Slu7/hSlu7 was reported to affect 3′ss choice in vitro (Chua and Reed, 1999a; Frank and Guthrie, 1992) and our previous observations indicated that hSlu7 is an alternative splicing regulatory protein that can also cause an effect on 3′ss selection in vivo (Lev-Maor et al., 2003; Shomron et al., 2004). As a result, we wanted to determine whether altering the physiological environment of the cell changes hSlu7 nuclear concentrations [as reported for other splicing proteins (van der Houven van Oordt et al., 2000; Xie et al., 2003)] and whether a different subcellular distribution of hSlu7 can affect alternative splicing.

HeLa cells were exposed to various stress treatments – heat shock or UV-C irradiation, followed by a recovery period of 3 hours. Cells were then fixed, stained with anti-hSlu7 and DNA stained with DAPI (Fig. 1Ai-v). Both heat shock and UV-C irradiation (at two dose levels) caused a shift of hSlu7 proteins from exclusively nuclear to nuclear and cytoplasmic staining. The speckled nuclear pattern observed without stress (Fig. 1Ai) differs from the dispersed nuclear pattern after stress (Fig. 1Aii-v). The latter is composed of many small hSlu7 centers scattered evenly around the nucleus, similar to those seen in other proteins following stress (Dye and Patton, 2001). As a control, staining with anti-Sm antibody against snRNPs and with anti-hPrp8 antibody (U5-specific protein – p220) demonstrated that this effect was specific and not caused by the generation of nuclear leakage, because no cytoplasmic accumulation of Sm and hPrp8 proteins were observed at either dose level (Fig. 1B and data not shown). Similar experiments using only the second antibodies gave no background fluorescence (data not shown). Addition of transcription and protein-synthesis inhibitors (actinomycin-D and puromycin, respectively) before the irradiation had no effect on hSlu7 cytoplasmic accumulation (Fig. 1C). This demonstrates that the accumulation of hSlu7 in the cytoplasm does not represent newly synthesized protein concomitant with nuclear import blockage. A time-scale analysis of hSlu7 export out of the nucleus following 0.3 J cm^–2 UV-C irradiation shows that the effect begins 120 minutes after irradiation and reaches its maximum level (48%) after 300 minutes (Fig. 1D). Recovery is not seen even after 24 hours (data not shown).

The slightly different subcellular distribution observed after 0.3 J cm^–2 and 0.9 J cm^–2 UV-C irradiation suggests that it is a UV-dose-dependent response and not a general failure of any nuclear localization signals. This implies that the UV response is a unique effect of individual proteins to a stress-induced environment. The cytoplasmic accumulation is not cell-type specific, because UV-C irradiation caused similar subcellular distribution in three other cell lines (293T, HepG2 and U2OS; data not shown). The above results indicate that stress causes a shift in the cellular localization of hSlu7 – from nuclear to cytoplasmic accumulation. Other stress conditions, such as peroxidase, osmotic shock, cisplatin and neocarzinostatin, had no or very little effect on hSlu7 cytoplasmic accumulation (Fig. 1D and data not shown).

To test whether the cellular localization of hSlu7 has a functional significance in alternative splicing, we used two minigenes containing the genomic sequence of the gene encoding adenosine deaminase, ADAR2, from exon 7 to exon 9, in which exon 8 is an alternatively spliced Alu exon (Lev-Maor et al., 2003). The 3′ end of the intron preceding exon 8 is composed of a polypyrimidine tract of 18 bases, followed by a 3′ss motif of GAGACAG. In the 3′ss motif, the distal AG is selected (bold), and there is a delicate interplay between the two AGs. The proximal AG (which is not selected as a 3′ss; underlined) is essential to weaken the selection of the distal AG, thus maintaining alternative splicing. The 4-nucleotide distance between the proximal and distal AGs also ensures the mode of alternative splicing: Increasing that distance leads to exon 8 skipping, and exon 8 inclusion can be restored by a high concentration of hSlu7 when the distance between the two AGs is over 8 nucleotides (Lev-Maor et al., 2003). We therefore reasoned that, if hSlu7 proteins partially shift from the nucleus to the cytoplasm, this might alter the alternative splicing pattern. Thus, we transfected cells with the plasmids containing the ADAR2 or ADAR2+10 minigene. After 48 hours, UV-C stress was induced followed by a 3 hour recovery period and total RNA extraction. The splicing pattern was examined by RT-PCR using specific primers to the flanking exons (Fig. 2A, arrows). In the absence of stress conditions, the average inclusion/skipping ratio of the alternatively spliced Alu exon was 35%:65% (and the distal AG 3′ss was selected; this was confirmed by sequencing). In the ADAR2+10 minigene, alternative exon inclusion was a rare event observed in 11% of the splicing events and the proximal 3′ss was selected [average of three experiments (Lev-Maor et al., 2003)]. The alternative splicing pattern of both ADAR2 and ADAR2+10 minigenes was affected by UV-C irradiation, 21% and only 2% Alu-exon inclusion was observed, respectively (average figures of three experiments; Fig. 2B, lanes 2-5, quantified in Fig. 2C). This suggests that native hSlu7 nuclear levels are required for alternative splicing and 3′ss selection of some genes (which still remain to be elucidated). It also suggests that lower concentrations of hSlu7 in the nucleus (and probably in the spliceosome) leads to suboptimal 3′ss selection and thus to exon skipping.

To test whether the different alternative splicing patterns were caused by reduced nuclear hSlu7 levels we used RNAi (for a review, see Novina and Sharp, 2004) to knockdown hSlu7 cellular concentrations. Cells with reduced levels of hSlu7 (Fig. 2D) showed a similar effect on alternative splicing of the ADAR2 and ADAR2+10 minigenes, thus supporting the notion that the alternative splicing pattern caused by UV-C irradiation is attributed to the reduced levels of nuclear hSlu7 concentrations (Fig. 2B, lanes 6-9; quantified in Fig. 2C; average figure of three experiments). We note that a control plasmid and other unrelated RNAi plasmids had no or little effect on the alternative splicing pattern (Fig. 2B,C and data not shown). In addition, reduced levels of hSlu7 had no effect on hPrp18 protein levels (Fig. 2D; see also Discussion).

To test whether the effect on splicing acts on an endogenous gene as well, we assayed the alternative splicing of the D-aspartate-oxidase (DDO) gene. DDO has a weak 3′ss upstream of its alternative exon 3 (see Fig. 2A for the sequence of the 3′ss). UV-C irradiation and RNAi treatment leading to reduction in hSlu7 nuclear concentration affect the inclusion/skipping ratio, changing it from mostly exon inclusion to mostly exon skipping (Fig. 2E, quantified in Fig. 2F). These results indicate that the reduced nuclear levels of hSlu7 are not restricted to the mutated minigene and might imply a broad effect of hSlu7 levels on alternative splicing of other exons as well.

To ensure that the effect of hSlu7 is on alternative 3′ss selection, we looked at the splicing pattern of the adenovirus E1A gene, which uses three alternative 5′ss spliced to a single 3′ss (Fig. 3A). If UV-C irradiation caused hSlu7 partial shift to the cytoplasm and this subsequently caused a different alternative splicing pattern, 5′ss selection should not be altered. In fact, E1A spliced products were selected differently only as a response to UV-C irradiation and not following hSlu7 knockdown or transfection of a control plasmid (Fig. 3B,C). We note that the effects on E1A splicing can be explained by stress-induced relocalization of hnRNP A1 (van der Houven van Oordt et al., 2000).

The signaling pathways activated by stress have been extensively studied. Stress stimuli such as UV-C irradiation activate two stress-response kinase cascades: JNK and p38 (Canman and Kastan, 1996; Karin and Delhase, 1998; Kyriakis and Avruch, 1996). We therefore analysed the potential role of the JNK (SAPK) and p38 kinase (MKK) pathways in controlling hSlu7 subcellular redistribution after stress signaling. We used two specific inhibitors, SB205823 to inhibit the p38 pathway and SP600125 to inhibit the JNK pathway. Following a 1 hour incubation with the inhibitor, cells were UV-C irradiated and then allowed to recover for 3 hours, followed by fixation and staining with anti-hSlu7 (Fig. 4A, quantified in Fig. 4B). Blocking of both pathways was effective in reducing phosphorylated states of p38 and JNK kinases by ten- and sixfold after 1 hour, and nine- and fourfold after 3 hours, respectively (Fig. 4C). A considerable reduction in cytoplasmic localization of hSlu7 was observed upon treatment with the p38 pathway inhibitor (from an average of 45% to 18%), whereas the JNK pathway inhibitor had a greater impact on reducing cytoplasmic accumulation (from an average of 44% to 1%). Thus, the JNK pathway showed a stronger involvement than the p38 pathway in preventing the cytoplasmic accumulation of hSlu7 after UV-C irradiation (even at the lowest concentration of 100 nM; Fig. 4A). This suggests that the JNK signaling pathway might possibly be the significant player in the subcellular distribution of hSlu7.

To test whether the alternative splicing activity of hSlu7 (Fig. 2) can be reversed by inhibiting the JNK pathway and thus inhibiting the shift in hSlu7 localization, we assayed alternative splicing of the DDO and ADAR+10 minigene, this time by also preventing hSlu7 shift to the cytoplasm using the JNK pathway inhibitor. As expected, preventing hSlu7 shift from the nucleus to the cytoplasm abrogates the effect caused by reduced hSlu7 nuclear concentrations (Fig. 4D and data not shown). This supports the link between stress induced by UV-C, activation of the JNK pathway and hSlu7 nuclear levels, which cause an effect on alternative splicing.

To asses whether JNK1 phosphorylation is accountable for the induction of hSlu7 relocalization and, as a consequence, affect alternative splicing, we increased the cellular levels of active JNK1 protein by transfecting the cells with a JNK1-cDNA-containing plasmid and then inducing its phosphorylation using mild UV-C irradiation (0.01 J cm^–2) (Hutter et al., 2002). By increasing active JNK1 levels we observed a shift in DDO alternative splicing towards the exclusion isoform, an opposite effect caused by the JNK inhibitor (Fig. 4D,E). Also, a significant boost in hSlu7 shuttling from the nucleus to the cytoplasm is observed (98% compared with 48%; Fig. 4Fi,iii) supporting the involvement of active JNK1 in hSlu7 localization. We note that the inclusion/skipping ratio of the DDO gene in these experiments is slightly different because of various UV-C doses used (compare lanes 3 in Fig. 4D,E) and that a mutant JNK1 protein had no effect (data not shown).

In this study, we identify physiological conditions that alter the nuclear concentration of second-step splicing protein hSlu7. Our results strongly suggest that UV-C stress stimuli trigger changes in the alternative splicing patterns of cellular genes by decreasing the nuclear concentration of hSlu7 through the modulation of its nuclear to cytoplasmic transport. Therefore, this spatial and temporal regulatory mechanism might be the method by which hSlu7 protein levels are regulated within the nucleus upon exposure to environmental stimuli.

Two known mechanisms control the levels of splicing proteins within the nucleus: a regulation on the translational level, for SF2/ASF (Hanamura et al., 1998); and exclusion from the nucleus, for PTB (Xie et al., 2003) and hnRNP A1 (van der Houven van Oordt et al., 2000). All three proteins have been shown to affect alternative splicing of specific genes. The latter mechanism, which was also observed for hSlu7, seems to provide a finer regulation of nuclear protein levels that control splicing.

Several of the stress condition's effects might seem to be contradictory. Osmotic shock and UV-C irradiation, among other factors, stimulate the p38 and JNK stress-response kinase cascades (Canman and Kastan, 1996). In our experiments, we observed that UV-C irradiation and not osmotic shock resulted in cytoplasmic accumulation. However, the events upstream of these pathways are poorly defined and this inconsistent stress-induced localization effect has been reported elsewhere (van der Houven van Oordt et al., 2000). It is also intriguing that UV-C irradiation and not cisplatin elicits a change in hSlu7 cellular localization, because they both exert similar physiological effects (Hayakawa et al., 2003). UV-C light produces bulky adducts, pyrimidine dimers and other photoproducts in DNA, whereas cisplatin forms primarily 1,2-intrastrand cross-links between adjacent purines in DNA and also introduces DNA 1,3-intrastrand cross-links, interstrand cross-links and monoadducts (Kartalou and Essigmann, 2001). These DNA obstructions must be removed or bypassed to prevent arrest of the DNA polymerase during replication and thus are expected to activate the same repair cascade. However, a recent report showed that the mechanisms by which cells sensitize to UV and cisplatin are likely to be different (Bowman et al., 2000). In the case of UV, global genome nucleotide-excision repair (GG-NER) is enhanced without affecting transcription-coupled nucleotide-excision repair (TC-NER), which is activated by cisplatin. We could not find any reasonable explanation why only UV-C and heat shock (out of the treatments we tested) affected hSlu7 cellular localization, yet we should note that different splicing proteins react differently to different stress stimuli and this can change considerably in different cell lines (van der Houven van Oordt et al., 2000).

In a previous study, van der Houven van Oordt and colleagues first showed that signal-transduction pathways regulate not only gene transcription but also the correct processing of mRNA (van der Houven van Oordt et al., 2000). These studies showed a remarkable link between a signaling cascade route (p38) and a splicing protein (hnRNP A1). They demonstrated that the subcellular localization of hnRNP A1 modulated by the p38 pathway in response to stress and caused an increase in cytoplasmic and a decrease in nuclear hnRNP A1 levels, leading to changes in alternative splicing activity. If the JNK signaling pathway is indeed involved in the subcellular distribution of hSlu7 (with or without p38), as suggested by our results, this represents a new facet to the hypothesis that the subcellular distribution of splicing factors is a regulated process (because, in the case of hSlu7, the JNK-kinase cascade appears to be mostly involved). The mechanism whereby hSlu7 cytoplasmic accumulation is regulated by JNK kinases is currently unknown, yet the observed considerable reduction in cytoplasmic localization of hSlu7 upon treatment with the p38 pathway inhibitor suggests a possible cooperation between these cascades in controlling hSlu7's cellular localization.

We assess the affect of hSlu7 nuclear concentration on alternative splicing of the endogenous gene DDO. In the presence of endogenous hSlu7, two mRNA isoforms are observed in 293T cell line. Following a decrease in hSlu7 nuclear concentrations, using either RNAi or UV-C, the amount of the inclusion isoform is reduced (Fig. 4D). Interestingly, following hSlu7 cDNA transfection, the shift of the alternative isoforms is towards the inclusion isoform. The two forms of the alternatively spliced DDO mRNA (DDO-1 and DDO-2) have been previously described (Setoyama and Miura, 1997). The inclusion isoform encodes the DDO-1 protein which consists of 341 amino acids. The skipped isoform, DDO-2, is identical to that of DDO-1, except for the absence of 59 amino acids covering residues 95-153 of DDO-1. Expression of human DDO-1 and DDO-2 showed that the 37-kDa DDO-1 protein is the active form and that the expression of DDO-2 produces an insoluble protein whose functional role could not be determined (Setoyama and Miura, 1997). It is suggested that the protein is involved either generally in animal development or specifically in the regulation of the central nervous system (Tedeschi et al., 1999). We have shown here that hSlu7 concentrations might have significant consequences on cellular processes.

A possible redundancy in function of hSlu7 with another protein might explain how the spliceosome can function in vivo under low levels or in the absence of hSlu7. The most probable candidate is hPrp18 (another second-step splicing protein) because, in yeast, these proteins have shown some redundant function in vitro (Zhang and Schwer, 1997). However, the fact that hPrp18 cellular concentration did not compensate for reduced hSlu7 nuclear levels after depletion of hSlu7 using RNAi suggests that their cellular levels are independent of one another. We should note that the anti-hPrp18 antibody identifies two bands: hPrp18 and a non-specific 55 kDa band that was reported by the producers of the antibody (Horowitz and Krainer, 1997) (Fig. 2D). Hence, there was no point in performing immunofluorescence experiments with these antibodies owing to the interaction with the two different proteins. HeLa and 293T cells knockdown of hSlu7 did not show any phenotype detectably different from wild-type and plasmid-only-transfected cells (data not shown), or any change in cell cycle (data not shown), indicating that cells with reduced levels of hSlu7 are viable. One might argue that using the RNAi mechanism might bias the splicing results because the RNAi system was shown to exert (in some cases) a broad, nonspecific downregulating effect (Jackson et al., 2003). However, the similar results obtained with UV-C irradiation argue against this idea.

Before this study, comprehensive research was performed on hSlu7 to elucidate its function during the second step of splicing and its 3′ss selection mechanism (Chua and Reed, 1999a; Chua and Reed, 1999b). Here, we have elucidated a possible mechanism by which hSlu7 can regulate alternative splicing or 3′ss selection in a specific environmental cellular context.

We thank V. Meltser, K. Molakandov, A. Shaul, Y. Vaisbuch, G. Nakash and N. Asia for technical assistance, K. Hollander for commenting on the manuscript, and the referees whose insightful suggestions helped to strengthen our conclusions. This work was supported by grants from the Israel Science Foundation (1449/04 and 717/01) and, in part, by a grant from the Israel Cancer Association, MOP India-Israel, FD Hope, and the Chief Scientist of Israel Health Ministry to G.A.