ATP-facilitated chromatin assembly with a nucleoplasmin-like protein from Drosophila melanogaster.

To gain a better understanding of the factors that can mediate chromatin assembly, we have purified and cloned a core histone-binding protein from Drosophila melanogaster embryos. This protein resembles Xenopus laevis nucleoplasmin, and it has therefore been termed dNLP, for Drosophila nucleoplasmin-like protein. dNLP is a nuclear protein that is present throughout development. Both purified native and recombinant dNLP bind to core histones and can function in the assembly of approximately regularly spaced nucleosomal arrays in a reaction that additionally requires DNA, purified core histones, ATP, and a partially purified fraction (containing at least one other assembly activity). We also analyzed the properties of an N-terminally truncated version of dNLP, termed dNLP-S, and found that the deletion of the N-terminal 31 residues of dNLP results in a loss of the specificity of the interaction of dNLP with core histones. We then compared the abilities of dNLP and Drosophila nucleosome assembly protein-1 (dNAP-1) to promote the decondensation of Xenopus sperm chromatin, a process that can be mediated by nucleoplasmin. We observed that dNAP-1, but not dNLP, was able to promote the decondensation of sperm chromatin. These and other data collectively suggest that dNLP may participate in parallel with other histone-binding proteins such as dNAP-1 in the assembly of chromatin.

To gain a better understanding of the factors that can mediate chromatin assembly, we have purified and cloned a core histone-binding protein from Drosophila melanogaster embryos. This protein resembles Xenopus laevis nucleoplasmin, and it has therefore been termed dNLP, for Drosophila nucleoplasmin-like protein. dNLP is a nuclear protein that is present throughout development. Both purified native and recombinant dNLP bind to core histones and can function in the assembly of approximately regularly spaced nucleosomal arrays in a reaction that additionally requires DNA, purified core histones, ATP, and a partially purified fraction (containing at least one other assembly activity). We also analyzed the properties of an N-terminally truncated version of dNLP, termed dNLP-S, and found that the deletion of the N-terminal 31 residues of dNLP results in a loss of the specificity of the interaction of dNLP with core histones. We then compared the abilities of dNLP and Drosophila nucleosome assembly protein-1 (dNAP-1) to promote the decondensation of Xenopus sperm chromatin, a process that can be mediated by nucleoplasmin. We observed that dNAP-1, but not dNLP, was able to promote the decondensation of sperm chromatin. These and other data collectively suggest that dNLP may participate in parallel with other histonebinding proteins such as dNAP-1 in the assembly of chromatin.
Chromatin assembly is a fundamental process that is involved in a broad range of biological phenomena such as gene regulation, recombination, DNA repair, and progression through the cell cycle, and it is therefore important to investigate the factors that participate in the formation of chromatin (for reviews, see Refs. [1][2][3][4][5][6][7]. The analysis of chromatin assembly has revealed that the deposition of the core histones H3 and H4 precedes the incorporation of H2A and H2B into chromatin, and that both the pre-existing and newly synthesized histones are randomly distributed among the daughter DNA strands. Moreover, the newly synthesized histones, which are more highly acetylated in their lysine-rich N-terminal tails relative to bulk histones, become deacetylated after assembly into chromatin. Chromatin assembly commences immediately after DNA replication, and DNA replication and chromatin assembly appear to be coupled, although possibly by an indirect mechanism, as chromatin assembly appears to occur preferentially, but not obligatorily, with newly replicated DNA (see, for instance, Ref. 8).
The mechanism of chromatin assembly is likely, however, to be more complex than the random deposition of histones that is mediated by the core histone-binding factors alone. For instance, it is known from studies of chromatin assembly activities in crude extracts derived from Xenopus laevis oocytes (25), HeLa cells (26), or Drosophila melanogaster embryos (27,28) that the assembly of approximately regularly spaced nucleosomal arrays is an ATP-dependent process. Biochemical fractionation of a chromatin assembly extract from Drosophila embryos led to the identification of two fractions, termed dCAF-1 and dCAF-4, which, when combined, were able to reconstitute the ATP-facilitated assembly of nucleosomal arrays, as was observed with the crude extract (29). One component in the dCAF-4 fraction was purified and cloned, and found to be the Drosophila homologue of NAP-1 (dNAP-1) (21). Purified recombinant dNAP-1 was observed to function in a cooperative manner with the active component(s) in the dCAF-1 fraction to mediate the ATP-facilitated assembly of nucleosomal arrays. The ATP-utilizing chromatin assembly factor(s) in the dCAF-1 fraction has not yet been identified, although it is known that the Drosophila homologue of CAF-1 (termed dCAF-1 protein, which should not be confused with the dCAF-1 fraction) is present in the partially purified fraction. These and other data collectively suggest a model for chromatin assembly wherein nucleoplasmin and/or NAP-1 function as a carrier for H2A-H2B, N1/N2 and/or CAF-1 function as a carrier for H3-H4, and an additional factor(s) (in the dCAF-1 fraction) mediates the ATP-dependent assembly of nucleosomes by acting in conjunction with the histone chaperones.
In this study, we focus upon the further investigation of components in the dCAF-4 fraction, and describe the purification, cloning, and characterization of a second core histonebinding protein. The primary amino acid sequence of this protein resembles that of X. laevis nucleoplasmin, and it has therefore been termed dNLP, for Drosophila nucleoplasminlike protein.

EXPERIMENTAL PROCEDURES
Purification of dNLP-All operations were performed at 4°C. The S-190 chromatin assembly extract (28,30) was sequentially subjected to chromatography on DEAE-Sepharose FF (Pharmacia Biotech Inc.), SP-Sepharose FF (Pharmacia), and Q-Sepharose FF (Pharmacia) resins to give the dCAF-4 fraction (29). To this fraction, 4 M ammonium sulfate (adjusted to pH 7 by the addition of NaOH) was added to a final concentration of 2.75 M. The mixture was incubated for 20 min and then subjected to centrifugation (Sorvall SS-34 rotor; 10,000 rpm for 10 min). The supernatant (containing the dNLP) was applied to a phenyl-Sepharose CL-4B (Pharmacia) column (column volume ϭ 0.5 ml; column dimensions (diameter ϫ length) ϭ 0.5 ϫ 2.5 cm; flow rate ϭ 0.1 ml/min; fraction size ϭ 0.5 ml). The column was washed with three column volumes of buffer R (10 mM Hepes (K ϩ ), pH 7.6, containing 10 mM KCl, 1.5 mM MgCl 2 , 10% (v/v) glycerol, 10 mM ␤-glycerophosphate, 1 mM dithiothreitol, 0.2 mM phenylmethylsulfonyl fluoride) containing 2.5 M ammonium sulfate, and protein was eluted with a linear gradient (over 10 column volumes) from 2.5 M to 0 M ammonium sulfate in buffer R. Fractions were analyzed by SDS-polyacrylamide gel electrophoresis, and the peak dNLP-containing fractions were pooled and then dialyzed against buffer R.
Isolation of cDNA Encoding dNLP-The purified dNLP was subjected to electrophoresis on a 12% SDS-polyacrylamide gel, and the protein was visualized by staining with Coomassie Brilliant Blue G. The dNLP band was excised, and the protein was digested with a lysylendopeptidase (Achromobacter protease I; Wako Chemicals). The resulting peptides were purified by high performance liquid chromatography with a C-18 column (Vydac) and sequenced by automated Edman degradation (Applied Biosystems model 470, 473, or 477). A fragment of the cDNA was obtained by using reverse transcription-PCR with oligonucleotides corresponding to the coding sequences of the peptides, and six independent cDNAs (four of which were of the dNLP type, and two of which were of the dNLP-S type) were isolated by screening a Drosophila embryo cDNA library (embryos were collected from 0 to 4 h after egg deposition) in ZAPII with the radiolabeled, PCR-amplified fragment. Both the dNLP and dNLP-S cDNAs were sequenced completely from both strands. The dNLP cDNA was directly subcloned into the NcoI and BamHI sites of pET15b (Novagen) to give the bacterial expression plasmid pETdNLP. The dNLP-S cDNA was amplified by PCR and then subcloned into the NcoI and XhoI sites of pET15b to give the plasmid pETdNLP-S, and the coding region of this plasmid was completely resequenced to confirm the absence of mutagenesis during the PCR process.
Purification of Recombinant dNLP and dNLP-S-A freshly transformed colony of pETdNLP (or alternatively, pETdNLP-S) in Escherichia coli strain BL21(DE3) was grown at 37°C in a volume of 2 liters to A 600 nm of approximately 0.6, and synthesis of dNLP (or dNLP-S) was induced by the addition of isopropyl-␤-D-thiogalactopyranoside to 0.4 mM final concentration. The cultures were incubated for an additional 3 h at 37°C. Unless stated otherwise, all subsequent operations were performed at 4°C. The bacteria were pelleted by centrifugation (Sorvall GSA rotor; 5000 rpm for 5 min) and resuspended in 40 ml of Buffer R. The bacteria were lysed by sonication, and the insoluble material was removed by centrifugation (Sorvall SS-34 rotor; 10,000 rpm for 10 min). Then, 4 M ammonium sulfate (pH 7.0) was added to the supernatant to final concentration of 2.25 M (for dNLP; for dNLP-S, the final concentration of ammonium sulfate was 1.75 M). The mixture was incubated for 20 min, and then subjected to centrifugation (Sorvall SS-34 rotor; 10,000 rpm for 10 min). The supernatant was applied to a phenyl-Sepharose CL-4B (Pharmacia) column (column volume ϭ 30 ml; column dimensions (diameter ϫ length) ϭ 1.6 ϫ 15 cm; flow rate ϭ 1 ml/min; fraction size ϭ 5 ml). The column was washed with three column volumes of buffer R containing 2.25 M ammonium sulfate (for dNLP; for dNLP-S, the buffer R contained 1.75 M ammonium sulfate), and protein was eluted with a linear gradient (10 column volumes) from 2.25 to 0 M ammonium sulfate (for dNLP; for dNLP-S, the gradient was from 1.75 M to 0 M ammonium sulfate). Fractions were analyzed by SDS-polyacrylamide gel electrophoresis, and the peak dNLP-containing (or dNLP-S-containing) fractions were pooled and dialyzed against buffer R containing 0.2 M NaCl. The resulting sample was applied onto tandemly linked P-11 (first column; phosphocellulose, Whatman; column volume ϭ 30 ml; column dimensions (diameter ϫ length) ϭ 1.6 ϫ 15 cm; flow rate ϭ 0.2 ml/min) and Mono Q (second column; HR 5/5; Pharmacia) columns. After application of the protein sample, the P-11 column was removed, and the Mono Q resin was washed with four column volumes of buffer R containing 0.2 M NaCl. dNLP (or dNLP-S) was eluted with a linear gradient (10 column volumes) from 0.2 to 1 M NaCl in buffer R. dNLP (or dNLP-S) eluted from the Mono Q resin at roughly 250 mM NaCl. The protein content of the fractions was analyzed by 12% SDS-polyacrylamide gel electrophoresis. Nuclease activity in the fractions was detected by incubation of plasmid DNA with the samples at 37°C for 1 h in the presence of 10 mM MgCl 2 , followed by analysis of the DNA by 1% agarose gel electrophoresis. Bacterial nucleic acid contamination was monitored by 1% agarose gel electrophoresis and staining with ethidium bromide. The peak dNLP-or dNLP-S-containing fractions, which were not contaminated with nucleases or bacterial nucleic acids, were dialyzed against buffer R (with no additional salt), frozen in liquid nitrogen, and stored at Ϫ80°C.
Assembly and Analysis of Chromatin-Chromatin assembly reactions were performed under the conditions (at 57 mM KCl) described by Bulger et al. (29) by using circular plasmid DNA (470 ng, previously relaxed by incubation with purified, recombinant Drosophila topoisomerase I), dCAF-1 fraction (19 g of total protein; from the 0.25-0.5 M potassium phosphate fraction of the hydroxylapatite column; Ref. 29), purified recombinant dNLP, dNLP-S, or dNAP-1 (amounts as indicated in the figure legends), purified Drosophila core histones (410 ng), 3 mM ATP, and an ATP regeneration system (30 mM phosphocreatine and 1 g/ml creatine phosphokinase). The core histones were purified from Drosophila embryos (collected from 0 to 12 h after egg deposition) by using a nondenaturing procedure, as described by Bulger and Kadonaga (30). Recombinant dNAP-1 was purified as described elsewhere (21). DNA supercoiling and micrococcal nuclease digestion assays were carried out as described previously (28 -30). All experiments were performed a minimum of two times (but typically, several times) to ensure reproducibility of the data.
Analysis of Protein-Histone Interactions-Polyclonal antibodies against purified recombinant dNLP were generated in rabbits. The immunoprecipitation of dNLP and dNAP-1 from a crude Drosophila embryo extract was performed as described previously (21). The in vitro co-immunoprecipitation of dNLP and dNLP-S with core histones was carried out as follows. Purified dNLP proteins (native dNLP, 6 g; recombinant dNLP, 6 g; recombinant dNLP-S, 5 g; or no protein, as a control) were incubated with purified core histones (4 g) at 4°C for 1 h in buffer R in a final volume of 80 l. Under these conditions, the molar ratio of dNLP:dNLP-S:core histone octamer is 10:10:1 (note that a protein:octamer ratio of 10:1 corresponds to a protein:histone ratio of 1.25:1). Immunoaffinity-purified anti-dNLP (20 l; these antibodies recognize both dNLP and dNLP-S) was then added to each of the samples, which were incubated at 4°C for 1 h. Then, a 50% (v/v) slurry (20 l) of protein A-Sepharose CL-4B (Pharmacia) was added to each sample, each of which was incubated for 1 h at 4°C. The protein A-Sepharose beads were washed twice with dilution buffer (10 mM Tris-HCl, pH 7.9, 0.14 M NaCl, 0.1% Triton X-100, 0.1% bovine serum albumin), once with TSA (10 mM Tris-HCl, pH 7.9, 0.14 M NaCl, 0.04% (w/v) sodium azide), and once with 50 mM Tris-HCl (pH 7.9). The beads were suspended in SDS sample buffer and boiled for 5 min. A portion of the supernatant was then subjected to 15% SDS-polyacrylamide gel electrophoresis and staining with Coomassie Brilliant Blue R-250.
Immunostaining of Drosophila Schneider Cells-Schneider cells were grown on round coverslips that were coated with poly-D-lysine and then fixed with 4% paraformaldehyde for 10 min in 1 ϫ phosphatebuffered saline (1 ϫ PBS) at room temperature. The cells were permeabilized by treatment with methanol at Ϫ20°C for 10 min and dried by rinsing with acetone. The coverslips were incubated in buffer B (5% calf serum in 1 ϫ PBS containing 0.1% Tween 20), and then with affinitypurified rabbit polyclonal anti-dNLP for 1 h at room temperature in buffer B. The coverslips were washed with 1 ϫ PBS at room temperature; incubated with rhodamine-conjugated, anti-rabbit antibody (Boehringer Mannheim) for 1 h at room temperature in buffer B; and rinsed with 1 ϫ PBS. The samples were then mounted in 80% glycerol and placed on a glass slide. The cells were observed by using a 100ϫ oil immersion objective with a Nikon microscope.
Assay for Decondensation of Xenopus Sperm Chromatin-Demembranated Xenopus sperm nuclei (5 ϫ 10 4 nuclei/l) were prepared by the method of Lohka and Masui (31) and were the generous gift of Drs. H. Yan and J. Newport (University of California, San Diego). The demembranated sperm nuclei (final concentration of 5 ϫ 10 3 nuclei/l) were incubated with purified native dNLP, purified recombinant dNLP, or purified recombinant dNAP-1 (at a final concentration of 700 g/ml for each protein) in a final volume of 10 l in 50 mM Hepes (K ϩ ), pH 7.6, buffer containing 75 mM potassium acetate, 0.5 mM spermidine, and 0.15 mM spermine. The samples were mixed gently, and at each time point, a 2-l aliquot was removed and mixed with an equal volume of 50% glycerol in 1 ϫ PBS containing 10 g/ml Hoechst 33258 and 1% formaldehyde. The slides were examined by using a Nikon microscope.

RESULTS
Purification and Cloning of dNLP-Further analysis of the dCAF-4 fraction (29), from which we had purified and cloned dNAP-1 (21), led to the identification of a second, less abundant protein with an apparent molecular mass of 22 kDa that appeared to bind to core histones. To investigate the function of this histone-binding protein in chromatin assembly, we purified the 22-kDa protein from the dCAF-4 fraction to greater than 90% homogeneity, generated and purified peptides, obtained partial amino acid sequence of the peptides, and isolated six independent cDNAs that encoded the protein (Fig. 1A). Comparison of the predicted amino acid sequence of this histone-binding protein with those of known proteins (Fig. 1B) revealed a similarity (31% identity) to another core histonebinding protein, nucleoplasmin, which has been cloned from X. laevis (12, 13). Therefore, the 22-kDa protein from the dCAF-4 fraction was termed Drosophila nucleoplasmin-like protein (dNLP). Examination of dNLP, nucleoplasmin, and dNAP-1 reveals that all three factors have a C-terminal stretch of 24 -28 amino acid residues that contains a high proportion (71-81%) of acidic residues (Fig. 1, B and C), which may be important for the binding of histones.
Among the six independent cDNAs that were obtained, there were two different types. Four of the cDNAs correspond to the protein that we have denoted as dNLP, while two of the cDNAs encode an N-terminally truncated version of dNLP that we have designated as dNLP-Short (dNLP-S). The calculated molecular mass of dNLP is 16,985 daltons (which is significantly less than its apparent molecular mass of 22 kDa; see, for instance, Fig. 2A), while the calculated mass of dNLP-S is 13,522 daltons (which has an apparent molecular mass of 18 kDa; Fig. 2A). Both of the independently isolated dNLP-S cDNAs differ from the dNLP cDNAs only upstream of nucleotide 112, as in the nomenclature depicted in Fig. 1A; thus, the dNLP-S cDNAs may have been generated from alternately spliced mRNA species. Northern blot analysis with a probe that is complementary to both dNLP and dNLP-S mRNAs revealed an apparent single band at approximately 1.2 kilobases. Southern blot analysis indicated that there is a single gene encoding dNLP/dNLP-S, but unfortunately, we were not able to determine the cytogenetic location of the dNLP gene in Drosophila   FIG. 1. The 22-kDa core histone-binding protein in the dCAF-4 fraction is a Drosophila nucleoplasmin-like protein (dNLP). A, nucleotide sequence of a cDNA encoding dNLP. The predicted amino acid sequence is given in the single-letter amino acid code. The Nterminal three amino acid residues of dNLP-S are indicated in bold type. The amino acid sequences of peptides that were derived from purified native dNLP are underlined. The dNLP and dNLP-S cDNA sequences have been submitted to GenBank (accession numbers U59498 and U59497, respectively). B, alignment of the amino acid sequence of dNLP with that of nucleoplasmin from X. laevis (12, 13).
The sequence alignment was compiled by using the PileUp program of the GCG sequence analysis package. The asterisks denote amino acid residues that are identical in both proteins. A stretch of acidic amino acid residues that is present in both proteins is highlighted with bold type. The nuclear localization signal (NLS) of nucleoplasmin (48, 49) is indicated in bold italics. C, comparison of the distribution of acidic and basic amino acid residues in dNLP and dNAP-1.
polytene chromosomes. 2 Assembly of Nucleosomal Arrays with dNLP-Next, we tested the ability of dNLP and dNLP-S to function in the ATP-facilitated assembly of nucleosomal arrays in conjunction with the assembly factor(s) in the dCAF-1 fraction. First, we purified native dNLP, recombinant E. coli-synthesized dNLP, and recombinant E. coli-synthesized dNLP-S to greater than 90% homogeneity ( Fig. 2A). (The preparation of native dNLP did not contain any detectable amount of dNLP-S.) Then, we compared the relative activities of native dNLP, recombinant dNLP, recombinant dNLP-S, and purified recombinant dNAP-1 (synthesized in E. coli; Ref. 21) in assembly reactions that contained dCAF-1 fraction, DNA, purified core histones, and ATP. With the DNA supercoiling assay (Fig. 2B), dNAP-1 exhibited the highest assembly activity, whereas native and recombinant dNLP possessed assembly activity that was consistently lower than that of dNAP-1. In contrast, dNLP-S was inactive. We further characterized the assembly of chromatin with dNLP by using DNA supercoiling and micrococcal nuclease digestion assays. As depicted in Fig. 2 (C and D), the efficient assembly of nucleosomal arrays requires dNLP, dCAF-1 fraction, and ATP. In addition, when both dNAP-1 and dNLP were used simultaneously, there was no apparent synergism between the factors (data not shown). Therefore, both dNAP-1 and dNLP are able to function with the dCAF-1 fraction in the ATP-facilitated assembly of nucleosomal arrays, although dNAP-1 appears to mediate chromatin assembly more efficiently than dNLP.
Binding of dNLP and dNLP-S to Core Histones-Because dNLP was initially identified as a histone-binding protein in the dCAF-4 fraction, we examined the binding of dNLP and dNLP-S to core histones. First, we tested whether core histones will co-immunoprecipitate with dNLP or dNLP-S when either protein is combined with core histones in an in vitro binding assay. In this experiment, native dNLP, recombinant dNLP, or recombinant dNLP-S were combined with purified core histones at a 10:1 molar ratio of protein:octamer (which is identical to a 1.25:1 ratio of protein:histone, at which dNLP functions in chromatin assembly; Fig. 2B), and then the dNLP species were immunoprecipitated with polyclonal antibodies that recognize both dNLP and dNLP-S (or with preimmune serum, as a control). As shown in Fig. 3A, all four core histones coimmunoprecipitated in the presence, but not in the absence of either dNLP or dNLP-S. It thus appears that both dNLP and dNLP-S can bind to the core histones.
We then investigated whether core histones will co-immunoprecipitate with dNLP from a whole Drosophila embryo extract. A crude extract (prepared from embryos collected from 0 to 2 h after egg deposition) was therefore subjected to immunoprecipitation with polyclonal antibodies that recognize dNLP (or with preimmune serum as a control). Because it is known that the core histones H2A-H2B will co-immunoprecipitate with dNAP-1 from a whole embryo extract (21), we had also used antibodies against dNAP-1 as a positive control. The resulting immunoprecipitates were subjected to Western blot analysis with antibodies that recognize H2A, H2B, H3 (but not H4), dNLP, and dNAP-1. As seen in Fig. 3B, there was no detectable co-immunoprecipitation of core histones with dNLP, while under the same conditions, the core histones H2A-H2B co-immunoprecipitated with dNAP-1, as observed previously (21). Moreover, we had performed additional experiments with embryos at later stages of development, and did not observe any co-immunoprecipitation of histones with dNLP (data not shown). Hence, as reported previously for X. laevis nucleoplasmin in oocytes (Refs. 32 and 33, but see also Refs. 15 and 17), there is no detectable co-immunoprecipitation of core histones with dNLP in Drosophila embryos. It is, in addition, unlikely that the antibodies against dNLP inhibit binding of core histones to dNLP because the histones are able to co-immunoprecipitate with dNLP when the purified proteins are combined (Fig. 3A). These results therefore suggest that the majority of dNLP in embryos is not stably associated with core histones.
To characterize further the binding of dNLP to core histones, 2 T. Laverty and G. Rubin, unpublished data. we carried out glycerol gradient sedimentation analyses (Fig.  4). In these experiments, we were particularly interested in the relative affinities of the binding of dNLP, dNLP-S, and dNAP-1 to the core histones. For instance, when dNLP-S was incubated with core histones for 30 min (at a protein:histone ratio of 1.25:1) and then combined with dNLP (at a dNLP-S:dNLP ratio of 1:1) and subjected to glycerol gradient sedimentation, the core histones were associated with the dNLP and not, to any detectable extent, with the dNLP-S (Fig. 4E). Hence, it appears that dNLP binds to histones with higher affinity than dNLP-S. We then performed a similar experiment with dNLP and dNAP-1 and found that dNAP-1 binds to histones with a higher affinity than dNLP (Fig. 4F). (Note that these experiments were performed with an excess of histone-binding proteins relative to core histones. At a higher core histone:histonebinding protein ratio, both dNAP-1 and dNLP were bound to the histones.) Thus, these data suggest that the relative affinity of the proteins for core histones is as follows: dNAP-1 Ͼ dNLP Ͼ dNLP-S. Moreover, the results shown in Fig. 4F, in which the core histones were associated with dNAP-1 in a mixture of dNAP-1 and dNLP, are consistent with the coimmunoprecipitation studies with the embryo extracts (Fig.  3B), wherein the core histones were bound to dNAP-1 but not to dNLP.
The glycerol gradient sedimentation analysis also revealed that dNLP-S appears to form a high molecular mass aggregate with core histones (Fig. 4D), which is in contrast to the behavior of dNLP, which forms a soluble histone complex (Fig. 4E).
These results suggest that the co-immunoprecipitation of core histones with dNLP-S, as observed above (Fig. 3A), might be due to nonspecific interaction of the histones with dNLP-S. Thus, although dNLP-S differs from dNLP only by deletion of 31 amino acid residues at the N terminus and retains the majority of dNLP that is related to nucleoplasmin (Fig. 1, A and  B), the ability of dNLP-S to interact with core histones and to function in chromatin assembly is distinct from that of dNLP.
dNAP-1, but Not dNLP, Can Mediate Decondensation of Sperm Chromatin-Xenopus nucleoplasmin has been shown to mediate the decondensation of sperm chromatin (34,35) by a mechanism that appears to involve the loss of two spermspecific proteins and the incorporation of H2A and H2B into the chromatin (36). In addition, two Drosophila proteins, termed FIG. 3. Characterization of dNLP binding to core histones by co-immunoprecipitation analysis. A, in vitro co-immunoprecipitation of core histones with dNLP or dNLP-S. Purified dNLP proteins (native dNLP, 6 g; recombinant dNLP, 6 g; recombinant dNLP-S, 5 g; or no protein, as a control) were incubated with purified core histones (4 g), and the mixture was subjected to immunoprecipitation by sequential addition of immunoaffinity-purified anti-dNLP antibodies (which recognize both dNLP and dNLP-S; the corresponding preimmune serum was used as a control) and protein A-Sepharose. The resulting samples were washed and then analyzed by 15% SDS-polyacrylamide gel electrophoresis and staining with Coomassie Brilliant Blue R-250. The positions of dNLP, dNLP-S, and core histones are indicated. The major background bands are the immunoglobulin heavy and light chains. The protein molecular mass markers are similar to those used in Fig. 2A. CH, core histones. B, immunoprecipitation of dNLP and dNAP-1 from a crude, whole embryo extract. A crude extract derived from ϳ50 mg of Drosophila embryos (collected from 0 to 2 h after egg deposition) was subjected to immunoprecipitation with preimmune serum control (this preimmune serum corresponds to the dNLP antiserum), anti-dNLP, or anti-dNAP-1, as indicated. The resulting immunoprecipitates were analyzed by Western blot with antibodies that recognize dNLP, dNAP-1, H2A, H2B, and H3 (but not histone H4).

FIG. 4.
Relative binding affinities of dNLP, dNLP-S, and dNAP-1 to core histones. Protein-core histone (CH) mixtures were subjected to 20 to 50% glycerol gradient sedimentation in a SW55 (Beckman) rotor at 50,000 rpm for 17 h at 4°C, and the gradient fractions were analyzed by 15% SDS-polyacrylamide gel electrophoresis and staining with Coomassie Brilliant Blue R-250. A, core histones only (10 g). B, dNLP-S only (12 g). C, dNLP only (15 g). D, core histones (10 g) and dNLP-S (12 g) were incubated at 4°C for 30 min before applying the sample to the gradient. E, core histones (10 g) and dNLP-S (12 g) were incubated at 4°C for 30 min, and then dNLP (15 g) was added to the mixture immediately before applying the sample to the gradient. In lanes 1 and 2, the bands below dNLP-S are a contaminant in the dNLP-S preparation (see, for example, panel B), and are not core histones that are associated with dNLP-S. F, core histones (10 g) and dNLP (15 g) were incubated at 4°C for 30 min, and then dNAP-1 (39 g) was added to the mixture immediately before applying the sample to the gradient. In these experiments, the relative molar proportion of dNLP:dNLP-S:dNAP-1:histone was 1.25:1.25:1. The extent of sedimentation of the dNLP-histone complexes, as seen in panel E, is identical to that observed in the absence of dNLP-S (data not shown), and is greater than that of dNLP in the absence of histones (panel C). Likewise, the extent of sedimentation of the dNAP-1-histone complexes, as seen in panel F, is identical to that observed in the absence of dNLP, and is greater than that of dNAP-1 in the absence of histones (as shown in Ref. 21). p22 (37) and DF 31 (38), were also observed to decondense Xenopus sperm chromatin. Both p22 and DF 31 exhibit some characteristics, such as heat stability and sperm decondensation activity, that are similar to those of nucleoplasmin. Moreover, because p22 (37) and dNLP (this study) migrate on SDSpolyacrylamide gels with the same apparent molecular mass, it is possible that these proteins are related.
We therefore examined the ability of dNLP and, for comparison, dNAP-1 to mediate decondensation of demembranated Xenopus sperm chromatin (Fig. 5). In contrast to the behavior of nucleoplasmin or p22, neither native dNLP nor recombinant dNLP was able to promote the decondensation of sperm chromatin under various conditions and concentrations (Fig. 5, panels A-F; data not shown). However, dNAP-1 was highly active for sperm chromatin decondensation (Fig. 5, panels G-L). These results indicate that dNAP-1, but not dNLP, is able to mediate the decondensation of Xenopus sperm chromatin. The degree to which NAP-1 participates in sperm chromatin decondensation in vivo remains to be determined, but the involvement of NAP-1 in this process should at least be considered. Furthermore, it should be noted that the activity of either native or recombinant dNLP is not identical to that of p22, and therefore, either the proteins are not related or some aspect of the preparation of p22 (relative to that of dNLP) renders the protein active for sperm chromatin decondensation.
Subcellular Localization and Presence of dNLP throughout Development-We have also examined the subcellular localization of dNLP. dNLP is located in the nucleus of Drosophila Schneider cultured cells (Fig. 6) and Drosophila embryos (data not shown). By comparison, in Drosophila embyos, dNAP-1 appears to be predominantly cytoplasmic during G 2 phase and in the nucleus and cytoplasm during S phase (21). Moreover, we did not observe nucleolar localization of dNLP, as has been observed with the nucleoplasmin-related NO38 protein (39 -41). Thus, dNLP appears to be exclusively localized to the nucleus, as is Xenopus nucleoplasmin (32,33,42).
Finally, we determined the amount of dNLP and dNAP-1 proteins that were present throughout Drosophila development (Fig. 7). Both dNLP and dNAP-1 were at the highest levels in early embryos and remained at reduced, but significant levels throughout development. This property of dNLP and dNAP-1 is different from that seen with Xenopus nucleoplasmin, which is present in embryos to the feeding tadpole stage, but then decreases to undetectable levels in advanced tadpoles and adults (12,43,44). The presence of dNLP and dNAP-1 at their highest levels in early embryos and at lesser but detectable amounts during later stages of development are consistent with a function, such as chromatin assembly/disassembly or condensation/ decondensation (as in mitosis), that would be needed throughout the life cycle, but at especially high levels in the early embryo during which every cycle of DNA replication and nuclear division is completed approximately every 10 min. DISCUSSION We have described the purification, cloning, and characterization of a nucleoplasmin-like protein from D. melanogaster, termed dNLP, which is a nuclear protein (Fig. 6) that is present throughout development (Fig. 7). dNLP binds to core histones in vitro (Figs. 3A and 4E) and functions in the ATP-facilitated assembly of nucleosomal arrays in conjunction with an activity (or activities) in a partially purified dCAF-1 fraction (Fig. 2). The properties of dNLP resemble those of nucleoplasmin as well as those of two Drosophila proteins named p22 (37) and DF 31 (38), with the major exception that dNLP, unlike the other three proteins, is not able to promote the decondensation of sperm chromatin (Fig. 5). These and other data suggest that dNLP may participate in parallel with other histone-binding proteins such as dNAP-1 in the assembly of chromatin.
dNLP-S-Two independently isolated cDNA clones encode an N-terminally truncated version of dNLP, designated as dNLP-S, which is identical to dNLP except that it is lacking amino acid residues 2-32. However, dNLP-S is not present in our purified preparations of native dNLP ( Fig. 2A), is not detectable by Western blot analysis (Fig. 7), and is inactive for chromatin assembly (Fig. 2B). dNLP-S is able to bind to core histones in vitro in a co-immunoprecipitation assay (Fig. 3A), but it appears to form a high molecular mass aggregate with the core histones, as analyzed by glycerol gradient sedimenta- tion (Fig. 4D). Because dNLP-S protein is not detectable by Western blot analysis and does not appear to possess any biologically significant activity, the function, if any, of the protein encoded by the dNLP-S cDNA is not apparent. It is interesting, however, to compare the differences in the properties of histone-dNLP complexes with those of histone-dNLP-S complexes. Although dNLP and dNLP-S are closely related proteins, dNLP forms a soluble complex with histones (Fig. 4E) and is active for chromatin assembly (Fig. 2, B-D), whereas dNLP-S forms a high molecular mass aggregate with the core histones (Fig. 4D) and is inactive for chromatin assembly (Fig.  2B). Hence, the deletion of the N-terminal 31 amino acid residues of dNLP results in a loss of specificity of the interaction of dNLP with core histones.
Comparison of Nucleoplasmin with dNLP, p22, and DF 31-Among dNLP, p22, and DF 31, the biochemical properties of DF 31 most closely resemble those of nucleoplasmin. Specifically, characteristics of Drosophila DF 31 that are similar to those of Xenopus nucleoplasmin, which are apparently dissimilar to those of dNLP and/or p22, are as follows: (i) apparent molecular mass: nucleoplasmin has an apparent molecular mass of 29 -30 kDa (32,42), while DF 31 has an apparent mass of 31 kDa (38); (ii) phosphorylation: both nucleoplasmin (32, 45) and DF 31 (38) are phosphorylated; (iii) abundance: both nucleoplasmin (9, 42) and DF 31 (38) are among the predominant thermo-stable proteins in Xenopus eggs and oocytes and Drosophila embryos, respectively. Thus, although dNLP is similar to nucleoplasmin at the level of their primary amino acid sequences (Fig. 1B), it is possible that DF 31 may resemble nucleoplasmin more closely than dNLP.
Because of the similar apparent molecular masses of dNLP and p22, it is tempting to suggest that the dNLP cDNA encodes p22, but this postulate is inconsistent with the inability of either native or recombinant dNLP to mediate sperm chromatin decondensation, which is the characteristic biochemical activity of p22 (37). To attempt to clarify this issue, we had subjected dNLP as well as dNLP-containing fractions to Western blot analysis with antibodies that were generated against Drosophila p22 (kindly provided by Dr. P. Fisher, SUNY Stony Brook), but the results were inconclusive (data not shown). Thus, the available data do not suggest a relation between dNLP and p22, although this point would be resolved if the gene(s) encoding p22 were cloned. It should be further noted that the methodology for the purification of dNLP is distinct from that for the purification of p22, and hence, if these proteins are in fact identical, differences in the properties of dNLP and p22 may also be due to some aspect of the treatment or the purity of the proteins.
dNLP Versus dNAP-1-dNAP-1 binds to core histones with higher affinity than dNLP. For instance, when core histones were incubated with dNLP (for 30 min at a dNLP:histone ratio of 1.25:1; under these conditions, the majority of the histones bind to dNLP; see Fig. 4E) and then dNAP-1 was added to the dNLP-histone complexes (at a dNAP-1:dNLP ratio of 1:1), the histones transferred from the dNLP to the dNAP-1 (Fig. 4F). Thus, when dNLP and dNAP-1 are present in excess relative to the core histones, the histones preferentially bind to dNAP-1, even if they are prebound to the dNLP. This preferential binding of histones to dNAP-1 relative to dNLP might also explain why core histones co-immunoprecipitate from a whole embryo extract with dNAP-1, but not with dNLP (Fig. 3B). Moreover, when compared directly, dNAP-1 was more active than dNLP for chromatin assembly (Fig. 2B) as well as for sperm chromatin decondensation (Fig. 5). It seems likely that the greater activity of dNAP-1 relative to dNLP is due to the higher affinity of dNAP-1 for binding to core histones (and, presumably, sperm-specific proteins/protamines) relative to dNLP. Given the complexity of these processes in vivo, however, it is not possible to predict whether the higher affinity of dNAP-1 for histones relative to dNLP necessarily reflects a more significant role for dNAP-1 in chromatin assembly relative to dNLP.
Because of the common ability of both dNLP and dNAP-1 to participate in the ATP-facilitated assembly of nucleosomal arrays, it is reasonable to consider that there is some redundancy in the function of these factors. Consistent with this hypothesis, there is another NAP-1-related protein termed SET in Drosophila as well as in humans (46,47). Moreover, a search of the yeast genome data base revealed a putative NAP-1/SETrelated protein in Saccharomyces cerevisiae, designated "ORF YNL246w" (GenBank accession number Z71522 (1996)), which exhibits 21% amino acid identity with Drosophila SET (data not shown). (On the other hand, a search of the S. cerevisiae genome data base did not reveal any apparent homologues of dNLP or Xenopus nucleoplasmin in yeast (data not shown), which is consistent with the notion that these proteins have an important function in preblastoderm embryos (a stage in development for which there is no yeast counterpart).) This NAP-1/SET-related ORF YNL246w protein may compensate for the absence of NAP-1 and thus be responsible, at least in part, for the observation that NAP-1 is not essential for viability of S. cerevisiae (46). FIG. 7. Western blot analysis of dNLP and dNAP-1 at different stages of Drosophila development. Equivalent amounts of total protein extracts (25 g) derived from Drosophila at the indicated stages of development were subjected to SDS-polyacrylamide gel electrophoresis and Western blot analysis. The upper panel shows the presence of dNLP, as detected with affinity-purified anti-dNLP and 125 I-labeled protein A. The lower panel reveals the presence of dNAP-1, as detected with immunoaffinity-purified anti-dNAP-1 and 125 I-labeled protein A. The relative protein levels (with 0 -3-h embryos as the reference ϵ 1.0) were determined by using a PhosphorImager (Molecular Dynamics). To correct for variation in the total protein concentration in each of the lanes, the value obtained for each test sample (either dNLP or dNAP-1) was normalized to the amount of an ␣-tubulin internal control that was quantitated in a parallel analysis.
In addition, we have found that dNLP and dNAP-1 exhibit distinct subcellular localization in Drosophila cells, i.e. that is, dNLP is present only in the nucleus (Fig. 6), while dNAP-1 is predominantly cytoplasmic during G 2 phase and both nuclear and cytoplasmic during S phase (21). It is thus possible that the relative degree to which dNLP or dNAP-1 (as well as perhaps SET) participates in chromatin assembly may vary with the phase of the cell cycle. Therefore, the available data suggest that dNLP and dNAP-1 may function by parallel and/or complementary mechanisms in the assembly of chromatin in vivo.
Conclusions-We are in the early stages in the analysis of the mechanism of chromatin assembly. For instance, there are assembly factors that have yet to be purified and characterized, such as the ATP-utilizing activity in the dCAF-1 fraction. In addition, the specific functions of known factors, such as NAP-1, dNLP, nucleoplasmin, N1, and CAF-1, remain to be determined. The purification of these factors and the isolation of the genes encoding the proteins are, however, important steps toward addressing these issues. It is our hope that the further analysis of the factors that participate in the assembly of chromatin will provide new and significant insight into a fascinating and important biological process.