The ratio of cell to nuclear size is relatively constant across eukaryotes, but the ratio of genome size to nuclear volume is more variable, with large (defined as > 1 Gbp) genomes packed into nuclei only slightly larger than those of small (< 100 Mbp) eukaryotic genomes. Macadangdang, et al. supply a mechanism for the underbalance between genome size and nuclear volume, showing that arginine residues at conserved positions in the N terminal domain of histone H2A are associated with, and can recapitulate in vitro and in vivo, greater condensation of chromatin.

Histone 2A is one of the four subunits that make up the octamer of the nucleosome. While it’s well established that linker histones, like H1, have sequence variation and posttranslational modifications that impact chromatin structure, the core histones show more sequence conservation across eukaryotes. Macadangdang, et al. started their search for histone variants by extracting each of the four core histone protein from 160 sequenced eukaryotic genomes. Splitting these into three groups based on genome size (small < 100 Mpb, medium 100 Mbp-1 Gbp, large > 1 Gbp), they scanned for amino acids that showed significantly different frequencies across the three groups separately for three protein domains. The increased frequency of arginine in larger genomes in the N-terminal domain of H2A are presented in this paper.

We felt that the evolutionary context was a bit weak, and had a few awkward phrasings (a lysine ‘evolves to an arginine’). We were conflicted on whether phylogenetic correction would be appropriate, since genome size is pretty labile even within species, but think their uncorrected analysis probably works best since it’s exceedingly difficult to decide on a model of evolution. It’s also unclear to what degree alternate alignments might have impacted their findings. Muscle with (presumably) default settings might not be appropriate for considerations across all eukaryotes. I’m not familiar enough with protein substitution matrices, but it seems pan-eukaryote alignment might need some real thought about gap penalties and alignment scores.

Beyond the evolutionary evidence, experiments both in vivo and in vitro corroborate the important role of arginines at these conserved positions in chromatin condensation.

Using a yeast strain with H2A on a plasmid and the chromosomal copies removed, they can generate a slew of mutants that make the H2A sequence of yeast look more like that of a large-genomed organism. To measure the effect of these mutations on chromatin compaction, they use FISH to identify the proximity between sets of probes ~100-300 kb apart. If the mutation compacts chromatin, the interprobe distance will decrease, as DNA is more tightly wound. They find arginines at positions 3 and 11 are sufficient to decrease the probe distance for four different probe sets, especially when these residues are spaced closer to the histone fold domain, as they are in larger-genomed organisms. These strains were also imaged using tagged nuclear pore proteins, and smaller nuclear volumes were observed in R11, despite constant cell size. We wondered whether the observation across eukaryotes of constant ratio of nuclear volume to cellular volume results from selection for compensatory mutations that shift cell size, as these results suggest nuclear volume can be altered independently of cellular volume.

They show that in an exponential growth phase, these yeast mutants have highly correlated gene expression. Of the small set of genes with more than two-fold differential expression, they can’t identify a gene ontology that is significantly enriched. They infer that chromatin compaction doesn’t signficantly alter global gene expression, but we wondered whether there were absolute changes in the level of global transcription.

We were a bit puzzled by the fact that while mutant vs. wild-type strains showed similar growth rates, mutant strains, regardless of the mutation’s effect on chromatin compaction, were always outcompeted. This may simply be due to a low power to distinguish small differences in growth rate which would be magnified in any competition experiment.

Their yeast studies recreate a large genome H2A in a small-genomed organism. To mimic small genome H2A in a large-genomed organism, they overexpress wt and mutant H2A in human cell lines and measure chromosome compaction again by FISH. When R3 and R11 are mutated, interprobe distance increases, again supporting the role of arginine in compaction.

To further show this is not just some artifact of in vivo systems, they generate chromatin fibers with repeats of the 601 nucleosome positioning sequence, a sequence typically used for in vitro nucleosome assembly, and histone octamers from Xenopus laevis. This large-genomed organism has an arginine at position 11, and they generated mutants with R11 deleted. H2A mutants without R11 were less compacted, and this bulkier confirmation resulted in a lower sedimentation coefficient. This was good evidence that there wasn’t some kind of posttranslational modification to H2A that could account for differences in chromatin configuration.

Cellular phenotypes of cancer often present less chromatin compaction than wild type cells. While this may be due to global hypomethylation of DNA and linker histones, the majority (71%) of missense mutations in the N-terminal domain of H2A are within an arginine motif (the arginine itself or the immediately adjacent amino acids, often nonpolar). When three of these arginine mutations are expressed in human fibroblasts, they decrease chromatin compaction, but don’t alter nuclear volume. It is unclear when these H2A mutations occur in the progression of cancer, but this provided yet more evidence for the role of arginine residues in the NTD impacting chromatin configuration.

Something that seemed underexplained in the text was that different isoforms of H2A exist in many of the genomes they tested. Histone proteins are not single copy genes - for example, there are 17 canonical H2A genes in the human genome (plus H2A genes coopted for other purposes, like H2AX for DNA repair). A supplementary spreadsheet (available in the Additional Files tab in the zip file) documents the 56 (of 160) species which have 2-7 canonical isoforms of H2A. For example, Mus musculus and Zea mays have isoforms with and without arginine at position 3, and Zea mays contains 5 isoforms with variants at three of the four important arginine residues. It’s unclear whether these isoforms are differentially expressed, and whether this variation has an impact on chromatin compaction within a genome.

Michelle Stitzer