Multiple ISWI ATPase Complexes from Xenopus laevis

The nucleosomal ATPase ISWI is the catalytic subunit of several protein complexes that either organize or perturb chromatin structure in vitro. This work reports the cloning and biochemical characterization of a Xenopus ISWI homolog. Surprisingly, whereas we find four complex forms of ISWI in egg extracts, we find no functional homolog of NURF. One of these complexes, xACF, consists of ISWI, Acf1, and a previously uncharacterized protein of 175 kDa. Like both ACF and CHRAC, this complex organizes randomly deposited histones into a regularly spaced array. The remaining three forms include two novel ISWI complexes distinct from known ISWI complexes plus a histone-dependent ATPase complex. This comprehensive biochemical characterization of ISWI underscores the evolutionary conservation of the ACF/CHRAC family.

In Xenopus laevis, early development proceeds from fertilization through a series of rapid embryonic cleavages in the absence of zygotic transcription (35)(36). The structural components and enzymatic machinery for embryonic chromatin assembly are stored in the egg (37), representing a rich biochemical source. Historically, Xenopus egg (38 -40) and oocyte extracts (41) have been utilized to assemble spaced chromatin in vitro for studies of transcription (42)(43) and DNA repair (44). They contain activities competent for both spacing of nucleosomal arrays (45) and mononucleosome disruption (46). We have utilized classical biochemical techniques for a survey of ISWI from these extracts. This report outlines the purification of four ISWI complexes. Consistent with the demands for chromatin assembly during early Xenopus development, an ACF/CHRAC functional homolog was the most abundant form identified. In addition, we report the surprising absence of detectable quantities of NURF.

EXPERIMENTAL PROCEDURES
Isolation of an ISWI cDNA-RT-PCR 1 was performed under standard conditions (47) with Xenopus ovary RNA as template using the following oligonucleotides: sense, 5ЈCCCAAGCTTCATATGAACTA(T/ C)GC(A/T)GTGGATGCCTA(C/T)TT3Ј; antisense, 5ЈAAGCTTCTCGAG-TTCTTCATATAC(A/G)TT(C/T)TC(C/T)CT(A/G)TCAAA3Ј. The resulting PCR product was cloned into pGEM-T (Promega) and sequenced. The RT-PCR product was used as a probe for screening an oocyte cDNA library using standard methods (47). Two overlapping clones were isolated covering the entire ISWI open reading frame. These clones were sequenced on both strands using an ABI Prism 377 automated DNA sequencer.
Production of Anti-ISWI Antiserum-The RT-PCR clone was excised from pGEM-T with NdeI and XhoI and cloned into the same sites of * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. pET21a (Novagen). Protein production and purification under denaturing conditions were performed according to the manufacturer's instructions (Novagen). Antibodies were produced in New Zealand White rabbits using a standard immunization protocol (48).
Purification of ISWI Complexes-Egg extract was prepared and fractionated through Bio-Rex 70 and MonoQ essentially as described (49). Eggs were obtained from mature female X. laevis and extracted by the crushed cell method (49). The resulting extracts were applied to Bio-Rex 70 (Na ϩ form, Bio-Rad) previously equilibrated in Buffer A (100 mM) (20 mM HEPES, pH 7.5, 5 mM KCl, 1.5 mM MgCl 2 , 1 mM EGTA, 10 mM ␤-glycerophosphate, 0.5 mM dithiothreitol, 10% glycerol, 0.2 mM phenylmethylsulfonyl fluoride, 1 g/ml leupeptin, numbers in parentheses indicate millimolar concentration of NaCl) at a ratio of 10 mg of protein per ml packed bed volume. The column was washed with 4 volumes of Buffer A (100) and eluted with 3 volumes of Buffer A (500). Fractions containing approximately 80% of the eluted protein from the Buffer A (500) elution were pooled and dialyzed versus Buffer A (0) to a conductance equivalent to 100 mM NaCl. The protein pool was cleared by centrifugation and applied to a MonoQ 10/10 column previously equilibrated in Buffer A (100). The column was washed with 5 volumes of Buffer A (100) and eluted with a linear gradient from Buffer A (100) to A (500) in 20 column volumes. Fractions were analyzed for ISWI content by immunoblot.
The ISWI-A peak was further purified by sedimentation on a 5-20% sucrose gradient in Buffer A (49) for 28 h at 41,000 rpm in an SW 41 rotor.
The ISWI-B, ISWI-C, and ISWI-D complexes were each purified through gradient elution on MonoS and heparin-Sepharose. Peak fractions from the MonoQ column were dialyzed versus Buffer A (0 mM) to a conductance equivalent to 100 mM NaCl and applied to a MonoS 5/5 column previously equilibrated in Buffer A (100). The column was washed with 5 column volumes Buffer A (100) and eluted with a linear gradient from Buffer A (100) to A (500) in 20 column volumes. Fractions were analyzed for ISWI content by immunoblot. Elution positions for the individual ISWI complexes were as follows: ISWI-B, 360 mM NaCl; ISWI-C, 350 mM NaCl, ISWI-D, 330 mM NaCl. Each ISWI MonoS pool was dialyzed versus Buffer A (0 mM) containing 0.1% Triton X-100 to a conductance equivalent to 100 mM NaCl. Protein samples were cleared by centrifugation and applied to 1-ml heparin-Sepharose columns (Hi-Trap columns, Amersham Pharmacia Biotech) previously equilibrated in Buffer A (100) plus 0.1% Triton X-100. Columns were washed with 5 volumes A (100) and eluted with linear gradients from Buffer A (100) to Buffer A (1000) in 20 column volumes, buffers contained 0.1% Triton X-100. Elution positions for the respective ISWI complexes were as follows: ISWI-B, 540 mM NaCl; ISWI-C, 570 mM NaCl; ISWI-D, 540 mM NaCl.
ATPase assays were performed as described (49,50). Octamer Mobilization and Spacing Assays-Nucleosome reconstitution and histone octamer mobilization experiments were performed as described (50) using the quantities indicated in the figure legends.
Chromatin assembly and spacing assays were performed as described (45). In the indicated reactions, 1 l of anti-ISWI antiserum or preimmune serum from the same rabbit was included. Additional spacing assays with salt-dialyzed chromatin were performed essentially as described (25).

Cloning of Xenopus ISWI and Fractionation into Multiple
Complexes-We began our studies of ISWI in Xenopus by cloning an ISWI cDNA from an oocyte library. A single class of cDNA coding for a protein of 1037 amino acids was isolated (GenBank TM accession number AF292095). Comparison of the predicted amino acid sequence of the Xenopus cDNA (see Fig. 1) revealed closest similarity to the human ISWI homolog hSNF2H (51). The Xenopus protein contains the conserved sequence blocks diagnostic of the SNF2 superfamily (52) in addition to considerable sequence identity with both human ISWI homologs in the carboxyl-terminal portion of the protein including the SANT domain (53). We utilized this highly conserved carboxyl-terminal region to elicit antibodies (amino acids 724 -955). The antibody detects a protein of the predicted size in egg and oocyte extracts (data not shown); we have subsequently utilized it as a tool for the purification of ISWI from egg extracts.
The first two steps of ISWI purification are outlined in Fig.   2A. All the detectable ISWI bound the initial cation exchange column and was eluted at moderate salt. Surprisingly, gradient elution of the 0.5 M Bio-Rex fraction on MonoQ reproducibly yielded four distinct peaks of ISWI by immunoblot that we provisionally named ISWI-A, -B, -C, and -D (see Fig. 2B). The ISWI-C peak was the most abundant on the MonoQ gradient. RbA p48/p46, a subunit of the Drosophila NURF complex (54), did not copurify with ISWI at this step. In fact, the prominent RbA p48/p46 peak on the MonoQ gradient is associated with the Xenopus Mi-2 complex (49). ATPase activity assays of the gradient fractions revealed a major peak of DNA-stimulated ATPase activity coincident with the peak of RbA p48/p46 in fractions 31-33. Although we do not currently know the identity of this ATPase, it is not associated with either Mi-2 or ISWI (data not shown (49)). Surprisingly, the ISWI-A peak coincided with ATPase activity stimulated by free histones. ISWI-D, as expected, copurified with nucleosomestimulated ATPase activity. The ISWI-B and -C complexes were associated with minor peaks of nucleosome-stimulated ATPase, although their activities were partially obscured by the robust DNA-stimulated ATPase in this region of the gradient.
Purification of a Histone-dependent ATPase Complex-ISWI-A was further purified by sedimentation on a sucrose gradient. Histone-stimulated ATPase activity was found in fractions 22-24 (see Fig. 3A) whereas immunoblots revealed ISWI in fractions 23-25 (Fig. 3B). ATPase activity precisely cosedimented with three polypeptides of 200, 21, and 15 kDa at a position consistent with a molecular mass of 250 kDa ( Fig. 3A and data not shown). Although ISWI protein closely cosedimented with histone-dependent ATPase activity, it was not present at comparable stoichiometry to p200, p21, and p15. We are currently investigating the identity of p200, p21, and p15 in order to determine whether their copurification with ISWI is functionally relevant.
To determine the relationship between these three polypeptides and ATPase activity, we performed a cross-linking assay using a radiolabeled ATP analog (8-azido-[␥-32 P]ATP). The 200-kDa polypeptide cross-linked to ATP in the presence of UV light in a reaction stimulated by core histones (see Fig. 3D). As histones represent the vast majority of protein in the reaction (roughly 5-6 orders of magnitude more abundant than ISWI-A), it is not surprising that they are nonspecifically crosslinked. We conclude that p200 represents a candidate ATPase stimulated by free histones. To our knowledge, this is the first description of an ATPase stimulated solely by free histones and not by DNA or nucleosomes. The biological relevance of this activity is currently unknown.
ISWI-C Represents a Functional Homolog of ACF-ISWI-C was purified through the chromatographic steps summarized in Fig. 4A. Nucleosome-stimulated ATPase activity coeluted with ISWI in fractions 31-34 of the final column (data not shown). Three polypeptides of 185, 175, and 135 kDa, respectively, were present in the purified complex (see Fig. 4A), and higher percentage gels fail to reveal additional subunits (data not shown). This polypeptide composition is essentially identical to that reported for the Drosophila ACF complex (35)(36). We confirmed the identity of the 135-kDa subunit as ISWI using immunoblots and by mass spectrometry (data not shown).
Peptide sequence from the 185-kDa protein yielded a single peptide (KEREEAAARIRK) with an exact match to a human EST clone (GenBank TM accession number AA112166). This EST clone was subsequently found to be contained in its entirety within a protein recently described by different groups as either WCRF180 (29), BAZ1A (55), or hAcf1 (27). We conclude that ISWI-C, based on the similarity in polypeptide profile as well as the chemical identification of p185 as a sequence relative of human and Drosophila Acf1, represents a new member of the ACF-CHRAC complex family.
The primary difference between the Xenopus and Drosophila ACF complexes and the recently described WCRF and hACF complexes is the presence of a p175 species in the Xenopus and Drosophila ACF complexes. The p175 component of Drosophila ACF has been identified as similar or identical to the Acf1 protein based on peptide chemistry and on immunoreactivity (26). We have compared peptide maps obtained by mass spectrometry of Xenopus p185 and p175. We find, in contrast to Drosophila ACF, that these two proteins are distinct (data not shown). We are currently pursuing the identity of the Xenopus p175 protein; there are no candidate sequences in the current data bases.
To assess the ability of Xenopus ACF (hereafter referred to as xACF) to perturb histone-DNA interactions, we utilized an octamer mobilization assay (reviewed in Ref. 56) previously used for Drosophila NURF and CHRAC (57)(58). Nucleosomes were assembled by salt dialysis on a 250-base pair fragment of the Xenopus TR␤A promoter (59). The resulting uniquely positioned nucleosomes have been characterized by micrococcal nuclease mapping (50), and the positions are summarized in cartoon form in Fig. 4B. Following isolation of the unique position termed N2 by native polyacrylamide gel electrophoresis, we incubated this nucleosome with purified xACF complex. Recombinant Drosophila ISWI served as a positive control FIG. 1. Xenopus ISWI is homologous to hSNF2H. The deduced amino acid sequence of Xenopus ISWI (xISWI) was aligned manually with the amino acid sequences of hSNF2H, hSNF2L, and Drosophila ISWI (dm ISWI). The boxed regions numbered I-VI represent sequence blocks characteristic of the SNF2 superfamily (52). The conserved SANT domain (53) is boxed and labeled. (57)(58). As previously reported (57)(58), recombinant ISWI shifted the position of the histone octamer to a new position, namely to the end of the DNA fragment (see Fig. 4C). In contrast, purified xACF shifted the octamer position toward the center (see Fig. 4C). This octamer mobilization required ATP hydrolysis because a non-hydrolyzable ATP analog did not promote this reaction nor did GTP. Additionally, as the reaction did not result in the accumulation of free DNA (see Fig.   FIG. 2. Xenopus egg extracts contain multiple forms of ISWI. A, the fractionation scheme from egg extract is summarized in the diagram. The inset immunoblot depicts the elution of ISWI on the Bio-Rex 70 column. B, the panels depict immunoblots of selected fractions from the gradient elution on MonoQ. The individual ISWI peaks are indicated in the figure. C, ATPase assays were performed as described under "Experimental Procedures" using free DNA, purified core histones, or chicken erythrocyte mononucleosomes as substrates. The graph depicts picomoles of ATP hydrolyzed as a function of fraction number.

FIG. 3. A novel histone-dependent ATPase activity copurifies with a three-polypeptide complex.
A, histone-stimulated ATPase activity of sucrose gradient fractions is depicted in the graph. ATPase assays were performed as described under "Experimental Procedures." B, an immunoblot of selected sucrose gradient fractions was probed with anti-ISWI antibody. C, aliquots of the indicated sucrose gradient fractions were resolved on SDS-PAGE (4 -20% acrylamide gradient) and stained with Coomassie Blue. Arrows indicate the 200-, 21-, and 15-kDa proteins cosedimenting with ATPase activity. D, an ATP crosslinking assay was performed with 8-azido-ATP. Two microliters of sucrose fraction 23 were incubated with or without core histones and 8-azido-ATP for 30 min at room temperature. Cross-linking was initiated with a Stratalinker, and samples were resolved in 4 -20% gradient SDS gels that were subsequently dried and exposed to x-ray film. Arrowheads on the right indicate the migration of DNA fragments resulting from the production of mono, di, tri, tetra, and penta-nucleosomal particles. 4C), octamer movement must occur in cis. The Drosophila CHRAC complex mobilizes nucleosomes in a similar manner, albeit on an unrelated DNA fragment (58). The Drosophila NURF complex, on the other hand, moves octamers bidirectionally on DNA and prefers, on a fragment of the Drosophila hsp70 promoter, to reposition histones near the end of the fragment (57). Recombinant Drosophila ISWI preferentially repositions the octamer at the fragment end (57)(58).
As Drosophila ACF was initially characterized as an energydependent chromatin assembly factor (25-26), we have investigated the ability of Xenopus ACF to organize nucleosomal arrays. We first utilized a chromatin assembly system from Xenopus oocyte extracts consisting of a partially purified H3⅐H4-N1⅐N2 complex, purified H2A/H2B heterodimers, and spacing activity provided by either crude oocyte extracts or the previously defined spacing fraction (45). As seen in Fig. 4D, this reconstituted system directs the assembly of a nucleosomal array with physiologic spacing (see 1st and 2nd lanes). The addition of anti-ISWI antibody, but not preimmune serum, results in a loss of chromatin organization. The spacing fraction can rescue the ISWI-depleted reaction (Fig. 4D), and immunoblot analysis reveals that ISWI is present in the spacing fraction (data not shown). Like the original spacing fraction, Xenopus ACF is capable of rescuing the spacing and assembly properties of the ISWI-depleted reaction (see Fig. 4D). Drosophila ACF can reorganize nucleosomes deposited by salt gradient dialysis (25). We tested xACF for similar activity in a system consisting solely of purified core histones deposited by salt dialysis and purified xACF1. Treatment of the resulting minichromosomes with xACF1 resulted in a regularly spaced nucleosomal array with up to five or six nucleosomes visible following micrococcal nuclease digestion (Fig. 4E). In addition, octamer spacing on the xACF-treated minichromosomes changes from a close-packed to a uniformly spaced configuration in an ATP-dependent manner (Fig. 4E).
ISWI-B and ISWI-D Represent Novel ISWI Complexes-ISWI-B was purified through the chromatographic steps outlined in Fig. 5A. ISWI copurified with a single polypeptide of 200 kDa (Fig. 5A) and with nucleosome-stimulated ATPase activity in the final purification step, gradient elution on heparin-Sepharose (data not shown). Sedimentation analysis indicated a molecular mass of approximately 500 kDa (data not shown). Peptide mapping of p200 by tryptic digestion and mass spectrometry failed to generate matches to any proteins in the current data base (data not shown).
The final peak, ISWI-D, was purified through chromatography on MonoS and heparin (see Fig. 6A). A sharp peak of nucleosome-stimulated ATPase activity exactly coeluted with ISWI on the final purification step (data not shown). Polypeptides of 200, 135, and 70 kDa copurified in the ISWI-D complex (shown in Fig. 6A). However, electrophoresis of a greater quantity of protein on higher percentage gels revealed the presence of additional polypeptides of 55 and 17 kDa (see Fig. 6A). Peptide mapping for the p200, p70, and p55 ISWI-D polypeptides failed to generate matches in the current data bases. The p200 polypeptide found in the ISWI-B complex is distinct from the ISWI-D p200 protein by mass spectrometry (data not shown). Additionally, mass spectrometry rules out copurification of ISWI with Xenopus topoisomerase II as initially reported for Drosophila CHRAC (24). We were also unable to detect copurification of ISWI with topoisomerase II in any of these complexes by immunoblot analysis (data not shown) using ␣-Xenopus topoisomerase II (60).
We have also examined the capacity of the ISWI-B and ISWI-D complexes to mobilize histone octamers in our standard mobilization assay. Like xACF, both ISWI-B and ISWI-D move histone octamers in an ATP-dependent fashion (see Fig.  5B and Fig. 6B). Additionally, both complexes, like xACF, reposition octamers toward the center of the DNA fragment (see Fig. 6B). Comparison of the mobilization activities of these three complexes failed to reveal significant qualitative or quantitative differences.

DISCUSSION
This report describes the first systematic characterization of all major forms of ISWI from a vertebrate, X. laevis. The ISWI protein presented considerable biochemical complexity, fractionating into four forms with distinct biochemical properties and subunit composition. In Saccharomyces cerevisiae, the two ISWI genes form only two types of protein complex (32). In Drosophila embryos, three complexes containing ISWI have been described (reviewed in Refs. 12, 19, and 20), but two, ACF and CHRAC, are clearly related (26,28). Undoubtedly, the developmental program of Xenopus, with an early period of rapid nuclear divisions, places a special burden on the machinery for chromatin assembly. The multiple ISWI complexes described here may reflect those specialized demands. Human cells, Xenopus eggs, and Drosophila embryos all contain ISWI complexes featuring an Acf1 homolog (26 -30). In addition, S. cerevisiae contains a homolog (YGL133w) suggesting that an ACF-like complex also exists in this organism. ACF is thus likely to represent a fundamental activity in eukaryotic nuclei. The differences in polypeptide composition found in the various members of the ACF/CHRAC family likely represent variations on a common theme. As depicted in Fig. 7, ISWI and Acf1 associate to form a common structural and functional module. Functionality, such as differential partitioning within the genome, may be added to this module by association with accessory proteins like the histone fold proteins of CHRAC or p175 of xACF. Elucidation of the precise variations in composition and function for this interesting family of chromatin-dependent ATPases should provide important clues into mechanisms for assembly of specialized chromatin structures.
We have initiated an attempt to understand the functions of the Xenopus ISWI complexes by defining their nucleosome mobilization properties. Each complex, regardless of polypeptide composition, mobilized histone octamers toward the center of the DNA fragment like the Drosophila CHRAC complex (58) and unlike the Drosophila NURF complex (57). This repositioning of histones relative to DNA sequence required a hydrolyzable form of ATP, operated at moderate monovalent and divalent cation concentrations, and occurred in cis on a single DNA fragment. Mechanistically, we favor the model recently proposed by Wu and colleagues (57), which stipulates that nucleosome motion occurs in small, discrete steps. Central to this proposal is the notion that ATP hydrolysis by the ISWI ATPase leads to lowered activation energy for short translational sliding by the histone octamer. The NURF complex, based on the bidirectional character of its mobilization activity, is predicted to generate a relatively unrestricted random walk of histones to a thermodynamically favorable position (57). In marked contrast, the CHRAC complex (58) and all three Xenopus complexes exhibit strong directionality in their mobilization activity. We propose that proteins other than ISWI in the Xenopus  ISWI complexes and in CHRAC restrict the potential locations available for octamer motion. This restriction serves functionally to maximize the amount of DNA protruding from each end of the nucleosome. This hypothesis predicts that proteins in the Xenopus ISWI complexes other than ISWI interact with extranucleosomal DNA. Activity of this nature on a mononucleosome is also predicted to translate into spacing in the context of a nucleosomal array where adjacent nucleosomes rather than DNA ends provide the limits for available positions. The definition, in molecular terms, of the remaining ISWI-associated proteins described in this work should prove valuable in the mechanistic determination of histone octamer mobilization by this class of enzymes as well as their roles in influencing chromatin structure and DNA function in vivo.