CENP-A: the key player behind centromere identity, propagation, and kinetochore assembly

Abstract

Chromosome segregation is the one of the great problems in biology with complexities spanning from biophysics and polymer dynamics to epigenetics. Here, we summarize the current knowledge and highlight gaps in understanding of the mechanisms controlling epigenetic regulation of chromosome segregation.

Fig. 1

Centromeric DNA sequences are not conserved through species. With the exception of S. cerevisiae, most evidence shows that DNA sequence has no role in centromere identity. Species comparison reveals vast differences in nucleotide composition as well as the centromere length. S. cerevisiae has a 125-bp centromere which is divided in three centromere DNA elements (CDE), with CDEII being the sequence important for Cse4 incorporation. It is still somewhat controversial whether this centromere is composed of a single or two to three CENP-A nucleosomes.(Lawrimore et al. ; Furuyama and Biggins 2007) S. pombe 10 kbp centromere consists of inner and outer repeats located outside the core region and in a head-to-head orientation. D. melanogaster has relatively large centromeres made up of DNA repeats and transposon elements for a genomic size of ∼420 kbp. A. thaliana and Homo sapiens centromeres are made of head-to-tail 171–178 bp repeats that can go up to 1.4 Mbp for the plant and 5 Mbp for human. General centromere structures display two general types of centromere: monocentric for S. cerevisiae to H. sapiens, and holocentric (whole length of chromosome) for C. elegans (generally nematodes and several other species) (Maddox et al. ; Melters et al. 2012) Elongated centromeric chromatin may have different arrangements of CENP-A and H3 nucleosomes arrays, which are repetitive and exclusive from one another (Blower et al. 2002). When compacted, the CENP-A arrays form a hypothetical centromeric plate required for kinetochore formation in mitosis. Green circles H3 nucleosome, red circles CENP-A nucleosome

Fig. 2

Compacted CENP-A chromatin promotes kinetochore assembly and centromere propagation. An inactive centromere, which is a chromosomal locus having alpha-satellite sequence without CENP-A protein, demonstrated that centromeric sequence is not sufficient for kinetochore assembly. Kinetochore proteins are recruited to this chromosomal locus only when CENP-A is present (active centromere) (Warburton et al. 1997). Conversely, when CENP-A is overexpressed in cells, it incorporates on chromosome arms. However, this mislocalized CENP-A does not recruit kinetochore proteins in mitosis. Ultimately, by targeting CENP-A on chromosome arms using the lac operon technique, this creates a high-density array of CENP-A protein and forms a neocentromere: it recruits kinetochore components and is able to propagate for a limited time. Those observations suggest that CENP-A density is important for centromere identity and its essential role of kinetochore assembly

Fig. 3

CENP-A incorporation is cell cycle regulated and depends on epigenetic marks. a CENP-A incorporation is a replication-independent mechanism; CENP-A-H4 tetramers are loaded in G1 instead of S-phase as for H3-H4 tetramers. The incorporation is a three-step mechanism: I—Licensing, II—Loading, and III—Maintenance. First, the licensing complex (KNL-2, Mis18α and β) recognizes and binds the centromeric chromatin. This will license the centromere for loading of newly synthesized CENP-A. Next, the licensing complex recruits, by an unknown mechanism, the CENP-A chaperone HJURP which directly binds and stabilizes CENP-A/H4 complexes. Finally, when newly synthesized CENP-A is incorporated, a mark is removed or CENP-A conformation is changed in order to change the newly synthesized identity to that of an old one. Cdc42 or Rac1 might be part of this cellular process through the action of the GAP MgcRacGAP and the GEF ECT-2. CENP-A loading to centromeres is regulated through the cell cycle by the CDK1/2 kinases, which phosphorylate KNL-2 in mitosis and block its localization to centromeres. At anaphase onset, CDK activity diminishes, KNL-2 is dephosphorylated and is able to localize to centromeres. b CENP-A incorporation to centromeres also depends on a post-translational modification of histone H3. H3K4 dimethylation is important to eventually HJURP, which localizes to centromeres and loads newly synthesized CENP-A. H3K9 trimethylation by the methyltransferase Suv39h1 inhibits CENP-A loading to centromeres, whereas H3K9 acetylation by the histone acetyl transferase (HAT) p300 or pCAF triggers CENP-A loading at centromeres. We hypothesize that KNL-2 to binds the linker DNA in the centromeric chromatin and, together with its partners Mis18α and Mis18 β, this protein complex acts as a licensing mark of centromeres for CENP-A loading. Also, a DNA methyltransferase DNMT3A and DNMT3B interacts with Mis18α, and depletion of the protein leads to a decrease of other epigenetic marks such as H3K9me2, H3K9m3, and H3K4me2 at centromeric loci

Fig. 4

What is the first epigenetic mark of centromeres? Information provided by the literature brings the question of what is the first epigenetic mark of centromeres. Some evidence has shown that CENP-A nucleosome structure might be the feature for centromere identity. However, CENP-A incorporation and localization to centromeres are not passive cellular processes; therefore, CENP-A might not be the first epigenetic mark. Centromere identity could be through genetic features of DNA sequence (nucleic acid composition) or motif (CENP-B box, DNA methylation), but strong experimental evidence disputes this model. One marker that could link the two other hypotheses is KNL-2. This protein was shown to be essential for CENP-A localization to centromeres and genetically, it is the most upstream protein for this important cellular process. Thus, KNL-2 could be the first epigenetic mark of centromeres. For KNL-2 predicted structure: Orange is the SANTA domain and in purple is the Myb domain. (Published CENP-A nucleosome structure, Protein Data Bank reference # 3AN2 (Tachiwana et al. 2012); KNL-2 predicted structure (Zhang 2008))