Structural and functional dissection of Mif2p, a conserved DNA-binding kinetochore protein

Abstract

Mif2p is the budding-yeast orthologue of the mammalian centromere-binding protein CENP-C. We have mapped domains of Saccharomyces cerevisiae Mif2p and studied the phenotyptic consequences of their deletion. Using chromatin immunoprecipitation (ChIP) and electrophoretic mobility shift assays, we have further shown that Mif2p binds in the CDEIII region of the budding-yeast centromere, probably in close spatial association with Ndc10p. Moreover, ChIP experiments show that Mif2p recruits to yeast kinetochores a substantial subset of inner and outer kinetochore proteins, but not the Ndc80 or Spc105 complexes. We have determined the crystal structure of the C-terminal, dimerization domain of Mif2p. It has a "cupin" fold, extremely similar both in polypeptide chain conformation and in dimer geometry to the dimerization domain of a bacterial transcription factor. The Mif2p dimer seems to be part of an enhanceosome-like structure that nucleates kinetochore assembly in budding yeast.

Figure 1.

Domain organization of the Mif2p and CENP-C polypeptide chains. Top, the Mif2p polypeptide chain. Yellow, green, red, and blue blocks indicate regions of distinct function (assigned, in part, through results described herein). The two regions recognizably conserved in human CENP-C (middle) are a short “CENP-C signature sequence” (light green) and a C-terminal domain (blue) shown here to be a dimerizing element. The pink box in the human CENP-C diagram represents the presumptive DNA-binding domain. Various signature-motif sequences are shown in the bottom panel.

Figure 2.

Phenotypes of Mif2p domain deletants. (A) amino-acid residue boundaries of domains. (B) Growth at 30°C of yeast expressing various MIF2 deletions as sole copy. Color code corresponds to A. (C) Temperature sensitivity of deletion mutants lacking the N-terminal, PEST, and dimerization domains, respectively. (D) Temperature sensitivity of various point mutations (see text for details). (E) Effects of deleting parts or all of the DNA-binding region (red block in A). The deletion Δ366-436 removes the region between the AT-hook (residues 346–365) and the dimerization domain. For details of each deletion, see Supplemental Table S1.

Figure 3.

Structure of the Mif2p dimerization domain. (A) Two views of the domain in ribbon representation. The two β-sheets of the cupin domain are in red (ABIDG) and blue (CHEF), respectively. The red sheet is at the dimer interface. On the right is a schematic representation of the strand connectivity in a cupin fold. (B) Amino acid sequence of the Mif2p dimerization domain (residues 438–530) from S. cerevisiae (top line), with secondary-structure elements indicated by arrows. For comparison, sequences of the homologues from Kluyveromyces walti, Schizosaccharomyces pombe, Stongylocentrotus purpuratus, and Homo sapiens (CENP-C) are aligned over the relevant regions. (C) Ribbon representation of the domain dimer (left), compared with the C-terminal domain of Hth-3 from V. cholerae (right).

Figure 4.

Chromatin immunoprecipitation of CENIV DNA with Cbf1p-GFP, Cep3p, Ndc10p-myc12, and Mif2p (A) and H3 and Cse4p (B). DNA in immunoprecipitated fractions was amplified by PCR with contiguous, nonoverlapping primers spanning CENIV as shown at the bottom of the figures. Bar graphs show the percentage of IP of each PCR fragment relative to total input DNA.

Figure 5.

Binding of Mif2p to S. cerevisiae CENIV. Recombinant proteins were incubated with radiolabeled DNA fragments, and complexes were resolved on nondenaturing bandshift gels (see Materials and Methods). (A) Binding of Mif2p to CENIV. Recombinant Mif2p(256-549) (lanes A–E), Mif2p(365-549) (lane G), or CBF3 complex (lane F), was mixed with ³²P-radiolabeled 88-base pair wild-type CDEIII alone or in addition to unlabeled wild-type 88bp CDEIII (lane D) or 88-bp random plasmid DNA (lane E). Recombinant Mif2p(256-549) (lane H) or CBF3 complex (lane I) was mixed with ³²P-radiolabeled 88-base pairs random plasmid DNA. (B and C) DNA requirements for Mif2p binding. Recombinant Mif2p(256-549) was incubated with ³²P-radiolabeled DNA probes of indicated sequence and binding evaluated as described in A. The boxed CCG is the conserved binding triplet for one of the Cep3p Zn clusters (cf. Figure 6; McGrew et al., 1986, Ng and Carbon, 1987; Espelin et al., 1997; Bellizzi et al., 2007). (D) Recombinant Mif2p(256-549) was incubated with ³²P-radiolabeled 88-bp CDEIII DNA alone, in the presence of increasing amounts of unlabeled 88-bp wild-type CDEIII DNA, or increasing amounts of sonicated salmon sperm DNA. Percentage of binding in the presence of unlabeled WT CDEIII DNA or salmon sperm DNA is shown relative to binding to radiolabeled CDEIII probe alone. (E) Binding of Mif2p to CENIV (88bp CDEIII) in the presence of CBF3. Recombinant Mif2p(365-549) (left lanes) or Mif2p(256-549) (right lanes) was incubated with ³²P-radiolabeled 88-bp CDEIII DNA in the presence of CBF3. The two CBF3-shifted bands, CDEIII:CBF3(1) and CDEIII:CBF3(2) contain one and two Ndc10p dimers, respectively. Mif2p(365-549) does not bind DNA, either on its own or with CBF3 bound; Mif2p(256-549) binds both on its own (to excess free DNA, lower shifted band) or together with CBF3 (supershift of CDEIII:CBF3(1) band).

Figure 6.

Diagram summarizing binding data in this article, in previous work on CBF3–centromere interactions (Espelin et al., 1997), and in the literature. CBF3 comprises a Skp1p:Ctf13p heterodimer associated closely with a Cep3p dimer; addition of one or two dimers of Ndc10p generates “core” and “extended” complexes, respectively. The Cse4p nucleosome contains Scm3p, Cse4p, and H4 in 1:1:1 proportion, but it seems to lack H2a and H2b (Mizuguchi et al., 2007). An interaction of Scm3p and Ndc10p has been proposed to be the mechanism by which CBF3 recruits the Cse4p nucleosome (Camahort et al., 2007). We discuss in the text whether adding a second Ndc10p dimer, to form an extended complex, is compatible with recruitment of Mif2p; the possibility, that the two are mutually exclusive is suggested by a white hatching across the second Ndc10p dimer. The guide at the bottom of the figure relates the spatial organization diagrammed in the upper part to the sequence and organization of centromeric DNA in yeast.