Background

Lysine residues on histone H3, which DNA wraps around, undergo post-translational modifications such as methylation and acetylation to regulate gene expression and chromatin structure. The function of specific modifications has primarily been inferred by inhibiting enzymes that add or remove them. However, a single enzyme can modify multiple proteins and perform non-catalytic functions, and different enzymes may compensate for the same site, making it difficult to attribute observed phenotypes solely to the effect of a specific histone mark.

Direct substitution of histone residues can reduce such ambiguity, but implementation in mammals is challenging. In mice, the canonical H3.1 and H3.2 are encoded by 9 and 3 genes, respectively, scattered across multiple chromosomal clusters, and the non-canonical H3.3 is produced by two genes, H3f3a and H3f3b. Conventional base editing has limited editing windows and may inadvertently alter unintended neighboring bases, limiting precise manipulation of multiple H3 paralogs. The research team from the Nature Genetics original paper addressed this issue using CRISPR prime editing.

Key Findings

The team combined Cas9 nuclease with reverse transcriptase in PEmax, used MLH1dn to temporarily suppress mismatch repair, and employed enhanced prime editing guide RNA (epegRNA). Among PE2, PE3b, PE4, and PE5b, PE4 and PE5b with MLH1dn showed the highest editing efficiency and lowest insertion/deletion frequency. They then engineered mouse embryonic stem cell lines inducible with doxycycline to express PEmax and MLH1dn, reducing the burden of long-term mismatch repair suppression.

In experiments replacing H3K23 with arginine, 35% of single-cell clones achieved over 80% editing efficiency. Two clones completely substituted not only the 18 alleles of H3.1 but also H3.2, and the H3K23 acetylation signal was nearly eliminated. Reverse editing to restore the lysine and combinatorial editing of H3K14, K18, and K23 simultaneously were also feasible, though triple editing efficiency was approximately 4–5 times lower than single residue editing.

By comparing cells with lysine-to-arginine substitutions and isogenic controls maintaining the original amino acid, the team confirmed that H3K4, K9, K14, K18, and K79 are critical for embryonic stem cell fitness. Isogenic editing of H3K14 and K18 reached 70–100% and 75–90%, respectively, but the corresponding arginine variants were selectively lost in highly edited cells. H3K4R showed poor persistence, with only two out of 36 clones retaining 11–12% editing after 13 rounds of re-introduction.

When canonical H3K18 was altered, H3.3 was more frequently incorporated into promoters and enhancers, compensating for acetylation loss. This redistribution disappeared when H3.3 was also substituted. H3K56R cells were more sensitive to the DNA-damaging agent etoposide, and cells with combined H3K27R and H3K36R or H3K37R showed reduced colony formation and self-renewal capacity compared to single mutants.

Implications and Outlook

This study provides a platform to precisely test residues where modifications occur, rather than relying on histone modification enzymes. Its strength lies in the ability to dissect compensatory interactions between canonical and non-canonical H3 variants and functional interactions between different lysines. It also allows validation of how histone variants found in cancer alter cell growth, transcription, and DNA repair within the same framework.

Safety concerns remain. In clones targeting multiple genomic loci, histone cluster copy number loss ranged from 4–34% depending on the condition, and large deletions (25 kilobases) and megabase-scale losses on one side of chromosome 13 histone clusters were observed. Removing nicking RNA did not fully eliminate the risk at all loci. The results are also limited to cultured mouse embryonic stem cells. For expansion to human cells, differentiated tissues, and in vivo models, long-range genomic analysis and independent clone validation must become standard procedures.