dMi-2 plays a role in RNA processing and splicing of hsp genes

Expression of several hsp genes is strongly abrogated in flies with depleted levels of dMi-2 or flies expressing its inactive form. This raises the possibility that dMi-2 might be involved in chromatin remodeling on these genes and thus facilitate their expression.

However, several results do not support this hypothesis. First, heat shock genes undergo rapid nucleosome disruption across the entire locus within two minutes after gene

177 activation and this requires at least three factors: HSF1, GAF factor and PARP1 enzyme (Petesch and Lis 2008). However, the first signal for dMi-2 is observed earliest at two minutes of heat shock and it increases in time (Fig. 5.29). Second, no significant differences in histone H3 removal on hsp genes in dMi-2 knockdown flies were observed (data not shown). Thus, it is unlikely that dMi-2 plays a role in nucleosome removal at heat shock loci.

The low transcript levels of hsp genes in dMi-2 depleted or mutant overexpressing flies may reflect disturbances in RNA processing. Indeed, the ratio of unprocessed or unspliced forms of hsp70 and hsp83 to total RNA levels of these transcripts are significantly increased in both, dMi-2 depleted and catalytic mutant overexpressing flies (Fig. 5.25).

Similar defects have been reported previously for several factors involved in RNA processing. For instance, depletion of THO complex subunits in SL2 cells increases the ratio of unprocessed transcripts to total hsp70 RNA while decreasing the heat shock gene response (Rehwinkel et al. 2004; Kopytova et al. 2010). Another report has shown that pharmacological inhibition of P-TEFb, a kinase that phosphorylates CTD of RNAP II at Ser2, leads to reduction of hsp transcripts and significant increase of unprocessed hsp RNA species. The authors postulated that inhibition of CTD phosphorylation at Ser2 abrogates recruitment of RNA processing factors which in turn leads to inefficient RNA processing and rapid degradation of hsp transcripts (Ni et al. 2004).

Several other experiments suggest dMi-2´s role in mRNA 3‟end processing. First, it crosslinks throughout the entire transcribed hsp70 gene and it is associated with the 3‟ end of the gene, where the cleavage and polyadenylation site is present (Fig. 5.21). Second, a recent ChIP deep sequencing on heat shocked cells performed in our laboratory has confirmed the binding of dMi-2 downstream of the transcription termination sites at most of hsp genes (Eve-Lyne. Mathieu, data not published). It has been reported that factors involved in transcription elongation and RNA processing display different crosslinking patterns at coding regions and downstream of the polyadenylation site. Some of them bind through the entire transcribed regions, others drop downstream of the polyadenylation site or conversely crosslink just downstream of the polyadenylation site (Kim et al. 2004;

Mayer et al. 2010). Thus, there is an evidence for a transition of the elongation and processing factors at 3‟ ends of transcribed genes. Finally, dMi-2 interacts with nascent unprocessed hsp70 and hsp83 transcripts, which suggests that it can be involved in RNA processing more directly (Fig. 5.26). For example, dMi-2 might facilitate proper substrate

178 formation or access of processing factors to pre-mRNA 3‟ end. In this context, it would be important to determine whether any interactions between processing factors and dMi-2 occur in vivo.

mRNA 3‟ end processing is functionally coupled to transcription termination downstream of protein coding genes. This interplay between mRNA 3‟ end processing and transcription termination, makes it difficult to distinguish in which of these two processes dMi-2 plays a direct role. Despite extensive studies, still little is known about the mechanism of transcription termination and mRNA 3‟ end processing. There are currently two models which explain how transcription is terminated. The first model, known as the “anti-terminator model”, suggests that the appearance of the polyadenylation sequence on the mRNA triggers an exchange in factors associated with elongating RNAP II which would decrease its processivity and eventually lead to transcription termination. The second scenario, called the “torpedo model” proposes that the mRNA cleavage at polyadenylation site could act as an entry point for an enzyme (helicase or exonuclease) that would track along the RNA and dissociate the RNAP II. Recent discoveries provide a support for both models (reviewed in (Buratowski 2005; Kuehner et al. 2011)). In addition, pausing of RNAP II downstream of the polyadenylation site facilitates transcription termination (Gromak et al. 2006).

Studies in yeast suggest a link between chromatin remodeling and transcription termination. yChd1 and Hrp1 have been implicated in transcription termination regulation at several genes via regulation of the chromatin structure at the 3‟ end of these genes. It has been proposed that chromatin structure at 3‟ end of the gene may enhance RNAP II pausing and thus facilitate the switch of RNAP II mode from elongation to termination (chapter 2.5.4.4) (Alén et al. 2002). In order to determine the role of dMi-2 in transcription termination, it would be interesting to test how far RNAP II transcribes beyond the polyadenylation site at the hsp70 gene upon dMi-2 depletion. This question can be addressed by performing RNAP II ChIP experiments and compare its association with downstream regions of the polyadenylation site between dMi-2 depleted and wild type flies. Additionally, analysis of the transcription readthrough at the 3‟ end of hsp70 gene could be also performed in wild type and mutant flies. Moreover, it would be interesting to monitor the chromatin structure at the 3‟ end of heat shock genes upon dMi-2 depletion by histone H3 and micrococcal nuclease accessibility assay. It has been reported that there is a nucleosomal free region at the 3‟ end of hsp70 (Petesch and Lis 2008). Thus, it is plausible

179 that chromatin remodeling activities are involved in proper chromatin structure maintenance at the 3‟ end and consequently proper transcription termination.

Finally, it has been reported that defective transcription termination at the 3‟ end of a gene leads to decreased splicing and is required for optimal gene transcription (West and Proudfoot 2009). In this respect, the transcription and splicing defects of hsp genes could be an indirect effect of affected transcription termination. It has been shown that transcription termination involves protein components that bind to DNA and/or RNA polymerase and employ ATPase activity to dissociate polymerase complexes from the DNA template (Deng and Shuman 1998; Liu et al. 1998). Thus, it is plausible that ATPase activity of dMi-2 contributes to coupling ATP hydrolysis to transcription termination.

Undoubtedly, the fascinating link between chromatin remodeling and RNA processing has to be elucidated in future experiments.

Recent years have also revealed interplay between splicing and chromatin remodeling.

Two chromatin remodelers, hCHD1 and Brahma, have been implicated in pre-mRNA splicing, although the role of their ATPase activity is not clear in this context. First, hCHD1 is involved in regulation of pre-mRNA splicing by recruiting components of the splicing machinery to the transcribed RNA via recognition of the H3K4me3 mark (chapter 2.5.4.5) (Sims et al. 2007). Second, human Brahma associates with components of the spliceosome and favours inclusion of variant exons in the mRNA of several genes independent of its enzymatic activity. It has been found that Brahma decreases RNAP II elongation rate and thus facilitates recruitment of the splicing machinery to variant exons with suboptimal splice sites (Batsché et al. 2006). Moreover, Brahma is incorporated into nascent pre-mRNPs co-transcriptionally in Chironomus tentans. Depletion of SWI/SNF complex subunits in Drosophila SL2 cells changes the relative abundance of alternative transcripts from a subset of genes (Tyagi et al. 2009). Altogether, these results suggest that Brahma is involved in splicing regulation in different species. However, the molecular mechanism still remains to be determined.

The role of dMi-2 in splicing also has to be elucidated in further studies. It would be important to figure out whether dMi-2 plays a direct role in this process. One possibility would be to test if dMi-2 interacts with splicing machinery by making immunoprecipitation experiments followed by mass spectrometry. Although dMi-2 complex purification with classic chromatography did not show association of any splicing

180 factors, it is plausible that such associations exist in only substoichiometric amounts or are transient and become lost during traditional purification steps. Therefore it would be worth to try to perform coimmunoprecipitation experiments directly from crude nuclear extracts using milder purification conditions. Interestingly, in the recent release of The Drosophila Protein Interaction Mapping (DPiM) project that applied purification of transiently expressed tagged proteins followed by identification of associated peptides by mass spectrometry, an association of a splicing factor U2af50 with dMi-2 has been identified (https://interfly.med.harvard.edu). It would be interesting to test whether this interaction occurs between endogenous proteins. Another experiment, which would allow to study dMi-2 involvement in splicing, could be in vitro splicing assay in the presence of extracts containing or depleted of dMi-2. Similar experiments have been applied to test hCHD1 role in splicing (Sims et al. 2007). In addition a splice reporter minigene could be also adapted to Drosophila cell lines in order to study dMi-2 function in splicing in vivo (Lallena et al. 2002; Batsché et al. 2006).

Hsp genes usually do not possess introns, but hsp83 is an exception. The role of dMi-2 in splicing regulation of this gene raises the question whether dMi-2 is involved in splicing regulation on a more global, genome wide scale. Currently there is no microarray platform designed for splicing analysis in Drosophila, thus in order to address this question other methods should be utilized. For instance, a recent development of next-generation deep sequencing technology is extensively used to study splicing events on a genome wide scale (Pan et al. 2008; Fox et al. 2009; Filichkin et al. 2010). Comparison of transcriptomes at high resolution between wild type and dMi-2 depleted cells or flies, could shed light onto its potential role in splicing genome-wide.

What could be the role of dMi-2 in splicing? dMi-2 might similarly to hCHD1 link or recruit splicing machinery to RNA via binding to nascent transcripts. However, the role of the enzymatic activity of dMi-2 for splicing suggests a more active role of this remodeler.

In one possible scenario, dMi-2 might function as an RNP remodeling factor to modulate interactions between RNA splicing factors and their target RNA. It is also plausible that dMi-2 influences splicing less directly, for instance by remodeling chromatin structure during transcription. Indeed, recently, chromatin structure and histone modifications have been linked to splicing regulation. Mapping nucleosome positions at a genome-wide scale from various organisms has shown that nucleosomes are particularly enriched at intron-exon junctions (Andersson et al. 2009; Nahkuri et al. 2009; Schwartz et al. 2009; Dhami et

181 al. 2010). Nucleosomes can act as barriers that modulate RNAP II density by inducing its pausing (Hodges et al. 2009). Thus, nucleosome positioning might affect splicing efficiency. It is possible that chromatin remodelers, like dMi-2, might influence splicing by remodeling nucleosomes in the way of RNAP II. Although nucleosomes are severely disrupted at hsp genes upon gene induction there are still some histones left, besides histone variant H3.3 is deposited in the body of hsp genes which indicates that chromatin remodeling might occur to some extent on these genes upon their activation (Schwartz and Ahmad 2005). Whether dMi-2 plays any of these roles in splicing, remains to be determined in the future.

Finally, one reason for inefficient transcription of hsp genes in dMi-2 depleted flies could be that transcription elongation by RNAP II is affected. This possibility cannot be formally excluded and demands further investigation. Western blot analysis revealed no differences in the level of RNAP II phosphorylation at Ser2 in wild type and dominant negative mutant overexpressing flies (data not shown). This suggests that at least at the global scale dMi-2 depletion does not affect RNAP II elongation. One experiment to test this more directly on hsp genes would be to probe for elongating RNAP II occupancy at hsp genes upon gene induction by ChIP in dMi-2 depleted flies.

To sum up, the results presented in this doctoral thesis provide a first link of a catalytic activity of the ATP dependent chromatin remodeler in RNA processing and splicing. Many experiments remain to be done in order to clarify the function of dMi-2 in this context.