Epigenetics is the alteration of phenotype without affecting the genotype. An underlying molecular mechanism of epigenetics is the changes of chromatin structure by covalent histone modifications and nucleosome reorganization. In the yeast, Saccharomyces cerevisiae, two of the most well-studied macromolecular complexes that perform these epigenetic changes are the ATP-dependent Swi/Snf chromatin-remodelling complex and the SAGA histone acetyltransferase complex. To understand fully the mechanism by which these large protein complexes perform their functions in the cell, it is crucial that all the subunits of these complexes are identified. In an attempt to identify new subunits associated with SAGA and Swi/Snf, we used tandem affinity purification, followed by a multidimensional protein identification technology to analyse the subunit composition. Our analysis identified two novel proteins, one associated with SAGA, YPL047W (Sgf11), and another associated with Swi/Snf, Rtt102.
Cells use several types of chromatin-remodelling complexes to alter or modify chromatin in the process of gene activation or repression. The ability to regulate transcription through DNA accessibility is crucial for the maintenance of proper gene activation. The process by which chromatin regulates this accessibility occurs through two distinct mechanisms, acetylation and ATP-dependent chromatin remodelling. Among the more well-characterized complexes involved in these two processes are the ATP-dependent Swi/Snf complex and the SAGA HAT (histone acetyltransferase) complex.
The SAGA complex is a multisubunit HAT complex that is conserved from yeast to humans [1,2]. SAGA serves as a transcription co-activator for a large group of transcription activators ranging from the yeast GAL4 protein to the human c-myc proto-oncogene product [3,4]. Gcn5 is the catalytic subunit of SAGA and is also present in at least three other HAT complexes in yeast: ADA, SLIK and HAT-A2 (reviewed in ). Interestingly, although the SAGA and SLIK HAT complexes are distinct, they share every subunit except that SLIK lacks Spt8 and has a truncated form of the Spt7 protein and SLIK has been proposed to have distinct roles from SAGA in transcription [5,6].
The Swi/Snf complex is one of several ATP-dependent chromatin-remodelling complexes found in Saccharomyces cerevisiae. The Swi/Snf complex is conserved from yeast to humans. In yeast Swi/Snf, 11 polypeptides have been characterized as being associated with the complex . Interestingly, two of these subunits, Arp7 and Arp9, are shared by the Swi/Snf complex and the RSC chromatin-remodelling complex .
The ability to understand fully the underlying mechanism by which macromolecular complexes, such as SAGA and Swi/Snf, function to regulate various processes in the cell depends on knowing the composition of the complex itself . A variety of techniques have been employed in an attempt to characterize the make up of various protein complexes. Many of these techniques focus on one target gene or protein at a time . Recent advances in MS have allowed a more rapid way of identifying complex mixtures of proteins without having to separate individual subunits and purifying them to homogeneity [10–12]. In contrast with other methods of characterizing protein–protein interactions, such as genome-wide two-hybrid screens  and synthetic lethality screens  that require the development of resources, MS simply requires a complex protein mixture to return consistent and reproducible results defining the interactions over an entire proteome [15,16].
Many of the problems associated previously with protein analysis using MS have been resolved with the development of MudPIT (multidimensional protein identification technology) as an unbiased method that allows for rapid, large-scale proteomic analysis, while decreasing the sample handling [10–12]. MudPIT relies on the digestion of protein mixtures in solution into peptide fragments, which are then resolved by multidimensional capillary chromatography and directly applied to an ion-trap mass spectrometer . The results are subsequently analysed using the SEQUEST algorithm .
We used MudPIT along with TAP (tandem affinity purification) to characterize further and clearly define all of the subunits of the SAGA and Swi/Snf chromatin-modifying complexes in S. cerevisiae. Using the standard procedure for TAP of the Swi/Snf complex followed by MudPIT, we found all the previously characterized subunits, in addition to a new subunit, RTT102, which has just recently been reported by Graumann et al. . Using a modified TAP procedure that allowed for the separation of multiple Gcn5-containing HAT complexes, followed by MudPIT, again we found all the previously characterized SAGA and SLIK subunits, in addition to a new subunit, YPL047W, which has just recently been renamed SGF11 (SAGA-associated factor 11) .
The ability to identify new subunits in these two well-characterized complexes reveals the power this technique of coupling TAP with MudPIT MS brings to protein biochemistry.
Materials and methods
TAP tag strains, Ada2, RTT102, YPL047W (SGF11) and Snf6, were obtained from Open Biosystems (Hunstville, AL, U.S.A.) as part of the TAP-tagged library described by Ghaemmaghami et al. . Ada2TAP;sgf11Δ was generated by crossing the Ada2TAP strain (a) with an sgf11Δ (α) obtained from the Matα deletion library from Open Biosystems.
TAP of Snf6 and Sgf11 was performed essentially as described in  except for the following changes: all incubations were performed overnight at 4°C and final elutions were performed in batches at room temperature (22°C) in 250 μl fractions repeated nine times.
Ada2 was purified as described previously  except for the following changes: after cleavage with TEV protease (Invitrogen, Carlsbad, CA, U.S.A.), 3 ml of TEV cleavage buffer [10 mM Tris, pH 8.0, 150 mM NaCl, 0.1% Nonidet P40, 0.5 M EDTA, 10% (v/v) glycerol, 1 mM PMSF and 1 mM DTT (dithiothreitol)] was added, followed by Buffer B [50 mM Tris/HCl, pH 8.0, 10% (v/v) glycerol, 0.1% Tween 20, 1 mM PMSF and 1 mM DTT], to bring the NaCl concentration to 0.1 M. The samples were then loaded on to an anion-exchange column (MonoQ), and the bound proteins were eluted with a 25 ml salt gradient of 100–500 mM NaCl in Buffer B. All other fractions were assayed for HAT activity and peak fractions were pooled. Salt concentration was adjusted to 0.3 M, along with the addition of 1 mM imidazole, 1 mM magnesium acetate and 2 mM calcium chloride. These samples were then bound to 500 μl of calmodulin resin (Amersham Biosciences, Uppsala, Sweden) by the standard TAP procedure .
Whole cell extracts were prepared from 50 ml of exponential phase culture in 1 ml of lysis buffer (40 mM Hepes/KOH, pH 7.5, 350 mM sodium chloride, 10% (v/v) glycerol, 0.1% Tween 20, 0.5 mM DTT, 1 mM PMSF, 1 μg/ml Pepstatin A and 1 μg/ml leupeptin). 1 mg of proteins were incubated overnight at 4°C with 10 μl of IgG–Sepharose (Amersham Biosciences) to immunoprecipitate TAP-tagged proteins. The immunoprecipitated material was washed three times in lysis buffer and then resolved by SDS/PAGE (8% gel), transferred on to a PVDF membrane and probed for the presence of Swi3 or Sth1. These antibodies also reveal the precipitated TAP-tagged proteins. Western blots were developed using ECL®+ reagent and then scanned using a Typhoon 9400 (Amersham Biosciences).
HAT assays were performed as described previously .
MudPIT analyses of TAP-purified complexes were performed as described previously [11,12] with usage of the three-phase MudPIT column as described by McDonald et al. . Data analysis commenced with the application of the software algorithm 2 to 3  to determine the charge state and to delete spectra of poor quality. SEQUEST  was used to match MS/MS spectra to peptides in a database containing S. cerevisiae protein sequences downloaded from the NCBI. The validity of peptide/spectrum matches was assessed using the SEQUEST-defined parameters, cross-correlation score (XCorr) and normalized difference in cross-correlation scores (DeltCn). Spectra/peptide matches were retained only if they had a DeltCn of at least 0.08 and a minimum XCorr of 1.8, 2.5 and 3.5 for singly, doubly and triply charged spectra. In addition, the peptides had to be at least seven amino acids long. DTASelect  was used to select and sort the peptide/spectrum matches passing this set of criteria. Peptide hits from multiple runs were compared using CONTRAST .
Results and discussion
Identification of YPL047W as a novel subunit of both SAGA and SLIK HAT complexes
Using a modified procedure for TAP of the Ada2 protein, a subunit shared by SAGA and SLIK (reviewed in ), we separated and individually purified the two complexes (Figures 1A and 1B). This purification was followed by MudPIT of both SAGA and SLIK, in which we identified a previously uncharacterized open reading frame, YPL047W, present in both the SAGA and SLIK HAT complexes (Figure 1C). Concomitant with our discovery, Powell et al.  annotated YPL047W as SGF11 (SAGA-associated factor) and we will refer to it as such for the remainder of this paper. SGF11 encodes a 99-amino-acid open reading frame with no identifiable homologues in higher eukaryotes.
Purification of SAGA and SLIK and identification of YPL047W (Sgf11) as a novel subunit present in both complexes
To determine whether SGF11 is a stable component of SAGA and SLIK, we reciprocally tagged SGF11 using the standard tandem-affinity tag and purified the complexes associated with SGF11. The proteins purified from the Sgf11TAP strain were identified using MudPIT MS. An aliquot of the Sgf11TAP purification was compared with the Ada2TAP purification and analysed by silver staining before MS analysis (Figure 2A, lane 3 versus lane 5). All the proteins known to be associated with SAGA and SLIK were also found in the Sgf11TAP purification. Therefore Sgf11 is indeed a stable component of both SAGA and SLIK.
Deletion of SGF11 does not affect the integrity or acetyltransferase activity of either SAGA or SLIK
Additionally, we tested the enzymic activity of SAGA purified through SGF11. Since HAT activity is a major function of SAGA in transcriptional regulation, we reasoned that, if Sgf11 is a stable component of SAGA, the complex purified through Sgf11 should have HAT activity. Indeed, when we performed HAT assays for acetylation, we found that the Sgf11TAP complex was able to acetylate both nucleosomes and, to a lesser extent, core histones (Figure 2B and results not shown). The observation that nucleosomes are a better substrate for acetylation by SAGA is consistent with what other studies have shown .
Defining the role of Sgf11 in SAGA/SLIK complex integrity and activity
Of the 18 polypeptides characterized for the SAGA and SLIK complexes, a small number of these proteins are required for the structural integrity of the complex; these include Ada1, Spt7 and Spt20, which are supposed to play a role in stabilizing the complex .
A number of other proteins are known to be essential for the enzymic activity of SAGA, including Gcn5, the catalytic subunit, as well as Ada3, which are required for all HAT activities, and TAF12, which is required for HAT activity on nucleosomes [26,27].
To determine if SGF11 plays a role in the stability or enzymic activity of SAGA and SLIK, we purified both SAGA and SLIK from a yeast strain lacking SGF11. MudPIT analysis of the complexes purified revealed all of the subunits associated with SAGA and SLIK with the exception of the recently described deubiquitinating enzyme Ubp8 (Figure 2C) [28,29]. Therefore SGF11 is not important for the overall integrity of SAGA or SLIK.
Next, we wanted to determine the role of Sgf11 in the acetyltransferase activity and, hence, performed HAT assays on SAGA and SLIK purified from wild-type and sgf11Δ strains. Using both nucleosomes and core histones as a template, we see no effects on the ability of SAGA or SLIK to acetylate nucleosomes or core histones (Figure 2B, cf. lanes 1 and 2 with lanes 4 and 5). Therefore we conclude that the HAT activity of both SAGA and SLIK is SGF11-independent. However, it remains to be seen whether the other enzymic activity recently associated with SAGA and SLIK, namely the deubiquitination activity associated with UBP8, is affected in the sgf11Δ [28,29]. Since UBP8 is lost from SAGA and SLIK in the absence of SGF11, SGF11 may be important for deubiquitination by maintaining the association of UBP8 with SAGA and SLIK.
RTT102 is a novel subunit present in the Swi/Snf complex
Since our MS analysis of the well-characterized SAGA complex yielded a previously unknown subunit, we decided to test the hypothesis that other chromatin-modifying complexes still have subunits that are yet to be identified. Using the same method of TAP followed by MudPIT analysis, we purified the Swi/Snf complex from a yeast strain containing the TAP tag on the Snf6 subunit. MudPIT analysis of the purified Swi/Snf complex revealed a number of peptide hits for RTT102, which had not been previously characterized as a subunit of Swi/Snf (Figure 3A). Furthermore, a number of peptides for the RSC chromatin-remodelling complex subunit, Sth1, were also identified (results not shown).
Purification of Swi/Snf and identification of RTT102 as a novel subunit of both the Swi/Snf and RSC chromatin-remodelling complexes
To validate RTT102 as a bona fide subunit of Swi/Snf, we performed pulldown experiments taking advantage of the Protein A module of the TAP tag. Using whole-cell extract derived from the Rtt102TAP-tagged strain and IgG resin, we were first able to show that Swi3, a component of the Swi/Snf complex, was pulled down with the Rtt102TAP tag (Figure 3B, lane 1). Additionally, we found that IgG pulldown of Rtt102 also revealed an interaction with Sth1, the catalytic subunit of the Rsc chromatin-remodelling complex, indicating that Rtt102 is also part of RSC (Figure 3B, lane 4). However, IgG pulldowns of Snf6 did not reveal any interaction with Sth1, indicating that the interactions are specific and Rtt102 is part of both the Swi/Snf and RSC chromatin-remodelling complexes (Figure 3B, lanes 2 and 5). We further tested whether Rtt102 was important for the integrity of Swi/Snf by isolating Swi/Snf from a yeast strain lacking RTT102. We found that the complex appears intact in the absence of Rtt102 (results not shown). Graumann et al.  have also recently shown that Rtt102 is part of both the Swi/Snf and RSC chromatin-remodelling complexes and that the loss of RTT102 causes phenotypes consistent with the loss of other Swi/Snf subunits .
Genes: Regulation, Processing and Interference: A Focus Topic at BioScience2004, held at SECC Glasgow, U.K., 18–22 July 2004. Edited by I. McEwan (Aberdeen, U.K.), B. White (Glasgow, U.K.), S. Graham (Glasgow, U.K.), S. Roberts (Manchester, U.K.), A. Sharrocks (Manchester, U.K.), D. Black (Organon, U.K.), S. Newbury (Oxford, U.K.), J. Sayers (Sheffield, U.K.) and A. Lloyd (University College London, U.K.).
This work was supported by a post-doctoral fellowship from the Damon Runyon Cancer Research Foundation to K.K.L. (1751-03), a post-doctoral fellowship from the Leukemia and Lymphoma Society to P.P. and NIH grant no. GM46787 from NIGMS to J.L.W.