mChIP-KAT-MS, a method to map protein interactions and acetylation sites for lysine acetyltransferases

Significance

Recent proteomic studies have revealed that lysine acetylation is a global and ubiquitous posttranslational modification. However, in the vast majority of cases the lysine acetyltransferases (KATs) responsible for individual modifications remain unknown. Here we present a unique methodology that connects KATs to their substrates. To validate the methodology, we use the yeast KAT nucleosome acetyltransferase of histone H4 (NuA4) and identify both protein interactions and acetylation targets. Importantly, this methodology can be applied to any KAT and should aid in the linking of KATs to their cellular targets.

Abstract

Recent global proteomic and genomic studies have determined that lysine acetylation is a highly abundant posttranslational modification. The next challenge is connecting lysine acetyltransferases (KATs) to their cellular targets. We hypothesize that proteins that physically interact with KATs may not only predict the cellular function of the KATs but may be acetylation targets. We have developed a mass spectrometry-based method that generates a KAT protein interaction network from which we simultaneously identify both in vivo acetylation sites and in vitro acetylation sites. This modified chromatin-immunopurification coupled to an in vitro KAT assay with mass spectrometry (mChIP-KAT-MS) was applied to the Saccharomyces cerevisiae KAT nucleosome acetyltransferase of histone H4 (NuA4). Using mChIP-KAT-MS, we define the NuA4 interactome and in vitro-enriched acetylome, identifying over 70 previously undescribed physical interaction partners for the complex and over 150 acetyl lysine residues, of which 108 are NuA4-specific in vitro sites. Through this method we determine NuA4 acetylation of its own subunit Epl1 is a means of self-regulation and identify a unique link between NuA4 and the spindle pole body. Our work demonstrates that this methodology may serve as a valuable tool in connecting KATs with their cellular targets.

Lysine acetyltransferase (KAT) enzymes catalyze the transfer of an acetyl group from acetyl CoA onto the ε-amino group of lysine residues. Acetylation then regulates protein function in a number of ways, including altering the localization, activity, stability, and physical interactions of the target protein (1, 2). Through the acetylation of histone proteins, KATs have traditionally been associated with a variety of chromatin-based cellular processes, such as transcription, silencing, and DNA repair. More recently, systematic screens aimed at identifying acetylated lysine peptides in both prokaryotic and eukaryotic systems have established acetylation as a ubiquitous and conserved posttranslational modification occurring on thousands of proteins (3–10). Furthermore, these screens revealed fundamental properties associated with lysine acetylation, such as the abundance of acetylation sites found on metabolic enzymes and mitochondrial proteins, the tendency of multisubunit protein complexes to be abundantly acetylated, and that most acetylated proteins do not have obvious roles in chromatin-directed processes. However, in the vast majority of cases the biological consequences of lysine acetylation and the KAT responsible for catalysis have yet to be determined. To fully elucidate the cellular functions of KATs in vivo will require a detailed understanding of the direct pathways in which a KAT functions, and a focused analysis of the role of acetylation within those pathways. (Abbreviations for genes and proteins are provided in Table S1.)

A recent proteome-wide survey of acetyl lysine residues in Saccharomyces cerevisiae identified more than 4,000 lysine acetylation sites; however, the KATs responsible for these modifications were not discerned (4). Global genetic screens that identify proteins whose overexpression cause fitness defects in either KAT mutants (11) or lysine deacetyltransferase mutants (12) have successfully predicted proteins whose function is regulated by reversible acetylation. Together, synthetic dosage lethal genetic screens have led to the identification of 96 acetylated proteins (11, 12). In another systematic analysis, protein microarray technology, encompassing more than 90% of the proteins encoded by the S. cerevisiae genome, was used to identify in vitro acetylation substrates of the KAT complex nucleosome acetyltransferase of histone H4 (NuA4) (13). This work identified 91 NuA4 targets in vitro and ultimately succeeded in confirming 13 substrates in vivo (13). Despite these successes, the genetic and proteomic approaches currently in use each have unique technological challenges and further share the drawback that the specific lysine residues acetylated by a particular KAT are not immediately identified. Thus, complementary techniques need to be developed to connect KAT enzymes to their substrates and acetyl lysine residues in vivo to fully elucidate the pathways governed by acetylation.

To this end, we have developed a unique proteomic method to generate a KAT-associated protein interaction network in which the level of acetylated lysine residues is enriched in vitro. Here we describe the methodology we have named mChIP-KAT-MS (modified chromatin-immunopurification coupled to an in vitro KAT assay with mass spectrometry) designed to link KAT enzymes to new substrates and cellular pathways. We showcase this method using the essential S. cerevisiae KAT NuA4, a highly conserved, multifunctional enzyme complex (14). Using mChIP-KAT-MS, we define the NuA4 interactome and in vitro-enriched acetylome, identifying over 70 unique physical interaction partners for the complex and over 150 acetyl lysine residues, of which 108 are NuA4-specific in vitro sites. To validate our method, we perform a series of directed follow-up experiments to decipher the impact of NuA4 acetylation of one of its own subunits, Epl1, and link the catalytic function of NuA4 to the spindle pole body (SPB) and spindle dynamics. Taking these data together, this work demonstrates the utility and the flexibility of the mChIP-KAT-MS approach as a unique methodology to study KATs in vivo.

Results

mChIP-KAT-MS as a Unique Method to Study KAT Function.

To gain insight into the mechanisms of action and specific pathways in which a KAT functions, we developed a method that can identify the network of proteins that physically interact with a KAT as well as simultaneously identify acetyl lysine residues arising from either preexisting in vivo or KAT-dependent in vitro catalysis within that network. The methodology, called mChIP-KAT-MS, consists of three steps: (i) isolating the KAT and its associated protein network from cells; (ii) enriching the level of acetylated lysine residues within the network using an in vitro KAT reaction; and (iii) identifying interacting proteins and acetylation sites by LC-MS/MS (Fig. 1). To isolate NuA4 and interacting proteins, we purify NuA4 through Esa1-TAP (tandem-affinity purification), its catalytic subunit (15), using a variant of the traditional immunopurification strategy, termed modified chromatin immunopurification (mChIP) (16). The mChIP method, originally developed to study chromatin-associated proteins, has now been used successfully to increase the depth of coverage of protein interactions of wide range of bait proteins in vivo (17, 18). Briefly, yeast whole-cell lysates are subjected to mild sonication followed by gentle centrifugation, thereby promoting retention of poorly soluble cellular components in solution. Immunopurification is performed in a single step using magnetic beads coated with IgG antibodies that specifically recognize the protein A component of the TAP tag. Next, to boost the level of acetyl lysine residues on proteins within the network, an in vitro KAT assay is performed in which exogenous NuA4, stringently purified from yeast, and isotopically labeled acetyl CoA (¹³C₂-acetyl CoA; herein referred to as heavy acetyl CoA) are incubated with the immunopurified bead matrix. Finally, the acetyl lysine enriched network is analyzed by LC-MS/MS. Because of the shift in mass-to-charge ratio, heavy acetyl groups resulting from NuA4 in vitro KAT activity can be distinguished from unlabeled, preexisting acetyl moieties (herein referred to as light acetyl CoA). Therefore, this methodology enables: the (i) identification of the network of proteins associated with the KAT of interest under normal growth conditions; (ii) generation of a list of light or in vivo acetylation sites, thereby increasing our general knowledge of yeast acetylation; and (iii) definition of a set of KAT-specific in vitro acetylation sites on proteins that physically copurify with the KAT of interest.

Fig. 1.

mChIP-KAT-MS Methodology. Step 1: NuA4 and its associated protein interaction network are purified from yeast using mChIP technology. Briefly, yeast whole-cell lysate is mildly sonicated and gently clarified before NuA4 immunopurification through endogenously TAP-tagged Esa1 with magnetic beads coupled to IgG antibodies. Step 2: NuA4 and copurifying proteins are subjected to an in vitro KAT assay. Highly purified exogenous NuA4 and isotopically labeled acetyl CoA (¹³C₂-Acetyl CoA) are added to the NuA4-bead matrix under conditions promoting NuA4 KAT activity. Step 3: Proteins and acetylation sites are identified by LC-MS/MS. Acetyl lysines identified on sequenced peptides may derive from preexisting, in vivo catalysis (green “Ac” tag; light) or in vitro catalysis, which will yield isotopically labeled acetyl lysine residues (red “Ac” tag; heavy).

NuA4-Associated Protein Interaction Network.

To generate a high-confidence NuA4-associated protein network (Fig. 2A), the Esa1-TAP mChIP assay was repeated six times. Interacting proteins identified in at least two experimental replicates were deemed reproducible and included in the final dataset. Proteins identified in only one mChIP experiment but found in complexes with reproducible interactors were also included. Finally, all proteins modified by lysine acetylation are also presented in Fig. 2A. Excluded proteins were those previously identified as mChIP contaminants or common mChIP preys (17) (Dataset S1), as well as any protein functioning in a complex with a filtered protein (Dataset S1). The protein network includes the 13 core members of the NuA4 KAT complex, as well as an additional 84 proteins (Fig. 2A and Dataset S1). As expected, this network contains proteins previously shown to interact with one or more NuA4 subunits, such as histone proteins H4 (Hhf1), H2A (Hta1 and Hta2), H2B (Htb2), and H3 (Hht1) (19); stress-responsive transcription factors (Msn2, Msn4, and Yap1) (20); subunits of the Chaperonin Containing Tcp1 (CCT) complex (Tcp1, Cct8, Cct4) (21); and the 14-3-3 protein Bmh1 (22). The unique NuA4 interacting proteins dramatically expand our knowledge of pathways in which NuA4 may function. New interactions include multiple members of protein complexes such as the SPB [8 of 17 core components (23)], the COPI [4 of 7 (24)] and COPII [3 of 13 (25)], transport complexes, and all four members of a multifunctional mRNA binding stress granule complex [4 of 4 (26)] (Fig. 2A). Furthermore, we identified the copurification of the 26S proteasome: 8 of 33 core members of both the 19S regulatory particle and 20S catalytic core particle (27), supporting the recent report of this interaction (28). In total, more than half of the proteins in the network are nonnuclear (Fig. 2B), suggesting a broad localization pattern for NuA4 within the yeast cell. This work represents an in depth analysis of proteins that copurify with NuA4 and substantiates the hypothesis that NuA4 participates directly in a diverse array of cellular processes beyond chromatin-related functions.

Fig. 2.

The acetyl lysine-enriched NuA4-associated protein network. (A) The 13 NuA4 subunits, including the bait protein Esa1-TAP, are represented by black nodes (Top Left). Interacting proteins are grouped by cellular process (node color/label) and further organized into complexes where appropriate (circles). Eighty-four protein interactions are represented in the map, 57 of which copurified in a minimum of two of six experimental replicates (large nodes); and 27, although only identified in one of six replicates, belong to reproducibly copurifying protein complexes (i.e., Spc29 of the SPB complex). Furthermore, all proteins harboring one or more acetyl lysine residues are included in the map. Previously identified NuA4 protein interactions are indicated by a black circle around the node (as published at www.theBiogrid.org v3.1.70). Only physical interactions for Esa1, Yng2, Epl1, and Eaf1 were considered, because these proteins function solely within NuA4 or the related PicNuA4 complex in vivo. (B) Cellular localization of proteins in the network. Localization annotation is based on a global study (62) (see Dataset S2 for individual annotations). The 13 NuA4 subunits were excluded in this analysis. (C) Frequency distribution of amino acids surrounding heavy acetyl lysine residues. Frequency of amino acids (y axis) spanning positions −6 to +6 (x axis) surrounding heavy acetyl lysine residues identified within the NuA4-associated protein network (red tag). Residues in green: basic; red: hydrophobic; pink: small; blue: S/T; black: all other residues.

Lysine Acetylation Identified Within the NuA4-Associated Protein Network.

The mChIP-KAT-MS methodology enables the identification of acetylation sites resulting from either in vivo acetylation (light acetyl groups) or in vitro catalysis (heavy acetyl groups). Although the mChIP-MS was performed six times, only two replicates included the NuA4 KAT assay with isotopically labeled acetyl CoA (Fig. S1). In total, we identified 66 acetyl lysine residues on 23 proteins (Fig. 2A, Table 1, and Dataset S1). Twenty-eight lysines were modified only by a light acetyl group, including most previously identified acetylation sites on histone proteins (29, 30), the known autoacetylation site of Esa1 (K262) (31), plus an additional 12 sites on six nonhistone proteins. Five of these 12 acetylation sites had also been identified in a recent yeast acetylome study (4). Because these proteins copurify with NuA4 it suggests, but does not confirm, that their acetylation could depend on NuA4. Thirty-three lysines were modified by only a heavy acetyl group; as we detected no other KAT enzymes in the NuA4 protein interaction network (Fig. 2 and Dataset S1), heavy acetylation implicates NuA4-dependent catalysis in the acetylation of these sites in vitro. The set of heavy acetylation sites includes: 13 lysines on seven NuA4 interacting proteins, of which two were previously identified in vivo (4) (Table 1); two heavy sites on histone H2A (Hta1 K124, K127), a protein acetylated by NuA4 in vivo at K5 (32) and an in vivo modification reported on the conserved K127 residue of human histone H2A (30); and 18 heavy sites on five NuA4 subunits themselves. The remaining six acetyl lysine residues, identified on five proteins, were represented by peptides modified by both light and heavy acetyl groups. These sites represent strong candidates for NuA4-dependent acetylation targets in yeast cells, as their in vivo occurrence suggests biological relevance, and their heavy acetylation in vitro indicates NuA4 can catalyze the reaction. All but one of these proteins (Pab1, K7) belongs to the NuA4 complex. Importantly, this set includes the Yng2 K170 site, a previously described acetylation target of Esa1 in vivo (33). Taken together these data serve to expand our knowledge of lysine acetylation in yeast by identifying in vivo acetylated proteins and putative targets of NuA4.

Table 1.

Acetyl lysine residues in the NuA4-associated protein network

Protein	Acetylation sites	Description
Light (in vivo) acetylation
H2A (Hta2)	K5*,†, K9*,†	Histone
H2B (Htb2)	K7*,†, K8†, K12*,†, K17*,†, K22*,†, K23*,†	Histone
H3 (Hht1)	K19*†, K24*†, K28†	Histone
H4 (Hhf1)	K6*,†, K9*,†, K13*,†, K17*,†	Histone
Rps12	K114, K125	Ribosome
Sec7	K1237, K1238	Transport
Ssb1	K466, K538 or K539	Protein folding
Ssb2	K428†	Protein folding
Esa1	K262*,†	NuA4
Epl1	K345†, K353†, K376, K379	NuA4
Tra1	K552†	NuA4
Heavy (in vitro) acetylation
Gds1	K343 or K345, K348, K351, K354	Unknown
Hca4	K570	RNA processing
Nop4	K144	RNA processing
Rpl31A	K86, K102	Ribosome
Rps12	K95	Ribosome
Sas10	K10	RNA processing
Ssb1	K90 or K95, K571 or K573	Protein folding
H2A (Hta1)	K124, K127*	Histone
Esa1	K82, K96 or K97, K97	NuA4
Epl1	K96, K100, K116, K118, K395, K446, K470, K496, K512, K569, K721	NuA4
Eaf1	K280	NuA4
Arp4	K350	NuA4
Eaf3	K45, K54	NuA4
Light and heavy (in vivo and in vitro) acetylation
Esa1	K135	NuA4
Eaf5	K3	NuA4
Epl1	K39	NuA4
Yng2	K170*, K208	NuA4
Pab1	K7	RNA processing

Boldface K represents acetyl lysine residues identified in S. cerevisiae acetylome study (4); ‘or’ indicates that unambiguous assignment of the acetyl group to one of two lysine residues within a single peptide was not possible.

^*Acetyl lysine residues previously identified in vivo (30, 33, 60, 61).

^†Cannot be distinguished from trimethylation. Note, position number of lysine corresponds to position listed in SGD (www.yeastgenome.org).

A unique feature of the acetylation dataset is that it directly links specific acetylation sites to the catalytic activity of NuA4. To identify preferences that may contribute to NuA4 acetylation site selection, we performed an enrichment analysis on the amino acid sequence surrounding heavy acetyl lysine residues. Previous enrichment analyses have focused on high-throughput acetyl lysine datasets where the KAT responsible for each identified modification was unknown (3, 7, 30). Using all heavy acetylation sites presented in this work, which includes the acetylation sites outlined above as well as those identified using the inverse KAT-mChIP described below (Tables 1 and 2 and Tables S2 and S3), we compared the local amino acid sequences surrounding heavy acetyl lysine residues (six residues to the left and right) for biases at each position. We identified significant enrichments (Table S2) for lysine residues and small amino acids (serine and alanine) immediately surrounding heavy acetyl lysines (Fig. 2C). In the absence of a clearly defined acetylation motif, one interpretation of this result is that NuA4 may recognize some other cognate feature of its substrate rather than a specific motif surrounding the targeted lysine. Indeed, noncatalytic NuA4 subunits have been implicated in targeting the complex to specific genomic loci (34), supporting multiple modes of recognition of acetylation targets.

Table 2.

Acetyl lysine residues identified using the inverse mChIP-KAT-MS approach on Bait proteins and NuA4

Protein	Acetylation sites	Sequence coverage* (%)	Description
Light (in vivo) acetylation
Spc110	K331	66	SPB
Spc72	K590	35	SPB
Gds1	K87	43	Unknown
Cdc10	K128, K166	56	Septin
Cdc12	K251	80	Septin
Epl1	K345	—	NuA4
Heavy (in vitro) acetylation
Cnm67	K128, K433	72	SPB
Nud1	K35	54	SPB
Sfi1	K868, K869	28	SPB
Spc42	K13	66	SPB
Bfa1	K328	46	SPB
Msn4	K357, K557	33	Transcription
Gds1	K348, K351, K354, K365, K366, K408	43	Unknown
Shs1	K488†	73	Septin
Eaf1	K848	—	NuA4
Eaf3	K156	—	NuA4
Eaf7	K343†, K381, K399, K409	—	NuA4
Epl1	K16, K429, K821	—	NuA4
Swc4	K345, K346, K350	—	NuA4
Yng2	K145, K146, K170	—	NuA4
Esa1	K97	—	NuA4
Light and heavy (in vivo and in vitro) acetylation
Cdc3	K3	76	Septin
Epl1	K427	—	NuA4

^*Sequence coverage is not given for NuA4 subunits as they were identified in multiple MS runs and coverage varied. Boldface K represents acetyl lysine residues identified in S. cerevisiae acetylome study (4).

^†Acetyl lysine residues previously identified in vivo (30, 33, 60, 61).

Inverse Application of mChIP-KAT-MS Provides an Alternative Strategy to Identify Acetylation Sites.

More than half of the acetylation sites we identified by NuA4 mChIP-KAT-MS analysis occurred on NuA4 subunits (Fig. 2A, Table 1, and Dataset S1), likely resulting from their relative abundance due to tight copurification with Esa1-TAP and additional supplementation for the in vitro KAT reaction. We hypothesized that acetylation sites on other non-NuA4 physical interactors, generally identified at much lower abundance (Fig. S2), may have been missed. To address this theory, we chose four preys from the NuA4 interactome (Msn4, Gds1, Cnm67, and Spc72), and two additional proteins, Cdc11 and Shs1, subunits of the septin protein complex we recently linked to NuA4 function (11), and used an inverse mChIP-KAT-MS strategy. In this approach mChIP samples of the six TAP-tagged proteins were individually subjected to NuA4 in vitro KAT assays, followed by LC-MS/MS to identify acetylation sites. For Msn4, Gds1, and the septin proteins, only silver-stained bands corresponding to the prey proteins were subjected to tryptic digest and LC-MS/MS; but in the case of the SPB proteins, each silver-stained lane was subdivided into 12 bands and all were processed. From our initial NuA4 mChIP-KAT-MS dataset (Fig. 2 and Table 1), only Gds1 harbored acetylation sites, all four of which were heavy. We hypothesized that the increased abundance resulting from the inverse mChIP-KAT-MS approach would enable identification of previously undiscovered sites of light or heavy acetylation. Indeed, we identified multiple acetylation sites on the bait proteins, as well as additional sites on NuA4 subunits and on proteins that copurified with the bait proteins (Tables 2 and 3, and Dataset S2). On the stress responsive transcription factor Msn4, the transcriptional activity of which we previously demonstrated to be repressed by NuA4 under nonstress conditions (20), we identified two heavy acetyl lysine residues (Table 2). On the uncharacterized protein Gds1, we identified six heavy and one light acetyl lysine residues, importantly reproducing three of the four heavy sites found in the initial NuA4 mChIP-KAT-MS experiment (Table 2). Although it is likely that the heavy acetylation sites identified on Gds1 and Msn4 are because of NuA4, as peptides of Sgf73 and Spt20 (subunits of the KAT SAGA/SLIK) were identified (Table 3) it is possible that the in vitro acetylation sites detected on these baits could be dependent on SAGA/SLIK. Through Cnm67 and Spc72 we purified almost all subunits of the SPB complex and many associated proteins (Dataset S3). Nine unique acetylation sites on seven SPB or SPB-associated proteins, including seven heavy sites, were identified (Table 2). Two of the heavy sites were identified on Cnm67, which was previously implicated as a NuA4 target in the in vitro protein microarray analysis (13), and more recently shown to be acetylated in vivo (12). Notably, application of the inverse mChIP-KAT-MS approach to the two yeast septin proteins (Cdc11, Shs1) revealed two heavy acetylation sites (Table 2), including one that overlaps with a previously identified known in vivo site (Shs1 K488) (11). Furthermore, this analysis enabled the identification of an additional 18 sites on NuA4 subunits, including one Epl1 lysine residue (K427) represented by peptides harboring both heavy and light acetyl moieties (Table 2) and one Eaf3 lysine residue (K156) that was recently identified as an in vivo site (4). In addition to the acetylation sites found on target preys and NuA4, 53 additional lysine acetylation sites were identified on copurifying proteins, including 16 in vivo and 37 in vitro (Table 3 and Dataset S2). In summary, the identification of 20 acetylation sites on 12 target proteins, 18 acetylation sites on NuA4 subunits (Table 2 and Dataset S2), and 53 acetylation sites on copurifying proteins, validates the inverse approach as an effective variant of mChIP-KAT-MS to identify acetyl lysine residues.

Table 3.

Acetyl lysine residues identified using the inverse mChIP-KAT-MS approach on copurifying proteins

Protein	Acetylation sites	Purification	Description
Light (in vivo) acetylation
Bir1	K920	SPB	Chromosome segregation
Cdc1	K451	Septin	Cell cycle
Dbp10	K76	SPB	Ribosome biogenesis
Hac1	K38	SPB	Transcription factor
Hul5	K24, K31	SPB	Ubiquitin ligase
Itc1	K317, K319	SPB	Chromatin remodeling
Ltv1	K453	Septin	Ribosomal processing
Mss2	K348	SPB	Electron transport chain
Nam2	K200	SPB	tRNA synthetase
Nup159	K5	SPB	Nuclear pore complex
Rad24	K291	Gds1	cell cycle
Rps18A	K56	SPB	Ribosome
Srf1	K206, K214	Septin	Phospholipase D regulation
Heavy (in vitro) acetylation
Nop1	K85, K102	SPB/Septin	Ribosome biogenesis
Nup116	K796	SPB	Nuclear pore complex
Nup53	K392, K394, K399	SPB	Nuclear pore complex
Pbp1	K433	SPB	RNA processing
Rpl2A	K155	SPB	Ribosome
Rpl35B	K5, K14	SPB	Ribosome
Rpl4A	K343	SPB	Ribosome
Rpl8A	K15	SPB	Ribosome
Rpl8B	K23	SPB	Ribosome
Rps1A	K45	SPB	Ribosome
Rps25A	K21	SPB	Ribosome
Rtt106	K315	SPB	Histone modification
Sfi1	K868, K869	SPB	SPB
Sgf73	K180, K199, K211, K212	Msn4	SAGA complex
Skt5	K634	SPB	Chitin biosynthesis
Spt20	K570, K573, K574	Msn4/Gds1	SAGA complex
Spt5	K317, K318, K803	SPB/Septin	DSIF complex
Stm1	K74	SPB	Stress response
Tef2	K454, K457	SPB	eEF1A
Tif4631	K153	SPB	eIF4G
Vid31	K141	SPB	Transcription regulation
Xrn1	K1272	SPB	RNA processing
Yer138C	K394, K396	SPB/Septin	Retrotransposon

Boldface K represents acetyl lysine residues identified in S. cerevisiae acetylome study (4).

NuA4 Autoacetylation.

To date, the biological significance of Esa1-dependent acetylation has been reported on two NuA4 subunits (Yng2 K170 and Esa1 K262) (31, 33). Additionally, a radioactive in vitro KAT assay using piccolo NuA4 (PicNuA4) suggested Epl1 may also be an Esa1 target (35). Here we confirm these results and identify a plethora of previously undescribed acetylation sites on seven other NuA4 subunits. In total, we identified 42 acetyl lysine residues (30 heavy sites, 6 light sites, and 6 sites modified by both heavy and light acetyl groups) (Tables 1 and 2, and Tables S2 and S3). This identification includes both previously reported sites on Esa1 and Yng2 (among other acetylation sites on these proteins), as well as a total of 20 acetylation sites on Epl1. Furthermore, we identified acetylation sites on Eaf1, Eaf3, Eaf5, Eaf7, Swc4, Arp4, and Tra1 (Tables 1 and 2, and Tables S2 and S3). To confirm hyperacetylation of NuA4 subunits by Esa1, we performed an in vitro KAT reaction on NuA4, stringently purified from yeast, using radiolabeled acetyl CoA. We detected acetylation on Yng2, Eaf5, Eaf1, and Epl1, and one or both of the comigrating proteins Esa1-TAP/Eaf7 and Arp4/Swc4 (Fig. 3A). Using antiacetyl lysine antibodies to detect acetylation, we observed similar results (Fig. S3).

Fig. 3.

NuA4 subunits are acetylated in vivo and in vitro. (A) Radioactive KAT assay to assess NuA4 autoacetylation in vitro. [³H]-acetyl CoA was added directly to bead matrix of a stringently immunopurified NuA4 preparation [Esa1-TAP (YKB440)] versus an untagged control (YPH499) immunopurification. Histone proteins were added to the reaction to serve as a positive control for acetylation. The reactions were separated on a gradient gel, Coomassie-stained to visualize proteins (Right), treated for fluorography, and finally exposed to film for 2 wk (Left). Proteins are identified on the right side and protein size is indicated on the left (kilodaltons). (B) Epl1 acetylation is dependent on Esa1 in vivo. NuA4 was purified from cells grown at 25 °C through Eaf5-TAP (lanes 2–6) relative to an untagged control sample (lane 1, YPH499). Epl1 was expressed from its endogenous locus (lane 2, YKB1042), as a C-terminal HA fusion protein in the presence (lane 4, YKB2876) or absence (lane 3, YKB2862) of the acetyltransferase-deficient esa1-L254P allele, or as a lysine-to-arginine or -glutamine multipoint mutant [EPcA-R; EPcA-Q (K39,345,376,379R or Q) YKB2781 and YKB2782]. Immunopurified (IP) products and whole cell extract (WCE) were separate by SDS/PAGE (7.5%) and subjected to Western blot using the indicated antibodies: antiacetyl lysine (α-AcK), antihistone H4 acetyl lysine (α-AcK H4), antiglyceraldehyde 6-phosphate dehydrogenase (α-G6PDH). (C) Epl1-EPcA-Q reduces NuA4 in vitro KAT activity. NuA4 was purified through Eaf5-TAP (strains described in B). KAT assays were performed using NuA4 preparations equalized for Esa1 and nucleosome purified from HeLa cells. Error bars represent SD from duplicate reactions. NuA4 complexes used in assay were separated by SDS/PAGE and subjected to Western blot using the anti-Esa1 antibody (α-Esa1) (Lower).

Epl1 was by far the most abundantly acetylated protein in vitro in the mChIP-KAT-MS data and we also identified two overlapping heavy/light acetylation sites (K39 and K427) (Tables 1 and 2). To assess the in vivo dependence of Epl1 acetylation on Esa1, we introduced a mutant allele of the essential ESA1 gene, esa1-L254P (19), into a strain expressing EAF5-TAP. esa1-L254P exhibits reduced KAT activity both in vivo and in vitro at the permissive temperature of 25 °C and is catalytically inactive at 37 °C (19). We purified the NuA4 complex through Eaf5-TAP and discovered that acetylation of Epl1 was higher in the ESA1 strain background compared with the level observed in the esa1-L254P mutant, even at the permissive temperature of 25 °C (Fig. 3B, lane 4). Because equal amounts of Epl1-HA (hemagglutinin) copurified with Eaf5-TAP (Fig. 3B, lanes 3 and 4), this finding suggests that Epl1 is a bona fide acetylation target of Esa1 in vivo. Epl1 contains the highly conserved enhancer of polycomb A (EPcA) domain, spanning residues 50–380, which serves to bridge the physical interaction between Esa1 and Yng2 and its expression is sufficient for cell survival (36, 37). To assess the contribution of acetyl lysine residues identified within this domain, we generated arginine and glutamine point mutants to block and mimic lysine acetylation, respectively. The detectable acetylation signal on the Epl1 mutants EPcA-R (K39, 345, 376, 379R) and EPcA-Q (K39, 345, 376, 379Q) was virtually eliminated (Fig. 3B). This result suggests that these acetylation sites account for the majority of the signal (detected by this antiacetyl lysine antibody) in the fraction of Epl1 that copurifies with Eaf5-TAP under these growth conditions. We next sought to determine the function of Epl1 acetylation within the EPcA domain. Although the EPcA-Q and EPcA-R mutants do not impact coimmunopurification with Eaf5-TAP, global histone H4 acetylation was reproducibly decreased in these strains, with the greatest reduction seen in the esa1-L254P and EPcA-Q strains (Fig. 3B, lanes 4 and 6). To confirm that the defect in histone acetylation is direct, we performed an in vitro KAT assay using purified oligonucleosomes from HeLa cells with NuA4 complexes purified from yeast through Eaf5-TAP. In agreement with the decrease in global H4 acetylation, NuA4 containing Epl1-EPcA-Q displayed the greatest reduction in in vitro KAT activity (Fig. 3C). Furthermore, the strain expressing Epl1-EPcA-Q displayed mild temperature sensitivity (Fig. S4), which is consistent with a decrease rather than an abolishment of acetylation activity. This finding suggests the acetylation state of one, all, or a subset or the lysines in the EPcA domain (K39, 345, 376, 379) are contributing to the catalytic activity of NuA4, which may reflect subtle defects in Yng2 association with or misorientation, with Epl1 resulting in reduction in catalysis. Furthermore, although autoacetylation has be largely associated with activation of MYST KAT catalytic activity (31), our work suggests that some lysine acetylation found on KAT complexes may play subtler roles in regulating function, including negatively regulating activity.

NuA4 Is Functionally Connected to the SPB in Yeast.

Copurification of multiple SPB and SPB-associated proteins with NuA4 (Fig. 2A), coupled with NuA4-dependent acetylation on several of these proteins (Table 2), suggests that NuA4 may regulate some aspect of SPB function via direct acetylation. The SPB, equivalent to the mammalian centrosome, is the sole microtubule-organizing center in budding yeast and serves as a platform for nucleation of both nuclear and cytoplasmic microtubules (38), playing a critical role in chromosome segregation (23) (Fig. 4A). A connection between NuA4 and chromosome segregation has been established. Some NuA4 mutant genes are sensitive to the microtubule-destabilizing drug Benomyl (32, 39–44), have elevated rates of chromosome loss (39), and display synthetic genetic interactions with genes that impact microtubule dynamics (BIK1, CIN8, BIM1) and the spindle assembly checkpoint (BUB1, BUB2, BUB3, MAD1, MAD2) (20, 33, 45, 46). To confirm the interaction between NuA4 and SPB, we performed a reciprocal coimmunopurification with cells expressing endogenously tagged SPB components, Cnm67-TAP and Spc72-TAP, as well as a MYC-tagged (c-Myc epitope) NuA4 subunit, Eaf7. IgG-coated magnetic beads were used to immunopurify the TAP-tagged proteins, and Western blot analysis confirmed the copurification of Eaf7-MYC with both SPB bait proteins (Fig. 4B). Taking these data together with the unbiased identification of multiple SPB components through Esa1-TAP in the NuA4 interaction network (Fig. 2A), and a previously reported yeast two-hybrid interaction between Yaf9 and the SPB protein Mps2 (42), we conclude that NuA4 and the SPB physically interact.

Fig. 4.

NuA4 acetylates spindle pole body proteins in vitro and regulates spindle dynamics in vivo. (A) Cartoon of the yeast SPB. The relative positions of all core SPB components are shown with respect to the nuclear and cytoplasmic faces of the nuclear lipid bilayer. Astral and nuclear microtubules (MT) emanate into the cytoplasm and nucleus, respectively. SPB proteins identified in the Cnm67 and/or Spc72 SPB mChIP-KAT-MS experiments are indicated by blue text, otherwise protein names are in black. Acetylated proteins are noted (as described in the legend). (B) NuA4 interacts with the SPB in vivo. Protein extracts expressing the indicated tagged proteins [Bbp1-TAP (YKB1996), Spc72-TAP (YKB1999), Eaf7-MYC (YKB518), Bbp1-TAP Eaf7-MYC (YKB1296), Spc72-TAP Eaf7-MYC (YKB1306)] or an untagged control (no TAP tag) (YPH499) were immunoprecipitated with magnetic beads coated with IgG antibodies that recognize the protein A component of the TAP tag. Total protein extracts (WCE) and immunoprecipitates (α-TAP IP) were resolved by 7.5% SDS/PAGE and subjected to Western blot analysis with anti-MYC and anti-TAP (α-MYC, α-TAP, respectively), as indicated at the right side of the panels. (C) esa1-L254P mutants have microtubule morphology defects. Fixed cells expressing GFP-Tub1 encoding either a wild-type (YKB1233) or mutant allele of ESA1 (YKB1250) (wild-type or esa1-L254P) were examined by fluorescence microscopy. Cells were grown to midlog phase in standard YPD medium supplemented with adenine at 25 °C or 30 °C, as indicated. The average of three experimental replicates is shown and at least 50 large-budded cells with extended microtubules were scored for each replicate. (D) Acetylation regulates Benomyl sensitivity. Wild-type (YPH499), eaf1Δ (YKB44), esa1 (esa-L254P, YKB859), rpd3Δ (YKB1130), rpd3Δeaf1Δ (YKB1154), and rpd3Δ esa1 (YKB2158) cultures were diluted to an OD₆₀₀ of 0.1 and 10-fold serial dilutions were plated on YPD plates containing vehicle control (DMSO) or Benomyl, as indicated.

As the sole microtubule-organizing center in yeast, disruption of SPB function by mutation of any subunit can lead to specific defects in nuclear or astral microtubule dynamics that can be monitored by fluorescence microscopy in cells expressing a green fluorescent protein (GFP)-tagged tubulin protein. To assess specific defects associated with loss of NuA4 acetylation, we examined cells expressing GFP-Tub1 using fluorescence microscopy in either a wild-type strain or a strain harboring the mutant allele esa1-L254P described above (19). In unbudded and small-budded cells, we observed no difference in microtubule morphology between cells expressing the wild-type or mutant alleles of ESA1. However, in large-budded cells in which the mitotic spindle had begun to extend, we observed multiple defects in esa1-L254P mutant cells (Fig. 4C). Specifically, although 90% of large-budded wild-type cells had straight anaphase mitotic spindles, extending continuously to opposing edges of the mother and daughter bud, only 40% of large-budded mutant cells displayed this expected morphology. Rather, in about 5% of cells, the mitotic spindle had a “hooked” phenotype, 30% of cells exhibited bent or broken spindles, and in about 20% of cells the spindle was characterized as “mis-extended,” as it had elongated within the mother cell (Fig. 4C). Similarly, a high-throughput screen to identify mutants with microtubule defects uncovered the hooked microtubule phenotype in multiple nonessential NuA4 mutants (47). In agreement with a role for reversible acetylation playing a relevant role in microtubule function, deletion of RPD3, a proposed NuA4 KDAC antagonist (33), suppresses the Benomyl sensitivity of eaf1∆ mutant cells and partially suppresses the Benomyl sensitivity of esa1 mutant cells, and the single rpd3Δ mutant displays remarkably robust growth on this microtubule-destabilizing drug (Fig. 4D). Further deletion of RPD3 partially rescues the microtubule defects of esa1-L254P mutant cells (Fig. S5). Importantly, our work connects NuA4-dependent acetylation to this phenotype, and moreover indicates the critical importance of lysine acetylation, as only a moderate reduction in NuA4 acetyltransferase activity results in dramatic, pleiotropic defects.

Discussion

Acetylome studies aimed at identifying acetylated lysine peptides in vivo have dramatically expanded our knowledge of acetylation sites (3–10); however, the identity of the KATs responsible for these acetylation events remain elusive. Even in studies comparing the acetylome of wild-type versus KAT mutant strains, differentially acetylated peptides may not represent direct targets of the mutant KATs. Here we describe a unique method called mChIP-KAT-MS that enables the generation of a KAT protein-interaction map enriched for lysine acetylation. Importantly, we have incorporated an isotopically labeled acetyl CoA into the in vitro KAT reaction, thus allowing the differentiation of preexisting in vivo acetyl lysine residues from those resulting from in vitro catalysis. In a nutshell, the mChIP-KAT-MS connects KATs to cellular pathways, targets and acetylated lysine residues and should be universally applicable to the study of KATs in all model systems.

Validation of the mChIP-KAT-MS Methodology.

We chose to validate the mChIP-KAT-MS methodology on the S. cerevisiae KAT NuA4. Before this work, the known protein interactions of NuA4 were minimal. High-throughput affinity purification surveys in yeast (17, 48–50) had identified a limited number of NuA4 interaction partners, and thus most published interactions were defined in the course of directed experiments (20–22, 51). The work we present here dramatically increases our knowledge of physical interaction partners for the NuA4 complex (Fig. 2A), as it encompasses most previously known interactors and identifies over 70 hitherto unknown ones (Fig. 2A). It is important to note that although some of the proteins in the network directly interact with at least one NuA4 subunit [e.g., Msn4 (20)], many others—in particular those belonging to protein complexes (e.g., the SPB)—could have copurified through an intermediate protein in the network. Moving forward, additional studies will be able to distinguish these possibilities; however, proteins modified by heavy acetylation represent an excellent candidate list of direct protein interaction with Esa1. Further, we cannot exclude the possibility that at least some interactors presented in this network copurify with Esa1 as part of PicNuA4, the KAT-competent trimer complex (Esa1-Epl1-Yng2) thought to globally acetylate chromatin (36). Finally, it is also possible that some interactions may be mediated through DNA, although the nonnuclear localization pattern of more than half of the proteins in the network (Fig. 2B) suggests this is not the case for the majority of the interactors.

NuA4 has been linked to a wide variety of cellular processes, but the molecular pathways in which the complex directly functions are largely unclear. Importantly, the physical interactions identified in this study may help to explain the phenotypes and genetic interactions associated with the complex and also implicate NuA4 in new biological pathways. For example, NuA4 has been implicated in chromosome stability (39, 52) and microtubule dynamics (47), but our identification of the physical interaction between NuA4 and the SPB strongly suggests that NuA4 is not mediating its pleotropic effects on the spindle solely through histone acetylation or transcriptional events. Rather, NuA4 may be regulating many aspects of chromosome segregation through multiple nonhistone targets, for which our interactome/acetylome may provide valuable clues for deciphering this complex cellular event.

In addition to the unique NuA4 interactions we confirmed, the network also suggests additional roles for NuA4. For example, we identified a large number of proteins involved in RNA processing, including four proteins that together bind to and regulate polyadenylation of mRNA (Pab1, Pbp1, Pbp4, Lsm12) (26), as well as proteins involved in both mRNA decay (Lsm1) and rRNA maturation (Sik1, Hca4, Pwp1, Rcl1) (Fig. 2A). Intriguingly, Esa1 acetylates Pab1 in vitro at K7, which is a modification we also detect in vivo. These physical interactions are the first indication that NuA4 may participate directly in one or more aspects of mRNA processing. It is of interest to note that both Esa1 and Eaf3 contain chromodomains, best known for their methylated histone-binding capability, but are also able to bind RNA (53). Another pair of protein complexes identified in the NuA4-associated network are the COPI and COPII transport complexes, which promote vesicle formation and cargo transport at the Golgi and endoplasmic reticulum membranes, respectively (54). These interactions may provide functional insight into previously reported synthetic lethal genetic interactions that suggest a role for NuA4 in vesicle-mediated transport, and phenotypic analysis indicating that acetyltransferase-deficient NuA4 mutants have defects in vacuole morphology (20).

mChIP-KAT-MS to Identify Putative Acetylation Targets.

Our mChIP-KAT-MS methodology addresses two hurdles associated with dissecting pathways mediated by lysine acetylation: the reduced likelihood of detecting low-abundance acetyl lysine residues by mass spectrometry and the need to identify the KAT responsible for catalysis. Specifically, the heavy in vitro KAT reaction, making use of ¹³C₂-acetyl CoA, addresses these issues by: (i) enriching the pool of acetyl-lysine residues within the copurified proteins; and (ii) covalently attaching an isotopically labeled acetyl group, which unequivocally distinguishes in vitro catalysis from preexisting modifications that originated in vivo. Therefore, the mChIP-KAT-MS technique not only builds on the success of the KAT in vitro protein acetylation microarray (13) in predicting novel substrates, but also introduces significant advantages. Specifically, our methodology defines both a list of substrates and the specific lysine residues modified by acetylation, because mass spectrometry is used to detect the modification. Moreover, because enzymatic activity demands a physical interaction between the substrate and enzyme in vivo, acetylated proteins identified using mChIP-KAT-MS inherently meet this important constraint, thereby providing a highly relevant functional connection in vivo between NuA4 and the acetylated proteins. As proof for the relevance of the KAT-mChIP-MS in finding relevant in vivo substrates, we show that the in vivo acetylation status of Epl1 (Fig. 3) is dependent on NuA4 acetyltransferase activity.

We postulate the low abundance, and hence minimal MS protein sequence coverage of many NuA4 copurifying proteins, might result in undetected acetyl lysine residues. To this end, we used an inverse application of the mChIP-KAT-MS to identify both in vivo acetylation sites and NuA4-dependent in vitro acetylation sites on SPB and septin subunits, plus Msn4 and Gds1 (Table 2). Although the majority of the in vitro sites we identified have, as of yet, no known biological relevance in vivo, similar approaches using kinases have yielded biologically relevant targets (55). Indeed of the 158 lysine acetylation sites identified in this study (Dataset S4), more than 20% of sites were identified as in vivo sites in a S. cerevisiae acetylome study (4). As one would predict, the overlap between the sites identified in the acetylome and ours was greater for sites that we identified as being both heavy and light peptides (50%). The overlap decreased to 34% for in vivo or light sites and 14% for NuA4-dependent in vitro sites. As it has been predicted that most posttranslational modifications likely have no biological role (56), increasingly it will be important to focus on the acetylation sites identified by multiple approaches. Our ability to identify previously known acetyl lysine residues, including those from both directed studies and unbiased global proteomic approaches, indicates that in vitro acetylation sites identified by mCHIP-KAT-MS should uncover relevant target sites. Directed studies assessing the biological consequence of additional acetylation sites will provide greater insight into this.

We postulate that the mChIP-KAT-MS methodology presented here can serve as a valuable complementary analysis tool in the elucidation of pathways governed by lysine acetylation because of its ability to connect KATs to their substrates and modified lysine residues. Pairing this information with phenotypes previously associated with KATs through genetic analyses provides a powerful system toward unraveling the biological consequences of lysine acetylation. Overall, our data suggest that subjecting purifications of novel interactors to KAT-MS represents a promising tool to identify novel KAT-specific acetylation sites and to ultimately define novel enzyme–substrate relationships.

Materials and Methods

Yeast Strains.

Yeast strains used in this study are listed in Table S3. Genomic deletions or epitope tag integrations made for this study were designed with PCR-amplified cassettes, as previously described (57).

NuA4 mChIP.

NuA4 and its associated protein network were isolated from exponentially growing yeast cultures through mChIP (16) of endogenously TAP-tagged ESA1. Six replicates of the experiment were performed and a flowchart is presented in Fig. S1 with specific details of each replicate. Briefly, cells from 400- to 700-mL midlog phase cultures grown in YPD at 30 °C were collected by centrifugation, washed in 25 mL of ice-cold water, transferred to 1.5-mL Eppendorf tubes, and frozen on dry ice. Cell pellets were resuspended in 300 μL of lysis buffer [100 mM Hepes pH 8.0, 20 mM magnesium acetate, 10% glycerol (vol/vol), 10 mM EGTA, 0.1 mM ETDA, 300 mM sodium acetate, and fresh protease inhibitor mixture (Sigma; P8215)] plus an equal volume of glass beads, and cells were lysed through vortexing (six 1-min blasts with incubation on ice between vortexing). Lysates were subjected to sonication (3 × 20 s; 1-min incubation on ice between each pulse) using a Misonix Sonicator 3000 at setting four. Before centrifugation (10 min, 800 × g, 4 °C), Nonidet P-40 was added to a final concentration of 1% (vol/vol). Next, 40–150 mg of whole-cell extract was incubated with 100–600 μL magnetic beads (Invitrogen; 143.02D) cross-linked to rabbit Ig (IgG) (Chemicon; PP64) as per the manufacturer’s instructions. Following 2 h of end-over-end rotation at 4 °C, the beads were collected on a magnet and washed three times with 1 mL of ice cold wash buffer [100 mM Hepes pH 8.0, 20 mM magnesium acetate, 10% glycerol (vol/vol), 10 mM EGTA, 0.1 mM EDTA, 300 mM sodium acetate, 0.5% Nonidet P-40]. At this point, immunopurified proteins were either eluted from the magnetic beads in 1× loading dye [50 mM Tris, pH 6.8, 2% SDS, 0.1% bromophenol blue, 10% (vol/vol) glycerol] with gentle heating (65 °C for 10 min) (three replicates); eluted from the magnetic beads by incubating in 1 mL of elution buffer (0.5 M NH₄OH, 0.5 M EDTA) at room temperature for 20 min (one replicate); or washed once in 1 mL 1× KAT buffer (50 mM Tris pH 8.0, 50 mM NaCl, 5 mM MgCl₂, 1 mM DTT) and subjected to a KAT reaction using isotopically labeled acetyl CoA (see below; two replicates). The loading dye eluate samples were transferred to new tubes and boiled for 10 min at 95 °C following the addition of β-2-mercaptoethanol to 100 mM. Proteins were separated by SDS/PAGE (NuPAGE Novex 4–12% Bis⋅Tris Gel; Invitrogen, NP0321), visualized by silver stain, and bands were excised and processed for mass spectrometry (see below). The sample in elution buffer was transferred to a new tube and evaporated to dryness using a speed vac and processed on the proteomic reactor (see below).

High-Stringency Purification of NuA4 from Yeast for in Vitro KAT Assays.

NuA4 immunopurified from yeast was carried out using Esa1-TAP as previously described (20), except the complex was eluted from magnetic beads by enzymatic cleavage in TC Buffer [50 mM Tris, pH 8.0, 1 mM DTT, 0.1% Nonidet P-40, 150 mM NaCl, 10% (vol/vol) glycerol] using tobacco etch virus (TEV) protease, which was prepared and generously provided by the laboratory of Jean-François Couture (Ottawa Institute of Systems Biology, University of Ottawa, Ottawa, Canada). Briefly, 1 L of exponentially growing yeast cells (in YPD, at 30 °C) expressing endogenously TAP-tagged ESA1 were lysed and NuA4 was purified in a single step using 600 μL of magnetic beads coupled to IgG. After washing, the NuA4-bead matrix was resuspended in TC buffer (100 μL) to which was added 20 μL of tobacco etch virus (TEV). The cleavage reaction was incubated overnight at 4 °C with end-over-end rotation. Finally, the supernatant was isolated from the beads, aliquoted, and stored at −80 °C. The purity of each NuA4 preparation was assessed by silver stain using 2 μL of TEV-cleaved NuA4 separated by SDS/PAGE (7.5%). Activity of all high-stringency NuA4 complex preparations was confirmed by performing a KAT reaction using 2 μg of chicken core histones (Upstate; 13–107) and 2 μg of standard, unlabeled acetyl CoA (Sigma; A2056) in a final volume of 15 μL. The acetylation signal was assessed by Western blot using an antiacetyl lysine antibody (Upstate; 06–933). An untagged control strain was also taken through immunopurification procedure to ensure the purity of the purification (by silver stain) and to confirm that KAT activity did not nonspecifically associate with the IgG-coated magnetic beads.

In Vitro Heavy KAT Reactions for NuA4 mChIP.

NuA4 in vitro heavy KAT assays using isotopically labeled acetyl CoA (herein referred to as “heavy”) were carried out with immunopurified proteins still bound to the magnetic beads by adding to the protein-magnetic bead matrix 5× KAT buffer (250 mM Tris pH 8.0, 250 mM NaCl, 25 mM MgCl₂, 5 mM DTT), ¹³C₂-acetyl CoA (Isotec; 658650), and stringently purified, exogenous NuA4 (see above). As the exogenous NuA4 preparation included the TEV enzyme, immunopurified proteins still bound to magnetic beads were enzymatically cleaved during the KAT reactions. Heavy KAT reactions were performed on two NuA4 mChIP replicates (Fig. S1) and were carried out in a total volume of 100 uL using 6 μg of ¹³C₂-acetyl CoA and 10 μL of exogenous NuA4 at 30 °C for 1 h with end-over-end rotation. Samples were separated from the beads and processed directly on the proteomic reactor, as described below (58).

Inverse mChIP-KAT-MS.

Cnm67-TAP, Spc72-TAP, Msn4-Tap, Gds1-TAP, Cdc11-TAP, and Shs1-TAP were isolated from 700 mL of exponentially growing yeast cultures in YPD at 30 °C using strains expressing endogenously tagged genes. The mChIP procedure was carried out identically as described above for NuA4 except 100 μL of magnetic beads coupled to IgG and 100 mg of whole-cell lysate were used. Following immunopurification and washing, beads were equilibrated in 1× KAT buffer, and then subjected to a heavy KAT reaction in a final volume of 20 μL including 5 μL highly purified exogenous NuA4, 2 μg of ¹³C₂-acetyl CoA, and 4 μL 5× KAT buffer. Following incubation at 30 °C for 1 h with end-over-end rotation, an equal volume of 2× loading dye was added directly to the beads and the samples were heated at 65 °C for 10 min. The loading dye eluate samples were transferred to new tubes and boiled for 10 min at 95 °C following the addition of β-2-mercaptoethanol to 100 mM. Inverse mChIP-KAT-MS assays were performed once and proteins were separated by SDS/PAGE (NuPAGE Novex 4–12% Bis⋅Tris Gel; Invitrogen, NP0321), and visualized by silver stain. For the Cnm67-TAP and Spc72-TAP, entire lanes were excised and digested (23 bands in total for the two samples). For Cdc11-TAP and Shs1-TAP, 11 bands were processed, and for Gds1-TAP and Msn4-TAP, only the bait protein band was analyzed. To identify the proteome of Gds1-TAP, the mChIP procedure was repeated without the addition of KAT assay, and the entire lane was excised and digested before MS.

MS to Detect Protein Interactions and Acetyl Lysine Residues.

For three replicates of NuA4, samples were separated by SDS/PAGE (NuPAGE Novex 4–12% Bis⋅Tris Gel; Invitrogen, NP0321), the entire lane was excised into 15–20 bands, reduced, alkylated, and digested as previously described (59). The other three replicates of NuA4 mChIP, including the two subjected to the heavy KAT assay, were prepared using a proteomic sample processing device termed the “proteomic reactor” (58). Briefly, the proteomic reactor enables the enrichment, clean up, and chemical and enzymatic processing within capillary tubing packed with either strong anion or strong cation exchange (SAX or SCX, respectively) beads. Processed peptides were eluted using 10-step pH buffers as described previously (58). One of three replicates was subjected to both SAX and SCX proteomic reactor conditions, and the remaining two subjected to only SAX conditions (Fig. S1). LC-MS/MS was performed as previously described (16) using an Agilent 1100 HPLC system (Agilent Technologies) coupled to either an LTQ or an LTQ-Orbitrap XL mass spectrometer (Thermo-Electron) as indicated (Dataset S1). Acetylated lysine residues were identified as previously described (3). MS/MS corresponding to putative lysine acetylation sites were all manually validated.

Lysine Acetyltransferase Assay on Nucleosomes.

NuA4 complexes were purified from the indicated Eaf5-TAP strains as described above and eluted from the IgG magnetic beads by incubation with TEV protease (an untagged strain was used for mock purification). Histone acetyltransferase reactions were performed in 15-µL final volume of 50 mM KCl, 50 mM Tris (pH 8), 1 mM DTT, 5% (vol/vol) glycerol, 10 mM Na-Butyrate, and 0.1 mM EDTA with 500 ng of purified H1-depleted oligonucleosomes (from HeLa cells) and 1.25 µL of [³H] acetyl-CoA (0.1 µCi/µL, 4.9 Ci/mmol) for 60 min at 30 °C. Amounts of wild-type and mutant complexes used in the reaction were normalized by Western blot based on the Esa1 signal. Each reaction was spotted onto p81 filters, which were then washed three times with 50 mM Na Carbonate (pH 9.2). The amount of incorporated [³H] acetyl was determined using a scintillation counter. Error bars represent SD from duplicate reactions.

Additional methods are provided in the SI Materials and Methods.