The Drosophila DAXX-Like Protein (DLP) Cooperates with ASF1 for H3.3 Deposition and Heterochromatin Formation

ABSTRACT

Histone variants are nonallelic isoforms of canonical histones, and they are deposited, in contrast to canonical histones, in a replication-independent (RI) manner. RI deposition of H3.3, a histone variant from the H3.3 family, is mediated in mammals by distinct pathways involving either the histone regulator A (HIRA) complex or the death-associated protein (DAXX)/α-thalassemia X-linked mental retardation protein (ATRX) complex. Here, we investigated the function of the Drosophila DAXX-like protein (DLP) by using both fly genetic approaches and protein biochemistry. DLP specifically interacts with H3.3 and shows a prominent localization on the base of the X chromosome, where it appears to act in concert with XNP, the Drosophila homolog of ATRX, in heterochromatin assembly and maintenance. The functional association between DLP and XNP is further supported by a series of experiments that illustrate genetic interactions and the DLP-XNP-dependent localization of specific chromosomal proteins. In addition, DLP both participates in the RI deposition of H3.3 and associates with anti-silencing factor 1 (ASF1). We suggest, in agreement with a recently proposed model, that DLP and ASF1 are part of a predeposition complex, which is recruited by XNP and is necessary to prevent DNA exposure in the nucleus.

INTRODUCTION

In the nucleus, DNA is packaged under the form of chromatin. The basic repeating unit of chromatin is the nucleosome, and the nucleosome core particle consists of approximately 146 bp of DNA wrapped around a histone octamer composed of two copies of each of the histones H2A, H2B, H3, and H4. (1, 2). Gene-rich domains are packaged into euchromatin, which is decondensed in interphase nuclei and is enriched for histone modifications characteristic of transcriptionally active regions. Moreover, euchromatin often exhibits high DNA accessibility. In contrast, gene-poor domains and repetitive sequences are packaged into condensed heterochromatin that carries histone modifications associated with transcriptional repression.

In addition to the canonical core histones, cells express nonallelic-isoform histone variants (3). Histone variants have emerged as essential contributors to the regulation of chromatin structure, and they are involved in multiple processes, including chromatin stability, DNA repair, and transcriptional regulation. Canonical histones are almost exclusively expressed during the S phase of the cell cycle and are incorporated into chromatin in a DNA replication-dependent fashion, whereas replication-independent (RI) histone variants are expressed throughout the cell cycle.

One of the best-studied histone variants is H3.3, which can replace the major species H3 (4). H3.3 is expressed and incorporated at all phases of the cell cycle (5). The available data suggest that H3.3 is a marker of active chromatin and associated with the epigenetic maintenance of chromatin status (6, 7). This hypothesis is further supported by the finding that H3.3 is enriched with posttranslational modifications specific to active genes (8, 9). The chromatin structure is routinely disrupted in active regions of the genome, and these regions are repackaged using H3.3. Consequently, this process results in the enrichment of H3.3 in active regions where nucleosomes are unstable or disrupted.

Biochemical studies have identified factors associated with H3.3 that mediate RI assembly of nucleosomes (10–12). These factors include histone chaperones and chromatin remodelers and potentially target H3.3 to active regions of the genome. Two major histone chaperone complexes have been characterized as being responsible for H3.3 incorporation: the histone regulator A (HIRA)/Yemanuclein (Yem)–Ubinuclein-1 complex (10, 13–15), which incorporates H3.3 into genic, euchromatic regions, and the mammal death-associated protein (DAXX)/α-thalassemia X-linked mental retardation protein (ATRX) (10, 11) complex, which incorporates H3.3 into pericentromeric and telomeric heterochromatin regions. So far, no obvious counterpart for mammal DAXX has been identified in Drosophila.

Surprisingly, mutants of some of these factors have only limited phenotypes. For example, Drosophila HIRA is essential only for de novo assembly of H3.3-containing nucleosomes in the male pronucleus during fertilization. (16, 17). XNP, the Drosophila homolog of ATRX, is not essential, and XNP-deficient flies are viable and fertile (18–20). This suggests that H3.3 assembly factors may be redundant, or alternatively, there may be additional factors involved in H3.3 deposition. In somatic cells, mutant animals can compensate for the loss of H3.3 with the major H3 histone. However, the loss of H3.3 leads to reduced viability and has dramatic impacts on the fertility of both males and females (21, 22). XNP and HIRA bind DNA at genomic gaps when nucleosomes are disassembled, marking the sites for RI assembly. It has been hypothesized that a separate predeposition complex containing anti-silencing factor 1 (ASF1) and H3.3/H4 heterodimers is recruited by XNP and HIRA and that the heterodimers are subsequently incorporated into new nucleosomes (23).

In the present study, we dissected the function of Drosophila DAXX-like protein (DLP) by both fly genetic and protein biochemistry approaches. We show that DLP (i) specifically associates with the histone variant H3.3, (ii) functionally interacts with XNP, and (iii) is involved in in vivo heterochromatin formation through modification of position effect variegation (PEV). Taken as a whole, these data suggest that DLP and ASF1 form the central part of a predeposition complex involved in the recruitment of H3.3 to nucleosome-depleted chromatin gaps by XNP.

RESULTS

DLP is widely expressed during development.

dlp encodes a protein of 184 kDa (1,659 amino acids), which contains a C-terminal DAXX homology region (DHR) (Fig. 1A). This DLP domain is 347 amino acids long (residues 1125 to 1472) and exhibits 28% sequence identity and 46% similarity to the high-affinity H3.3-interacting domain of DAXX, suggesting that it might be the fly homolog of mammalian DAXX. If this is the case, DLP should exhibit functions similar to those of DAXX.

FIG 1.

Schematic representation of dlp and description of the mutants used in the present study. (A) Intron/exon structure of Drosophila dlp. Introns are represented by solid lines, while exons are represented by rectangles. The coding sequence is shaded, and the black portion is highly homologous to the H3.3-interacting domain of DAXX (DHR). The triangle below NP4778 denotes the insertion site of P(GawB)DaxxNP⁴⁷⁷⁸, carrying the coding sequences of yeast GAL4. The dlp⁴⁵ allele was generated by the imprecise excision of Mi(ET1)Daxx^MB03646 (triangle below MB03646). 1 and 2 denote the domains of DLP used to generate the 1D6 and 1C11 antibodies, respectively. (B) Immunoblot analysis of DLP in wild-type (w¹¹¹⁸) (lane 4) and mutant (lane 1, dlp⁴⁵/+; lane 2, dlp⁴⁵/dlp⁴⁵; lane 3, dlp^G/dlp^G) embryo protein extracts using the 1D6 antibody. β-Tub is used as a loading control. (C and D) Testis (C) and ovarian follicle at stage 10 (D) from wild-type flies were stained to visualize DNA (gray in panel C and blue in panel D) and DLP (red in panel C and green in panel D). The arrow in panel C denotes the hub cells (HC) located at the apical tip of the testis. White frames in panel D show nurse cells (NC). GV, germinal vesicle. No signal was detected in dlp^Gal4 ovaries. 4′,6-diamidino-2-phenylindole (DAPI)-stained DNA is in blue.

To analyze the function of Drosophila DLP in vivo, we generated mutant alleles of dlp by the imprecise excision of the Minos element present in the Mi(ET1)Daxx^MB03646 line (24, 25). The Minos element was inserted in the middle of dlp (Fig. 1A). We isolated dlp⁴⁵, an allele carrying a 5-bp insertion (TCGTA), as evidenced by PCR analyses, causing a frameshift in the open reading frame (ORF) and introducing a new stop codon. dlp⁴⁵ is predicted to encode a truncated protein of 869 amino acids, lacking the C-terminal DHR. However, the homozygous dlp⁴⁵ allele is both viable and fertile.

Next, we analyzed the P(GawB)Daxx^NP4778 enhancer trap line, which carries a Gal4-containing transposon inserted into the 5′ untranslated sequences of dlp (Fig. 1A). Hence, GAL4 expression is thought to recapitulate some traits of dlp expression. Surprisingly, the P(GawB)Daxx^NP4778 line (here termed the dlp^G [dlp^Gal4] line) was found not to express DLP, and dlp^G can consequently be considered a null dlp allele (see below).

Analysis of DLP by immunoblotting using the 1D6 and the 1C11 antibodies (Fig. 1B and data not shown) revealed that wild-type embryos expressed a full-length protein with a molecular mass of ∼184 kDa (Fig. 1A, lane 4), while dlp⁴⁵ encodes a truncated version of DLP with a molecular mass of ∼100 kDa (Fig. 1A, lanes 1 and 2). In contrast, DLP was absent in protein extracts made from homozygous dlp^G embryos (Fig. 1A, lane 3).

The mRNA of dlp is widely expressed during embryogenesis, with the peak of expression in 0- to 2-h-old embryos. It is also present during larval stages and in adults. In adult tissues, the messenger is particularly abundant in brain, testis, ovary, and salivary glands (http://flybase.org/). Immunostaining of wild-type testis (Fig. 1C) with the 1D6 antibody revealed the presence of DLP in the germ line, with prominent expression in primary spermatocytes and meiotic spermatocytes. In wild-type ovaries (Fig. 1D), DLP expression was observed in nurse cells and also in the germinal vesicle of the ovarian follicle at stage 10. In contrast, no DLP expression was observed for the ovarian follicles of homozygous dlp^G females (Fig. 1D).

DLP specifically interacts with H3.3.

DLP and DAXX share a C-terminal DHR known to mediate interactions with H3.3 in mammals. We then investigated whether Drosophila DLP also physically interacts with histone H3.3.

We used the tandem-immunoaffinity purification method (11, 26) to isolate the proteins associated with histones H3 and H3.3 in a total soluble extract made from 0- to 3-h-old embryos from transgenic lines, in which histones were expressed as fusion proteins with C-terminal Flag and hemagglutinin (HA) epitope tags. Proteins associated with epitope-tagged H3 (e-H3) and e-H3.3 were purified by sequential immunoprecipitation with anti-Flag antibody, followed by anti-HA antibody. The proteins were separated by SDS-containing 4%-to-12% gradient polyacrylamide gels and silver stained (Fig. 2A). Numerous proteins were found to be associated with e-H3 and e-H3.3. Mass spectrometry and immunoblot analyses led to the identification of several common components of the e-H3 and e-H3.3 complexes, including the core histones H2A, H2B, H3, and H4 and the well-characterized histone chaperones ASF1 (27), NAP1 (nucleosome assembly protein 1) (28), Drosophila NASP (nuclear autoantigenic sperm protein) homolog (29), and Drosophila CAF-1 (chromatin assembly factor 1) (p105/p180 subunits) (30) (Fig. 2A, left).

FIG 2.

Immunopurification of e-H3 and e-H3.3 protein complexes from soluble extracts of Drosophila embryos. (A) Silver staining of proteins associated with e-H3 and e-H3.3. The protein complexes containing e-H3 and e-H3.3 were purified by double immunoaffinity from soluble extracts of 0- to 3-h-old transgenic embryos expressing either H3 or H3.3. H3 and H3.3 were expressed as fusion proteins with C-terminal Flag and HA epitope tags. Polypeptides identified by mass spectrometry analysis and the positions of molecular mass markers (in kilodaltons) are indicated. An extract made from w¹¹¹⁸ embryos was similarly processed and used to assess the specificity of the antibodies (mock lane). The e-H3 and e-H3.3 complexes were analyzed by immunoblotting with the indicated antibodies. It should be noted that each polypeptide identified by immunoblotting was present in an equivalent amount in the crude extracts (crude soluble extracts) containing e-H3 and e-H3.3. (B) DLP and XNP do not physically interact. A protein extract was prepared from baculovirus-infected Sf9 insect cells coexpressing DLP and XNP. DLP was expressed as a fusion protein carrying a C-terminal Flag epitope tag (DLP-F). XNP was expressed as a fusion protein carrying an N-terminal HA epitope tag (HA-XNP). Half of the extract was immunoprecipitated with the anti-Flag antibody, the other half was immunoprecipitated with the anti-HA antibody, and selected proteins were eluted with the Flag peptide (elu-Flag) or the HA peptide (elu-HA), respectively. After elution, proteins were fractionated on SDS-PAGE gels and revealed by Coomassie staining (lanes 1 and 2). The bracket with the asterisk denotes degradation products of DLP. Eluted proteins were also analyzed by immunoblotting (lanes 4 and 5) with the anti-DLP antibody 1D6 (α-DLP) and the anti-XNP antibody 1D9 (α-XNP). The input fraction is shown (lane 3). (C) Specific interactions of DLP with H3.3. Protein extracts made from Sf9 cells infected with baculovirus and coexpressing DLP and either H3.3-H4 or H3-H4 were immunoprecipitated with the anti-Flag antibody. DLP was expressed as a fusion protein carrying a C-terminal Flag epitope tag. H3.3 and H3 were expressed as fusion proteins carrying an N-terminal HA epitope tag (HA-H3.3 and HA-H3). After elution, proteins were fractionated on SDS-PAGE gels and revealed by Coomassie staining (lanes 1 and 2). DLP proves to be very unstable, and degradation products are depicted by brackets with asterisks. Eluted proteins (lanes 4 and 6) were also analyzed by immunoblotting with the anti-DLP antibody 1D6 (DLP-F) and the anti-HA antibody (HA-H3.3). Input fractions are shown (lanes 3 and 5). (D) Truncated DLP⁴⁵ no longer interacts with H3.3. Extracts made from Sf9 cells infected with baculovirus and coexpressing H3.3-H4 and either DLP or DLP⁴⁵ were immunoprecipitated by the anti-Flag antibody. Selected proteins were revealed as described above for panel C. Brackets with asterisks depict degradation products.

Importantly, we identified Yem (13), HIRA (31), and DLP as unique partners of H3.3 (Fig. 2A, left). Immunoblotting confirmed that DLP and Hira are found only in the e-H3.3 complex (Fig. 2A, right). However, DLP is poorly visible (Fig. 2A, left), reflecting the small amount of DLP in the protein extract and suggesting that DLP is a minor component of H3.3-containing complexes. Alternatively, this may reflect a low stability of the protein.

Surprisingly, XNP (18–20), the Drosophila counterpart of ATRX, was not found among the proteins associated with e-H3.3, whereas DAXX and ATRX physically interact and belong to the e-H3.3 preassembly complex isolated from HeLa cells (11). The absence of XNP in immunopurified H3.3 and H3 complexes was confirmed by immunoblotting (Fig. 2A, right). This suggests a distinct relationship between ATRX and DAXX in Drosophila and in mammals. This hypothesis is further supported by the observation that DLP and XNP were not found to associate in extracts made from baculovirus-infected Sf9 insect cells coexpressing DLP and XNP (Fig. 2B). XNP lacks the C-terminal plant homeodomain (PHD) fingers that are present in ATRX and that mediate ATRX binding to chromatin. In Drosophila, interaction with chromatin is achieved by a separate protein, called dADD1 (32). Mass spectrometry analysis revealed that dADD1, similarly to XNP, was not found among the proteins present in the immunopurified H3.3 complexes.

In addition, our data revealed that the ASF1 chaperone is more abundant in the protein complex associated with H3.3 than in the one associated with conventional H3 (Fig. 2A). ASF1 was identified as a subunit of the RCAF (replication-coupling assembly factor) complex that facilitates the assembly of nucleosomes onto newly replicated DNA in vitro (27). RCAF, in addition to ASF1, contains H3 and H4, but the molecular nature of H3 is not documented. Bearing in mind the higher abundance of ASF1 in the H3.3 complex, we hypothesize that Drosophila ASF1 may have a greater affinity for H3.3 than for H3.

As DLP was found only in the e-H3.3 complexes but not in the e-H3 complexes, we next investigated the interactions between DLP and H3.3. We used Sf9 insect cells expressing C-terminal Flag epitope-tagged DLP and either HA-H3.3/H4 or HA-H3/H4. The protein extracts were immunoprecipitated with the Flag antibody, and selected proteins were eluted with the Flag peptide, prior to electrophoresis on SDS-containing 9% polyacrylamide gels and Coomassie staining. We observed that only HA-H3.3/H4 associates with DLP-Flag (Fig. 2C, left). This was further confirmed by Western blotting (Fig. 2C, right). We conclude that under physiological conditions, only tagged H3.3 consistently associates with DLP-Flag.

We carried out the same type of experiments but with dlp⁴⁵, which encodes a truncated protein lacking the C-terminal DHR. Both electrophoretic and Western blotting analyses revealed no interaction of the truncated protein with HA-H3.3, as expected, demonstrating the requirement for the C terminus of the protein for H3.3 binding and strongly suggesting the functional conservation of the DHR from mammals to flies (Fig. 2D).

In summary, these data demonstrate that DLP is a component of e-H3.3-containing complexes but not of e-H3-containing complexes. DLP specifically interacts with H3.3, and the binding of H3.3 is mediated by the C-terminal DHR domain of DLP, a region conserved in all known DAXX homologs.

DLP is associated with regions located at the base of both the X and the fourth chromosomes.

In order to gain more insights into DLP function, we examined DLP localization with high resolution, using the giant polytene chromosomes of larval salivary glands. Immunostaining showed two bright foci of DLP, close to the heterochromatic chromocenter (Fig. 3A, panels 1 to 3). One of them is the base of the fourth chromosome (33), which is mostly heterochromatic. Staining with an anti-male-specific lethal 1 (anti-MSL-1) antibody (Fig. 3A, panel 4) revealed that the other focus corresponds to the base of the X chromosome. MSL-1, together with MSL-2, MSL-3, and Maleless (MLE), is an essential subunit of the male-specific lethal (MSL) complex (34), which localizes at hundreds of sites along the euchromatic part of the male X chromosome and does not stain the heterochromatin at the base of the X chromosome. This was further supported by double staining for DLP and heterochromatic protein 1 (HP1) (35, 36), which showed that DLP and HP1 overlap on the fourth chromosome, while DLP is located next to HP1 on the X chromosome (Fig. 3A, panel 5).

FIG 3.

DLP marks pericentric heterochromatin associated with the X chromosome and with the fourth chromosome. Polytene chromosome squashes show immunostaining for DLP with the control w¹¹¹⁸ strain (A and D), the mutant dlp^G strain (B), the ln(1)w^m4h strain (C and E), and the mutant dlp⁴⁵ strain (F). DLP staining is in green, and DAPI-stained DNA is in blue (A, C, and F, panels 3) or in gray (B and D to F, panels 1). (A) The arrow and arrowhead indicate DLP staining associated with pericentric heterochromatin of the X chromosome and DLP staining associated with pericentric heterochromatin of the fourth chromosome, respectively. In panel 2, the chromocenter is highlighted by a white frame and is enlarged in panel 3. Panel 4 is an enlargement of the chromocenter of polytene chromosomes costained for DLP and MSL-1 (red). The X chromosome is visualized by MSL-1 staining. Panel 5 is an enlargement of the chromocenter costained for DLP and HP1. The chromocenter and the fourth chromosome are HP1 rich (red). (B) No DLP signal is observed in spreads from dlp^G mutants. The chromocenter is highlighted by a white frame and is enlarged in panel 2. (C) Immunostaining for DLP and HP1 with polytene chromosomes from ln(1)w^m4h larvae. (D) Enlargement of the chromocenter of polytene chromosomes from w¹¹¹⁸ larvae costained for DLP (green) and XNP (red). The arrow and arrowhead (panel 2) indicate DLP staining associated with pericentric heterochromatin of the X chromosome and DLP staining associated with pericentric heterochromatin of the fourth chromosome, respectively. (E) Distal tip of the X chromosome from ln(1)w^m4h larvae costained for DLP (green) and XNP (red). DLP was detected with the 1D6 antibody, and XNP was detected with polyclonal antiserum described previously (19). (F) Truncated DLP⁴⁵ is not detected in spreads from dlp⁴⁵ mutants. Panel 1 shows DAPI-stained DNA (gray). Panels 2 and 3 are enlargements of the chromocenter seen in panel 1. Panel 2 is stained for DLP (green). Panel 3 shows HP1 staining (red) and DAPI-stained DNA (blue).

Notably, no signal is observed on spreads of dlp^G polytene chromosomes, which do not express DLP (Fig. 3B, panels 1 and 2). This demonstrates that the staining of the base of both the X and the fourth chromosomes is specific and reflects the presence of endogenous DLP. No consistent staining was observed throughout the euchromatic arms. We attribute this to either the complete absence of DLP or the very small amount of DLP associated with the euchromatic arms.

We further visualized the DLP focus by immunostaining of polytene chromosomes from ln(1)w^m4h larvae (Fig. 3C). The ln(1)w^m4h rearrangement inverts most of the X chromosome and moves the euchromatin-heterochromatin junction to the end of the chromosome. In agreement with the above-described data, doubly stained polytene chromosomes revealed that the DLP focus lies near the tip of the X chromosome, next to HP1 (Fig. 3C).

DLP, like XNP, is associated with pericentric heterochromatin of the X chromosome (19), suggesting that DLP and XNP may be located in the vicinity of each other and may functionally interact. This observation prompted us to compare the localization of DLP with that of XNP. We costained the chromosomes for DLP and XNP (Fig. 3D and E) and observed overlaps at the base of the X chromosome of w¹¹¹⁸ larvae and also at the distal tip of the X chromosome of larvae carrying the ln(1)w^m4h inversion. Schneiderman and colleagues importantly demonstrated that XNP also coincides with H3.3 at this heterochromatic locus, which was shown to be a major site of H3.3 deposition, implying that nucleosomes are continuously being disassembled and reassembled at this site (19). These observations provide a link between the association of DLP with specific heterochromatin loci and DLP as an H3.3 chaperone.

Immunoblot analysis of protein extracts made from homozygous dlp⁴⁵ embryos with the 1D6 antibody revealed a truncated protein with a molecular mass of ∼100 kDa (Fig. 1B). Although DLP⁴⁵ contains the epitope recognized by 1D6, we were unable to detect it on polytene spreads of dlp⁴⁵ larvae (Fig. 3D). As we have evidence that DLP⁴⁵ can nevertheless be incorporated into chromatin (see below), we speculate that the truncated protein might exhibit some conformational changes, which may mask the epitope recognized by the antibody. Alternatively, the incorporation of truncated DLP⁴⁵ into chromatin is reduced, and the protein concentration is consequently too low to be detected.

These data established that DLP is associated with heterochromatin of both the X and the fourth chromosomes. Prominent DLP accumulation at the base of the X chromosome is of particular interest, as DLP and XNP colocalize at this heterochromatic locus, which is a major site of H3.3 deposition and where nucleosomes are continuously being disassembled and reassembled (19).

DLP is involved in heterochromatin formation.

Immunostaining revealed prominent DLP accumulation in pericentric heterochromatin of the X chromosome. Hence, we investigated whether DLP is involved in heterochromatin formation, using the ln(1)w^m4h inversion of the X chromosome (Fig. 4A). The rearrangement places the white gene (w) into pericentric heterochromatin, causing variegated expression in the eye (Fig. 4A, panel 1). We observed a mild dominant suppression of variegation for both heterozygous dlp⁴⁵/+ and dlp^Gw/+ animals, while homozygous dlp⁴⁵ and dlp^Gw animals deficient for wild-type dlp function had greatly derepressed w expression (Fig. 4A, panels 2 to 5). As previously described (18–20), we also observed that animals with reduced xnp function (xnp⁵⁶/+) or lacking xnp function (xnp⁵⁶/xnp⁵⁶) display derepressed w expression (Fig. 4A, panels 6 and 7). Therefore, both dlp and xnp are involved in the formation of heterochromatin of the X chromosome. As XNP and DLP are present at the base of the X chromosome, we next analyzed the consequences of simultaneous losses of function for dlp and xnp (Fig. 4A, panels 8 and 9). We observed that dlp^Gw/+; xnp⁵⁶/+ animals exhibit a greater derepression of w than do animals heterozygous for xnp (xnp⁵⁶/+) or dlp (dlp^Gw/+), suggesting that DLP and XNP may act in concert during heterochromatin formation. This hypothesis is further reinforced by the observation that doubly heterozygous dlp⁴⁵/+; xnp⁵⁶/+ animals also display a greater derepression of w than do singly heterozygous dlp⁴⁵/+ or xnp⁵⁶/+ animals.

FIG 4.

Mutations of dlp suppress heterochromatic silencing. +/+ depicts w expression seen in adult eyes of ln(1)w^m4h males (A, panel 1) and in adult eyes of wild-type males carrying a single copy of the variegating insert (B, panels 1, 3, 5, and 7). (A) Effect of dlp mutations on variegation of heterochromatin-silenced genes of the X chromosome. Heterozygotes (dlp⁴⁵/+ [panel 2], dlp^Gw/+ [panel 4], and xnp⁵⁶/+ [panel 6]) display moderate derepression, while double heterozygotes (dlp⁴⁵/+; xnp⁵⁶/+ [panel 8] and dlp^Gw/+; xnp⁵⁶/+ [panel 9]) have greatly derepressed silencing. Derepressed w expression is also shown for dlp⁴⁵ (panel 3), dlp^Gw (panel 5), and xnp⁵⁶ (panel 7) animals. (B) dlp⁴⁵ mutation also suppresses silencing of the transgenic insertions 118E-15 (telomeric region of chromosome 4) (panel 2), 118E-10 (pericentric region of chromosome 4) (panel 4), and 39C-3 (pericentric region of chromosome 2) (panel 8). In contrast, it has no effect on the silencing of the transgenic insertions 39C-12 (medial region of chromosome 4) (panel 6). Alleles and crosses are described in Materials and Methods. All flies are male and are heterozygous for the particular P-element insert.

Since DLP is also present at the base of the fourth chromosome, we further studied the implication of DLP in heterochromatin formation of the fourth chromosome. As dlp⁴⁵ has greater effects on the suppression of variegation than does dlp^Gw, we examined the consequences of introducing dlp⁴⁵ into lines 118E-10, 39C-12, and 118E-15 (37). These lines have variegating insertions near the base, in a medial position, and near the telomere of the fourth chromosome, respectively. We observed suppression of variegation for lines 118E-10 and 118E-15 (Fig. 4B, panels 1 to 4). In contrast, variegated expression in line 39C-12 is not modified in heterozygous dlp⁴⁵/+ animals (Fig. 4B, panels 5 and 6).

Finally, DLP appears to be a general modifier of heterochromatin silencing because dlp⁴⁵/+ animals exhibit mild dominant derepression of variegating hsp70–mini-w inserted at the base of the second chromosome (line 39C-3) (37) (Fig. 4B, panels 7 and 8). A similar effect on the second chromosome has been described for xnp, where loss-of-function alleles are dominant derepressors of brown^Dominant (bw^D)-mediated silencing (19). Hence, both dlp and xnp are also involved in the formation of heterochromatin of the second chromosome.

These results demonstrate that DLP, like XNP, is a modifier of heterochromatin silencing and suggest that DLP and XNP functionally interact in heterochromatin formation and maintenance.

Depletion of H3.3 alters the DLP nuclear distribution pattern.

As described above, immunostaining of polytene chromosomes from wild-type larvae revealed two bright foci of endogenous DLP associated with the chromocenter, but no consistent staining was observed along the chromosome arms. However, new binding sites for DLP appear along chromosome arms and on nucleolar chromatin in H3.3 knockdown cells (Fig. 5A). Hence, the suppression of H3.3 expression has a profound effect on the DLP distribution pattern, which further confirms the close relationship between these two proteins. The nucleolar staining in H3.3-depleted cells is reminiscent of the one described previously for XNP (23), suggesting that DLP and XNP may have some common target sites in the genome. Indeed, the major site for XNP binding is at the base of the X chromosome, where DLP is found. Intense XNP and DLP staining is also localized to the repeated ribosomal DNA (rDNA) in the nucleolus of H3.3-deficient cells, while no staining is observed in the nucleolus of wild-type cells.

FIG 5.

Reduced viability of dlp-overexpressing animals. (A) Polytene chromosome squashes of H3.3 knockdown cells (H3.3 KD). DLP staining is in green and is observed across the chromosome arms. Intense staining is also observed on chromatin in the nucleolus (dotted outline). Nucleolar DLP staining is not seen in wild-type cells. DAPI-stained DNA is in blue (left) or gray (right). (B) Polytene chromosome squashes showing immunostaining for overexpressed dlp for female or male larvae. uas-dlp expression is induced by the salivary gland-specific sgs3^Gal4 driver. DLP staining, in green, is shown for a line characterized by weak dlp expression (WEDL). DAPI-stained DNA is in blue. White arrows indicate pericentric heterochromatin of the X chromosome, while the red arrow shows the X chromosome. Color splits show the correspondence between DLP and the banding pattern along the arms of polytene chromosomes. (C) Female embryos are more sensitive than male embryos to harmful effects of overexpressed dlp. dlp^G; Sxl-GFP females were crossed with males of different uas-dlp lines. Results are presented for two lines characterized by weak DLP expression (WEDL1 and WEDL2) and two lines characterized by high DLP expression levels (HEDL1 and HEDL2). One hundred male embryos and 100 female embryos were selected by means of the Sxl-GFP reporter specifically expressed in females and were allowed to develop at 25°C. Numbers of male and female adults are presented and are based on data from 4 independent experiments. Error bars represent standard errors of the means. Analysis of uas-dlp⁴⁵ lines revealed that overexpressed DLP⁴⁵ behaves similarly to overexpressed DLP and provokes severe lethality of developing embryos, with a greater sensitivity of female embryos. (D) DLP staining, in green, is shown for a HEDL. Msl-1 staining is in red and shows the X chromosome. DAPI-stained DNA is in blue. The white arrow indicates pericentric heterochromatin of the X chromosome, and the red arrow shows the X chromosome. (E) Number of adult males when dlp is overexpressed in the presence of a reduced function of either msl-2 (msl2²²⁷) or mle (mle¹). dlp^G/Cy, msl-2²²⁷dlp^G/Cy, or mle¹dlp^G/Cy females were crossed with either w¹¹¹⁸ males or males of a WEDL. For each cross, 200 embryos of the progeny were allowed to develop at 25°C. The number of Cy⁺ adult males is presented and is based on data from 4 independent experiments. Error bars represent standard errors of the means. (F) Polytene chromosome squashes showing immunostaining for overexpressed uas-dlp⁴⁵ driven by sgs3^G. HP1 staining (in red) shows the chromocenter.

These results demonstrate that DLP behaves similarly to XNP in H3.3-deficient cells, suggesting a direct functional link between these proteins.

The reduced viability of DLP-overexpressing flies reflects the association of DLP with euchromatin.

We next investigated whether overexpressed DLP may associate with chromosome arms. We first used a uas-dlp line characterized by low DLP expression levels (see below). Immunostaining of polytene chromosomes revealed that overexpressed uas-dlp, driven by the salivary gland-specific driver sgs3^G, localized to multiple euchromatic sites on different chromosomes when spreads were made from female larvae (Fig. 5B). Surprisingly, no DLP staining was observed on the X chromosome when spreads were made from male larvae. This observation prompted us to investigate the consequences of overexpressed DLP for the development of male and females embryos.

To isolate separate populations of female and male embryos, we took advantage of the Sex-lethal (Sxl) promoter that is specifically active in females (38). Thus, dlp^G; Sxl-GFP females were crossed with uas-dlp males, and we selected green fluorescent protein (GFP)-positive female embryos and GFP-negative male embryos. When we let embryos develop to adults, we observed a reduced number of Sxl-GFP/+; dlp^G/uas-dlp adults in comparison to the number of Sxl-GFP/+; dlp^G/+ adults resulting from a control cross between Sxl-GFP; dlp^G females and w¹¹¹⁸ males (Fig. 5C), suggesting that overexpressed DLP reduces the viability of developing embryos. In addition, female embryos appear more sensitive to the effects of overexpressed DLP than do male embryos (Fig. 5C). For a given line, we also observed that increased uas-dlp expression at 30°C leads to a reduced viability of developing embryos in comparison to the viability observed at 25°C. Conversely, reduced expression at 18°C is associated with better viability. Based on the effects on viability at 25°C, we have arbitrarily classified the uas-dlp lines from weakly expressing dlp lines (WEDL), characterized by low DLP expression levels associated with little affected viability of developing male embryos, to highly expressing dlp lines (HEDL), characterized by the absence of adult females and a severely reduced number of adult males. Finally, we performed immunostaining of polytene chromosomes from sgs3^G/+; uas-dlp/+ male larvae for a HEDL. Under these conditions, we also revealed DLP staining on the X chromosome (Fig. 5D). Hence, the low viability of DLP-overexpressing males is likely associated with misregulated X-linked genes.

Multiple evidences underscore that MSL-2 is probably the most central regulator of dosage compensation. MSL-2 is expressed only in males (39) and induces the assembly of the MSL complex on chromatin and a 2-fold upregulation of the single X chromosome to reach the level of the two X chromosomes in females (33). We then investigated the consequences of overexpressed dlp for male viability when msl-2 function is reduced (Fig. 5E). msl-2²²⁷dlp^G/Cy females were crossed with uas-dlp males from a WEDL, and 200 embryos from the progeny were allowed to develop at 25°C. Reduced msl-2 function has a moderate effect on the viability of msl-2²²⁷dlp^G/+ males compared to the viability of dlp^G/+ males. In contrast, the viability of msl-2²²⁷dlp^G/+; uas-dlp/+ males was severely impaired with regard to the viability of dlp^G/+; uas-dlp/+ and msl-2²²⁷dlp^G/+ males (Fig. 5E). Hence, harmful effects on viability are aggravated when msl-2 function is reduced, suggesting that DLP and MSL-2 may compete for binding to common sites on the X chromosome. This hypothesis is further supported by data from experiments where uas-dlp was overexpressed in the presence of reduced mle function (Fig. 5E). The MLE subunit of the MSL complex is an ATP-dependent RNA/DNA helicase, which is peripherally associated with the MSL complex and is involved in the spread of the complex on the X chromosome from high-affinity sites (40). We observed that the effects of overexpressed dlp on male viability are not changed when mle function is reduced (Fig. 5E).

Overexpressed uas-dlp⁴⁵ encodes a truncated DLP protein lacking the DHR and is consequently thought to induce a loss-of-function phenotype. Overexpressed DLP⁴⁵, like overexpressed DLP, provokes severe lethality of developing embryos, with a greater sensitivity of female embryos (Fig. 5C). Accordingly, intense staining on the chromosome arms of polytene chromosomes is observed when uas-dlp⁴⁵ expression is conducted by the salivary gland-specific driver sgs3^G (Fig. 5F). We consequently hypothesize that both DLP and DLP⁴⁵, when highly expressed, behave as a loss of dlp function.

Together, these results suggest that the DLP concentration should be strictly regulated, as overexpressed DLP has toxic effects during Drosophila development.

DLP functionally interacts with XNP.

Previous studies have shown that loss-of-function alleles of xnp affect fly viability (18–20). We similarly observed that the development of both xnp⁵⁶/Df(3R)Exel6202 and homozygous xnp⁵⁶ embryos is affected, leading to reduced viability (Table 1).

TABLE 1.

Genetic interactions between dlp and xnp^a

Genotype	Mean % of embryos surviving to adults ± SEM
w1118	95.5 ± 3
dlpG/+	94.0 ± 4.5
dlp45/+	96.5 ± 3
xnp56/+	93.5 ± 3.5
dlpG/+; xnp56/+	57.5 ± 6.5
dlp45/+; xnp56/+	55.0 ± 6.2
dlpG	92.0 ± 6.3
dlp45	95.0 ± 3.3
xnp56/Df(3R)Exel6202	40.0 ± 5
dlpG; xnp56/Df(3R)Exel6202	3.5 ± 1.5
dlp45; xnp56/Df(3R)Exel6202	2.5 ± 1.9
uas-dlp/+	98.0 ± 1.3
dlpG/+; uas-dlp/+	31.0 ± 3.2
dlpG/+; uas-dlp/xnp56	3.0 ± 1.6

^aFor each genotype, 100 embryos were selected and allowed to develop at 25°C. Adults were scored. Results of overexpression experiments are shown for a representative HEDL. Results are based on data from 4 independent experiments.

As evidences in both mammals and Drosophila suggest that ATRX/XNP and DAXX/DLP functions may be linked, we then investigated whether dlp may genetically interact with xnp during development. The development of both heterozygous dlp^G/+ and homozygous dlp^G embryos is not affected compared to w¹¹¹⁸ wild-type embryos, suggesting that DLP has no crucial roles during development (Table 1). Alternatively, a loss of dlp function may be compensated for by a redundant factor(s). In spite of the fact that the development of dlp^G/+ and xnp⁵⁶/+ embryos is not affected, we observed a significantly reduced viability of developing dlp^G/+; xnp⁵⁶/+ embryos (Table 1). Moreover, dlp^G; xnp⁵⁶/Df(3R)Exel6202 animals are poorly viable, suggesting that DLP and XNP may act together on common target sites. Similarly, the development of dlp⁴⁵; xnp⁵⁶/Df(3R)Exel6202 embryos was severely impaired (Table 1).

We next investigated the consequences of overexpressed dlp for viability when induced in the presence of reduced xnp function (Table 1). We observed that the harmful effects of overexpressed dlp in dlp^G/+; uas-dlp animals are aggravated when induced in the loss of xnp function. dlp^G/+; uas-dlp/xnp⁵⁶ adults are extremely rare (Table 1), reinforcing the hypothesis that xnp and dlp may act together on common developmentally regulated targets.

In summary, the loss and the gain of dlp function similarly affect the viability of loss-of-xnp-function alleles, supporting the notion that overexpressed dlp behaves as a loss-of-function allele. DLP is likely involved in a multiprotein complex, and accurate stoichiometry of the different subunits required for proper functioning is disrupted when DLP is overexpressed.

The lethality of dlp^G; xnp⁵⁶ animals is greater than the lethality of xnp⁵⁶ animals, suggesting that DLP might also be involved in a pathway distinct from that mediated by XNP. As previously suggested (23), Xnp and HIRA recognize exposed DNA and serve as a binding platform for the efficient recruitment of H3.3 predeposition complexes to chromatin gaps. We consequently investigated genetic interactions between dlp and Hira. Hira is located on the X chromosome, and we analyzed the viability of developing male embryos simultaneously lacking dlp and Hira functions (Table 2). We observed that the development of male embryos lacking either dlp (Table 1) or Hira (Hira^HR1/Y) (Table 2) is not significantly affected, while the development of Hira^HR1/Y; dlp^G male embryos is severely impaired. We concluded that DLP function is connected not only to that of XNP but also to that of HIRA.

TABLE 2.

Genetic interactions between dlp and Hira^a

Cross	Adult genotype	Mean % of embryos surviving to adults ± SEM
HiraHR1/FM7 × w1118	Female HiraHR1/+	20 ± 1.9
	Male HiraHR1/Y	17 ± 2.5
HiraHR1/FM7; dlpG × w1118	Female HiraHR1/+; dlpG/+	19 ± 3.9
	Male HiraHR1/Y; dlpG/+	16 ± 3.6
HiraHR1/FM7; dlpG × dlpG	Female HiraHR1/+; dlpG	17 ± 4.5
	Male HiraHR1/Y; dlpG	4 ± 2.1

^aHira^HR1/FM7 females were crossed with w¹¹¹⁸ males. Hira^HR1/FM7; dlp^G females were crossed with w¹¹¹⁸ or dlp^G males. For each cross, 100 embryos were selected and allowed to develop at 25°C. Hira^HR1 adult males were scored, and results are based on data from 4 independent experiments.

In summary, our data reveal genetic interactions between dlp and xnp consistent with functional interactions of DLP and XNP during Drosophila development.

DLP is involved in RI deposition of H3.3.

A previous study showed that DAXX facilitates H3.3 deposition on DNA in vitro (11), while our present report demonstrates that DLP specifically interacts with H3.3. These observations prompted us to investigate whether DLP is involved in the RI deposition of H3.3 in vivo. Hence, we assayed the incorporation of H3.3^core-red fluorescent protein (RFP) in chromatin in wild-type and mutant genotypes. H3.3^core-RFP labels active genes and can be incorporated only in an RI manner.

The expression of H3.3^core-RFP was induced by heat shock, and polytene chromosome spreads were prepared 2 h later to visualize RI nucleosome assembly. In wild-type cells, H3.3^core-RFP efficiently labels chromosome arms (Fig. 6A). In dlp^G-null mutant cells, we observed reduced incorporation into chromosomes. In addition, nonincorporated H3.3^core-RFP in the nucleolus was not observed (Fig. 6A).

FIG 6.

DLP is involved in RI deposition of H3.3, probably through interaction with ASF1. (A) Expression of H3.3^core-RFP was induced by heat shock, and spreads of polytene chromosomes were prepared after 2 h of recovery. RI deposition of H3.3^core-RFP (red) into polytene chromosomes is reduced in dlp^G and Hira^HR1 cells in comparison to the deposition seen in wild-type control cells. Deposition of H3.3^core-RFP is reduced but not abolished in Hira^HR1; dlp^G cells, suggesting that the pathway mediated by XNP/DLP remains active in the absence of DLP. DAPI-stained DNA is in blue. (B and C) DLP can interact with ASF1. (B) Protein extracts made from baculovirus-infected Sf9 insect cells coexpressing ASF1 and full-length DLP were immunoprecipitated with the anti-Flag antibody. ASF1 was expressed as a fusion protein carrying an N-terminal His epitope tag (His-Asf). DLP was expressed as a fusion protein carrying a C-terminal Flag epitope tag (DLP-F). After elution, proteins were fractionated on SDS-PAGE gels and revealed by Coomassie staining (lane 1). The bracket with the asterisk denotes degradation products of DLP. Eluted proteins were also analyzed by immunoblotting with the anti-DLP antibody 1D6 and the anti-ASF1 antibody (lanes 2 and 3). The input fraction is shown (lane 2). (C) Eluted proteins (elu Flag) obtained as described above for panel B were used as an input for immobilized Co²⁺ affinity chromatography. Selected proteins were then eluted with imidazole (elu His). Input proteins (elu Flag) and eluted proteins (elu His) were fractionated on SDS-PAGE gels and revealed by Coomassie staining (lanes 1 and 2) and by immunoblotting (lanes 3 and 4). Both Coomassie staining and immunoblot analyses reveal that the ratio of DLP-F/His-Asf observed before affinity chromatography (elu Flag) was equivalent to the ratio of DLP-F/His-Asf observed after affinity chromatography (elu His) (compare lane 1 with lane 2 and compare lane 3 with lane 4 in panel C). This suggests that all DLP molecules immunoprecipitated with the anti-Flag antibody were associated with ASF1.

A previous report (23) described the reduced incorporation of H3.3^core-GFP into chromosomes and the accumulation of nonincorporated H3.3^core-GFP within the nucleolus of mutant cells deficient for either xnp, Hira, or both (xnp; Hira). However, when we prepared spreads of Hira-deficient cells (Fig. 6A), we similarly observed reduced H3.3^core-RFP incorporation into the chromosomes, but the nucleolus, including nonincorporated histones, was never observed. This discrepancy may result from different stabilities of nonincorporated GFP and RFP fusion proteins. In particular, it was proposed that H3.3^core-GFP may form aggregates within the nucleolus. Nonincorporated H3.3^core-RFP could be differently processed and degraded after synthesis. Since we were able to visualize accumulating DLP in the nucleolus of H3.3-deficient cells, this discrepancy cannot be explained by the systematic degradation of the nucleolus during spread preparation.

We next investigated H3.3^core-RFP incorporation into the chromosomes of Hira^HR1, Dlp^G cells. We found that double mutant animals are viable and that H3.3 RI nucleosome assembly still occurs although at a lower level than that in wild-type cells (Fig. 6A). Since the incorporation of H3.3^core is abolished in Hira^HR1; xnp⁻ mutant cells (23), this suggests that the contribution of DLP to RI deposition is not equivalent to that of XNP.

A previously reported model suggested that XNP recruits a predeposition complex containing H3.3/H4 dimers, ASF1 (23), and additional factors. We anticipated that DLP may be such an additional factor, and we investigated the molecular interaction between DLP and ASF1 (41). DLP carrying a C-terminal Flag epitope tag was expressed in Sf9 insect cells in the presence of ASF1 carrying an N-terminal His epitope tag (Fig. 6B). The protein extracts were immunoprecipitated with the Flag antibody, and proteins were eluted with the Flag peptide prior to electrophoresis on SDS-containing polyacrylamide gels and Coomassie staining. We observed that His-ASF1 associates with DLP-Flag (Fig. 6B, lane 1). Interactions were also investigated by immunoblot analysis of eluted proteins, and we confirmed molecular interactions between His-ASF1 and DLP-Flag (Fig. 6B, lanes 3 and 4). Thus, the functioning of the predeposition complex is disrupted when DLP is absent and RI deposition of H3.3 is reduced.

Taken together, these data established that DLP is involved in the RI deposition of H3.3, likely as a subunit with ASF1 of a predeposition complex recruited to nucleosome-depleted chromatin gaps by XNP.

DISCUSSION

We have identified DLP as the Drosophila homolog of DAXX. We have shown that DLP is involved, likely in concert with XNP/dATRX, in the formation of pericentric heterochromatin of the X chromosome. Moreover, DLP is implicated in RI deposition of the histone variant H3.3 and, with ASF1, may constitute the central core of a predeposition complex recruited to chromatin gaps by XNP. The existence of such a complex was recently suggested (23).

In spite of the fact that both proteins do not molecularly associate as their mammal homologs do, we provide evidence that DLP and XNP functions are closely linked. DLP and XNP are located on the base of the X chromosome, and analysis of animals with simultaneous mutations of both dlp and xnp revealed that DLP and XNP likely act together during heterochromatin formation. In addition, both DLP and XNP are located next to the distal heterochromatic marker HP1 on the X chromosome of larvae carrying the ln(1)w^m4h rearrangement. Functional interactions between DLP and XNP were also supported by the similar behaviors of DLP and XNP in H3.3-deficient cells. In wild-type cells, in addition to the base of the X chromosome, the expression of XNP is detected at many sites across the chromosome arms where DLP is not observed (23). In H3.3 knockdown cells, DLP and XNP are present at many euchromatic sites of the chromosomes and are simultaneously associated with nucleolar chromatin of rDNA. Finally, overexpressed DLP binds to many interbands on the polytene chromosomes, suggesting that DLP may also be involved in chromatin organization at euchromatic sites in addition to the pericentric heterochromatin. However, the latter observation should be viewed with caution since it cannot be ruled out that overexpressed DLP is not present in its usual complex and is consequently mistargeted. Additional support for functional interactions between XNP and DLP is provided by genetic interactions between xnp and dlp. Indeed, a loss of xnp function is characterized by reduced viability, which is further aggravated when dlp function is simultaneously reduced, strongly indicating that xnp and dlp may functionally cooperate during the regulation of common targets. How XNP is recruited to nucleosome-depleted chromatin remains an important issue. XNP may be recruited to active genes by transcriptional machinery. Alternatively, XNP may bind structural motifs common to chromatin gaps or may simply bind exposed DNA. The homolog ATRX contains a PHD domain that can bind DNA or histone tails (42, 43). Recent work demonstrates that mammalian Hira may bind exposed DNA at chromatin gaps (44). Moreover, Hira and XNP bind active regions independently of one another (23). Hence, there may be multiple ways by which RI assembly factors recognize exposed DNA.

In Drosophila, the loss of H3.3 has a large impact on the viability and fertility of both males and females (21, 22). The Drosophila genome encompasses two single-copy genes, H3.3A and H3.3B, which code for the same protein. H3.3 is highly expressed in mitotic, meiotic, and postmeiotic male germ cells, probably reflecting high transcriptional activity (45–47). Interestingly, high dlp expression levels are observed in primary spermatocytes, in meiotic spermatocytes, and also in the germinal vesicle, suggesting that it may have important functions during the development of germ cells. In Drosophila testis, H3.3 disappears with the bulk of histones, prior to the accumulation of protamine and other sperm-specific nuclear basic proteins, leading to sperm DNA compaction at late stages of spermiogenesis (48). At fertilization, the assembly of nucleosomes on paternal DNA immediately follows the rapid loss of protamines from the decondensing male nucleus and is dependent on maternally provided factors like Hira and YEM (14, 16, 17). HIRA and YEM are crucial since male pronuclei fail to decondense at the pronuclear stage in eggs derived from female HIRA and YEM mutants. Hence, the function of HIRA/YEM at fertilization represents a unique example where deficient chromatin activity cannot be compensated for by other redundant factors.

In contrast, many examples suggest that H3.3 chaperones/chromatin remodeling complexes may display functional redundancy, as mutants of these factors have limited phenotypes (16, 18–20). In this context, DLP may be viewed as a typical example. The null allele dlp^G and the dlp⁴⁵ allele encoding a truncated protein lacking the C-terminal DHR necessary for H3.3 binding are viable and fertile, indicating that DLP and other chromatin factors may share common functions. Alternatively, DLP may display accessory functions during the development of germ cells. Characterization of the phenotypes of double mutant animals during germ cell development would help to resolve this important issue.

H3.3 was initially seen as a characteristic of active genes, with histone turnover occurring as a consequence of transcription (5, 49, 50). More recent studies revealed that H3.3 is widespread within the genome. In particular, H3.3 is deposited by ATRX/DAXX at telomeres and pericentric repeats. Interestingly, ATRX and DAXX are components of the same chromatin-remodeling complex and physically interact. Recently, Schneiderman and colleagues (23) proposed a model for how XNP, the Drosophila homolog of ATRX, and HIRA identify nucleosome-depleted DNA following gene activation and promote nucleosome assembly through a three-step process. Initially, XNP and HIRA bind exposed DNA at chromatin gaps where nucleosomes have been displaced. Subsequently, they cooperate to recruit a predeposition complex, including ASF1 and histones. In the final step, XNP and HIRA assist in the transfer of histones from delivery factors to DNA and are released when nucleosome assembly is complete. Even if XNP and DLP do not physically interact, here, we provide several evidences suggesting that DLP could be a component of the predeposition complex recruited by XNP/HIRA (Fig. 7).

FIG 7.

Model for how a H3.3/H4 predeposition complex including DLP is recruited at chromatin gaps by XNP/Hira. When old nucleosomes (red) have been disassembled (1), Hira and XNP recognize and bind exposed DNA at chromatin gaps (2). A separate predeposition complex containing DLP, ASF1, and the H3.3/H4 heterodimer is then recruited to chromatin-bound XNP and/or Hira (3). In a final step, XNP and Hira pry histones off ASF1-DLP and wrap them with DNA to rebuild new nucleosomes (green).

Both HIRA and XNP have been implicated in RI nucleosome assembly, but mutants of these factors have only limited phenotypes, revealing that they have redundant functions. Thus, either xnp or hira single mutants weakly affect H3.3 deposition, which is abolished in a double mutant of xnp and hira. This observation highlights the need for two distinct pathways during RI nucleosome assembly, one mediated by HIRA/YEM and the other mediated by XNP. Our study assigns a role to DLP during RI H3.3 deposition since H3.3 incorporation is affected in animals lacking DLP. DLP is thought to cooperate with XNP, and we were surprised to observe H3.3 deposition in animals lacking HIRA and DLP. Hence, we speculated that the pathway mediated by XNP is always functional although less efficient. It was recently proposed that XNP recognizes exposed DNA when a nucleosome has been displaced and serves as a binding platform for the recruitment of H3.3 predeposition complexes to chromatin gaps. Such complexes are believed to contain H3.3-H4 heterodimers, ASF1, and additional factors. In line with this, our data revealed physical interactions between ASF1 and DLP in protein extracts made from baculovirus-infected Sf9 cells coexpressing DLP and ASF1, suggesting that DLP may be one of these additional factors.

Our data provide additional links between HIRA, DLP, ASF1, and XNP during H3.3 incorporation (Fig. 7), but how they functionally interact during development remains an open question. Identification of their genomic targets and characterization of their activities during H3.3 deposition would obviously help to resolve this important issue.

MATERIALS AND METHODS

Fly stock, mutant isolation, and crosses.

All flies were raised at 25°C on standard cornmeal medium.

All crosses were performed at 25°C, unless stated otherwise.

The Mi(ET1)Daxx^MB03646 and P(EP)xnp(EP635) lines were obtained from the Bloomington Drosophila Stock Center. The dlp⁴⁵ and xnp⁵⁶ alleles were generated by imprecise excisions of the Mi(ET1)Daxx^MB03646 and P(EP)xnp(EP635) insertions using a Minos transposase and a P transposase, respectively. The P(hsILMiT)2.4 line (obtained from the Bloomington Stock Center) expresses the Minos transposase upon heat shock. P-element excision of P(EP)xnp(EP635) was performed according to standard methods by using the P(Delta2-3) transposase. Excision lines were analyzed by PCR across the dlp and xnp loci. Sequences of PCR primer pairs are available upon request.

Among the xnp-mutated lines, we selected xnp⁵⁶ for further analysis. xnp⁵⁶ removes 2,774 bp from the P-element insertion site in the 5′ untranslated region (UTR) of xnp into the protein-coding sequences. xnp⁵⁶, like previously described xnp alleles (19), deletes the two start codons of the xnp isoforms and lacks XNP, as revealed by immunoblot analysis of protein extracts of xnp⁵⁶/+ embryos and by polytene chromosome squashes of homozygous xnp⁵⁶ salivary glands. The Df(3R)Exel6202 genomic deficiency (obtained from the Bloomington Drosophila Stock Center) uncovers numerous genes in 3R, including xnp. When heterozygous xnp⁵⁶ lines were crossed with heterozygous Df(3R)Exel6202 lines, we observed a reduced number of xnp⁵⁶/Df(3R)Exel6202 progeny with regard to the expected Mendelian number. Hence, xnp⁵⁶, like previously described xnp alleles, has a strong effect on viability (18, 20).

The P(GawB)Daxx^NP4778 enhancer trap line (also termed dlp^G), obtained from Drosophila Genomics and Genetic Resources (Kyoto, Japan), is homozygous viable and carries a Gal4-containing transposon inserted into the 5′ untranslated sequences of dlp. The dlp^Gw line is a modified version of P(GawB)Daxx^NP4778 in which the mini-white gene has been deleted by imprecise excision. uas-dlp expression driven by either the original dlp^G or the modified dlp^Gw allele induces similar lethality of developing embryos. Moreover, no DLP staining is detected on spreads of dlp^Gw polytene chromosomes.

The GAL4-inducible knockdown line 5825R-3 (51) was obtained from the National Institute of Genetics Fly Stock Center (Japan) and was expressed by using the salivary gland-specific P(SGS3-GAL4)TP1 driver line (sgs3^G) (Bloomington Drosophila Stock Center). This genotype efficiently knocks down the new production of H3.3 (23).

mle¹ is a spontaneous loss-of-function allele of mle (52). msl-2²²⁷ is a loss-of-function allele of msl-2 (39). mle¹ and msl-2²²⁷ were obtained from the Bloomington Drosophila Stock Center.

Hira^HR1 is a loss-of-function allele described previously (17).

Lines 118E-15, 118E-10, 39C-12, 39C-3, and 118E-12 carry a variegating hsp70–mini-w insertion and were described previously (37). To test the effects of dlp and xnp mutants on the w expression of the variegating lines, homozygous females carrying variegating inserts were crossed to dlp⁴⁵/CyO, dlp^Gw/CyO, xnp⁵⁶/TM3Sb, dlp^Gw/CyO; xnp⁵⁶/TM3Sb, or dlp⁴⁵/CyO; xnp⁵⁶/TM3Sb males or to wild-type w¹¹¹⁸ males. Flies were collected at 1-day intervals, and progeny of different genotypes were directly compared by using side-by-side visual analysis. The CyO and TM3Sb balancer chromosomes were not found to substantially affect the silencing of the w reporter. Male progeny were collected from crosses, aged for 3 days, and photographed. Effects on gene silencing were consistent for all progeny from each cross, and representative pictures are shown.

Line G78b carries an Sxl-Pe-GFP reporter specifically expressed in female embryos and was provided by the Bloomington Drosophila Stock Center.

Baculovirus expression constructs and recombinant protein synthesis and purification.

cDNA fragments that encompass the complete coding sequences of Drosophila DLP, DLP⁴⁵, and ASF1 were cloned into pFastBac1 (Invitrogen) to express DLP-Flag, DLP⁴⁵-Flag, and His-ASF1. Coding sequences of Drosophila HA-H3 or HA-H3.3 were cloned together with coding sequences of Drosophila H4 into the bicistronic vector pFastBac1IR to express HA-H3/H4 and HA-H3.3/H4 heterodimers in Sf9 insect cells. pFastBac1IR is a pFastBac1 derivative carrying a 378-bp fragment of Rhopalosiphum padi virus containing an internal ribosome entry site (IRES) element shown to be active in Sf9 insect cells (53). Details of the constructs are available upon request.

Recombinant polypeptides were expressed in Sf9 cells and purified by either Flag or HA immunoaffinity chromatography.

Antibodies.

The following commercial primary antisera were used: rabbit anti-HP1 (catalog number 923901; Covance) (1/500 dilution), rabbit anti-CAF-1 (catalog number 53610; Abcam) (1:4,000), rat anti-HA (clone 3F10; Roche) (1/1,000), and mouse anti-β-tubulin (β-Tub) (clone KMX-1; Millipore) (1/800). We also used the following primary antisera: rabbit anti-NAP1 (from C. P. Verrijzer) (1/5,000), rabbit anti-ASF1 (from F. Karch) (1/2,000), rat anti-MSL-1 (from A. Akhtar) (1/400), and rabbit anti-XNP (from K. Ahmad) (1/2,000).

cDNA fragments of Drosophila DLP corresponding to amino acid residues 25 to 311 and 699 to 1002 were cloned in frame with an N-terminal His₆ tag into pET-28c (Novagen). The cDNA fragment of Drosophila XNP corresponding to amino acid residues 1003 to 1311 was cloned in frame with an N-terminal His₆ tag into pET-28c (Novagen). The antigens were expressed in bacteria, purified by using standard methods, and used to immunize mice.

Fluorescently labeled secondary antibodies (Thermo Fisher Scientific) were used at a 1:600 dilution for cytology, and horseradish peroxidase (HRP)-conjugated antibody (Jackson ImmunoResearch) was used at a 1:10,000 dilution for Western detection.

Transgenic lines.

Germ line transformants of P-element constructs were generated by using standard procedures (54). Details on the construction and sequences of the PCR primers are available upon request.

Wild-type dlp cDNA as well as mutant dlp⁴⁵ cDNA were PCR amplified by using the full-length cDNA clone SD 20887 received through the Drosophila Genomics Resource Center (Indiana University) as a template. They were subcloned into appropriate restriction sites of plasmid pUAST, which contains binding sites for the GAL4 activator upstream of the basal hsp70 promoter sequences.

The pCaSpeR (H3.3prom-H3.3-FHA) and pCaSpeR (H3.3prom-H3-FHA) plasmids were used to establish transgenic lines expressing histones H3.3 and H3 as fusion proteins carrying C-terminal Flag and HA epitope tags. Expression is conducted by a 2-kb H3.3 promoter fragment (16), which was PCR amplified from genomic DNA. The coding sequences of H3.3 and H3 were also PCR amplified from genomic DNA.

Sequences encoding the H3.3^core-monomeric red fluorescent protein (mRFP) fusion protein were cloned into pCaSpeRHS carrying the hsp70 promoter. The fusion protein was deleted for amino acid residues 4 to 36 of the histone N-terminal tail and can undergo only RI assembly. The expression of the fusion protein was induced at 37°C for 10 min and can be directly visualized on polytene spreads.

Double immunoaffinity purification.

Protein extracts from Drosophila embryos were prepared by using an S-190 extract protocol described previously (55), with slight modifications. Briefly, embryos 0 to 3 h after egg deposition were collected, rinsed in embryo wash buffer (0.7% [wt/vol] NaCl, 0.05% [vol/vol] Triton X-100) before dechorionation. Embryos were subsequently homogenized at 4°C in low-salt Ex10 buffer (10 mM HEPES K⁺ [pH 7.65], 10 mM KCl, 10% [vol/vol] glycerol, 1.5 mM MgCl₂, 0.2 mM EGTA, 1 mM β-glycerophosphate). The crude homogenate was supplemented with MgCl₂ (5 mM final concentration), and the salt concentration was adjusted up to 150 mM by the addition of high-salt Ex250 buffer (corresponding to Ex10 buffer with 250 mM KCl). The homogenate was centrifuged at 38,000 rpm during 2 h at 4°C in an SW41 rotor to remove floating lipids and to pellet yolk granules and cellular debris.

Tagged proteins were immunoprecipitated with anti-Flag M2-agarose (Sigma), eluted with Flag peptide (0.5 mg/ml), further affinity purified with anti-HA antibody-conjugated agarose (Sigma), and eluted with HA peptide (1 mg/ml). The peptides were diluted in Ex150 buffer (equivalent to Ex10 buffer containing 150 mM KCl), and during each step of immunopurification, beads were washed in Ex150 buffer. Complexes were resolved by SDS-PAGE and stained by using a Silver Quest kit (Invitrogen).

The identification of proteins was carried out by the Taplin Biological Mass Spectrometry Facility (Harvard Medical School, Boston, MA).

Indirect immunofluorescence.

Immunostaining of testis squashes was performed as described previously (56). Immunostaining of ovaries was performed as described previously (57). Immunostaining of polytene squashes was performed as described previously (58).

A protocol described previously (59) was used to directly visualize RI deposition of H3.3-mRFP on polytene chromosomes.

Viability test.

Embryos lacking xnp function were selected by means of the TM3SbSerGFP balancer chromosome. Embryos were allowed to develop at 25°C. Adults were counted, and for each genotype, four independent experiments were performed.