Mutations Synthetically Lethal with cep1 Target S. cerevisiae Kinetochore Components

Corresponding author: Richard E. Baker, Department of Molecular Genetics and Microbiology, University of Massachusetts Medical School, 55 Lake Avenue North, Worcester, MA 01655, rbaker{at}banyan.ummed.edu (E-mail).

Communicating editor: M. D. ROSE

	ABSTRACT

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

CP1 (encoded by CEP1) is a Saccharomyces cerevisiae chromatin protein that binds a DNA element conserved in centromeres and in the 5'-flanking DNA of methionine biosynthetic (MET) genes. Strains lacking CP1 are defective in chromosome segregation and MET gene transcription, leading to the hypothesis that CP1 plays a general role in assembling higher order chromatin structures at genomic sites where it is bound. A screen for mutations synthetically lethal with a cep1 null allele yielded five recessive csl (cep1 synthetic lethal) mutations, each defining a unique complementation group. Four of the five mutations synergistically increased the loss rate of marker chromosomes carrying a centromere lacking the CP1 binding site, suggesting that the cep1 synthetic lethality was due to chromosome segregation defects. Three of these four CSL genes were subsequently found to be known or imputed kinetochore genes: CEP3, NDC10, and CSE4. The fourth, CSL4, corresponded to ORF YNL232w on chromosome XIV, and was found to be essential. A human cDNA was identified that encoded a protein homologous to Csl4 and that complemented the csl4-1 mutation. The results are consistent with the view that the major cellular role of CP1 is to safeguard the biochemical integrity of the kinetochore.

CP1 (also known as Cbf1 or Cpf1), encoded by the gene CEP1, is an abundant, nonessential, sequence-specific DNA-binding protein of Saccharomyces cerevisiae (yeast) that recognizes the degenerate octanucleotide RTCACRTG (R = purine) (BAKER et al. 1989 ; BRAM and KORNBERG 1987 ; CAI and DAVIS 1989 ). CP1 binding sites are scattered more or less randomly throughout the yeast genome (BAKER et al. 1989 ) and are functionally important in two contexts. First, the site is present at the same position and orientation in all yeast centromeres where it is known as centromere DNA element I (CDEI) (HIETER et al. 1985 ). Neither CP1 nor the CDEI site are essential for centromere function; however, cep1 null mutants exhibit defects in chromosome segregation (BAKER and MASISON 1990 ; CAI and DAVIS 1990 ; MELLOR et al. 1990 ). Mitotic chromosome nondisjunction is increased 10- to 30-fold, and high rates of premature sister chromatid separation are observed at meiosis (MASISON and BAKER 1992 ). The quantitative effects of cis-acting CDEI mutations and trans-acting cep1 mutations are equal and nonadditive, indicating that the role of CP1 in chromosome segregation is executed at CDEI (BAKER and MASISON 1990 ). Second, CP1 binding sites are found near the promoters of genes of the methionine biosynthetic pathway (O'CONNELL and BAKER 1992 ; THOMAS et al. 1989 ). The best characterized example is MET16 where it has been shown that CP1 modulates two different activation pathways, and both CP1 and an intact CDEI site are required for full MET16 expression (O'CONNELL et al. 1995 ). (We refer to the RTCACRTG element as CDEI, regardless of context.) The cep1 null mutant is a methionine auxotroph (BAKER and MASISON 1990 ; CAI and DAVIS 1990 ; MELLOR et al. 1990 ), although the MET16 defect alone does not account for this phenotype (O'CONNELL et al. 1995 ). CP1, along with Met28 is required for the binding of Met4 to the MET16 UAS in vitro (KURAS et al. 1997 ). As Met4 is the positive transactivator for most if not all MET genes (THOMAS et al. 1992 ), cep1 methionine auxotrophy probably results from cumulative reductions in enzyme levels throughout the sulfate utilization pathway.

In addition to chromosome segregation and transcription defects, cep1 null mutants double more slowly than isogenic wild-type strains. The physiological basis for the 35% increase in generation time is not known, but measured rates of chromosome missegregation are insufficient to account for a growth defect of this magnitude (BAKER and MASISON 1990 ). The pleiotropic cep1 phenotype as well as the widespread distribution of CDEI sites and its cellular abundance have led to the notion that CP1 is a "general regulatory factor" whose function is to promote the assembly of higher order structures in yeast chromatin (BUCHMAN et al. 1988 ): at MET promoters, CP1 enhances assembly of the RNA polymerase II initiation complex, while at centromeres, CP1 facilitates kinetochore assembly.

The kinetochore is the centromere-associated organelle at which spindle microtubules attach to chromosomes during mitosis and meiosis. Besides providing the physical connection between spindle and chromosome, the kinetochore participates in checkpoint regulation of the cell cycle and receives the signal for sister chromatid disjunction upon anaphase onset (reviewed by PLUTA et al. 1995 ). The S. cerevisiae kinetochore is a multiprotein complex that assembles over the conserved 120-bp centromere (CEN) DNA sequence. In addition to CDEI described above, the yeast centromere contains two other defined sequence elements: CDEIII, a 25-bp partially palindromic sequence located at the right-hand end of the centromere; and CDEII, a 78- to 86-bp segment of highly A+T-rich (>87%) DNA (HIETER et al. 1985 ). CDEII acts as a spacer between CDEIII and CDEI, located at the left-hand end of the centromere. CDEIII is absolutely essential for centromere function; mutations of CDEI and CDEII are tolerated but can significantly compromise function (CUMBERLEDGE and CARBON 1987 ; GAUDET and FITZGERALD-HAYES 1987 ; HEGEMANN et al. 1988 ; MCGREW et al. 1986 ).

Yeast kinetochore proteins have been characterized by both biochemical and genetic means. CP1, the first centromere protein purified, was identified by virtue of its specific binding to CDEI (BAKER et al. 1989 ; CAI and DAVIS 1989 ). Cbf3 is a multiprotein complex that binds CDEIII (LECHNER and CARBON 1991 ). The Cbf3 complex consists of four subunits, p110 (NDC10/CBF2/CTF14), p64 (CBF3b,CEP3), p58 (CTF13), and p23 (SKP1) (CONNELLY and HIETER 1996 ; DOHENY et al. 1993 ; GOH and KILMARTIN 1993 ; JIANG et al. 1993 ; LECHNER 1994 ; STRUNNIKOV et al. 1995 ). Mif2p, an essential protein required for chromosome segregation, is presumed to be part of the kinetochore by virtue of observed genetic interactions with cep3, ndc10, and cep1 (MELUH and KOSHLAND 1995 ). Because it contains an "AT hook," Mif2p is presumed to interact with CDEII (BROWN 1995 ). Lastly, Cse4, a variant of histone H3 and a homolog of CENP-A, appears to play some role in kinetochore structure or assembly (STOLER et al. 1995 ).

To gain additional perspective on the biochemical function of CP1 at kinetochores and promoters and possibly to identify other pathways in which CP1 might play a critical role, we carried out a genetic screen for mutations that were synthetically lethal with cep1, i.e., mutations lethal in a cep1 genetic background but not in a wild-type background. Synthetic lethal gene interactions often define similarity or functional relatedness between gene products (BOTSTEIN 1988 ). Mutations synthetically lethal with cep1 (cep1 synthetic lethal, or csl) might identify gene products having functions related to, redundant with, or overlapping those of CP1. At least two different classes of csl mutations were anticipated. The first class were mutations in kinetochore components or kinetochore assembly factors. If a major role of CP1 is to facilitate kinetochore assembly, nonlethal mutations in other essential kinetochore components could become lethal in a cep1 background. Synthetic lethality would result from inefficient or failed kinetochore assembly and consequent catastrophic increases in chromosome loss. The second class of mutations would be related to the transcription function of CP1. If CP1 were to participate along with other transcription factors in the activation of one or more essential genes, mutations in the other factors could render transcription CP1-dependent. Synthetic lethality would result from the failure to express the essential gene(s). For both classes of mutation, we assumed that the Csl target would interact directly or indirectly with DNA-bound CP1 and, therefore, that the mutations could be distinguished by their sites of action, i.e., kinetochore-related csl mutations might exhibit genetic interactions with cis-acting centromere CDEI mutations, while transcriptional (or other) csl mutations would not.

Mutations giving rise to a Csl phenotype have already been described. FOREMAN and DAVIS 1996 screened for such mutations and identified a novel gene which they named CDP1 (Cbf1-dependent). Both cdp1-1 and cdp1 mutants exhibit defects in nuclear division and chromosome segregation, although the CDP1 gene product is probably not directly involved in kinetochore assembly. Additional cdp1 phenotypes include spindle abnormalities and altered sensitivities to benzimidazole compounds, suggesting that Cdp1 might interact directly with microtubules or tubulin (FOREMAN and DAVIS 1996 ). MELUH and KOSHLAND 1995 , in an analysis of genetic interactions between kinetochore genes, detected synthetic lethality between cep1 and mif2, ndc10, and cep3. Their findings confirm the prediction that kinetochore components will be targets of csl mutations.

This article describes the results of our independent Csl screen. Five csl complementation groups were identified and the corresponding genes cloned. Consistent with expectations, the csl mutations targeted both kinetochore and transcription components, but kinetochore mutations predominated. One of the Csl gene products, Csl4, is an uncharacterized yeast protein for which a human homolog exists, and genetic evidence is presented implicating Csl4 in centromere function. Overall, the results imply that CP1 is important for stabilizing kinetochore structure and are discussed in light of the model for the yeast kinetochore proposed by MELUH and KOSHLAND 1995 . Specifically, our findings confirm interaction between CP1 and Cbf3 and provide the first evidence suggesting interaction between CP1 and the presumed centromeric histone Cse4.

	MATERIALS AND METHODS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Strains, media, and plasmids:
The yeast strains used in this study are listed in Table 1. Strain construction and genetic analysis were carried out using standard methods (MORTIMER and HAWTHORNE 1969 ). All media were as described (BAKER and MASISON 1990 ). Color indicator plates were synthetic complete medium containing 6 µg/ml adenine (1/3 normal concentration). Cycloheximide plates contained 5 µg/ml cycloheximide. Unless noted otherwise, all strain growth was at 30°.

Plasmid pAP2 was derived from YEp352 (HILL et al. 1993 ) by inserting DNA fragments containing ade3-2p (SalI-SmaI) and CEP1 (BamHI-BglII). Plasmid pKM35 was derived from pRS328 (SIKORSKI and BOEKE 1991 ) by inserting the wild-type CEP1 gene (BamHI-BglII) at the unique BamHI site. Plasmid pRB267 was obtained by inserting the 1.3-kbp EcoRI-HindIII fragment carrying CSL4 into pRS306 (SIKORSKI and HIETER 1989 ). Additional 5'-flanking sequences were restored to the CSL4 gene in pRB267 using PCR to amplify DNA containing the 356-bp EcoRI fragment spanning positions -468 to -113 relative to the CSL4 initiator ATG and inserting this fragment into the EcoRI site of pRB267. The resulting plasmid, pRB289, has the EcoRI fragment in the native orientation. Plasmid pRB288 is identical to pRB289 except that the CDEI site located within the EcoRI fragment was mutated to GTCATTATG using bidirectional recombinant PCR. This mutation (CAT) was characterized previously (BAKER et al. 1989 ). PCR primers overlapping the CDEI site and containing the CAT mutation were used to amplify both halves of the EcoRI fragment in separate reactions with flanking primers. The two amplification products were gel-purified, mixed, and used as template in a third PCR reaction with only the flanking primers. The resulting product was cleaved with EcoRI and inserted into the EcoRI site of pRB267. The Expand High Fidelity PCR System (Boehringer Mannheim, Indianapolis) was used for the recombinant PCR reaction; otherwise, Taq polymerase (Boehringer Mannheim) was used.

To disrupt CSL4, the SphI-MluI fragment containing the CSL4 ORF (starting at codon 2 and including 30 bp of 3'-flanking DNA) was removed from pRB267 and replaced with the URA3 gene. The resulting plasmid, pRB274, was cleaved with EcoRI and HindIII and used to transform diploid strain R117. One of the Ura⁺ transformants (strain R211) was shown by Southern blot analysis to have one copy of CSL4 replaced by the csl4::URA3 deletion allele. To allow subsequent transformation with URA3 plasmids, the csl4-linked URA3 marker in R211 was changed to LEU2 using the marker change plasmid pDM2 (MASISON and BAKER 1992 ). The resulting strain (R229) retains ura3 sequences flanking LEU2 (csl4::ura3::LEU2).

Human Csl4 (hCsl4) was expressed in yeast using the cDNA expression vectors p416GPD and p426GPD (MUMBERG et al. 1995 ), which are single- and multicopy plasmids, respectively, containing the yeast glyceraldehyde 3-phosphate dehydrogenase promoter and terminator flanking a polylinker cloning site. The hCsl4 coding region and 16 bp of 3' untranslated sequences were obtained by PCR using I.M.A.G.E Consortium cDNA clone 511361 (Research Genetics, Inc., Birmingham, AL) as template and primers incorporating XbaI and XhoI restriction sites. The upstream primer also contained the sequence ATTATA immediately preceding the initiator ATG to serve as a ribosome binding site. This is the sequence found in yeast CSL4. The PCR product was cleaved with XbaI and XhoI and inserted between the SpeI and XhoI sites of p416GPD and p426GPD, generating plasmids pRB305 and pRB308, respectively. As controls, analogous expression plasmids containing the yeast CSL4 coding DNA were constructed in an identical fashion. The yeast CSL4 plasmids, pRB306 and pRB307, respectively, contained CSL4 nucleotides -6 to 1170 relative to the initiator ATG. The authors will gladly supply the sequences of all PCR primers upon request.

Csl screens:
Two screens were used to isolate csl mutants, one a visual screen based on colony color sectoring, the other based on cycloheximide sensitivity (Cyh^s). The first protocol, adapted from BENDER and PRINGLE 1991 , employed parental strains (D37-1A, D102-4A, or R95-1-1) of genotype MATa leu2 ura3 ade2 ade3 cep1 carrying an episomal plasmid (pAP2) bearing the wild-type CEP1 gene. The plasmid also contained the yeast ADE3 and LEU2 genes. The parent strains are phenotypically Cep1⁺ (Met⁺) Leu⁺ Ade3⁺ Ade2^- and form red colonies with multiple white sectors. The red color is the result of a block in adenine biosynthesis due to the ade2 mutation; the white sectors arise through mitotic loss of pAP2, uncovering the chromosomal ade3 mutation, which blocks formation of the red pigment. The csl mutants are unable to form sectored colonies, because plasmid loss also uncovers the cep1 allele and exposes the cep1 synthetic lethality. Cells were mutagenized with ethylmethane sulfonate to 30–50% survival and spread directly on indicator plates. Survivors were visually screened for a red, nonsectoring colony color phenotype (Sect^-). After verifying the Sect^- phenotype by restreaking, the candidate csl mutants were transformed with pKO56, an episomal plasmid carrying CEP1 and URA3 (but not ADE3). Mutants that regained the ability to sector (showing dependence on CEP1 but not LEU2 or ADE3) were backcrossed to strain D102-5D to test for dominance and for 2:2 segregation of the Sect^- phenotype.

As an alternative to the sectoring screen, a csl screen based on cycloheximide sensitivity was also used. In this case, the parent (D102-4A) contained the chromosomal cyh2^r mutation and carried a CEN plasmid (pKM35) bearing the wild-type CYH2 allele along with CEP1. Plasmid dependence (Sect^-) was scored by replicating mutagenized colonies onto cycloheximide plates. The csl mutants cannot give rise to Cyh^r papillae, which would normally arise from mitotic loss of pKM35. Except for scoring the Sect^- phenotype, other aspects of the Cyh^s csl screen were the same as for the color sectoring screen.

Results of the screens are summarized in Table 2. Each of the csl mutants was backcrossed at least three times to congenic parents. During this process, cslx segregants were obtained in both mating types and in CEP1 and ADE3 genetic backgrounds.

Identification of CSL genes:
Wild-type alleles of the csl genes were cloned by complementation. CSL3 was obtained by transforming a csl3-1 strain with the YEp24-based S. cerevisiae genomic library of CARLSON and BOTSTEIN 1982 and screening by replica plating for reversion of the Cyh^s phenotype. CSL2 was obtained by transforming a csl2-1 mutant with the YCp50-based S. cerevisiae library of ROSE (1987) and plating the transformants directly on indicator medium to score for reversion of Sect^-. It was found that the Cyh^s system was more efficient; therefore, the remaining CSL genes were cloned from the YCp50 library using the Cyh^s selection system after first introducing pKM35 into each mutant. In addition to csl-complementing DNAs, CEP1 clones were obtained in all cases. These were recognized by their Met⁺ phenotype. The csl-complementing plasmids were rescued into Escherichia coli for subsequent manipulation and analysis.

CSL1, CSL2, and CSL3 were obtained prior to completion of the S. cerevisiae genome sequencing project. The complementing DNAs were first localized within the genomic insert by subcloning and then partially sequenced. Matches to CSL1 and CSL2 were found in GenBank, accession numbers X81396 (CBF3b) and Z28049 (ORF YKL049c; Cse4), respectively. CSL3 was physically mapped by hybridizing the complementing DNA to ordered phage blots (ATCC 76269) and found to lie very near the RTF1 locus on chromosome VII. The sequence of RTF1 (GenBank accession no. U86702) was provided by KAREN ARNDT (University of Pittsburgh, PA) and found to be identical to CSL3. CSL4 and CSL5 were obtained after the yeast genome sequence was made public. Genomic inserts from the original CSL4- and CSL5-complementing library plasmids were partially sequenced and the data used to search the S. cerevisiae genome database. After identifying the chromosomal locations of the cloned DNAs, potential ORFs in the region were tested singly and in combination to determine which one complemented the csl mutation. The GenBank accession numbers for CSL4 (ORF YNL232w) and CSL5 (NDC10) are Z71508 and X69300, respectively.

The identity of the cslx-complementing ORFs was confirmed by introducing frameshift mutations and retesting complementation. The frameshift mutations were generated by cleaving the respective ORF-containing plasmids at unique restriction sites within the identified ORF and blunting the DNA ends with Klenow polymerase. After religation, the plasmids were transformed into the cslx strains to test for complementation. In all cases (csl1-1, csl2-1, csl3-1, csl5-1), the frameshift plasmids failed to complement. The restriction sites used to generate the mutations were as follows: CSL1, SacI (-4); CSL2, NdeI (+2); CSL3, BglII (+4); and CSL5, BglII (+4). (Numbers in parentheses indicate the number of base pairs added or deleted.) Assignment of CSL4 was confirmed by expressing YNL232w from a heterologous promoter and observing complementation of csl4-1 (plasmid pRB306, Figure 3).

View larger version (75K): In this window In a new window Download PPT slide   Figure 1. —Dissected tetrads of cslx diploids. Viable segregants were scored for Met prototrophy to determine cep1 genotype: csl1, 14/14 Met+; csl2, 18/18 Met+; csl3, 20/20 Met+ (tetrad #4 from left segregated Met 3+:1-); csl4, 19/19 Met+ (tetrad #3 from left segregated Met 3+:1-). To avoid the sporulation defect of csl5-1 homozygotes, csl5-1 synthetic lethality was analyzed using a csl5-1 cep1 double heterozygote, and csl5-1 was scored by its Ts phenotype. No Met- Ts- segregants were recovered (see text).

View larger version (47K): In this window In a new window Download PPT slide   Figure 2. —Sequences of the CSL4 gene product and 5' flanking DNA. (A) The translated sequence of the CSL4 ORF is displayed with the glutamate-rich motif underlined. Also shown is an alignment between Csl4 and its human homolog (hCsl4) made using the Genetics Computer Group (Madison, WI) GAP program. (B) The CSL4 5'-flanking DNA sequence is shown with the CDEI site boxed. The CDEII-like sequence is underlined and the proximal EcoRI site is overscored.

View larger version (61K): In this window In a new window Download PPT slide   Figure 3. —hCsl4 complements csl4-1. A cep1 csl4-1 cyh2r strain maintained by pKM35 (CEP1-CYH2) was transformed with hCsl4 (pRB305, pRB308) or CSL4 (pRB306, pRB307) expression vectors (single- and multicopy, respectively). Transformant colonies (Ura+) were picked, resuspended, and plated to test for cycloheximide resistance. Plasmids complementing csl4-1 allow the mitotic loss of pKM35 which is detected by the appearance of cycloheximide-resistant cells. Three independent transformants were tested for each plasmid, and "spots" of approximately 103 and 102 cells were plated in each case. Plasmid pRS306, which carries the selectable marker URA3 but no CSL4 sequences, was included as a control.

Restriction fragments from the cloned DNAs were inserted into the integrating vector pRS306 (SIKORSKI and HIETER 1989 ) and used to target URA3 to the corresponding chromosomal loci by homologous recombination. After confirming the integrations by Southern blotting, meiotic analysis was performed to test linkage between the integrated URA3 and the respective csl mutations. In all cases, direct linkage was observed, demonstrating that the cloned DNAs carried the respective CSL genes and not phenotypic suppressors. The restriction fragments used to target integration and the PD:T:NPD ratios for cslx-URA3 linkage were as follows: CSL1, 2-kbp BglII-ClaI, 7:0:0; CSL2, 2.7-kbp EcoRI-BglII, 13:0:0; CSL3, 2.7-kbp HindIII-SmaI (located 5 kbp downstream of CSL3), 9:1:0; CSL4, 1.3-kbp EcoRI-HindIII, 11:0:0; and CSL5, 2-kbp XbaI, 10:0:0.

Analysis of human Csl4 (hCsl4):
A BLAST search (ALTSCHUL et al. 1990 ) of the predicted yeast Csl4 amino acid sequence against the GenBank expressed sequence log (EST) database translated in all six reading frames (tblastn) was performed using the NCBI default parameters. The highest scoring homology (P = 1.1 x 10^-15) was to human cDNA clone zl84h09.r1 (accession no. AA088342). Several other EST homologies of lesser significance were also returned, all apparently the same or related cDNAs present in different libraries including colon, pregnant uterus, fetal heart, pineal gland, placenta, and parathyroid tumor. Fifteen EST sequences homologous to zl84h09.r1 were collected from the database and assembled by computer to generate a cDNA consensus sequence.

Chromosome loss assays:
Mitotic loss rates of nonessential chromosome fragments (CFs) were determined by fluctuation test as described (HEGEMANN et al. 1988 ). The CFs were derived from chromosome III and were generated in strain DJ147-1-2 by transformation with derivatives of plasmid pJS2 linearized with EcoRI (SHERO et al. 1991 ). The pJS2 derivatives used for the fragmentation had the CEN6 segment replaced with either wild-type CEN3 or CEN3 lacking the CDEI site (BAKER et al. 1989 ). The resulting CFs carry URA3 and SUP11 on their short arms and chromosome III sequences distal to D8B on their long arms. It was observed that ade2-1 haploid strains containing either CF (white) grew slightly faster than strains lacking the CF (red). This is probably related to the red pigment formation, because the opposite effect was observed in ade2-1 ade3 strains when the CF carried ADE3. In any event, the growth differential was not observed in diploid strains carrying only one copy of the CF (2N+1); therefore, 2N+1 diploids were used for all quantitative CF loss assays.

Determination of CSL mRNA levels:
Steady-state RNA levels were determined by northern blot analysis as described (O'CONNELL et al. 1995 ). Hybridization signals were quantitated using a Betascope Blot Analyzer or GS-525 Molecular Imager (Bio-Rad Laboratories, Hercules, CA). CSLX cpm are reported relative to the level of ACT1 RNA which was used as an internal standard to correct for differences in RNA recovery and gel loading. The following DNA fragments were labeled by random primed DNA synthesis and used as hybridization probes: CSL1, 690-bp PstI-DraI; CSL2, 640-bp NsiI-SphI; CSL3, 1050-bp SmaI-NsiI; CSL4, 1150-bp SphI-HindIII; CSL5, 1340-bp EcoRI; and ACT1, 300-bp BglII. The strains used for the analyses were as follows: wild-type, R203-8A; cep1, R95-1-1; csl1-1, R57-7B; csl2-1, R56-7B; csl3-1, R82-10D; csl4-1, BM3-40a; and csl5-1, BM3-64a. (The CSL2 probe was isolated from plasmid pRB190 in which the CSL2 gene had been modified to contain an NsiI site at the initiator ATG.)

	RESULTS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

cep1 synthetic lethal (csl) mutants:
Two screening systems were developed to isolate csl mutants; both utilized CEP1 plasmid dependence as the primary screening phenotype (see MATERIALS AND METHODS). Parental strains carried a disrupted or deleted chromosomal cep1 allele but harbored a plasmid containing wild-type CEP1. The csl mutants are plasmid-dependent, because plasmid loss uncovers the chromosomal cep1 mutation, exposing the cep1 synthetic lethality. A "plasmid shuffle" test was used to eliminate plasmid dependence due to sequences on the screening plasmid other than CEP1, i.e., candidate mutants were transformed with an unrelated CEP1 plasmid, and only mutants that regained the ability to segregate the original plasmid were characterized further. In three separate trials, 36,500 colonies were screened to obtain 22 candidate csl mutants that passed the shuffle test. Of those, five segregated the plasmid-dependent phenotype 2:2 when backcrossed to a cep1 parent, and these were designated csl1-csl5 (Table 2). All five csl mutations were recessive and each defined a unique complementation group.

The csl screens were based on CEP1 plasmid dependence, an indirect phenotype. Direct demonstration of the cep1 synthetic lethal phenotype was accomplished by tetrad analysis. Diploids of genotype CEP1/cep1 cslx/cslx yielded tetrads with two viable and two inviable spores (Figure 1). Inviability was genetically linked to cep1; CEP1 spores (Met⁺) were viable, cep1 spores (Met^-) were not. The csl5-1 homozygous diploids germinated very poorly (<1%); therefore, csl5-1 cep1 synthetic lethality was tested using cep1 csl5-1 double heterozygotes. In this case, 25% of spores are expected to cosegregate cep1 and csl5-1 and be inviable. This was the observed result. Assuming no linkage, two-thirds of the tetrads (tetratypes) will contain one csl5-1 cep1 spore, one-sixth (parental ditypes) will contain two csl5-1 cep1 spores, and one-sixth (nonparental ditypes) will contain no csl5-1 cep1 spores. The csl5-1 mutation causes a Ts phenotype, allowing it to be scored in the CEP1 background. No viable Ts^- Met^- (csl5-1 cep1) spores were recovered.

Microscopic examination revealed that most of the "inviable" spores were able to germinate and complete several cell divisions. In the case of csl1-1 and csl4-1, microcolonies became visible after 5–7 days, while csl2-1 and csl5-1 produced microcolonies of only 4–50 cells (not visible to the naked eye). The phenotype of csl3-1 was the most leaky, with microcolonies obvious after 3–4 days. No clear cell division cycle arrest phenotype was observed for any of the slow-growing cep1 cslx segregants. After backcrossing at least three times, strains containing each csl mutation in the CEP1⁺ background were tested for additional phenotypes. None of the mutations caused methionine auxotrophy or altered sensitivity to the microtubule inhibitor benomyl; however, csl5-1 strains were found to be temperature sensitive (Ts) for growth at 38°.

The cis-trans test:
As loss of CP1 leads to defects in both centromere function and gene transcription, the synthetic lethality of csl mutations could be due to defects in either of these essential genetic processes. To distinguish the subset of csl gene products that interact directly or indirectly with CP1 at centromeres, we tested the effect of a cis-acting CDEI mutation on chromosome stability in the csl mutants. In carrying out its CEN-related function, CP1 acts through the CDEI site (BAKER and MASISON 1990 ). If the synthetic lethality of a given csl were due to a gross defect in centromere function, e.g., the inability of the mutant to assemble a functional kinetochore in the absence of CP1, the same defect should exist for centromeres lacking the CDEI site even when CP1 is present. That is, with respect to an individual chromosome, the phenotypes of cslx cep1 (cep1 acting in trans) and cslx CDEI (CDEI acting in cis) should be equivalent, and centromere-related csl mutations should selectively and synergistically destabilize a CDEI-mutated chromosome, leading to a "synthetic acentric" phenotype (STRUNNIKOV et al. 1995 ).

The marker chromosome used for the cis-trans test was derived from chromosome III and carried either a wild-type or CDEI-deleted version of CEN3. Mitotic loss rates were determined for wild-type, cep1, and cslx strains (Table 3). The assays were carried out using diploids (2N+1) to avoid the slight growth advantage conferred by the marker chromosome in haploids (see MATERIALS AND METHODS). Loss rates were determined by fluctuation test. The phenotypic equivalence of cep1 and cen3CDEI mutations was confirmed for this marker chromosome by the results reported on the first three lines of Table 3. Deleting CDEI from the centromere caused a 5.8-fold increase in loss rate (R109 vs. R108), while deleting cep1 caused a not significantly different 6.8-fold increase (R100 vs. R108). Combining the cis and trans mutations led to no significant additional increase (R101 vs. R100). The csl mutations caused only moderate increases in the mitotic loss rate of the wild-type marker chromosome; the 1.7- to 8.4-fold increases were quantitatively similar to or less than the 6.8-fold effect of cep1. However, unlike cep1, all of the csl mutations except csl3-1 synergistically increased the loss rate of the cen3CDEI chromosome. Expressed relative to the wild-type rate (3.8 x 10^-4), the loss rate of the cen3CDEI chromosome in the csl1-1, csl2-1, csl4-1, and csl5-1 mutants was increased 89-, 166-, 46-, and 253-fold, respectively. In each case, the increase was significantly greater than the additive effect of the cslx and cen3CDEI mutations individually. The csl3-1 mutation behaved like cep1; the loss rate of the cen3CDEI chromosome (R111) was the same as that of the wild-type chromosome (R110). On the basis of these results, we concluded that the csl1-1, csl2-1, csl4-1, and csl5-1 synthetic lethalities probably resulted from kinetochore defects, while the synthetic lethality of csl3-1 appeared to be unrelated to centromeres.

View this table: In this window In a new window   Table 3. Mitotic loss rates of marker chromosomes containing either a wild-type (CEN3) or CDEI-deleted (cen3CDEI) centromere

Identification of csl genes:
The wild-type alleles corresponding to each of the csl mutations were cloned by complementation of the plasmid-dependent phenotype. In all cases, complementing activity was shown to be conferred by a single open reading frame (ORF), and genetic linkage between the complementing ORF and the original csl mutation was verified by targeted integration and tetrad analysis (see MATERIALS AND METHODS). The results are summarized in Table 4.

CSL3 is identical to RTF1, a recently characterized gene whose product appears to regulate transcription initiation by RNA polymerase II (STOLINSKI et al. 1997 ). Missense and loss of function rtf1 alleles suppress mutations in TATA-binding protein and affect TATA element utilization. Rtf1 appears to act directly or indirectly at many promoters in the genome and may exert its effect by modulating chromatin structure (STOLINSKI et al. 1997 ). The possible involvement of Rtf1/Csl3 in promoter recognition by RNA polymerase II is consistent with the conclusion that Csl3 does not act at centromeres and suggests that the Csl phenotype of csl3-1 is related to the transcription function of CP1.

The csl3-1 mutant is a methionine prototroph, indicating that MET gene transcription is not grossly defective. To test for a more subtle phenotype, genetic interaction between csl3-1 and a cis-acting MET16 promoter mutant was analyzed, i.e., the cis-trans test applied to an extracentromeric site of CP1 action. The met16-47 mutation inserts a TA dinucleotide at the center of the CP1 binding site in the MET16 promoter. As a result, the dissociation constant for CP1 binding is increased by 170-fold, steady-state MET16 RNA levels are reduced by 90%, and met16-47 strains grow slowly on media lacking methionine (Met^+/-) (O'CONNELL et al. 1995 ). Tetrads of a met16-47 csl3-1 double heterozygote yielded two Met⁺ spores and two Met^+/- spores (data not shown), indicating that csl3-1 neither enhanced nor suppressed the methionine bradytrophy of met16-47 (csl3-1 and met16-47 are unlinked). The same result was obtained with csl1-1 met16-47 and csl2-1 met16-47 double heterozygotes. If the synthetic lethal phenotype of csl3-1 is the result of perturbations in RNA polymerase II transcription, the defect apparently does not extend to the CP1-dependent gene MET16.

RTF1 (CSL3) is not an essential gene (STOLINSKI et al. 1997 ). To determine if the csl3-1 synthetic lethal phenotype was due to loss of Rtf1 function, the CSL3 gene was deleted in a diploid strain heterozygous for cep1 (the csl3 deletion was marked with URA3). Tetrads from the doubly heterozygous diploid yielded four viable spores, and Ura⁺ and Met⁺ phenotypes segregated 2:2 in all tetrads. Therefore, total loss of Rtf1 led to neither cep1 synthetic lethality nor methionine auxotrophy. Also, the chromosome loss phenotype of the csl3-1 mutant is not due to loss of Rtf1 function, as the mitotic loss rates of neither wild-type nor cen3CDEI marker chromosomes are significantly changed in the csl3 null mutant (Table 3). Thus, it appears that even though csl3-1 is recessive, the observed phenotypes result from altered Rtf1 function rather than loss of function.

CSL1 and CSL5 encode subunits of the CDEIII-binding complex Cbf3. CSL1 is identical to CEP3 (CBF3b) which encodes the 64-kD subunit (LECHNER 1994 ; STRUNNIKOV et al. 1995 ), while CSL5 is identical to NDC10 (CBF2, CEP2) encoding the 110-kD subunit (GOH and KILMARTIN 1993 ; JIANG et al. 1993 ). Significantly, these two genes were predicted to have centromere-specific functions on the basis of the cis-trans test. CSL2 is identical to CSE4, another essential gene associated with the centromere. CSE4 was first identified in a screen for mutations that enhanced the phenotypic effect of a cis-acting mutation in CDEII. Cse4 appears to be the S. cerevisiae homolog of the mammalian centromere protein CENP-A, and while biochemical evidence is lacking, the available genetic evidence supports the hypothesis that Cse4 is a centromere-specific histone (SMITH et al. 1996 ; STOLER et al. 1995 ). Thus, three of four Csl proteins predicted by the cis-trans test to have centromere-related functions are known or imputed kinetochore components. This supported the validity of the cis-trans test for distinguishing centromere-related csl targets and suggested that the fourth member of this class, CSL4, also encoded a gene product that acts at centromeres. The csl4-1 mutation had relatively little effect (less than 2-fold) on the loss rate of the wild-type marker chromosome; however, csl4-1 dramatically increased the loss rate of the cen3CDEI chromosome. In relative terms, the synergism between csl4-1 and cen3CDEI was similar to that observed for csl1-1 and csl5-1 (25-fold vs. 16-fold and 32-fold, respectively; Table 3).

CSL4 is an essential gene with a human homolog:
CSL4 corresponds to the uncharacterized S. cerevisiae ORF YNL232w located on the left arm of chromosome XIV. It potentially encodes a 292-amino-acid protein of molecular weight 32,000 (Figure 2A). A BLAST search against the GenBank protein databases produced numerous hits of low significance (0.1 < P < 0.9), with apparent homologies occurring mainly in a region of the hypothetical Csl4 that is rich in glutamic acid residues. When these regions were filtered (ALTSCHUL et al. 1994 ), no significant similarity between Csl4 and other proteins in the database was found. Nonetheless, the limited homology to other proteins with glutamic acid-rich regions may suggest that Csl4 is a chromatin protein, like many of the proteins in which such Glu-rich regions are found, e.g., AT motif binding factor (IDO et al. 1996 ), X-linked nuclear protein (GECZ et al. 1994 ), HMG protein (PENTECOST and DIXON 1984 ).

A BLAST search of the Csl4 ORF against the human EST database yielded several significant hits, all apparently to the same or closely related cDNAs. The most extensive homology was observed with EST zl84h09.r1 (GenBank AA088342) from the Stratagene (La Jolla, CA) human colon cDNA library (#937204). The homology to yeast ORF YNL232w (CSL4) had been noted when the EST was entered into the database. By collecting additional ESTs homologous to zl84h09.r1, a cDNA consensus sequence was assembled. The assembled sequence contained a good Kozak initiator ATG consensus sequence (KOZAK 1992 ) at the 5' end (AATCATGG) and encoded an open reading frame of 195 amino acids (M_r = 21,400) which was designated human Csl4 (hCsl4). The putative human protein is smaller than Csl4, but the two proteins share significant homology over their C-terminal regions (Figure 2A). In this 107-residue region, the proteins are 57% similar and 48% identical. The possible functional significance of this homology was assessed by testing whether hCsl4 could complement csl4-1. As shown in Figure 3, when expressed from either a single- or multicopy vector, hCsl4 rescued the synthetic lethal phenotype of csl4-1. The high copy hCsl4 vector was also able to partially complement the csl4 deletion mutation (data not shown).

The possibility that Csl4 might be a kinetochore component led us to determine if CSL4 was an essential gene. One copy of CSL4 was disrupted in a cep1 heterozygous diploid strain by deleting the presumptive CSL4 ORF and marking the deletion with either the URA3 or LEU2 gene. Tetrads obtained from these heterozygous diploids yielded two viable and two inviable spores (e.g., strain R229, Figure 4). The inviability was genetically linked to csl4 (no Ura⁺ or Leu⁺ prototrophs were recovered among 65 viable spores) and independent of cep1. Thus, CSL4 is an essential gene. To show that the observed lethality was complemented by CSL4, a URA3 plasmid carrying the 1.3-kbp genomic CSL4 EcoRI-HindIII fragment was integrated at the endogenous ura3 locus in strain R229. This CSL4 subclone complemented csl4-1 (not shown); therefore, it was somewhat surprising that the same fragment was unable to completely rescue csl4::LEU2. The csl4::LEU2 spores cosegregating the integrated plasmid (Leu⁺ Ura⁺) grew very slowly compared to the wild-type spores (Figure 4, strain R226), and the slow growth phenotype was independent of cep1 (3/8 Leu⁺ Ura⁺ spores were Met^-). The most likely explanation for the poor complementation was that the 1.3-kbp EcoRI-HindIII fragment lacked 5'-flanking sequences required for full CSL4 expression.

View larger version (123K): In this window In a new window Download PPT slide   Figure 4. —Effect of CSL4 gene disruption and dispensability of the upstream CDEI site. Shown are dissected tetrads of diploid strain R229 and its derivatives. R229 is heterozygous for csl4::LEU2. Strain R226 was obtained from R229 by integrating plasmid pRB267 (carrying the 1.3-kbp EcoRI-HindIII fragment of CSL4) at the ura3 locus on chromosome V. In strains R231 and R230, the integrated CSL4 alleles contained additional 5' sequences, either with (pRB289) or without (pRB288) the intact CDEI site, respectively. In the case of R230 and R231, the CSL4 alleles integrated at ura3 sequences flanking the LEU2 gene at csl4::LEU2 (see MATERIALS AND METHODS); therefore, only 4+:0- segregation is observed. The CSL4 locus from two independent Ura+ Leu+ segregants of R230 was amplified by PCR and subjected directly to DNA sequencing. The results confirmed that the CAT mutation was present.

Examination of the CSL4 5'-flanking DNA upstream of the proximal EcoRI site revealed sequences resembling an S. cerevisiae centromere. A perfect CDEI site is present at position -171 (relative to the initiator ATG) adjacent to a stretch of AT-rich DNA (Figure 2B). Although shorter in length than CDEII elements found at centromeres, the nucleotide composition of this 44-bp segment (86% AT) is similar. Since poly (dA:dT) is known to be a ubiquitous transcriptional activation sequence in yeast (IYER and STRUHL 1995 ), it seemed plausible that the CDEII-like element provides this function for CSL4, explaining the failure of the 1.3-kbp EcoRI-HindIII fragment to provide wild-type CSL4 function. To test this hypothesis, the AT-rich element, either with or without an intact CDEI site, was restored to the 1.3-kbp CSL4 EcoRI/HindIII fragment and the resulting constructs (pRB289 and pRB288, respectively) were reintegrated at the csl4 locus to test their ability to provide Csl4 function. In pRB288, the CDEI site was disabled by changing the CA dinucleotide at position 5/6 to TTA. This mutation (CAT) increases the dissociation constant for CP1 binding by 1400-fold and inactivates CDEI with respect to centromere function (BAKER et al. 1989 ). Both constructs complemented csl4::LEU2 (Figure 4; strains R230, R231), indicating that full expression of CSL4 requires sequences upstream of -113, but not a functional CDEI element.

CSL mRNA levels are independent of CP1:
A trivial explanation for the synthetic lethality of csl1-1, csl2-1, csl4-1, and csl5-1 would be that these mutations render transcription of the respective essential genes CP1-dependent. (This model does not explain csl3-1, because CSL3 is not an essential gene.) Such a model is inconsistent with the results of the cis-trans test, which argue that the Csl gene products interact with CP1 at the level of CEN DNA; however, CP1-dependent expression could contribute to the phenotypes observed. For example, if full expression of an essential kinetochore gene (CSLX) required CP1, a partial loss of function mutation (cslx) might be synthetically lethal in a cep1 genetic background due to reduced expression of the mutant Cslx gene product. In the CEP1⁺ background, the hypomorph would be viable but especially sensitive to cis-acting CEN mutations. To determine if CSL gene transcription was dependent on CP1, steady-state CSL RNA levels were measured in wild-type and cep1 strains. The results (Figure 5) showed no significant differences. In fact, except for CSL1, RNA levels were actually higher in cep1 cells. Also, the csl mutations themselves did not dramatically alter CSL RNA levels; increases or decreases of more than 50% were not observed. In sum, these results offer no evidence that CSL gene expression is defective in cep1 strains or altered in csl mutants.

View larger version (42K): In this window In a new window Download PPT slide   Figure 5. —Determination of CSLX RNA levels. Steady state levels of CSL RNAs were determined as described in MATERIALS AND METHODS and are expressed relative to levels found in the wild-type strain.

	DISCUSSION

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

The Csl screen is an efficient genetic method for identifying genes involved in centromere function. Of the five csl genes obtained so far, two encode Cbf3 subunits, while a third encodes Cse4, presumed to be an essential centromere histone. As only a single allele was obtained for each of the five csl complementation groups, the screen is not saturated. Also, no alleles of mif2 or cdp1 were obtained, and both of these genes are known to yield mutations conferring a Csl phenotype (FOREMAN and DAVIS 1996 ; MELUH and KOSHLAND 1995 ). Continued screening should yield additional Csl complementation groups, at least some of which may identify new kinetochore-related gene products. In this regard, the cis-trans test proved to be an effective means of classifying the csl mutants. All three of the csl mutations targeting known or imputed kinetochore components (csl1-1, csl2-1, csl5-1) showed synergism in the cis-trans test. On this basis, we propose that a fourth csl gene product, Csl4, also has a kinetochore-related function. Like Csl1/Cep3, Csl2/Cse4, and Csl5/Ndc10, Csl4 is essential for cell viability.

Obtaining an allele of rtf1 (csl3-1) in the Csl screen is further evidence that the function of CP1 is not restricted to centromeres. Rtf1 apparently acts at gene promoters to influence transcription site selection by RNA polymerase II (STOLINSKI et al. 1997 ). While the csl3-1 mutation does increase chromosome loss, the lack of interaction between csl3-1 and cen3CDEI implies that the cep1 synthetic lethality of csl3-1 is not the direct result of kinetochore defects. The csl3 null mutant has no phenotype that we detect, again arguing that Rtf1 plays no positive role in centromere function. Our favored explanation for synthetic lethality of csl3-1 is that pol II transcription is lethally misregulated by the mutant Rtf1 protein in the absence of CP1. The number and identity of the promoters involved is unknown. The chromosome loss phenotype of csl3-1 may be the cumulative result of multiple subtle defects in chromosome structure introduced by the mutant Rtf1 protein (possibly at all pol II promoters) and, in effect, be independent of kinetochore function. In this case, the presence or absence of CDEI at the centromere would be expected to have at most an additive effect.

The very high mitotic loss rates of the cen3CDEI marker chromosome in csl1-1, csl2-1, csl4-1, and csl5-1 mutants lead us to conclude that the Csl phenotype of these strains is due directly to chromosome missegregation. In absolute terms, loss rates of the CP1-deficient model chromosome ranged from 1.8 to 9.6%. Assuming that these csl mutations affect all 16 endogenous chromosomes to similar extents, loss rates of this magnitude would extrapolate to at least one loss event in a high proportion of cep1 mitoses (29–100%). An alternative explanation for the synthetic lethality of these mutations, that the expression of these essential genes is rendered CP1-dependent, is inconsistent with the finding that CSL RNA levels vary by less than twofold between wild-type, cep1, and csl backgrounds. Possible transcriptional regulation by CP1 is especially relevant to CSL4, which contains a CDEI site in its 5'-flanking DNA. Although this region of upstream sequence is apparently required for full expression of CSL4, the CDEI site is not. CSL4 carrying an inactivating mutation in the upstream CDEI site complements the csl4 null allele and actually produces higher than wild-type levels of CSL4 RNA (data not shown). The juxtaposition of a CDEI site next to the AT-rich element in the CSL4 promoter is interesting insofar as it creates a CEN-like sequence; however, at present, there is no evidence to suggest any functional relevance.

MELUH and KOSHLAND 1995 have proposed a model for the yeast centromere-kinetochore complex invoking multiple cooperative interactions between proteins bound at CDEI and CDEIII. Our findings reinforce and extend this model. The new alleles csl1-1 and csl5-1 confirm the proposed interactions between CP1 and Cbf3 subunits. More significantly, the synthetic lethality detected between csl2-1 and cep1 extends the network of genetic interactions to include Cse4. Genetic evidence suggests that Cse4 interacts with CDEII (STOLER et al. 1995 ). That an allele of cse4 (csl2-1) exhibits strong synthetic interactions with both cep1 and cen3CDEI can be interpreted to mean that Cse4 also interacts with CP1/CDEI. If so, a complete network of protein-protein and protein-DNA interactions would exist to interconnect CDEI, CDEII, CDEIII, and the proteins associated with them. Cse4 is a variant of histone H3 and the S. cerevisiae homolog of the mammalian centromere protein CENP-A (STOLER et al. 1995 ). SMITH et al. 1996 isolated CSE4 as a high copy suppressor of hhf1-20, a Ts lethal mutation in histone H4 that leads to G₂-M arrest at the restrictive temperature. This finding has been interpreted to suggest that Cse4 and H4 interact and that this interaction is essential for cell cycle progression through mitosis. The observed genetic interactions between cse4 alleles and cis-acting centromere mutations (cen3-130, cen3CDEI) argue that Cse4-H4 acts at the centromere, perhaps taking the form of a novel nucleosome required for wrapping the AT-rich CDEII sequence. We note that BLOOM et al. 1989 proposed some years ago that a modified nucleosome might serve as the structural foundation for the yeast kinetochore. CP1 might provide the targeting signal for a CEN-specific Cse4 histone complex (nucleosome?).

Histones generally impede the interaction of DNA binding proteins; however, recent findings show that many transcription factors are capable of binding to their cognate sites on the surface of a nucleosome (OWEN-HUGHES and WORKMAN 1994 ). Therefore, it is not unreasonable to propose that CEN DNA be packaged in a nucleosome with CDEI and CDEIII accessible for binding CP1 and Cbf3. In fact, assuming that the Cse4-modified nucleosome retained the approximate molecular dimensions of a conventional nucleosome, CDEI and CDEIII, separated by an average of 82 bp, would be brought adjacent to each other on the surface of the nucleosome, allowing direct interaction between CP1 and Cbf3. Cbf3 appears to be the kinetochore component responsible for binding microtubules (SORGER et al. 1994 ); therefore, the mechanical forces transferred to the DNA fiber by kinetochore microtubules during metaphase and anaphase must be equally opposed by the sum of forces binding Cbf3 to the CEN DNA. A configuration, in which CEN DNA is wrapped around a specialized nucleosome with CDEI and CDEIII linked through CP1-Cbf3 interaction, would allow that force to be distributed over a full turn of nucleosomal DNA. In addition to the energy of sequence-specific Cbf3-CDEIII and CP1-CDEI binding, multiple histone-DNA interactions would be brought to bear in maintaining kinetochore attachment to the centromere DNA.

If a unique nucleosomal substrate containing Cse4 is indeed present at yeast centromeres, we would expect that specialized chromatin proteins are needed to position, assemble, and/or maintain the centromere-specific structure. Mif2 and Csl4 are candidates for such roles. Both proteins possess amino acid sequence characteristics of chromatin components. Mif2 has a highly acidic domain and an AT hook motif found in chromatin proteins that bind AT-rich DNA (BROWN 1995 ), the latter attribute leading MELUH and KOSHLAND 1995 to place Mif2 at CDEII in their kinetochore model. Protein database homology searches yield little insight to the function of Csl4. The most tangible clue is a 25-amino-acid region of charged residues, mostly glutamate (40%). Such motifs are common in transcription factors and chromatin proteins, but they are by no means restricted to this class of proteins. Csl4, like Mif2, is an essential gene product, and we interpret the interaction between csl4-1 and cen3CDEI to mean that Csl4 acts at the level of CEN DNA. It may be a bona fide component of the kinetochore, or Csl4 may be a kinetochore assembly factor. The analysis of conditional alleles of csl4 is in progress and should be informative in elucidating Csl4 function. Csl4 joins Cse4 and Mif2 as likely yeast centromere proteins having structural mammalian homologs. As mentioned above, Cse4 is homologous to CENP-A, while Mif2 is homologous to the mammalian centromere protein CENP-C (BROWN 1995 ), a component of the inner kinetochore plate (PLUTA et al. 1995 ). It is tempting to speculate that hCsl4 might also be a centromere protein. As the hCsl4 cDNA complements csl4-1, yeast and human Csl4 share at least one function in common.

The proposal that CP1 forms a network of interactions with centromere proteins bound at CDEII and CDEIII explains how CP1 could facilitate kinetochore assembly and confer stability to the complex; however, this CP1-centric view of the kinetochore is seemingly at odds with the fact that CP1 is essential for neither kinetochore assembly nor function. What is the selective advantage for maintaining CP1 at the centromere? First, CP1 decreases the rate of mitotic chromosome loss ~10-fold (BAKER and MASISON 1990 ). As chromosome loss is lethal in haploids and aneuploidy deleterious to diploids, the improved accuracy of chromosome transmission would be strongly selected. Second, any factor that promotes the timely assembly and/or maintenance of the kinetochore complex indirectly reduces cell generation times by eliminating potential cell cycle delays. Kinetochore integrity is monitored by the spindle assembly cell cycle checkpoint (WANG and BURKE 1995 ). A single defective centromere is sufficient to trigger the checkpoint and induce a cell cycle delay at G₂-M (SPENCER and HIETER 1992 ; WELLS and MURRAY 1996 ), implying that kinetochore assembly is or can readily become rate-limiting for cell cycle progression through mitosis. Centromeres lacking CDEI-bound CP1 to orchestrate kinetochore assembly might lag in establishing spindle attachment, trip the checkpoint, and delay mitosis. This hypothesis would explain the lengthened cell generation times of cep1 mutants. It also predicts that the spindle integrity checkpoint is constitutively activated in cep1 mutants and that cell cycle times would decrease in cep1 double mutants defective in checkpoint regulation (e.g., mad, bub). Of course, the cost of abrogating spindle checkpoint control in cep1 strains would be increased segregation errors and a consequent decrease in cell viability.

The only known process for which CP1 is absolutely essential is methionine biosynthesis, where CP1 is involved in the transcriptional activation of MET genes. Is CP1 a transcription factor whose function was co-opted to centromeres or vice versa? The results of the Csl screen focus attention to the centromere. That four of five csl mutations target known or imputed kinetochore proteins implies, in our opinion, that the major biological role of CP1 is to safeguard the biochemical integrity of the kinetochore. When CP1 is bound at CDEI, defects in kinetochore structure or function are tolerated that otherwise would be lethal.

	ACKNOWLEDGMENTS

The authors thank MOLLY FITZGERALD-HAYES for many helpful discussions and KAREN ARNDT for communicating results prior to publication. We also thank JOSEPH CONNOLLY for technical assistance and, especially, GÜL BÜKÜOLU for performing the initial trial of the Csl screen. This work was supported by a grant to R.B. from the National Science Foundation (MCB-9406050).

Manuscript received November 14, 1997; Accepted for publication February 13, 1998.

	LITERATURE CITED

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

ALTSCHUL, S. F., M. S. BOGUSKI, W. GISH, and J. S. WOOTTON, 1994  Issues in searching molecular sequence databases. Nat. Genet. 6:119-129[Medline].

ALTSCHUL, S. F., W. GISH, W. MILLER, E. W. MYERS, and D. J. LIPMAN, 1990  Basic local alignment search tool. J. Mol. Biol. 215:403-410[Medline].

BAKER, R. E., M. FITZGERALD-HAYES, and T. C. O'BRIEN, 1989  Purification of the yeast centromere binding protein CP1 and a mutational analysis of its binding site. J. Biol. Chem. 264:10843-10850[Abstract/Free Full Text].

BAKER, R. E. and D. C. MASISON, 1990  Isolation of the gene encoding the Saccharomyces cerevisiae centromere-binding protein CP1. Mol. Cell. Biol. 10:2458-2467[Abstract/Free Full Text].

BASRAI, M. A. and P. HIETER, 1995  Is there a unique form of chromatin at the Saccharomyces cerevisiae centromere? Bioessays 17:669-672[Medline].

BENDER, A. and J. R. PRINGLE, 1991  Use of a screen for syntheric lethal and multicopy suppressee mutants to identify two genes involved in morphogenesis in Saccharomyces cerevisiae.. Mol. Cell. Biol. 11:1295-1305[Abstract/Free Full Text].

BLOOM, K., A. HILL, M. KENNA, and M. SAUNDERS, 1989  The structure of a primitive kinetochore. Trends Biochem. Sci. 14:223-227[Medline].

BOTSTEIN, D., 1988  Why study the cytoskeleton in yeast? Harvey Lec. 82:157-167.

BRAM, R. J. and R. D. KORNBERG, 1987  Isolation of a Saccharomyces cerevisiae centromere DNA-binding protein, its human homolog, and its possible role as a transcription factor. Mol. Cell. Biol. 7:403-409[Abstract/Free Full Text].

BROWN, M. T., 1995  Sequence similarities between the yeast chromosome segregation protein Mif2 and the mammalian centromere protein CENP-C. Gene 160:111-116[Medline].

BUCHMAN, A. R., W. J. KIMMERLY, J. RINE, and R. D. KORNBERG, 1988  Two DNA-binding factors recognize specific sequences at silencers, upstream activating sequences, autonomously replicating sequences, and telomeres in Saccharomyces cerevisiae.. Mol. Cell. Biol. 8:210-225[Abstract/Free Full Text].

CAI, M. and R. W. DAVIS, 1989  Purification of a yeast centromere-binding protein that is able to distinguish single base-pair mutations in its recognition site. Mol. Cell. Biol. 9:2544-2550[Abstract/Free Full Text].

CAI, M. and R. W. DAVIS, 1990  Yeast centromere binding protein CBF1, of the helix-loop-helix protein family, is required for chromosome stability and methionine prototrophy. Cell 61:437-446[Medline].

CARLSON, M. and D. BOTSTEIN, 1982  Two differentially regulated mRNAs with different 5' ends encode secreted and intracellular forms of yeast invertase. Cell 28:145-154[Medline].

CONNELLY, C. and P. HIETER, 1996  Budding yeast skp1 encodes an evolutionarily conserved kinetochore protein required for cell cycle progression. Cell 86:275-285[Medline].

CUMBERLEDGE, S. and J. CARBON, 1987  Mutational analysis of meiotic and mitotic centromere function in Saccharomyces cerevisiae.. Genetics 117:203-212[Abstract/Free Full Text].

DOHENY, K. F., P. K. SORGER, A. A. HYMAN, S. TUGENDREICH, and F. SPENCER et al., 1993  Identification of essential components of the S. cervisiae kinetochore. Cell 73:761-774[Medline].

FOREMAN, P. K. and R. W. DAVIS, 1996  CDP1, a novel Saccharomyces cerevisiae gene required for proper nuclear division and chromosome segregation. Genetics 144:1387-1397[Abstract].

GAUDET, A. and M. FITZGERALD-HAYES, 1987  Alterations in the adenine-plus-thymine-rich region of CEN3 affect centromere function in Saccharomyces cerevisiae.. Mol. Cell. Biol. 7:68-75[Abstract/Free Full Text].

GECZ, J., H. POLLARD, G. CONSALEZ, L. VILLARD, and C. STAYTON et al., 1994  Cloning and expression of the murine homologue of a putative human X-linked nuclear protein gene closely linked to PGK1 in Xq13.3. Hum. Mol. Genet. 3:39-44[Abstract/Free Full Text].

GOH, P.-Y. and J. V. KILMARTIN, 1993  NDC10: a gene involved in chromosome segregation in Saccharomyces cerevisiae.. J. Cell Biol. 121:503-512[Abstract/Free Full Text].

HARTWELL, L. H., 1980  Mutants of Saccharomyces cerevisiae unresponsive to cell division control by polypeptide mating hormone. J. Cell Biol. 85:811-822[Abstract/Free Full Text].

HEGEMANN, J. H., J. H. SHERO, G. COTTAREL, P. PHILIPPSEN, and P. HIETER, 1988  Mutational analysis of centromere DNA from chromosome VI of Saccharomyce cerevisiae.. Mol. Cell. Biol. 8:2523-2535[Abstract/Free Full Text].

HIETER, P., D. PRIDMORE, J. H. HEGEMANN, M. THOMAS, and R. W. DAVIS et al., 1985  Functional selection and analysis of yeast centromeric DNA. Cell 42:913-921[Medline].

HILL, J. E., A. M. MYERS, T. J. KOERNER, and A. TZAGOLOFF, 1993  Yeast/E. coli shuttle vectors with multiple unique restriction sites. Yeast 2:163-167.

IDO, A., Y. MIURA, M. WATANABE, M. SAKAI, and Y. INOUE et al., 1996  Cloning of the cDNA encoding the mouse ATBF1 transcription factor. Gene 168:227-231[Medline].

IYER, V. and K. STRUHL, 1995  Poly(dA:dT), a ubiquitous promoter element that stimulates transcription via its intrinsic DNA structure. EMBO J. 14:2570-2579[Medline].

JIANG, W., J. LECHNER, and J. CARBON, 1993  Isolation and characterization of a gene (CBF2) specifying a protein component of the budding yeast kinetochore. J. Cell Biol. 121:513-519[Abstract/Free Full Text].

KOZAK, M., 1992  Regulation of translation in eukaryotic systems. Annu. Rev. Cell Biol. 8:197-225.

KURAS, L., R. BARBEY, and D. THOMAS, 1997  Assembly of a bZIP-bHLH transcription activation complex: formation of the yeast Cbf1-Met4-Met28 complex is regulated through Met28 stimulation of Cbf1 DNA binding. EMBO J. 16:2441-2451[Medline].

LECHNER, J., 1994  A zinc finger protein, essential for chromosome segregation, constitutes a putative DNA binding subunit of the Saccharomyces cerevisiae kinetochore complex, Cbf3. EMBO J. 13:5203-5211[Medline].

LECHNER, J. and J. CARBON, 1991  A 240 kd multisubunit protein complex, CBF3, is a major component of the budding yeast centromere. Cell 64:717-725[Medline].

MASISON, D. and R. E. BAKER, 1992  Meiosis in Saccharomyces cerevisiae mutants lacking the centromere binding protein CP1. Genetics 131:43-53[Abstract].

MCGREW, J., B. DIEHL, and M. FITZGERALD-HAYES, 1986  Single base-pair mutations in centromere element III cause aberrant chromosome segregation in Saccharomyces cerevisiae.. Mol. Cell. Biol. 6:530-538[Abstract/Free Full Text].

MELLOR, J., W. JIANG, M. FUNK, J. RATHJAN, and C. A. BARNES et al., 1990  CPF1, a yeast protein which functions in centromeres and promoters. EMBO J. 8:4017-4026.

MELUH, P. B. and D. KOSHLAND, 1995  Evidence that the MIF2 gene of Saccharomyces cerevisiae encodes a centromere protein with homology to the mammalian centromere protein CENP-C. Mol. Biol. Cell 6:793-807[Abstract].

MORTIMER, R. K., and D. C. HAWTHORNE, 1969 Yeast Genetics, pp. 385–460 in The Yeasts, edited by A. H. ROSE and J. S. HARRISON. Academic Press, New York.

MUMBERG, D., R. MULLER, and M. FUNK, 1995  Yeast vectors for the controlled expression of heterologous proteins in different genetic backgrounds. Gene 156:119-122[Medline].

O'CONNELL, K. F. and R. E. BAKER, 1992  Possible cross-regulation of phosphate and sulfate metabolism in Saccharomyces cerevisiae.. Genetics 132:63-73[Abstract].

O'CONNELL, K. F., Y. SURDIN-KERJAN, and R. E. BAKER, 1995  Role of the Saccharomyces cerevisiae general regulatory factor CP1 in methionine biosynthetic gene transcription. Mol. Cell. Biol. 15:1879-1888[Abstract].

OWEN-HUGHES, T. A. and J. L. WORKMAN, 1994  Experimental analysis of chromatin function in transcription control. Crit. Rev. Eukaryotic Gene Expression 4:403-441[Medline].

PENTECOST, B. T. and G. H. DIXON, 1984  Isolation and partial sequence of bovine cDNA clones for the high-mobility-group protein (HMG-1). Biosci. Rep. 4:49-57[Medline].

PLUTA, A. F., A. M. MACKAY, A. M. AINSZTEIN, I. G. GOLDBERG, and W. C. EARNSHAW, 1995  The centromere: hub of chromosomal activities. Science 270:1591-1594[Abstract/Free Full Text].

ROSE, M. D., P. NOVIK, J. H. THOMAS, D. BOTSTEIN, and G. R. FINK, 1987  A Saccharomyces cerevisiae genomic plasmid bank based on a centromere-containing vector. Gene 60:237-243[Medline].

SHERO, J. H., M. KOVAL, F. SPENCER, R. E. PALMER, and P. HIETER et al., 1991  Analysis of chromosome segregation in Saccharomyces cerevisiae.. Meth. Enzymol. 194:749-773[Medline].

SIKORSKI, R. S. and J. BOEKE, 1991  In vitro mutagenesis and plasmid shuffling: from cloned gene to mutant gene. Meth. Enzymol. 194:302-318[Medline].

SIKORSKI, R. S. and P. HIETER, 1989  A system of shuttle vectors and yeast host strains designed for efficient manipulation of DNA in Saccharomyces cerevisiae.. Genetics 122:19-27[Abstract/Free Full Text].

SMITH, M. M., P. YANG, M. S. SANTISTEBAN, P. W. BOONE, and A. T. GOLDSTEIN et al., 1996  A novel histone H4 mutant defective in nuclear division and mitotic chromosome transmission. Mol. Cell. Biol. 16:1017-1026[Abstract].

SORGER, P. K., F. F. SEVERIN, and A. A. HYMAN, 1994  Factors required for the binding of reassembled yeast kinetochores to microtubules in vitro.. J. Cell Biol. 127:995-1008[Abstract/Free Full Text].

SPENCER, F. and P. HIETER, 1992  Centromere mutations induce a mitotic delay in Saccharomyces cerevisiae.. Proc. Natl. Acad. Sci. USA 89:8908-8912[Abstract/Free Full Text].

STOLER, S., K. C. KEITH, K. E. CURNICK, and M. FITZGERALD-HAYES, 1995  A mutation in CSE4, and essential gene encoding a novel chromatin-associated protein in yeast, causes chromosome nondisjunction and cell cycle arrest at mitosis. Genes Dev. 9:573-586[Abstract/Free Full Text].

STOLINSKI, L. A., D. M. EISENMANN, and K. M. ARNDT, 1997  Identification of RTF1, a novel gene important for TATA site selection by TATA box-binding protein in Saccharomyces cerevisiae.. Mol. Cell. Biol. 17:4490-4500[Abstract].

STRUNNIKOV, A. V., J. KINGSBURY, and D. KOSHLAND, 1995  CEP3 encodes a centromere protein of S. cerevisiae.. J. Cell Biol. 128:749-760[Abstract/Free Full Text].

THOMAS, D., H. CHEREST, and Y. SURDIN-KERJAN, 1989  Elements involved in S-adenosylmethionine-mediated regulation of the Saccharomyces cerevisiae MET25 gene. Mol. Cell. Biol. 9:3292-3298[Abstract/Free Full Text].

THOMAS, D., I. JACQUEMIN, and Y. SURDIN-KERJAN, 1992  MET4, a leucine zipper protein, and centromere-binding factor 1 are both required for transcriptional activation of sulfur metabolism in Saccharomyces cerevisiae.. Mol. Cell. Biol. 12:1719-1727[Abstract/Free Full Text].

WANG, Y. and D. J. BURKE, 1995  Checkpoint genes required to delay cell division in response to nocodazole respond to impaired kinetochore function in the yeast Saccharomyces cerevisiae.. Mol. Cell. Biol. 15:6838-6844[Abstract].

WELLS, W. A. and A. W. MURRAY, 1996  Aberrantly segregating centromeres activate the spindle assmbly checkpoint in budding yeast. J. Cell Biol. 133:75-84[Abstract/Free Full Text].

This article has been cited by other articles:

				A. C. Graham, D. L. Kiss, and E. D. Andrulis Core Exosome-independent Roles for Rrp6 in Cell Cycle Progression Mol. Biol. Cell, April 15, 2009; 20(8): 2242 - 2253. [Abstract] [Full Text] [PDF]

				W.-C. Au, M. J. Crisp, S. Z. DeLuca, O. J. Rando, and M. A. Basrai Altered Dosage and Mislocalization of Histone H3 and Cse4p Lead to Chromosome Loss in Saccharomyces cerevisiae Genetics, May 1, 2008; 179(1): 263 - 275. [Abstract] [Full Text] [PDF]

				R. E. Baker and K. Rogers Genetic and Genomic Analysis of the AT-Rich Centromere DNA Element II of Saccharomyces cerevisiae Genetics, December 1, 2005; 171(4): 1463 - 1475. [Abstract] [Full Text] [PDF]

				L. Morey, K. Barnes, Y. Chen, M. Fitzgerald-Hayes, and R. E. Baker The Histone Fold Domain of Cse4 Is Sufficient for CEN Targeting and Propagation of Active Centromeres in Budding Yeast Eukaryot. Cell, December 1, 2004; 3(6): 1533 - 1543. [Abstract] [Full Text] [PDF]

				N. J. Krogan, K. Baetz, M.-C. Keogh, N. Datta, C. Sawa, T. C. Y. Kwok, N. J. Thompson, M. G. Davey, J. Pootoolal, T. R. Hughes, et al. Regulation of chromosome stability by the histone H2A variant Htz1, the Swr1 chromatin remodeling complex, and the histone acetyltransferase NuA4 PNAS, September 14, 2004; 101(37): 13513 - 13518. [Abstract] [Full Text] [PDF]

				N. A. Kent, S. M. Eibert, and J. Mellor Cbf1p Is Required for Chromatin Remodeling at Promoter-proximal CACGTG Motifs in Yeast J. Biol. Chem., June 25, 2004; 279(26): 27116 - 27123. [Abstract] [Full Text] [PDF]

				J.-m. Hsu, J. Huang, P. B. Meluh, and B. C. Laurent The Yeast RSC Chromatin-Remodeling Complex Is Required for Kinetochore Function in Chromosome Segregation Mol. Cell. Biol., May 1, 2003; 23(9): 3202 - 3215. [Abstract] [Full Text] [PDF]

				G. Wieland, P. Hemmerich, M. Koch, T. Stoyan, J. Hegemann, and S. Diekmann Determination of the binding constants of the centromere protein Cbf1 to all 16 centromere DNAs of Saccharomyces cerevisiae Nucleic Acids Res., March 1, 2001; 29(5): 1054 - 1060. [Abstract] [Full Text] [PDF]

				M. R. Volkert, N. A. Elliott, and D. E. Housman Functional genomics reveals a family of eukaryotic oxidation protection genes PNAS, December 8, 2000; (2000) 260495897. [Abstract] [Full Text]

				P. Hemmerich, T. Stoyan, G. Wieland, M. Koch, J. Lechner, and S. Diekmann Interaction of yeast kinetochore proteins with centromere-protein/transcription factor Cbf1 PNAS, November 7, 2000; 97(23): 12583 - 12588. [Abstract] [Full Text] [PDF]

				A. van Hoof, R. R. Staples, R. E. Baker, and R. Parker Function of the Ski4p (Csl4p) and Ski7p Proteins in 3'-to-5' Degradation of mRNA Mol. Cell. Biol., November 1, 2000; 20(21): 8230 - 8243. [Abstract] [Full Text]

				K. C. Keith and M. Fitzgerald-Hayes CSE4 Genetically Interacts With the Saccharomyces cerevisiae Centromere DNA Elements CDE I and CDE II but Not CDE III: Implications for the Path of the Centromere DNA Around a Cse4p Variant Nucleosome Genetics, November 1, 2000; 156(3): 973 - 981. [Abstract] [Full Text]

				Y. Chen, R. E. Baker, K. C. Keith, K. Harris, S. Stoler, and M. Fitzgerald-Hayes The N Terminus of the Centromere H3-Like Protein Cse4p Performs an Essential Function Distinct from That of the Histone Fold Domain Mol. Cell. Biol., September 15, 2000; 20(18): 7037 - 7048. [Abstract] [Full Text]

				L. Glowczewski, P. Yang, T. Kalashnikova, M. S. Santisteban, and M. M. Smith Histone-Histone Interactions and Centromere Function Mol. Cell. Biol., August 1, 2000; 20(15): 5700 - 5711. [Abstract] [Full Text]

				C. Allmang, E. Petfalski, A. Podtelejnikov, M. Mann, D. Tollervey, and P. Mitchell The yeast exosome and human PM-Scl are related complexes of 3' right-arrow 5' exonucleases Genes & Dev., August 15, 1999; 13(16): 2148 - 2158. [Abstract] [Full Text]

				L. I. Pietrasanta, D. Thrower, W. Hsieh, S. Rao, O. Stemmann, J. Lechner, J. Carbon, and H. Hansma Probing the Saccharomyces cerevisiae centromeric DNA (CEN DNA)-binding factor 3 (CBF3) kinetochore complex by using atomic force microscopy PNAS, March 30, 1999; 96(7): 3757 - 3762. [Abstract] [Full Text] [PDF]

				J. A. Chekanova, R. J. Shaw, M. A. Wills, and D. A. Belostotsky Poly(A) Tail-dependent Exonuclease AtRrp41p from Arabidopsis thaliana Rescues 5.8 S rRNA Processing and mRNA Decay Defects of the Yeast ski6 Mutant and Is Found in an Exosome-sized Complex in Plant and Yeast Cells J. Biol. Chem., October 13, 2000; 275(42): 33158 - 33166. [Abstract] [Full Text] [PDF]

				R. Brouwer, C. Allmang, R. Raijmakers, Y. van Aarssen, W. V. Egberts, E. Petfalski, W. J. van Venrooij, D. Tollervey, and G. J. M. Pruijn Three Novel Components of the Human Exosome J. Biol. Chem., February 23, 2001; 276(9): 6177 - 6184. [Abstract] [Full Text] [PDF]

				A. Kobayashi, T. Miyake, Y. Ohyama, M. Kawaichi, and T. Kokubo Mutations in the TATA-binding Protein, Affecting Transcriptional Activation, Show Synthetic Lethality with the TAF145 Gene Lacking the TAF N-terminal Domain in Saccharomyces cerevisiae J. Biol. Chem., January 5, 2001; 276(1): 395 - 405. [Abstract] [Full Text] [PDF]

				M. R. Volkert, N. A. Elliott, and D. E. Housman Functional genomics reveals a family of eukaryotic oxidation protection genes PNAS, December 19, 2000; 97(26): 14530 - 14535. [Abstract] [Full Text] [PDF]