Steven Henikoff

There has been extraordinary progress in molecular biology during the 50-year span that began with the discovery of the DNA double helix and culminated with the nearly complete specification of our genetic inheritance. In contrast, the inheritance of differences between cells and tissues is poorly understood. To better understand inheritance that does not depend on DNA sequence, we apply genomic and evolutionary tools to the study of epigenetic markers, such as histone variants and DNA methylation.

The bulk of the eukaryotic genome is packaged into nucleosome particles, each of which comprises an octamer of four core histones--H2A, H2B, H3, and H4--which wrap nearly two turns of DNA. Nucleosomes can be differentiated both by numerous post-translational histone modifications and by incorporation of a few histone variants. H3 variants fall into three categories: canonical H3, which is deposited during replication; H3.3, which is the general constitutive form; and CenH3 (CENP-A in mammals), which is deposited exclusively at centromeres.

CenH3 nucleosomes wrap only a minute fraction of the genomic landscape and have received little attention relative to bulk nucleosomes. We wondered whether the presence of CenH3 might result in a profoundly different nucleosome, and so we biochemically characterized CenH3 nucleosomes and directly visualized them in their native form. We found that, in stark contrast to octameric bulk nucleosomes, centromeric nucleosomes are stable heterotypic tetramers with one copy of CenH3, H2A, H2B, and H4 each, wrapping only one turn of DNA, the first example of a stable half-nucleosome. In addition, our studies of CenH3 nucleosome assembly have led to our discovery of another unexpected difference between chromatin at centromeres and chromatin on chromosome arms. Using the classical plasmid supercoiling assay for nucleosome assembly, we found that CenH3 induces positive DNA supercoils both in vitro and in vivo, implying a right-handed wrap. This is the opposite direction of wrapping of conventional nucleosomes. The right-handed wrapping of DNA around the histone core implied by positive supercoiling means that interaction surfaces between histones that prevent the nucleosome core from springing apart would be facing away from one another. The mutual incompatibility of nucleosomes with opposite topologies can potentially explain how centromeres are efficiently maintained as a unique loci on chromosomes.

Our studies of H3.3 have revealed that it is deposited by a distinct nucleosome assembly pathway. Whereas canonical H3 is incorporated strictly during DNA replication, amino acid changes toward H3.3 allow replication-independent (RI) deposition. Active genes are the predominant sites of RI deposition, and so H3.3 is a marker for active chromatin. This paradigm is general, and we find that H3.3 marks active chromatin in the germline and during the early cleavage divisions of C. elegans embryos. We have also applied genome-wide profiling to H3.3 to gain a better understanding of chromatin dynamics. This has revealed a correspondence to sites responsible for epigenetic regulation and to promoter regions genome-wide. In addition, we have introduced a salt fractionation procedure to isolate active chromatin and profile likely intermediates in nucleosome dynamics. By directly interrogating chromatin based on its physical properties, our method has been able to detect unstable nucleosomes in regions that have been generally thought to be nucleosome-free.

Another mode of epigenetic inheritance is DNA methylation. We have introduced a method for profiling DNA methylation in Arabidopsis using microarrays, and provided strong support for the role of DNA methylation in silencing transposons. In addition, we have discovered that a large fraction of Arabidopsis genes are methylated. Unlike transposon methylation, which occurs on nearly all cytosines, genic methylation is exclusively found at CG dinucleotides and is present in short dense clusters. What is responsible for these DNA methylation patterns? We find that DNA methylation and the histone H2A.Z variant are mutually antagonistic in Arabidopsis, and we are currently exploring the possibility that this antagonism is more general, for example in cancer. We have recently turned our attention to DNA methylation in developing Arabidopsis seeds, where the endosperm and embryo co-exist as independent products of double fertilization in the female gametophyte. We have profiled Arabidopsis DNA methylation genome-wide in the embryo and endosperm. We found that large-scale methylation changes accompany endosperm development and endosperm-specific gene expression. Transposable element fragments are extensively demethylated in the endosperm. We discovered several new imprinted genes by identifying candidates associated with differentially methylated regions. Our data suggest that imprinting in plants evolved from genome defense against transposable elements.

In addition to our chromatin studies, we continue to develop tools to facilitate our own research and that of others. These have included computational tools for interpreting sequence information, such as SIFT (sorting intolerant from tolerant) for predicting deleterious mutations, and the DamID method for mapping in vivo binding targets of chromatin proteins and transcription factors. We have also introduced a general reverse genetics strategy, called TILLING (targeting induced local lesions in genomes), whereby chemical mutagenesis is followed by screening for point mutations. The resulting allelic series can be used to determine gene function in the context of a whole organism. In our TILLING technique, heteroduplexes formed from denatured and annealed PCR (polymerase chain reaction) products are cleaved at single-base mismatches and the resulting fragments detected on electrophoretic gels. We have established TILLING services for several organisms, including Arabidopsis, Drosophila and maize.