What makes a histone variant a variant: Changing H2A to become H2A.Z

Chromatin structure and underlying DNA accessibility is modulated by the incorporation of histone variants. H2A.Z, a variant of the H2A core histone family, plays a distinct and essential role in a diverse set of biological functions including gene regulation and maintenance of heterochromatin-euchromatin boundaries. Although it is currently unclear how the replacement of H2A with H2A.Z can regulate gene expression, the variance in their amino acid sequence likely contributes to their functional differences. To tease apart regions of H2A.Z that confer its unique identity, a set of plasmids expressing H2A-H2A.Z hybrids from the native H2A.Z promoter were examined for their ability to recapitulate H2A.Z function. First, we found that the H2A.Z M6 region was necessary and sufficient for interaction with the SWR1-C chromatin remodeler. Remarkably, the combination of only 9 amino acid changes, the H2A.Z M6 region, K79 and L81 (two amino acids in the α2-helix), were sufficient to fully rescue growth phenotypes of the htz1Δ mutant. Furthermore, combining three unique H2A.Z regions (K79 and L81, M6, C-terminal tail) was sufficient for expression of H2A.Z-dependent heterochromatin-proximal genes and GAL1 derepression. Surprisingly, hybrid constructs that restored the transcription of H2A.Z-dependent genes, did not fully recapitulate patterns of H2A.Z-specific enrichment at the tested loci. This suggested that H2A.Z function in transcription regulation may be at least partially independent of its specific localization in chromatin. Together, this work has identified three regions that can confer specific H2A.Z-identity to replicative H2A, furthering our understanding of what makes a histone variant a variant.


Introduction
In eukaryotic cells, chromatin structure is a major modulator of genomic processes that require DNA accessibility. The basic packaging unit of chromatin is the nucleosome, consisting of approximately 146 base pairs of DNA wrapped around a protein octamer composed of two copies of each of the core histones: H2A, H2B, H3, and H4 [1]. One modulator of chromatin structure is the incorporation of histone variants [2].Unlike the core histones (also called replicative histones), which are produced in equal amounts during S-phase by multiple genes, histone variants are encoded by one or two replication-independent genes, allowing variantspecific transcription and deposition throughout the cell cycle [3]. Histone variants are also distinct in their protein sequence and structure, which can lead to different post-translational modifications, recruitment of proteins to chromatin that specifically interact with the variant, and changes to the stability and mobility of the nucleosome into which they are incorporated [4]. Through these properties, histone variants are capable of establishing unique chromatin neighbourhoods with distinct character and function in a regulated manner.
H2A.Z's primary protein sequence is remarkably similar to H2A's, with 60% identical amino acids between the two proteins, including in some of the core structural features [50]. Despite this similarity, H2A.Z is encoded by a single non-essential gene in S. cerevisiae (HTZ1) whereas the deletion of both H2A genes (HTA1 and HTA2) is lethal [51,52]. However, the htz1Δ mutant has various growth phenotypes, including drug sensitivities that cannot be rescued by overexpression of H2A or by H2A expressed from an H2A.Z promoter [51,53]. This suggests that amino acid composition, not expression dynamics, is the main driver of H2A.Z's unique function from H2A. The overall structure of a H2A.Z-containing nucleosome is very similar to that of a nucleosome containing H2A, which is unsurprising given the similarities between the two histone's 3D conformations [50,54]. Regardless, in vitro, H2A.Z nucleosomes have a lower physical breaking force and thermal stability than H2A nucleosomes [54][55][56], and salt washes indicate that H2A.Z is more readily lost from chromatin than H2A [37,54,57,58]. Therefore, while these histones have similar structures, the regions of primary sequence divergence between H2A and H2A.Z likely do give rise to their unique functions.
Distinguishing which of H2A.Z's unique regions are responsible for its identity is key to understanding its functional specialization and may provide new insights into how the replacement of replicative H2A with H2A.Z can regulate gene expression and overall chromatin biology. Previous studies have explored the function of H2A.Z's primary structure using a variety of methods including gene truncations, alanine point mutation libraries, and genetic screens of H2A.Z mutants that are not incorporated into chromatin [53,59,60]. However, while these methods are effective at identifying amino acids required for H2A.Z function, they do not determine which ones are responsible for H2A.Z's unique identity. To determine which regions distinguish H2A.Z from H2A, we built a plasmid library consisting of various H2A/H2A.Z hybrid constructs created by replacing regions of the H2A gene (HTA1) with the corresponding H2A.Z gene (HTZ1) and vice versa. We identified three regions that were able to confer H2A.Z identity: the M6 region, the C-terminal tail, and two evolutionarily conserved amino acids in the α2 helix (K79 and L81). Remarkably, the combination of only 9 amino acid changes, the H2A.Z M6 region, K79 and L81, was sufficient to fully rescue htz1Δ growth phenotypes. However, the addition of the H2A.Z C-terminal tail enabled wild-type transcription of H2A.Z-dependent heterochromatin-proximal genes and GAL1 after derepression. Surprisingly, hybrid constructs that restored the mRNA levels of H2A.Z-dependent genes had chromatin enrichment patterns that more closely resembled H2A than H2A.Z, suggesting that H2A.Z function in transcription regulation may be at least partially independent of its specific pattern of localization in the genome. More generally, we took advantage of the functional differences between H2A and H2A.Z, as well as the power of yeast genetics, to broadly explore what makes a histone variant a variant.

H2A/H2A.Z hybrid constructs revealed regions that contribute to H2A.Z identity
In order to determine which protein regions are involved in conferring H2A.Z-specific functions, we grouped all of the amino acids that differ between H2A.Z and H2A into nine regions (Fig 1). One of these regions, M6, is a group of 11 amino acids immediately preceding the Cterminal tail that has previously been shown to be a defining feature of H2A.Z identity in S. cerevisiae and Drosophila [61,62]. Apart from the M6 region, the boundaries of each of the regions in this study were determined by H2A.Z's primary and secondary structure. We constructed hybrid constructs using two previously characterized parent plasmids, H2A.Z and H2A, which are both expressed from the H2A.Z gene promoter (HTZ1) and C-terminally FLAG-tagged for determination of protein levels and to enable functional genomic studies (Fig 2A) [53]. To determine if a region was sufficient to confer H2A.Z identity onto H2A, we built a hybrid construct of H2A containing the corresponding region of H2A.Z (H2A-H2A.Z), while a hybrid construct of H2A.Z with the equivalent H2A region (H2A.Z-H2A) was used to determine if a region was necessary for H2A.Z function. In order to assess the ability of each region to recapitulate several aspects of H2A.Z identity, these hybrid constructs were placed into an htz1Δ background. While the H2A.Z mutant showed a marginal increase in H2A.Z abundance in comparison to wild-type (S1A Fig), all hybrid constructs were present in comparable levels to the H2A.Z construct (S1B and S1C Fig).
Given that the htz1Δ mutant exhibits slow growth when exposed to a variety of genotoxic stress conditions [23,24], we used growth assays to assess how exchanging each of the nine H2A.Z and H2A regions impacted cell fitness. Consistent with previous reports [51,53], we found that the htz1Δ mutant had growth defects in formamide, caffeine, and hydroxyurea (HU), which could not be rescued by expressing H2A from the H2A.Z promoter (Figs 2B and S2A). In contrast, the H2A.Z mutant restored cell growth and was indistinguishable from the wild-type empty vector control. A visual comparison of the H2A-H2A.Z hybrid mutants growth phenotypes indicated that several regions could partially recapitulate H2A.Z identity. mutants all showed improved growth over the H2A mutant in various conditions. However, no single region was able to fully rescue growth phenotypes of the htz1Δ mutant and thus fully recapitulate H2A.Z identity. Therefore, some combination of these nine unique regions must be required to completely confer H2A.Z function.
To determine which of the nine regions were necessary for H2A.Z's identity, we compared the growth of each H2A.Z-H2A hybrid mutant to the H2A.Z mutant. The H2A.Z-H2A[M6] mutant had growth defects in all three conditions in comparison to the H2A.Z mutant (Figs 2C and S2B). All of the remaining H2A.Z-H2A hybrid mutants grew similar to wild-type. Therefore, of the nine divergent regions between H2A and H2A.Z, only the M6 region was necessary for H2A.Z dependent phenotypes.  The H2A.Z M6 region was necessary and sufficient for interaction with SWR1-C and histone chaperones The H2A.Z M6 region is a key binding region for several factors involved in H2A.Z's incorporation into chromatin, including SWR1-C and the histone chaperones Nap1 and Chz1 [27,61,63,64]. Therefore, the growth phenotypes observed for the H2A.Z-H2A[M6] and H2A-H2A.Z[M6] mutants may be associated with how the hybrid constructs interact with these proteins. To examine this possibility more directly, we purified the FLAG-tagged M6 hybrid constructs from cells containing VSV-tagged SWR1-C subunits: Swc2, Swc3, or Swc4. Consistent with previous work, all the tested subunits co-purified with the H2A and H2A. Z-H2A[M6] constructs at lower levels compared to the H2A.Z construct, confirming that the H2A.Z M6 region is required for interaction with SWR1-C (Figs 3A and S3A and S3B) [61, 63,64]. In contrast, the H2A-H2A.Z[M6] construct associated with each SWR1-C subunit in a manner that was reminiscent of the H2A.Z construct. Therefore, as judged by this analytical method, the H2A.Z M6 region on its own was sufficient for interaction with SWR1-C subunits.
constructs were used to create the hybrid H2A.Z-H2A and H2A-H2A.Z constructs. Each construct was expressed from the native H2A.Z gene promoter and C-terminally 3xFLAG tagged. (B) Growth assays of the H2A-H2A.Z mutants indicated that mutants containing either the H2A.Z N-terminus, M6, G3, or G4 region had improved growth over the H2A mutant, while (C) the H2A. Z-H2A mutants showed that of the nine regions of H2A.Z, only the M6 region could not be functionally replaced by the corresponding region from H2A. Cells expressing the indicated hybrid constructs were 10-fold serially diluted, spotted onto SC-TRP media with the indicated concentrations of formamide, caffeine, and hydroxyurea and grown for 3 days.
https://doi.org/10.1371/journal.pgen.1009950.g002 Purifications of the FLAG-tagged constructs in cells with VSV-tagged Chz1 showed that the H2A.Z-H2A[M6] construct closely resembled the H2A construct, which indicated that the H2A.Z M6 region was necessary for Chz1 interaction with H2A.Z (Figs 3B and S3C). The H2A-H2A.Z[M6] construct, however, had an increased interaction with Chz1 in comparison to the H2A.Z construct, a result which replicated in reciprocal purifications of VSV-tagged Chz1 (Figs 3C and S3D). This may suggest that H2A.Z amino acids outside the M6 region partially inhibited interaction with Chz1. Conversely, Nap1 interaction with H2A.Z was not entirely dependent on the H2A.Z M6 region (Figs 3B and S3C). While the H2A.Z-H2A[M6] construct did have reduced interaction with Nap1 in comparison to the H2A.Z construct, it also had increased interaction in comparison to the H2A construct. In addition, the H2A-H2A.Z[M6] construct did not fully recapitulate H2A.Z interaction with Nap1. Therefore, while the H2A.Z M6 region did play a role in interaction with Nap1, other H2A.Z regions may be required for complete wild-type Nap1 association.
In combination with M6, an evolutionarily conserved region in the α2-helix of H2A.Z confers H2A.Z-specific growth phenotypes As our H2A-H2A.Z hybrid constructs showed that no single H2A.Z region conferred upon a H2A background was able to fully rescue the htz1Δ mutant growth defects and recapitulate H2A.Z identity, a combination of these nine unique regions must be required. Having established that the M6 region was both necessary and sufficient for H2A.Z's binding to SWR1-C, we created a series of combination hybrids that contained the H2A.Z M6 region, plus one of the eight remaining H2A.Z regions, placed in the H2A protein background. After being transformed into the htz1Δ mutant, each hybrid again showed similar protein levels as the H2A. To gain a better understanding of the importance of K79 and L81 in H2A.Z identity, we compared the primary protein sequences of several H2A.Z homologs found in eukaryotic organisms. We found that both K79 and L81, which are located at the C-terminal proximal end of the α2-helix, are highly conserved in H2A.Z homologs and are consistently different from H2A ( Fig 4C)   This suggested that the presence of SWR1-C missing its H2A.Z substrate (apo-SWR1-C) could be affecting the transcription of these genes in the htz1Δ mutant. As substrates of SWR1-C, our hybrid constructs could then be simply alleviating apo-SWR1-C mischief at heterochromatin boundaries instead of recapitulating H2A.Z-specific function in transcription regulation. To address this important caveat, we performed RT-qPCR at GIT1, RDS1, and YCR100c for htz1Δ, swr1Δ, htz1Δ swr1Δ mutants (S7B Fig). The htz1Δswr1Δ mutant had comparable levels of mRNA to the htz1Δ mutant for each of these genes, which suggested that the decrease in expression for the htz1Δ mutant was primarily due to the loss of H2A.Z function in gene expression and not from the presence of apo-SWR1-C.
The expression of several inducible genes has been previously found to occur at significantly slower rates in a htz1Δ mutant than in wild-type cells [31,[46][47][48]. Given that both the

Discussion
Determining what makes a histone variant a variant is a fundamental question in chromatin biology. While characterizing the amino acids required for H2A.Z function has been the focus of several previous studies, determining which regions are actually sufficient to confer histone variant identity, and thus responsible for distinguishing H2A.Z from H2A, have remained unclear. In this study, we used hybrid constructs based on replicative histone H2A to systematically analyze all of the divergent regions between H2A.Z and H2A. Here we report that the M6 region, the C-terminal tail, and the G4 region in the α2-helix each contributed to H2A.Z's unique identity from H2A (Fig 8). While combining the M6 and G4 regions was sufficient for rescuing htz1Δ growth phenotypes, only the combination of the three unique H2A.Z regions (K79 and L81, M6, C-terminal tail) were sufficient for expression of H2A.Z-dependent heterochromatin-proximal genes and GAL1 after derepression.
Here we report that while the M6 region was not sufficient to fully recapitulate H2A.Z identity, it was required for H2A.Z-specific function. We found that replacing the H2A.Z M6 region with the corresponding H2A residues resulted in growth defects in comparison to wildtype H2A.Z. Therefore, the seven amino acid differences between H2A and H2A.Z M6 regions must be required for H2A.Z-specific growth phenotypes (Fig 8). We confirmed previous reports which found that the H2A.Z M6 region is required for H2A.Z to interact with SWR1-C [61], which is unsurprising given that H2A.Z M6 region directly binds two of the [M6,G4,C] constructs had significantly increased enrichment in comparison to H2A.Z. FLAG-tagged hybrid enrichment levels determined by ChIP-qPCR for three replicates were normalized to their respective inputs. Asterisks indicate all the significant comparisons determined by unpaired twotailed Student's t-tests. � = p-value <0.05, �� = p-value <0.01, ��� = p-value <0.001. All constructs were significantly enriched over the untagged control (p-value <0.05). All other unlabelled comparisons had a p-value > 0.05.
https://doi.org/10.1371/journal.pgen.1009950.g006  [63,64]. Furthermore, we demonstrated that the H2A.Z M6 region was sufficient to fully recapitulate interaction with SWR1-C subunits (Fig 3A). This is consistent with work that compared crystal structures of H2A.Z-H2B dimers bound to Swr1 and found that a H2A hybrid construct containing the H2A.Z M6 residues I105-A107, had binding affinities for Swr1 that were similar to wild-type [63]. However, an additional structural study found that T87 (found in the L2 region in this study), D102 (in the M6 region) and the lack of a glycine in the H2A.Z M6 region, are important for differentiating H2A and H2A. Z's binding affinity with Swc2 [64]. Despite this, we found that the H2A-H2A.Z[M6] construct had similar levels of association with Swc2 as the wild-type, indicating that only the M6 residues were necessary for association. Further supporting the M6 regions role as a binding motif, we also report that interchanging the H2A and H2A.Z M6 regions affected association with the histone chaperones, Chz1 and Nap1 (Fig 3B and 3C). This is consistent with previous work which has implicated G97 and the absence of a glycine in M6, as defining features in H2A.Z's specific interaction with histone chaperones in yeast and mammals [27,68].
In this study we identified the two residues in the G4 region as key contributors to H2A.Z's unique identity from H2A (Fig 8). The combination of K79 and L81 with the H2A.Z M6 HMRa locus and the right telomere of chromosome III. Arrows represent open reading frames (ORFs) and point in the direction of transcription. (B) Expression defects observed in the htz1Δ mutant were gradually improved by the stepwise addition of the H2A.Z M6, G4, and C-terminal regions. RT-qPCR analysis of heterochromatin-proximal genes mRNA levels from three replicates were normalized to ACT1 mRNA levels. Matrices of significant comparisons for GIT1, RDS1, and YCR100c were determined by unpaired two-tailed Student's t-tests. (C) RT-qPCR analysis of GAL1 mRNA levels performed on the indicated mutants, which were grown long-term in SC-TRP-glucose (2%) prior to being transferred to SC-TRPgalactose (2%) and collected at 2-hour intervals. During GAL1 derepression, the addition of the M6, G4, and C-terminal regions incrementally increased GAL1 expression, were the H2A-H2A.Z[M6,G4,C] mutant showed comparable levels to the H2A.Z mutant. GAL1 mRNA levels were normalized to ACT1. Error bars indicate the standard deviation between the three replicates. Significance comparisons determined by unpaired two-tailed Student's t-tests are indicated: � = p-value <0.05, �� = p-value <0.01, ��� = p-value <0.001. All unlabelled comparisons within each time point had a p-value >0.05.
https://doi.org/10.1371/journal.pgen.1009950.g007 region resulted in a complete rescue of htz1Δ growth phenotypes, an outcome not observed for any of the other regions when combined with M6 ( Fig 4A). We also noted that the majority of G4's impact on growth phenotypes was driven by L81. These data extend and contextualize previous results showing that a substitution of L81 with alanine results in a slight sensitivity to methyl methanesulfonate (MMS), while a K79 mutation does not [60]. This is largely unsurprising given that L81 (non-polar) has substantial differences in biochemical properties from its H2A counterpart (asparagine, a polar amino acid) while K79 has relatively similar properties to its H2A counterpart (arginine). Relatively little is known about the specific mechanistic or structural roles of K79 or L81 in H2A.Z function. There are currently no confirmed posttranslational modifications for these amino acids in yeast, however tandem mass spectrometry indicates that the homologous lysine of H2A.Z K79 in humans (K74) is ubiquitinated [69]. It is possible that G4's location just prior to the unstructured L2 loop region of the histone fold may be of significance given that the L2 region acts together with the H2B L1 loop to create a DNA-binding motif in the nucleosome [70]. Even small differences in the biochemical properties of adjacent amino acids could result in changes to the L2 loop positioning or flexibility. In humans, for example, the L1 loop regions of two isoforms of H2A.Z (H2A.Z.1 and H2A.Z.2) have identical amino-acid sequences, but crystal structures of nucleosomes containing either isoform indicate that these flexible regions have distinct structures that are at least partially caused by a single amino acid difference immediately prior to the L1 loop region (T38) [58]. We therefore propose that K79 and L81 may influence the flexibility or structure of the L2 loop region in yeast in a similar manner.
The large diversity of C-terminal tails is a key characteristic of the H2A family of histones and is thought to contribute to the distinct functions of many H2A variants, including H2A.Z [4,71]. Here we report that, in combination with the M6 and G4 regions, the C-terminal tail conferred H2A.Z driven gene expression for genes located at a heterochromatin-boundary as well as GAL1 after derepression (Figs 7 and 8). This is consistent with previous work which has implicated the C-terminal tail as being important for wild-type transcription of both inducible genes and heterochromatin-boundary genes [46,47,53]. Despite this, our growth assay results for the H2A.Z-H2A[C] mutant suggested that the C-terminal tail can be functionally replaced with the H2A counterpart, which is consistent with previously published findings ( Fig 2C) [53]. However, given that our results indicate that growth assay phenotypes do not necessarily equal complete wild-type function, it is possible that replacing the H2A.Z C-terminal tail with the H2A counter part could still result in expression defects for certain genes. In addition to having a role in transcription, together with M6, the H2A.Z C-terminal tail has been repeatedly shown to play an important role in H2A.Z incorporation and retention in chromatin [50,53,72,73]. The crystal structure shows that the C-terminal tail is anchored in the nucleosome, at least in part through hydrophobic interactions of side chains and backbone residues with histone H3 [50]. Truncations of this region result in the complete loss of H2A.Z in chromatin, and individual mutations of the H2A.Z specific residues S111, H117, and N119, also result in defects in chromatin association [53,72]. Finally, the H2A.Z C-terminal tail may also play a role in DNA damage repair. SUMOylation of K125 and K132 in the H2A.Z C-terminal tail in yeast is required for the recruitment of double strand breaks to the nuclear periphery [74]. As SUMO modifications have been shown to facilitate the exchange of H2A. Z.2 at DNA damage sites in human cells, a role in DNA damage repair may in fact be a conserved function of the H2A.Z C-terminal tail [75].
Throughout this study we used a broad variety of established htz1Δ phenotypes to determine if various amino acid regions are able to confer H2A.Z-specific functions when inserted into a H2A backbone. However previous studies have shown that mischief caused by the presence of apo-SWR1-C may be responsible for many of the defects observed in htz1Δ mutants [31][32][33]. Given that the presence of the H2A.Z M6 region was sufficient for wild-type levels of association with SWR1-C subunits (Fig 3A), it is possible that the hybrids containing the M6 region were simply recapitulating H2A.Z's ability to alleviate apo-SWR1-C mischief. However, our results suggested that the hybrids did confer H2A.Z function in chromatin in addition to counteracting apo-SWR1-C. Specifically, and consistent with previous work, we showed that both the swr1Δ and htz1Δswr1Δ mutants grew better than the htz1Δ mutant, but still had minor grow defects in comparison to wild-type (S7A Fig) [31 -33]. As this slow growth phenotype cannot be attributed to apo-SWR1-C, it likely represents defects caused by H2A.Z's absence from chromatin. Therefore, since both the H2A-H2A.Z[M6,G4] and H2A-H2A.Z[M6, G4,C] mutants grew similar to wild-type, these regions likely contributed to H2A.Z-dependent functions in chromatin. Furthermore, the htz1Δswr1Δ mutant had similar levels of mRNA for the heterochromatin-boundary genes as the htz1Δ mutant, which suggested the stepwise recapitulation of expression observed in the hybrid mutants was not caused the alleviation of apo-SWR1-C mischief. However, previous work has shown that the GAL1 expression defect in the htz1Δ mutant is partially repressed by the deletion of Swr1 [31], which suggests the impact of apo-SWR1-C on expression is gene dependent. Given the complexity of the circuitry between H2A.Z and SWR1-C, determining exactly what fraction of the htz1Δ mutant phenotype is caused directly by the of loss of H2A.Z function in chromatin is likely unattainable; however our results suggest that, at least in part, the M6, G4, and the C-terminal regions do confer H2A.Z function beyond alleviating apo-SWR1-C mischief.
During our examination of the enrichment of the H2A-H2A.Z hybrid constructs at gene promoters and their respective ORF regions we found that the addition of the M6, G4, and Cterminal tail regions were not sufficient to confer H2A.Z's selective localization (Fig 6A and  6B). Given that H2A.Z's distinct pattern of enrichment along chromatin is one of the definitive features differentiating between H2A and H2A.Z [3,5,7], we were surprised that hybrids lacking H2A.Z's-specific localization still recapitulated H2A.Z dependent growth and gene expression (Figs 5A and 7). There is no doubt that H2A.Z-specific enrichment is necessary for many aspects of H2A.Z biology, such as promoting H3K27me3 deposition at promoters in Arabidopsis [76], yet our results suggested that general incorporation of the histone variant into chromatin may be sufficient for some H2A.Z functions. For example, based on our RT-qPCR data, H2A.Z's role in restricting the spread of heterochromatin into euchromatin [66,[77][78][79][80] may not be dependent on H2A.Z-specific enrichment. In the absence of H2A.Z, the SIR protein complex spreads from the silent mating loci, and a subset of telomeres, into the ORFs of adjacent regions causing decreased expression of the affected genes, including GIT1, YCR100c, and RDS1 [66,80]. While the specific mechanism by which H2A.Z contributes to blocking heterochromatin spread is still poorly understood, studies using purified nucleosome arrays have found that H2A.Z containing chromatin is inherently resistant to condensation in vitro [81,82]. Therefore, the presence of H2A.Z in chromatin, and not its specific location, may be sufficient to prevent heterochromatin spread in yeast. This would also explain why the H2A-H2A.Z[M6,G4] and H2A-H2A.Z[M6,G4,C] mutants recapitulated the H2A.Z mutants mRNA levels for YCR100c despite more closely resembling the enrichment of the H2A construct at the YCR100c promoter (Figs 6B and 7B). In addition, our growth assay results also suggested that H2A.Z function in resistance to genotoxic stress may not be completely dependent on its specific enrichment levels, given that the H2A-H2A.Z[M6,G4] and H2A-H2A.Z [M6,G4,C] mutants had wild-type growth phenotypes despite not recapitulating H2A.Z-specific enrichment at all tested loci. Ultimately genome-wide high-resolution enrichment and expression analysis of H2A.Z mutants with defects in localization may yield insights into how H2A.Z might regulate transcription distinct from its specific localization patterns.
By utilizing a hybrid gene approach, we systematically analyzed each region of H2A.Z that is unique from H2A. While a multitude of previous studies have identified amino acids required for H2A.Z function [53,59,60], our study design has enabled us to distinguish which of these regions confer H2A.Z unique identity. In addition, while the exact mechanism by which H2A.Z affects gene expression still remains poorly understood, our results narrow down the amino acids responsible for this function, hopefully informing future research on the topic. Altogether, this study identified K79 and L81 as a new region that contributes to H2A.Z identity while also adding to our understanding of two key regions in H2A.Z biology, M6 and the C-terminal tail. As many of the amino acids in these regions are highly conserved in the mammalian homologs of H2A and H2A.Z, there is a strong possibility that these regions also serve to define H2A.Z unique identity in humans.

Yeast strains and plasmids
All strains used in this study, listed in Table 1, were made using standard yeast genetic techniques. Gene deletions and the integration of a vesicular stomatitis virus (VSV) tag [83] at the 3' end of genes were achieved using the one step gene replacement method [84]. All double mutant stains were generated via mating and tetrad dissection. Table 2 lists all plasmids used in this study. Both the parent H2A.Z and H2A pRS314 plasmids were expressed from the HTZ1 promoter and C-terminally tagged with 3xFLAG [53]. These parent plasmids were used to generate each of the hybrid constructs using standard molecular techniques. The QuickChange site directed mutagenesis technique (Agilent) was adapted to replace both the N-terminal and C-terminal tail regions. Multiple rounds of site-directed mutagenesis were performed to replace the remaining regions (primers available upon request).

Protein extraction and protein blotting
Overnight cultures grown in synthetic complete media without tryptophan (SC-TRP) were diluted to 0.3 OD 600 and grown to 1.0 OD 600 . Whole-cell extracts were prepared by glass bead lysis in the presence of trichloroacetic acid [86]. Immunoblotting was performed using anti-FLAG (Sigma), anti-HTZ1 (Active Motif) and anti-PGK1 (Life Technologies) antibodies. Immunoblots were scanned with the Odyssey Infrared Imaging System (Life Technologies) and analyzed using the Image Studio software (LI-COR). Three biological replicates were performed for each culture.

Growth assays
Overnight cultures grown in SC-TRP were diluted to 0.5 OD 600 . Cells were then 10-fold serially diluted and spotted onto SC-TRP plates with or without the indicated amount of caffeine (Sigma), hydroxyurea (HU) (Sigma), and formamide (Sigma). Plates were incubated for three days at 30˚C. Three biological replicates were performed for each culture.

Liquid growth curves
Overnight cultures were diluted to 0.1 OD 600 and then grown to 0.5 OD 600 . Cultures were diluted to 0.1 OD 600 in 200 μL SC-TRP media plus 1.5% formamide (Sigma). OD 600 readings were measured every 30 minutes over a period of 35 hours in a TECAN M200 plate reader. Plates were shaken for 10 minutes before each reading. Strains were tested in three biological replicates per plate and the area under the curve (AUC) was calculated for each replicate.

Chromatin association assays
Chromatin association assays were performed in triplicate as previously described [87]. Overnight cultures were diluted to 0.2 OD 600 in 50 mL of SC-TRP and grown to 0. 5  Following crosslinking reversal and DNA purification, both the immunoprecipitated and input DNA were analyzed by qPCR using a Rotor-Gene 6000 (Corbett Research) and PerfeCTa SYBR green FastMix (Quanta Biosciences). All samples were analyzed from three independent biological replicates and normalized to percent input (0.7%). Primers used are listed in Table 3.

RT-qPCR
Overnight yeast cultures were diluted to 0.15 OD 600 in 40 mL of SC-TRP. Cells were grown to an OD 600 of 0.5 and a total of 10 OD units were collected. RNA was extracted using the RNeasy Mini Kit (Qiagen) and converted to cDNA using the QuantiTect Reverse Transcription Kit (Qiagen). cDNA was analyzed using a Rotor-Gene 6000 (Corbett Research) and PerfeCTa SYBR green FastMix (Quanta Biosciences). Samples were analyzed from three independent biological replicates and normalized to ACT1. Primer sequences are listed in Table 3.

Galactose derepression
The procedure for growth of yeast cultures for GAL1 derepression was adapted from a previous report [31]. Overnight cultures were diluted to 0.1 OD 600 in 175 mL in SC-TRP (glucose, 2%). Once cells had reached an OD 600 of 0.45, 7.5 OD units were collected (time point "0"). The remaining cells were washed with sterile ddH 2 O and resuspended in 150 mL of SC-TRP containing 2% galactose. Samples were taken 2, 4, 6, and 8 hours after cells were switched to galactose media, with 7.5 OD units collected at each timepoint. RT-qPCR was carried out for three independent biological replicates as described above.