Nucleosomes and centromeric DNA packaging

The eukaryotic chromosome is a conserved structure, with the DNA double-helix wrapping around octamers of histone proteins to form the chromatin, which is further packaged into chromosomes. The centromere defines the kinetochore, the region of spindle microtubule attachment that pulls the two replicated chromatids of each chromosome apart during cell division (Fig. 1), leading to fidelity in transmission of genetic information. Like the telomere, centromeres are well defined morphologically and functionally, but their DNA sequence shows no conservation between species, and their coiling into chromatin is still poorly understood. In PNAS, Zhang et al. (1) give unique insight into the centromeric CenH3-nucleosomes associated with the rice CentO satellite sequence and adjacent regions of DNA, having implications for centromere specification, activity, and evolution.

Fig. 1.

Features of centromeric DNA from different viewpoints. (A) An electron micrograph of a section of a metaphase chromosome of a wild wheat species, showing two arms with the centromere at the bend. (B) Metaphase chromosomes of triticale fluorescing blue in the light microscope. Constrictions at the centromeres are visible on each chromosome, with the two chromatids that will separate as the cell divides. (C) Chromosomes from a cell culture line of the model species Arabidopsis thaliana, labeled with a centromeric histone antibody. (D) A diagram of a metaphase chromosome showing the two arms each of two chromatids, separated at the centromere (E) and dividing into chromatids which segregate and are pulled by spindle microtubules (red) attached via the kinetochore at the centromere. (F) DNA motifs found in many centromeres, with blocks of tandemly repeated satellite DNA monomers interspersed with single copy DNA and transposable elements. (G) A diagram of the packaging of double stranded DNA (blue) into nucleosomes, with 147 bp of DNA wrapping 1.67 times around each octamer of the canonical histone proteins (olive) and fixed phase of the nucleosome within the repeat monomer. (H) The unique packaging reported by Zhang et al. (1) with ∼100 bp of the rice CentO tandem repeat sequence (red) folding once around the nucleosome core that includes CenH3 (yellow). (I) A key method for nucleosome analysis involving micrococcal nuclease digestion of chromatin and size separation of the resultant DNA fragments; the enzyme cuts DNA in the linker regions and, over the time course shown, isolates more mononucleosomes, and trims overhanging DNA not protected from digestion by the histone proteins (12). (Scale bars for A–C, 2 µm.)

Centromeres are readily observed as partial constrictions on the metaphase chromosome by light or electron microscopy (Fig. 1). However, few sequence-related characteristics of the DNA at the centromeres are conserved. In contrast, nucleosome proteins (2) and kinetochores linking centromeres and spindle microtubules with a multiprotein complex are highly conserved across all kingdoms (3).

Typically, there are tandemly repeated satellite DNA sequences at the centromeres of chromosomes of animals and plants (4–6). Detailed work on the budding yeast Schizosaccharomyces pombe and brewer’s yeast Saccharomyces cerevisiae identified relatively short DNA sequences and binding protein counterparts that direct chromosome segregation (7, 8). The search was then on for critical “boxes” in plants and animals but, although some motifs were found to be involved in protein–DNA interactions, no motif shared across diverse phyla was found within the centromeric tandem repeats.

As sequencing technology advanced, both variants of tandem repeats and almost any other class of DNA—retrotransposons, transposons, genes, transcription factors, and microsatellites—were revealed (9) underlying the kinetochore and microtubule attachment sites. Evidence from comparisons of species, mutants that still showed centromeric function, the lack of necessity for tandemly repeated satellite sequence, and the presence of facultative neocentromeres forming at unusual points along chromosomes started to suggest there was no required sequence motif for centromere activity.

From the 1980s, the nucleosomal packaging of DNA was being investigated, and in 1991, a protein called centromere protein A (CENP-A) was found exclusively associated with active centromeres in human cells (10). The protein was characterized as a histone H3 variant, with two monomers replacing those of the non–division-associated or canonical form of H3 present in the histone octamers found in interphase nuclei. It is now clear that presence of the CenH3 histone [terminology here follows (1), CenH3 rather than CENP-A] is specifying centromere location independently of DNA sequence, and centromeric function is an epigenetic character of the DNA sequence, where under some conditions a particular DNA sequence associates with the histone and becomes the centromere. Thus, centromere function joins many other aspects of epigenetic control of the nucleus where modification of histone proteins is critical to the DNA behavior (11).

Many centromeric satellite DNA sequences have a repeat motif length of 150–180 bp, the same as the DNA length forming the 1.67 turns around the histone octamer core (147 bp) plus the linker DNA between the nucleosomes, a variable distance that is typically 10–70 bp (Fig. 1 F and G). When chromatin is extracted from cells with gentle methods, the DNA remains associated with the histones. The enzyme micrococcal nuclease (MNase) cuts free double-stranded DNA, except when it is protected by association with the nucleosomal proteins. Digestions and sequence analysis show how tandemly repeated DNA is packaged around nucleosomes (Fig. 1G). For one family of satellites, there is a strictly defined association (phasing) of DNA and nucleosomes with the same sequences in all linkers (4, 6, 9, 12). By increasing time of digestion with MNase, DNA adjacent to the nucleosome is trimmed, leaving only the core particle (Fig. 1I). The CenH3-nucleosomes occur in multiple blocks intermingled with nucleosomes with the canonical H3 (9). Some structural models suggest this is due to the positioning of the CenH3 nucleosomes at the periphery of the chromatid, interacting with kinetochore proteins, while canonical nucleosomes are folded to the inside.

Zhang et al. (1) analyze the CenH3-nucleosome structures associated with the rice centromeric CentO satellite sequence and adjacent regions of DNA in unprecedented detail. Rice has proved valuable for understanding centromere structure, where the assembly of the DNA sequence across several centromeres, with exceptional accuracy, has shown the presence of a 155-bp-long tandemly repeated sequence, CentO, with array lengths varying from 65 to 2,000 kb on the different chromosomes, interspersed with transposable element and single-copy sequences (9). Now, the authors digested chromatin to single nucleosomes using the MNase method. They then precipitated the nucleosomes containing CenH3 by ChIP (immunoprecipitation with a CenH3 antibody), and sequenced the DNA that was associated, thus revealing the length and sequence of the CenH3-nucleosome DNA. Remarkably, only 90–100 bp of the CentO was protected from digestion on the centromeric nucleosomes, representing only one turn of DNA around the octamer (Fig. 1H), compared with the 147 bp typically protected on canonical nucleosomes. The periodicity of the CenH3 nucleosomes was exactly the same as the 155 bp length of the satellite and therefore was in phase with the sequence. A final feature of the analysis was a 10-bp periodicity of CentO, one turn of the DNA double helix A 10bp rotational phasing minimizes the bending energy or strain on the folded DNA and increases stability. The unexpected 100-bp result in rice has now also been found in human cells (13) where the CENP-A (CenH3) nucleosome protects only 110 bp of the 171-bp-long α-satellite from MNase digestion. Could there be an alternative packaging structure involving CENH3? Zhang et al. (1) cannot rule out a tetrameric arrangement of histone proteins in CENH3 nucleosomes, reported by Dimitriadis et al. (14), although Hasson et al. (13) conclude that the phased satellite at human centromeres wraps around octameric nucleosomes, with loose and extended spacers.

In rice, Zhang et al. (1) show how ChIP sequencing on a relatively large scale has revealed the phasing and nature of spacing of CentO- and nonsatellite CenH3-nucleosomes. This has made a significant advance in our understanding of the DNA:histone variant interaction and the nature of centromeric chromatin packaging, crucial to chromosome structure and the specification of centromeres, assigning a functional role to the abundant but enigmatic class of centromeric satellite DNA. It will now be important to integrate these results with a dynamic picture of CenH3 loading (15). New, super-resolution microscopy methods (16), careful assembly of satellite DNA arrays (9, 17), and comparative studies of different organisms will help with answering the long-standing questions of what defines a centromere. How and when is the DNA underlying the centromeric nucleosomes defined in the cell? Why and how does this DNA become the target to associate with the centromeric histone variant? These large questions must be partitioned to tractable experiments, and it needs to be considered whether imaging, mutant, sequencing/informatics, or biochemical approaches are going to be most informative. Exquisite control of centromere function, and the highest stability of the process, is essential for correct chromosome segregation. This requires robust and redundant controls, including evolutionary selection for satellite and other DNA sequences with their interacting counterparts. Application of knowledge of centromere specification and control has potential not only for understanding aneuploidy-related diseases, but also may be exploited in generation of new hybrid plant varieties.