key: cord-1011602-byxd4o4m
authors: Doheny, Kimberly Floy; Sorger, Peter K.; Hyman, Anthony A.; Tugendreich, Stuart; Spencer, Forrest; Hieter, Philip
title: Identification of essential components of the S. cerevisiae kinetochore
date: 1993-05-21
journal: Cell
DOI: 10.1016/0092-8674(93)90255-o
sha: 36d81dfe6a86cf8bfe344eeb9fdf0dc32c40894c
doc_id: 1011602
cord_uid: byxd4o4m

Abstract We have designed and utilized two in vivo assays of kinetochore integrity in S. cerevisiae. One assay detects relaxation of a transcription block formed at centromeres; the other detects an increase in the mitotic stability of a dicentric test chromosome. ctf13-30 and ctf14-42 were identified as putative kinetochore mutants by both assays. CTF14 is identical to NDC10 CBF2 , a recently identified essential gene that encodes a 110 kd kinetochore component. CTF13 is an essential gene that encodes a predicted 478 amino acid protein with no homology to known proteins. ctf13 mutants missegregate chromosomes at permissive temperature and transiently arrest at nonpermissive temperature as large-budded cells with a G2 DNA content and a short spindle. Antibodies recognizing epitope-tagged CTF13 protein decrease the electrophoretic mobility of a CEN DNA-protein complex formed in vitro. Together, the genetic and biochemical data indicate that CTF13 is an essential kinetochore protein.

The term chromosome cycle describes a fundamental aspect of the cell division cycle in which each of the chromosomal DNA molecules first is replicated and then undergoes a series of morphological changes and complex movements to ensure its faithful distribution at mitosis. The gene products responsible for execution of the chromosome cycle include structural components, such as those that assemble into the kinetochore, and regulatory components, such as those that establish checkpoints monitoring the proper completion of ordered events within the cell cycle.

Saccharomyces cerevisiae offers two major advantages as an experimental organism in which to study the chromosome cycle. First, it is possible to combine classical §Present address: European Molecular Biology Laboratory, Meyerhoff Strasse 1, Heidelberg 6900-DE, Federal Republic of Germany. lIPresent address: Center for Medical Genetics, Department of Medicine, Johns Hopkins School of Medicine, Baltimore, Maryland 21205. genetics (isolation and phenotypic analysis of mutants) with recombinant genetics (manipulation of cloned DNA segments by recombinant DNA methods and subsequent reintroduction into the yeast host). Second, all of the cisacting DNA sequence elements required for chromosome maintenance are cloned and well characterized, including functional centromere DNA, chromosomal origins of DNA replication, and telomere DNA (reviewed by Newlon, 1988) .

One structure clearly essential to chromosome distribution is the kinetochore (centromere DNA and associated proteins), providing the site of attachment of spindle microtubules. The kinetochore is a relatively simple structure in S. cerevisiae in comparison with the large and complex trilaminar structures seen in higher eukaryotes (Rieder, 1982; Pluta et al., 1990) . In electron microscopic studies of S. cerevisiae chromosomes, one microtubule is seen to interact with each chromatin molecule, but astructurally differentiated kinetochore is not visible (Peterson and Ris, 1976) . The kinetochore of S. cerevisiae is composed of an approximately 160-220 bp nuclease-resistant core that is centered around the centromere (CfN DNA) sequence and flanked by an ordered array of nucleosomes (Bloom et al., 1964; Funk et al., 1989) .

The CEN DNA sequence requirements for S. cerevisiae have been rigorously and extensively characterized through mutational analysis (reviewed by Carbon and Clarke, 1990) . Approximately 125 bp is sufficient for centromere function (Cottarel et al., 1989) . Comparison of centromeres from different chromosomes reveals that they consist of three centromere DNA sequence elements (CDEI, CDEII, and CDEIII) (Fitzgerald-Hayes et al., 1962; Hieter et al., 1985) . CDEI (8 bp) and CDEIII (25 bp) exhibit dyad symmetry and are separated by CDEII, a 76-86 bp sequence of over 90% AT content. Deletions of CDEI and CDEll reveal that they are important but not essential for chromosome segregation, while single-nucleotide point mutations in CDEIII can completely destroy centromere function.

Although a great deal is known about CEN DNA sequence determinants in S. cerevisiae, little is known about the proteins required for kinetochore activity or its regulation within the cell cycle. Biochemical purification of kinetochore proteins through sequence-specific affinity purification with CEN DNA (Ng and Carbon, 1967; Lechner and Carbon, 1991) has proven to be difficult, apparently owing to the low abundance of the kinetochore proteins and the requirement of accessory factors for binding in vitro. To date, only one CEN DNA-binding protein, CPFl (also known as CPl or CBFl), has been extensively characterized (Baker and Masison, 1990; Mellor et al., 1990; Cai and Davis, 1990) . CPFl is a member of the helix-loophelix family of DNA-binding proteins and binds as a homodimer to CDEI. A null mutation in cpfl results in only a IO-fold decrease in chromosome segregation, indicating that it is important but not essential for kinetochore function. Lechner and Carbon (1991) have described the isola- The amino-terminal actin ORF is represented by the stippled boxes, separated by a line representing the actin intron. The in-frame LacZ coding sequence is shown as a hatched box. CEN DNA is indicated by open boxes, with roman numerals I, II, and III indicating CDEI, CDEII, and CDEIII, respectively.

Transcription initiation from GAL70 is indicated by the solid arrow, and the length and number of transcripts by the length and width of the dashed arrow. (A) Control experiments (CEN DNA mutation in cis). CEN DNA transcriptional blocks (wild-type and CDEIII-1% mutant) were tested in a wild-type strain. Q-galactosidase activity levels were normalized to 100% with a strain containing the reporter with no CEN transcriptional block (top). (6) Proposed relaxation of the wild-type CEN transcriptional block due to mutation of a kinetochore protein component. Perier and Carbon (1992) recently described a reporter with CEN DNA within the GAL7 promoter. This situation presumably sets up a competition for binding between kinetochore proteins and transcription initiation factors. The reporter used here is different in that it assays for the relief of a transcriptional block caused by a CEN placed downstream of the GAL10 promoter.

tion of a multicomponent protein-CEN DNA complex, CBF3, which is defined as an in vitro activity that can bind CEN DNA sequences in a CDEIII sequence-specific manner. Three major protein species of 110 kd, 84 kd, and 58 kd apparent molecular weight are present in approximately equimolaramounts in the most highly purified preparations, although many substoichiometric species are also present (Lechner and Carbon, 1991) . The CBF3 preparation has recently been shown to exhibit a minus end mechanochemical motor activity in vitro, observed as translocation of a latex bead covalently attached to CEN DNA along polymerized microtubules (Hyman et al., 1992) . Classical genetic approaches have also been undertaken to identify S. cerevisiae genes required for chromosome transmission, some of which are expected to encode kinetochore components.

Several mutant collections have been isolated with the primary criterion of chromosome missegregation, including the ctf(chromosome transmis-sion fidelity; Spencer et al., 1990) , ch/(chromosome loss; Kouprina et al., 1993), tin (chromosome instability; Hoyt et al., 1990) , mcm (minichromosome maintenance; Maine et al., 1984) , and M/F (mitotic fidelity; Meek+Wagner et al., 1988) mutants. These mutants could have defects in any of the many components necessary for the chromosome cycle to proceed with high fidelity. Secondary criteria can be applied to identify those mutants defective in a particular structure or process. For example, sensitivity to Benomyl (a microtubule-destabilizing drug) was used as a secondary screen for the tin collection to identify mutants involved in microtubule function. This recently resulted in the identification of ClN8 and KlPl (CIN9) (Hoyt et al., 1992; Saunders and Hoyt, 1992; Roof et al., 1992) two kinesin-related proteins that are involved in mitotic spindle function.

We have designed two secondary screens in order to identify kinetochore mutants. One assay monitors transcriptional readthrough of a centromere, and the other monitors the stability of a test dicentric chromosome. These in vivo assays of the integrity of a test kinetochore were used to screen the cff mutant collection. The cff collection consists of 138 independent mutants that exhibit increased loss of a nonessential chromosome. This collection represents approximately 50 genes whose products are required for high fidelity chromosome transmission in the mitotic cell cycle (Spencer et al., 1990) . Two mutations, cff13-30 and ctf14-42, tested positive in both secondary screens. We found that CTF14 was identical to NDClOI CBF2, recently shown to encode a 110 kd kinetochore protein (Goh and Kilmartin, 1993; Jiang et al., 1993) . Through a combination of genetic and biochemical approaches, we have shown that CTF13 is a previously unidentified essential protein that is a component of the S. cerevisiae kinetochore.

Readthrough Assay and Secondary Screen of the ctf Collection When transcription from a strong promoter is initiated toward a CEN DNA sequence, the mitotic segregational function of the centromere is destroyed (Hill and Bloom, 1987) without disruption of its 180-220 bp nuclease protected region (Bloom et al., 1984; Hill and Bloom, 1987) , indicating that at least some of the kinetochore complex remains intact. Furthermore, it has been observed that the majority of transcripts terminate at the border of the CEN sequence (P. Phillipsen, personal communication). These observations suggest that the CEN DNA-protein complex is responsible for this transcriptional block. In the reporter plasmid used to test this hypothesis (Figure l) , the GAL70 promoter initiates transcription of an actin-laczfusion gene. A wild-type CENG(185 bp) inserted in the actin intron allowed only 1% of the j3-galactosidase levels seen when no CEN was present ( Figure 1 ). The structurally dicentric reporter plasmid was maintained in a functionally monocentric state by keeping transformed strains on medium containing galactose. Transcription initiated from the GAL70 promoter inactivates the segrega- 

tional function of the test CEN (Hill and Bloom, 1987) . To test the hypothesis that a CfN DNA-protein complex was responsible for the transcriptional block, we replaced the wild-type CENG sequence with a CEN6 sequence containing a single-nucleotide point mutation (CDEIII-15C) in the central element of the palindrome of CDEIII (CCG). This transversion from G to C causes a 250-fold increase in the rate of mitotic missegregation of a chromosome fragment (Hegemann et al., 1988; Jehn et al., 1991) . Similar central element mutations have been shown by in vivo footprinting to result in decreased methylation protection of CEN DNA (Densmore et al., 1991) . The CDEIII-15C CEN mutant inserted in the actin intron allowed approximately 20% of the 8-galactosidase levels seen when no CfN was present ( Figure 1 ). Thus, a CEN DNA mutation affecting kinetochore integrity caused an increase in transcriptional readthrough that was detectable by increased levels of 8-galactosidase activity. This increase in @galactosidase activity could also be detected as blue colony color when strains were grown on solid medium containing X-gal (see Experimental Procedures), providing a sensitive visual assay for rapid screening of the ctf collection.

We proposed that the transcriptional block provided by the full-length wild-type CfN6 might be relaxed in cff strains with mutant kinetochore proteins. The reporter minichromosomes used in the control experiments would very likely be present in highly variable copy number in these ctf strains, owing to increased rates of nondisjunction and/or loss. This could result in the appearance of false positives and false negatives. To maintain the reporter in single copy, we integrated it into the cff strains. Two independent transformants of each cff strain, containing an integrated wild-type CEN reporter, were plated on medium containing X-gal and monitored for the appearance of blue colony color (see Experimental Procedures). Of 34 cff mutants screened (see Experimental Procedures for list), 7 were identified as putative kinetochore mutants because they produced an intermediate level of blue colony color between the levels of the CTF+ strains carrying the wild-type and mutant CEN reporters. Five of these mutant strains (designated "s" followed by an isolate number) are members of complementation groups, ~10 (ctf7) s9 (ctf8), s30 (ctfl3), ~42 (ctfl4), and s61 (ctfl7) and two contain independent mutations, ~26 and s58 (Table 1 ).

Quantitative measurement of P-galactosidase activity levels in protein extracts from these strains verify the identifi-cation of a relaxed transcriptional block by the colony color assay (Table 2) .

Dicentric Chromosome Stabilization Assay and Secondary Screen The second assay we developed to screen for kinetochore mutants among the cffcollection is based upon the behavior of dicentric chromosomes as they undergo mitotic segregation. If a chromosome has two centromeres, kinetochores on the same chromatid may become attached to opposite poles of the mitotic spindle ( Figure 2A ). When this occurs, the DNA molecule usually breaks, and the dicentric chromosome is rapidly lost or is rearranged to a stable form (Mann and Davis, 1983; Haber and Thorburn, 1984) . A kinetochore mutant might assemble kinetochores that have a weakened attachment of chromosomal DNA to microtubules. This could lead to microtubule detachment before chromatid breakage ( Figure 28 ) resulting in stabilization of the dicentric chromosome.

The artificial chromosome fragment present in the c?f strains was an appropriate substrate for the construction of a dicentric test chromosome. The chromosome fragment, a nonessential disome, possesses all the sequences required for proper chromosome segregation. Its stability can be visually monitored by the degree of colony color sectoring (see Figure 38 ) (Spencer et al., 1990; Shero et al., 1991) , and selective pressure for rearrangement to a stable form is absent because the chromosome fragment is not essential for viability. The GAL&EN constructs developed in the transcriptional readthrough assay allow control of the mitotic activity of a centromere by the choice of carbon source in the medium. We constructed a vector that would direct integration of these test conditional centromeres to the /euBd 7 locus present approximately 23 kb from the centromere on the chromosome fragment (see Experimental Procedures). In control experiments, we examined the stability of dicentric chromosome fragments containing either a nearly wild-type (ACDEI) or a highly defective (CDEIII-15C) secondary conditional CfN ( Figures 3A and 38 ). We predicted that upon activation Assays were performed on strains grown at 30%. CENT, wild-type CENG. B s16(ctf9) was not identified as a putative kinetochore mutant by the transcriptional readthrough assay and serves as a negative control. D s42 (ctf14) is inviable at 30%. In (B), the arrowhead indicates release of the microtubule attachment to the chromosome, allowing the dicentric chromatid to proceed intact to the spindle pole.

The hypothesis that a kinetochore mutant might result in thestabilization of a linear dicentric chromosome is based on a previous study of the behavior of dicentric minichromosomes (Koshland et al., 1987) . A circular minichromosome carrying a single wild-type centromere is quite stable in S. cerevisiae (maintained in 98%-990/o of the population under selection), whereas a minichromosome with two wild-type centromeres is highly unstable (maintained in only 6% of the population). However, when two identical partially defective centromeres (which by themselves allowed maintenance of minichromosomes in 91% of the population) were placed on the same minichromosome, the plasmid was not destabilized to the same degree (maintained in 49% of the population).

In this case, defective kinetochore function was due to mutation of the CEA! DNA sequence. By analogy, it is possible that kinetochore dysfunction due to a defective or aberrant protein will also result in stabilization of a test dicentric chromosome.

of the secondary CEN, the dicentric chromosome fragment containing a strong secondary CEN would be highly unstable, resulting in a frequent sectoring phenotype, and the dicentric chromosome fragment containing one wild-type and one defective CfNwould be relatively stable, resulting in fewer sectors per colony ( Figure 3B ). The actual sectoring phenotypes that resulted, shown in Figure 4 , were consistent with our hypothesis, indicating that the dicentric stability assay was a feasible screen for kinetochore mutants among the cff collection ( Figure 3C ). To screen the cff mutants for stabilization of a dicentric chromosome, the conditional nearly wild-type CEN, ACDEI, was integrated into the chromosome fragment present in each strain. The strains were maintained on galactose to induce the GAL.70 promoter and inactivate the conditional CEN (see Experimental Procedures and Figure 3A ). Two transformants of each cff strain were streaked onto medium containing dextrose to activate the dicentric state by repressing transcription from the GAL70 promoter. The stability of the dicentric chromosome fragment in the crf strains was visually assessed and compared with the stability of the dicentric chromosome frag-ment in the CTF+ strain. If the dicentric chromosome fragment was as unstable as in the wild-type background ( Figure 3B ), the cti strain was scored negative in this assay. With 27 cff mutants tested (see Experimental Procedures for list), 2 exhibited a reduction in sectoring frequency relative to the wild-type control and were thus identified as putative kinetochore mutants: ~30 (ctfl3), and ~42 (ctfl4) (see Table 2 ). The sectoring phenotypes exhibited by ~30 (ctfl3) and a representative mutant that was scored negative, ~16 (ctfg), are shown in Figure 4 . The sectoring phenotype of ~42 (ctfl4) carrying the dicentric chromosome fragment is similar to that seen for ~30 (ctfl3). The sectoring phenotypes of these mutants are similar to that seen with a test dicentric chromosome carrying a weak secondary CEN.

Two &strains, ~30 (ctfl3) and ~42 (ctfl4), were scored as putative kinetochore mutants in both secondary screens. The ctf74-42 mutation identifies a recently characterized kinetochore component, NDClOICBF2 (see below). We therefore explored further whether the cff13-30 and active on dextrose medium (the test chromosome behaves as a dicentric). (B) The stability of the test dicentric chromosome fragment can be monitored visually. A tRNA suppressor gene (SUP1 1) present on the chromosome fragment partially suppresses the accumulation of red pigment caused by the ade2-101 ochre mutation in our strain backgrounds. If the chromosome fragment is present, the strain is white; if it is lost, the strain is red. Thus, the numberof red sectors that develop during the growth of a colony founded by a haploid cell containing the chromosome fragment (white) is indicative of the rate of loss of the chromosome fragment in the strain. The lines within the circles represent the presence of such red sectors in a white colony. The sectoring phenotypes pictured are those predicted and observed (see Figure 4 ) when nearly wild-type (CEN*v) and highly defective (CEN-) CEN

The centromere originally present on the chromosome fragment is fully wild type (CENW). Labels at the left indicate the type and number of CEN DNAs present on the test chromosome fragment. Labels across the top indicate the relevant genotype of the pictured strain. sl6 (ctfg) was scored negative; its sectoring phenotype with the test dicentric chromosome fragment (column 2, row 2) was the same as that seen in the CTf+ background (column 1, row 2). s3g (ctfl3) was scored positive; its sectoring phenotype with the test dicentric chromosome fragment (column 3, row 2) was not as severe as that seen in the CTF+ background (column 1, row 2) and looked similar to that seen with the test dicentric chromosome carrying the weak secondary CEN (column 1, row 3). mutation had identified a gene encoding a kinetochore component.

Molecular Cloning cff73-30 is completely deficient for growth at 37%, and this temperature sensitivity was shown to cosegregate with its moderate sectoring phenotype at 25%. CTf73 was cloned by complementation of lethality at 37% (Spencer et al., 1988) . A 2.2 kb Sau3A fragment that complemented the temperature sensitivity of ctf73-30 was shown to correspond to CTF73 by the directing of an integration event in a heterozygous diploid. This event introduced a prototrophic marker at the genomic site of the cloned DNA segment and deleted approximately half of the 2.2 kb genomic sequence (almost the entire CTF73 open reading frame [ORF]; see Experimental Procedures and Figure  58 ). When the diploid transformants were dissected, it was found that viability segregated 2:2 (see Experimental Procedures). We concluded that the cloned DNA encodes wild-type CTF73 and that CTF73 is an essential gene in S. cerevisiae. CTf 73 was localized to the right arm of chromosome XIII using both physical and genetic mapping methods ( Figure 5A ). From the mapping data, we concluded that CTF73 is a previously unidentified gene in S. cerevisiae.

The nucleotide sequence of the 2.2 kb CTF73 clone contains a 1.4 kb ORF that encodes for a protein of 478 amino acids with a predicted molecular weight of 58 kd ( Figures 5B and 8 ). The CTF73 protein shows no significant overall homology at the amino acid level to entries in Gen-Bank, GenPept, GPUpdate, SwissProt, PIR, EMBL, and EMBLUpdate data bases as of January 1993. The homology searches were performed on the National Center for Biotechnology Information BLAST network (Altschul et al., 1990) . The CTF7 3 protein contains a short acidic serinerich region (amino acids 200-230) that is approximately 40% identical to the first acidic block found in a mammalian centromere-associated protein, CENP-B (Pluta et al., 1992) . The significance of this small region of similarity is unclear, and there are no other significant homologies found outside this area. Interestingly, there is a possible CDC28 phosphorylation site in the CTF73 protein (SSPSS, amino acids 224-228) (Figure 8 ).

Analysis Lechner and Carbon (1991) have described a multiprotein complex (CBF3) present in nuclear extracts of S. cerevisiae cells that binds in vitro to a 350 bp fragment of CEN DNA. DNA footprinting reveals that CBF3 interacts with the CDEIII sequence element. Using a modification of the methods of Lechner and Carbon (1991) we were able to detect the binding of CBF3 complexes present in wholecell extracts to an 88 bp DNA probe that spans CDEIII but lacks CDEI and CDEII.

To determine whether CTF13 is a component of the CDEIII-binding complex, the CTF13 ORF was fused to peptide epitopes against which antibodies had been pre- by Southern hybridization to an etectrophoretic karyotype (Spencer et al., 1966; Gerring et al., 1990a) . CTF73 was positionally mapped by the method of chromosome fragmentation (Gerring et al., 199Oa) and was localized to the right arm of chromosome XIII, 475 kb from the right arm telomere and 445 kb from the left arm telomere (see Experimental Procedures). This physical location was verified by using the 2.2 kb SauM fragment to probe a set of filters containing contiguous overlapping 1 clones that cover 66% of the yeast genome (L. Riles and M. Olson, unpublished data). CTF73 was localized to overlapping clones 4199 and 6643, placing it on an -15 kb segment of DNA located on the right arm of chromosome XIII between adh3 and i/v2 The temperature-sensitive cff7530 mutation was meiotically mapped and found to be located 34 CM proximal to the cin4 locus, which agrees well with the physical mapping data (see of CTF13 (see Experimental Procedures). In the first construct, an 11 amino acid epitope derived from the HA1 protein of influenza virus (Field et al., 1988) was inserted in frame into the amino terminus of CTF13 ( Figure 5C ). In the second construct, two tandem copies of the El epitope (Pluta et al., 1992) , derived from the carboxy-terminal 25 amino acids of an avian coronavirus glycoprotein (Machamer and Rose, 1987) were placed in frame at the amino terminus of the CTF13 ORF under the transcriptional control of the GAL7 promoter ( Figure 5C ). Both epitope-tagged CTF13 derivatives were able to rescue viability in a ctfl3 Al::HIS3 null strain.

Extracts were prepared from cells carrying either wildtype or epitope-tagged CTF13, reacted with 32P-labeled CDEIII DNA, and complexes were resolved on a nondenaturing gel (Figure 7) . A single band corresponding to a CDEIII-protein complex was observed (Figure 7 , lanes 1, 7, and 10); no complex was observed with a nonfunctional CDEIII variant (data not shown). The addition of antiepitope antibodies to binding reactions containing extracts from cti73 mutant strains rescued by the respective CTFl3-epitope fusion protein resulted in the appearance of a complex with significantly decreased electrophoretic mobility (Figure 7, lanes 2-4 and 11 ). This supershift is DNA-protein complexes formed with YP-labeled CDEIII probe and whole-cell extracts were analyzed on a nondenaturing acrylamide gel. Antibodies were added to preformed complexes, and samples were incubated for 20 min at room temperature before gel analysis. Unbound probe was run off the bottom of the gel. Lanes 1-6, extracts from cff73d7::H/S3 null cells carrying an HA epitope-CTF13 fusion (see Figure 5C ) and incubated with antibodies at various dilutions; lanes 7-9, extracts from cff13dl::H/S3 null cells carrying a CTF13 plasmid reacted with the indicated antibody (controls for lanes l-6); lanes 10-12, extracts from ctf73-30 cells carrying an El epitope-CTF13 fusion (see Figure 5C ) and incubated with indicated antibodies (control is lane 6). HA indicates the addition of 12CA5 monoclonal antibody, which is directed against the HA epitope; El indicates the addition of a polyclonal serum directed against the El epitope; peptide indicates the addition of HA peptide to 1 mM prior to the addition of 12CA5 antibody. clearly antibody specific, because antibodies directed against the El epitope did not recognize the HA-CTF13 fusion protein (Figure 7, compare lanes 6 and 4) and antibodies directed against the HA epitope did not recognize the El -CTF13 fusion protein (Figure 7 , compare lanes 12 and 11). As expected, the supershift was also shown to require the presence of El-CTF13 (Figure 7 , compare lanes 6 and 11) or HA-CTF13 (Figure 7 , compare lanes 8 and 9 with lanes 3 and 4) and to be competed with HA peptide (Figure 7, compare lanes 5 and 3) . These results show that the supershifted band shown in lanes 2-4 and 11 of Figure 7 is composed of a complex containing proteins, DNA, and antibody.

These results demonstrate that CTF13 is present in the protein complex that binds to the essential CDEIII region of S. cerevisiae CfN DNA in vitro. Because all of the CDEIIIprotein complex formed in our reactions were able to be supershifted by antibodies directed against epitope-tagged CTF13, the stoichiometry of CTF13 and DNA in the complexes must be at least 1 to 1. We conclude from these data that CTF13 is a major component of the yeast kinetochore, which, probably in combination with other proteins, interacts with CDEIII.

Phenotypic Analysis The cff73-30 mutant allowed transcriptional readthrough of a test CEN and stabilized a test dicentric chromosome fragment. Further phenotypic analysis of this mutant revealed defects consistent with defective kinetochore function.

The colony color assay for chromosome fragment stability can be used to monitor the rates of chromosome fragment loss and nondisjunction events in diploids. These rates were measured for a ctf13-30 homozygous diploid and its wild-type parent at permissive temperature (25OC). The cti73-30 homozygous diploid exhibited an approximately 50-fold elevation both in the rates of nondisjunction and in loss of the chromosome fragment (Table 3A) . The rates of mitotic missegregation and recombination of a suitably marked endogenous chromosome Ill were also measured. The mitotic missegregation rate of chromosome Ill was elevated 1 O-fold in the ctf73-30 homozygous diploid, while the mitotic recombination rate was only elevated 4-fold (Table 38) . We conclude that the cti73-30 mutation confers mitotic segregation and recombination phenotypes consistent with a role in the segregational machinery.

cti73-30 causes cells to arrest at the G2/M phase of the cell cycle when shifted to the nonpermissive temperature. Flow cytometric analysis of DNA content per cell revealed an accumulation of cells with a G2 DNA content during log phase growth at the permissive temperature and a single peak of G2 content DNA after arrest at the nonpermissive temperature ( Figure 8A ). Quantitation of cell and nuclear morphology at the permissive temperature also indicated an accumulation of cells with a G2 content; 13% of cells were large budded with the nucleus at the neck in a cff73-30 background, while only 2% of wild-type cells had this morphology ( Figure 8C ). cff73-30 is a cdc-like mutation that arrests with a cell morphology indicative of the G2/M preanaphase portion of the cell cycle. After 3 hr at nonpermissive temperature (38X), approximately 80% of cff73-30 homozygous diploid cells had arrested as large-budded cells with an undivided nucleus positioned at or near the neck between the mother and daughter cells. The mitotic spindle was very short in virtually every cell; a medium or long (anaphase B-like) spindle is never seen (Figure 86, upper panels) .

The cdc arrest is leaky in cff73-30 at 38OC, and the uniform cell morphology decays with time (see Figures 88  and 8C ). The mitotic spindle phenotype also becomes less uniform with the appearance of misaligned and aberrantlooking spindles. Interestingly, after 5 hr at the nonpermissive temperature, a "cut"-like phenotype is observed in approximately 10% of the population (though present in only 2% of the population at the 2 hr time point). This morphology is reminiscent of the phenotype of Schizosaccharomyces pombe cut mutants (Hirano et al., 1986) as well as of the phenotype observed in topoisomerase II mutants of S. cerevisiae (Holm et al., 1985) . We define this cut-like cell morphology as a very narrow-necked, large-budded cell in which the nucleus straddling the neck has a pinched appearance (see Figure 8B , lower panel).

These data demonstrate that the cff73-30 mutation results in a defect revealed in the G2/M phase of the cell cycle, consistent with a defect in kinetochore function.

The ctistrain s42 (ctfl4) was also identified by both secondary screens as a putative kinetochore mutant. A clone that complemented the temperature sensitivity of clf74-42 was obtained and mapped to chromosome VII essentially as described for clf73-30 (data not shown). Nuclear division cycle 70 (NDCIO), recently identified by Goh and Kil-cells is slightly tighter (a single G2 peak) when incubated at 36% for 3 hr (data not shown). shown above the columns. The numbers shown represent the percentage of total cells scored; small-budded cells with a single nucleus were quantitated (percentage is 100 minus the sum indicated) but are not shown. At 25°C 1500 cells were scored; 200 cells were scored for each time point at 36%. martin (1993) as an essential gene involved in chromosome segregation in S. cerevisiae, is identical to CBF2, a gene recently identified by Jiang et al. (1993) that encodes the 110 kd subunit of the CBF3 complex (Lechner and Carbon, 1991) . Multiple internal restriction fragments from the NIX70 clone were found to comigrate with fragments from the cti74-42 complementing clone. Moreover, the temperature-sensitive mutation cff74-42 failed to complement the temperature sensitivity of n&70-7. We conclude that the cff74-42 mutation is present at the NDC70/ CBF2 locus. Thus, the only two cff mutants identified by both secondary screens as putative kinetochore mutants have now been shown to be defective in essential kinetochore components.

Although the CEN DNA sequence elements from budding yeast have been cloned for over 10 years (Clarke and Carbon, 1980) , identification of the genes encoding proteinsessentialforcentromerefunction hasprovendifficult. We describe a genetic approach using two independent in vivo genetic assays to screen an existing large reference set of mitotic segregation mutants (the cffcollection; Spencer et al., 1990) for altered kinetochore integrity. In combination, these assays identified two mutant strains, ~30 (ctf 13) and ~42 (ctfl4), as putative kinetochore mutants. Biochemical and further phenotypic analysis indicated that the CTF73 gene product was indeed an essential structural component of the kinetochore, and the CTF74 gene product was shown to be identical to NDC70KBF2, a recently characterized essential kinetochore component (Goh and Kilmartin, 1993; Jiang et al., 1993) .

Mutants Theoretically, the transcriptional readthrough assay might result in both false negatives (e.g., kinetochore protein mutations that fail to relieve a transcription block) and false positives(e.g., mutationsthat affect transcriptional regulation or chromatin structure). Similarly, a dicentric chromosome could be stabilized by mutants affecting DNA metabolism or spindle integrity and assembly. In light of these caveats, we used these assays to screen a set of mutants previously shown to have defects in mitotic chromosome segregation. It is not known how efficient either of these secondary assays would be in a primary screen.

Our experience suggests that kinetochore mutants can be recognized by the combined phenotype of transcriptional readthrough and dicentric chromosome stabilization. In theory, the degree to which the integrity of the kinetochore must be compromised to result in either of these phenotypes could be quite different. In a simplified view, mutations affecting the interaction of the kinetochore complex with the CEN DNA should be detected by both assays, while mutations affecting kinetochore to microtubule interactions may only be detected by the dicentric stabilization assay. However, we note that kinetochores participate in several distinct processes in vivo, including microtubule capture, congresaion to the metaphase plate, and poleward migration. In addition, a viable kinetochore mutation would most likely be a leaky mutation, which might exhibit complex consequences following the primary defect. Thus, in reality, it is quite possible that some of the mutations identified by only one of these secondary screens indeed disrupt kinetochore integrity but perhaps lead to more subtle alterations than cff73-30 or cff74-42. It is clear that these two screens, whether used alone or in combination, have the potential to aid in the identification of additional regulatory and structural components of the S. cerevisiae kinetochore. They may also be adaptable to other organisms.

CTF13 Is an Esserttial Kinetochore Compomnt We have presented a combination of in vivo and in vitro evidence demonstrating that the CTF13 protein is a cornponent of the S. cerevisiae kinetochore. In vivo, tf& ctf73-30 mutation confers relaxation of a transcriptional Mock mediated by the kinetochore and stabilizes a test dioentric chromosome fragment. In vitro, we demonstrate otat the CTF13 protein is a component of the CEN DNA-@rot&n complex and, specifically, that it interacts withCDElll. The predicted molecular mass of 56 kd for the CTF13 protein is the approximate size seen on a Western blot (data not shown). Therefore, the CTF13 protein seemed to be a very good candidate for the 58 kd subunit of the CBF3 complex (Lechner and Carbon, 1991) , and in fact, the predicted amino-terminal amino acid sequence of the CTF13 protein was found to be identical to tryptic peptide sequence obtained from the purified 58 kd protein component of the CBF3 complex (J. Lechner, personal communication).

The CTF13 protein appears to be limiting for CDEIIIprotein complex formation in vitro. When extracts derived from a strain overproducing CTF13 are used in the band shift assays (see Figure 7 , lanes lo-12), the amount of CDEIII-protein complex formed is increased relative to the amount seen with extracts from nonoverproducing cells (see Figure 7 , lanes l-9). Also, a cti73 heterozygous diploid strain (cti7347::H/%/CTF13) exhibits a mild but detectable sectoring phenotype. This indicates that the amount of CTF13 protein produced by one copy of the CTF73 locus is not sufficient to keep the fidelity of chromosome segregation at a wild-type level. These observations suggest that CTF13 may be limiting for kinetochore function in vivo.

Phenotypic analysis of the temperature-sensitive cff73-30 mutation is consistent with a kinetochore defect. cti73-30 causes an increase in the mitotic rate of chromosome missegregation and results in a terminal phenotype indicative of a defect in the G2/M phase of the cell cycle. It has been previously proposed that missegregation mutants will fall into two broad groups: those affecting the pathways of DNA metabolism and those affecting the mitotic segregational machinery. A mutation affecting DNA metabolism was expected to cause increased rates of both chromosome loss and mitotic recombination, while a mitotic segregation mutant was expected to cause only an increase in chromosomal loss events. Phenotypic analysis of a known DNA metabolic mutant, DNA polymerase a (c&77; Hartwell and Smith, 1985) , and a known spindle mutant, B-tubulin (r&2; Huffaker et al., 1988), supported this model. Examination of these phenotypes for cff73-30 strains revealed a significant increase in the rate of chromosomal missegregation with only a very slight elevation in the rate of mitotic recombination (Table 38 ). In addition, we now have the ability to distinguish between loss (1:O) and nondisjunction

(2:0) missegregation events, and we find that the rates of both of these events are significantly elevated in.& ctf73-30 background (Table 3A) . Thus a known kinetochore mutation, cti73-30, has been shown to result in phenotypes consistent with previous expectation, and we can perhaps extend this expectation to include a predicted increase in the rates of both chromosomal loss and nondisjunction in mitotic segregation mutants.

The ctf73.30 Kinetochore Defect May Be Recognized by a Cell Cycle Checkpoint Critical events in the cell cycle are temporally ordered and coordinated by a series of dependency pathways in which late events are dependent on the successful completion of earlier events. These dependencies can result from a substrate-product mechanism or from extrinsic control by a monitoring function termed cell cycle checkpoint control (Hartwell and Weinert, 1989) . Checkpoints are responsible for a subset of observed cell cycle arrests or delays. Cell cycle arrests or delays associated with kinetochore defects have been reported in several systems. In animal cells, a delay in the initiation of anaphase is correlated with the failure of chromosomes to achieve bipolar attachment to the spindle (Rieder and Alexander, 1989; Zirkle, 1970) and a metaphase arrest is observed with kinetochore disruption by injection of anti-centromere antibodies (Bernat et al., 1990) . In yeast, one abberant kinetochore on a single chromosome can cause a mitotic delay (Spencer and Hieter, 1992) . ctf73-30 strains exhibit a G2/M phase accumulation in logarithmic cultures at permissive temperature and a preanaphase arrest morphology at nonpermissive temperature. Cell morphology and DNA content do not critically distinguish G2 and M phases in yeast. However, at permissive temperature, crf73-30 strains exhibit a detectable increase in Hl kinase activity relative to CTF73 controls, and at nonpermissive temperature, Hl kinase activity levels in cff73-30 strains are equivalent to nocodazole-arrested strains (data not shown). Thus, Hl kinase activity measurements suggest that the cti73-30 mutation causes an accumulation in M phase. It is tempting to speculate that this cell cycle alteration is similar to those described above and that these are a result of checkpoint control exerted in the presence of defective kinetochores.

Checkpoints are defined by two experimental criteria: first, identification of mutations or conditions that allow bypass of an arrest or delay (resulting in the accumulation of errors) and second, an observed error correction when cell cycle delay is reintroduced experimentally.

Alternatively, defective substrate-product conversion that becomes rate limiting for progress may also result in cell cycle delay. These alternatives have not been distinguished for the delays seen associated with kinetochore defects. Conditional mutations in kinetochore proteins will provide important tools for exploring the relationship be-tween kinetochore structure and cell cycle progression.

Examination of the terminal phenotype of cti73-30 mutants raises several interesting questions. The fact that the ctf73-30 defect does not lead to a permanent and uniform arrest morphology may simply be a result of the presence of a small amount of active CTF13 protein that eventually allows completion of mitosis, or mitosis may never be completed but cytokinesis may still eventually be attempted in some cells. Consistent with the latter possibility, we have observed the accumulation of cells with a cut-like phenotype: 10% of all cells exhibit this phenotype after 5 hr at the nonpermissive temperature. Bernat et al. (1990) describe a similar cut-like phenotype after injection of mammalian cells with anti-centromere antibodies and propose that it is a result of the cells' eventual attempt to undergo cytokinesis after prolonged mitotic arrest. Because it is not known whether this subset of the cff73-30 population is still capable of dividing, it is unclear whether cytokinesis has trapped nuclei in these cells or whether they are caught undergoing nuclear transits at the time of fixation (Palmer et al., 1989) .

The terminal phenotype of ~7773-30 is quite different from the terminal phenotype of the other described temperature-sensitive kinetochore mutant, n&70-7 (Goh and Kilmartin, 1993). n&70-7 mutants exhibit detatchment of the chromosomes from one spindle pole and progression through the cell cycle in the absence of chromosome segregation (most cells produce one aploid daughter and one daughter of increased ploidy). If there is checkpoint control exerted in response to events at the kinetochore, the ndclO-7 defect is not recognized. Perhaps this is because the NDC70 protein itself is involved in the recognition and/ or signaling of a dysfunctional complex, or alternatively, checkpoint control may be disabled by complete disruption of kinetochore structure. Future experiments addressing the relationships of kinetochore proteins to the control of progression through mitosis should help define important molecular determinants of the temporal order of events in chromosome segregation.

Will the Molecular Dissection of the S. cerevisiae Kinetochore Aid the Understanding of Kinetochore Function in More Complex Eukaryotes? At this time, analysis of the DNA sequence and protein component requirements of the kinetochore is significantly more advanced in S. cerevisiae than in any other eukaryotic organism, although there is great speculation about the relevance of these studies to the understanding of the much larger and morphologically more elaborate kinetochores present in other eukaryotes. While there may be a need for additional components to ensure fidelity in more complex eukaryotes, we think it is probable that the basic mechanisms of the segregational process, including those involved in centromere function, will have been conserved through evolution. A repeat subunit model for the centromere-kinetochore complex has recently been proposed by Zinkowski et al. (1991) . This model describes the kinetochore as organized in multiple small repeat units that fold together into a contiguous plate-like structure when condensed at metaphase. Zinkowski et al.

propose that each unit is capable of microtubule binding and segregational function. In this context, the S. cerevisiae kinetochore, which binds a single microtubule, could represent the simplest ancestral unit of the eukaryotic kinetochore (Fitzgerald-Hayes et al., 1982; Koshland et al., 1987) .

Identification and characterization of the S. cerevisiae kinetochore components will facilitate the definition of the activities necessary for the completion of proper mitotic segregation in this organism and may well provide substrates for the identification of kinetochore components in other eukaryotic organisms.

Yeast Strains and Media

The c/f and wild-type parental strains containing chromosome fragments that can be monitored by a visual assay have been previously described (Spencer et al., 1990; Shero et al., 1991) . The et/collection of 136 originally isolated mutants can be represented in 18 complementation groups and 41 single isolates. All cff mutant isolates that are members of a complementation group retain the original isolate number as an allele number (e.g., 930 contains cff/3-30). One member of each complementation group (the isolate with the most severe sectoring phenotype) and 19 single isolates (those that were his3-) were tested in the two kinetochore screens. Media for yeast growth and sporulation were as described (Rose et al., 1990 ) except that where sectoring was examined, adenine was added to 6 us/ml to minimal (SD) medium to enhance the development of red pigment in ada2-101 strains. X-gal plates were made as described for synthetic complete (SC) medium (Rose et al., 1990) except for the addition of 0.1 M NaPOl(pH 6.6) and 40 pg/ml X-gal (5-bromo4chloro-5indolyl-6-D-galactopyranoside) (use a 20 mg/ml stock in dimethylformamide). All yeast transformations were done by the method of Ito et al. (1963) .

Readthrough Assay The reporter construct schematically pictured in Figure 1 was modified from pGAB (U. Vijayraghavan and J. Abelson, unpublished data) obtained from R. Parker. (pGAB is a modification of pYAHB2 [Vijayraghavan et al., 19661, a CEN-ARSplasmid containing an actin-/acZ fusion gene that has been used extensively to study splicing in S. cerevisiae [Vijayraghavan et al., 1966; Cellini et al., 19861 mann et al., 1988) . The DNA sequence and orientation of the test CEN6 sequences in pKF16 were.verified by sequencing (Sanger et al., 1977; Hattori and Sakaki, 1966) ,The resulting plasmids, pKF19 and pKF44, contain wild-type and mutant (CDEIII-15C) CEN8 sequences, respectively, in the prientatipn placing CDEI 5'to CDEIII. In control experiments, pKFI 6, pKF19, and pKF44 were transformed into the wild-type strain YPH102 tiAi$ ura3-52 /ys2-801 ada2-101 his3-A200 /eu2-A 1, and p-galactosidase assays were done on independent URA+ transformants (see Figure 1 ). The structurally dicentric plasmids (all YCpBO derivatives) were maintained in a functionally monocentric state by keeping transformed strains on medium containing galactose as the carbon source causing transcriptional inactivation of the test centromeres (Hill and Bloom, 1967) . 6galactosidase assays were performed essentially as described (Rose et al., 1990) following the protocol for assay of crude extracts, except that cells were grown in SCGal-His liquid medium or scraped off SCGal-His plates. The values for optical density at 420 nm were zeroed to an isogenic yeast strain that did not contain a reporter plasmid.

For screening of the ctfcollection, the reporter was integrated into chromosome XV as follows: the GAL&IO-actin-test CENG-/acZ fragment described above was inserted into a genomic Xhol site immediately 3' to the HIS3 gene contained on a pBR322-based plasmid, pSZ62-Xbal (McCleod et al., 1966) kindly provided by J. Broach. The resulting BamHl fragment containing HIS3 and the test reporter fragment was transformed into YPH276 selecting for replacement of the his3A200 locus on chromosme XV. Independent His+ transformants were picked and analyzed by Southern blotting toverify insertion of the reporter construct into chromosome XV at the HIS3 locus. pKF71 contains the wild-type CEN6 reporter, and pKF72 contains the mutant CDEIII (15C) CEN6 reporter inserted into the HIS3 BamHl fragment. YPH977 and YPH976 contain the pKF71-and pKF72derived BamHl fragments, respectively, and were maintained in medium containing galactose.

Strains were tested for the production of blue color on medium containing the chromogenic substrate of 6galactosidase, X-gal. YPH977 colonies appear white (this progresses to a very faint blue color after several days), while YPH978 colonies develop a deep blue color.

The reporter containing the wild-type CEN6 (pKF71) was inserted into chromosome XV in each of the cff mutants as follows. Each cff strain was made competent for transformation in SC medium containing 2% galactose and transformed with the BamHl fragment of pKF71. Transformants were selected, and the colony was purified on SCGal-His plates at 25OC. Two independent transformants of each cff strain tested were then plated at a low density (-200 Two p679 and two pKF77 transformants of each cff strain tested were streaked onto synthetic complete dextrose plates containing a limiting amount of adenine. The switch to dextrose as a carbon source causes the GAL70 promoter to be turned off, resulting in activation of the second conditional centromere and a functionally dicentric chromosome fragment. Sectoring phenotypes were directly compared with those of a YPH276 p679 or pKF77 transformant streaked onto the same plate.

The &strains tested with the stabilization of a dicentric assay were: 659 (ctf2), ~50 (ctf4), ~31 (ctf5), ~53 (ctf6), ~10 (ctfi'), s9 (ctf6), ~16 (ctf9), ~67 (ctfll), ~16 (ctfl2), s30 (ctf13), ~42 (ctfl4), YPH960 MATa his3-A200 ade2-101 lys2-801 leu2-Al cdl&124

CFVll (RADLd.YPtf 275) URA3 SUP1 1, ~61 (ctflir), YPH961 MATa ~6~3-52 lys2-801 adeB 101 his3-A200 trpl-A 1 leulA 1 ctf18-160 CFlll (CEN3.L. YPH278) URA3 SUPIl, S3, S4, S12, S17, ~20, ~22, ~41, s.47, ~55, ~56, ~62, ~63, and ~64. ~31 (ctf5) and ~20 were unscorable in this screen because the chromosome fragment present in the p679 derivative strains was extremely unstable. True positives were verified in multiple independent transformants. One source of false positives was transformants containing two chromosome fragments, only one of which carried the conditional secondary centromere. These false positives were easily identified by the demonstration that sectored colonies were His+. For ~30 (ctft3), the phenotype of stabilization of the test dicentric chromosome was shown to be due to a mutation in the CTF73 gene product by transformation of a pKF76 derivative of ~30 with pKF1 I, a pRS3lCbased (Sikorski and Hieter, 1969) plasmid carrying the CTFf3 locus. The presence of the wild-type CTF73 gene product resulted in destabilization of the dicentric chromosome fragment back to the level seen in the wild-type parent, YPH276.

Characterization of CTFl3 The 2.2 kb Sau3A subclone that rescues the temperature sensitivity of S30 (ctfl3) was inserted into the polylinker of pRS314 (Sikorski and Hieter, 1969) , resulting in pKF1 I, Deletion derivatives of the pKFt 1 insert were made using existing restriction sites (see Figure 5) . A Bglll to polylinker deletion, as well as a Clal to polylinker deletion, was unable to rescue the temperature sensitivity of ~30 (ctfl3). The DNA sequence encoding the entire ORF was determined using a set of unidirectional deletions (Henikoff, 1967) by standard methods (Sanger et at., 1977; Hattori and Sakaki, 1966) . The sequence of the second strand of the ORF was obtained using synthetic oligonucleotides as primers.

The CTF13 clone was shown to correspond to the cff73-30 locus, and CTF73 was shown to be an essential gene in S. cerevisiae by using the CTFl3 clone to direct an integration event (Sikorski and Hieter, 1969) that replaced a majority of the CTF73 ORF with vector and HIS3 sequences.

The integration vector, pKF93, was constructed by inserting the -600 bp Bglll (polylinker)-Bglll fragment and the -200 bp Clal-EcoRI (polylinker) fragment from pKFl1 (see Figure  58 ) into the BamHl site and Clal-EcoRI sites of pRS303 (Sikorski and Hieter, 1969) respectively. pKF93 was linearized with EcoRl and transformed into a cff73-30BTF13 heterozygous diploid strain, YPH974, selecting for His+ transformants.

Integration of pKF93 should delete the CTFl3 ORF from amino acid 57 to amino acid 467 (see Figure 58 ). Approximately half of the His+ diploid transformants obtained exhibited the cff73-30 sectoring phenotype, indicating that the cff73-30 locus was being targeted by pKF93. The integration of pKF93 and deletion of CTF73 sequences was confirmed by Southern analysis (data not shown). Two sectoring diploid isolates (CTFl3 locus deleted) and two nonsectoring diploid isolates (cff1530 locus deleted) were sporulated, and tetrads were dissected. Viability segregated 22 in all 31 tetrads dissected, and all viable spores were his-. All viable spores resulting from the sectoring diploids were temperature sensitive, and all of the viable spores resulting from the nonsectoring diploids were not temperature sensitive. CTF73 was physically mapped by the previously described method of chromosome fragmentation (Gerring et al., 199Oa) , using the 2.2 kb CTF13 fragment.

The sizes of the resulting stable chromosome fragments were determined by orthogonal field-alteration gel electrophoresis (OFAGE) analysis (Carle and Olson, 1964) . and assignment of CTF13 to an arm of chromosome XIII was accomplished by hybridization of a left arm telomere-adjacent probe, TUB3, to a Southern blot of the OFAGE gel. TUB3 was obtained from P. Schatz, and the probe used was a 1.2 kb Hindlll fragment, radioactively labeled with 'P (Feinberg and Vogelstein, 1964) . TUB3 hybridized to the 445 kb proximal fragment. To obtain a meiotic map position, a diploid strain was constructed that was heterozygous for cff73 and cin4 (~61530/+, +/chC:URA3).

The meiotic distance was calculated from the following data by using the formula of Perkins: cff73cin4 34 CM (parental ditypel nonparental ditypeitetratype = 42/2/56). CTFl3 was placed proximal to cin4 by probing the CTF73 chromosome fragmentation OFAGE blots with a 2 kb Sacl-Kpnl C/N4 fragment obtained from A. Hoyt. C/N4 hybridized to the 475 kb distal CTF73 chromosome fragment, placing CTF73 proximal to CIN4.

Analysis The plasmid containing the El tag fused to CTF13, pKF60, was constructed from the base plasmid p414GEUl (J. Kroll, unpublished data). p414GEUl has a 460 bp GAL7 promoter fragment cloned into the Kpnl site and two tandem copies of the El tag sequence, described by Pluta et al. (1992) , inserted in frame into the Apal and Xhol sites of pRS414 (Sikorski and Hieter, 1969) . The GAL7 promoter directs transcription from its own ATG toward the polylinker. An EcoRl fragment containing the entire 2.2 kb insert of pKFl1 was cloned into the EcoRl site of p414GEUl in the appropriate transcriptional orientation. The 5' -600 bp of the CTF73containing fragment (up to the Bglll site; see Figure 58 ) were removed and replaced with an -200 bp polymerase chain reaction product containing sequences from the ATG of CTF13 to the Bglll site. This allowed the in-frame fusion of the tandem El tags to CTF73 under the transcriptional control of GAL7 (see Figure 5C ). pKF60 was transformed into YPH972 and shown to rescue the temperature sensitivity caused by the ctf73-30 mutation on both galactose-and dextrosecontaining media. The plasmid containing the HA epitope fused to CTF13, pSFl97a, was constructed by using a synthetic oligonucleotide to fuse the HA epitope and linker sequences to the amino terminus of CTF13 (see Figure 5C ). The fusion protein and -200 bp of 3'noncoding sequence from the CTFl3 locus were cloned into pRS315 (Sikorski and Hieter, 1969) downstream of a 625 bp fragment of 5' flanking DNA that is presumed to include the CTF73 promoter. pSF197a was transformed into YPH975. Transformants were streaked onto medium containing 5-fluoroorotic acid to select against the CTFl3-URA3 plasmid (Boeke et al., 1967) and it was shown that pSF197a would rescue viability in the resulting ctf73Al::H/S3 strain. The preparation and analysis of CBF3-DNA complexes was performed using a modification of procedures previously described by Lechner and Carbon (1991) . Cells in log phase were harvested by centrifugation, frozen in liquid nitrogen, and mechanically disrupted by fragmentation with a liquid nitrogen-cooled mortar and pestle in 30 mM sodium phosphate (pH 7.0). 60 mM ftglycerophosphate, 1 M KCI, 6 mM EGTA, 6 mM EDTA, 6 mM NaF, 10% glycerol, 1 mM phenylmethytsulfonyi fluoride, and 10 &ml (each) leupeptin, pepstatin, and chymostatin.

Whole-ceil extract (40 ug) was incubated for 30 min at room temperature with 20 fmol of "P-labeled DNA probe, 5 ug of salmon sperm DNA, 5 pg of poly(dl-dC), and 10 pg of bovine serum albumin in 30 nl of 10 mM HEPES (pH 6.0) 1 mM NaF, 6 mM MgCI,, 10% glycerol, and KCI at a final concentration of 125 mM. The 66 bp DNA probe was derived from CEN3 and spans the core region of CDEIII, from 5 bp to the left of CDEIII to 59 bp to the right of CDEIII. Binding reactions were electrophoresed on 4% polyacrylamide gels as described (Ng and Carbon, 1967) and visualized by autoradiography.

Basic local alignment search tool

Isolation of the gene encoding the Saccharomyces cerevisiae centromere-binding protein CPI

Chromatin conformation of yeast centromeres

5-Fluoroerotic acid as a selective agent in yeast molecular genetics

Yeast centromere binding protein

of the helix-loop-helix protein family, is required for chromosome stability and methionine prototrophy

We thank C. Connelly for significant contributions to this work. We would like to thank Ft. Parker for sending pGAB, J. Kroll for allowing us to use p414GEU1, and A. Pluta for kindly providing us with poly clonal El antibodies.We would like to acknowledge S. Holloway, R. Sikorski, and N. Kouprina for helpful theoretical discussions and H. Varmus and T. Mitch&on for support during this project. We are also grateful to J. Kilmartin, J. Carbon, and J. Lechner for communicating results prior to publication.We thank D. Koshland, W. Earnshaw, and T. Kelly for critical reading of the manuscript.K. F. D. is a student in the predoctoral training program in human genetics at Johns Hopkins (National Institute of General Medical Sciences grant P32GM07614). S. T. is supported by the National Institutes of Health Departmental Training Grant 5T32CA09139. P. K. S. and A. A. H. are biomedical scholars of the Lucille P. Markey Charitable Trust. This work was supported by a National Institutes of Health grant (CA16519) to P. H. and an American Cancer Society grant (CD-509) to F. S.

Carbon, J., and Clarke, L. (1990) . Centromere structure and function in budding and fission yeast. New Biologist 2, 10-19. Carle, G. F., and Olson, M. (1964) . Separation of chromosomal DNA molecules from yeast by orthogonal-field-alteration gel electrophoresis. Nucl. Acids Res. 12, 5647-5665.Cellini, A., Parker, R., McMahon, J., Guthrie, C., and Rossi, J. (1966) . Activation of a TACTAAC box in the Saccharomyces cerevisiae actin intron. Mol. Cell. Biol. 6, 1571 -1576 . Clarke, L., and Carbon, J. (1960 . Isolation of a yeast centromere and construction of functional small circular chromosomes. Nature 287, 504-509.Cottarel, G., Shero, J., Hieter, P., and Hegemann, J. (1969) . A 125base-pair CEN8 DNA fragment is sufficient for complete meiotic and mitotic centromere functions in Saccharomyces cerevisiae. Mol. Cell. Biol. 9, 3342-3349. Densmore, L., Payne, W., and Fitzgerald-Hayes, M. (1991) . In vivo genomic footprint of a yeast centromere. Mol. Cell. Biol. 77, 154-165. Feinberg, A. P., and Vogelstein, B. (1964) . A technique for radiolabeling DNA restriction endonuclease fragments to high specific activity. Anal. Biochem. 732, 6-13. Field, J., Nikawa, J., Broek, D., MacDonald, B., Rogers, L., Wilson, I., Lerner, R., and Wigler, M. (1966) . Purification of a RAS-responsive adenylyl cyclase complex from Saccharomyces cerevisiae by use of an epitope addition method. Mol. Cell. Biol. 8, 2159 -2165 . Fitzgerald-Hayes, M.. Clarke, L., and Carbon, J. (1962 . Nucleotide sequence comparisons and functional analysis of yeast centromere DNAs. Cell 29,235-244. Funk, M., Hegemann, J., and Philippsen, P. (1969) . Mellor, J., Jiang, W., Funk, M., Rathjen, J., Barnes, C., Hiz, T., Hegemann, J., and Philippsen, P. (1990) . CPFl, a yeast protein which functions in centromeres and promoters. EMBO J. 9, 4017-4028. Mullis, K. B., and Faloona, F. A. (1987) . Specific synthesis of DNA in vitro via a polymerase catalysed chain reaction. Meth. Enzymol. 755, 335-350. Newlon, C. (1988) . Yeast chromosome replication and segregation. Microbial. Rev. 52, 588-801. Ng, R., and Carbon, J. (1987) . Mutational and in vitro protein-binding studies on centromere DNA from Saccharomyces cerevisiae. Mol. Cell, Biol. 7, 4522-4534.Palmer, R., Koval, M., and Koshland, D. (1989) . The dynamics of chromosome movement in the budding yeast Saccharomyces cerevisiae. J. Cell Biol. 109, 3355-3388. Perier, F., and Carbon, J. (1992) . A colony color assay for Saccharomyces cefevisiae mutants defective in kinetochore structure and function. Genetics 132, 39-51.Peterson, J., and Ris, H. (1978) . Electron microscope study of the spindle and chromosome movement in the yeast S. cerevisiae. J. Cell Sci. 22, 219-242.Pluta, A. F., Cooke, C. A., and Earnshaw, W. C. (1990) . Structure of the human centromere at metaphase. Trends Biochem. Sci. 75, 161-185.Pfuta, A. F., Saitoh, N., Goldberg, I., and Earnshaw, W. C. (1992). Identification of a subdomain of CENP-B that is necessary and sufficient for focalization to the human centromere.J. Cell Biol. 716,1081-1093. Rieder, C. L. (1982) . The formation, structure and composition of the mammalian kinetochore fiber. Int. Rev. Cytol. 79, l-88. Rieder, C. L., and Alexander, S. P. (1989) . The attachment of chromosomes to the mitotic spindle and the production of aneuploidy in newt lung Cells. In Mechanisms of Chromosome Distribution and Aneuploidy, M. Resnick and B. Vig, eds. (New York: Liss), pp. 185-194. Roof, D. M., Meluh, P. B., and Rose M. D. (1992) . Kinesin-related proteins required for assembly of the mitotic spindle. J. Cell Biol. 118, 95-108. Rose, M. D., Winston, F., and Hieter, P. (1990) . Methods in Yeast Genetics (Cold Spring Harbor, New York: Cold Spring Harbor Laboratory Pressj. Sanger, F., Nicklen, S., and Coulson, A. R. (1977) . DNA sequencing with chain-terminating inhibitors. Proc. Natl. Acad. Sci. USA 74,5483-5487. Saunders, W. S., and Hoyt, M. A. (1992) 

The accession number for the CTF73 sequence reported in this paper is L10083.