CHL1 is a nuclear protein with an essential ATP binding site that exhibits a size-dependent effect on chromosome segregation

Abstract

Saccharomyces cerevisiae chl1 mutants have a significant increase in the rate of chromosome missegregation. CHL1 encodes a 99 kDa predicted protein with an ATP binding site consensus, a putative helix–turn–helix DNA binding motif, and homology to helicases. Using site-directed mutagenesis, I show that mutations that are predicted to abolish ATP binding in CHL1 inactivate its function in chromosome segregation. Furthermore, overexpression of these mutations interferes with chromosome transmission of a 125 kb chromosome fragment in a wild-type strain. Polyclonal antibodies against CHL1 show that CHL1 is predominantly in the nuclear fraction of S.cerevisiae. CHL1 function is more critical for the segregation of small chromosomes. In chl1Δ1/chl1Δ1 mutants, artificial circular or linear chromosomes <150 kb in size exhibit near random segregation (0.12 per cell division), whereas all chromosomes tested >225 kb were lost at rates (5 × 10^–3 per cell division) comparable to that observed for endogenous chromosome III. These results reveal an important role for ATPases/DNA helicases in chromosome segregation. Such enzymes may alter DNA topology to allow loading of proteins involved in maintaining sister chromatid cohesion.

INTRODUCTION

The CHL1 gene product of Saccharomyces cerevisiae plays an important role in maintaining a high fidelity of chromosome transmission during mitosis. The original allele, chl1-1, was identified in a screen for mutants exhibiting unusual mating phenotypes (1). chl1-1/chl1-1 diploids were shown to exhibit a bimating phenotype due to the missegregation of one copy of chromosome III carrying the mating-type locus (2). We identified 30 additional alleles of chl1 in a screen for chromosome transmission fidelity (ctf) mutants (3) using an assay that monitored the rate of missegregation of a non-essential 145 kb chromosome fragment. The chl1-1 mutation appeared to cause non-random chromosome loss (2), significantly raising the frequency of recovery of 2n-1 diploids monosomic for chromosome III or chromosome I, but not for all chromosomes tested. It has been suggested that these results may indicate a chromosome size-dependent loss phenotype associated with chl1 mutants. This phenomenon could be related to the increase in mitotic chromosome transmission fidelity observed as a function of increased chromosome length in wild-type cells (4–6). CHL1 is also important for the proper selection of mating type locus, HML or HMR, during mating type switching in yeast (7). Rather than selecting the opposite mating type, Chl1 mutants randomly select mating type (7).

Sequence analysis of a genomic clone of CHL1 (8) revealed a 2.6 kb open reading frame (ORF) encoding a 99 kDa predicted protein with 23% identity to the RAD3 gene (9,10) involved in nucleotide excision repair in yeast (11). Three short domains of high homology between these two genes contained sequence motifs associated with known biochemical functions: an A box and B box consensus found in ATP binding proteins (typically ATPases) (12,13), and a helix–turn–helix DNA binding motif (14). In addition to this, the CHL1 sequence contains all seven consensus motifs found in helicases such as the RAD3 helicase (15,16). A human homolog of CHL1 has been identified which is 33% identical to CHL1 and has DNA helicase activity (17–20). Purified RAD3 protein also exhibits single-stranded DNA-dependent ATPase and helicase activities in vitro (21–23). Furthermore, both of these activities, but not ATP or DNA binding, are abolished by a Lys→Arg mutation in the ATP binding site A-box consensus of RAD3 (24). Unlike RAD3, none of the 30 chl1 mutants tested have any detectable defect in DNA synthesis or DNA repair functions (2,3,8). chl1Δ1 null mutants exhibit near wild-type rates of mitotic recombination and a cell cycle delay in G₂/M which is independent of the RAD9 DNA damage checkpoint control, but may be dependent on the MAD spindle checkpoint (8,25). Thus, CHL1 may be a DNA-associated ATPase (or helicase) that is required to ensure chromosome transmission fidelity.

In this paper, I provide evidence for the importance of the ATP binding site to CHL1 function and for the prediction that CHL1 is in the nucleus of S.cerevisiae. Site-directed conservative mutations in the putative ATP binding site of CHL1 disrupt CHL1 function and form a semi-dominant interfering variant of the CHL1 protein when overexpressed. Antibodies specific for CHL1 when used in conjunction with cell fractionation techniques show that CHL1 is in the nucleus of S.cerevisiae. Finally, I have used human DNA containing artificial chromosomes to show that CHL1 exhibits a size-dependent chromosome missegregation phenotype, providing support for the hypothesis that the non-random chromosome missegregation phenotype originally observed in chl1-1 is due to a chromosome size-dependent requirement for CHL1 (2).

MATERIALS AND METHODS

Strains

Escherichia coli strain CJ236 has the genotype dut, ung, thi, relA1, pCJ105 (Cmr). FZ392 is a mutant derivative of LE392 (kindly provided by Richard Janis at Abbott Laboratories, Abbott Park, IL). Strains containing chl1Δ1::TRP1, chl1Δ1::HIS3 and chl1Δ1::URA3 were constructed by one-step gene replacement as previously described (8). In these strains, the first AUG and 1.7 kb of downstream CHL1 sequence is replaced with pRS304, pRS303 and pRS306, respectively (26). All yeast strains are indicated in Table 1.

Table 1. Yeast strains used in this study.

YPH277	MATa ura3-52 lys2-801 ade2-101 trp1Δ1 leu2-Δ1 CFVII (RAD2.d.YPH277) URA3 SUP11
YPH491	MATa ura3-52 his3-Δ1 leu2-Δ1 trp1D63 prbl-1122 pep4-3 prcl 407
YPH516	Mata/Matα ura3-52/ura3-52 lys2-801/lys2-801 ade2-101/ade2-101 his3-Δ200/his3-Δ200 trp1-Δ1/trp1-Δ1 leu2-Δ1/leu2-Δ1 LAC (H.365kb.YPH425) CEN4 URA3 SUP11 TRP1
YPH851	YPH491 transformed with pRS129
YPH852	YPH491 transformed with p129C
YPH853	YPH491 transformed with p129CGA
YPH854	YPH491 transformed with p129CGV
YPH855	YPH491 transformed with p129CKR
YPH862	MATα, ura3-52, lys2-801, ade2-101, his3-Δ200, trp1Δ1, leu2-Δ1, chl1Δ1::HIS3, CFVII (RAD2.d.YPH275) TRP1 SUP11
YPH863	YPH864 chl1-Δ1::HIS3/chl1-Δ1::URA3 pRS317LYS2
YPH864	Mata/Matα ura3-52/ura3-52 lys2-801/lys2-801 ade2-101/ade2-101 his3-Δ200/his3-Δ200 trp1-Δ1/trp1-Δ1 LEU2/leu2-Δ1 pRS316URA3 CHL1/CHL1 CAC (λ.50kb.YPHS133) TRP1 SUP11
YPH865	YPH866 chl1-Δ1::HIS3/chl-Δ1::URA3
YPH866	Mata/Matα ura3-52/ura3-52 lys2-801/lys2-801 ade2-101/ade2-101 his3-Δ200/his3-Δ200 trp1-Δ1/trp1-Δ1 LEU2/leu2-Δ1 pRS317LYS2 CHL1/CHL1 CAC (λ.100kb.YPHS138) TRP1 SUP11 URA3
YPH867	YPH868 chl1-Δ1::HIS3/chl1-Δ1::TRP1
YPH868	Mata/Matα ura3-52/ura3-52 lys2-801/lys2-801 ade2-101/ade2-101 his3-Δ200/his3-Δ200 TRP1/trp1-Δ1 LEU2/leu2-Δ1 CHL1/CHL1 CIII (HMR HML.355kb.YPHS187) URA3 SUP11
YPH869	Mata/Matα ura3-52/ura3-52 lys2-801/lys2-801 ade2-101/ade2-101 his3-Δ200/his3-Δ200 trp1-Δ1/trp1-Δ1 LEU2/leu2-Δ1 chl1-Δ1::HIS3/chl1-Δ1::TRP1 pYCP50 URA3 SUP11
YPH870	Mata/Matα ura3-52/ura3-52 lys2-801/lys2-801 ade2-101/ade2-101 his3-Δ200/his3-Δ200 trp1-Δ1/trp1-Δ1 LEU2/leu2-Δ1 CHL1/CHL1 pRS317LYS2 pYCP50 URA3 SUP11
YPH871	YPH872 chl1-Δ1::TRP1/chl1-Δ1::URA3
YPH872	Mata/Matα ura3-52/ura3-52 lys2-801/lys2-801 ade2-101/ade2-101 his3-Δ200/his3-Δ200 trp1-Δ1/trp1-Δ1 LEU2/leu2-Δ1 CHL1/CHL1 LAC (H.75kb.YBP304) CEN4 URA3 SUP11 HIS3
YPH873	YPH874 chl1-Δ1::TRP1/chl1-Δ1::URA3
YPH874	Mata/Matα ura3-52/ura3-52 lys2-801/lys2-801 ade2-101/ade2-101 his3-Δ200/his3-Δ200 trp1-Δ1/trp1-Δ1 LEU2/leu2-Δ1 CHL1/CHL1 LAC (H.225kb.YBP321) CEN4 URA3 SUP11 HIS3
YPH875	YPH876 chl1-Δ1::TRP1/chl1-Δ1::URA3
YPH876	Mata/Matα ura3-52/ura3-52 lys2-801/lys2-801 ade2-101/ade2-101 his3-Δ200/his3-Δ200 trp1-Δ1/trp1-Δ1 LEU2/leu2-Δ1 CHL1/CHL1 LAC (H.360kb.YBP364) CEN4 URA3 SUP11 HIS3

Materials

T4 DNA ligase and T4 DNA polymerase were obtained from New England Biolabs (Beverly, MA). Uridine and lyticase were from Sigma Biochemicals (St Louis, MO). All protease inhibitors were obtained from Boehringer Mannheim Biochemicals (Indianapolis, IN), with the exception of Benzamidine and phenylmethylsulfonyl fluoride which were purchased from Sigma. Zymolyase 60T was obtained from ICN Immunobiologicals (Irvine, CA). The BCA protein determination reagents were obtained from Pierce Biochemicals (Rockford, IL).

Media

Media for yeast growth and sporulation were as described (27) except that adenine was added at 6 µg/ml to minimal (SD) media to enhance red pigment formation in ade2 backgrounds (4). SG medium is the same as SD medium except that 2% galactose replaces 2% dextrose.

Plasmids

pRS129 (abbreviated to p129 in Figs 3 and 4) contains the galactose inducible GAL1–GAL10 promoter (28) subcloned into the KpnI site of the CEN6 ARSH4 TRP1 plasmid pRS314 (26). p129C carries a 3.5 kb SpeI fragment containing 2.6 kb of CHL1 predicted ORF, 9 bp of sequence upstream of CHL1, and 900 bp of sequence downstream of CHL1 subcloned into the SpeI site of pRS129. p316C contains a 6.0 kb NotI–PstI fragment derived from pS35 (8) and subcloned into the NotI–PstI sites of the CEN6 ARSH4 URA3 plasmid pRS316 (26). The NotI–PstI fragment in p316C contains the CHL1 ORF and flanking genomic DNA sequence, along with ∼200 bp of pS35-derived polylinker sequence. pATH2C was constructed by cloning the SpeI fragment that contains the CHL1 ORF into the XbaI site of pATH2 (29,30).

Figure 3.

Gal1 promoter directed overexpression of CHL1 point mutant proteins carried on the p129-series of constructs. Western analysis was performed with affinity purified CHL1 antiserum on cell extracts made from YPH491 cells transformed with pRS129 (lanes 9 and 10), p129C (lanes 7 and 8), p129CGA (lanes 5 and 6), p129CGV (lanes 3 and 4) and p129CKR (lanes 1 and 2). Transformants were grown in dextrose (lanes 1, 3, 5, 7 and 9) or galactose (lanes 2, 4, 6, 8 and 10). Western analysis was performed with affinity-purified RJH87 antiserum. Positions of molecular weight markers are indicated on the left.

Figure 4.

Analysis of dominant-negative phenotype of ATP binding site point mutants. A CHL1 haploid (YPH277) was transformed with the plasmid constructs indicated and transformants were replated on galactose plates to induce GAL1 promoter driven overexpression of ATP binding site mutants of CHL1. Chromosome transmission fidelity of a 125 kb chromosome fragment was assessed by the colony color sectoring assay. ATP binding site mutant alleles on p129CGA, p129CKR and wild-type CHL1 on p129C are as indicated in Figure 2A. YPH277 transformed with p129CGV is similar in phenotype to p129CGA under these conditions (data not shown).

Site-directed mutagenesis

Plasmid pS4 consists of a 752 bp SpeI–PvuI fragment of the CHL1 ORF (8) that encompasses the A-box consensus of ATP binding proteins (12) subcloned into the SmaI site of the phagemid pBluescript KS (Stratagene, La Jolla, CA). Uracil-containing single-stranded pS4-derived template was produced in E.coli strain CJ236 infected with helper phage VCS-M13 (Stratagene) essentially as described by the manufacturer. Oligonucleotide-directed mutagenesis was performed using the procedure of Kunkel (31). Oligonucleotides (22 bp) with a single mismatch were used to change GGC at position 139 in the CHL1 nucleotide sequence to GCC or GTC. Similarly, a 21 bp oligonucleotide was used to mutate AAG at position 142 to AGG. These single base changes were confirmed by sequence analysis of the entire 752 bp between SpeI and PvuI. Mutations were placed into the full-length, wild-type CHL1 ORF by replacing the 655 bp PflM1 fragment in wild-type CHL1 with the mutant PflM1 fragment derived from mutagenized pS4. Each of the three mutations generated was subcloned into p129C and p316C.

Phenotypic analysis of CHL1 point mutants

YPH862 (Table 1) carries a deletion of CHL1 in an ade2-101 background which allows visual monitoring of the missegregation of a 125 kb non-essential chromosome fragment bearing the ochre suppressing tRNA (SUP11) (4). YPH862 was transformed selecting Ura⁺ (and Trp⁺) by the LiOAc procedure (32) with each of the members of the p316-series. Transformants were replated onto medium (containing tryptophan), grown at 30°C, and sectoring frequencies scored. Dominant-negative effects of the chl1 point mutations were examined in a CHL1 strain (YPH277) transformed with each of the p129-series and plated onto SG-T (lacking tryptophan) medium.

Induction and purification of TrpE-CHL1 fusion protein

Insoluble proteins were prepared as described previously (33) from FZ392 E.coli (Abbott Labs) strains containing the TrpE-CHL1 expression vector pATH2C (see above). The Trp promoter was activated (34) by the addition of indole acrylic acid (35) to a final concentration of 10 µg/ml. The TrpE-CHL1 fusion protein was cut out of preparative SDS–PAGE gels stained with 3 M copper sulfate (36), and the gel slice was macerated and sent to Hazelton Couance Company (Denver, PA) for the production of antibodies.

Western analysis and cellular fractionation

Cultures of YPH491 transformed with pRS129 (YPH851) or p129C (YPH852) were grown in 2% galactose to induce Chl1 expression and then fractionated into nuclear and post-nuclear fractions as previously described (37). Cellular protein fractions (and whole yeast cells) were extensively sonicated and boiled in Laemmli buffer to ensure maximal solubilization of proteins. Total protein concentration in these solubilized fractions was determined after a TCA DOC protein precipitation to remove interfering substances (38–42). Protein representative of equal numbers of cells from each fraction was loaded by assuming 100% yield of each fraction. The nuclear protein is therefore most likely to be underrepresented. Western blot analysis was performed using standard procedures (43) except that 0.01% sodium dodecyl sulfate was added to the western transfer buffer to replace methanol. The D77 nucleolar antibodies were a kind gift of Dr John P. Aris.

Affinity purification of CHL1 antibodies

CHL1 antibodies were affinity purified from CHL1-TrpE fusion protein that was immobilized on nitrocellulose, using an acid shock treatment essentially as described (44). A 100 kDa protein was recognized in strains overexpressing CHL1 protein, but not in strains expressing wild-type levels of CHL1 (data not shown). Identical results were obtained with affinity purified antibodies derived from the sera of both rabbits (data not shown).

Chromosome missegregation rate calculations

Half sector analysis of various artificial chromosome constructs (linear or circular, containing yeast, plasmid or human DNA) was performed as previously described (4,45). Chromosome missegregation rates of SUP11-containing constructs lost too frequently to score by half sector analysis were calculated by fluctuation analysis essentially as described (46). Approximately 200 cells from each of seven to 10 colonies were replated for each mutant and visually scored for presence of the chromosome fragment. Rates were calculated by the method of the median (47). All circular artificial chromosomes were as described previously (4). YCp50 is a 10 kb pBR322-derived circular minichromosome. The 75 kb yeast artificial chromosome (YAC; in YPH871 and YPH872), 225 kb YAC (in YPH873 and YPH874) and 360 kb YAC (in YPH875 and YPH876) are serial deletion derivatives (48) of the distal arm of a 365 kb human DNA-derived YAC (in YPH516). The sizes of all YACs were confirmed by OFAGE gel analysis (data not shown).

RESULTS

Nuclear localization of CHL1

Attempts to interpret CHL1 localization to nuclei in whole yeast cells by immunofluorescence with affinity-purified CHL1 antibodies were ambiguous due to a weak nuclear signal in chl1Δ1 strains. I, therefore, took advantage of a nuclear purification technique (37) to determine the localization of CHL1 in yeast cells. Cytoplasmic (post-nuclear supernatant) and nuclear extracts were made from a protease-deficient strain grown under conditions leading to GAL1-driven overproduction of CHL1 (YPH 852). Cytoplasmic and nuclear fractions from equal numbers of cells were separated by SDS–PAGE and analyzed for the presence of CHL1 or a control nucleolar protein (D77) by western analysis (Fig. 1). The 39 kDa D77 nucleolar antigen was present in nuclear subcellular fractions (Fig. 1A, lane 2), but not in cytoplasmic subcellular fractions (Fig. 1A, lane 1) of cells expressing elevated (Fig. 1) or wild-type levels of CHL1 (data not shown). A similar subcellular distribution was found using affinity-purified CHL1 antibodies (Fig. 1B). A 100 kDa CHL1-specific band was detected in nuclear extracts (Fig. 1B, lane 2), and is weakly detectable in cytoplasmic fractions (Fig. 1B, lane 1) made from a strain overexpressing CHL1. It is not detectable in any of the fractions from a strain expressing wild-type levels of CHL1 (not shown). A 67 kDa cross reacting band is present in western blots analyzed with either CHL1 or D77 antibodies. This band is also present in cells completely lacking CHL1 and is probably due to non-specific antibody binding to an antigen of this size. I conclude that CHL1 is located predominantly in the nucleus of S.cerevisiae. Its low abundance in the cytoplasmic fraction may indicate a role for CHL1 in the cytoplasm in addition to the nucleus, may simply be due to overexpression of Chl1 or may be due to leakage of Chl1 from nuclei when extracts are made.

Figure 1.

Localization of CHL1 protein by western analysis of cell fractions. Whole yeast cells overexpressing CHL1 were fractionated into nuclear (lanes 2) and cytoplasmic (lanes 1) fractions. Protein extract representative of equal cell numbers was loaded in each lane and subject to western analysis with D77 antiserum in (A) or affinity purified RJH87 anti-CHL1 antiserum in (B). Positions of molecular weight markers are indicated on the left.

Conservative mutations in the ATP binding site consensus of CHL1 disrupt CHL1 function

To determine whether the putative ATP binding site was important for CHL1 function, I made site-directed changes in highly conserved residues of the ATP binding site of CHL1 (Fig. 2A, ‘C’) and analyzed the phenotypes of strains carrying these mutations on a plasmid. All amino acid changes maintained the charge approximate size, and hydrophobic character of the wild-type amino acid. Gly47 of CHL1 was changed to Ala or Val, and Lys48 was changed to Arg (Fig. 2A, ‘CGA, CGV and CKR’, respectively). Single copy (CEN ARS) plasmids were made which allowed either CHL1-promoter driven synthesis (p316-series), or GAL1-driven overexpression (p129-series) of CHL1 point mutant derivatives (see Materials and Methods).

Figure 2.

(A) ATP binding site point mutations made in CHL1. (B) Analysis of CHL1 function of ATP binding site mutants using the colony color sectoring assay. A chl1Δ1 null haploid strain (YPH862) was transformed with the plasmid constructs indicated and transformants were plated to visualize their chromosome missegregation phenotype by the colony color sectoring assay. p316 is vector alone. The p316-series of plasmid constructs carry wild-type or ATP binding site point mutants of CHL1 driven by their own promoter on a single copy CEN ARS plasmid pRS316 (26).

To determine whether the ATP binding site mutants retained CHL1 activity, the p316-series of constructs (CHL1 promoter-driven) were transformed into a chl1 deletion mutant carrying a 125 kb non-essential chromosome fragment for monitoring chromosome transmission fidelity (4,45). Wild-type (p316C) CHL1 was able to fully complement the dramatic chromosome missegregation phenotype of chl1Δ1 mutants (Fig. 2B). Vector alone (p316) and plasmid constructs carrying each of the three ATP binding site mutations (p316CGA, p316CGV and p316CKR) were unable to complement the chromosome transmission defect of chl1Δ1 (Fig. 2B).

I also determined whether overexpression of CHL1 ATP binding site mutants (p129-series) could rescue the chl1Δ1 chromosome fragment missegregation phenotype. Only strains overexpressing wild-type CHL1, but none of the mutants, could complement the chl1Δ1 mutation (data not shown). Western analysis (Fig. 3) revealed that the chl1 and CHL1 proteins were being made. A 100 kDa band was clearly detectable in extracts from cells overexpressing CHL1 (lane 7), and to various extents in extracts overexpressing each of the three chl1 mutant constructs (lanes 1, 3 and 5). This 100 kDa band could not be detected in extracts from cells containing vector alone (lanes 9 and 10), or those grown in dextrose (lanes 2, 4, 6 and 8). Thus, the CHL1 ATP binding site mutant proteins are being expressed, but they do not possess CHL1 function. I conclude that the ATP binding site of CHL1 is essential to CHL1 function.

CHL1 ATP binding site mutants exhibit a dominant-negative effect on chromosome transmission

Changes in the GKT consensus of some nucleotide binding proteins have created functionally altered proteins with dominant interfering biological effects (49). I therefore analyzed whether ATP binding site mutations in CHL1 could interfere with chromosome transmission in a CHL1 strain. YPH277 is a CHL1 strain that carries a 125 kb non-essential chromosome fragment in a background designed to monitor the chromosome missegregation rate of the fragment by a colony color sectoring assay (4). This strain was transformed with the p129-series (GAL1 promoter driven constructs) and transformants were analyzed for any chromosome segregation defects associated with CHL1 mutant overexpression. The fidelity of chromosome transmission was not affected by overexpression of CHL1 on p129C, or the presence of the vector (p129) alone (Fig. 4). Overexpression of each of the point mutant alleles of CHL1, especially the Lys47→Arg47 mutation, interfered with the fidelity of transmission of the 125 kb chromosome fragment (Fig. 4). This effect was not detectable in strains transformed with a CEN ARS vector carrying these point mutations expressed from the CHL1 promoter (data not shown). This and the above data suggest that ATP binding site mutant proteins of CHL1 are still competent to bind a protein or DNA in wild-type cells but that the lack of ATPase activity causes the mutant protein to interfere directly or indirectly with the process of chromosome segregation.

Null alleles of CHL1 destabilize circular chromosomes

Previous work (8,50) has shown that the absence of CHL1 function has a dramatic effect on the fidelity of transmission of linear chromosomes, including chromosome III. I analyzed the effect of a chl1 null mutant on the segregation of a circular derivative of chromosome III. I reasoned that if CHL1 plays a role in telomere structure or function, I might expect the transmission fidelity of circular chromosomes (which lack telomeres) to be unaffected in chl1 mutants. Chromosome missegregation rates were determined for a SUP11-marked 335 kb circular derivative of chromosome III (4,50) using half-sector analysis (4). In CHL1/CHL1 (wild-type) diploids this circular chromosome is lost at a rate of 4.2 × 10^–4 per cell division (Table 2), and in chl1Δ1/chl1Δ1 mutant diploids it is lost at a rate of 1.2 × 10^–3. Therefore, the chl1Δ1 mutation has a minor but measurable effect on circular chromosome III, increasing its missegregation rate 3-fold. Despite this small fold effect, the missegregation rate of circular chromosome III in chl1Δ1 strains was raised to the same level as that observed for natural chromosome III in chl1Δ1 strains (Table 2).

Table 2. Analysis of chromosome missegregation rates in CHL1/CHL1 and chl1Δ1/chl1Δ1 diploids.

Structure	Length (kb)	Absolute missegregation rate in CHL1/CHL1 diploids	Absolute missegregation rate in chl1Δ1/chl1Δ1 diploids	Fold effect
Circle	10	7.9 × 10–2	1.5 × 10–1	2
Circle	50	9.0 × 10–3	1.5 × 10–1	16
Circle	100	4.6 × 10–3	1.5 × 10–1	33
Circle III	335	4.2 × 10–4	1.2 × 10–3	3
Human DNA
Linear HYAC	75	4.2 × 10–4	1.4 × 10–1	340
Linear HYAC	225	4.9 × 10–4	3.8 × 10–3	8
Linear HYAC	360	2.7 × 10–3	7.3 × 10–3	3
Yeast DNA
CEN 3L	145	4.0 × 10–4	1.0 × 10–1	3000
CHR III	365	1.5 × 10–5	3.1 × 10–3	210

I also analyzed the chromosome missegregation rates of circular artificial chromosomes of different sizes in CHL1/CHL1 and chl1Δ1/chl1Δ1 diploid strains using half sector- and fluctuation analysis (see Materials and Methods). Consistent with previously published results (4,5), larger circular artificial chromosomes are more stable in wild-type strains (Table 2). The 10, 50 and 100 kb chromosomes showed missegregation rates of 7.9 × 10^–2, 9.0 × 10^–3 and 4.6 × 10^–3 per cell division, respectively (Table 2). However, in chl1Δ1/chl1Δ1 mutants, all three circular artificial chromosomes missegregate at the same high rate of 0.15 missegregation events per cell division (Table 2). Thus, CHL1 is essential to the transmission of circular chromosomes, suggesting that it does not strictly play a role in telomere maintenance.

Effect of size on missegregation rates of linear chromosomes in a chl1 null mutant

It has been observed that chl1-1 generates 2n-1 karyotypes with a bias towards missegregation of the smaller chromosomes (2,50). I wanted to distinguish between a possible chromosome size-dependent requirement for CHL1 and lethality or slow growth associated with specific aneuploidies. To do this systematically, I took advantage of the availability of human DNA-based non-essential YACs of different sizes. The absence of yeast DNA in these chromosomes circumvents any potential for gene dosage associated toxicity which might bias missegregation rates for particular chromosomes. All YACS used were sequential deletion derivatives (48) of a 365 kb YAC. The short arm of the YAC is ∼10 kb in length and contains a yeast auxotrophic marker, ARSH4, SUP11 and yeast telomeric sequences. The long arm consists of ∼340 kb of human DNA with a yeast auxotrophic marker and yeast telomeric sequences at its end. Deletion derivatives of this YAC sequentially delete sequences from the telomere on the long arm.

Chromosome missegregation rates were determined (see Materials and Methods) for a 75, 225 and 360 kb YAC in CHL1/CHL1 and chl1Δ1/chl1Δ1 diploids (Table 2). In CHL1/CHL1 diploids, the missegregation rates of the 75, 225 and 360 kb YACs were 4.2 × 10^–4, 4.9 × 10^–4 and 2.7 × 10^–3 per cell division, respectively (Table 2). In chl1Δ1/chl1Δ1 diploids the 75 kb YAC was destabilized 340-fold to a level approaching random segregation. In contrast, the 225 and 360 kb YACs were destabilized 8- and 3-fold, respectively, to missegregation rates of 3.8 × 10^–3 and 7.3 × 10^–3, approaching that previously observed for the missegregation of endogenous chromosome III in chl1Δ1/chl1Δ1 diploids (Table 2) (8). Although the analysis of the fold increase in chromosome missegregation rates in CHL1/CHL1 versus chl1Δ1/chl1Δ1 diploids is complicated (see Discussion), the analysis of the absolute missegregation rates in chl1Δ1/chl1Δ1 diploids for the various chromosomes is more clear. These results suggest that the lack of CHL1 cripples the segregation of a small chromosome to the point that it almost segregates randomly. In contrast, the lack of CHL1 affects the segregation of larger chromosomes more moderately.

We showed previously that the 365 kb full-length chromosome III missegregates at a rate of 1.5 × 10^–5 in CHL1/CHL1 cells and is destabilized 210-fold in chl1Δ1/chl1Δ1 diploids, to a missegregation rate of 3 × 10^–3 (8). To determine the effect of deleting CHL1 on smaller chromosome III derivatives, I measured the chromosome missegregation rates of a 145 kb chromosome fragment in chl1Δ1/chl1Δ1 and CHL1/CHL1 strains. This chromosome fragment exhibited a chromosome missegregation rate of 4 × 10^–4 in CHL1/CHL1 strains. In chl1Δ1/chl1Δ1 strains, however, the chromosome missegregation rate rose dramatically (3000-fold) to 1 × 10^–1 (Table 2) (8). Once again, this suggests that the loss of CHL1 affects the segregation of smaller chromosomes more than larger chromosomes.

DISCUSSION

The CHL1 gene product is critical to the high fidelity of mitotic chromosome transmission. Although CHL1 is not essential for viability, chl1 mutants exhibit extremely high rates of sister chromatid missegregation. Our previous work (8) suggested that the CHL1 gene product acts as an ATP binding protein that functions after DNA synthesis and before the completion of mitosis. In the present work, I present experimental support for the importance of the putative ATP binding site, localize the CHL1 gene product to the nucleus, and analyze the effect of chromosome length and structure on the chl1Δ1-associated chromosome missegregation phenotype.

Conservative site-directed changes (Gly47 to Ala or Val, and Lys48 to Arg) made in the predicted ATP binding site A-box consensus of CHL1 inactivated its chromosome transmission fidelity function. Mutations within these three highly conserved amino acids (GKT/S) of known ATP and GTP binding proteins generally alter the biological function of the protein (49,51–55). Changes identical to those introduced into CHL1 which maintain the approximate size, charge and hydrophobic characteristics of the wild-type amino acid residue have been introduced into the mdr (51) and RAD3 (24) genes. A Gly→Ala mutation or a Lys→Arg mutation in either of the two ATP binding site A-box consenses in mdr abolishes its ability to confer multidrug resistance, yet retains the ability of the protein to bind ATP (51). A Lys→Arg mutation in the A-box consensus of RAD3 abolishes its ATPase and helicase activities, but also still allows ATP and DNA binding. Thus, proteins with these conservative ATP binding site mutations appear to retain wild-type structure to the extent that they can bind ATP, yet biological functions which depend on the energy of hydrolysis of ATP are prevented. Proteins with these characteristics may form irreversible associations in the absence of ATP hydrolysis and thus act as dominant interfering mutants. For example, a Lys→Asn mutation in the GKT amino acid consensus of the Ha-ras GTPase causes a dominant temperature-dependent lethality in S.cerevisiae (49). Similarly, Chl1 proteins containing mutations in the ATP binding site exhibited no detectable CHL1 function. Furthermore, these ATP binding site mutations in CHL1 exhibited a dominant-negative effect on chromosome segregation when overexpressed in wild-type cells. These results suggest that the ATP binding site in CHL1 is essential to the role of CHL1 in chromosome segregation. Furthermore, by drawing a parallel between the mdr, rad3 and chl1 ATP binding site point mutant data and sequence homologies between these proteins, I suggest that the CHL1 ATP binding site mutants retain partial CHL1 structure, and may bind their normal substrate (DNA perhaps). Once bound, however, ATPase- (or helicase-)defective Chl1 protein may then hinder the recognition or loading of protein complexes essential for proper chromosome segregation during mitosis. Several explanations could account for the need to overexpress CHL1 point mutants to observe a dominant phenotype. For example, the interacting component may be relatively abundant. Alternatively, the association of CHL1 with this component may be weak, or transient (if CHL1 is an ATPase), or the fraction of mutant chl1 protein that is interaction-competent may be small.

A chl1Δ1 mutation causes dramatic increases in the missegregation rates of small (<100 kb) circular artificial chromosomes, showing that structures that lack telomeres still have a requirement for CHL1 function for accurate chromosome segregation. Furthermore, a chl1 deletion mutant exhibits no measurable effect on telomere length (F.Spencer and V.Lundbladt, unpublished data). These results suggest that CHL1 does not play a role in telomere structure.

I have attempted to determine if chl1 mutants exhibit a chromosome-size dependent effect on the missegregation rates of yeast chromosomes, as previously suggested by Liras and colleagues (2,50). First, I measured the chromosome missegregation rates of three related YACs of different sizes (75, 225 and 360 kb) in chl1Δ1/chl1Δ1 and CHL1/CHL1 strains. Then, I measured the missegregation rates of a 145 kb chromosome fragment derivative of endogenous yeast chromosome III in chl1Δ1/chl1Δ1 and CHL1/CHL1 strains and compared that to our published chromosome missegregation rate for the 365 kb full-length endogenous chromosome III in the same strains (Table 2) (8).

Two numerical values emerge from our studies of chromosome missegregation in these strains (Table 2): the absolute missegregation rates of chromosomes of different sizes in chl1Δ1/chl1Δ1 strains, and the fold effect on chromosome missegregation caused by deleting CHL1 (rate in chl1 mutants/rate in wild-type cells).

When one focuses on the absolute missegregation rates, a clear trend emerges (Table 2). In chl1Δ1/chl1Δ1 strains the larger 225 and 360 kb YACs and 365 kb endogenous chromosome are still able to segregate quite well, with average missegregation rates of 5 × 10^–3 per cell division. In contrast, however, in chl1Δ1/chl1Δ1 strains the smaller 75 kb YAC and 145 kb chromosome III derivative segregate very poorly, with average missegregation rates of 0.12 per cell division. Thus, deleting CHL1 results in more severe defects in the segregation of small chromosomes than large chromosomes.

When one focuses on the fold effect on chromosome missegregation caused by deleting CHL1, the analysis is more complex (Table 2). In wild-type cells the 225 and 360 kb YACs were less stable than the shorter 75 kb YAC. The relative instability of the two larger YACs was a surprise to us, since previous studies with λ DNA-based YACs showed that larger chromosomes were more stable than smaller chromosomes (4–6). A significant difference between these studies and my study lies in the length of the chromosome arm containing the foreign DNA insert (<150 kb, in contrast to 225 and 360 kb in our study). Thus, the relative instability in wild-type cells of our longer human YACs may be related to difficulties associated with replicating and segregating long chromosome arms composed of foreign DNA. The loss of CHL1 had a small fold effect on these two YACs (3–8-fold). This may be because CHL1 acts in the same pathway that is already defective in these YACs. In contrast to the unusual missegregation behavior of the two long YACs in wild-type cells, the smaller 75 kb YAC, the 145 kb chromosome III fragment and the 365 kb chromosome III behaved as expected from the literature. They exhibited missegregation rates that decreased with increasing size, and these missegregation rates were close to those measured for similar-sized chromosomes in the literature (4–6,56–59). Deletion of CHL1 had a similar and large fold effect (200–3000-fold) on these three chromosomes (Table 2).

Why does deleting CHL1 have similar fold effect on chromosomes of all sizes yet have such a significant effect on the absolute missegregation rate of small chromosomes? The lack of CHL1 may cause a similar defect on all chromosomes. However, this defect may be additive (or synergistic) with a cis defect already associated with small chromosomes, and that causes smaller chromosomes to missegregate more often than large chromosomes in wild-type cells (4–6,56).

There is some evidence for cis defects in small chromosomes. First, very short chromosomes (∼15 kb) have irreparable defects that activate the spindle damage checkpoint (60). The exact nature of this defect is not yet known. However, it must affect the mitotic spindle or chromosome structure, and alter the forces on the kinetochore during metaphase chromosome alignment, since these are the types of defects that activate the spindle checkpoint (61–64). I do not know if my 75 kb human DNA based YAC or 145 kb chromosome III derivatives activate the spindle checkpoint. Second, chromosomes below a size threshold of ∼200 kb, may not have, or may not be able to maintain, sister chromatid intercatenation after DNA replication (65,66). These intercatenanes may be required, in conjunction with protein complexes such as the ‘cohesins’ (67), to hold sister chromatids together until the metaphase to anaphase transition (5,68,69). At this transition, only associated sisters with bipolar spindle attachment will segregate accurately during the ensuing anaphase. If these intercatenanes are not present in chromosomes <200 kb, then they will be more sensitive to defects in the sister chromatid cohesion complex, leading to a greater missegregation frequency in wild-type cells, and significant missegregation rates in mutants that affect that complex.

If removing CHL1 is additive (or synergistic) with these types of cis defects in small chromosomes, what role might Chl1 play during chromosome segregation? It is intriguing that the chromosome size range below which removing CHL1 causes almost random chromosome segregation, coincides with the size range within which sister chromatids may no longer be held together by intercatenation. Perhaps CHL1 normally regulates some function that compensates for the absence of intercatenation of small chromosomes. One model that can be tested is that Chl1 acts as a helicase that modifies DNA structure during the establishment or maintenance of sister chromatid cohesion.