Sometimes one just isn't enough: do vertebrates contain an H2A.Z hyper-variant?

How much functional specialization can one component histone confer on a single nucleosome? The histone variant H2A.Z seems to be an extreme example. Genome-wide distribution maps show non-random (and evolutionarily conserved) patterns, with localized enrichment or depletion giving a tantalizing suggestion of function. Multiple post-translational modifications on the protein indicate further regulation. An additional layer of complexity has now been uncovered: the vertebrate form is actually encoded by two non-allelic genes that differ by expression pattern and three amino acids. See research articles http://www.biomedcentral.com/1741-7007/7/86 and http://www.biomedcentral.com/1471-2148/9/31.

All species have to perform a balancing act with their genome: cram it into the cell (in the case of eukaryotes, a small part of that cell: for example, about 2 m of human DNA in a nucleus of about 5 μm diameter) yet make the appropriate regions readily available for replication and expression (and repair if something goes wrong). Eukaryotes achieve this feat by wrapping their DNA into chromatin, a highly ordered complex with a simple repeat ing unit of about 146 bp DNA plus eight histone proteins, termed the nucleosome. Th is beads-on-a-string array is further assembled into a variety of higher-order structures all the way up to the metaphase chromosome. Th e chromatin fi eld exploded with the discovery that the monotonous-looking nucleosomes are actually hugely variable, with post-translational modifi cations of numerous residues on the major histones, their substitution for histone variants, and even post-translational modifi cations of the variants. Furthermore, the highly regulated use of these marks serves to distinguish regions of DNA (such as promoters, centromeres or damaged DNA) to the appropriate enzymatic machineries.
Histone variants are non-allelic isoforms of the canonical histones that can be assembled into nucleosomes in their place, and are thought thereby to provide the basis for regulation of biological processes that require local access to DNA. In contrast to the S-phasecoupled synthesis of the major histones (timing availability to the peak demand of genome replication), variants are generally expressed throughout the cell cycle. Histone H2A has one of the largest variant families, and includes H2A.Z, a protein that is highly conserved across eukaryotes but diff ers considerably from the major H2A in each species (Figure 1) [1]. H2A.Z has been ascribed a large number of roles, including most recently suppressing antisense RNAs [2] and stabilizing the association of condensin with mitotic chromosomes [3]. Although we still have a poor understanding of how the variant mediates any specifi c function, it is likely that diff erential enrichment at specifi c locations and distinct post-translational modifi cations contribute. H2A.Z at the inactive X chromosome of mammalian female cells is monoubiquitinated [4], that in budding yeast (Saccharomyces cerevisiae) is sumoylated [5], and in all tested species it is subject to multiple amino-terminal acetylations (Figure 1), primarily by the Kat5 family of acetyltransferases [1,3]. Mutation of the S. cerevisiae sumoylation sites impairs movement of DNA double-strand breaks to the nuclear periphery [5], whereas an unacetylatable allele in fi ssion yeast (Schizosaccharomyces pombe) recapitulates many of the phenotypes of a complete deletion [3]. However, it is still unknown whether the eff ect of any of these modifi cations is direct (such as steric hindrance or charge modulation infl uencing the formation of higher-order structures) or indirect (such as generating sites for the recruitment of regulatory proteins).

The impact of H2A.Z on individual nucleosomes
Th e major route for H2A.Z into chromatin is via Swr1, the Snf2-family ATPase at the catalytic center of the SWR

Abstract
How much functional specialization can one component histone confer on a single nucleosome? The histone variant H2A.Z seems to be an extreme example. Genome-wide distribution maps show non-random (and evolutionarily conserved) patterns, with localized enrichment or depletion giving a tantalizing suggestion of function. Multiple posttranslational modifi cations on the protein indicate further regulation. An additional layer of complexity has now been uncovered: the vertebrate form is actually encoded by two non-allelic genes that diff er by expression pattern and three amino acids. chromatin remodeling complex. Th is is certainly the case in S. cerevisiae (Swr1), S. pombe (Swr1), humans (SRCAP), Drosophila (Domino) and Arabidopsis (PIE1) [2,3,[6][7][8].
Th e presence of H2A.Z in a nucleosome facilitates intramolecular folding to higher-order arrays, particularly 30 nm chromatin fi bers, although these resist the formation of more highly condensed structures resulting from intermolecular association [6]. H2A.Z has also been reported to have a subtle destabilizing eff ect on the nucleosome in which it is incorporated, although this has been disputed. Th e apparent contradictions in the data can however be reconciled if the total histone composition of a nucleosome octamer is considered: variants seem to subtly alter nucleosome stability, so the order of stability is H3/ H2A.Z = H3/H2A > H3.3/H2A > H3.3/H2A.Z [9]. Furthermore, although it was originally predicted that H2A.Z was unlikely to form hybrid nucleosomes (Figure 1), both homotypic (containing two H2A.Z:H2B dimers) and hetero typic (containing H2A:H2B and H2A.Z:H2B dimers) forms have been observed, adding yet another level of structural (and possibly functional) heterogeneity [6,10].

Genome distribution maps of H2A.Z are suggestive of function
High density maps of H2A.Z across genomes as diverse as S. cerevisiae [6], S. pombe [8], Arabidopsis thaliana [7] and Caenorhabditis elegans [2] show the variant to be widely but non-randomly distributed. In budding yeast H2A.Z occupancy peaks in the single nucleosomes directly fl anking (-1/+1) a nucleosome-free region over promoters, a pattern apparently induced by the nucleosome-free region itself [6]. Fission yeast, in contrast, shows enrichment in the +1 but not the -1 nucleosome around the nucleosome-free region [8]. Th e reason for (or outcome of ) this diff erence is unknown, although in each organism enrichment depends on the SWR complex and is inversely correlated with transcriptional activity. H2A.Z has been linked to both transcriptional activation and repression at various genes, although consensus seems to be building towards a role in marking and/or poising promoters for expression. Th e presence of H2A.Z might commit local chromatin to a state competent for activation by other factors or, in higher eukaryotes, it might protect the region from inactivating DNA methylation [7]. Or it might be that the cell uses the inherent instability of H2A.Z-containing nucleosomes to regulate promoter accessibility.

H2A.Z is found in 'hot' nucleosomes at promoters and euchromatin-heterochromatin boundaries
Studies on the dynamics of replication-independent histone turnover in budding yeast suggest that H2A.Zcontain ing nucleosomes have signifi cantly higher turnover rates [6]. Th is 'hotness' of the nucleosomes around transcription start sites could aid promoter function by making it easier to expose these DNA elements to the H2A.Z proteins across species (green) and between H2A.Z and H2A (blue). The relative location of the nuclear localization signal (NLS) and regions of S. cerevisiae (Sc) H2A.Z (called Htz1) that mediate contact with the Nap1 chaperone and the SWR-complex (SWR-C) ATPase complex are also shown. All H2A.Z post-translational modifi cations identifi ed so far are on the relatively divergent amino and carboxyl termini, so it is unclear whether each specifi c modifi cation is invariably used to regulate variant function across species. Addition of post-translational modifi cations generally depends on the SWR complex, indicating that each modifi cation occurs after the variant is assembled into chromatin [3,5]. A major region of diff erence between H2A and H2A.Z is in the Loop 1 domain, which regulates interaction between the two H2A molecules in a nucleosome. This has led to the suggestion that nucleosome core particles can only be homotypic, containing either H2A or H2A.Z. However, hybrid nucleosomes containing H2A:H2B and H2A.Z:H2B dimers have been observed [6]. The sequence of human H2A.Z-1, including the three residues that diff er in H2A.Z-2, is also indicated. transcriptional machinery. Such rapid fl ux might also abrogate the spread of a propagating domain (such as that of the budding yeast Sir2 deacetylase complex), which could explain how H2A.Z mediates a heterochromatin-euchromatin boundary function in this organism [6]. H2A.Z enrichment at the regions between euchromatin and heterochromatin is a feature also found in other organisms, despite the dramatic biochemical diff erences between the heterochromatin of budding yeast and that of many other species. In this manner the mammalian variant is a component of pericentric heterochromatin and fl anks sites occupied by the insulator binding protein CTCF [6,11].

A vertebrate H2A.Z hyper-variant
As if the complexity described above wasn't enough, recent mass spectrometry analyses of chicken erythrocytes identifi ed two forms of H2A.Z that diff er by just three amino acids: H2A.Z-1 (previously H2A.Z) and H2A.Z-2 (previously H2A.F/Z or H2A.V) ( Figure 1) [11]. Th ese proteins are encoded by two non-allelic genes, which phylogenetic analyses indicate are present in all vertebrates and have a common origin early in chordate evolution [12]. Both isoforms are incorporated into chromatin, and both seem to be acetylated on the same three lysine residues within the amino terminus (Lys4, Lys7 and Lys11) to a similar degree (Figure 1) [11]. Why would vertebrates need two copies of a protein that diff er by just three amino acids, whereas invertebrates do just fi ne with just one? Mouse studies ostensibly showing that H2A.Z is indispensable deleted only H2A.Z-1, indicating non-redundancy: that is, H2A.Z-2 cannot compensate. At this stage, we can only speculate about the relevance of each isoform. Th e three-aminoacid diff erence is not expected to have any major structural implication for nucleosomes [12]. Th e H2A.Z isoforms could have a diff er en tial affi nity for various chaperones and/or deposi tion machineries, which could explain their subtly diff er ent chromatin occupancy patterns [11]. Although this might sound unlikely, there is a comparable precedent: three of the four amino acids that diff er between histones H3.1 and H3.3 regulate the usage of the respective proteins in the replicationdependent and -independent deposition pathways [13]. However, the most important diff erence between H2A.Z-1 and H2A.Z-2 may be their highly divergent promoter sequences [11]. Th is opens up the possibility of dramatically diff erent temporal and/or spatial expression patterns for the two isoforms. Indeed, preliminary studies suggest some diff erences in mRNA expression levels depending on the developmental stage of a variety of tissues [11].
Plants also have multiple H2A.Z isoforms. Th ree have been reported in A. thaliana -HTA8, HTA9 and HTA11which share about 90% identity but have distinct expression patterns, with HTA9 alone being cell-cycleindependent [7]. PIE1 (the plant homolog of Swr1) interacts with all three variants, but not with H2A. Single knockouts have no distinct phenotypes, but double hta9/ hta11 knockouts show developmental abnormalities. It remains to be seen whether these variants of the variant have any tissue or developmental function [7]. However, the parallel with vertebrates is striking and may suggest that more complex organisms need more fi nely tuned chromatin than one H2A.Z can provide.