Poly(ADP-ribose) Binds to the Splicing Factor ASF/SF2 and Regulates Its Phosphorylation by DNA Topoisomerase I*

Human DNA topoisomerase I plays a dual role in transcription, by controlling DNA supercoiling and by acting as a specific kinase for the SR-protein family of splicing factors. The two activities are mutually exclusive, but the identity of the molecular switch is unknown. Here we identify poly(ADP-ribose) as a physiological regulator of the two topoisomerase I functions. We found that, in the presence of both DNA and the alternative splicing factor/splicing factor 2 (ASF/SF2, a prototypical SR-protein), poly(ADP-ribose) affected topoisomerase I substrate selection and gradually shifted enzyme activity from protein phosphorylation to DNA cleavage. A likely mechanistic explanation was offered by the discovery that poly(ADP-ribose) forms a high affinity complex with ASF/SF2 thereby leaving topoisomerase I available for directing its action onto DNA. We identified two functionally important domains, RRM1 and RS, as specific poly(ADP-ribose) binding targets. Two independent lines of evidence emphasize the potential biological relevance of our findings: (i) in HeLa nuclear extracts, ASF/SF2, but not histone, phosphorylation was inhibited by poly(ADP-ribose); (ii) an in silico study based on gene expression profiling data revealed an increased incidence of alternative splicing within a subset of inflammatory response genes that are dysregulated in cells lacking a functional poly(ADP-ribose) polymerase-1. We propose that poly(ADP-ribose) targeting of topoisomerase I and ASF/SF2 functions may participate in the regulation of gene expression.

DNA topoisomerase I (topo I) 2 is a constitutively expressed multifunctional enzyme that localizes at active transcription sites (1,2). Its best known function is to control the topological state of DNA by relieving torsional stress that is generated following DNA strand separation during transcription, replication, and repair (3)(4)(5). The catalytic mechanism involves the formation of a DNA⅐topo I complex (cleavage complex) with the enzyme being covalently bound to the 3Ј-end of the cleaved DNA strand through a tyrosine-phosphate ester bond. Cleavage complexes are usually short-lived; their stabilization by compounds of the camptothecin (CPT) family of anticancer drugs may cause DNA strand break accumulation and eventually lead to cell death. Human topo I can relax both negative and positive supercoils by controlled rotation of the DNA strand downstream of the cleavage site followed by break resealing and restoration of an intact DNA duplex.
In addition to relaxing supercoiled DNA, human topo I also plays a major role in pre-mRNA splicing, being endowed with a protein kinase activity targeted at a group of splicing factors of the serine-arginine (SR)-rich protein family (6). SR-proteins function both as components of the basal RNA splicing machinery and as regulators of alternative splicing (7,8). Moreover, SR-proteins are involved in the control of mRNA transport and stability (8,9) and contribute to the maintenance of genomic stability (10). SR-proteins are structurally characterized by having one or two domains that include a RNA recognition motif (RRM) at the N terminus, and an SR-rich C-terminal domain (RS domain) containing a variable number of SR dipeptidic repeats. Phosphorylation at serine residues in such sequences regulates SR-protein functions as well as their subnuclear localization (7-9, 11, 12). Besides topo I, other kinases are also involved in SR-protein phosphorylation; these include the SR-protein-specific kinases 1 and 2 (SRPK1 and SRPK2) and the cell cycle-dependent dual specificity kinase Clk/Sty (13). Moreover, Akt/protein kinase B phosphorylation of SR-protein family members appears to play a critical role in signal transduction pathways linking extracellular stimuli (hormones and mitogens) to changes in gene expression via regulation of alternative splicing and mRNA translation (14,15).
Depletion of topo I results in the hypophosphorylation of SR-proteins and impaired exonic enhancer-dependent splicing (16). Likewise, inhibition of topo I-dependent SR-protein phosphorylation by indolocarbazole antitumor drugs has been shown to interfere with the spliceosome assembly pathway, leading to altered gene expression and eventually to cell death (17).
Thus, human topo I may either bind DNA and catalyze its relaxation or bind SR-proteins and ATP and play the role of a kinase. The two activities are mutually exclusive, and they are most likely the functional expression of distinct conformational states (18). What regulates such structural and functional transitions is unknown. Human topo I is a member of the poly-* This work was supported by the Swiss National Science Foundation, the Polish Ministry for Science and Higher Education (Grant N031 053 32/1969), and the Italian Ministry for Education, University and Research. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. 1 To whom correspondence should be addressed: (ADP-ribose)-binding family of proteins (19). Poly(ADP-ribose) (PAR) is the product of a class of enzymes known as PAR polymerases (PARPs) (20). PARPs utilize NAD ϩ as a source of ADP-ribose units and catalyze the covalent modification of a number of proteins (heteromodification), including themselves (automodification), with an array of linear or branched ADPribose chains of variable lengths; these polymers are then degraded by a specific PAR glycohydrolase, thus making the reaction reversible (20,21). PARP-1 and PARP-2, the best known members of the PARP family, depend on DNA strand breaks for activity and are responsible for most PAR synthesized in the nucleus of eukaryotic cells both under physiological and DNA damage conditions (20,21). Protein targeting by PARP-bound polymers via non-covalent, yet specific interactions, is emerging as an important regulatory mechanism for diverse biological functions, including transcription, DNA damage signaling and checkpoint activation, proteasomal histone degradation, and mitotic spindle formation (22,23). topo I bears three PAR-binding sites localized in domains that are critical for the catalytic activity of the enzyme on DNA and for its regulation (24). In fact, PAR has a dual effect on topo I: it inhibits DNA cleavage (thus preventing initiation of new catalytic cycles) while it stimulates the re-ligation activity of the enzyme blocked in a ternary complex with nicked DNA and CPT (thus counteracting the poisoning effect of the drug) (24). In this study we addressed the question of whether PAR could also affect topo I kinase activity and/or act as a molecular switch of distinct topo I functions. ASF/SF2, a prototype of the SR-protein family, was used as a specific substrate for the topo I kinase activity.

MATERIALS AND METHODS
Purified human topo I was obtained from TopoGen. This enzyme undergoes spontaneous conversion into a 70-kDa form lacking the N-terminal domain (⌬N-topo I). Full-length Histagged human topo I expressed in a baculovirus system was from Jena Bioscience (distributed by Alexis). topo I homogeneity was Ͼ95%. GST-SRPK1 was from Upstate. Highly purified recombinant PARP-1 and PARP-2 were purchased from Alexis. Protein-free, affinity-purified [ 14 C]PAR, [ 32 P]PAR, and 32 P-5Јend-labeled ds-oligonucleotide, used as a topo I substrate in DNA cleavage assays, were prepared as previously described (24). All experiments were repeated at least three times and confirmed with different preparations of topo I, SR-proteins, PAR.
topo I Activity Assays-topo I DNA cleavage activity was assayed using a 32 P-5Ј-end-labeled ds-oligonucleotide containing a single topo I binding/cleavage site, as previously reported (24). Reaction mixtures (15 l) were assembled on ice and contained 50 mM Tris-HCl, pH 8.0, 0.1 mM EDTA, 1 mM DTT, 0.01% Triton X-100, 20 M CPT, 0.2 mg/ml bovine serum albumin, 2% glycerol, human topo I (0.2-0.27 pmol). Reaction was started by addition of 32 P-5Ј-end-labeled ds-oligonucleotide (0.02-0.04 pmol, 0.5-1.1 ϫ 10 6 dpm/pmol) and carried on for 10 -20 min at 37°C. After termination by addition of concentrated Laemmli buffer, cleavage complexes were separated by SDS-PAGE on 7.5% polyacrylamide gels and visualized by autoradiography. When present, affinity-purified PAR or ASF/SF2 were added to the reaction mixture at the amounts indicated in the figures. topo I kinase activity was assayed in the same buffer as indicated above in the presence of 10 mM MgCl 2 and 1 M ␥-[ 32 P]ATP (0.3 ϫ 10 6 dpm/pmol), in a final volume of 15 l. 0.2 pmol of yeast-expressed human recombinant topo I and 0.7-2.8 pmol of recombinant His-ASF/SF2 were used in each assay. After incubation at 37°C for 20 min, reaction was stopped by adding concentrated Laemmli buffer; ASF/SF2 phosphorylation was visualized by autoradiography, after electrophoretic separation on 10% polyacrylamide gels (SDS-PAGE). ASF/SF2 phosphorylation by SRPK1 was assayed under the same conditions using 0.05 pmol of the recombinant kinase.
PAR Binding Assay-The PAR binding assay was carried on essentially as described by Panzeter et al. (26). Proteins were immobilized on nitrocellulose either by Western blotting after electrophoretic separation on polyacrylamide gels, or by slot blotting. Duplicate samples were either gold/Coomassie Blue stained for protein visualization or probed with [ 32 P]PAR, washed with 10 mM Tris-HCl, 0.15 M NaCl, 0.05% (v/v) Tween 20, pH 7.4 (TBST), and analyzed by autoradiography. Where indicated, 0.5 M NaCl was added to the washing buffer (0.5 M NaCl-TBST).
Protein Extracts-HeLa S3 cells were cultured in complete Dulbecco's modified Eagle's medium under standard culturing conditions and harvested at a subconfluent stage to be used for nuclear and cytoplasmic protein extract preparations, as previously described (27).
ASF/SF2 or histone phosphorylation by endogenous kinase(s) was assayed in the conditions described by Rossi et al. (6) using 0.5 l of extract (1 g of protein), in the presence or absence of PAR. Phosphorylated proteins were analyzed by SDS-PAGE and autoradiography.
Quantitative Analyses-Autoradiographic bands were quantified by scanning densitometry using the GS-710 Bio-Rad densitometer and the image analysis software QuantityOne (Bio-Rad). Data are expressed as mean of at least three independent experiments Ϯ S.D.
Bioinformatics-Peer-reviewed literature was surveyed for comparative genome-wide studies of the transcriptomes of parp-1 knock-out mammalian cells and their wild-type counterparts, both under physiological conditions and after exposure to cytotoxic stimuli. Then, gene expression profiling data were matched against the Alternative Splicing Annotation Project II data base (28). Only genes coding for at least one expressed alternatively spliced isoform were taken as positive and expressed as the percentage of total analyzed genes.

RESULTS
topo I Protein Kinase Activity Is Inhibited by PAR-In a reconstituted system consisting of PAR, topo I, and an oligonucleotide substrate, we first demonstrated that site-specific DNA cleavage by either yeast-or baculovirus-expressed recombinant human topo I is inhibited by PAR in a dose-dependent manner (Fig. 1A), as previously reported by us for topo I purified from human placenta (24). This inhibition is accompanied by the formation of a PAR⅐topo I complex (24); degradation of PAR to monomeric ADP-ribose abolished the inhibition (Fig. 1C). To investigate whether PAR had any influence on topo I kinase activity, we performed an ASF/SF2 phosphorylation assay in the presence or absence of the polymer. topo I kinase targets the C-terminal RS domain of ASF/SF2 (29). His-tagged ASF/SF2, expressed in bacteria and purified by metal ion affinity chromatography, was used in this study. The recombinant protein migrated as a doublet in SDS-polyacrylamide gels (Fig. 2C); the shorter form is generated by proteolysis at the C terminus of the protein, which does not affect the capacity of the protein to be phosphorylated by SR-protein-specific kinases (29). Recombi-nant ASF/SF2 was phosphorylated by both topo I and the SRprotein kinase SRPK1 (Fig. 2, A and B). However, when PAR was present in the kinase assay reaction mixture, topo I-catalyzed ASF/SF2 phosphorylation was dramatically reduced (Fig.  2, B and E); the extent of inhibition was dependent on PAR concentration (Fig. 2, B and E). Thus, PAR appears to function as a negative regulator not only of the DNA cleavage ( Fig. 1A) but also of the protein phosphorylation activity of topo I (Fig. 2, B and E). Neither PARP-1 nor PARP-2 in their native state affected topo I-catalyzed ASF/SF2 phosphorylation (Fig. 2D). Furthermore, the polymer at severalfold higher concentrations than that inhibitory on topo I had only modest consequences on ASF/SF2 phosphorylation by SRPK1 (Fig. 2, A and E), thus pointing at a topo I-specific PAR effect. Noteworthy, although both topo I and SRPK1 phosphorylate serine residues in the RS-domain of ASF/SF2, the reaction catalyzed by topo I has different specificity and kinetics from that of SRPK1 (29,30), implying the involvement of distinct mechanisms.  PAR and ASF/SF2 Reciprocally Antagonize Their topo I Inhibitory Action-ASF/SF2 is known to inhibit DNA relaxation by topo I by interfering with the DNA cleavage step of the catalytic cycle (25,(31)(32)(33). In Fig. 1B inhibition of topo I activity was achieved at an approximate topo I:ASF/SF2 molar ratio of 1:10.
Because both PAR and ASF/SF2 are negative regulators of topo I-catalyzed DNA cleavage (Fig. 1), and ASF/SF2 phosphorylation is inhibited by PAR as well (Fig. 2), would PAR and ASF/SF2 together cause a complete silencing of topo I functions? To address this question, we set up an in vitro assay that allows simultaneous detection of topo I⅐DNA cleavage complex and phosphorylated ASF/SF2. Surprisingly, we observed a full restoration of topo I-catalyzed DNA cleavage while the inhibitory effect of PAR on topo I-dependent ASF/SF2 phosphorylation persisted (Fig. 3A). Reversal of topo I activity on DNA in the presence of both PAR and ASF/SF2 was clearly dependent on PAR concentration (Fig. 3A, right panel) and also occurred in the absence of ATP (Fig. 3B, left panel). It is noteworthy that cleavage complex formation by a N-terminally truncated form of topo I (⌬N-topo I, 70 kDa), which maintained full proficiency to relax DNA but had dramatically reduced protein kinase activity, was also inhibited by either PAR or ASF/SF2 individually (Fig. 3B, right panel). This is in agreement with our previous identification of PAR binding sites outside of the N-terminal domain of topo I (24). Like full-length topo I, cleavage complex formation by ⌬N-topo I was also restored in the presence of both PAR and ASF/SF2 (Fig. 3B, right panel).
ASF/SF2 Is a Novel Member of the PAR-binding Protein Family-We next hypothesized that PAR could bind to ASF/ SF2 and that the formation of a stable PAR⅐ASF/SF2 complex might prevent either of the two interaction partners from binding topo I; as a consequence, topo I would be able to express its DNA cleavage activity. PAR binding was assessed using recombinant ASF/SF2, either His-tagged or as a fusion protein with GST. To identify the specific domain(s) potentially involved in polymer binding, we constructed ASF/SF2 deletion mutants, each comprising only one of the ASF/SF2 functional domains (RRM1/RRM2/RS) (Fig. 4A). PAR binding assays revealed that ASF/SF2 does indeed bind PAR (Fig. 4, C and D). We identified the RRM1 and RS domains as potential targets of such interaction, whereas the RRM2 domain did not appear to possess any PAR binding activity (Fig. 4, C and D). Identical results were obtained both when proteins were first separated by SDS-PAGE and then transferred onto nitrocellulose by Western blotting (Fig. 4, B and D) and when native proteins where immobilized on the membrane by slot-blotting (Fig. 4C), before being probed with radioactive PAR. Importantly, ASF/SF2⅐PAR binding appeared to be considerably stronger than topo I-PAR interactions: nearly 90% topo I binding to PAR was destroyed by 0.5 M NaCl; in contrast, most ASF/SF2⅐PAR complexes resisted high salt treatment (Fig. 5A).
Next, the inhibitory effect of ASF/SF2 on DNA cleavage by a constant amount of topo I was titrated against increasing concentrations of PAR. We found that restoration of topo I DNA cleavage activity was strictly dependent on the relative amounts of the two negative effectors (Fig. 5B): 10 pmol of polymeric ADP-ribose were sufficient to fully remove inhibition caused by 2.8 pmol of ASF/SF2; at higher PAR concentrations however, cleavage complex formation decreased again, probably as a consequence of topo I targeting by PAR molecules that exceeded ASF/SF2 binding capacity.
ASF/SF2, but Not Histone, Phosphorylation by HeLa Nuclear Extracts Is Inhibited by PAR-Is PAR able to modulate ASF/SF2 phosphorylation in a complex protein environment such as a nuclear extract? Fig. 6A demonstrate that PAR inhibits ASF/ SF2 phosphorylation activity in HeLa nuclear extracts in a dosedependent manner. Notably, preincubation with DNA and  CPT, but not with CPT alone, strongly reduced ASF/SF2 phosphorylation (Fig. 6B). Such a treatment has been demonstrated to cause specific inhibition of topo I kinase activity in HeLa nuclear extracts as a consequence of drug-induced topo I trapping in an inactive cleavage complex (6). In addition, the inhibitory action of PAR appeared to be selectively targeted to ASF/ SF2 kinase(s); histone H1 and H2B phosphorylation was not inhibited by PAR over the same dose range (Fig. 6C). Phosphorylation of endogenous nuclear extract proteins also appeared to be unresponsive to PAR (Fig. 6, A and C, uppermost part of the gels).

DISCUSSION
Human DNA topoisomerase I plays an important role in the regulation of gene expression by controlling DNA supercoiling on one side, and pre-mRNA splicing, on the other. In fact, in addition to relaxing supercoiled DNA, topo I exhibits a kinase activity targeted at the serine-arginine-rich family of splicing factors (SR-proteins). How these functions are regulated is still unknown. Significantly, pharmacological inhibition of either of the two enzyme activities alters gene expression and eventually leads to cell death (3)(4)(5)17).
Likewise, in the absence of ATP, PAR and ASF/SF2 each caused a dose-dependent silencing of topo I activity on DNA (Fig. 1). However, when these negative effectors were present simultaneously, the outcome was different and entirely unexpected: PAR and ASF/SF2 neutralized each other's inhibitory actions and allowed restoration of the enzyme's DNA cleavage activity (Fig. 3B). Thus PAR enabled topo I to direct its activity onto DNA under conditions that would otherwise be strongly inhibitory. In a more complex setting, with DNA, ATP, and ASF/SF2 being incubated simultaneously with topo I, we observed that the enzyme phosphotransferase activity was by far favored over the DNA cleavage activity; under such conditions topo I appeared in fact to be a more efficient kinase than a topoisomerase (Fig. 3A); this is in agreement with observations reported by others (31). In addition, we found that PAR is able to restore topo I activity on DNA while at the same time reducing (but not abolishing) topo I-dependent ASF/SF2 phosphorylation; the extent to which distinct topo I functions are either reactivated or inhibited appeared to be dependent on the amount of PAR (Fig. 3A, right panel). A mechanistic explanation of such an effect is offered by the finding that ASF/SF2 is a novel PAR binding partner, with the RRM1 and RS domains as specific PAR interaction targets (Fig. 4). Conceivably, the formation of a PAR⅐ASF/SF2 high affinity complex leaves topo I available for directing its action on DNA. Indeed, PAR⅐ASF/SF2 titration experiments (Figs. 3 and 5B), as well as the finding that PAR binds to ASF/SF2 more tightly than to topo I (Fig. 5A), are consistent with such a model.  The fact that SRPK1, another SR-protein specific kinase, is only partially sensitive to PAR (Fig. 2) further emphasizes the specificity of PAR effects on topo I activities. Thus, PAR appears to be the first known physiological modulator of topo I kinase activity. Moreover, we found that ASF/SF2, but not histone phosphorylation by HeLa nuclear extracts, is inhibited by PAR (Fig. 6), suggesting that PAR might play a role as a regulator of ASF/SF2 phosphorylation status in the nucleus as well. Noteworthy, endogenous topo I could be identified as the main ASF/SF2 kinase in the extract: in fact, enzyme trapping in a ternary complex with CPT and cleaved DNA abolished most of the ASF/SF2 phosphorylating activity (Fig. 6B). In future studies it will be of interest to determine whether other members of the SR-protein family, that are also phosphorylated by topo I (6), may become targets of a PAR-reliant control mechanism.
How might regulation of topo I functions by PAR affect biological processes? We propose a model by which PAR, through modulation of topo I substrate selection (either DNA or ASF/ SF2), may participate in the coordination of DNA transcription and RNA splicing. Indeed it is now well established that RNA splicing and transcription are spatially and functionally coupled (34,35). This integration ensures that pre-mRNA processing is efficient and accurate and at the same time may prevent formation of potentially lethal DNA-RNA hybrids involving template DNA and nascent mRNA precursors (10). By virtue of its dual activity, involving control of DNA topology on one side and phosphorylation of splicing factors on the other, topo I may take part in transcription-splicing coupling mechanisms (31). Besides, several lines of evidence support a likely functional link between topo I and PARP-1 in the regulation of gene expression. In fact, both topo I and PARP-1 localize to active transcription sites (1,2,36,37); furthermore, the two enzymes may form a complex resulting in a severalfold enhancement of topo I DNA relaxation activity and/or loss of interaction with other protein partners (37)(38)(39). Additionally, ASF/SF2, topo I, and PARP-1 share a high affinity for the phosphorylated C-terminal domain of RNA polymerase II and can be associated with the elongating polymerase in vivo (40). In the light of such evidences, we propose that topo I targeting by PAR may play a role during transcription elongation, when topo I activity is required both to relieve torsional stress generated by RNA polymerase II translocation along the DNA template (41), and to phosphorylate ASF/SF2 engaged in co-transcriptional splicing events (16,17). The nick introduced onto the DNA template by topo I itself might constitute the activating signal for PARP-1, in analogy with the mechanism underlying PARP-1 and DNA topoisomerase II ␤ cooperation at the level of transcription initiation (42). Alternatively, PAR synthesis might be stimulated by PARP-1 binding to altered DNA structures (43) that are formed during transcription. Newly synthesized PAR might then exert its regulatory role at two levels: 1) by attenuating (but not abolishing) topo I kinase activity that would otherwise be FIGURE 7. Regulation of topo I functions by PAR. The model illustrates PAR effects on distinct topo I activities and the likely mechanism underlying PARinduced topo I functional switch. PAR interacts both with topo I and ASF/SF2 (double pointed arrows), with the latter forming more stable complexes (thicker double pointed arrow). In the absence of ASF/SF2, PAR targeting causes topo I to lose its DNA cleavage activity (A). ASF/SF2 is also a binding partner for topo I and a negative regulator of topo I-catalyzed DNA cleavage; moreover, in the presence of ATP, ASF/SF2 reactivates a latent topo I kinase activity and becomes itself phosphorylated (B). When PAR, topo I, and ASF/ SF2 are present simultaneously, the two proteins compete for PAR binding: the formation of a PAR⅐ASF/SF2 high affinity complex leaves topo I free to direct its action on DNA (C). In such a case, PAR acts as a modulator of the relative expression levels of topo I activities; in fact, whether topo I functions as a topoisomerase or as a kinase depends on PAR amounts: at low PAR concentrations, both DNA cleavage and protein phosphorylation can occur (C, left); DNA cleavage activity is fully restored (at the detriment of the kinase activity) at higher PAR concentrations that titrate off ASF/SF2 from its interaction with topo I (C, right). Lastly, several lines of evidence hint at a likely more direct involvement of the poly(ADP-ribosyl)ation system in pre-mRNA processing: (i) PAR binds to components of the RNA splicing machinery, i.e. the splicing factor ASF/SF2 (this study, Fig. 4) and several heterogeneous nuclear ribonucleoproteins (46); (ii) both PARP and PAR glycohydrolase activities have been found in cytoplasmic ribonucleoprotein particles (47,48); and (iii) PARP-1 has been shown to possess RNA-binding ability (49). The observation, that ASF/SF2 phosphorylation, but not histone phosphorylation, in HeLa nuclear extracts is sensitive to PAR (Fig. 6), is also consistent with such a scenario. Moreover, the ASF/SF2 domains targeted by PAR, i.e. the RRM1 and RS domains, are crucial for splicing functions, suggesting that PAR effects may extend beyond regulation of ASF/ SF2 phosphorylation. Notably, the RS domain, which is present in several splicing factors, when tethered to a pre-mRNA is sufficient alone to promote pre-spliceosome assembly and to support splicing (50). A speculation along this line is also supported by our results of an in silico study on published expression profiling data from wild-type and parp-1 knock-out cells and animal tissues (51)(52)(53)(54)(55). We found that, within a group of genes involved in the response to cytotoxic stimuli, the incidence of alternative splicing increases specifically in the subset of parp-1-dependent genes ( Table 1). The fraction of alternatively spliced genes in this subset was 67.7-72.2% compared with 53-59% in the group of stress-regulated genes as a whole. Noteworthy, of the 18 genes that were reported to be differentially involved in the tumor necrosis factor-␣ response (54), 10 were not responsive either in wild-type or knock-out cells (possibly because of altered signaling), whereas 8 genes changed their expression levels in both cell types, but at different extents; the incidence of alternative splicing in this subgroup of genes was found to be 87.5% and reached 100% when considering only those genes (5 of 8) that were expressed at lower levels in cells lacking a functional parp-1; moreover, all these genes contained ASF/SF2 binding sites (data not shown). It should be emphasized that the number of published data that could be included in this study does not allow any conclusive statement yet. Nevertheless, our observations suggest there is a possible link between PAR/PARP-1 and alternative splicing that warrants further validation and investigations.
In conclusion, PAR may function as a molecular switch of topo I DNA cleavage/SR-protein kinase activities. We speculate that PAR targeting of topo I and/or SR-protein functions may play a role as a novel regulatory mechanism of gene expression. Such a mechanism might involve a specific set of genes and entail regulation of topo I-dependent phosphorylation of SRproteins and/or modulation of SR-proteins' interaction with splicing coactivators/corepressors. Important clues may come from the identification of target genes, an achievement that should be greatly accelerated as more differential expression profiling data become available.