Yeti, an essential Drosophila melanogaster gene, encodes a protein required for chromatin organization

The Bucentaur (BCNT) protein family, conserved from yeast to humans, is characterized by the so-called BCNT domain, a stretch of 80 amino acids located at the C-terminal end (Fig. 1). Despite its widespread distribution in animals and plants, however, the functions of this protein family remain largely unknown (Iwashita et al., 2003). Sparse information in the literature indicate that BCNT proteins might perform important cellular functions; the yeast BCNT protein, Swc5, is a component of the ATP-dependent chromatin remodeling complex SWR1, involved in the exchange of the histone variant H2A.Z with the canonical H2A within euchromatin (Kobor et al., 2004; Mizuguchi et al., 2004). The Drosophila melanogaster BCNT protein, YETI, was originally defined as a kinesin-binding protein able to bind both subunits of the microtubule-based motor kinesin-I (Wisniewski et al., 2003). Related BCNT proteins have been found in both cytoplasmic and nuclear fractions (Iwashita and Naoki, 2011), and they have been shown to asociate with kinetochores in chicken (Ohta et al., 2010). Additional studies have linked the BCNT proteins to craniofacial development and osteogenesis, though limited mechanistic understanding is available. First, the mouse BCNT protein, called CP27 or CFDP1, plays essential cellular functions and is required for teeth and bone development (Diekwisch et al., 1999; Diekwisch et al., 2002). Second, in mouse, downregulation of the Cfdp1 gene correlates with the induction of cleft palate due to persistent expression of the PAX3-encoding gene in neural crest cells (Wu et al., 2008); upregulation of Cfdp1 was also observed after experimentally induced osteodifferentiation (Bustos-Valenzuela et al., 2011). Third, the Zebrafish BCNT gene, called Rltpr, is developmentally expressed in the branchial arches that will give rise to the craniofacial structures (Thisse, 2004). Finally, the human Cfdp1 gene has been shown to be a target of TFII-I transcription factors, which in turn are encoded by genes suggested to be primary candidates for Williams–Beuren disease, which causes – among others symptoms – craniofacial abnormalities (Makeyev and Bayarsaihan, 2011).

In this work, we have carried out in-depth functional studies of the Yeti gene and its encoded product, YETI, in Drosophila melanogaster. We show that the YETI protein is required for higher-order chromosome and/or chromatin organization during development. In addition, our results provide evidence suggesting that YETI interacts with HP1a and with the H2A.V-exchanging machinery and participates in the step of H2A.V deposition.

We previously found that l(2)41Aa, a vital heterochromatin gene of Drosophila melanogaster (Hilliker, 1976; Dimitri et al., 2003; Rossi et al., 2007), is required for proper chromosome organization in both mitosis and meiosis (Cenci et al., 2003a). To identify the actual gene at the molecular level, we screened for P-element-induced mutations that failed to complement l(2)41Aa alleles and recovered two non-complementing lethal insertions, denoted LP1 and LP2 (see Materials and Methods). Molecular analysis showed that LP1 is a P-insertional allele that, when homozygous, results in prolonged larval development followed by lethality before pupation. Mutant larvae exhibited severely reduced imaginal discs and brains, with melanotic masses in their hemocoel (Fig. 2A). Cytological examination of DAPI-stained salivary gland cells from LP1/LP1 homozygous and LP1/Df(2R)B hemizygous larvae revealed decondensed and disorganized chromosomes (Fig. 2B); all examined polytene cells displayed such aberrant chromosome morphology, albeit at different degrees. Moreover, polytene chromosomes and nuclei of LP1/LP1 homozygous salivary glands are about 40% smaller than those present in wild-type salivary glands.

The flanking sequences of the LP1 insertional allele were recovered by inverse PCR, sequenced and used to search the Drosophila melanogaster genome database. BLAST analysis using the 3′ end flanking sequences of LP1 revealed 100% identity with the 5′ end region of CG40218, also called Yeti, a 967-bp-long gene located in scaffold AABU01002199. The P-element was found to be inserted within the predicted start codon ATG of Yeti at position +2. To further verify the connection between the genetic locus l(2)41Aa and the Yeti gene sequence, we performed PCR amplification and sequencing of the Yeti region using genomic DNA from larvae homozygous for EMS-31, a lethal allele of l(2)41Aa. Sequence inspection revealed that EMS-31 carried a single base pair mutation (from A to T) at position 106, where codon AAA for Lys is changed into a TAA stop codon, most likely producing a truncated protein lacking ∼60% of the protein including the evolutionary conserved C-terminal region (Fig. 1). Taken together, these results indicate that the l(2)41Aa vital gene corresponds to the Yeti gene.

Northern analysis using the 0.9-kb Yeti cDNA as a probe revealed a transcript of the expected size (∼0.9 kb) in larvae of the Oregon-R wild-type strain, which was absent in LP1/LP1 homozygotes (Fig. 2C) indicating that LP1 is a null allele of Yeti. The transcript was detected throughout development in embryos, larvae, pupae, adults and ovarian RNA (Fig. 2D). To further confirm whether the Yeti product is required for proper chromosome organization, we depleted the Yeti gene in S2 cells using double-stranded RNA (dsRNA)-mediated interference (RNAi) (Fig. 2E). The RNAi-treated cells displayed abnormally condensed chromosomes with fuzzy and loose appearance in ∼70% of examined metaphases (n = 457), versus the 3% baseline level recorded in mock-treated cells. Remarkably, these defects of chromosome organization strongly resemble those originally characterized in l(41A) mutants (Cenci et al., 2003a).

If Yeti actually corresponds to the vital heterochromatin gene l(2)41Aa, then reconstituting LP1 homozygous mutant strains with exogenous Yeti gene product should rescue the defective phenotypes observed in mutants. To assess whether that was the case, we generated transgenic flies carrying the full-length Yeti cDNA. Overexpression of the UAS-Yeti-M17 transgene transgene using tub-GAL4 to induce the UAS promoter rescued indeed both the larval lethality and the chromosome morphology phenotypes associated with LP1 null alleles (supplementary material Fig. S1A; details in Materials and Methods). To assess the expression of the YETI protein in LP1/LP1; UAS-Yeti-M17/tub-GAL4 larvae, we produced polyclonal antibodies directed against the conserved C-terminal portion of the YETI protein (see Materials and Methods) and analyzed protein extracts by western blotting. A single 32-kDa band was detected in extracts from LP1/LP1; UAS-Yeti-M17/tub-GAL4 larvae (Fig. 2F), but not in extracts from LP1/LP1; UAS-Yeti-M17/+ larvae, which do not express the M17 transgene in the absence of the tub-GAL4 driver, or from LP1/Df null hemizygotes lacking the functional Yeti gene. These results therefore demonstrate that the Yeti gene (1) fully complements the vital heterochromatin gene l(2)41Aa, and (2) encodes a protein required for proper chromosome organization, which is highly defective in mutant strains.

Our previous work with l(2)41Aa (Cenci et al., 2003a) and the results reported thus far implicate the YETI protein in chromatin organization. To gain more understanding of how YETI might perform this function, we sought to localize the protein in salivary gland polytene chromosomes using anti-YETI in immunostaining assays, but cytological analyses failed to reveal any clear staining, suggesting that the antibody, although performing well in western blot assays of denatured protein extracts (Fig. 2F), could not react with the protein in its native conformation in whole cells. To circumvent this problem, we generated transgenic Drosophila melanogaster lines carrying a UAS-Yeti-GFP transgene (see Materials and Methods) and induced its expression in salivary glands and brains using the elav-GAL4 driver (Fig. 3A). Anti-GFP antibody specifically detected the YETI–GFP fusion protein in total protein extracts from salivary glands of larvae carrying the UAS-Yeti-GFP transgene activated by elav-GAL4 driver (Fig. 3B). GFP immunofluorescence assays in these larvae show that the YETI–GFP fusion protein is present in all nuclei and is recruited to hundreds of sites along polytene chromosome arms (Fig. 3A), yielding a staining pattern that overlaps with that produced by DAPI-staining. On mitotic cells of neural ganglia isolated from third-instar larvae, anti-GFP staining was present in nuclei but non detectable on chromosomes (supplementary material Fig. S2). Moreover, in LP1/LP1; tub-GAL4/Yeti-GFP larvae, expressing the Yeti-GFP transgene activated by tub-GAL4 driver, a rescue of aberrant polytene chromosome morphology was seen (Fig. 3C) and the staining pattern was similar to that found in elav-GAL4; Yeti-GFP flies (Fig. 3C). This result shows that the YETI–GFP fusion protein binds to chromatin in the absence of the YETI wild-type protein and is able to rescue the aberrant chromosomal phenotype found in Yeti mutants.

The BCNT domain is a highly conserved stretch of about 80 amino acids located in the C-terminal portion of the Bucentaur protein family (Fig. 1), the function of which is unknown. To test whether the BCNT domain of YETI is required for chromatin binding, we constructed three different transgenic Drosophila melanogaster lines carrying diverse Yeti-3×FLAG transgenes: (1) the Yeti-3×FLAG transgene bearing a complete open reading frame of the Yeti gene, (2) the Yeti-Cter3×FLAG transgene (containing residues 145–241), and deleted for the 5′ end, and (3) the Yeti-Nter3×FLAG transgene (containing residues 1–166) and lacking the 3′ end of Yeti, which includes the BCNT domain (at 159–238) (Fig. 3D,E; see Materials and Methods). We next induced the expression of each transgene in salivary glands by using the elav-GAL4 driver and performed both immunostaining and western blot assays using anti-FLAG antibodies (Fig. 3E). A clear staining of salivary gland nuclei was observed using anti-FLAG in larvae expressing each of the three transgenes, showing that both truncated YETI proteins, like YETI–GFP, are able to enter the nucleus (see the inserts in Fig. 3E). However, when we immunostained polytene chromosomes, we found GFP fluorescent signals in larvae expressing both full-length Yeti-3×FLAG transgene and the Yeti-Cter3×FLAG-transgene, but not in larvae expressing the Yeti-Nter3×FLAG transgene (Fig. 3E). The results of these experiments, therefore, show that the BCNT domain of YETI is essential for chromatin binding.

The SWC5 protein, the yeast ortholog of YETI, is a component of the SWR1 ATP-dependent remodeling complex and is required for the recruitment of variant histone Htz1 into chromatin (Kobor et al., 2004; Mizuguchi et al., 2004; Wu et al., 2005). Such a role would be consistent with the observed defects in chromosome/chromatin organization in the absence of YETI function (Cenci et al., 2003a; this paper). We therefore examined the pattern of H2A.V histone variant in polytene chromosomes of wild-type and Yeti mutant strains by immunofluorescence methods. As shown in Fig. 4A,C, Drosophila melanogaster wild-type salivary glands show multiple H2A.V sites of varying fluorescence intensity along the chromosome arms and in the chromocenter, consistent with previous reports (Leach et al., 2000). Yeti mutants [both LP1/LP1 homozygotes and LP1/Df(2R)B hemizygotes] showed instead only residual background fluorescence, with rare faint signs of banding (Fig. 4A,C). This finding prompted us to assess whether other histone types are also affected in the mutant. To address that question we also studied the distribution of H2A, H2B, H3 and H4 chromosomal histones. We found that, indeed, the abundance of Histone H2A was dramatically reduced in Yeti mutant polytene chromosomes, similar to the H2A.V variant (Fig. 4A,C). The levels of the H2B, H3 and H4 species are also affected on polytene chromosomes of Yeti mutants, but to a lesser extent compared to those of H2A.V and H2A (Fig. 4A,C). These results indicate that loss of Yeti function preferentially impairs the chromosomal accumulation of H2A.V and H2A.

As loss of Yeti affects the abundance of nucleosomal histones in polytene chromosomes, we assessed the overall level of chromosomal epigenetic marks required for the establishment and maintenance of chromatin organization and function, e.g. histone acetylation and methylation, in wild-type and mutant strains. Immunofluorescence experiments showed that levels of H3K9ac, H3K9me3, H3K4me3 histone modification marks were indeed strongly reduced in Yeti mutants (Fig. 4A). However, the levels of H3K9me3, H3K4me3 in the chromocenter were comparable to that of the controls (Fig. 4A,C).

Post-translationally modified histones are recognized by and interact with different multiprotein complexes that ensure the organization of chromatin in open or closed states, leading, in turn, to either activation or repression of neighboring gene expression. These complexes contain one or more subunits responsible for histone modification, a chromatin remodeling ATPase and other chromatin proteins (Hopfner et al., 2012; Simon and Kingston, 2013). We therefore tested a panel of known major chromatin factors, such as ISWI and DOM-A remodelers, HOAP telomeric protein (also known as Caravaggio), HP1a and Histone Deacetylase RPD3 (James et al., 1989; De Rubertis et al., 1996; Corona et al., 1999; Ruhf et al., 2001; Shareef et al., 2003; Cenci et al. 2003b; Kusch et al., 2004), for their incorporation in polytene chromosomes. The chromosomal levels of ISWI and HOAP in Yeti mutants were comparable to those of wild-type strains; however, those of DOM-A (Fig. 4B,C) were reduced by ∼70%. In addition, we found reduced chromosomal abundance of HP1a and Histone Deacetylase RPD3. However, the HP1a levels at chromocenter were comparable to those of the controls (Fig. 4B,C).

The results thus far indicate a substantial decrease in the accumulation of both H2A.V and H2A, and of some histone modifications and chromatin factors, along polytene chromosomes of Yeti mutants. Western blotting assays of whole protein extracts (Fig. 5A) depicted a reduced overall protein amount of both H2A.V and H2A and specifically modified histones in Yeti mutants; HP1a and RPD3 also showed reduced protein levels in Yeti mutants.

In Drosophila melanogaster, the loss of chromatin remodelers, such as ISWI, leads to defects in higher-order chromosome structure similar to those found in Yeti mutants and affects transcriptional regulation (Corona et al., 2007). We next performed qRT-PCR assays for genes encoding H2A.V, nuclesomal histones, HP1a and RPD3, to test whether their transcript abundance is perturbed in LP1/LP1 mutants. The results of this analysis show that the level of transcripts of genes encoding H2A.V, nucleosomal histones, HP1a and RPD3 is indeed affected in LP1 mutants (Fig. 5B): decreased levels of transcripts for histones and chromatin marks were apparent and correlated with the reduced protein levels found by both immunofluorescence and western analyses (Fig. 4; Fig. 5A).

The Drosophila melanogaster histone variant H2A.V is known to play roles in HP1a recruitment (Swaminathan et al., 2005). Consistent with this, our present findings show, among other chromosomal defects, a poor content of both H2A.V and HP1a in polytene chromosomes of Yeti mutants (Fig. 4).

Interestingly, loss-of-function mutants for YETI, H2A.V, HP1a and DOM-A, all have a similar lethal phenotype characterized by prolonged larval development followed by lethality before pupation and by the presence of large melanotic masses in the larval hemocoel (Minakhina and Steward, 2006). Notably, DOM-A is the ATPase subunit of the Tip60 remodeling complex of Drosophila (Ruhf et al., 2001; Kusch et al., 2004), which catalyzes the exchange of phosphorylated H2A.V with H2A during the repair of DNA damage (Kusch et al., 2004) and is also proposed to exchange back H2A with H2A.V, thus facilitating incorporation of the latter in nucleosomes (Baldi and Becker, 2013).

Based on these observations, we decided to investigate whether YETI might interact with H2A.V, HP1a and DOM-A proteins. Using the GFP-TRAP system, we immunoprecipitated H2A.V–GFP fusion protein from extracts of salivary glands of larvae expressing H2A.V–GFP and then analyzed the immunoprecipitates by western blotting using antibodies against YETI or GFP (to visualize H2A.V). As shown in Fig. 6A, YETI was selectively found in the immunoprecipitate from H2A.V–GFP extracts, but was absent from control Oregon-R immunoprecipitate. Reciprocally, H2A.V was found in the YETI–GFP immunoprecipitate (Fig. 6B). However, HP1a was not detected in the same YETI–GFP immunoprecipitate (data not shown). Thus, we also performed GST pulldown experiments to test the interaction between YETI and HP1a. As shown in Fig. 6C,D, GST–HP1a, but not GST alone, precipitates YETI from YETI–GFP-expressing larval salivary gland extracts.

Next, we immunoprecipitated DOM-A from salivary glands extracts of elav-Gal4; UAS-Yeti-GFP larvae (expressing the YETI–GFP fusion protein). As shown in Fig. 7A, YETI–GFP was recovered in the immunoprecipitates with anti-DOM-A antibody. Taken together, these results suggest that DOM-A interacts with YETI. The present findings that YETI interacts both with DOM-A and with H2A.V, together with the known role of DOM-A in recruitment of H2A.V, would place DOM-A upstream of YETI and H2A.V in events involved in the recruitment of chromatin proteins.

We therefore directly asked whether DOM-A is actually responsible for the recruitment of YETI to chromatin. To this aim, we examined the distribution of YETI–GFP in polytene chromosomes of homozygous larvae for domino3 (dom3), a null allele of domino. The domino mutant larvae undergo prolonged larval development followed by lethality before pupation and show large melanotic masses in their hemocoel. By genetic crosses we obtained dom3/dom3; Tub-Gal4/UAS-Yeti-GFP larvae, such that YETI could be visualized by its GFP fluorescence, both in salivary glands in vivo and in fixed polytene chromosomes. Comparable amounts of YETI–GFP were present in both dom3/dom3; Tub-Gal4/UAS-Yeti-GFP (lane 1) and dom3/CyS; Tub-Gal4/UAS-Yeti-GFP control (lane 2) salivary glands (Fig. 7C). As shown in Fig. 7B, domino mutant salivary gland nuclei exhibit intensely fluorescent YETI signals, whereas, in contrast, only a residual background fluorescence is visible on polytene chromosomes (Fig. 7C). Thus, in domino mutants the YETI–GFP protein enters the nucleus but its deposition onto chromatin is abolished. Taken together, these results suggest that YETI is recruited to chromatin by DOM-A and is in turn implicated in regulating H2A.V, which is another downstream target of DOM-A.

In present study, we have carried out a functional analysis of the Drosophila melanogaster Yeti gene and its encoded product, YETI. The results show the Yeti is an essential gene and indicate a previously unrecognized role of YETI protein as a key chromatin factor, providing mechanistic clues on possible roles of the BCNT protein family in eukaryotes. In particular, based on the chromatin defects found here in Yeti mutants, as well as the newly emerging interactions of YETI–GFP with H2A.V, DOM-A and HP1a, we propose that YETI interacts with H2A.V-exchanging machinery, as a chaperone or as a new subunit of the Drosophila DOM/Tip60 complex, and participates in the step of H2A.V deposition.

The functions of the YETI protein were poorly understood thus far. We found that chromosome structure is drastically perturbed in Yeti mutants (Cenci et al., 2003a; Fig. 2B,E), suggesting that YETI is involved in chromosome and/or chromatin organization. YETI protein has also been proposed to bind both subunits of the microtubule-based motor kinesin-I (Wisniewski et al., 2003), and previous immunolocalization experiments have shown both a nuclear and cytoplasmic localization of epitope-tagged YETI in Drosophila S2 cultured cells (Wisniewski et al., 2003). In those experiments, however, it remained unclear whether YETI was located within the nuclei or associated with the nuclear membrane.

The present results clearly show that the YETI–GFP fusion protein is abundant in nuclei and associates with polytene chromosomes (Fig. 3A,C). Notably, our experiments demonstrate that the evolutionary conserved BCNT domain is indispensable for YETI binding to polytene chromosomes (Fig. 3D,E), providing the first direct evidence for an essential role of this domain in chromatin binding.

Furthermore, we find that YETI–GFP interacts with H2A.V, HP1a and DOM-A (Fig 6; Fig. 7A). DOM-A is the ATPase subunit of the Tip60 remodeling complex that, similar to the related SWR1 complex in yeast, is likely to incorporate H2A.V into nucleosomes by exchange of H2A with the variant (Baldi and Becker, 2013). Like the related yeast SWIR1 protein, DOM-A might be the platform required for the assembly of the other subunits of the Tip60 complex. In support of this view, we now show that YETI–GFP requires DOM-A to be recruited to chromatin (Fig. 7C).

The defects in higher-order chromosome structure characterized here in Yeti mutants are accompanied by reduced chromosomal levels of H2A.V, of nucleosomal histones and of epigenetic marks (Fig. 4). However, not all tested histones and epigenetic marks are equally affected in Yeti mutants compared to controls; the chromosomal levels of H2A.V and H2A are dramatically decreased, whereas those of H2B, H3 and H4 are reduced to a lesser extent (Fig. 4A,C). In addition, among tested chromatin proteins, the chromosomal levels of ISWI and HOAP remained unaffected, those of DOM-A were decreased by 30% and those of HP1a only decreased in euchromatin (Fig. 4B,C). Taken together, these results indicate that the loss of YETI protein does not generally affect chromatin and/or chromosome organization. We can also exclude a primary defect on DNA replication, as that would result in a significant percentage of chromosome breaks (Branzei and Foiani, 2008), which are not detected in the Yeti mutants (Cenci et al., 2003a; Fig. 4E). Finally, qRT-PCR analysis of Yeti mutants showed downregulation of the genes encoding the proteins tested by immunostaining and western assays.

It is worth mentioning that the lethal phenotype of Yeti mutants (prolonged larval development followed by lethality before pupation and presence of melanotic masses) is highly reminiscent of that produced by mutations in genes encoding H2A.V and chromatin proteins, such as DOM-A, HP1a and Drosophila Rsf1 (Minakhina and Steward, 2006; Hanai et al., 2008). Interestingly, the Rsf1 remodeler interacts with H2A.V and members of the Tip60 complex (Hanai et al., 2008), suggesting a functional link between those chromatin factors. The lack of essential chromatin regulators that control chromatin architecture and gene expression might well affect blood cell differentiation, the alterations of which result in melanotic tumors. Interestingly, it has been recently found that a network of epigenetic regulators controls the establishment of hematopoietic cells in zebrafish (Huang et al., 2013).

In S. cerevisiae the H2A histone variant, called Htz1, is incorporated into chromatin by the SWR1 ATP-dependent chromatin-remodeling complex (Wu et al., 2005). The Swc5 subunit of the SWR1 complex, the ortholog of YETI, has been suggested to be required for eviction of H2A from nucleosomes prior to Htz1 loading carried out by Swc2 and Swr1 (Lavigne et al., 2009; Wu et al., 2009; Morillo-Huesca et al., 2010). Our findings that H2A.V deposition is perturbed in Yeti mutants and that YETI interacts with both H2A.V and DOM-A (Fig. 6A,B; Fig. 7A), strongly suggest that YETI, like his yeast homolog Swc5, might contribute to loading the variant H2A.V onto nucleosomes. The dramatic defect in H2A deposition also found in Yeti mutants (Fig. 4) suggests that, in YETI defective cells, H2A is removed by the DOM/tip60 complex but not replaced with H2A.V, resulting in depletion of both histones and disrupting chromatin integrity and function. A similar scenario has been proposed in yeast (Morillo-Huesca et al., 2010).

Moreover, the observation that YETI–GFP interacts with HP1a in GST pulldown assays (Fig. 6C), suggests that YETI might be an important mediator in targeting HP1a to chromatin remodeling regions. Association between HP1a and other chromatin remodeling factors has been reported (Fischer et al., 2009; Lavigne et al., 2009). Human HP1α interacts with the BRG1 protein of the SWI/SNF ATP-dependent remodeling complex, suggesting a direct role for BRG1 in the modulation of HP1α-containing higher-order chromatin structures (Nielsen et al., 2002).

The loss of Drosophila melanogaster chromatin remodelers such as ISWI leads to defects in higher-order chromosome structure similar to those found in Yeti mutants and affects transcriptional regulation (Corona et al., 2007). H2A variants and HP1 play key roles in both transcriptional activation and repression control in different species (Stargell et al., 1993; Dhillon and Kamakaka, 2000; Piacentini et al., 2003; Jin et al., 2005; Swaminathan et al., 2005; Keogh et al., 2006; Mavrich et al., 2008; Petesch and Lis, 2008; Henikoff et al., 2009; Cryderman et al., 2011; Kotova et al., 2011; Baldi and Becker, 2013). The drastic depletion of both H2A.V and H2A found in Yeti mutants might either directly or indirectly impair chromatin organization and function. H2A.V knockdown in Drosophila melanogaster indeed results in inefficient processing of the histone mRNA (Wagner et al., 2007), giving rise to a reduction of histone gene transcription and histone protein synthesis; this can affect in turn the global transcriptional output (Celona et al., 2011).

In the light of our results, we suggest that the specific impairment of H2A.V exchange and HP1a recruitment in YETI-defective cells yields a global chromatin unbalance, which in turn reverberates on the epigenetic control of gene expression, thus producing a ‘boomerang’ effect with a generalized impact on chromatin organization and function.

In the past decade, growing lines of evidence indicate that mutations in genes encoding proteins directly or potentially implicated in chromatin organization and remodeling are crucial players in human developmental disorders, including craniofacial development (Bickmore and van der Maarel, 2003; Fog et al., 2012). Thus, the identification of new factors required for chromatin dynamics and organization can expand our understanding of mechanisms underpinning developmental diseases. Several studies implicate CFDP1, the human ortholog of YETI, in osteogenesis and craniofacial development (Diekwisch et al., 1999; Diekwisch et al., 2002; Thisse, 2004; Wu et al., 2008; Bustos-Valenzuela et al., 2011; Makeyev and Bayarsaihan, 2011). Future research can provide a better understanding of BCNT protein functions in chromatin organization and yield more insight into their role in developmental disorders.

Fly cultures and crosses were carried out at 25°C in standard cornmeal yeast medium. The fly line #5941 of the Bloomington Drosophila Stock Center expresses H2A.V–GFP. To induce expression of YETI–GFP we used the fly lines #5138 and #5112 containing the Tub-Gal4 and elav-Gal4 drivers, respectively. The T(2;3) translocation carrying the dominant Cy and Tb markers used as balancer of Yeti alleles is designated ‘ST’ in the text. Df(2R)B is a chromosome carrying a partial deletion of the heterochromatin of the right arm chromosome 2 (2R) that fails to complement l(2)41Aa (Hilliker, 1976; Rossi et al., 2007) and LP1. Df(2R)M41A10 is a large deletion lacking almost all the heterochromatin of 2R together with all known essential genes and gene models (Corradini et al., 2003; Rossi et al., 2007). The EMS-31 allele is an EMS-induced allele of l(2)41Aa (Hilliker, 1976; Cenci et al., 2003a; Dimitri et al., 2003). To recover insertional alleles of the l(2)41Aa gene, we induced mobilization of a homozygous viable and fertile P element [P(w+)19.74.3] that colocalizes cytologically with l(2)41Aa in region h41 of 2R mitotic heterochromatin (Corradini et al., 2003; Fig. 2). By inverse PCR and sequencing, we found that 19.74.3 is located in scaffold AABU01002199, about 2 kb from the 5′ end of CG40218. P(w+)19.74.3/Cy S; delta 2-3/TM3, Sb Sr males were generated carrying simultaneously the P(w+)19.74.3, heterochromatic element and the delta 2-3 element as a constitutive source of transposase. In the germline of P(w+)19.74.3/Cy S; delta 2-3/TM3 males, the single P element jumps all around the genome, with some preference for insertion in nearby heterochromatin (Zhang and Spradling, 1993). P(w+)19.74.3/Cy S; delta 2-3/TM3 males were crossed to Cy S/Sco females to recover P(w+)19.74.3*/Cy S males carrying putative new insertions [P(w+)19.74.3*] in the vital heterochromatin genes of chromosome 2. P-element-induced recessive lethal alleles of vital heterochromatic genes were selected over Df(2R)M41A10. Complementation tests were performed by crossing single P(w+)19.74.3*/CyS males to Df(2R)M41A10/CyS females. The absence of P(w+)19.74.3*/Df(2R)M41A10 Cy⁺ flies among the offspring indicated that the tested P(w+)19.74.3* chromosome carries a lethal mutation uncovered by Df(2R)M41A10. For each lethal mutation, balanced stocks were established. Lethal mutations were tested for their ability to complement a collection of EMS-induced lethal alleles of known genes of 2R heterochromatin. After mobilization of P(w+)19.74.3, out of 1077 tested chromosomes we recovered two chromosomes denoted LP1 and LP2 that resulted in lethality when heterozygous with Df(2R)M41A10 and alleles of l(2)41Aa vital gene.

Yeti-GFP transgenic lines of Drosophila melanogaster were made as follows: (1) the Yeti cDNA was cloned in-frame with the reporter GFP sequence in the transformation vector pUASt; (2) the pUAS-Yeti-GFP construct was injected in pre-blastoderm Drosophila embryos of w¹¹¹⁸ genotype; (3) transgenic flies were selected using the white⁺ phenotypic marker. Transgenic lines carrying the Yeti-cDNA (M17 and M7) under control of UAS sequences were also made using the same experimental strategy.

Yeti-3×FLAG transgenic lines of Drosophila melanogaster were made as follows: full-length, N-terminal and C-terminal sequences of Yeti cDNA were cloned in frame with the FLAG epitope into the p3xFLAG-CMV-10 expression vector (Sigma); the different Yeti-3×FLAG fragments were then subcloned into the pUASt destination vector. The same procedures for injections and selection of transgenic flies were used for YETI–GFP and YETI–cDNA constructs.

Diffuse expression of Yeti-GFP, Yeti-cDNA (line M17) and Yeti-3×FLAG transgenes was induced by crosses with enhancer trap lines carrying the ubiquitously expressed tub-GAL4 driver or the elav-GAL4 driver that activates expression in larval neuroblasts and salivary glands. To rescue the lethal phenotype of the LP1 null allele, w/w; LP1; tub-GAL4/T(2;3)ST, Cy Tb females were crossed to w/Y; LP1; UAS-Yeti-M17/T(2;3)ST, Cy Tb males. In the progeny of this cross, a 2∶1 ratio was found between Cy Tb and Cy⁺ Tb⁺ progenies, the latter corresponding to the LP1/LP1; UAS-Yeti-M17/tub-GAL4 flies.

Rabbit polyclonal antibodies against Drosophila YETI were raised against a synthetic peptide (CFEMEKKLRLSRRPY) located from position 227 to position 241 of the C-terminal region of the protein. Polyclonal antibodies against H2A.V, HP1a, DOM-A, ISWI, HOAP, RPD3 and GFP (for immunostaining) were kindly provided by Robert Glazer (Wadsworth Center, Div. Genetic Disorders New York State Department of Health & University at Albany, USA); Sarah Elgin (Department of Biology Washington University in St. Louis, USA); Marie-Laure Rufh; Davide Corona (Dipartimento di Biologia Cellulare e dello Sviluppo, Università degli Studi di Palermo, Italy); Laura Ciapponi (Dipartimento di Biologia e Biotecnologie, Sapienza Università di Roma, Italy); Thomas Rudolph (Institute of Genetics, Halle University, Germany); and Gianluca Cestra (Dipartimento di Biologia e Biotecnologie, Sapienza Università di Roma, Italy), respectively. The following commercial antibodies against Drosophila proteins were used: anti-H2A, anti-H2B, anti-H3 and anti-H4 (Millipore), anti-HP1a, anti-H3K9me3 and anti-H3K9ac (Abcam), anti-Tubulin (Sigma), anti-GFP-HRP (vector).

Polytene chromosome squashes and immunolocalization experiments to Drosophila polytene chromosomes performed as described previously (Deuring et al., 2000). Rabbit antibodies against GFP, H3K9ac, ISWI, H2A.V or H2A, and mouse antibodies against H3met K9 were used at 1∶100 dilution. Rabbit antibodies against HOAP were used at 1∶50 dilution; as a secondary antibody a rabbit monoclonal Alexa-Fluor-conjugated antibody (Life technologies) was used at 1∶300 dilution. Rat anti-DOM-A antibodies were used at 1∶500 dilution; secondary rat monoclonal Alexa-Fluor-conjugated antibody antibodies (Life technologies) was used at 1∶200 dilution. Chromosomes were analyzed by a computer-controlled Nikon Eclipse 50i epifluorescence microscope equipped with a CCD camera. Quantification of chromosomal levels of H2A.V, nucleosomal histones, histone modifications and chromatin proteins were performed using the NIS-Elements software provided by Nikon.

Drosophila S2 cells were cultured at 25°C in Shields and Sang M3 medium (Sigma-Aldrich, St. Louis, MO) supplemented with 10% heat-inactivated FBS (Sigma-Aldrich). RNAi treatments were carried out according to Clemens et al. (Clemens et al., 2000). Production of dsRNA and cytological analysis of RNAi-treated cells were performed according to Somma et al. (Somma et al., 2002).

Total RNA and proteins were extracted from larval brains and salivary glands. Total RNA preparations, and northern and western blotting were according to standard protocols (Crowley et al., 1984; Sambrook et al., 2001). Antibodies against histones and α-Tubulin were used at the following dilution: 1∶2500. Rabbit, mouse and rat secondary antibodies were used at 1∶5000 dilution. The bands were immunodetected using the ECL kit from BIORAD.

Total RNA was extracted from wandering third-instar larvae with the Tri Reagent (Sigma) according to the manufacturer's instruction. RNA (2 γ) were retrotranscribed with the RETROscript kit (AM1710) according to the manufacturer's instruction. RT-qPCR was performed on an ABI 7300 Real-Time PCR System, according to the supplier protocol and using the following PCR conditions: 10 min at 95°C, 40×(15 s at 95°C, 1 min at the annealing/extension temp of 60°C) 10 min at 72°C. Primers and annealing temperatures were as follows: H2A-F, 5′-CGGCCAGATATTCCATTACG-3′; H2A-R, 5′-GGCTGCTCCTACTGATCCTG-3′; H2B-F, 5′-TGACACCGGAATTTCGTCGA-3′; H2B-R, 5′-TTGGTTCCCTCACTGACAGC-3′; H2A.V-F, 5′-CGAAACCGAATTCCGTAGAA-3′; H2A.V-R, 5′-GCCAGCCATTTCTGTTCAAT-3′; H3-F, 5′-GCTCGGTGCTCTTTTGGTAG-3′; H3-R, 5′-ATGGCTCGTACCAAGCAAAC-3′; H4-F, 5′-GTGTGAAGCGCATATCTGGA-3′; H4-R, 5′-TAACCGCCAAATCCGTAGAG-3′; HP1-F, 5′-CAAGCGAAAGTCCGAAGAAC-3′; HP1-R, 5′-ACCATTTCTGCTTGGTCCAC-3′; RPD3-F, 5′-CAAGGACACCAACAGCAAC-3′; RPD3-R, 5′-TCGCTGCGGCATCATTATC-3′; Rpl32-F, 5′-GCGCACCAAGCACTTCATC-3′; Rpl32-R, 5′-TTGGGCTTGCGCCATT-3′.

Data were analyzed according to Livak and Schmittgen (Livak and Schmittgen, 2001). Relative expression was calculated using the comparative Ct method (2^−ΔΔCt); mRNA expression was normalized against the average values of RpL32. The values in Fig. 5 were calculated from four independent experiments, with three replicates each.

Genomic DNA from two to three flies was prepared as described previously (Eggert et al., 1998) and digested using either HhaI or HpaII. DNA equivalent to approximately one-fifth of a fly was ligated using the Rapid Ligation kit (Roche). Inverse PCR was performed according to standard procedures (Dobie et al., 2001) and sequencing was performed by the MWG company. BLASTN against nonredundant database was performed on flanking sequences >25 bp in order to identify homologous gene models or repeated sequences.

Wandering third-instar larvae were dissected in physiological solution; brains and salivary glands were resuspended in lysis buffer [20 mM HEPES-KOH pH 7.9, 1.5 mM MgCl₂, 10 mM KCl, 0.1 mM DTT, 420 mM NaCl, 30 mM NaF, 25 mM β-glycerophosphatase, 0.2 mM Na₃VO₄, 0.1% NP-40, 0.5 mM PMSF, 1× Protease Inhibitor Cocktail (Roche)] and incubated on ice for 1 hour. GFP-TRAP assays in Drosophila was performed according to manufacturer's instructions (Chromotek). The proteins of interest was detected using the specific antibodies against the following proteins at the given dilutions: GFP (1∶2500); Yeti (1∶250); H2A.V (1∶2500) and α-Tubulin (1∶2500). Horseradish peroxidase secondary antibodies were all used at 1∶5000. Immunodetection was performed using the ECL kit from BIORAD. For GST pulldown, 2 mg of either the GST–HP1 fusion protein or GST alone was incubated with protein extracts from 50 brains in incubation buffer (20 mM HEPES KOH, 20 mM NaF and 0.8% NP40), 1 hour at 4°C with agitation. After incubation, GST proteins were harvested and washed several times with incubation buffer and resuspended in 1× Laemmli Buffer. Co-immunoprecipitation assays were repeated at least twice.

We thank Thomas Kusch (Department of Molecular Biology and Biochemistry, Rutgers, The State University of New Jersey, Piscataway, USA), Patrizia Lavia, (Istituto di Biologia e Patologia Molecolare, Consiglio Nazionale delle Ricerche, Italy), Barbara Wakimoto (Department of Biology, University of Washington, Seattle, USA) and Marie-Laure Rufh for critical reading of the manuscript and suggestions, and Peter Becker and Kenneth Börner (Ludwig-Maximilians-University Munich Adolf Butenandt Institute, Munchen, Germany), and Steve Henikoff (Fred Hutchinson Cancer Research Center, Seattle, USA) for helpful comments. We are also grateful to Gianluca Cestra, Laura Ciapponi, Davide Corona, Sarah Elgin, Robert Glaser and Marie-Laure Ruhf for gifts of antibodies. We thank Enzo Iacueo for the technical support with the fly room. We also thank Giulia Guarguaglini and Patrizia Lavia, at Nikon Reference Centre, CNR Institute of Molecular Biology and Pathology (Istituto di Biologia e Patologia Molecolare, Consiglio Nazionale delle Ricerche, Italy).