Domain architectures of the Scm3p protein provide insights into centromere function and evolution

Abstract

Recently, Scm3p has been shown to be a nonhistone component of centromeric chromatin that binds stoichiometrically to CenH3-H4 histones, and to be required for the assembly of kinetochores in Saccharomyces cerevisiae. Scm3p is conserved across fungi, and displays a remarkable variation in protein size, ranging from approximately 200 amino acids in S. cerevisiae to approximately 1300 amino acids in Neurospora crassa. This is primarily due a variable C-terminal segment that is linked to a conserved N-terminal, CenH3-interacting domain. We have discovered that the extended C-terminal region of Scm3p is strikingly characterized by lineage-specific fusions of single or multiple predicted DNA-binding domains different versions of the MYB and C2H2 zinc finger domains, AT-hooks, and a novel cysteine-rich metal-chelating cluster that are absent from the small versions of Scm3. Instead, S. cerevisiae point centromeres are recognized by components of the CBF3 DNA binding complex, which are conserved amongst close relatives of budding yeast, but are correspondingly absent from more distant fungi that possess regional centromeres. Hence, the C-terminal DNA binding motifs found in large Scm3p proteins may, along with CenH3, serve as a key epigenetic signal by recognizing and accommodating the lineage-specific diversity of centromere DNA in course of evolution.

Figure 1. Diversity of domain architectures of Scm3p across fungal phylogeny

The domain architectures of Scm3p proteins have been superimposed on a maximum-likelihood phylogenetic tree of fungi based on six conserved proteins. The major lineages shown in the figure are supported by 80% or greater relative logarithm likelihood bootstrap support. The proteins are not rendered to scale because of their enormous size difference, though the conserved domains are approximately rendered to size. The actual size of each protein is shown to its left and the small versions are boxed. The notations are: C- Cysteine cluster; the orange box represents positively charged segments. Notably, the N-termini of Scm3p orthologs across much of the fungal tree begin with a conserved block with the signature MxxP (shown as a grey block), but those of most members belonging to Saccharomycotina lack this signature. Instead they possess a distinctive positively charged helical segment indicated with a “+”. There is some uncertainty regarding the predicted gene structure, and hence the predicted protein of Cryptococcus neoformans. The different independently acquired versions of the MYB domain are shaded differently. The Ndc10 column reflects the presence or absence of orthologs of the Ndc10p protein, where +: presence, −: absence and (-): species with a homologous GCR1 domain, either in paralogous transcription factors or crypton family transposons, but no Ndc10 ortholog. All domains noted here were detected with e-values <.01 in profile and HMM searches. They were confirmed through reciprocal searches using PSI-BLAST that recovered known representatives of each of the depicted domains with e <.01 prior to convergence.