Chromatin Regulation by the NuA4 Acetyltransferase Complex Is Mediated by Essential Interactions Between Enhancer of Polycomb (Epl1) and Esa1

Abstract

Enzymes that modify and remodel chromatin act in broadly conserved macromolecular complexes. One key modification is the dynamic acetylation of histones and other chromatin proteins by opposing activities of acetyltransferase and deacetylase complexes. Among acetyltransferases, the NuA4 complex containing Tip60 or its Saccharomyces cerevisiae ortholog Esa1 is of particular significance because of its roles in crucial genomic processes including DNA damage repair and transcription. The catalytic subunit Esa1 is essential, as are five noncatalytic NuA4 subunits. We found that of the noncatalytic subunits, deletion of Enhancer of polycomb (Epl1), but not the others, can be bypassed by loss of a major deacetylase complex, a property shared by Esa1. Noncatalytic complex subunits can be critical for complex assembly, stability, genomic targeting, substrate specificity, and regulation. Understanding the essential role of Epl1 has been previously limited, a limitation now overcome by the discovery of its bypass suppression. Here, we present a comprehensive in vivo study of Epl1 using the powerful tool of suppression combined with transcriptional and mutational analyses. Our results highlight functional parallels between Epl1 and Esa1 and further illustrate that the structural role of Epl1 is important for promotion of Esa1 activity. This conclusion is strengthened by our dissection of Epl1 domains required in vivo for interaction with specific NuA4 subunits, histone acetylation, and chromatin targeting. These results provide new insights for the conserved, essential nature of Epl1 and its homologs, such as EPC1/2 in humans, which is frequently altered in cancers.

EUKARYOTIC genomes are packaged into chromatin, which is composed of nucleosome units containing DNA wrapped around a histone octamer (Kornberg and Lorch 1999). Chromatin is subject to multiple, diverse modes of post-translational regulation that have many established roles, including functions in recombination, DNA damage repair, and transcription (Kouzarides 2007). Acetylation is one such post-translational modification that regulates chromatin function, mediated by the opposing enzymatic activities of lysine acetyltransferases (KATs/HATs) and deacetylases (KDACs/HDACs) (Campos and Reinberg 2009). HATs often exist in large multimeric complexes, such as the deeply conserved NuA4 complex (Doyon et al. 2004).

In humans, the essential catalytic subunit of NuA4, KAT5/Tip60, along with additional essential subunits such as EPC1/2, are associated with several carcinomas (Avvakumov and Côté 2007; Lafon et al. 2007; Nakahata et al. 2009; Biankin et al. 2012; Huang et al. 2014), suggesting their importance for controlled cellular growth. Much of the basic understanding of NuA4 comes from studies performed in Saccharomyces cerevisiae. NuA4 in yeast includes six essential subunits: Esa1 (Tip60 ortholog), Epl1 (EPC1/2 ortholog), Tra1, Arp4, Act1, and Swc4, all of which are broadly conserved. NuA4 primarily acetylates histones H4 and H2A in vivo (Smith et al. 1998; Clarke et al. 1999) along with noncanonical histones, such as H2A.Z (Keogh et al. 2006), and >250 nonhistone substrates (Lin et al. 2009; Yi et al. 2012; Mitchell et al. 2013; Downey et al. 2015), including 91 essential proteins.

There are two distinct smaller complexes containing NuA4 subunits: piccolo-NuA4, composed of Esa1, Epl1, Yng2, and Eaf6 (Boudreault et al. 2003; Mitchell et al. 2008; Rossetto et al. 2014), and the TINTIN triad of Eaf5/7/3 (Cheng and Côté 2014; Rossetto et al. 2014). Piccolo-NuA4 is thought to also exist alone (Ohba et al. 1999; Boudreault et al. 2003) and is sufficient for broad nucleosome acetylation in vitro, whereas the NuA4 holo-complex is required for more targeted NuA4 functions such as DNA damage repair and transcriptional activation (Figure 1A) (Bird et al. 2002; Boudreault et al. 2003; Selleck et al. 2005; Friis et al. 2009).

Figure 1.

The requirement for two essential NuA4 subunits is bypassed by disassembly of Rpd3L. (A) The NuA4 histone acetyltransferase complex contains six essential subunits (underlined). The NuA4 holo-complex has targeted functions including roles in transcription and DNA damage repair, whereas the smaller piccolo-NuA4 complex is a broadly acting acetyltransferase complex. Piccolo-NuA4 contains the catalytic subunit Esa1, along with the essential subunit Epl1 and two nonessential subunits. NuA4 complex schematic is based on: Boudreault et al. (2003); Bittner et al. (2004); Doyon et al. (2004); Mitchell et al. (2008); Chittuluru et al. (2011); and Rossetto et al. (2014). (B) A screen of all essential NuA4 subunits for bypass potential. Double mutant analysis of esa1∆ sds3∆ (LPY20724), epl1∆ sds3∆ (LPY20609), act1∆ sds3∆ (LPY20974), arp4∆ sds3∆ (LPY20617), swc4∆ sds3∆ (LPY20611), and tra1∆ sds3∆ (LPY20443) revealed that only epl1∆, like esa1∆, could be bypassed by loss of SDS3, which encodes a central component of the Rpd3L deacetylase complex. Serial dilutions on the plasmid counterselective medium at 24° (and 30°, Figure S1 in File S1) illustrate that epl1∆ sds3∆, like esa1∆ sds3∆, survived without a plasmid-based copy of its corresponding essential gene.

Because Esa1 is essential, much of our early understanding of it came from studying hypomorphic alleles, where Esa1 is only partially or conditionally functional (Clarke et al. 1999; Decker et al. 2008). Recently, the first bypass suppressor of Esa1 was identified, where esa1∆ is rescued by loss of the Rpd3L HDAC complex (Torres-Machorro and Pillus 2014). This bypass of Esa1 is promoted by establishing a relatively balanced cellular acetylation state. The discovery of this bypass allowed for the first studies in which cells were completely depleted of Esa1.

Among the six essential NuA4 subunits only Esa1 and Epl1 are found in the very active smaller piccolo complex (Galarneau et al. 2000). Epl1 was first reported as the yeast ortholog of Drosophila melanogaster Enhancer of Polycomb E(Pc), which can function as a suppressor of position-effect variegation and can increase the homeotic phenotype of Polycomb group mutations (Sinclair et al. 1998; Stankunas et al. 1998). Epl1 and E(Pc) are broadly conserved and are orthologous to the EPC1/2 paralogs in humans (Shimono et al. 2000; Doyon et al. 2004).

It is noteworthy that despite its conservation and discovery nearly two decades ago, Epl1 function has been only minimally characterized, primarily based on low-dosage variants, limited in vitro analyses, and most recently when its partial structure bound to nucleosomes was solved (Boudreault et al. 2003; Selleck et al. 2005; Chittuluru et al. 2011; Huang and Tan 2012; Xu et al. 2016). Phenotypes of EPL1 depletion are quite similar to those of impaired ESA1. These include roles in cell-cycle progression through G2/M, H4 acetylation, DNA damage repair, telomeric silencing, and autophagy (Boudreault et al. 2003; Yi et al. 2012).

Epl1 bridges Esa1 and the Yng2 and Eaf6 subunits to the larger NuA4 complex (Boudreault et al. 2003; Mitchell et al. 2008; Rossetto et al. 2014). The C-terminus of Epl1 contacts the NuA4 holo-complex through Eaf1 (Auger et al. 2008), but only the N-terminus (the EPcA domain) is essential for viability (Boudreault et al. 2003), suggesting that integrity of piccolo-NuA4 is crucial.

Despite progress made in earlier studies, the essential function of Epl1 in vivo has remained unknown in S. cerevisiae and metazoans alike. Here, we report that Epl1 can be bypassed by the same loss of the Rpd3L deacetylase complex observed for Esa1 and present a comprehensive in vivo analysis of Epl1 made possible only by its bypass suppression. Although Epl1 has no known catalytic activity, we find striking phenotypic and transcriptional similarity between esa1∆ and epl1∆ mutant strains under bypass conditions, suggesting coordinated function and activity. Through mutational analysis of Epl1, we provide evidence that Epl1’s essential function is directly linked to physical contact with Esa1, such that without the Epl1-Esa1 structural interaction, Esa1 is no longer fully active. These new findings thus help to illuminate the essential coordinated activity of a MYST-family acetyltransferase and its broadly conserved binding partner.

Materials and Methods

Yeast strains and plasmids

Strains, plasmids, and oligonucleotides are listed in Supplemental Material, Tables S1–S3 in File S1. EPL1 and ESA1 mutant strains were constructed initially with covering plasmids (pLP3189 or pLP796). EPL1-13MYC-HISMX6 was derived from LPY21686 (QY237), and integrated at the endogenous EPL1 locus in LPY79 by amplification of the 13MYC-HISMX6 tag with oLP2196 and oLP2172. EPL1-13MYC-HISMX6 was similarly cloned into pLP74, as pLP3337, by amplifying EPL1-13MYC-HISMX6 from LPY21686 with oLP2169 and oLP2180, digested with HinDIII, and ligated into pLP74. Epl1 mutant plasmids were constructed using NEB Q5 site-directed mutagenesis on pLP3337. Mutants were tested for dominance by transforming into a wild-type (WT) strain. The mutations were then integrated at the EPL1 locus in diploid WT W303 and dissected. Mutagenesis was verified by sequencing both prior to and after integration. Strains were backcrossed prior to use.

Growth assays

Plate-based assays were performed using fivefold serial dilutions on standard media as described (Chang and Pillus 2009). For temperature and DNA damage assays, cultures were grown at 24° in SC for 1–3 days and then plated with starting concentrations normalized to one A₆₀₀ unit and imaged after 2–5 days. Methyl methanesulfonate (MMS) sensitivity was assayed at 0.0075% in SC. Hydroxyurea (HU) sensitivity was assayed at 0.05 M in SC. Camptothecin (CPT) sensitivity was assayed at 7 μg/ml in SC (DMSO as vehicle control) prepared with 100 mM phosphate buffer (pH 7.5). Cultures for 5-fluoroorotic acid (5-FOA) assays were grown for 2 days at 30° to reach saturation, normalized to starting dilutions of 5–7 A₆₀₀ units, and imaged after 4–6 days after plating. 5-FOA assays performed in the W303 background were plated on 10% glucose; all other plate-based assays were performed with standard 2% glucose.

Flow cytometry

Strains grown at 24° in SC for 1–2 days were diluted and grown to mid-log. One milliliter of exponentially growing cells (∼3 × 10⁷ cells) was fixed with cold 70% ethanol and prepared for flow cytometry, staining with propidium iodide (Chang et al. 2012). Thirty thousand cells were analyzed using a BD Accuri C6 Flow Cytometer.

Lysate preparation

Strains grown at 24° in SC were collected in mid-log for whole-cell extract preparations by bead-beating as described (Clarke et al. 1999). Fractionation was performed using spheroplasting, detergent-based lysis, and differential centrifugation (Liang and Stillman 1997) to yield whole-cell extract, and soluble and crude chromatin fractions. Lysates were briefly sonicated prior to immunoblot analysis.

Immunoprecipitations

Strains were grown for 1–2 days in 3 ml of SC at 24°, expanded to 10 ml, then diluted into 200 ml for growth and collected in mid-log phase. After pelleting and a phosphate-buffered saline (PBS) wash, cells were lysed by bead-beating in 1 μl of cold immunoprecipitation (IP) lysis buffer per A₆₀₀ OD of cells (50 mM HEPES-KOH pH 7.5, 100 mM NaCl, 0.25% NP-40, 1 mM EDTA, 10% glycerol, and protease, phosphatase, and deacetylase inhibitors). The lysate was cleared then incubated with rotation for 3 hr with 5 μl of anti-Myc. IP mixtures were incubated for 50 min with 75 μl of Dynabeads Protein G (Thermo Fisher Scientific), prewashed with lysis buffer. Protein–antibody–bead conjugates were washed twice with lysis buffer and twice with wash buffer (50 mM HEPES-KOH pH 7.5, 150 mM NaCl, 1 mM EDTA) prior to elution by boiling for 10 min in 40 μl of sample loading buffer (250 mM Tris-HCL pH 6.8, 10% SDS, 30% glycerol, 5% β-mercaptoethanol, 0.02% bromophenol blue).

Immunoblots

To evaluate histones, proteins were separated using 15% SDS-PAGE, transferred to a 0.2-μm nitrocellulose membrane, and probed with: anti-H4K8Ac (1:2000, EMD Millipore, Darmstadt, Germany), anti-H4K5Ac (1:2000, Millipore), anti-H4K12Ac (1:2000, Active Motif), anti-H4 (1:2000, Active Motif), anti-H3K9/K14Ac (1:10,000, Upstate), and anti-H3 (1:2500, Abcam). Other proteins were separated on 7.5, 8, or 10% SDS-PAGE or for IP samples, 8–16% Novex Wedgewell Tris-Glycine gels (Thermo Fisher Scientific), transferred to a 0.2-μm nitrocellulose membrane and probed with: anti-Myc (1:2500 for detection of Epl1, 1:5000 for detection of all other Myc-tagged proteins) (Evan et al. 1985), anti-HA (1:1000, Covance), anti-Yng2 (1:1000, graciously provided by S. Tan), anti-Sir2 (1:10,000) (Garcia and Pillus 2002), anti-Pgk1 (1:20,000), and anti-β-tubulin (1:20,000) (Bond et al. 1986).

RNA-seq sample preparation and analysis

RNA was prepared in biological triplicate using hot-phenol extraction from mid-log cells grown in SC at 24°. RNA was DNase treated (Ambion). Quality was evaluated by gel electrophoresis and bioanalyzer (Agilent). Samples were depleted of rRNA (Ribo-zero Magnetic Gold Yeast, Epibio), and libraries were prepared (Tru-seq Stranded total RNA, Illumina). Twenty-four samples were sequenced with 50-bp single-reads on one lane of the HiSeq 2500 (Illumina), yielding a total of 287.52 million reads passing the quality filter.

Upon data generation, library adaptors were trimmed computationally with Cutadapt (Martin 2011), and reads were mapped to Repbase (Bao et al. 2015). Any reads mapping to Repbase were excluded from further analysis. The remaining reads were mapped to SacCer3 (Engel et al. 2014) with STAR (Dobin et al. 2013). Differential expression was assessed with DESeq2 (Love et al. 2014), and transcripts with a log₂(fold change) ≥1 or ≤−1 and P-adj ≤0.05 were called as differentially expressed. Further data analysis and visualization was completed using R computing software (R Development Core Team 2015) and the ggplot2 package (Wickham 2009).

qPCR validation

Select transcripts were validated using RT-qPCR. Briefly, cDNA was synthesized in biological triplicate from the RNA samples (TaqMan Reverse Transcriptase kit, Life Sciences) and qPCR was performed using EvaGreen qPCR Master Mix (Lambda bio) on an MJ Research Opticon 2 to determine levels relative to the SCR1 control. Significance was tested and assigned based on P-values calculated by a Student’s t-test.

Data availability

Strains and plasmids are available upon request. Gene expression data have been deposited in the Gene Expression Omnibus with accession number GSE92774.

Results

Bypass and function of essential piccolo-NuA4 subunits

The finding that the essential requirement for Esa1 could be bypassed by loss of the Rpd3L deacetylase due to deletion of SDS3 (Torres-Machorro and Pillus 2014) was significant because it marked the first condition where cellular viability was maintained without an essential NuA4 subunit. Similar to bypass suppression of ESA1, identification of other NuA4 bypass suppressors could facilitate in vivo analysis of these essential chromatin factors.

To test the extent to which disruption of Rpd3L by sds3∆ could bypass loss of genes encoding the essential NuA4 subunits (Esa1, Epl1, Act1, Arp4, Swc4, Tra1, underlined in Figure 1A), double mutants were constructed. Initially, each double mutant was recovered with a URA3-marked plasmid carrying the corresponding wild-type NuA4 gene. The strains were then challenged by plating on 5-FOA, which is toxic to cells expressing URA3. Growth on 5-FOA reveals mutant cells that can survive without the corresponding wild-type covering plasmid. Of the five new double mutants tested, only epl1∆ could be bypassed by sds3∆; all remaining essential NuA4 subunits were still required for viability (Figure 1B and Figure S1 in File S1). The recovery of epl1∆ was of particular interest because of Epl1’s limited in vivo characterization in any species and its close structural proximity to the catalytic Esa1 in piccolo-NuA4/NuA4.

Previous in vitro and in vivo studies of Epl1 were reported using two hypomorphic alleles and repressible expression. Epl1 was shown to have roles similar to Esa1, such as in histone H4 acetylation, DNA damage repair, and cell-cycle progression (Boudreault et al. 2003). To evaluate potential distinctions between Epl1 and Esa1 function in vivo, we examined phenotypes of the bypass strains. The esa1∆ epl1∆ sds3∆ triple mutant was viable (Figure 2A) and thus included in the phenotypic analysis.

Figure 2.

Bypass of EPL1 is phenotypically akin to esa1∆ sds3∆. (A) The epl1∆ sds3∆ (LPY21299) bypass strain shared growth defect and temperature-sensitivity phenotypes of esa1∆ sds3∆ (LPY21631), relative to both WT (LPY79) and sds3∆ (LPY20877). Likewise, in the triple mutant, loss of both EPL1 and ESA1 (LPY21751) had similar growth defects to loss of either essential subunit alone. (B) Bypass strains were surveyed at 24° for DNA damage, revealing sensitivity to all agents tested, relative to growth control (A), and the DMSO-vehicle control for CPT. (C) Histone H4 acetylation is significantly reduced upon loss of ESA1 and/or EPL1 relative to WT and sds3∆. Two acetylation isoforms were probed as representatives for acetylation. Histone H3 acetylation remained unchanged relative to WT upon Esa1 or Epl1 mutation, highlighting the effect on histone H4 acetylation as a NuA4 target rather than the H3–H4 tetramer. (D) Cell cycle profiles demonstrated that loss of Esa1 and Epl1 resulted in a G2/M delay. All experiments were completed in three or more independent assays. Representative results from each are shown here.

The NuA4 bypass strains were surveyed for growth across a range of temperatures: all showed extreme sensitivity to high temperatures, and general growth defects at lower temperatures (Figure 2A). The epl1∆ sds3∆ and esa1∆ epl1∆ sds3∆ mutants were sensitive to DNA-damaging agents (Figure 2B) as shown previously for esa1∆ sds3∆ (Torres-Machorro and Pillus 2014). These strains were also sensitive to the vehicle control for CPT, DMSO, which has been shown to broadly decrease cellular proliferation (Kakolyri et al. 2016). This sensitivity mirrors that which has been identified for mutants of other chromatin regulators (Gaytán et al. 2013; Sadowska-Bartosz et al. 2013). As illustrated by H4K8 and H4K12 acetylation, EPL1 bypass strains had low levels of histone H4 acetylation relative to WT and sds3∆. By contrast, H3 acetylation remained unaffected (Figure 2C). Finally, loss of EPL1 resulted in a similar defect in cell-cycle progression as loss of ESA1, characterized by a G2/M delay (Figure 2D).

Thus, loss of EPL1, despite not encoding acetyltransferase activity, had similar phenotypic and functional consequences as loss of ESA1. The observation that no distinct phenotypes were found when both ESA1 and EPL1 were lost, as compared to when only a single subunit was bypassed, further emphasized a high degree of functional overlap.

ESA1 and EPL1 bypass strains have nearly identical gene expression profiles

NuA4, and Esa1 specifically, contribute to the transcriptional regulation of ribosomal protein genes and many other targets genome-wide (Reid et al. 2000; Durant and Pugh 2006; Uprety et al. 2015). ESA1 and its metazoan counterparts have roles in heterochromatin regulation, gene expression, and DNA damage repair (Clarke et al. 2006). Mutation or transcriptional repression of EPL1 leads to similar phenotypes as those of ESA1 mutants (Sinclair et al. 1998; Boudreault et al. 2003).

We asked if loss of EPL1 mirrored loss of ESA1 during bypass at the level of transcription. We performed RNA-sequencing and found that epl1∆ sds3∆, esa1∆ sds3∆, and esa1∆ epl1∆ sds3∆ had extremely similar transcriptomes. In fact, hierarchical clustering analysis illustrates that the similarity between these mutants is nearly equivalent to that of biological replicates, such that the different mutants cluster in the same group as, and interspersed within, the replicates of each mutant (Figure 3A). It should be noted that this intermixed clustering of mutants and replicates is not due to high variability between biological replicates, as the given correlation coefficients are >0.99. Rather, the clustering highlights the striking similarity between the three NuA4 bypass mutants.

Figure 3.

ESA1 and EPL1 bypass strains have nearly identical gene expression profiles. (A) Transcriptome analysis of bypass strains (LPY21299, LPY21631, and LPY21751) demonstrated a significantly high degree of similarity. Pairwise correlation analysis by Spearman’s correlation coefficient was performed among strains and biological replicates shown by hierarchical clustering and a correlation heatmap. The three bypass strains clustered with near-perfect correlation coefficients. The biological replicates had analogous degrees of similarity, yet were clearly distinct from WT (LPY79) and sds3∆ (data not shown). (B) Differential expression analysis depicted by Venn diagram, highlights the similarity between ESA1 and EPL1 mutants. Analysis of 7126 transcripts that passed quality-control filters demonstrated that only five were differentially expressed above/below the threshold of log₂(fold change) ±1, respectively, and a false discovery rate (FDR) adjusted P-value ≤ 0.05. (C) Volcano plot illustrating the fold change and significance of transcripts in analysis of differential expression between esa1∆ sds3∆ and epl1∆ sds3∆. A negative fold change indicates down-regulation in esa1∆ sds3∆ relative to epl1∆ sds3∆. Transcripts meeting the significance threshold are in red with gene name indicated. EPL1 (gray) was not differentially expressed above threshold. (D) Fold change between WT and epl1∆ sds3∆ and between WT and esa1∆ sds3∆ is plotted in a smooth scatter plot, with color intensity corresponding to density of individual points. Linear regression analysis is indicated by R², with the major outlying transcripts labeled. All differential expression analysis is of three biological replicates for each strain: WT (LPY79), esa1∆ sds3∆ (LPY21631), epl1∆ sds3∆ (LPY21299), and esa1∆ epl1∆ sds3∆ (LPY21751).

Expression analysis of 7126 transcripts in the ESA1 and EPL1 bypass strains revealed that just over 1000 transcripts are differentially expressed between WT and esa1∆ sds3∆ and a similar number between WT and epl1∆ sds3∆. However, only five transcripts were differentially expressed between ESA1 and EPL1 bypass strains (Figure 3B). Notably, these five transcripts were only differentially expressed between esa1∆ sds3∆ and epl1∆ sds3∆; there were no transcripts differentially expressed between the triple esa1∆ epl1∆ sds3∆ and either esa1∆ sds3∆ or epl1∆ sds3∆. Further analysis of these five differentially expressed transcripts by volcano plot (Figure 3C) illustrates that two of the differentially expressed transcripts, ESA1 and HIS3, were expected due to the genetic background of the strains: these strains are auxotrophic for histidine and contain a his3-11 mutation, affecting the expression of HIS3. However, in the esa1∆ sds3∆ strain, ESA1 is replaced with HIS3, thereby restoring HIS3 transcription and explaining the observed differential expression. Although ADE17 was not differentially expressed at statistical significance by RT-qPCR, expression trended toward its down-regulation in esa1∆ sds3∆ as compared to epl1∆ sds3∆. The ATG19 and YHK8 differential expression was validated by RT-qPCR (Figure S2 in File S1), and in fact, YHK8, a largely uncharacterized open reading frame (ORF), is greater than sixfold up-regulated in epl1∆ sds3∆ as compared to esa1∆ sds3∆ by RT-qPCR. However, all three of these transcripts are only differentially expressed by one- to twofold by RNA-sequencing, and there is no functional theme underlying and unifying their differential expression, nor are the corresponding genes directly bound by Esa1 (Robert et al. 2004).

An analysis examining the differential expression of transcripts between WT and esa1∆ sds3∆ plotted against the differential expression of transcripts between WT and epl1∆ sds3∆ was also telling. Plotting the log₂ (fold change) of all transcripts relative to WT in esa1∆ sds3∆ vs. that in epl1∆ sds3∆ illustrated a high correlation between differential expression in esa1∆ sds3∆ and in epl1∆ sds3∆, both relative to WT (Figure 3D). As such, transcripts that differ between WT and esa1∆ sds3∆ also differ, and to a similar magnitude, between WT and epl1∆ sds3∆. Thus, the transcriptional profiles of Epl1 and Esa1 bypass conditions are virtually identical, despite their distinct noncatalytic and catalytic roles in NuA4.

Epl1 promotes the chromatin association of Esa1

Both phenotypic and transcriptional analyses of epl1∆ and esa1∆ emphasize their similarity, despite the overt difference of Esa1’s catalytic activity. To further probe distinctions between the roles of Epl1 and Esa1, and to determine the nature of EPL1’s essential function, we considered the in vitro characterization of Epl1, which reported that it associates with the nucleosome core particle to promote Esa1’s enzymatic activity (Chittuluru et al. 2011). We could ask for the first time if Epl1 drives Esa1’s chromatin association in vivo, and if Esa1 would remain chromatin associated in the absence of Epl1.

To test the role of Epl1 in targeting Esa1 to chromatin, subcellular fractionation (Liang and Stillman 1997) and immunoblotting were performed (Figure 4). Controls included probes for the chromatin-associated protein Sir2 and the glycolytic enzyme Pgk1, a predominantly cytoplasmic protein. In WT, sds3∆, and epl1∆ sds3∆ strains, Sir2 was primarily localized to the chromatin (C) fraction, whereas Pgk1 was more enriched in the soluble (S) fraction (Figure 4, A and B). In contrast, whereas Esa1 is largely localized to the chromatin fraction in WT and sds3∆, it becomes shifted to the soluble fraction upon loss of EPL1 and depleted from chromatin. Notably, this shift in association is specific for Esa1, as Sir2 remains chromatin associated.

Figure 4.

Epl1 is required for stable chromatin association of Esa1. (A) Subcellular fractionation assays reveal that in the absence of Epl1, a fraction of Esa1 is released from chromatin. Cells were collected and lysed for whole-cell extracts (W). Additional fractionation was performed to yield soluble (S) and crude chromatin (C) fractions. In WT cells (LPY21568), the majority of Esa1 is associated with the chromatin fraction, much like Sir2. However, in epl1∆ sds3∆ (LPY21596), Esa1 is shifted to the soluble fraction, analogous to the Pgk1 control. A brief chemical cross-link prior to lysis and fractionation was performed in parallel (Figure S3 in File S1). (B) The sds3∆ single mutant (LPY21579) alone does not alter Esa1 chromatin association, as illustrated by subcellular fractionation followed by immunoblotting for Esa1, and the Sir2 and Pgk1 controls. (C) Swc4 remains chromatin associated upon loss of EPL1. Subcellular fractionation demonstrates that Swc4, another essential NuA4 subunit, remains chromatin associated in epl1∆ sds3∆ (LPY21942), consistent with WT (LPY22201) and much like the Sir2 control. (D) The sds3∆ single mutant (LPY22202) alone also does not affect Swc4 chromatin association.

NuA4 contains subunits that have chromatin activity independent of NuA4, including several that contain their own chromatin targeting domains. We sought to determine if the newly defined role for Epl1 in promoting chromatin association of Esa1 in vivo was extended to other NuA4 subunits, and therefore, if its loss might have more widespread consequences. To test this possibility, we selected Swc4 for its essential nature, dual-role in NuA4 and the Swc4 chromatin-remodeling complex, and its SANT domain (Krogan et al. 2004). We found that Swc4 remained chromatin associated in the absence of EPL1 (Figure 4C), demonstrating that loss of Epl1 did not broadly affect all NuA4 subunits. Like Esa1 localization, Swc4 is unaffected by sds3∆ alone, and localization patterns in sds3∆ mirror WT (Figure 4D). Thus, Epl1 is important specifically for the association of Esa1 with chromatin, and its loss does not generally disrupt chromatin association of two other chromatin proteins with distinct functions in transcription and remodeling.

Defining the critical regions of Epl1 in vivo

Due to its essential nature and a limited number of hypomorphic alleles (Boudreault et al. 2003), much of Epl1’s characterization has been performed in vitro. Accordingly, we wanted to determine if Epl1’s chromatin-association function was essential, and concurrently, which regions were most critical for promoting Epl1’s essential role. Several prior studies defined regions of Epl1 essential for viability and in vitro activity, such as the conserved EPcA N-terminal domain (Boudreault et al. 2003; Selleck et al. 2005; Chittuluru et al. 2011; Huang and Tan 2012). This is in contrast to the more variable, nonessential C-terminus. We used mutational analysis to construct four distinct Epl1 mutants that targeted the EPcA domain, and one mutant targeting the C-terminus (Figure 5A).

Figure 5.

Defining functional regions of Epl1 in vivo. (A) Epl1 contains a conserved and essential EPcA domain, and a more variable and nonessential C-terminus. EPcA contains three subdomains that were previously classified by in vitro assays (Boudreault et al. 2003; Selleck et al. 2005; Chittuluru et al. 2011) and validated in recent structural studies (Xu et al. 2016). NP (nucleosome core particle) interacts with the nucleosome core particle, E (Esa1) makes physical contact with Esa1, and Y (Yng2) makes contact with the nonessential piccolo-NuA4 subunits Yng2 and Eaf6. Although the nomenclature for these domains follows that set by previous studies, it should be noted that the residues in the defined domains are not identical to past studies, varying by one or two amino acids. The C-terminus does not contain any conserved domains; however, in vitro it has a structural role in tethering the piccolo-NuA4 subunits to the NuA4 holo-complex by interacting with Eaf1. (B) Evaluation of dominance and viability of the Epl1 mutants. Serial dilution assays reveal that at 24° the mutants are not dominant (Control). In the epl1∆ mutant, only the EPL1-NP (LPY22120) construct supports viability of epl1∆, although cells have a significant reduction in fitness. Epl1 mutants for each of the other putative subunit-interaction domains fail to support viability, demonstrating an essential in vivo function for each (LPY22012, LPY22001, LPY22084). Confirming previous results, epl1-Ct∆ (LPY22010) is viable and robust. The epl1∆ sds3∆ (LPY21071) and epl1∆ (LPY20759) strains are plated as viable and inviable controls, respectively. (C) The EPL1 mutations do not have gross effects on protein levels of either Epl1 or Esa1 in whole cell lysates prepared from exponentially growing cells. A representative blot for one of at least three independently prepared lysates is shown. Eight strains were assayed: EPL1-13MYC ESA1-3HA (LPY22231), EPL1-13MYC sds3∆ ESA1-3HA (LPY22232), epl1-NP∆-13MYC sds3∆ ESA1-3HA (LPY22213), epl1-E∆-13MYC sds3∆ ESA1-3HA (LPY22208), epl1-Y∆-13MYC sds3∆ ESA1-3HA (LPY22226), epl1-Nt∆-13MYC sds3∆ ESA1-3HA (LPY22209), epl1-Ct∆-13MYC sds3∆ ESA1-3HA (LPY22211), and the WT no-tag control (LPY79).

Given that, to our knowledge, this represents the most comprehensive in vivo structure–function mutational analysis of Epl1 to date, we next moved to assess the essential nature of each of the subdomains. The EPcA domain was shown earlier to be essential, to interact with the nucleosome core particle in vitro, to contribute to substrate specificity, and together with Yng2, to position Esa1 to acetylate nucleosomes (Boudreault et al. 2003; Selleck et al. 2005; Chittuluru et al. 2011; Huang and Tan 2012; Lalonde et al. 2013). To date, no in vivo assessment has been reported for the requirement for all subdomains within EPcA. We found that among the mutants in the essential N-terminus, only epl1-NP∆ is viable (Figure 5B). However, its growth was not as robust as the epl1-Ct∆ strain. Therefore, although important, the NP subdomain of Epl1 is not essential.

As many of the Epl1 mutations were not viable in an otherwise WT-background (epl1-E∆, epl1-Y∆, epl1-Nt∆), and those that were viable were not robust (epl1-NP∆, epl1-Ct∆), we capitalized on the resource of bypass suppression, using the sds3∆ background to further study the functional consequences of the EPL1 mutations in vivo. To begin, we found that the mutations did not significantly disrupt either Epl1 or Esa1 protein levels (Figure 5C). This suggests that there are no gross changes in protein stability, although effects due to changes in protein conformation remain possible.

We next evaluated the phenotypic consequences of the EPL1 domain mutants in the sds3∆ background, such that the only Epl1 that is expressed is the mutant version, integrated at the genomic locus. In these bypass conditions, mutants of the Epl1 subunit interaction domains (epl1-E∆ and epl1-Y∆) are sensitive to high temperature (Figure 6A), and DNA damage (Figure 6B). Accordingly, complete loss of EPL1 or loss of the entire essential EPcA domain (epl1-Nt∆) is phenotypically similar to loss of either of the subunit interaction domains (epl1-E∆ and epl1-Y∆) alone. In contrast and consistent with epl1-NP∆ sufficiency for viability, this mutant has the most robust growth in bypass conditions when challenged with higher temperatures and DNA-damaging agents. These results suggest that the residues of Epl1 that interact with other subunits in vitro (Epl1-E and Epl1-Y) are most critical for both viability and function in vivo, and that in bypass conditions, loss of either of these regions is as detrimental to cellular fitness as loss of the entire gene. In contrast, the domain previously defined as critical for nucleosome targeting (Epl1-NP) in vitro, although important, is less critical during bypass suppression.

Figure 6.

Subunit interaction domains are critical for Epl1 function in vivo. (A) Under sds3∆ bypass conditions, EPL1 mutants were surveyed for growth on SC medium: epl1-NP∆ (LPY22111), epl1-E∆ (LPY22017), epl1-Y∆ (LPY22185), epl1-Nt∆ (LPY22091), epl1-Ct∆ (LPY22033). Growth at increasing temperatures is shown in comparison to WT (LPY22004) and sds3∆ (LPY22006). (B) Sensitivity to a spectrum of DNA-damaging agents at 24°. The growth control 24° SC plate is shown in (A), and DMSO is included as the vehicle control for CPT. (C) Histone H4 acetylation is low among mutants in the essential EPcA domain of EPL1 relative to WT, whereas H4 acetylation is at WT levels in the epl1-Ct∆ mutant.

Because nucleosomal H4 acetylation by Esa1 in the piccolo-NuA4 complex is one of its defining features (Boudreault et al. 2003), we evaluated H4 acetylation as a proxy for NuA4 catalytic activity in the EPL1 mutants. For the lysines probed, we observed that mutants of all three EPcA subdomains (epl1-NP∆, epl1-E∆, epl1-Y∆) were defective for H4 acetylation in the bypass state (Figure 6C). Accordingly, loss of the entire EPcA domain (epl1-Nt∆) leads to similarly low levels of H4 acetylation. This is a striking distinction from the growth assays where epl1-NP∆ was more robust than epl1-E∆ or epl1-Y∆, thus pointing to the idea that substrates in addition to H4 may be critical for full biological function.

One of the key findings from the initial Epl1 bypass analysis was that Epl1 promotes the stable chromatin association of Esa1 (Figure 4A). Because our mutational studies revealed that the most critical Epl1 residues (Epl1-E and Epl1-Y) were required for growth at high temperature, response to DNA damage, and histone H4 acetylation, we initially hypothesized that these same residues might be important for promoting chromatin association. We performed fractionation assays as above, in this case with each of the Epl1 mutants in the sds3∆ bypass background (Figure 7A). We found that each of the mutants in the essential EPcA domain retained chromatin association and, likewise, Esa1 remained chromatin associated in each of these mutants.

Figure 7.

Viability of epl1 mutants is linked to stable Epl1–Esa1 interaction, not chromatin association. (A) Both Epl1 and Esa1 remain chromatin associated like WT (LPY22231) in all mutants of the essential EPcA domain [epl1-NP∆-13MYC sds3∆ ESA1-3HA (LPY22213), epl1-E∆-13MYC sds3∆ ESA1-3HA (LPY22208), epl1-Y∆-13MYC sds3∆ ESA1-3HA (LPY22226), epl1-Nt∆-13MYC sds3∆ ESA1-3HA (LPY22209)]. However, upon loss of the C-terminus of Epl1 [epl1-Ct∆-13MYC sds3∆ ESA1-3HA (LPY22211)], both Epl1 and Esa1 are shifted to occupy both soluble and chromatin-bound pools. The observed shift of Epl1 association in esa1∆ sds3∆ here also controls for a possibility of the MYC-tag causing unintended association. (B) The physical interaction of Esa1–Epl1 is disrupted in the epl1-E∆ and epl1-Y∆ mutants, but not in the epl1-NP∆ and epl1-Ct∆ mutants. Immunoblots following immunoprecipitation illustrate the loss of the physical interaction in the two essential domain mutants as seen in WT, epl1-NP∆, and in epl1-Ct∆, and lack of nonspecific binding at the relevant molecular weights in the no-tag control (LPY79). Additionally, physical interaction with Yng2 is lost only in epl1-Y∆. ♦ marks cross-reactivity with IgG-heavy chain of the antibody. Whole-cell lysate was prepared for yng2∆ (LPY22421) in no-tag control background and is included as a negative control for the Yng2 antibody.

Previous in vitro studies suggested a key role for the Epl1-NP region of the protein in nucleosomal binding (Chittuluru et al. 2011; Xu et al. 2016). However, the observed (Figure 7A) epl1-NP∆ mutant protein associated with chromatin in vivo. In contrast, a small amount of the Epl1-Ct∆ protein shifted to the soluble pool (S), with a similar shift observed for Esa1 in this background. Sir2 remained chromatin bound regardless of EPL1 mutations. The shift to the soluble pool in epl1-Ct∆ sds3∆ for both Epl1 and Esa1 supports the idea that the C-terminus acts to stabilize both Esa1 and Epl1 in chromatin, thus defining a new role for this most divergent region of Epl1 and its orthologs.

Physical association between Esa1 and Epl1 is required for activity

From earlier in vitro studies, Epl1 was divided into two domains: the EPcA domain that physically interacts with the Esa1, Yng2, and Eaf6 piccolo-NuA4 subunits and the C-terminus that tethers Epl1 and the piccolo subunits to the NuA4 holo-complex through Eaf1 (Boudreault et al. 2003; Auger et al. 2008; Rossetto et al. 2014). These regions had not yet been evaluated in vivo, so we sought to determine which are essential for the interaction with Esa1, and simultaneously which are required for interaction with Yng2 and, by extension, Eaf6.

We immunoprecipitated Epl1 in WT, in each of the three EPcA subdomain mutants, and in the C-terminal deletion mutant, and then immunoblotted for Esa1 and Yng2. We found that only Epl1-NP∆ and Epl1-Ct∆ retained interaction with both Esa1 and Yng2 (Figure 7B). This connection of Epl1-NP∆ to the NuA4 holo-complex offers an explanation for its ability to survive under nonbypass conditions, and upon DNA damage and high-temperature stress. Consistent with recent structural analysis (Xu et al. 2016), we found that Epl1-Y∆ lost physical interaction with Yng2 in vivo. Interestingly, we found that Epl1-E and Epl1-Y, the two most critical domains of Epl1 defined phenotypically, were both essential for robust physical interaction with Esa1. Our results, in tandem with earlier in vitro studies illustrating that HAT activity of Esa1 is directly augmented by Epl1 (Boudreault et al. 2003), support the idea that Epl1 is a critical NuA4 subunit due to a role as an Esa1-cofactor. Thus, Epl1 is a central regulator that is as crucial for NuA4 complex function as the Esa1 enzyme itself.

Discussion

Defining the function of a noncatalytic component of a macromolecular complex can be a challenge, particularly when that component is essential for viability. Such has been the case for Enhancer of polycomb, originally identified as a suppressor of position-effect variegation in Drosophila (Sinclair et al. 1998). Shortly after its genetic discovery, E(Pc) was cloned and found to be both deeply conserved from yeast (Epl1) to humans (EPC1/2) (Stankunas et al. 1998; Doyon et al. 2004) and to be essential for chromatin-directed functions. Specifically, Epl1 was identified as a critical subunit of the conserved MYST-family histone acetyltransferase NuA4 complex (Galarneau et al. 2000) and appears to be dedicated to NuA4/piccolo-NuA4.

Progress made toward understanding the functions of Epl1 include identification of the essential EPcA domain of Epl1, characterization of essential in vitro functions, and most recently structural analysis of Epl1 as part of nucleosome-bound NuA4 core complex (Boudreault et al. 2003; Selleck et al. 2005; Chittuluru et al. 2011; Huang and Tan 2012; Xu et al. 2016). Despite this progress, analysis of Epl1 in vivo has been relatively modest. The discovery reported here, that the requirement for EPL1 could be bypassed by deletion of a component of a histone deacetylase complex, provided a unique advantage for performing in vivo studies in epl1∆ strains.

We found that the essential requirement for Epl1 and Esa1 could be bypassed by loss of the Rpd3L deacetylase, but not for the other four essential subunits in NuA4 (Figure 1B). Earlier studies suggested that among the essential subunits, only Epl1 and Esa1 appear to be dedicated to NuA4/piccolo-NuA4, whereas the others participate in additional chromatin-modifying complexes or cellular structures. These include Tra1, an ATM-family cofactor, which serves as a recruitment module in SAGA and SLIK/SALSA complexes (Grant et al. 1998) and Swc4, Arp4, and Act1, which are components of the SWR1 chromatin-remodeling complex (Krogan et al. 2004; Mizuguchi et al. 2004) and serve other cellular roles. Specifically, Arp4 and Act1 are also found in the INO80 ATP-dependent chromatin-remodeling complex (Shen et al. 2000) and Act1 is an essential cytoskeletal protein (Shortle et al. 1982). Given this context, the bypass suppression of Epl1, but not the other essential subunits, underscores its exclusive importance as a NuA4 subunit.

We have demonstrated that Epl1 is important for promoting the stable chromatin association of Esa1 through the nonessential C-terminus of Epl1. This was counter to expectations because Esa1 nucleosomal association was reported previously to occur via the N-terminal EPcA subdomain (Epl1-NP), essential for H4 but not H2A acetylation (Boudreault et al. 2003; Selleck et al. 2005; Chittuluru et al. 2011; Huang and Tan 2012; Lalonde et al. 2013; Xu et al. 2016). An important distinction is that in contrast to in vitro experiments utilizing recombinant piccolo-NuA4 components, epl1-NP∆ sds3∆ retains an assembled NuA4 holo-complex. Thus, it is possible that whereas an isolated Epl1 requires Epl1-NP for nucleosomal association, in epl1-NP∆ sds3∆, other chromatin-interacting subunits in NuA4 that are still attached to Epl1 may efficiently target Epl1 (and Esa1) to chromatin.

We found that without the Epl1 C-terminus (epl1-Ct∆), H4 acetylation remained at WT levels. Additionally, in Epl1 mutants with disrupted Epl1-Esa1 physical interactions, Esa1 remained chromatin-targeted. Whereas these findings may at first appear to be at odds, there are several key considerations. It is possible that, by default, Esa1 is associated with chromatin. When loss of the C-terminus dissociates Epl1 from chromatin, it may bring along Esa1. This possibility is supported by the fact that in epl1-E∆ sds3∆ and epl1-Y∆ sds3∆, Esa1 remains chromatin associated (Figure 7A) despite not physically interacting with Epl1 (Figure 7B). Even without Epl1, Esa1 may transiently associate with chromatin (Figure S3 in File S1), allowing the dynamic and rapid process of acetylation to occur. It is also possible that the small amount of chromatin association that remains is sufficient for Esa1 catalytic activity, especially in the bypass state, where Rpd3L does not actively deacetylate histone H4. The NuA4 holo-complex may also be required for stable chromatin association, such that the interaction of Esa1 is facilitated by other subunits, where loss of the C-terminus of Epl1 specifically represents dissociation of piccolo-NuA4. Further studies involving the nonessential NuA4 Eaf1 subunit and the dynamics between Epl1 and Eaf1 in vivo may provide insight into these possibilities.

Our results support a model in which Epl1 is physically required for promoting Esa1 enzymatic activity as a part of the piccolo-NuA4 and/or the NuA4 holo-complexes, much like Ada2 and Ada3, which act in a catalytic core to potentiate the activity of the Gcn5 acetyltransferase (Balasubramanian et al. 2002). In WT or sds3∆ cells, Epl1 acts as an anchor for piccolo-NuA4 subunits, including Esa1, and also tethers these subunits to the NuA4 holo-complex (Figure 8A). With both NuA4 and piccolo-NuA4 intact, as they are in WT and in sds3∆, there are normal levels of acetylation. However, upon loss of the nonessential C-terminus (epl1-Ct∆ sds3∆), piccolo-NuA4 becomes untethered, and the NuA4 holo-complex is disrupted, leaving only the broad nucleosomal HAT function of piccolo-NuA4 (Figure 8B). This is sufficient for global, less targeted acetylation of histone H4 but perhaps not for acetylation of nonhistone substrates that contribute to fitness. In epl1-NP∆ sds3∆, all piccolo-NuA4 subunits, including Esa1 and Epl1, still physically interact, with epl1-NP∆ still permitting both NuA4 and piccolo-NuA4 integrity; however, this mutation causes a reduction in acetylation relative to WT and sds3∆ (Figure 8C), perhaps due to inefficient nontargeted chromatin binding.

Figure 8.

Model: Epl1 is a core NuA4 regulator in tandem with Esa1. (A) Epl1 is a central component of NuA4 and piccolo-NuA4 (abbreviated pic-NuA4, shown here only as a part of NuA4). In WT and sds3∆, both complexes are intact and active, resulting in normal levels of acetylation, both of histone H4, as shown in the tails of the H3–H4 tetramer, and of nonhistone substrates, with two representative substrates illustrated, out of over 250 reported in proteomic studies. (B) Loss of the Epl1 C-terminus results in loss of the NuA4 holo-complex, but largely uncompromised pic-NuA4 function, and slightly reduced acetylation levels of nonhistone substrates only in the sds3∆ background. (C) epl1-NP∆ keeps NuA4 intact, and the essential components of pic-NuA4 remain tethered, promoting robust fitness in the bypass state; however, low acetylation levels are present. The assembly of the NuA4 holo-complex is critical here, such that upon loss of EAF1, epl1-NP∆ is no longer viable in nonbypass conditions (Figure S4 in File S1). (D) Loss of the subunit-interaction domains [epl1-E here and epl1-Y in (E)] results in Esa1 no longer structurally bound to Epl1, causing both NuA4 and pic-NuA4 to be compromised. This results in low acetylation and overwhelming loss of cellular fitness similar to epl1∆ sds3∆. (E) By comparison, epl1-Y∆ sds3∆ results in the similar loss of physical contact with Esa1, but also loss of physical interaction with Yng2, and therefore by extension, Eaf6. This mutant has the same severe fitness-deficits as epl1-E∆ sds3∆ and epl1∆ sds3∆, underscoring the primary importance of the Esa1–Epl1 interaction. Although not illustrated in the model, epl1∆ sds3∆ would be similar to epl1-E∆ sds3∆ and epl1-Y∆ sds3∆, with low levels of acetylation; however in the complete absence of Epl1, all piccolo-NuA4 subunits would be disassociated from the NuA4 holo-complex. Note that in our experiments, analysis of H4 acetylation is a proxy for NuA4/piccolo-NuA4 activity. Proteomic studies define many additional substrates for NuA4 activity, some of which are likely to contribute to processes affected upon loss of Epl1 or Esa1 functions. Of note, both Yng2 and Epl1 were identified as substrates of Esa1, where acetylation has already been demonstrated to have a significant functional impact (Yi et al. 2012; Mitchell et al. 2013; Downey et al. 2015). Therefore, we believe that nonhistone substrates, though not specifically analyzed here, are a critical part of the observed phenotypes and model presented.

Despite reduced acetylation levels in epl1-NP∆, we hypothesize that the presence and targeted activities of the NuA4 holo-complex may be sufficient to promote cellular viability and in bypass conditions, response to cellular stresses such as DNA damage. In fact, we found that disruption of the NuA4 holo-complex by eaf1∆ results in lethality of epl1-NP∆ in nonbypass conditions (Figure S4 in File S1), highlighting the importance of the NuA4 holo-complex for growth in the epl1-NP∆ mutant background. Further, we found that if Esa1 was simply disassociated from Epl1, such that it was no longer a component of NuA4 or piccolo-NuA4, as was the case in both epl1-E∆ sds3∆ and epl1-Y∆ sds3∆, cellular growth was severely compromised, with low levels of acetylation, and death at elevated temperatures or with DNA damage (Figure 8, D and E). These findings complement the recently published structural insights for piccolo-NuA4 (Xu et al. 2016), which illustrate that residues within those deleted in epl1-E∆ and epl1-Y∆ were most critical for contacting both Esa1 and the nucleosome.

Overall, our results support the concept that Epl1 is required to function in tandem with Esa1, tethering Esa1 to other subunits for full and robust function. This concept is supported by in vitro experiments demonstrating negligible HAT activity of Esa1 in the absence of the Epl1 and Yng2 piccolo-NuA4 subunits (Boudreault et al. 2003). We have shown that if Epl1 is not present (epl1∆ sds3∆), or unable to physically interact with Esa1 (epl1-E∆ sds3∆ and epl1-Y∆ sds3∆), Esa1 becomes ineffectual.

The majority of the studies reported here have been performed where the requirement for Epl1 is conditionally bypassed by sds3∆, serving to balance cellular acetylation, as in our studies of Esa1 (Torres-Machorro and Pillus 2014). Historically, bypass suppression has promoted fundamental understanding of multiple and diverse pathways. This includes, for example, studies of cell-cycle checkpoints where suppression of mec1∆ and rad53∆ lethality is bypassed by concurrent loss of SML1 (Zhao et al. 1998), and more recent studies of transcription regulation, where the requirement for the COMPASS methyltransferase subunit Swd2 is bypassed by set1∆ (Soares and Buratowski 2012). Likewise, bypass suppression served as a powerful tool here that has allowed comprehensive functional assessment of EPL1 in vivo. However, the concurrent loss of a major deacetylase should be kept in mind in the interpretation of data. It is only in this context that the relative comparisons between the mutants in vivo can be made with the earlier biochemical analyses to provide a deeper, valuable, and more holistic understanding of an essential protein-modifying activity. Studies in Drosophila and humans alike illustrate that Epl1 orthologs play key roles in development and cancer, akin to Epl1’s essential role in yeast. In addition to established roles in DNA damage (Figure 2B) (Boudreault et al. 2003), human EPC1 potentially has critical roles in DNA damage repair both within and independently of NuA4 (Attwooll et al. 2005; Wang et al. 2016). Failures in DNA damage repair are associated with genomic instability, which is a major driving force in cancer. The observation that mutational profiles of cancer patients reflect frequent alterations in EPC1/2 highlights the importance of these proteins in human biology and disease. Analysis of genomic cancer data illustrates, for example, that EPC1 is frequently amplified in neuroendocrine prostate cancer, but deleted in prostate adenocarcinoma (Cerami et al. 2012; Gao et al. 2013). Additionally, independent analysis demonstrates that EPC1 and EPC2 are often mutated across the gene body in many cancer subtypes, including in the critical domains studied here (Forbes et al. 2014). These frequent yet diverse alterations underscore the importance of understanding the critical functions of specific residues and domains of Epl1, along with the consequences of complete deletion of this essential gene. Our results provide new insights into both aspects of altered function and will be instrumental in deepening the understanding of Epl1 orthologs in development and disease.

Supplementary Material

Supplemental material is available online at www.genetics.org/lookup/suppl/doi:10.1534/genetics.116.197830/-/DC1.