The genetics and epigenetics of satellite centromeres

Abstract

Centromeres, the chromosomal loci where spindle fibers attach during cell division to segregate chromosomes, are typically found within satellite arrays in plants and animals. Satellite arrays have been difficult to analyze because they comprise megabases of tandem head-to-tail highly repeated DNA sequences. Much evidence suggests that centromeres are epigenetically defined by the location of nucleosomes containing the centromere-specific histone H3 variant cenH3, independently of the DNA sequences where they are located; however, the reason that cenH3 nucleosomes are generally found on rapidly evolving satellite arrays has remained unclear. Recently, long-read sequencing technology has clarified the structures of satellite arrays and sparked rethinking of how they evolve, and new experiments and analyses have helped bring both understanding and further speculation about the role these highly repeated sequences play in centromere identification.

Figure 1.

Models of amplification of higher-order repeats (HORs). (Left) In the unequal exchange model, reciprocal recombination between out-of-register tandem repeats can either duplicate or delete individual monomers. As variations accumulate in particular monomers, unequal exchange can generate higher-order repeats (HORs). (Right) In the break-induced replication (BIR) model, replication fork stalling can lead to one-ended double-strand breaks (DSBs). Resection yields a free single-strand 3′ end that can invade a homologous sequence and reinitiate replication. Reinitiating at an out-of-register repeated sequence ahead of the fork will lead to deletion, whereas reinitiating at one behind the fork will lead to duplication (insets with blue outlines). Duplication appears to be favored, perhaps because the chromatin behind the fork is more accessible to strand invasion owing to the new acetylated histones and/or the relaxed torsional state in contrast to the overtwisted DNA ahead of the fork (inset with red outline).

Figure 2.

Human centromere 8. (Top) Human centromere 8 shows successive evolutionary layers (1–5), with the oldest monomer layer on the edges of the array and the youngest, most uniform HOR in the middle (Logsdon et al. 2021). (Bottom) Similar structures are seen in great apes, whereas rhesus macaque centromere 8 consists entirely of dimers most closely related to the old monomers.

Figure 3.

Speculative model of replication through α-satellite. Holliday junction recognition protein (HJURP) associates with CENPA nucleosomes before S-phase and recruits the condensin II complex. At the replication fork, HJURP and the MCM2 subunit of the replication machinery work together to assure that CENPA nucleosomes reassemble behind the fork. DNA secondary structures form on single-stranded repetitive DNA behind the fork, and HJURP and mismatch repair proteins (MSH4 and MSH5 are shown) bind to them and resolve them. Condensin II complexes extrude positively supercoiled DNA loops, and the positive torsion inhibits the binding of replication protein A (RPA), which binds single-stranded DNA and must accumulate in order for the ATR serine/threonine kinase (ATR) to signal that DNA damage has occurred and to arrest replication. This inhibition by condensin II allows time for secondary structures to be resolved. Condensin II is also needed with HJURP to assemble new CENPA nucleosomes in G1, and condensin-mediated loops may play a role in the organization of the kinetochore.