To really understand how a biological structure works, we need to know not just which, but also how many, molecules are present in it. For kinetochores—the structures that link centromeric DNA to spindle microtubules—protein components have been tallied by comparing the fluorescence of GFP-tagged kinetochore proteins to fluorescent versions of the centromeric protein CENP-A, two copies of which are believed to assemble at every budding yeast centromere. Two studies now report that CENP-A is actually much more abundant at yeast centromeres, meaning that kinetochore protein numbers have previously been underestimated (1, 2).
CENP-A—known as Cse4 in budding yeast—is a variant of histone H3 that assembles into specialized nucleosomes that mark the position of centromeric chromatin. S. cerevisiae chromosomes have very small “point” centromeres that are thought to incorporate a single centromeric nucleosome containing two copies of Cse4 (3). This assumption has been used to estimate the numbers of kinetochore proteins that assemble on point centromeres and attach each budding yeast chromosome to a single spindle microtubule (4). Although fission yeast and vertebrates have much larger “regional” centromeres that attach to multiple microtubules, the Cse4 standard has been used to quantify kinetochore proteins in these cell types as well (5, 6).
“We sat down and asked: What's the data underlying the dogma?”
Jian-Qiu Wu, Valerie Coffman, and their colleagues, from The Ohio State University in Columbus, wanted to count the number of proteins in cytokinesis nodes, precursors of the cytokinetic contractile ring. But when they compared fluorescent node proteins to Cse4-GFP, their results didn't make sense. “For some proteins, there was too few, or even less than one molecule, in each node!” Wu explains. At the same time, Kerry Bloom and Ted Salmon, whose labs at the University of North Carolina, Chapel Hill had developed Cse4 as a standard for counting kinetochore proteins, became aware that several research groups were having similar problems.
Both sets of researchers decided to directly count the number of Cse4 molecules at centromeres using several alternative standards, ranging from individual GFP molecules in vitro (1) to an E. coli motor complex that contains 22 copies of a GFP-tagged protein (1, 2). Using these new references, the researchers found that there was more Cse4 in the vicinity of each budding yeast centromere than previously thought—5 or 6 copies according to Bloom, Salmon, and their colleague Josh Lawrimore, or even 8 molecules according to Coffman et al.
“We sat down and asked: What's the data underlying the dogma that there's one CENP-A nucleosome [at budding yeast centromeres]?” Bloom recalls. Chromatin immunoprecipitation experiments had suggested that a single CENP-A nucleosome occupies budding yeast point centromeres (3). But computer simulations showed that this approach would miss additional CENP-A nucleosomes located at random chromatin positions on either side of the centromere (1).
Meanwhile, Coffman et al. found that fission yeast have around 40 times more CENP-A at their centromeres than expected (2); previous estimates had suggested that, although S. pombe centromeres are much larger than budding yeast point centromeres, they only incorporate 2–3 CENP-A nucleosomes to nucleate attachments to 2 or 3 spindle microtubules (5). “But there's actually many more CENP-A nucleosomes in fission yeast,” says Coffman, “which makes them more similar to human centromeres.”
Budding and fission yeast kinetochore protein quantities also increased when re-counted using the new standards, which helps explain how kinetochores hold on to dynamic spindle microtubule ends. “[The kinetochore] needs many transient attachment sites so that it rarely lets go,” Salmon explains. Previous estimates had suggested that kinetochore microtubule-binding proteins such as Ndc80 and S. pombe Dam1 were too few in number to maintain robust dynamic attachments to a growing or shrinking microtubule, but the new figures indicate these proteins are present in sufficient quantities.
“We've learned a great deal about kinetochores in recent years,” Salmon continues. “Now we have to build more detailed mechanistic models of protein function.”