Drosophila Prp40 localizes to the histone locus body and regulates gene transcription and development

In eukaryotic cells, the nucleus is compartmentalized and contains several dynamic nonmembrane-bound structures referred to as nuclear bodies or nuclear compartments, which are essential for the correct maintenance of nuclear architecture and the gene-regulatory processes that occur within the nucleus (Dundr, 2012; Dundr and Misteli, 2001; Mao et al., 2011; Zhao et al., 2009). The study of the constituents and the spatial and dynamic properties of these nuclear bodies is essential for understanding the regulation of gene expression programs, which are critical for cell stability and survival. Given the importance of nuclear bodies in controlling how gene expression is exerted, alterations in the regulation or biosynthesis of these structures can lead to pathological consequences (Morimoto and Boerkoel, 2013; Sleeman and Trinkle-Mulcahy, 2014).

Because of the absence of a delineated membrane, the structural integrity of nuclear bodies is mediated by protein–protein and/or protein–RNA interactions. This property and the rapid dynamics of nuclear bodies are consistent with a self-organization model in which the structure of a body is determined by the global interactions among its constituents (Kaiser et al., 2008; Misteli, 2001). Although significant progress has been made regarding the role of these nuclear bodies in gene expression, we are still far from understanding how they are assembled in the cell. Many studies have led to two main distinct but not exclusive assembly models. While one model posits that assembly occurs through an ordered, hierarchical process through which constituents are assembled around a primordial scaffolding factor or RNA, the other model considers that self-organization is accomplished randomly without any particular ordered or hierarchical nuclear body assembly (reviewed in Mao et al., 2011; Matera et al., 2009; Staněk and Fox, 2017). More recently, a third model for nuclear body formation involving intracellular phase separation to promote the assembly of droplets of nuclear protein/RNA has been proposed (Zhu and Brangwynne, 2015; Boeynaems et al., 2018). This model posits the existence of different nucleoplasmic phases with distinct physical properties through which proteins may transition to gain favorable thermodynamic states so that nuclear body assembly is mediated by this phase transition.

Some nuclear bodies are also associated with specific gene loci, and this association with a specific nuclear function or activity may be important for their formation and function (Arias Escayola and Neugebauer, 2018). The histone locus body (HLB) is a chromatin-associated nuclear body that specifically associates with replication-dependent histone gene clusters to coordinate the transcription and 3′ end processing of histone pre-mRNA (Duronio and Marzluff, 2017; Marzluff and Koreski, 2017; Marzluff et al., 2008). In Drosophila, the histone gene cluster is composed of ∼100 copies of tandemly arranged histone H1, H2a, H2b, H3 and H4 gene cassettes (Lifton et al., 1978). Histones play a crucial role in the packaging of DNA into chromatin. Consistent with this role, histone expression is restricted to the early S phase of the cell cycle, which is tightly coupled to DNA synthesis (Marzluff et al., 2008). Defects in histone biosynthesis result in genomic instability, which may promote oncogenesis (Morgan and Shilatifard, 2015). Since the initial characterization of the HLB by the Gall laboratory (Liu et al., 2006), many factors have been identified as components of this nuclear body. Some of these factors are constitutively present in these nuclear bodies throughout the cell cycle, whereas others are recruited to the HLB only during the S phase when histone transcription is active. The first group of factors includes Multi sex combs (Mxc) (known as NPAT in mammals) (Ma et al., 2000; White et al., 2011; Zhao et al., 2000), FLASH (White et al., 2011; Yang et al., 2009), the U7 snRNP (Frey and Matera, 1995; Liu et al., 2006; Wu and Gall, 1993) and Mute (Bulchand et al., 2010), whereas general and elongation transcription factors, such as RNA polymerase II (RNAPII), TBP, Spt6 (Guglielmi et al., 2013) and Myc (Daneshvar et al., 2011; White et al., 2011), and factors regulating histone pre-mRNA processing, such as Symplekin (Tatomer et al., 2014) and other proteins (Duronio and Marzluff, 2017), associate with the HLB upon the activation of histone gene transcription. The emerging picture is that the Drosophila HLB assembles through the hierarchical recruitment of components; Mxc and FLASH form the foundational HLB that is detected in the early embryo at cycle 10, and U7 snRNP and Mute are recruited at cycle 11 in the absence of histone mRNA transcription (White et al., 2011). A sequence located between the histone H3 and H4 genes contains the shared H3 and H4 promoter (hereafter, denoted as the H3/H4 promoter), which is essential for histone gene expression, and is required for the recruitment of Mxc and FLASH (Salzler et al., 2013). A significant number of proteins are subsequently joined to the HLB in a manner coupled to active histone gene transcription (Duronio and Marzluff, 2017). How the initial interaction of Mxc and FLASH with the histone loci occurs and what the actual composition of a fully formed HLB is remain to be resolved.

Prp40 was initially identified as an essential yeast factor that participates as a scaffold in the early steps of spliceosome complex formation (Abovich and Rosbash, 1997; Kao and Siliciano, 1996). Prp40 has a characteristic domain organization, with two WW domains in the N-terminus and five FF domains in the C-terminus, which is a structure shared by a relatively small number of proteins (Becerra et al., 2016). Strikingly, most of these structurally related proteins have been implicated in transcription and splicing regulation. There are two putative mammalian orthologs of Prp40, PRPF40A and PRPF40B. Based on phylogenomic data, PRPF40A appears to be more closely related to Prp40 than does PRPF40B, which emerged much later in evolutionary history probably due to a gene duplication event from an ancestral PRPF40A (Becerra et al., 2016). PRPF40A and PRPF40B interact with the transcription and splicing machineries, and at least for PRPF40B, the modulation of alternative splice site selection in apoptosis-related genes has been shown (Becerra et al., 2016, 2015). The Drosophila ortholog of Prp40, herein denoted dPrp40, encoded by the CG3542 gene (Flybase.org), shares 23% and 41% sequence identity with the yeast Prp40 and the human PRPF40A proteins, which suggests that the function of these proteins in forming bridges between the 5′ and 3′ splice sites in the first spliceosomal complex might be conserved (Mount and Salz, 2000). In fact, the regulation of alternative pre-mRNA splicing of the glial-specific cell-adhesion molecule Neurexin IV by dPrp40 has been reported (Rodrigues et al., 2012).

Herein, we characterize dPrp40 and identify a putative new role for this protein in histone mRNA transcription. We show that dPrp40 localizes to the Drosophila HLB during prophase after the incorporation of the HLB primary protein components. We demonstrate that dPrp40 is essential for Drosophila development. Moreover, dPrp40 and its human orthologs can rescue the phenotype of dPrp40 mutant flies, demonstrating a functional conservation of eukaryotic Prp40 activities in vivo. We also show an essential requirement for dPrp40 in endocycling cells, highlighting a role for this factor in the replication efficiency in vivo. In a molecular context, we show that the depletion of dPrp40 from fly cells inhibits histone mRNA transcription without affecting the 3′ maturation of histone mRNA. Furthermore, H3/H4-dependent transcription, which is essential for HLB assembly and high-level histone gene expression (Salzler et al., 2013), is rescued by overexpressing dPrp40 in the depleted cells. Together, our results establish that dPrp40 is required for normal embryonic development and that dPrp40 can localize to the HLB and might regulate histone gene transcription, which could have important consequences for the cell cycle and maturation, development and viability.

Prp40 was originally identified as an essential splicing factor in yeast (Kao and Siliciano, 1996). To generate tools for the biochemical analysis of this protein in Drosophila, we expressed dPrp40 in E. coli, purified it, and used it to generate rabbit and guinea pig polyclonal antibodies. These antibodies react specifically with a protein of a molecular mass ∼90 kDa (which matches the predicted size of the dPrp40 protein) in fly extracts. To investigate the subcellular localization of dPrp40, we performed immunofluorescence experiments on cells from various tissues of Drosophila using confocal laser microscopy with anti-dPrp40 antibodies. Our experiments showed that dPrp40 was ubiquitously expressed in the nucleus of cells in all stages of D. melanogaster development and was distributed throughout the nucleoplasm, with greatly increased signal, usually localized in a single dot. The increased accumulation of dPrp40 at these foci was reminiscent of the nuclear pattern observed for the HLB, an evolutionarily conserved body involved in the expression of replication-dependent histone genes. To determine whether dPrp40 colocalizes with the HLB, we performed double-labeling immunofluorescence experiments using dPrp40-specific antibodies and antibodies against proteins that commonly define the HLB (Mxc, MPM2 and Lsm11). Using these HLB markers, we found that dPrp40 was concentrated at the HLB in salivary gland and wing disc tissues (Fig. 1A). Interestingly, dPrp40 stained the HLB in most MPM2-positive cells (Fig. S1). From these data, we conclude that dPrp40 is located in the nucleoplasm and associates with the Drosophila HLB.

Prp40 family members contain tandem repeats of WW and FF domains. Using EMBOSS global and local alignments, we identified two WW and four FF domains in dPrp40 with different degrees of similarity to the mammalian Prp40 domains. To identify the sequence elements responsible for the nuclear localization of dPrp40 and subsequently characterize the regions in dPrp40 responsible for association with the HLB, we first analyzed the primary amino acid sequence of dPrp40 for the presence of potential nuclear localization signals (NLSs) using the WoLF PSORT, cNLS Mapper and NucPred computer programs (Brameier et al., 2007; Horton et al., 2007; Kosugi et al., 2009). The three NLSs with the highest scores were located in the C-terminal region of the protein, whereas other sequences with lower scores were located outside this region (Fig. S2A). Then, we generated and transiently expressed epitope V5-tagged full-length dPrp40, a dPrp40 mutant with a C-terminal deletion removing the FF domains, a dPrp40 mutant with an N-terminal deletion removing the WW domains, and a truncated Prp40 derivative that lacked the putative NLS with the highest score (dPrp40, ΔFF-dPrp40, ΔWW-dPrp40 and ΔNLS-dPrp40, respectively; diagrams of the constructs are shown in Fig. S2B). The localization of the different dPrp40 protein versions was assessed by immunofluorescence in wing disc tissues, overexpressing the UAS transgenic lines using the broadly expressed ptc-Gal4 driver. In contrast with full-length dPrp40 and ΔFF-dPrp40, which accumulated at the HLB compartment in almost every cell that we screened, ΔWW-dPrp40 accumulation at the HLB was observed in less than half of the analyzed cells (Fig. 1B). We conclude that the WW domains are important for the localization of dPrp40 to the HLB; we do not rule out, however, a possible role of the FF domains in localization to the HLB, perhaps through increasing the binding efficiency. Cells expressing the ΔNLS-dPrp40 mutant construct displayed diffuse cytoplasmic and nuclear localization of the protein (Fig. 1B), suggesting that the deleted region contains the signals for proper dPrp40 nuclear localization. The partial nuclear staining observed upon ΔNLS-dPrp40 expression may be due to the absence of the NLSs from the deleted protein. Nuclear ΔNLS-dPrp40 expression also showed accumulation at the HLB. The quantification of the data confirmed the HLB distribution of these proteins (Fig. 1C).

According to the hierarchical recruitment of components during Drosophila HLB formation (White et al., 2011), Mxc and FLASH are present in the HLB before the expression of histone mRNA, and other components, such as Spt6 or RNAPII, are localized to the HLB region only when histone genes are transcribed (White et al., 2011; Yang et al., 2009). To explore whether dPrp40 participates in the ordered recruitment of HLB components, we used antibodies against Mxc and FLASH to assess HLB formation after dPrp40 depletion. Using a transgenic UAS-RNAi-dprp40 line containing the wing disc ptc-Gal4 driver, we observed normal staining of Mxc and FLASH at the HLB compartment in the dPrp40-depleted cells (Fig. 2A). This result was corroborated in wild-type embryo cells, where dPrp40 expression was absent in the HLB region (Fig. 2B, see a representative cell marked with an arrow). These data indicate that dPrp40 is not required for the initial steps in the formation of the HLB. Evidence for the reduction of dPrp40 RNA and protein levels due to RNAi is shown in Fig. S3. The lack of enrichment of dPrp40 at the HLB in certain cells (Fig. 2B) may indicate cell cycle-dependent dPrp40 localization to this region. To determine the localization of dPrp40 to the HLB in relation to the cell cycle, we assessed whether dPrp40 localization to the HLB is cell cycle-dependent. We observed that dPrp40 localization to the HLB occurs in prophase but not in metaphase, anaphase or telophase (Fig. 2C). The data in Fig. 2 is consistent with dPrp40 being present in the HLB throughout interphase and being lost in mitosis as the HLB dissociates (White et al., 2011).

In a gain-of-function screen to search for modifiers of fibroblast growth factor (FGF) signaling in Drosophila, Zhu et al. reported that dPrp40 is essential for oocyte development, influencing the growth and survival of cells in the eye disc (Zhu et al., 2005). The human dPrp40 homolog PRPF40B is also implicated in cell survival by altering the alternative splicing of relevant apoptotic genes in cultured human cells (Becerra et al., 2015). To investigate the effects of dPrp40 depletion in vivo, we used a mutant strain containing a single P element insertion (CG3542^EP719) in the coding region of the dprp40 gene (Zhu et al., 2005). Homozygous gene insertion of CG3542^EP719 was lethal in embryonic stages. To reduce the levels of the full-length dPrp40 protein, we used the transgenic UAS-RNAi-dprp40 line with the ubiquitous act5c-Gal4 driver (Dietzl et al., 2007). Reduced levels of dPrp40 were also lethal in embryonic stages, which indicates that dPrp40 is essential for Drosophila viability.

To analyze the role of dPrp40 in specific Drosophila tissues, we used several Gal4 drivers to induce the expression of the UAS-RNAi-dprp40 construct. Through the use of several imaginal disc drivers (en-Gal4, dpp-Gal4, ptc-Gal4, 638-Gal4 and ey-Gal4), we observed evidence for characteristic phenotypes with reduced wings and reduced and disorganized eye differentiation (Fig. 3A). With the 638-Gal4 driver, whose expression is restricted to the wing blade in third-instar wing discs, adult wings were extremely reduced in size, losing most of their pattern elements (Fig. 3A). To confirm dPrp40 depletion, we used in vivo detection of GFP to directly screen for Gal4-directed GFP expression in UAS-RNAi-dprp40 expressing cells (Fig. 3B). Reduced levels of dPrp40 correlated with increased apoptosis in the Drosophila wing imaginal disc in specific areas, as shown by immunostaining using an antibody against active Caspase-3 to detect apoptotic cells (Fig. 3B). Taken together, these data indicate that dPrp40 plays important roles during the development of Drosophila tissues.

To further confirm the specificity of the RNAi effects, we used one of the six P element-induced deletions of CG3542 described by Zhu and coworkers (Zhu et al., 2005). The imprecise excision (termed 18.2 and henceforth referred to as df18.2dprp40), which does not alter the reading frame and removes the three potential NLSs present in the C-terminal region, generated an internal deletion of 135 amino acids in dPrp40 (Fig. S4A). Western blot analysis of extracts from recombinant flies revealed the presence of the truncated dPrp40 protein of ∼70 kDa (Fig. S4B). Next, we analyzed the localization of the truncated protein in mitotic wing disc clones generated using the allele Df18.2dprp40. We observed that Df18.2prp40 clones were smaller than control clones at 48–72 h after clone induction (Fig. 3C). Strikingly, analysis of wing discs 24 h after clone induction revealed the absence of surviving homozygous Df18.2prp40 cells (Fig. 3C). In homozygous Df18.2prp40 cells, following clone induction, a predominantly cytoplasmic localization was observed (Fig. S4C), supporting our previous results with the ΔNLS-dPrp40 mutant construct that indicated the relevance of these NLSs for proper dPrp40 nuclear localization. Importantly, the dPrp40 constructs lacking the three C-terminal NLS sequences localized to nuclei and to the HLB when overexpressed (Fig. 1B). These data are consistent with our results and further support a role for dPrp40 in Drosophila development.

To characterize the dPrp40 regions with the observed phenotype in the mutant flies, we tested whether the UAS-dprp40, UAS-ΔFFdprp40, UAS-ΔWWdprp40 and UAS-ΔNLSdprp40 lines could restore the UAS-RNAi-dPrp40 phenotype observed with the 638-Gal4 driver (Fig. 4A,B). The dPrp40 overexpression in transgenic flies fully rescued the phenotype resulting from the loss of dPrp40 (Fig. 4C). The wing structure was also restored to a normal phenotype upon constitutive overexpression of ΔWWdPrp40 and ΔNLSdPrp40 in transgenic flies (Fig. 4D,E). Overexpression of ΔFFPrp40, however, was unable to rescue the wing phenotype (Fig. 4F). These data showed that the FF domains of dPrp40 are critical for rescuing the phenotype resulting from the loss of dPrp40, while the WW domains had little effect.

We also sought to test whether human Prp40 orthologs could replace dPrp40 in vivo by using UAS-PRPF40A and UAS-PRPF40B transgenic fly lines. When PRPF40A or PRPF40B was overexpressed in the developing wing, the wing shape and the overall pattern of the veins were almost completely restored to the control phenotype (Fig. 4G,H), although some differences in the wing shape were observed. We thus conclude that the full-length dPrp40 protein, through its FF domains, and both human Prp40 orthologs can rescue the loss of dPrp40 function in vivo. The quantification of these data is shown in Fig. 4I. Similar results were obtained with the env-Gal4 driver (Fig. S5A–H). These results posit the question of whether the mammalian Prp40 orthologs localize to the Drosophila HLB. We examined the distribution of these proteins by immunofluorescence and observed that PRPF40A clearly localized to the HLB (Fig. S5I); staining with anti-PRPF40B antibodies, however, led to a diffuse signal that precludes any positive conclusion about the presence of PRPF40B at the HLB. Overall, these data support the existence of functional traits shared among these related Prp40 proteins.

The phenotypic changes in cell size observed upon dPrp40 depletion suggested that there is a cell-cycle-specific role for this protein. Thus, we assessed the role of Prp40 in endoreplication. Many types of Drosophila tissues undergo endoreplication, which is considered a mechanism for cell growth (Fox and Duronio, 2013). Endoreplication consists of genome duplication through endocycles, which are defined as cell cycles consisting of S and G phases without cell division. Endocycles use the same molecular machinery as mitotic cell cycles to regulate successive rounds of DNA replication. The best-characterized system of endoreplication is found in the giant salivary gland. Salivary gland cells of Drosophila larvae undergo ten asynchronous endocycles in which repeated cycles of DNA synthesis (S phase) result in very high levels of polyploidy and large cell sizes (Hammond and Laird, 1985). Thus, we examined the role of dPrp40 in endoreplication in salivary glands. We used the ptc-Gal4 construct to drive the expression of the dPrp40-specific UAS-RNAi construct in this tissue. This driver is not expressed in fat body cells, which are adjacent to the salivary gland. Therefore, fat body cells express dPrp40 and can be used as a negative control in this experiment. The size of the salivary gland cells in dPrp40-knockdown flies was significantly lower than that of wild-type flies (Fig. 5A,B). The quantification of the data confirmed the reduction in cell size (Fig. 5C). In addition, nuclei in dPrp40-depleted salivary gland cells displayed reduced DAPI staining (Fig. 5D), which suggests that endocycling is affected by dPrp40 knockdown. We next generated mosaic salivary glands using the FRT-FLP technique to analyze the effect of the dfprp4018.2 allele in the endocycling process. Clones were induced during embryogenesis, in which the salivary gland precursor cells undergo mitosis. For dfprp4018.2-homozygous salivary gland cells, which do not express the GFP marker, the cell size and DAPI staining intensity were highly reduced compared to these parameters for the wild-type surrounding cells and nuclei (Fig. 5E–G). The quantification of the data confirmed the reduction in DAPI staining upon dPrp40 inhibition in these cells (Fig. 5H). Taken together, these data demonstrate a relevant role for dPrp40 in robust endocycling in endoreplicating salivary gland cells.

The expression of histone genes reaches a maximum during the S phase of the cell cycle (Marzluff et al., 2008), as these genes are components of the HLB necessary for proper histone mRNA biosynthesis. Based on our results and the reported regulation of transcription and mRNA processing by Prp40 family members (Becerra et al., 2016, 2015; Rodrigues et al., 2012), we investigated whether dPrp40 was involved in the expression of histone mRNAs. To this end, dPrp40 siRNA was ubiquitously expressed in third-instar larvae using the UAS-RNAidprp40 construct and the T80Gal4/TubGal80TS TARGET system (McGuire et al., 2003). In this system, the existence of a Gal80 temperature-sensitive mutant (Gal80^ts), which represses Gal4 activity at 18°C but not at temperatures of 29°C or higher, allows temporal control of Gal4 activity. After RNAi induction, total mRNA from the third-instar larval stage was collected and prepared for analysis. Replication dependent histone genes do not contain introns and the formation of a mature histone mRNA depends on endonucleolytic cleavage of the pre-mRNA, which produces a unique 3′ stem-loop structure in place of a poly(A) tail. To investigate the role of dPrp40 in histone expression, we first examined the histone mRNA species produced under dPrp40 knockdown conditions. Defects in histone pre-mRNA processing in Drosophila provoke the activation of cryptic polyadenylation signals located downstream of the normal processing site, resulting in the cytoplasmic accumulation of longer, polyadenylated histone mRNA (Tatomer et al., 2016 and references therein). To detect processed and misprocessed polyadenylated histone mRNA species, we used an S1 nuclease protection assay (Tatomer et al., 2016). As previously reported, FLASHPBac/Df mutant third-instar larvae expressed small amounts of properly processed histone mRNA with a length of 348 bp (Fig. 6A, see arrowhead) and large amounts of misprocessed, polyadenylated histone mRNA (Fig. 6A, see RNA fragments between the 713 bp probe and the 348 bp processed histone mRNA), confirming that FLASH is involved in histone mRNA 3′ end formation (Tatomer et al., 2016). In contrast, RNAi-dprp40 mutant third-instar larvae expressed lower amounts of properly processed histone mRNA, whereas no misprocessed histone mRNA was observed (Fig. 6A). These data suggest that dPrp40 affects histone mRNA transcription but is not required for proper mRNA 3′ end formation.

To test whether dPrp40 was important for histone gene transcription, we used quantitative reverse transcription PCR (RT-qPCR) to evaluate the expression of histone genes (Fig. 6B) under dPrp40 knockdown conditions in third-instar larvae. We observed that the transient reduction in dPrp40 levels resulted in lower levels of histone H1, H2a, H2b, H3 and H4 mRNAs (Fig. 6C). Similar results were obtained for late embryos and S2 cells under conditions of gene knockdown, whereas the opposite results were obtained under conditions of protein overexpression (Fig. S6). Expression of unrelated mRNAs (α-tubulin, actin and RPL32) was not altered under conditions of dPrp40 gene knockdown (data not shown). These data support the hypothesis that dPrp40 regulates the transcription of histone genes. To assess whether dPrp40 regulates histone mRNA transcription when recruited to the DNA, we performed chromatin immunoprecipitation (ChIP) assays and quantitative PCR (qPCR) on the H1, H2b-H2a and H3/H4 core promoter regions in Drosophila S2 cells using anti-dPrp40 antibodies. dPrp40 was found to be enriched at these sequences, particularly at the sequences encompassing the H3/H4 promoter (Fig. 6D). To further support these data, we compared the recruitment of dPrp40 to these promoter regions by qPCR in cells in early and late S phase, during which most histone biosynthesis occurs. To obtain a synchronous population of cells arrested in late S phase, cells were first synchronized at the G1/S transition by treatment with hydroxyurea (HU) and were then allowed to progress through the S phase for 6 h (Li et al., 2016). We observed an increase in dPrp40 binding at the promoter sequences in cells in late S phase compared to that in cells at the G1/S transition. The highest increase was observed at the H3/H4 core promoter region (Fig. 6E).

The formation of the Drosophila HLB depends on the H3/H4 intergenic region, which contains a bidirectional promoter that is also essential for the expression of other histone genes within the same cluster (Salzler et al., 2013). To investigate the relationship between the dPrp40 histone H3/H4 promoter occupancy and the transcriptional activity of histone H3/H4 genes, we used a luciferase reporter construct controlled by a 287 bp fragment of the essential H3/H4 promoter region under conditions of dPrp40 gene knockdown using double-stranded RNA (dsRNA) interference in Drosophila S2 cells. Inhibition of dPrp40 reduced H3/H4 promoter activity by 50% compared with that in the control (Fig. 6F, comparison of dPrp40-dsRNA to control GFP-dsRNA). Under these conditions, luciferase activity was completely restored upon dPrp40 expression (Fig. 6F, dPrp40). We also examined the capacity of dPrp40 derivatives to rescue H3/H4 promoter activity. Luciferase activity was partially restored by overexpression of the dPrp40 mutant containing the FF domains but not by overexpression of a dPrp40 mutant containing the WW regions (Fig. 6F, ΔWWdPrp40 and ΔFFdPrp40). Overexpression of the truncated dPrp40 derivative that lacked the putative NLS and had cytoplasmic and nuclear distribution also restored promoter activity (Fig. 6F, ΔNLS-dPrp40). These results positively correlate with the rescue of the dPrp40 null mutant phenotype (Fig. 4) and indicate that dPrp40 regulates histone mRNA levels at the transcriptional level. The FF domains of dPrp40, which were sufficient for phenotypic rescue (Fig. 4C), were also sufficient to restore histone gene transcription. Next, we sought to reproduce these results in vivo. To this end, we tested whether the UAS-dprp40, UAS-ΔFFdprp40, UAS-ΔWWdprp40, UAS-ΔNLSdprp40, UAS-PRPF40A and UAS-PRPF40B Drosophila transgenic lines could restore the H3 and H4 mRNA level when dPrp40 was silenced using 638-Gal4 drivers in wing disc cells. As shown in Fig. 6G, H3 and H4 transcripts were completely restored upon dPrp40, ΔWWPrp40 or ΔNLSdPrp40 protein expression. In agreement with our above luciferase reporter results, expression of the ΔFFPrp40 protein was unable to restore the H3 and H4 mRNA level. These data support the in vitro results and may suggest a role for dPrp40 in the regulation of histone mRNA transcription.

Herein, we performed experiments to characterize the function of Prp40 (dPrp40) in Drosophila. We showed that dPrp40 is essential for Drosophila viability and development via siRNA-mediated depletion and single P element-mediated gene disruption approaches. Conditional knockdown of dPrp40 using different drivers resulted in abnormal phenotypes and increased apoptotic cells in certain regions of wing imaginal discs. These abnormal phenotypes were rescued by expression of the human orthologs of Prp40, indicating that the fly and human proteins have shared functions that affect cell viability. In agreement with this result, we found previously that PRPF40B depletion increased both the number of Fas/CD95 receptors and cell apoptosis in mammalian cells, thus suggesting a role for this protein in programmed cell death (Becerra et al., 2015). The pleiotropic effects caused by the lack of dPrp40 expression, however, may indicate that dPrp40 also regulates other genes. In fact, transgenic PRPF40A expression resulted in a faint Notch-like phenotype with wing margin ‘notches’ (Fig. 4G), which may suggest either an effect of the overexpressed protein on Notch function or an involvement of dPrp40 in the Notch signaling pathway. We recently performed a transcriptome analysis of dPrp40 fly mutants, and the preliminary results support the involvement of dPrp40 in the Notch signaling pathway. Further study is required to determine the molecular targets and signaling pathways regulated by dPrp40.

This study also identified a putative function for dPrp40 in the regulation of histone gene transcription. The localization of dPrp40 to the HLB pointed to a possible role of dPrp40 in the regulation of histone gene expression. This hypothesis is supported by the observation that dPrp40 loss-of-function mutants exhibited altered S phase progression and decreased histone gene mRNA expression. The 3′ end processing of pre-mRNAs plays an important role in the regulation of histone mRNAs, and HLB components are required for the 3′ end maturation of histone mRNAs (Duronio and Marzluff, 2017 and references therein). Our results showed that dPrp40 depletion in Drosophila cells does not result in the polyadenylation of histone mRNAs, indicating that dPrp40 is not required for the 3′ end processing of histone pre-mRNAs in vivo. Our experiments, however, suggest that dPrp40 regulates histone mRNA expression by modulating transcription. We favor an effect of dPrp40 on transcription synthesis based on data using histone promoters and ChIP analysis (Fig. 6). A possible effect of dPrp40 on RNA stability remains to be studied. The regulation of histone gene expression at the level of transcription by dPrp40 was an unexpected finding. The only function described for Prp40 in Drosophila is the regulation of alternative pre-mRNA splicing of the glial-specific cell adhesion molecule Neurexin IV (Rodrigues et al., 2012). This role of dPrp40 in the splicing process agrees with the proposed role for yeast Prp40 in the early steps of spliceosome formation (Abovich and Rosbash, 1997; Kao and Siliciano, 1996) and with our own data supporting a role for the mammalian Prp40 ortholog PRPF40B in pre-mRNA splicing (Becerra et al., 2015). Other data challenge this view and suggest alternative mechanisms of action, including a role for this protein in the later steps of spliceosome assembly and in transcriptional regulation (reviewed in Becerra et al., 2016), which would be consistent with our unexpected data suggesting a role for dPrp40 in histone mRNA transcription. However, we do not exclude the possibility that dPrp40 is regulating other cellular processes and causing phenotypic defects by modulating the pre-mRNA alternative splicing of important Drosophila genes. In fact, the FF domains that are critical for dPrp40 function are responsible for binding to Luc7 and Snu71 (Ester and Uetz, 2008), two proteins within the U1 snRNP complex. The effect of dPrp40 on splicing, however, seems not to be prominent in Drosophila tissues, according to the results of our genome-wide analysis of transcript- and exon-level changes in siRNA flies. Further studies will be required to fully characterize the function of the dPrp40 protein in mRNA synthesis and processing.

Our data demonstrate that dPrp40 depletion results in growth defects and that dPrp40 localizes to the HLB and regulates histone gene transcription. Although it is tempting to link the phenotypic defects resulting from dPrp40 loss of function with histone gene expression, we believe that dPrp40 may regulate cell growth and proliferation by mechanism(s) other than the regulation of histone genes. Our data support this notion. First, we show here that dPrp40 associates with the HLB during interphase and early mitosis (Figs 1 and 2). Despite the colocalization of dPrp40 with MPM2 (Fig. S1), which is associated with the HLB only during S phase when the bulk of histone protein synthesis occurs, we also detected dPrp40 staining during the starting phase of cell division (Fig. 2), when DNA replication is over. These data are consistent with dPrp40 being present in the HLB throughout interphase and early mitosis and therefore disengaged from the activation of histone gene transcription. The small increase in dPrp40 binding at the promoter sequences in cells in late S phase compared to at the G1/S transition (Fig. 6E) is not in disagreement with this hypothesis. Second, whereas the expression of the ΔWWdPrp40 construct resulted in the deficient accumulation of dPrp40 in the HLB (Fig. 1B), indicating a less-important role of the FF domains in the localization of dPrp40 to this nuclear compartment, the FF domains of dPrp40 were essential for rescuing the phenotype resulting from the loss of dPrp40 (Fig. 4D; Fig. S4C) and were also important in the activation of the histone H3/H4 promoter (Fig. 6F,G). Therefore, and in the absence of convincing evidence of dPrp40 having a direct role in histone mRNA metabolism, these observations suggest that the growth defects resulting from dPrp40 loss of function were not linked to the localization of dPrp40 at the HLB and the regulation of histone gene expression.

An interesting question that arises from our study regards the means by which dPrp40 might be targeted to the HLB. Seminal work by Duronio and coworkers provided evidence for an ordered process in Drosophila HLB assembly. Mxc and FLASH are first recruited to the HLB, whereas the other components, including U7 snRNP, Mute and other transcription and mRNA factors, are subsequently recruited in a histone gene transcription-dependent fashion (Duronio and Marzluff, 2017). Because of the reported association of the WW and FF domains of Prp40 with the phosphorylated C-terminal domain (phospho-CTD) of RNAPII (Morris and Greenleaf, 2000), an exciting possibility is that dPrp40 might be recruited to the HLB via a mechanism involving the phospho-CTD. Importantly, phosphorylated RNAPII is highly associated with the HLB during the S phase, when histone mRNA transcription activation occurs (Guglielmi et al., 2013). Several other interpretations are also possible. Interactions among HLB components are necessary for the ordered recruitment of additional HLB factors. For example, the C-terminal region of FLASH is necessary for the recruitment of U7 snRNP to the HLB (Tatomer et al., 2016). Similarly, dPRP40 might be recruited to the HLB complex through interactions of its WW domains with other components of the complex. Another mechanism potentially collaborating in the formation of the HLB complex involves phosphorylation by Cyclin E–Cdk2, which is essential for histone mRNA expression (Lanzotti et al., 2004). Although Mxc is a target of this kinase, Mxc localization to the HLB does not require Cyclin E–Cdk2 activity (White et al., 2011). The Spt6 HLB component is specifically immunoprecipitated by the phosphoprotein epitope-specific MPM2 monoclonal antibody, and phosphate treatment of the extract disrupts the interaction of Spt6 with the HLB complex, thus suggesting a role of Cyclin E–Cdk2 activity in Spt6 localization to the HLB (White et al., 2011). Because of the cyclin-dependent kinase consensus motif at position 739 of dPrp40, assessing the localization of dPrp40 to the HLB with respect to Cyclin E–Cdk2 activity would be informative.

In summary, we have characterized the function of Prp40 in Drosophila and identified dPrp40 as a new component of the HLB. We also showed that dPrp40 is required for normal embryonic development and might participate in histone mRNA biosynthesis. Further study of dPrp40 will clearly be useful to define the detailed mechanism of its function.

The following Gal4 Drosophila lines were used in this work: act5c-Gal4, en-Gal4, ptc-Gal4 and ey-Gal4 [Bloomington Drosophila Stock Center (BDSC), Indiana University, Bloomington, IN]. To restrict the expression of Gal4 in these lines, we used tubG80^ts (BDSC), which inhibits Gal4 expression at 25°C. The following UAS Drosophila lines were used: UAS-GFP (BDSC); UAS-RNAi-dPrp40 [Vienna Drosophila Resource Center (VDRC), Vienna, Austria; stock ID #26227] (Dietzl et al., 2007), which encodes a hairpin RNA that is complementary to dPrp40 endogenous gene; UAS-PRPF40B and UAS-PRPF40A, which were engineered after the insertion of human PRPF40B (Becerra et al., 2015) and mouse PRPF40A (IMAGE Consortium, Source BioScience, Nottingham, UK; clone ID 100061646) cDNA sequences into the PUASattb plasmid (kindly provided by Mar Ruiz-Gómez, CBMSO, Madrid, Spain) and subsequent transformation into ZH-attP-86FB embryos; UAS-dPrp40, which was generated by subcloning dPrp40-V5 from pAC5.1V5/hisA (this work) into the PUASattb plasmid; and UAS-ΔWWdPprp40, UAS-ΔFFdPrp40 and UAS-ΔNLSdPrp40, which were generated by subcloning the appropriate fragments obtained from the dPrp40 parental vector.

Clones of dPrp40 tissue were induced by mitotic recombination using the FLP-FRT recombination system (Golic, 1991; Xu and Rubin, 1993). Wing disc and salivary gland deficiency clones were identified by the loss of GFP expression, which was driven by the ubiquitin reporter Ubi-GFP. The genetic cross was performed as follows: yw hsFLP122;Ubi-GFP FRT40A females were crossed to df18.2 FRT 40A/CyO males (kindly provided by Maria Leptin, EMBL, Heidelberg, Germany) and allowed to lay eggs for 24 h. The vials were then heat-shocked at 37°C for 40 min to induce somatic clones. Salivary gland clones were induced during embryogenesis when the salivary gland precursor cells were still dividing mitotically (0–24 h).

To express the dPrp40 protein in E. coli, we amplified a segment containing amino acids 1 to 806 of the dPrp40 cDNA and cloned it in-frame into the expression vector PGEX2TK. The protein was expressed as a GST fusion protein under the conditions previously described (Faber et al., 1998). The purified fusion protein was used to generate in-house polyclonal antibodies in guinea pigs and rabbits following standard protocols.

Third-instar imaginal discs were dissected in PBS and fixed in 4% paraformaldehyde for 20 min at room temperature. Discs were subsequently washed in PBS, blocked in blocking buffer (PBS; 0.3% Triton X-100, 1% BSA), and incubated with the primary antibodies in blocking buffer at 4°C overnight. Washes were performed with blocking buffer, and the appropriate fluorescent secondary antibody was added for 1 h at room temperature. After extensive washes with blocking buffer, the discs were mounted in VECTASHIELD antifade mounting medium (Vector Laboratories-Palex, Sant Cugat del Vallés, Barcelona, Spain). Images were acquired using a Leica SP5 spectral laser confocal microscope and processed using the LAS AF software. The images were digitally processed for presentation using Adobe Photoshop CS5 extended software. The antibodies and dilutions were as follows: anti-dPrp40 (1:1000, guinea pig), anti-Mxc (1:500, rabbit; kindly provided by Robert J. Duronio, UNC, Chapel Hill, NC), anti-Lsm11 (1:1000, rabbit; kindly provided by Joseph Gall, Carnegie Institution for Science, Baltimore, MD), rabbit anti-FLASH (1:1000 dilution, kindly provided by Zbigniew Dominski, UNC, Chapel Hill), anti-MPM2 (1:500, mouse, catalog #05-368; Millipore, Burlington, MA), anti-V5 (1:500, mouse, catalog #R960-25; Invitrogen, Carlsbad, CA), anti-PH3 (1:100, rabbit, kindly provided by Mar Ruiz-Gómez, CBMSO, Madrid, Spain), rabbit antibody against cleaved human Caspase-3 (1:50, Cell Signaling Technology; #9662), Alexa Fluor 488-conjugated goat anti-mouse-IgG (catalog #A-11029; Molecular Probes, Eugene, OR, USA), Alexa Fluor 647-conjugated goat anti-mouse-IgG (catalog #A-21235; Molecular Probes) and Alexa Fluor 647-conjugated goat anti-rabbit-IgG (catalog #A-21244; Life Technologies, Paisley, UK). Secondary antibodies were used at a 1:500 dilution.

For immunoblotting, anti-dPrp40 (1:10,000) and anti-α-tubulin monoclonal anti-α-tubulin (1:5000 dilution, clone DM1A, Sigma-Aldrich; #T6199) primary antibodies were detected using HRP-conjugated secondary antibodies (PerkinElmer Life Science, Waltham, MA) at a 1:5000 dilution.

Drosophila S2 cells (kindly provided by Fátima Gebauer, Centre for Genomic Regulation, Barcelona, Spain) were grown in Schneider's Drosophila medium (Gibco-Thermo Fisher Scientific) supplemented with 10% fetal bovine serum and 1% penicillin/streptomycin/glutamine (Gibco-Thermo Fisher Scientific) under standard conditions. To arrest cells at the G1/S transition, 0.2 mM hydroxyurea (HU; Alfa Aesar-Thermo Fisher Scientific) was added for 24 h (HU arrest population). To obtain a synchronous population of late-S phase cells, cells were released from the HU arrest and allowed to progress through the S phase to late-S phase for 5 h (HU release population) (Li et al., 2016).

ChIP assays were performed on ∼10⁷ S2 cells per experiment following the detailed protocol described by del Blanco et al. (2012). To analyze histone core promoter regions, the following primers were used: RTHis1ChIP-fwd, 5′-AAACCCATATTACATCCTT-3′; RTHis1ChIP-rv, 5′-TTAACAATGCTACTGACA-3′; RTHis2abChIP-fwd, 5′-AACACAATTCACTTATCGTAAT-3′; RTHis2abChIP-rv, 5′-TTGCTTTGCGTTTATTTCA-3′; RTHis34ChIP-fwd, 5′-TGTTCGTTCGCTTTTCGCTC-3′; RTHis34ChIP-rv, 5′-CGTGAAGTAGTGAACGTGAAC-3′. The Drosophila wispy gene was used as a negative control and was amplified using the following primers: WispyChIP-fwd, 5′-GAATAATAAGAACGGATAC-3′; and WispyChIP-rv, 5′-CGTGAATATAAGTAAGATTA-3′. Glyceraldehyde-3-phosphate dehydrogenase (GAPDH) was used as the internal control and was amplified using the following primers: GAPDH2-fwd, 5′-CAATGGATTTGGTCGCATCG-3′ and GAPDH2-rv, 5′-CGAACTTGAACAGGTAGACC-3′. The amplification efficiency of the primers was evaluated by performing a dilution series experiment using genomic DNA. The following PCR program was used: 95°C for 5 min, followed by 46 cycles at 95°C for 30 s, 59.5°C for 30 s, and 72°C for 45 s (CFX96 thermocycler, Bio-Rad).

To create the H3/4-PGL3 construct, a 260 bp fragment containing the H3/H4 promoter sequences was amplified from wild-type genomic Drosophila DNA by PCR using the following primers: fwd, 5′-GTTCTCGAGTATTATACACG-3′ and rv, 5′-GATTTGGGTTTCACTAAGCTT-3′. The derived PCR product was cleaved with XhoI and HindIII and ligated into the PGL3 luciferase reporter vector (Promega, Madison, WI) cut with the same enzymes.

Control GFP and CG3542 dsRNAs were generated with a MEGAscript T7 in vitro transcription kit (Life Technologies) using GFP and DRSC00612 [Drosophila RNAi Screening Center (DRSC), Harvard Medical School, MA] template amplicons, following the manufacturer's protocol.

Subconfluent Drosophila S2 cells (2×10⁶) in a 35 cm² plate were transfected with 1 µg of dsRNA using Effectene transfection reagent (Qiagen Iberia, Madrid, Spain) according to the manufacturer's protocol. At 48 h post transfection, 2×10⁶ S2 cells were transfected with 200 ng of H3/4-PGL3, 50 ng of pRL-TK Renilla luciferase for the internal control, and 300 ng of dsRNA by using an Amaxa Nucleofector system (Lonza, Barcelona, Spain). For the rescue experiments, we transiently transfected 1 µg of the dPrp40-pAC5.1, ΔFFdprp40-pAC5.1 and ΔWWdprp40-pAC5.1 plasmids using the Nucleofector system. At ∼48 h after transfection, cells were harvested, lysed and processed using a Dual-Luciferase Reporter Assay Kit (Promega). Drosophila S2 cells were tested regularly for contamination.

Total RNA was isolated from third-instar larvae (L3) of the following genotypes: Act5c-Gal4/Gal80ts;UAS-RNAi-dprp40/+ and Act5c-Gal4/Gal80ts +/+. Larvae were maintained for 24 h at 29°C to allow Gal4 expression. Total RNA was extracted with TRIzol (Invitrogen), and 1 µg of RNA was reverse transcribed with qScript cDNA SuperMix (Quanta Biosciences-VWR, Barcelona, Spain) using the manufacturers' protocols. The transcripts were quantified by RT-qPCR using iQ SYBR Green Supermix (Quanta Biosciences) and a CFX96 thermocycler with the following oligonucleotides: his1-fwd, 5′-GACTGCAGCGAAGCCAAAG-3′ and his1-rv, 5′-CCGATGCAATCTTCACCGTC-3′; his2b-fwd 5′-GCTGGTGTACTTGGTGACAG-3′ and his2b-rv, 5′-CATCACCAGTCGGGAGAT-3′; his3-fwd, 5′-TCTGCAGGAAGCTAGCGAAG-3′ and his3-rv, 5′-GGTGACACGCTTGGCATGA-3′; and his4-fwd, 5′-CTTGCCTCTTCAGAGCGTAC-3′ and his4-rv, 5′-ACCTACACGGAACACGCC-3′. For the RT-qPCR measuring transcripts of different lengths, the primers used were PCR1-fwd, 5′-CTTGCCTCTTCAGAGCGTAC-3′ and PCR1-rv, 5′-ACCTACACGGAACACGCC-3′; and PCR2-fwd, 5′-CCTTTACCACGACCAGTCATTT-3′ and PCR2-rv, 5′-CGTGCTGTGCGTGTATAATAGT-3′. Alpha-tubulin at 84D (alphaTub84D) was used as an internal gene control using the primers RTAlpha-tubulin-fwd, 5′-CTACAACTCCATCCTAAC-3′ and RTAlpha-tubulin-rv, 5′-CGATTCAGGTTAGTGTAA-3′. The expression values obtained with Act5c-Gal4/Gal80ts;UAS-RNAi-dprp40/+ larvae were compared to the value obtained with control larvae, which was set at one.

A total of 5 µg of RNA extracted with TRIzol reagent was used for each S1 nuclease protection reaction, which was performed as previously described (Tatomer et al., 2016).

All experiments were repeated at least three times, and statistical analysis was performed using Prism 5.0 software (GraphPad). Two-tailed Student's tests (unpaired t-test) were used to compare the samples and their respective controls. P-values are represented by asterisks (*P≤0.05, **P≤0.01 and ***P≤0.001). The absence of an asterisk indicates that the change relative to the control is not statistically significant.

We thank Robert Duronio for the anti-Mxc antibody, Zbigniew Dominski for the anti-FLASH antibody, Joseph Gall for the anti-lsm11 antibody, Olga Makarova for the anti-PRPF40A antibody, and Maria Leptin for df18.2 FRT 40A/CyO flies. We are grateful to Robert Duronio and Kaitlin Curry for conversations about experiments and S1 nuclease assay advice. We also thank Mar Ruiz, Marta Carrasco-Rando and Fátima Gebauer for reagents and experimental advice during the course of this work. Stocks obtained from the Bloomington Drosophila Stock Center (NIH P40OD018537) and the Vienna Drosophila Resource Centre were used in this study.

The peer review history is available online at https://jcs.biologists.org/lookup/doi/10.1242/jcs.239509.reviewer-comments.pdf