The same but different: reconciling the diversity of centromeres with their highly conserved function

From the Henikoff Lab, Basic Sciences Division

Centromeres are the locations at which protein machinery (the kinetochore) assembles to separate duplicated chromosomes during cell division. Centromeres are said to be both genetic and epigenetic in nature; that is, they are defined by their nucleic acid sequence and by the presence of specific histone protein variants. For nearly two decades, understanding the basis of centromere specification has been a primary focus of the Henikoff Laboratory in the Basic Sciences Division. Dr. Steve Henikoff and his colleagues were recently invited to document their contributions to the field of chromosome segregation in Cold Spring Harbor Symposia on Quantitative Biology. The resulting publication, which appeared last month, spans a large body of work conducted by many Henikoff lab members between 2001 and 2017.

the many types and sizes of centromeres
Figure 1: Centromeres (pink) are remarkably diverse in different organisms. In budding yeast, “point” centromeres consist of three DNA elements, CDE I–II–III. A single cenH3 nucleosome assembles on CDE II. In fission yeast, “regional” centromeres consist of a centromeric region that is usually flanked by pericentric heterochromatin (purple). Most animals and plants have “satellite” centromeres that occupy contiguous regions of chromosomes flanked by pericentric heterochromatin, but are composed of particular families of tandem repeats, and are occupied by arrays of cenH3 nucleosomes. Holocentromeres cover the entire length of the chromosomes. In most lineages, including that of nematodes, cenH3 preferentially incorporates at specific sites, whereas in holocentric insects, cenH3 has been lost. Image provided by Dr. Steve Henikoff

Early studies revealed that centromeric DNA and the centromere-specific histone variant cenH3 both evolve rapidly, leading to dramatic differences in centromeric organization even between closely related species (Figure 1). This realization was puzzling because of centromeres’ absolutely conserved function in recruiting kinetochore components. Dr. Harmit Malik, now a full member of the Basic Sciences Division but at the time a post-doc in the Henikoff lab, reported in 2001 that cenH3 is evolving adaptively. This finding suggested that cenH3 competes with other factors, such as satellite centromeres, which contain many repeats of centromere-like DNA that can selfishly compete for inclusion in the female egg. This “arms race” leads to an imbalance in centromere strength that can be restored by adaptive changes in cenH3.

To better understand the interplay between the genetic and epigenetic components of centromeres, the Henikoff lab investigated how centromere-associated proteins interact with DNA in different species. They began in budding yeast, which have a “point” centromere composed of a single Cse4 (cenH3)-containing nucleosome containing 80 base pairs (bp) of DNA. Because 147 bp is the minimum length of DNA required to wrap around a normal histone complex, the lab suspected that the composition of the yeast point centromere is atypical. Indeed, a series of studies conducted between 2007-2017 definitively established that the budding yeast point centromere contains a “hemisome”, composed of only four histone subunits instead of eight.

Interestingly, cenH3 nucleosomes in fission yeast are completely different from those in budding yeast; they contain the normal eight histone subunits and their positioning is less precise. Fission yeast centromeres are said to be “regional” because they encompass several kilobases of DNA and tens of cenH3 nucleosomes. These stark differences between yeast species highlights the rapid evolutionary divergence of centromere components.

In contrast to having one centromere per chromosome, some organisms like the nematode worm Caenorhabditis elegans have a “holocentromere”, meaning that the entire chromosome serves as a platform for microtubule attachments during separation of replicated DNA. Studies in the Henikoff lab revealed that the C. elegans holocentromere is discontinuous, or “polycentric”; there are approximately 100 locations along each chromosome that are enriched for cenH3-containing histones. Interestingly, many insects lack cenH3 and other centromere-associated proteins entirely. Like C. elegans, these organisms have holocentromeres, but it remains unknown whether they have specific DNA sequences or proteins that define polycentric regions.

Human centromeres are said to have a “satellite” structure that contains highly repetitive DNA sequences spread across hundreds of kilobases. The positioning of the human cenH3, CENP-A, has been a topic of debate because it is very difficult to map sequencing reads to repetitive regions. Using an unbiased, ChIP-based approach, the Henikoff lab determined in 2015 that CENP-A-containing nucleosomes are enriched at two specific dimeric repeats. These two repeat families are known to have been recently “homogenized”, in reference to the phenomenon by which repeats closer to the middle of tandem arrays (“repeats of repeats”) become fixed in sequence. Thus, the positioning of CENP-A is more precise in more homogeneous regions.

The amazing diversity of centromere architectures elucidated by the Henikoff lab and others suggests a role for selfish processes in driving centromere evolution. For example, satellite sequences can expand and compete with each other to be passed on to the next generation. Replacement of repetitive centromeric sequences with other selfish genetic elements has been observed in maize, potatoes and horses, which have all experienced the strongly selective process of domestication. Interestingly, some holocentric plants show no evidence of adaptive evolution of cenH3, suggesting that elimination of a fixed centromere position may provide a means of preventing invasion by selfish DNA elements.

Moving forward, Dr. Henikoff says that his lab will continue grappling with “unanswered questions, such as how to resolve the dual definition of centromeres as being genetic and/or epigenetic, and technical challenges, such as how to surmount the intractability of homogeneous repeats that characterize most centromeres, including our own.”



Henikoff S, Thakur J, Kasinathan S and Talbert PB. Remarkable Evolutionary Plasticity of Centromeric Chromatin. Cold Spring Harbor Symposia on Quantitative Biology. 2017 Dec 1. pii: 033605. doi: 10.1101/sqb.2017.82.033605

This research was supported by the Howard Hughes Medical Institute and the National Institutes of Health