The Rapidly Evolving Centromere-Specific Histone Has Stringent Functional Requirements in Arabidopsis thaliana

Centromeres control chromosome inheritance in eukaryotes, yet their DNA structure and primary sequence are hypervariable. Most animals and plants have megabases of tandem repeats at their centromeres, unlike yeast with unique centromere sequences. Centromere function requires the centromere-specific histone CENH3 (CENP-A in human), which replaces histone H3 in centromeric nucleosomes. CENH3 evolves rapidly, particularly in its N-terminal tail domain. A portion of the CENH3 histone-fold domain, the CENP-A targeting domain (CATD), has been previously shown to confer kinetochore localization and centromere function when swapped into human H3. Furthermore, CENP-A in human cells can be functionally replaced by CENH3 from distantly related organisms including Saccharomyces cerevisiae. We have used cenh3-1 (a null mutant in Arabidopsis thaliana) to replace endogenous CENH3 with GFP-tagged variants. A H3.3 tail domain–CENH3 histone-fold domain chimera rescued viability of cenh3-1, but CENH3''s lacking a tail domain were nonfunctional. In contrast to human results, H3 containing the A. thaliana CATD cannot complement cenh3-1. GFP–CENH3 from the sister species A. arenosa functionally replaces A. thaliana CENH3. GFP–CENH3 from the close relative Brassica rapa was targeted to centromeres, but did not complement cenh3-1, indicating that kinetochore localization and centromere function can be uncoupled. We conclude that CENH3 function in A. thaliana, an organism with large tandem repeat centromeres, has stringent requirements for functional complementation in mitosis.CENTROMERES are essential for chromosome inheritance, because they nucleate kinetochores, the protein complexes on eukaryotic chromosomes that attach to spindle microtubules. Despite the essential requirement for centromeres in chromosome segregation, their DNA sequences and the sequences of kinetochore proteins are highly variable. Kinetochores in Saccharomyces cerevisiae and related budding yeasts assemble on small, unique centromere DNAs (125 bp in S. cerevisiae) (Meraldi et al. 2006). Centromere DNAs in the fission yeast Schizosaccharomyces pombe are larger, consisting of a central core sequence of 4–5 kb, which binds kinetochore proteins, flanked by large inverted repeats whose heterochromatic nature is important for centromere function (the total size of the S. pombe centromere DNA is 35–110 kb). At the other extreme from small yeast centromeres are holocentric organisms, such as Caenorhabditis elegans, in which kinetochore proteins bind along the entire length of mitotic chromosomes (Dernburg 2001). Most plants and animals have extremely large centromere DNA tracts consisting of megabases of simple tandem repeats. The repeat sequence evolves extremely rapidly, and only a small fraction of the repeat array is likely to be bound by kinetochore proteins. Furthermore, kinetochores can be nucleated by noncentromeric DNA sequences in plant and animal cells (Amor and Choo 2002; Nagaki et al. 2004; Nasuda et al. 2005; Heun et al. 2006; Wade et al. 2009). Despite these findings, the maintenance of massive centromere repeat arrays in both animal and plant taxa suggests that repeats are a central feature of centromere biology in these organisms.Although centromere DNAs are extremely diverse, all eukaryote kinetochores contain the centromere-specific histone H3 variant CENH3 (originally described as CENP-A in human) (Henikoff and Dalal 2005; Black and Bassett 2008). CENH3 replaces conventional H3 specifically in a subset of centromere nucleosomes. It is essential for kinetochore function in all eukaryotes where this requirement has been tested. Conventional histones are among the most conserved proteins in eukaryote genomes. In contrast, CENH3 is rapidly evolving. The C-terminal histone-fold domain, which complexes with other histones to form the globular nucleosome core, can be aligned with conventional H3''s but evolves rapidly and shows signatures of adaptive evolution in some residues (Malik and Henikoff 2001; Talbert et al. 2002; Cooper and Henikoff 2004). The N-terminal tail domain of conventional histone H3 protrudes from the nucleosome core and is not resolved in the structure solved by X-ray crystallography (Luger et al. 1997). In CENH3, the tail domain evolves so rapidly that its sequence can barely be aligned between closely related species.Experiments in yeast and in animals have delineated functionally important regions within CENH3. S. cerevisiae kinetochores contain only a single CENH3/Cse4p nucleosome (Furuyama and Biggins 2007). In S. cerevisiae Cse4p, amino acid residues required for normal function are distributed throughout the histone-fold domain (Keith et al. 1999). The N-terminal tail of Cse4p contains an essential region termed the END domain, but overexpression of a Cse4p lacking the tail altogether can rescue a cse4 deletion mutant (Chen et al. 2000; Morey et al. 2004). In Drosophila melanogaster cells, CENH3/Cid from the distantly related D. bipectinata did not localize to kinetochores unless a specific region of the histone-fold domain, loop 1, was swapped with the corresponding region from D. melanogaster CENH3/Cid (Vermaak et al. 2002). In human, the histone-fold domain is important for centromere targeting (Sullivan et al. 1994). The functionally important region within the histone-fold domain was further defined by inserting loop 1 and the α-2 helix from CENH3/CENP-A (termed the CENP-A targeting domain, or CATD) into conventional H3 (Black et al. 2004). H3 containing the CATD acquires several functions of CENP-A when expressed in human cells. It localizes to kinetochores, binds the kinetochore protein CENP-N, has a rigid secondary structure when assembled into nucleosomes, and can restore normal chromosome segregation in cells depleted for CENP-A using RNA interference (RNAi) (Black et al. 2004, 2007a,b; Carroll et al. 2009).Despite these extensive studies, questions about structure–function relationships within CENH3 remain. CENH3 function may differ between small yeast centromeres and the large tandem repeat centromeres of animals and plants, particularly because larger centromere DNAs are likely to contain many more CENH3 nucleosomes and may require a higher level of organization. Experiments in D. melanogaster and in human cells have used RNAi to downregulate the endogenous protein, and a conditional knockout has been made in chicken DT-40 cells (Blower and Karpen 2001; Goshima et al. 2003; Regnier et al. 2005; Black et al. 2007b). These experiments are challenging because CENH3 is very stable. If preexisting CENH3 is partitioned equally between duplicated sister centromeres, its amount will be approximately halved at each cell division. Therefore the protein may persist for many cell divisions after induction of RNAi, as shown by Western blots indicating that ∼10% of endogenous CENH3 remains in human cells subjected to two rounds of RNAi (Black et al. 2007b).We have chosen to study CENH3 in the model plant A. thaliana, which combines facile genetics and transgenesis with centromere DNA structure that is similar to most plants and animals (megabases of tandem repeats with a repeating unit of 178 bp) (Murata et al. 1994; Copenhaver et al. 1999). Although Drosophila and mouse CENH3 knockout mutants have been characterized (Howman et al. 2000; Blower et al. 2006), a large-scale structure–function analysis of CENH3 has not been attempted in these organisms. A cenh3 null mutant in A. thaliana allows us to completely replace the endogenous protein with transgenic variants (Ravi and Chan 2010). Here we report four major conclusions regarding CENH3 function in A. thaliana: (1) CENH3 function requires an N-terminal histone tail domain, although either the CENH3 tail or the H3 tail can support mitotic chromosome segregation. (2) Inserting the CENP-A targeting domain of CENH3 into H3 does not confer CENH3 function. (3) Complementation of cenh3 by heterologous CENH3 requires that the species of origin be closely related to A. thaliana. (4) Localization of a heterologous CENH3 protein to kinetochores in the presence of native CENH3 does not necessarily indicate that it can complement a cenh3 mutant. Overall, our results indicate that requirements for CENH3 function in A. thaliana are more stringent that those obtained in human cells. They underscore the usefulness of comparative studies of centromere function using genetically tractable experimental organisms.