In this study, we identified a unique 3-amino-acid patch near the C terminus of histone H4, composed of L97, Y98, and G99, which when mutated led to a severe growth defect accompanied by polyploidy and aneuploidy. The precise structure of this patch must be important for function. Mutating Y98 to Ala led to a severe growth defect, while changing it to Phe did not. Gly 99 was clearly an important residue, because even a conservative change to Ala led to a significant growth defect, while mutating it to a larger residue, such as Leu or Asp, led to a more severe defect. Strikingly, mutating residues on either side of this 3-amino-acid patch did not lead to the same phenotypes, nor did mutating nearby nucleosomal residues on H2A or H2B (). Thus, we do not think that the phenotypes seen for these three H4 mutants are due to an altered nucleosome structure.
Yeast mutations of H4 Y98 have been the subject of previous studies. Santisteban et al. found that a Y98G mutation was lethal and that a Y98H allele grew poorly at 25°C and was temperature sensitive, while a Y98W mutant had no observable phenotype (26
). The lack of a phenotype seen for the Y98W mutant was similar to what we observed for a Y98F mutant (). In another study, a Y98A mutation was found to be lethal in one strain background and slow growing in another background (7
). In human tumors and in cultured cells exposed to nitric oxide, Y98 is sometimes found to be modified by nitration, and this modification of tyrosine is considered a biomarker for nitric oxide-dependent oxidative stress (11
). It is intriguing to consider that the genome instability often associated with cancer cells may be partially due to this modification of H4 Y98.
Polyploidy and a chromosome segregation defect of the mutants.
A striking phenotype of the three H4 mutants was that the strains, initially haploid, showed a rapid increase in ploidy upon further growth, and this increase was closely linked to improved growth (). This observation was consistent with a previous report showing that polyploidy and aneuploidy are common genetic alterations in evolving poorly growing yeast mutants (24
). We found that the H4 mutants rapidly became diploid, and when examined carefully by microarray analysis, many isolates actually had more than two copies of one or more chromosomes. Interestingly, in every case of aneuploidy, cells had 2 or 3 extra copies of chromosome XI, suggesting that overexpression of one or more genes on chromosome XI led to improved growth of the mutants. There were several histone-related genes on chromosome XI, including CSE4
. Nap1 protein is a histone chaperone for H2A-H2B dimers (20
). However, overexpression of NAP1
from a 2μm plasmid had no effect on the growth of the H4 mutants (data not shown). Cse4 is the histone H3 variant present at centromeres (18
). Overexpression of CSE4
did improve the growth of the mutants, but the extent of improvement was much less than that observed for the evolved aneuploid strains (data not shown). Hence, other transcriptional changes associated with the extra copies of chromosome XI, probably in combination with the alterations on other chromosomes, are required to fully rescue the growth defect of the mutants.
In addition to polyploidy, the H4 mutants exhibited a chromosome loss phenotype. Both chromosome III and a centromere-based plasmid were lost at a high frequency compared to the wild type (). The chromosome instability exhibited by the mutants, as well as synthetic sickness/lethality seen with a mad2 mutation (), suggested that the H4 mutants might have a defect in attaching the mitotic spindle to centromeric chromatin. Indeed, we found that kinetochore assembly was defective in the mutants, and the defect was almost always seen in the newer of the two spindles in cells in which spindle pole duplication had occurred ( and ). It was particularly significant that we observed this defect in strains with a plasmid expressing the mutant H4 but with both wild-type chromosomal H4/H3 genes, relying on the semidominant phenotype of the mutants. Presumably we would have seen a higher percentage of abnormal kinetochores if we had deleted one or both of the chromosomal H4/H3 genes.
We also found that CEN III chromatin had increased sensitivity to DraI nuclease in the case of all three H4 mutants compared to the wild type (A). However, examination of the GAL10 gene showed the same increased sensitivity to DraI cutting, suggesting that this effect might be genome wide (B). The kinetochore is likely to interact not only with the Cse4-containing nucleosome, but also with surrounding H3-containing ones (K. Bloom, personal communication). Thus, the defective kinetochore assembly of the mutants may be due to a lower nucleosome density in the centromere region or, conceivably, to a direct interaction of kinetochore components with H4 residues 97 to 99.
Global lower nucleosome density in the mutants.
In view of increased DraI sensitivity of the mutants at two loci, we looked at the micrococcal nuclease sensitivity of bulk chromatin and found a larger fraction of the chromatin was digested to mono- and dinucleosomes in the mutants than in the wild type (C). We also looked at H4 occupancy by chromatin immunoprecipitation at two loci expressed at a very low level, CEN III and GAL10, and at the highly expressed PMA1 gene. We found a lower occupancy for the mutants than for the wild type at all three loci (D). The combination of all these results leads us to conclude that the mutants have a lower than normal nucleosome density throughout the genome. We also considered the possibility that the altered chromatin structure changed gene expression in the H4 mutants, thus causing the increase in ploidy and the slow growth. A microarray analysis comparing gene expression of the mutants with that of the wild type found that no clear pattern emerged that could explain how altered gene expression caused the chromosome segregation defects (data not shown).
The roles of histone chaperones Rtt106 and CAF-I.
Affinity purification of the H4/H3 chaperone Rtt106 from yeast yielded a surprising result. Much more mutant H4 than wild-type H4 was bound to Rtt106 in each of the three H4 mutants (A). The amount of H3 bound was also correspondingly higher in the case of the mutants, which was not unexpected, since H4 and H3 both bind to Rtt106 and are deposited together, probably as a dimer. Purification of CAF-I from yeast using TAP-tagged Cac2 yielded results similar to those seen for Rtt106 for the H4 Y98A mutant, the one with the most severe phenotypes (B). However, in vitro
binding experiments provided a different result. Regardless of the source of histones, purified from either yeast or E. coli
, the mutant H4 bound slightly more weakly to recombinant Rtt106 than the wild-type H4 (C and D). Thus, the large amounts of H4 and H3 bound to Rtt106 purified from yeast in the case of the mutants (A) were not due to a greater affinity of the mutant H4 for Rtt106. Instead, they were due to a defect in the transfer of H4/H3 (we assume dimers) from Rtt106 to the next step in the deposition pathway, whether it was to another chaperone or to DNA itself. These results are depicted in cartoon form in A and B. The interpretation that deposition by Rtt106 and CAF-I is defective in the H4 mutants would explain why they have a lower nucleosome density on the chromatin, as judged by three different criteria. The centromeric region of the chromosomes may be particularly sensitive to this chromatin alteration, which would explain the kinetochore assembly and chromosome segregation defects observed for the H4 mutants. A previous study found that two H2A single-amino-acid replacement mutants showed an increase-in-ploidy and a chromosome loss phenotype, similar to what we found for the three H4 mutants. In the case of the H2A mutants, both genetic and biochemical experiments suggested that the phenotypes were due to an altered centromeric chromatin (22
Fig. 10. Model for the deposition defect caused by the histone H4 mutants. (A) Rtt106 or CAF-I deposits wild-type H4/H3 dimers (or possibly tetramers) onto DNA. (B) Mutant (MT) H4/H3 dimers accumulate on Rtt106 or CAF-I and cannot be transferred to the next step. (more ...)
We also observed strong genetic interactions between the three H4 mutants and RTT106, CAC1, and CAC2. Strains expressing wild-type H4 and one of the H4 mutants from different plasmids exhibited much better growth in the presence of a rtt106Δ, cac1Δ, or cac2Δ mutation (A and D). Conversely, overexpression of Rtt106 exacerbated the poor growth of the H4 mutants (C). These genetic interactions can be explained with the following model based on the large accumulation of mutant H4 seen bound to Rtt106 purified from yeast and of the Y98A mutant in the case of CAF-I. We suggest that in cells expressing both wild-type and mutant histones, the deposition of wild-type H4/H3 through Rtt106 or CAF-I is also affected, since a certain fraction of Rtt106 and CAF-I proteins have mutant histones bound to them nonproductively (C). As argued above, poor histone deposition causes a lower nucleosome density on the chromatin and, hence, poor growth. We propose that when RTT106, CAC1, or CAC2 is deleted, the deposition of histones is taken over by one of the other chaperones present in cells, and these chaperones do not bind mutant H4/H3 nonproductively the way Rtt106 or CAF-I does (D). This can explain why the growth of the histone mutants was improved by deletion of RTT106, CAC1, or CAC2. Similarly, overexpression of Rtt106 bound even more mutant H4/H3 and thus caused greater toxicity. Surprisingly, no apparent improvement in growth was observed in rtt106Δ or cac1Δ after the plasmid expressing wild-type H4 was removed by 5-FOA treatment, leaving mutant H4 as the only source of H4. One possible explanation is that in that case, the whole chromosome is occupied by mutant H4, and that might cause other defects that cancel the positive effects of alternate, redundant chaperones.
In conclusion, we have identified a small domain near the C terminus of histone H4 that is important for genome stability. When this domain is mutated, it leads to less deposition of H4/H3 onto chromatin, which in turn causes genome instability. Structural studies of the interaction between H4/H3 and the histone chaperones Rtt106 and CAF-I should shed further light on why this domain of H4 is so important for H4/H3 deposition.