A salt bath and a haircut clean up centromeric chromatin

From the Henikoff Lab, Basic Sciences Division
Figure 1: Schematic of chromatin architecture Image provided by Dr. Steve Henikoff

In the cell, chromosomal DNA is wrapped around protein complexes in a structure often described as “beads on a string” (Figure 1). Each bead—called a nucleosome—contains roughly 140 base-pairs (bp) of DNA. Together, the DNA and its associated proteins are known as chromatin. On each chromosome, one region of chromatin—the centromere—is responsible for recruiting the machinery that faithfully separates replicated chromosomes during cell division. Centromeres contain extremely repetitive DNA sequences known as α-satellite arrays and are also defined by the presence of centromere protein (CENP) family members within their nucleosomes.

The Henikoff Laboratory in the Basic Sciences Division is working to elucidate the structure of centromeric chromatin in greater detail. However, efforts to map the exact positions of CENP proteins across centromeres have been repeatedly stymied. “In part this is due to the lack of assembly maps of repetitive α-satellite arrays on any chromosome, and in part to the difficulty of extracting centromeric chromatin that is embedded in constitutive centromere-associated network (CCAN) protein complexes,” explains Dr. Jitendra Thakur, a post-doctoral fellow in the Henikoff lab.

Researchers generally study chromatin structure by isolating nuclei and treating them with an enzyme called micrococcal nuclease (MNase), which digests all DNA that is not protected by bound proteins, effectively liberating individual nucleosomes. Next, techniques such as chromatin immunoprecipitation (ChIP) can be applied to identify where on the DNA particular proteins, such as the various CENPs, bind.

Chromatin extraction is typically done under low-salt conditions, which works well for actively transcribed chromatin. Drs. Thakur and Henikoff, however, recently observed that most centromeric chromatin is actually not soluble in low salt, suggesting that previous studies may have drawn conclusions from studying only a subset of centromeric chromatin and that there is a difference in the nature of soluble vs. insoluble centromeric chromatin.

In work published this year in Genes & Development, Drs. Thakur and Henikoff investigated the insoluble fraction that comprises the majority of centromeric chromatin. They extracted chromatin in a range of salt concentrations after MNase treatment and confirmed that centromeric (CENP-A-containing) chromatin was strongly enriched in the high-salt fraction. Paired-end sequencing of CENP-A-associated DNA from each fraction revealed that only ~100 bp fragments were present in the low-salt condition, while a mixture of 100-450 bp fragments was present in the high-salt fraction. This result was surprising because it was previously assumed that centromeric nucleosomes are regularly spaced.

To determine why MNase cleavage yielded such diverse fragment sizes in high salt, the authors mapped the positions of CENP-A molecules on various α-satellite arrays under this condition. They observed that the region of DNA protected by CENP-A was shorter in low salt and longer in high salt, and saw increased protection of the CENP-B binding site in high salt. These results suggest that the nucleosomes isolated in low salt have lost CCAN components, such as CENP-B, thus allowing MNase to “nibble” away the ends of the DNA. By contrast, high-salt conditions aid in the extraction of complexes containing more CCAN proteins, which together protect a longer stretch of DNA.

In an effort to minimize artifacts caused by MNase cleavage variability, the researchers turned to a new method developed in the Henikoff lab called CUT&RUN. In this approach, MNase is directed to cleave DNA only at locations where a protein of interest is bound. For example, to map CENP-A, chromatin is treated with a CENP-A-recognizing antibody, followed by treatment with a version of MNase that has an antibody-recognizing domain (Protein A). Using CUT&RUN in combination with salt fractionation (CUT&RUN.Salt), Drs. Thakur and Henikoff saw much more homogenous fragment sizes; most were ~170 or ~340 bp, suggesting that this method isolates intact particles.

With CUT&RUN.Salt, the authors again observed higher occupancy of CENP-B on fragments isolated under high-salt conditions. To determine whether CENP-B is directly responsible for stabilizing these complexes, they asked whether there was a correlation between the strength of the CENP-B binding site and binding of CENP-A/B/C within a given α-satellite. Human chromosomes provide a natural context for answering this question because young (recently duplicated) α-satellites tend to have a high density of optimal CENP-B binding sites, whereas older α-satellites have accumulated random mutations over time that weaken CENP-B binding. CENP-A enrichment indeed correlated with the strength and density of CENP-B boxes within a given α-satellite, suggesting that the youngest arrays might have a better ability to attract CENPs and form centromeres.

Finally, Drs. Thakur and Henikoff exploited the slight sequence variation within an α-satellite array to map CENP-A/B/C across a large section of the array, rather than averaging across the repeat units. Contrary to the expectation that CENP occupancy would be similar among the highly similar repeat units, the authors were surprised to observe a very irregular pattern (Figure 2). In particular, CENP-A/B/C were seen to bind in different amounts and with different orientations relative to the CENP-B binding site on adjacent repeats. Together, these results suggest that subtle sequence differences between α-satellites can affect their ability to recruit CCAN components.

Figure 2: Structural and conformational variations of the CCAN complex on a highly homogenous α-satellite array Image provided by Dr. Jitendra Thakur

Because α-satellites contain such repetitive sequences, researchers have long wondered how a given α-satellite array out-competes all of the others to recruit centromere components. The conceptual advances resulting from the Henikoff lab’s efforts represent an important step forward in answering this question. As stated by Dr. Thakur, “the discovery of structural and conformational variations of CENP-A/B/C complexes on adjacent α-satellite units that differ only slightly reveals that variations in centromeric chromatin are not only present between species but also within a single array on a single centromere. These variations provide a great opportunity for evolutionary forces select the variants that are most efficient in building stronger centromeres.” In the future, the Henikoff lab plans to examine how variations in CENP-A/B/C complexes affect their ability to recruit the machinery required for centromere formation.


Thakur J and Henikoff S. (2018) Unexpected conformational variations of the human centromeric chromatin complex. Genes & Development. 32(1):20-25.

This research was supported by the Howard Hughes Medical Institute