F-actin-dependent insolubility of chromatin-modifying components.

Many complexes involved in chromatin modification are difficult to isolate and commonly found associated with nuclear matrix preparations. In this study, we examine the elution properties of chromatin-modifying components under different extraction conditions. We find that most, but not all, histone acetyltransferases and histone deacetylases predominantly partition with the nuclear pellet during intermediate salt extraction. In attempts to identify a biological basis for the observed insolubility, we demonstrate that depolymerizing cellular actin, but not cellular tubulin, mobilizes a significant proportion of the insoluble pool into the intermediate salt-soluble nuclear extract. The disruption of cellular F-actin releases a specific subset of high molecular weight, active, nuclear histone deacetylase complexes that are not found under normal conditions. This study demonstrates that actin polymerization, a physiologically relevant process, is responsible for the observed insolubility of these components during nuclear extract preparation and establishes a simple strategy for isolating subsets of chromatin-modifying complexes that are otherwise depleted or absent under the same extraction conditions.

Many of the individual components involved in the modification of chromatin structure have now been identified biochemically (1)(2)(3)(4)(5). The current challenge lies in isolating and characterizing these complex, macromolecular assemblies in a form that is representative of the in vivo chromatin-modifying machines (6 -12). Nuclear proteins were originally isolated by salt extraction of hypotonically isolated nuclei (13). These procedures were specifically developed to generate RNA polymerase II transcription-competent nuclear extracts with the advantage of isolating the soluble, initiation-competent, hypophosphorylated pool of RNA polymerase II while the hyperphosphorylated, initiation-incompetent form remains insoluble (14). This procedure quickly became the standard methodology for releasing functional nuclear complexes from isolated nuclei. Although this technique has successfully isolated many soluble nuclear complexes with definable activities, other experiments indicate that this extract may only contain a subset of the functional nuclear complexes in the nucleus.
Our understanding of nuclear structure and organization has undergone a remarkable transformation from the sealed bag of chromatin bathed in a nucleoplasm rich in regulatory and catalytic proteins to the realization that the nucleus is a structure that compartmentalizes functionally related molecules into distinct domains (15). For example, several proteins involved in chromatin modification or transcriptional activation, including HDACs, 1 localize to several hundred discrete foci dispersed throughout the interchromatin space (16). The size and intensity of these foci is a reflection of tens to hundreds of individual HDAC molecules accumulating in small domains within the interphase nucleus. Notably, these domains do not correspond to regions where histone proteins or DNA are detected (16), illustrating that chromatin-modifying proteins may be found in complexes other than those directly associated with the chromatin. Most gene regulation and chromatin-modifying activities are thought to be mediated by large, multiprotein complexes. For instance, many purified HDAC complexes approach or exceed 1 MDa in total mass (6 -8, 17-22). Due to the nature and extreme size of some of these complexes, it is not surprising that the individual nuclear proteins are frequently difficult to extract (17,23). To date, the problem of effectively isolating nuclear proteins has largely been set aside, and the association with nuclear matrixes has been assumed to be "high salt-induced" protein precipitation (24). Insolubility during subcellular fractionation is also a feature that is common to proteins that comprise or associate with the cytoskeleton. In this context, it is notable that actin and tubulin have both been found to be present within the interphase nucleus (25)(26)(27)(28).
The role of actin in the nucleus has been controversial at best. Findings that nuclear matrix preparations contained actin (29,30) were originally dismissed as preparation artifacts. Actin is likely small enough to passively gain entry to the nucleus through the nuclear pore. This, along with the lack of phalloidin binding to actin to indicate the presence of any F-actin in the nucleus, added to the skepticism of nuclear actin having any significance. However, in the past several years, there has been a resurgence of interest in nuclear actin. There is now substantial evidence to suggest that actin plays an integral role in several nuclear functions as well as being associated with the nuclear matrix (25,28,31).
Microscopic and ultrastructural analysis have suggested that actin is present in a filamentous form in the nucleus (32)(33)(34), and the use of F-actin-disrupting or -stabilizing agents has suggested that at least a portion of the actin in the nucleus is functioning in a filamentous form (35)(36)(37). In addition, we recently discovered that nuclear actin is in a dynamic equilib-rium between monomeric and polymeric forms within living cells. 2 It is generally accepted that chromatin-modifying complexes contain actin, Arps, and/or actin-binding proteins (11, 37, 39 -47). The yeast SWI/SNF chromatin remodeling complex was found to contain Arps 7 and 9 (40,44), and similarly, the Arp, BAF53, and ␤-actin have been identified bound to Brg1 of the human BAF complex (37). Interestingly, actin binding to Brg1 is required for stable association of the complex and provides a link between the chromatin remodeling complex and the nuclear matrix (37). BAF53 and ␤-actin have also been identified as subunits of the human Tip60 HAT complex (42), whereas Act3/Arp4 and actin are components of the yeast Nu4A HAT complex (41). Isoforms of the HDAC-containing complex, N-CoR, contain Brg1 from the mammalian BAF complex (11) or a novel actin-binding protein, IR10 (47). These findings all highlight association with F-actin as a potential physiological mechanism for retaining chromatin-modifying complexes in an insoluble form during nuclear extraction.
In this study, we analyze the efficiency of the conventional salt extraction of hypotonically isolated nuclei (13,14) by monitoring the solubility of various histone-modifying enzymes. We found that, with the exception of HDAC4, the conventional salt extraction does not efficiently solubilize the HDACs or HATs examined. This finding is consistent with previous reports that these proteins are retained in nuclear matrix preparations (48,49). Studies examining the association of HDAC activity with nuclear pellets have also demonstrated that the activity was retained in a manner that was independent of DNA/chromatin association (48). Therefore, we chose to examine whether actin or tubulin played a role in the partitioning of HATs and HDACs during nuclear extraction. We found that disruption of F-actin, but not microtubules, results in a dramatic increase in the solubility of several HAT and HDAC proteins. The nuclear extract obtained from F-actin-depleted cells was rich in high molecular weight, HDAC-containing complexes, and the elution profile of HDAC activity from a molecular-sizing column revealed dramatic differences in the composition when compared with that of the control nuclear extract. We have both provided a simple method to isolate novel nuclear HDAC and HAT complexes and identified a biological mechanism that may provide a link between subnuclear compartmentalization and functional regulation by chromatin-modifying complexes.

EXPERIMENTAL PROCEDURES
Nuclear Extraction-HeLa or HeLa S3 cells were subcultured the previous day and used at 70 -80% confluency on the day of the experiment. Cells were incubated in serum-free Dulbecco's modified Eagle's medium with or without 20 M latrunculin B (Calbiochem) or 5 M nocodazole (Sigma) for 60 min at 37°C, 5% CO 2 and then harvested using 0.53 mM EDTA in phosphate-buffered saline and washed once with cold phosphate-buffered saline. Nuclear extracts were prepared as per the procedure described by Dignam et al. (14) with slight modifications (13,14). All steps were carried out on ice or at 4°C unless stated otherwise. Protease inhibitors and a reducing agent were added to each buffer just prior to use (1 mM dithiothreitol, 1 mM phenylmethylsulfonyl fluoride, 10 g/ml aprotonin, 10 g/ml leupeptin). Briefly, cells were incubated in 5 volumes of hypotonic buffer A (20 mM Hepes, pH 7.9, 1.5 mM MgCl 2 , 10 mM KCl) on ice for 15 min. Cells were centrifuged at 450 ϫ g and resuspended at 1.5 ϫ 10 6 cells/ml in buffer A ϩ 300 mM sucrose. Cells were homogenized using a Dounce homogenizer until Ͼ90% of the cells were disrupted as seen by trypan blue dye uptake under a light microscope. Nuclei were recovered by centrifugation at 3000 ϫ g for 15 min, and the supernatant was kept as the cytoplasmic extract. The nuclei were washed once using nuclei wash buffer (10 mM Hepes, pH 7.9, 0.2 mM MgCl 2 , 10 mM KCl). Washed nuclei were extracted using buffer C (20 mM Hepes, pH 7.9, 25% glycerol, 420 mM NaCl, 0.2 mM EDTA, 1.5 mM MgCl 2 ) for 30 min on ice. Insoluble material was removed by centrifugation at 21,000 ϫ g for 10 min. The supernatant was used as the nuclear extract. The insoluble material was kept, resuspended directly in 3ϫ SDS-PAGE sample loading buffer and sonicated to facilitate resuspension.
Western Blots and Dot Blots-Cytoplasmic and nuclear extracts as well as the insoluble cellular material were separated on 10% SDS-PAGE and transferred to 0.2-M polyvinylidene difluoride blotting membrane according to standard protocols. The distribution of various nuclear proteins was examined by immunoblotting following standard procedures using 5% low fat milk in TBST (Tris-buffered saline with 0.05% Tween) as the blocking buffer and antibody incubation buffer. Primary antibodies used are as follows: anti-TAF II 250 (catalog number sc-735, Santa Cruz Biotechnology); anti RNA polymerase II (kind gift from C. Spencer); anti-PCAF (catalog number sc-8999, Santa Cruz Biotechnology); anti-MYST family (catalog number 06-815, Upstate Cell Signaling Solutions); anti-HDAC2 (catalog number sc-7899, Santa Cruz Biotechnology); polyclonal anti-HDAC1, anti-HDAC3, and anti-HDAC4 (kind gifts from W. Fischle); anti-p62 (catalog number 06-513, Upstate Cell Signaling Solutions); anti-CBP-NT (catalog number 06-297, Upstate Cell Signaling Solutions); anti-SF2/ASF (catalog number 32-4600, Zymed Laboratories Inc.). The primary antibody was followed by incubation with the appropriate secondary antibody conjugated to horseradish peroxidase or Cy3 (Jackson ImmunoResearch Laboratories). Antibody binding was detected using enhanced chemiluminescence (ECL plus, Amersham Biosciences) and autoradiography or by fluorescent imaging using the Typhoon 9400 variable mode imager (Amersham Biosciences), respectively. Immunoblots were performed in triplicate and analyzed by densitometry. The increase in solubility in response to latrunculin treatment was determined by adding the densitometry values representing the amount of protein detected in the cytoplasmic and nuclear soluble fractions and dividing the sum obtained for the latrunculin extract by that for the control extract to give a -fold increase in solubility.
Chromatography-Proteins were separated by gel filtration chromatography using the SMART™ FPLC system (Amersham Biosciences) equipped with a Superose 6 PC 3.2/30 column (Amersham Biosciences). A nuclear extract from untreated or latrunculin B-treated cells (see above), equivalent to ϳ6 ϫ 10 6 cells, was concentrated using a Biomax centrifugal filter with a 10,000 molecular weight (Millipore). The column was extensively washed and pre-equilibrated with column buffer (20 mM Hepes, pH 7.9, 100 mM KCl, 0.2 mM EDTA) before loading the concentrated nuclear extract. Proteins were eluted using column buffer at a flow rate of 40 l/min, and 50 -l fractions collected. The Superose 6 PC 3.2/30 column has an exclusion limit of 4 ϫ 10 7 .
Fluorescent Activity Assay-HDAC activity present in the Superose 6 column fractions (described above) was determined using the HDAC Fluor de Lys TM (fluorescent activity assay (Biomol)) according to the manufacturer's protocol. Briefly, 10 l of each Superose 6 column fraction indicated was incubated in assay buffer containing 400 M Fluor de Lys TM substrate with or without 1 M trichostatin A for 60 min in a white microtiter plate. Activity for each fraction was measured in triplicate. HDAC activity was stopped by the addition of the Fluor de Lys TM developer and 1 M (final) trichostatin A and allowed to incubate for 10 min. The samples were then read using a CytoFluor™ II microplate fluorescence reader (Biosearch Inc.) (excitation 360 nm, emission 460 nm).

Solubility of Various HATS and HDACs-
The most common approach for isolating nuclear proteins from tissue culture cells is to employ mechanical disruption of cells swollen under hypotonic conditions followed by isolation and extraction of the released, intact nuclei (13,14). The isolation of the nuclei is easily achieved due to their ability to sediment at relatively low centrifugal forces. Following isolation, extraction is typically done with sodium or potassium chloride at concentrations ranging from 0.35 to 0.5 M. This procedure is extensively used to isolate and maintain nuclear proteins in their active or functional state (13,14), but the extent to which this extract represents the activities responsible for modifying chromatin in vivo is unknown.
To determine the efficiency of common extraction agents, we compared intermediate salt, non-ionic detergent, and high salt/ urea for their ability to extract nuclear proteins that partici-pate in reversible acetylation. These procedures included one whole cell lysis technique using a detergent-based extraction buffer (1% Nonidet P-40) as well as two hypotonic cell lysis protocols that extract the isolated nuclei with very different buffers (0.42 M NaCl (14) versus 2 M NaCl, 3 M urea). Soluble and insoluble fractions generated from each procedure were separated by SDS-PAGE and analyzed by immunoblotting for various HDAC and HAT proteins (Fig. 1). We found that the 0.42 M NaCl extraction is particularly poor at extracting HDAC1 and HDAC3. When the amounts of HDACs 1 and 3 are compared between the soluble (indicated by C and NS) and insoluble (indicated by NP) fractions recovered from the 0.42 M NaCl extraction, the majority of the proteins remain insoluble (Fig. 1A). As may be predicted from the enrichment of these proteins in nuclear matrix preparations (48,49), the lower salt conditions on average only solubilize 10 and 15% of HDAC1 and HDAC3, respectively. The presence of high salt and urea effectively extracted all of the detectable HDAC1 and -3 from purified nuclei, whereas non-ionic detergent was equally effective from whole cells (Fig. 1, B and C, respectively). Similar results were obtained for the HAT proteins examined (Fig. 1). Table I summarizes the observations for the nuclear proteins examined with respect to HDAC1. Generally, we found that 0.42 M salt extraction of purified nuclei releases only a small subset of the nuclear protein pool for the class I HDACs and HATs examined. CBP and the MYST family of HAT proteins along with the HAT/transcription factor, TAF II 250, are slightly more soluble than HDAC1 but still show significant resistance to extraction using 0.42 M NaCl. Notably, this was not true of all of the proteins that we examined. The class II HDAC, HDAC4, is more effectively solubilized by the 0.42 M NaCl extraction procedure.
The Role of Architectural Proteins in Nuclear Protein Solubility-The use of agents that disrupt protein-protein interactions allows for the efficient extraction of the HAT and HDAC proteins. Unfortunately, these procedures are not suitable for maintaining intact, active chromatin-modifying complexes. The established retention of HDAC and HAT activity in nuclear matrix preparations (17,48,49) demonstrates that chromatin binding is not the basis of the observed insolubility. While attempting to generate extracts that are rich in the native form of these proteins, we postulated that a biological mechanism exists that contributes to the retention of these proteins in nuclear pellets.
Both intermediate filament proteins, tubulin and actin, have been reported within the nucleus (50,51), and increasing evidence suggests that actin may be specifically involved in several nuclear functions (25,28,31). ␤-Actin is a component of at least some chromatin-modifying complexes (37,41,42,45) and provides a link for at least one of these complexes to the nuclear matrix (37). Therefore, actin is an excellent candidate protein to be involved in the observed insolubility of chromatin-modifying components in the conventional salt extraction of isolated nuclei. To test this hypothesis, cells were treated with latrunculin, an F-actin-disrupting drug, prior to extraction of nuclear proteins using the conventional salt extraction method described (see "Experimental Procedures"). The use of latrunculin dramatically increased the proportion of protein found in the soluble fractions when compared with untreated controls for several of the HDACs and HATs examined (Fig. 2, C and NS versus NP). Aside from being associated with histone-modifying complexes, actin has been found to bind to pre-mRNA molecules (52,53) and heterogeneous ribonucleoproteins (53)(54)(55). It is suggested that actin may be involved in transcription at the level of elongation (54) and with the transport of newly synthesized RNA molecules (35,53). For comparison with the basal transcription machinery, we examined the solubility of RNA polymerase II and the p62 subunit of TFIIH and found that disruption of cellular actin did increase their solubility as well (Fig. 2). This result was not entirely unexpected considering that the salt extraction of isolated nuclei was originally designed to selectively solubilize only the inactive but initiation-competent RNA polymerase II fraction. Although latrunculin treatment consistently improved the efficiency of 0.42 M NaCl-mediated elution of most of the proteins examined, there were substantial differences in the extent of the release of these proteins into the soluble nuclear extract (Fig. 2). For example, the -fold increase in solubility for HDAC1, -2, and -3 is 14.2 Ϯ 2.0, 1.6 Ϯ 0.14, and 7.8 Ϯ 0.12, respectively. The increase in solubility for CBP ranged from 5.6 Ϯ 0.68-to 8.8 Ϯ 4.7-fold, depending on which isoform was examined. We also found examples of both soluble and insoluble proteins in which the solubility was not altered. The partitioning of HDAC4 and PCAF between the subcellular fractions was not significantly affected by latrunculin treatment (data not shown).
Latrunculin treatment causes disassembly of the actin cytoskeleton. This raises the possibility that the observed increase in solubility may be due to a general loss of structural stability within the cell. The question then arises as to whether the increase of HDAC proteins detected in extracts from latrunculin-treated cells is in fact an increase in solubility or due to an influx from the cytoplasm to the nucleus as a result of diminished cell integrity. We believe that the latrunculin-de-  pendent effects observed are in fact a change in solubility. HDAC1 and -2 along with the HAT proteins we examined are known to be exclusively nuclear in their localization, whereas HDAC3 and -4 are both known to be present in both the cytoplasm and the nucleus. Of the HDACs, only HDAC1, -2, and -3 are affected by the cytoskeletal disruption. The fact that extraction of HDAC4 was unaffected by F-actin disruption implies that the loss of cellular integrity is not enough to explain the observed increase in soluble nuclear proteins examined.
To determine whether another cytoskeletal component contributes to the insolubility of these proteins during the preparation of nuclear extracts, we chose to examine the effect of nocodazole, a microtubule-disrupting agent. Nocodazole treatment prior to nuclear isolation and extraction had no significant effect on the solubility of any of the nuclear proteins examined in comparison with those from untreated, control cells (Fig. 3). This result demonstrates that merely disrupting cytoskeletal integrity is not sufficient to alter the solubility of nuclear proteins during nuclear isolation and subsequent salt extraction. Therefore, nocodazole treatment provided a specific, negative control and confirms that the effects observed with latrunculin treatment are specific to actin.
Size Fractionation of Histone Deacetylase Complexes and Histone Deacetylase Activity from Latrunculin-treated and Control Cells-Most HATs and HDACs have been described in more than one macromolecular complex, reflecting the potential of these proteins to interact with different binding partners in a mutually exclusive manner. The release of the HATs and HDACs following disruption of actin filaments may reflect changes in the solubility of specific HAT and HDAC complexes or a general release of all complexes that contain these proteins. To evaluate whether or not latrunculin liberated a specific subset of proteins from the nucleus, we characterized the nuclear extracts from both latrunculin-treated and control cells using gel filtration chromatography. Silver-stained SDS-PAGE gels of the fractions collected indicate that the proteins present in the nuclear extract from latrunculin-treated cells differ from the untreated control (data not shown). There is an overall increase in the amount of protein present in the fractions collected from the latrunculin-treated extract. However, there also appear to be unique proteins present in the column fractions that correspond to larger molecular weight complexes that are not found in the untreated, control fractions. Thus, the disruption of F-actin liberated very high molecular mass (Ͼ2 Mda) protein complexes that were underrepresented in the 0.42 M NaCl extract of untreated cells.
We chose to investigate whether or not there are specific differences in the species of chromatin-modifying complexes found in the two extracts. For these studies, we chose to focus upon HDAC activity as a marker of chromatin-modifying activities. We examined the elution of specific HDACs and HDAC activity from the gel filtration column and then compared the mass of the HDAC complexes to the elution of HDAC activity. HDAC activity was quantified using a commercially available fluorogenic HDAC substrate (see "Experimental Procedures" for details). Column fractions were analyzed by immuno-dot blotting using primary antibodies specific for HDACs 1-4 to identify the corresponding proteins that eluted with the detected deacetylase activity (Fig. 4, A and B). All four of the HDACs examined are present in the gel filtration fractions, and each likely contributes to the detected deacetylase activity.
HDAC1 is consistently detected in both the control and latrunculin-treated extracts in the first enzymatic activity peak eluting from the column (Fig. 4, A and B, fractions 6 -8). HDAC2 appears to be the major contributor to the activity detected in fractions 14 -18 and 20 -24. Notably, there is considerable HDAC2 present in the fractions located between and including activity peak 3 (Fig. 4, fractions 14 -18) and peak 4 (Fig. 4, fractions 20 -24), and yet there is a slight decrease in the percent of total activity detected in these fractions from the latrunculin-treated cells. This suggests that the HDAC2 population is heterogeneous and that more total activity is present in the high molecular weight complexes. The higher molecular weight HDAC2 complexes are clearly enriched in the nuclear extract from latrunculin-treated cells relative to the control extract (Fig. 4, fractions 6 -12). The most striking difference between the two extracts is the dramatic latrunculin-dependent enrichment of the HDAC1 complex co-eluting with activity peak 2 (Fig. 4, fractions 10 -12). It appears that a large proportion of the HDAC-containing complexes are inaccessible under common salt extraction conditions and that this is, at least in part, due to their association with F-actin.
F-actin-dependent Retention of a Specific HDAC1 Complex in Nuclear Pellets-Most of the differences in the HDAC com- plexes found in the latrunculin-treated and control cells are quantitative, differing in the amount and not the presence of a specific HDAC complex. However, activity peak 2 (Fig. 4) represents a qualitative difference in the HDAC1 complexes found in these two extracts. To further characterize these complexes, we analyzed the elution of proteins that have been reported to be core components of purified HDAC1 complexes. Fig. 5 shows the elution of two known HDAC1-associated proteins, RbAp46 and MTA2, from the control and latrunculin-treated extracts. There is a striking enrichment of the RbAp46 protein in the high molecular weight column fractions of the latrunculintreated nuclear extract in comparison with the control extract (Fig. 5A, fractions 7-15). The metastasis-associated protein, MTA2, also shows a dramatic enrichment in the corresponding high molecular weight fractions from the latrunculin-treated extracts (Fig. 5B, fractions 6 -18). Interestingly, the elution profiles of MTA2 from both control and latrunculin-treated extracts (Fig. 5B) resembles that of the corresponding HDAC1 elution profiles (Fig. 4, A and B). This is consistent with the presence of both of these proteins in the same complexes, as reported previously (6,12). Our findings, along with the fact that not all of the proteins examined are affected by latrunculin treatment, supports the hypothesis that disruption of cellular actin releases specific HDAC complexes as opposed to having a more generic effect. It is also important to note that latrunculin treatment is releasing each of the complex components examined in quantities that are more representative of the protein population as a whole than what is accessible from untreated cells under the same extraction conditions. DISCUSSION A major objective over the next many years is to define the specific protein-protein interactions that bring the proteome to life. Given that nuclear proteins are compartmentalized, it is necessary to develop more efficient methods of extracting these proteins in their intact, native assemblies. Here we demonstrate that several nuclear proteins involved in chromatinmodification and transcription are highly depleted in the conventional nuclear salt extract. Our data are consistent with previous reports that these proteins enrich in non-chromatin nuclear domains in vivo (16) and are retained in nuclear matrix preparations in vitro (48,49). More, importantly, this study identifies a novel biological mechanism that modulates the behavior of several nuclear proteins.
The disruption of F-actin resulted in an overall increase in the solubility of several chromatin-modifying proteins. In addition, latrunculin treatment released unique HDAC complexes not normally found in conventional nuclear extracts. HAT and HDAC proteins have been found in a myriad of nuclear complexes and play a pivotal role in transcription regulation through the modification of chromatin structure (56). Although we cannot rule out the possibility that the loss of cytoskeletal integrity in response to latrunculin treatment contributes to the observed increase in solubility, the disruption of microtubules, another cytoskeletal component, had no effect on the partitioning of any of the proteins examined into the subcellular fractions. This indicates the specificity of the results seen with latrunculin treatment and is consistent with the binding of chromatin-modifying components with F-actin in vivo. We are currently investigating the nature of this association.
The ability of F-actin to contribute to the insoluble nature of the HAT and HDAC proteins examined may reflect the involvement of actin, actin-binding proteins, or nuclear Arps in chromatin-modifying complexes (37,39,(43)(44)(45)(46)(47). In the yeast SWI/ SNF and RSC complexes, Arp7 and Arp9 are essential for chromatin remodeling (40) and are suggested to function in a structural capacity (40,46). In the complimentary mammalian BAF complex, actin binds to the Brg1 subunit and is necessary for the chromatin-dependent activation of the complex (37). In addition, it is this association of actin along with BAF53 that provides a link between the BAF complex and the nuclear matrix (37). It may be that actin filaments function to organize and/or properly localize the complex within the nucleus. Recently, it has also been demonstrated that the BAF complex is able to bind and stabilize actin filaments in vitro (45). It is not yet clear whether this stabilizing effect occurs in vivo.
BAF53 and actin have also been identified as components of the Tip60 HAT complex, which is involved in DNA repair and apoptosis (42), and BAF53 is found in a distinct HAT complex involved in c-Myc activation (38). In the case of the yeast Nu4A HAT complex, actin and Act3/Arp4 are essential for the structural integrity and activity of the complex (41). Although actin itself has not been identified in HDAC-containing complexes, potential links to actin have been established. The Brg1 subunit of the BAF complex (11) as well as a novel actin-binding protein, IR10 (47), have been identified as constituents of N-CoR complexes. Ultimately, all of these chromatin-modifying complexes then have the capacity to interact with actin filaments, which could easily contribute to the observed insolubility of the complexes and their components upon nuclear extraction.
It is unlikely that F-actin is the only component in the nucleus contributing to the insolubility of these proteins. Although latrunculin treatment was very proficient at increasing the solubility of several of the HATs and HDACs examined, the effectiveness of this treatment varied. For example, PCAF solubility remained unaffected. Nuclear lamins form an insoluble network that is present at the surface of the nucleus and throughout its interior (50), which likely could affect the solubility of nuclear proteins. Understanding the molecular basis of these interactions will be essential if we are to identify and characterize all the endogenous chromatin-modifying machines that function in vivo.
The precise function of actin in the nucleus is open for debate. Actin may prove to contribute to the functional compartmentalization of nuclear macromolecular complexes by providing a type of structural scaffold. Although bundled actin filaments have not been visualized in the nucleus, the recent discovery that a pool of nuclear actin does exist in polymer form 2 is consistent with nuclear actin playing a structural role. This role would be supported by our findings that disruption of F-actin alters the solubility of nuclear complexes. However, the actin-dependent activation of the BAF complex (37,45) along with the involvement of actin in transcription (54) indicate that nuclear actin may also play a regulatory role. Regardless, the current study links previous work demonstrating the discreet localization and nuclear matrix association of the nuclear proteins examined and the presence of actin as a common component of chromatin-modifying machineries.
The discovery that the disruption of cellular actin affects the solubility of several nuclear proteins has practical value to the study of nuclear complexes. We have known for some time that functionally related molecules are ordered into an intricate substructure within the nucleus, but the organizational mechanisms have been unknown. The contribution of biochemical studies is limited by the accessibility of the various components in a form that reflects their physiological state. Our data show that the disruption of cellular actin is sufficient to liberate significant quantities of otherwise insoluble HDAC complexes while notably retaining their enzymatic activity. Therefore, in addition to revealing a novel biological mechanism influencing nuclear complex solubility, we have developed a simple procedure, through the manipulation of cellular actin, to enrich for high molecular weight, active nuclear complexes that are otherwise difficult to isolate.