|Home | About | Journals | Submit | Contact Us | Français|
Three recent papers, including Mizuguchi et al. (2007) in this issue, show that the nonhistone protein Scm3 is required for the recruitment of the histone H3 variant Cse4 to centromeres in budding yeast. Scm3 forms a chromatin component with Cse4:histone H4 tetramers that appear to lack H2A/H2B histones. These studies provide key insights into the pathway that recruits Cse4 to centromeres and have important implications for other functions of chromatin.
Centromeres are special chromosome loci where kinetochores are assembled during mitosis and meiosis. All eukaryotes contain a centromere-specific histone H3 variant (CENP-A) that replaces canonical histone H3 at centromeric nucleosomes, forming a structural foundation for the kinetochore. CENP-A:H4 tetramers are structurally different from H3:H4 tetramers with respect to the accessibility of the interface with H4 (Black et al., 2004). Most centromere and kinetochore proteins require CENP-A for proper localization and function, suggesting that CENP-A is at or close to the top of the centromere/kinetochore assembly pathway (Mellone and Allshire, 2003). The key issue currently faced by “centromerophiles” is to identify the molecular components and mechanisms responsible for CENP-A deposition and centromere propagation. Recent reports, including a paper in this issue of Cell, now show that the nonhistone protein Scm3 in budding yeast is critical to the recruitment of Cse4 (the CENP-A homolog) to centromeres (Mizuguchi et al., 2007; Camahort et al., 2007; Stoler et al., 2007).
In organisms such as humans and flies, primary DNA sequences are neither necessary nor sufficient to form a functional centromere, and centromere identity and propagation appear to be determined epigenetically (Karpen and Allshire, 1997). These large “regional” centromeres contain thousands of CENP-A nucleosomes, which are thought to be segregated randomly to replicated sister chromatids. Thus, pre-existing CENP-A nucleosomes could act as an epigenetic mark for subsequent replenishment by an as-yet-undetermined replication-independent chromatin assembly mechanism. Surprisingly, recent studies have demonstrated that CENP-A recruitment occurs in late mitosis/G1 phase of the cell cycle (Jansen et al., 2007).
In contrast, the budding yeast Saccharomyces cerevisiae contains “point” centromeres thought to have only a single Cse4-containing nucleosome; random segregation to only one sister chromatid would require de novo Cse4 nucleosome assembly on newly replicated DNA. Centromere assembly in S. cerevisiae requires binding of a conserved DNA element (CDE III) by Ndc10, a component of the CBF3 complex. Ndc10 is required for Cse4 recruitment (McAinsh et al., 2003), but the mechanism has been unclear as there does not appear to be a physical interaction between Ndc10 and Cse4.
Resolution of this issue and new insights into centromere regulation are provided by three new studies that analyze the Scm3 protein (Mizuguchi et al., 2007; Camahort et al., 2007, Stoler et al., 2007). Scm3 was previously identified as a high-copy suppressor of mutations in the Cse4 histone fold (Chen et al., 2000), the details of which are provided by Stoler et al. This observation stimulated the Camahort et al. (2007) study. The analysis by Mizuguchi et al. (2007) was initiated by the observation that Scm3 was a chromatin-associated protein that immunoprecipitated with Cse4 and not histone H3. Both Camahort et al. (2007) and Mizuguchi et al. (2007) used immunofluorescence and chromatin immunoprecipitation to show that Scm3 is a new component of the centromere. Scm3 colocalizes with Cse4 but also displays independent localization by immunofluorescence to other undefined parts of the nucleus.
Most surprisingly, chromatin immunoprecipitation by Mizuguchi et al. (2007) revealed that centromeric chromatin containing Cse4 and Scm3 lacks histone H2A and H2B in vivo. They also demonstrate that Scm3 forms a stoichiometric complex with Cse4:H4 tetramers in vitro in the absence of DNA. In fact, Scm3 was able to replace H2A and H2B when mixed with preformed octamers containing H2A, H2B, Cse4, and H4. Yeast two-hybrid analysis presented by Stoler et al. (2007) shows that Scm3 is able to interact with itself. Together, these results support a model in which Scm3, Cse4, and histone H4 form nucleosome-like structures in vivo—most likely a hexamer containing two copies of each protein (Figure 1A).
What is Scm3’s role in centromere assembly? Camahort et al. (2007) demonstrate physical interactions in vivo between Scm3 and Ndc10, suggesting that Scm3 might be the “missing link” between the CBF3 complex and Cse4. Both Mizuguchi et al. (2007) and Camahort et al. (2007) used mutants and chromatin immunoprecipitation to determine the epistasis relationships between Scm3, Cse4, and other essential centromere components. Cse4 and Scm3 are dependent on each other for localization to the centromere, as are Ndc10 and Scm3. The levels of centromeric Cse4 were drastically reduced in cells lacking Scm3 or Ndc10, but Scm3 and Ndc10 were only slightly diminished in cse4 yeast mutants. Scm3 deletion also resulted in decreased binding of other kinetochore proteins such as Cbf1, Mif2, and Cep3. These results position Ndc10 and Scm3 upstream of Cse4 in the centromere assembly pathway and demonstrate that Scm3 is required for normal centromere and kinetochore formation.
Cell cycle and morphological analysis further demonstrated that Scm3 is also required for kinetochore function (Mizuguchi et al., 2007; Camahort et al., 2007). Elimination of Scm3 either by degron-mediated removal or transcriptional repression resulted in cell cycle arrest in G2/M as well as aberrant DNA content, consistent with defects in kinetochore function and chromosome segregation. The G2/M arrest suggests activation of the mitotic or spindle checkpoint due to defective kinetochore-microtubule interactions in the absence of Scm3. However, Camahort et al. (2007) show that Scm3 is essential in early S phase for centromere localization of Ndc10, whose presence is directly required for checkpoint activation during mitosis. When cells were arrested in G1 by α factor followed by Scm3 inactivation, the checkpoint was not activated, and Ndc10 localization to centromeres was reduced. In contrast, if centromeres were allowed to replicate before Scm3 inactivation, cells arrested in G2/M and contained normal levels of centromeric Ndc10. Therefore, it appears that Scm3 is required for Ndc10 loading during or following centromere replication and that Scm3 is not directly required for checkpoint activation. Mizuguchi et al. (2007) also observed a G1 arrest after Scm3 degradation, which is presumably related to the noncentromeric localization of Scm3 or the different experimental approaches used to abolish Scm3 activity.
The biochemical, genetic, and molecular analyses of Scm3 suggest a new model for how centromeres are assembled in budding yeast (Figure 1B). Scm3 forms a complex with Cse4: H4 tetramers in centromeric chromatin, suggesting the presence of a nucleosome that lacks H2A and H2B. Scm3 may be involved in delivery of newly synthesized Cse4 to centromeres, incorporation of Cse4 into chromatin, and/or maintenance of Cse4 at centromeres. It is also unclear whether Cse4:H4 tetramers are initially associated with H2A and H2B tetramers at some point in the cell cycle and then are replaced by Scm3. Regardless, as Scm3 and Cse4 are reciprocally required for centromere localization, two nucleosome “variations” (Scm3 and Cse4) need to be present for the centromere to function. In addition, the requirement for Scm3 in early S phase for centromere localization of Ndc10 suggests that it is necessary for sequence-specific binding of CBF3 to the CDEIII element during centromere formation. Thus, point centromere assembly likely requires coordination and cooperativity between sequence-specific binding through CBF3/Ndc10, association of Scm3 with Cse4:H4 and CDEII, and Ndc10-Scm3 binding.
The relevance of these exciting discoveries to primary-sequence independent regional centromeres found in most eukaryotes is unknown. The Scm3-CBF3 pathway may only be required in organisms with a single CENP-A nucleosome in order to ensure propagation of the centromere to replicated chromatids. All three groups point out that Scm3 homologs cannot be identified in eukaryotes outside of fungi. In addition, previous studies have shown that Drosophila and human centromeric nucleosomes contain H2A and H2B (Blower et al., 2002; Foltz et al., 2006). However, Mizuguchi et al. (2007) suggest that a divergent Scm3 ortholog could be present in a small subset of CENP-A:H4 nucleosomes in these organisms. It is interesting that fungi such as Schizosaccharomyces pombe, Candida albicans, and Neurospora crassa, which contain small regional centromeres, do have Scm3 homologs. However, in budding yeast Scm3 is also required for G1 progression and is not exclusively centromeric, suggesting that it may regulate functions other than centromeres. Thus, it will be important to determine if the Scm3 homolog physically interacts with, for example, S. pombe CENP-A (Cnp1) and if it is required for Cnp1 centromere localization.
Whether or not these findings point the way toward a better understanding of assembly of CENP-A into regional centromeres, the three new studies provide a major advance in our understanding of point centromere assembly and composition. In addition, the unprecedented indication of chromatin containing this unique nucleosome composition has important ramifications for other chromosome functions. The assumption that functions such as DNA repair, replication, transcription, and silencing are solely dependent on the presence of histone octamers needs to be reexamined in light of this work.