Figure 1. Design and testing of asymmetric nucleosomes.

(A) Schematic of asymmetric histone H3 design. Left: Wild-type H3/H4 tetramers are symmetrical, with H3-H3 interactions serving as the dimerization interface. The H3 N-terminus, site of many modifiable residues, is indicated protruding from the globular domain. Right: Obligate heterodimer H3s are comprised of distinct ‘X’ and ‘Y’ interaction partners which are altered to prevent either H3 from homodimerizing. (B) Computational models of the designed histone H3 heterodimer. (1) The four-helix bundle comprising the H3-H3 C-terminal dimerization interface (pdb 1KX3). Leucine and alanine residues at the hydrophobic interface are indicated. (2) The designed asymmetric mutations pack efficiently as a heterodimer, with partners colored in green (H3Y) and yellow (H3X). Indicated residues at the H3-H3 interface were engineered to form a bump-hole interface to increase interaction affinity. (3-4) Altered H3s are designed not to homodimerize due to van der Waals clashes (3) or voids (4) in the hydrophobic core. (C–D) Genetic analysis of heterodimeric H3X/H3Y pairs. C)H3Y alone cannot support growth. Images show growth of yeast carrying wild-type H3 on a URA3-marked plasmid as well as either empty plasmid, wild-type H3, H3Y alone, or both H3X and H3Y. Three independent transformants for each strain were grown on 5-FOA to select against the URA3 shuffle plasmid. (D) H3X alone cannot support growth. Left panels: growth of three independent transformants of various H3X strains (evolved second-generation mutations indicated at left) in the absence of H3Y. Bottom: positive control with a TRP1-marked wild-type histone plasmid. Right panels: growth of indicated H3X strains in the presence of H3Y. Synthetic complete (SC) plates were grown 3 days, FOA plates 9 days. No growth on FOA was observed for any of the four evolved H3X alleles in the absence of H3Y, even after extended incubation. (E) Sequences of final H3X-H3Y molecules. All four H3X variants have been validated genetically, while two variants – 126A/130V and 126V/130A – have been validated biochemically (see Figure 2 and not shown). 126A is used for all functional analyses.

Figure 1—figure supplement 1. Optimization of the H3X/H3Y design.

(A) Genetic analysis of the original H3X design. The top four panels are identical to Figure 1C, showing growth of H3X/H3Y-bearing yeast strains but no growth of H3Y-bearing strains, and are duplicated here for comparison to H3X growth phenotypes. Bottom two panels: Yeast bearing a wild type histone-URA3 shuffle plasmid were transfected with the indicated plasmids carrying a wild-type H4 gene and the original H3X (126V, 130V) gene. Three independent isolates of each strain were patched onto 5-FOA media to select for loss of the wild type histone-URA3 plasmid. Strains were grown for the indicated number of days and photographed. The growth of papillae in the absence of H3Y upon extended incubation led us to optimize H3X by randomization of four key residues, as shown below. (B) Sequencing traces for the four ‘H3X*’ libraries, each with one randomly mutagenized codon as indicated. Although the base utilization in the synthetic oligonucleotides is not completely equal, all four bases are readily detected at each position. (C) Schematic of the screening strategy. Transformants carrying the H3X* library alleles were plated on SC-Trp and replicated onto FOA plates. Strains that couldn’t survive on FOA with H3* alone were transformed with an H3Y-expressing plasmid, and then were tested on FOA media to assess the viability of the H3X*-H3Y combinations (as shown in Figure 1C). As an additional criterion, we tested whether the two plasmids expressing the new H3X/H3Y heterodimers were stable when cells were cultured in rich media in the absence of selection for either plasmid. Repeated rounds of streaking on rich (YPD) plates or culturing in YPD media yielded colonies that all maintained a Trp⁺Leu⁺ phenotype, indicating that neither plasmid could be lost. Furthermore, resequencing of plasmids isolated after growth on YPD detected no reversion of the engineered H3X or H3Y in 96 independent isolates. Together, the genetic data strongly suggest that the H3X/H3Y combinations are obligate heterodimers, and that neither H3X nor H3Y can homodimerize at sufficient levels to support viability.