Function of Tubulin Binding Proteins in Vivo

Corresponding author: Frank Solomon, Bldg. E17, Rm. 220, MIT, Cambridge, MA 02139., solomon{at}mit.edu (E-mail)

Communicating editor: M. D. ROSE

	ABSTRACT

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Overexpression of the ß-tubulin binding protein Rbl2p/cofactor A is lethal in yeast cells expressing a mutant -tubulin, tub1-724, that produces unstable heterodimer. Here we use RBL2 overexpression to identify mutations in other genes that affect formation or stability of heterodimer. This approach identifies four genes—CIN1, CIN2, CIN4, and PAC2—as affecting heterodimer formation in vivo. The vertebrate homologues of two of these gene products—Cin1p/cofactor D and Pac2p/cofactor E—can catalyze exchange of tubulin polypeptides into preexisting heterodimer in vitro. Previous work suggests that both Cin2p or Cin4p act in concert with Cin1p in yeast, but no role for vertebrate homologues of either has been reported in the in vitro reaction. Results presented here demonstrate that these proteins can promote heterodimer formation in vivo. RBL2 overexpression in cin1 and pac2 mutant cells causes microtubule disassembly and enhanced formation of Rbl2p-ß-tubulin complex, as it does in the -tubulin mutant that produces weakened heterodimer. Significantly, excess Cin1p/cofactor D suppresses the conditional phenotypes of that mutant -tubulin. Although none of the four genes is essential for viability under normal conditions, they become essential under conditions where the levels of dissociated tubulin polypeptides increase. Therefore, these proteins may provide a salvage pathway for dissociated tubulin heterodimers and so rescue cells from the deleterious effects of free ß-tubulin.

THE spatial and temporal control of microtubule assembly is an essential aspect of many cellular functions, including division, motility, and organization of the cytoplasm. The development of a robust in vitro assembly reaction of microtubule polymers from heterodimeric subunits of - and ß-tubulin has had a major impact on the field. That assay led to identification of factors—structures, proteins, and small molecules—that influence the extent and organization of microtubule assembly. Reverse genetic techniques have enabled evaluation of the relevance of some of those factors to in vivo conditions.

Direct genetic approaches can also identify functions that directly modulate the assembly of subunits into microtubules (PASQUALONE and HUFFAKER 1994 ). What is notable, however, is the growing list of gene products that affect microtubule-dependent processes but do not interact directly with the polymer. For example, several genes that affect chromosome instability (HOYT et al. 1990 ), sensitivity to either microtubule depolymerizing drugs (STEARNS et al. 1990 ) or excess ß-tubulin (ARCHER et al. 1995 ), dependence upon a mitotic motor (GEISER et al. 1997 ), or phenotypes of mutants in -tubulin (GEISSLER et al. 1998 ) or -tubulin (VEGA et al. 1998 ) encode proteins that act on microtubule control at some step other than the polymerization reaction.

It is also striking that the screens enumerated above have, despite their diverse designs, frequently identified the same genes. For example, certain CIN (chromosome instability) genes affect not only chromosome instability and sensitivity to benomyl—the contexts in which they originally were identified—but also yeast cells' ability to function without the kinesin Cin8p (GEISER et al. 1997 ). Similarly, mutations in the PAC genes perish in the absence of Cin8p, but some also participate in cellular responses to excess ß-tubulin and to -tubulin function (ALVAREZ et al. 1998 ; GEISSLER et al. 1998 ).

Mammalian homologues of some of these proteins are essential factors in an in vitro reaction that mediates exchange of unfolded tubulin polypeptides into -ß-tubulin heterodimers (GAO et al. 1992 , GAO et al. 1994 ; MELKI et al. 1996 ; TIAN et al. 1996 ). Under the conditions of this reaction, the tubulin polypeptides released from the chaperonin complex do not exchange efficiently into preexisting heterodimers (TIAN et al. 1997 ). The assay identifies five factors that interact with monomeric - and ß-tubulin chains, finally bringing them together in a large complex that can be the precursor of heterodimer (TIAN et al. 1997 ). Four of those factors have homologues in budding yeast, three identified by independent genetic investigations—Cin1p/cofactor D (HOYT et al. 1990 ; STEARNS et al. 1990 ), Rbl2p/cofactor A (ARCHER et al. 1995 ), and Pac2p/cofactor E (GEISER et al. 1997 )—and a fourth, Alf1p/cofactor B identified by homology to the vertebrate protein (TIAN et al. 1997 ). None of these proteins is essential in Saccharomyces cerevisiae, although the homologue of CIN1 in Schizosaccharomyces pombe, ALP1, is essential and has been shown to bind microtubules (HIRATA et al. 1998 ). The results suggest that the in vitro assay does not fully represent early steps of microtubule assembly in vivo.

The RBL2 (rescue excess ß-tubulin lethality) gene has properties that make it particularly valuable for exploring the processing of tubulin polypeptides in vivo (ARCHER et al. 1995 , ARCHER et al. 1998 ). Overexpression of Rbl2p efficiently rescues the microtubule disassembly and cell death that are the consequences of ß-tubulin overexpression, and it interacts genetically with several conditional mutants of -tubulin. Like its mammalian homologue cofactor A in vitro (MELKI et al. 1996 ), Rbl2p binds ß-tubulin both in vivo and in vitro to form a heterodimer that excludes -tubulin. Unlike cofactor A, which binds only to a form of ß-tubulin that is not competent to bind -tubulin, Rbl2p can bind to ß-tubulin molecules both before and after they have been incorporated into heterodimer (ARCHER et al. 1998 ). RBL2 is not essential for mitotic growth but is essential for normal meiosis and normal resistance to microtubule depolymerizing drugs. In addition, its synthesis may be upregulated at the G2/M stage of the cell cycle although ß-tubulin expression is apparently unchanged (VELCULESCU et al. 1997 ).

These properties suggest that Rbl2p functions may affect processes other than folding of ß-tubulin, and the genetic interactions indicate what those processes may be. Deletion of RBL2 is lethal in cells that have depressed ratios of - to ß-tubulin, either because they lack the minor -tubulin gene, TUB3 (A. SMITH, M. MAGENDANTZ and F. SOLOMON, unpublished results) or because they lack PAC10, which regulates that ratio (ALVAREZ et al. 1998 ; GEISSLER et al. 1998 ). Conversely, overexpression of RBL2 is lethal in cells expressing a mutant -tubulin that forms a weaker -ß-tubulin heterodimer (VEGA et al. 1998 ).

To understand those interactions, we have applied RBL2 overexpression to identify nontubulin genes that influence heterodimer stability. We show here that cells lacking either of two of the proteins noted above, Cin1p/cofactor D or Pac2p/cofactor E, die upon overexpression of Rbl2p. Physiological and biochemical analyses presented here support a role for these proteins in promoting heterodimer formation. That role could be important for de novo heterodimer formation. However, since these genes are not essential under normal conditions, they may instead participate in a salvage pathway acting on dissociated tubulin polypeptides to protect cells from the toxicity of free ß-tubulin.

	MATERIALS AND METHODS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Strains, plasmids, and media:
All yeast strains are derivatives of FSY185 (WEINSTEIN and SOLOMON 1990 ) with the exception of the tub1 mutants (SCHATZ et al. 1988 ). We used standard methods for yeast manipulations (SCHATZ et al. 1986 ; SOLOMON et al. 1992 ). All the relevant strains are listed in Table 1.

Screen for erl mutants:
Wild-type cells containing pGAL-RBL2:URA3:CEN (pA5) were mutagenized with ethyl methanesulfonate (EMS) to 25% survival. The mutagenized strains were grown in YPD media for 4 hr. The cells were frozen at -70° in 25% glycerol. Cells were plated from frozen stocks to SC -ura glucose plates (200/plate) and after 40 hr growth were replica plated to SC -ura galactose plates. Replica plated colonies that were unable to grow on galactose were retested by streaking to SC -ura galactose and SC -ura raffinose plates. Cells unable to grow on galactose were streaked to SC 5-fluoroorotic acid (5-FOA) plates to select for loss of the plasmid. Positive erl (enhancer of Rbl2p lethality) mutants were able to grow on galactose after loss of the pA5 plasmid, but were unable to grow on galactose after retransformation with the same pA5 plasmid.

Construction of CIN1, PAC2, and TUB1 knockouts:
To disrupt CIN1, the primers 5'-GCACGACGTCGATAATATTTTTGGAAAGAACGCC and 5'-GCAGAGATCTGTTGATCGCGGCAATCGTCTGTTGGTGC were used to amplify DNA in the 5' untranslated region (UTR) of CIN1. The primers 5'-GACCGTCGACGAGATAAAGAAATGCGGAATGAAGC and 5'-GACCGCATGCGAGATAAAGAAATGCGGAATGAAGC were used to amplify DNA in the 3' UTR of CIN1. The PCR products from these primers were ligated into pNKY51 (ALANI et al. 1987 ) on opposite ends of the hisG-ura3-hisG sequence. The plasmid was digested with AatII and EagI and the CIN1 5' UTR-hisG-URA3-hisG -CIN1 3' UTR DNA fragment was isolated (Qaiex II from QIAGEN, Chatsworth, CA) after electrophoresis on a 1% agarose gel. This DNA was then transformed into FSY185 to create a disruption of the entire CIN1 open reading frame (ORF). The disruption was confirmed by Southern blot analysis of the diploids and of their haploid segregants and by the phenotypic analysis of the haploid segregants.

To disrupt PAC2, pPA14 containing 1050 bp of 5' PAC2 UTR and 800 bp of 3' PAC2 UTR (ALVAREZ et al. 1998 ) joined together at a BamHI site in pGEM (Promega, Madison, WI) was digested. The BamHI-BglII fragment containing hisG-ura3-hisG from PNK51 was cloned into the BamHI site of digested pPA14. The resulting plasmid pLV59 was digested with BglII and NotI to release a 5.7-kb disruption fragment to transform FSY183. The disruption was confirmed by PCR analysis.

To disrupt TUB1, the primers 5'-CGCGACGTCTATCAATGGCGGGCAC and 5'-GTTTACAGATCTTGGGTGGC were used to amplify DNA in the 5' UTR of TUB1. The primers 5'-GCGGCATGCCGACTCATACGCTGAGG and 5'-GCGCGGCCGGCATCAACTGTGACATCG were used to amplify DNA in the 3' UTR of TUB1. The PCR products from these primers were ligated into pNKY51 (ALANI et al. 1987 ) on opposite ends of the hisG-ura3-hisG sequence. The plasmid was digested with AatII and EagI and the TUB1 5' UTR-hisG-URA3-hisG-TUB1 3' UTR DNA fragment was isolated (Qaiex II from QIAGEN) after electrophoresis on a 1% agarose gel. This DNA was then transformed into FSY182 to create a disruption of the entire TUB1 open reading frame marked with hisG-URA3-hisG. The disruption was confirmed by loss of HIS⁺ and by PCR analysis.

Viability measurements:
JFY203 (cin1 containing pA5), JFY3 (wild-type containing pA5), and LTY500 (pac2 containing pA5) were grown overnight in SC -ura raffinose media. Log phase cells were then induced with 2% galactose and at various time points aliquots of cells were taken and counted using a hemocytometer. Known numbers of cells were then plated to SC -ura glucose plates. Cell viability was measured as the percentage of cells able to form colonies on the SC -ura glucose plates.

In vivo His₆-Rbl2p-ß-tubulin association experiments:
Yeast strains LTY503(pac2), JFY253(cin1), and LTY573 (wild-type) containing a GAL-RBL2-HIS₆ plasmid (pGHR) were grown overnight at 30° in selective media containing raffinose to about 2 x 10⁹ cells (log phase) per experiment. To induce His₆-RBL2 expression, 2% galactose was added. After 3 hr, protein was harvested by glass bead lysis in 1 ml PME buffer plus protease inhibitors. We applied 0.85 ml of protein extract to 50 µl Ni-NTA beads (QIAGEN). We washed and eluted the bound proteins as previously described (MAGENDANTZ et al. 1995 ). Eluted proteins were subjected to SDS-PAGE analysis and probed by immunoblotting for -tubulin, ß-tubulin, and Rbl2p.

Immune techniques:
Immunoblots: Modifications of standard procedures (SOLOMON et al. 1992 ) were used to assay for Rbl2p-His₆-ß-tubulin association. After gel electrophoresis and transfer to nitrocellulose membranes, we blocked with TNT (0.025 M Tris, pH 7.5, 0.17 M NaCl, 0.05% Tween-20) for 30–120 min. Primary antibodies were incubated for >12 hr at 1/3500 (#206 or #345; WEINSTEIN and SOLOMON 1990 ) or at 1/100 (#250; ARCHER et al. 1995 ) and then washed five times (5 min each) in TNT. Bound antibody was detected by ¹²⁵I Protein A (New England Nuclear, Boston).

In other experiments, after gel electrophoresis (as above), we blocked with milk (5% Carnation) TBST (0.05 M Tris, pH 8.0, 0.15 M NaCl, 0.1% Tween-20) overnight. Primary antibodies were incubated for 1–2 hr at 1/3500 for #206 and #345 and at 1/5000 for 12CA5 (Boehringer Mannheim, Indianapolis) in milk TBST. The blots were washed six times (two 20 sec, one 15 min, three 5 min) in TBST alone. Blots were incubated with 1/3000 dilutions of horseradish peroxidase conjugated goat anti-rabbit (Jackson ImmunoResearch, West Grove, PA) for #206 and #345 and horseradish peroxidase conjugated rabbit anti-mouse (Jackson ImmunoResearch) for 12CA5, in milk TBST, washes were done in TBST as above, and detected by chemiluminescence (Renaissance NEN).

Immunofluorescence: We used standard techniques (SOLOMON et al. 1992 ). Primary antibody was #206 (anti-ß-tubulin) and secondary antibody was fluorescein conjugated goat anti-rabbit IgG (Cappel). To visualize DNA, 4',6-diamidino-2-phenylindole (DAPI; Boehringer Mannheim) was used.

Analysis of -tubulin mutations synthetic lethal with cin1 and pac2:
cin1 tub1 tub3 (JFY474) or pac2 tub1 tub3 (LTY479) strains containing a plasmid with a genomic copy of TUB1 on a URA3 CEN vector or TUB3 on a URA3 2 µm vector, respectively, were transformed with LEU2 CEN plasmids containing the various -tubulin mutations. The strains containing both the wild type and a mutant form of TUB1 were grown on 5-FOA plates to select for cells that have lost the wild-type -tubulin plasmid, since 5-FOA kills URA3⁺ but not ura3^- cells. pac2 or cin1 strains that are synthetic lethal with the -tubulin mutations will be unable to lose the wild-type -tubulin plasmid and cannot survive on 5-FOA. However, strains that are viable without the wild-type -tubulin allele are able to lose this plasmid along with the URA3 gene and form colonies.

Construction of GAL-CIN1, GAL-CIN2, and GAL-CIN4:
To construct the GAL-CIN1 URA3 CEN plasmid, the CIN1 ORF and additional 5' and 3' UTR were amplified by PCR. The 5' primer (5'-GACACGCGTCATGAACAATATTCGGGCCTTGC) contained a MluI site and the 3' primer (5'-CAGCCGCGGATTATATGTAAAATTTGCCGTTTAC) contained a SacII site. The PCR product was ligated into the pT7-Blue plasmid from Novagen. This DNA was then digested with MluI and SacII and ligated into the pRS316-GAL1 plasmid (LIU et al. 1992 ). The construct (pJF10) suppressed the benomyl supersensitive phenotype of cells deleted for CIN1.

To construct the GAL-CIN1 HIS3 CEN plasmid, the pJF10 plasmid was digested with ApaI and SacII (both cut in the polylinker), which liberates a fragment that contains the GAL promoter and the entire CIN1 open reading frame, and ligated into pRS313. The construct (pJF17) suppresses the benomyl supersensitive phenotype of cells deleted for CIN1.

To construct the GAL-CIN2 HIS3 CEN plasmid, the CIN2 ORF and additional 3' UTR were amplified by PCR. The 5' primer (TAGGCCGTCGACATGGACTTTACTGCGAAGATAAAGGGTA) contained a SalI site and the 3' primer (CGACTAGCGGCCGCCTATAAGTAAGCGCGAAACAACTGCA) contained a NotI site. The PCR product was digested with SalI and NotI and ligated into pRS316-GAL1 plasmid (LIU et al. 1992 ). The contruct (pAS56) suppresses the benomyl supersensitivity of cells deleted for CIN2.

To construct the GAL-CIN4 HIS3 CEN plasmid, the CIN4 ORF and additional 5' and 3' UTR were amplified by PCR. The 5' primer (GCCGGATCCATGGGACTACTAAGTATTATC) contained a BamHI site and the 3' primer (CGGCCGCGGGTAATGAACTATCACGC) contained a SacII site. The PCR product was digested with BamHI and SacII and ligated into pRS316-GAL1 plasmid (LIU et al. 1992 ). The construct (pJF18) suppresses the benomyl supersensitivity of cells deleted for CIN4.

Characterization of cin1, rbl2 double mutants:
A cin1:: hisG/CIN1, rbl2::hisG-URA3-hisG/RBL2 diploid containing a RBL2 covering plasmid marked by HIS3 was sporulated. The cin1 null allele was followed by its benomyl supersensitivity and the rbl2 deletion allele by the URA3 marker. Haploid segregants harboring both null alleles and the covering plasmid were transformed with either a pCIN1 URA3 or a pRBL2 URA3 plasmid. Cells were grown in YPD, plated to SC -ura glucose plates, and replica plated to SC -his glucose plates. Strains that had lost the HIS3-marked plasmid were plated to 5-FOA plates to select for loss of the URA3-marked covering plasmid.

Interactions of -tubulin mutant alleles with overproduced CIN1:
tub1, tub3 strains containing mutant alleles of the TUB1 gene on LEU2:CEN plasmids (listed in Table 2) were transformed with pJF10 and YCpGAL. These strains were grown to saturation overnight in SC -ura glucose liquid media. The cultures were serial diluted in 96-well dishes and spotted onto SC -ura galactose plates containing 10 µg/ml benomyl, onto SC -ura galactose plates incubated at 25° (a semipermissive temperature for the growth of tub1-724 mutant strains), and also onto galactose and glucose plates at 30° as a growth control.

View this table: In this window In a new window   Table 2. -Tubulin alleles synthetic lethal with nulls in RBL2, C1N1, and PAC2

Effect of CIN1 overproduction on excess Pac2p tub1-724 lethality:
tub1, tub3 strains containing the tub1-724 mutant -tubulin allele on a LEU2 CEN plasmid were transformed with the following plasmids: pGAL-CIN1 CEN URA3 and pGAL-PAC2 CEN LYS2 (JFY475), pGAL-CIN1 CEN URA3 and pCEN LYS2 (JFY476), or pCEN URA3 and pGAL-PAC2 CEN LYS2 (JFY477). The strains were grown overnight in SC -lys -ura glucose liquid media. The cultures were serial diluted in 96-well dishes and spotted onto SC -lys -ura galactose plates. Cells were also spotted onto glucose plates as a growth control.

Overproduction of combinations of the CIN genes in tub1-724:
JFY531 was transformed with the following combinations of plasmids: pGAL-CIN1 CEN HIS3 and pGAL-CIN2 CEN URA3 (JFY525), pGAL-CIN1 CEN HIS3 and pGAL-CIN4 CEN URA3 (JFY526), pGAL-CIN1 CEN HIS3 and pGAL CEN URA3 (JFY527), pCEN HIS3 and pGAL-CIN2 CEN URA3 (JFY528), pCEN HIS3 and pGAL-CIN4 CEN URA3 (JFY529), or pCEN HIS3 and pGAL CEN URA3 (JFY530). The strains were grown overnight in SC -his -ura -leu glucose liquid media. The cultures were serial diluted in 96-well dishes, spotted onto SC -his -ura -leu galactose plates, and incubated at various temperatures. Cells were also spotted onto glucose plates as a growth control.

Construction of GAL-CIN1-HA and GAL-CIN1-HA-His₆:
The 3' third of the CIN1 open reading frame was amplified using PCR. The 5' primer (5'-GATGTAGGACGTCTGGTAAGAATACAGGC) contained an AatII site. Two 3' primers were used. To make the GAL-CIN1-HA construct we used the 3' primer (5'-CTCACCGCGGCTAGCGGCCGCCTAAAGTGATATCAGACTCTAATATATTCGC) containing a NotI site followed by two stop codons and a SacII site. To make GAL-CIN1-HA-His₆ we used the 3' primer (5'-CTCACCGCGGCTAGTGATGGTGATGGTGATGGCGGCCGCCTAAAGTGATATCAGACTCTAATATATTCGC) containing six in-frame histidine residues, a NotI site, two stop codons, and a SacII site. The PCR products were ligated into pT7-Blue plasmid (Novagen). The AatII SacII fragments were then ligated into pJF10 to create pJF11 and pJF12, respectively. A 111-bp NotI fragment containing the triple HA epitope from B2385 (provided by G. Fink, MIT) was cloned into the NotI site of pJF11 and pJF12 to create pJF14 and pJF15, respectively. These alleles of CIN1 rescue the benomyl phenotype of cin1 cells and suppress the tub1-724 phenotypes as well as does the unmodified CIN1.

In vivo Cin1p-HA-His6 and Pac2p-HA-His6 association experiments:
We grew yeast strains overnight in selective raffinose media at 30°. Galactose (2%) was added to induce the tagged constructs for 4 hr. A total of 6.0 x 10⁹ cells (log phase) were harvested by glass bead lysis per experiment in 1.1 ml PME buffer plus protease inhibitors. We applied 1 ml of protein extract to 25 µl Ni-NTA beads. We washed and eluted the bound proteins as previously described (MAGENDANTZ et al. 1995 ). Eluted proteins were subjected to SDS-PAGE analysis and probed for -tubulin, ß-tubulin, and HA(12CA5). For Cin1p-ß-tubulin association experiment we used strains JFY470 (pGAL1-10 CIN1-His₆-HA CEN URA3) and JFY471 (YCpGAL). For Pac2p-Cin1p association experiment we used strains LTY564 (pGAL1-10 PAC2-His₆-HA CEN LYS2, YCpGAL), LTY566 (pGAL1-10 PAC2-His₆-HA CEN LYS2, pGAL1-10 CIN1-HA CEN URA3), JFY478 (YCpGAL, pGAL1-10 PAC2-HA CEN LYS2), and JFY481 (pGAL1-10 CIN1-HA CEN URA3, pGAL1-10 PAC2-HA CEN LYS2).

	RESULTS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Mutations in the CIN genes are sensitive to Rbl2p overproduction:
We searched for mutations that affect the formation of tubulin heterodimers. Our strategy was based on the observation that overexpression of the ß-tubulin binding protein Rbl2p kills tub1-724 cells expressing an -tubulin with relatively weak affinity for ß-tubulin (ARCHER et al. 1995 ; VEGA et al. 1998 ). This lethal interaction is probably explained by formation of the Rbl2p-ß-tubulin complex at the expense of heterodimer containing the mutant -tubulin protein. We reasoned that other mutations affecting heterodimer stability or formation might be similarly affected by excess Rbl2p.

We mutagenized wild-type haploid cells containing a GAL-RBL2 plasmid (pA5) to find mutants that could not survive when RBL2 was overexpressed (see MATERIALS AND METHODS). This screen identified one mutant that was unable to live when overproducing Rbl2p. The mutant strain is extremely sensitive to benomyl and cold sensitive for growth, two properties associated with many mutations affecting microtubules in yeast.

By several criteria, we demonstrated that this mutation is an allele of CIN1. A library plasmid containing the entire CIN1 open reading frame rescued both the benomyl supersensitivity of the mutant and the lethality upon RBL2 overexpression. To confirm that loss of cin1 function is sufficient to confer these phenotypes, we deleted the entire open reading frame of CIN1 by integrative transformation in a wild-type strain and tested the effect of RBL2 overexpression. As shown in Fig 1, cin1 cells overproducing Rbl2p start to lose viability 4 hr after induction and, after 12 hr, <0.1% of the cells are viable. Finally, we confirmed that the original mutation is indeed an allele of CIN1 by both complementation and linkage analysis (data not shown) using a cin1 null allele (STEARNS et al. 1990 ).

View larger version (14K): In this window In a new window Download PPT slide   Figure 1. The lethality of RBL2 overexpression is enhanced in CIN1 and PAC2 nulls. cin1 (), pac2 (), and wild-type strains () containing a GAL-RBL2 plasmid were grown overnight in selective noninducing media. At t = 0 hr, Rbl2p overproduction was induced by addition of galactose to 2%. Cell viability is determined as the percentage of cells able to form colonies on glucose plates. Results shown are representative of three independent trials.

The S. cerevisiae genes CIN2 and CIN4 have functions similar to that of CIN1, and their products are thought to act as components of a complex or in a common pathway (STEARNS et al. 1990 ). We tested whether that functional relationship extends to interaction with Rbl2p. Overexpression of Rbl2p in cin2 and cin4 null strains also causes loss of viability (data not shown). The results suggest that the common microtubule-related functions of these three CIN genes are affected by Rbl2p levels. However, Cin2p or Cin4p homologues have not been identified in the in vitro heterodimerization reaction.

We also tested for interactions between Rbl2p levels and the absence of the two other yeast homologues of the in vitro factors, Pac2p/cofactor E and Alf1p/cofactor B. Pac2p/cofactor E binds to -tubulin in vitro (TIAN et al. 1997 ), in vivo (VEGA et al. 1998 ), and by two-hybrid assay (FEIERBACH et al. 1999 ). In the in vitro assay, it participates in a quaternary complex with both tubulin polypeptides and Cin1p/cofactor D, which dissociates in the presence of cofactor C to release heterodimer. As shown in Fig 1, overexpression of Rbl2p in cells lacking Pac2p causes rapid loss of viability, at a rate comparable to that of cells lacking Cin1p. In contrast, there is no effect of excess Rbl2p in cells that lack Alf1p/cofactor B (data not shown).

Rbl2p-ß-tubulin formation and microtubule depolymerization in pac2 and cin1 cells overexpressing Rbl2p:
The lethality of excess Rbl2p in tub1-724 cells is accompanied by enhanced formation of Rbl2p-ß-tubulin complex and loss of assembled microtubules (VEGA et al. 1998 ). We tested whether the lethality of excess Rbl2p in cin1 and pac2 cells has the same properties.

To assay the levels of Rbl2p-ß-tubulin complex, extracts were prepared from cells transformed with a plasmid encoding His₆-Rbl2p under the control of the galactose promoter and grown for 4 hr in inducing medium. His₆-Rbl2p and bound proteins were specifically purified on nickel-agarose beads. The amount of ß-tubulin associated with the His₆-Rbl2p fraction was two- to threefold higher in pac2 and cin1 cells than in wild-type cells (Fig 2). For comparison, the formation of His₆-Rbl2p-ß-tubulin complex in tub1-724 cells under the same conditions is fivefold greater than that in wild type (VEGA et al. 1998 ). There is no significant binding of -tubulin to His₆-Rbl2p in any of the strains (data not shown).

View larger version (18K): In this window In a new window Download PPT slide   Figure 2. Enhanced formation of Rlb2p-ß-tubulin complex in CIN1 and PAC2 nulls. Protein extracts from cin1, pac2, and wild-type strains containing a GAL-RBL2-HIS6 plasmid were obtained from cells grown 3 hr in selective inducing media. The tagged Rbl2p and bound proteins were purified using nickel-agarose. Nickel eluates were analyzed by immunoblotting. The Rbl2p and ß-tubulin signals were quantitated by densitometry and normalized to Rbl2p signal. The values are expressed as fold increase above the wild-type control. Error bars represent the standard deviation of three independent trials.

The effect on microtubule structures was assayed by immunofluorescence after 3 hr of Rbl2p overexpression (Fig 3). At this time, 78% of wild-type cells have intranuclear microtubules, 18% show a dot representing the spindle pole body, and 4% have no detectable staining. In contrast, only 25% of either cin1 or pac2 cells overexpressing Rbl2p have short or long spindles; the remainder have either single dots (28% for cin1; 38% for pac2) or no staining at all.

View larger version (54K): In this window In a new window Download PPT slide   Figure 3. Microtubule staining in cin1 and pac2 cells overexpressing RBL2. After 3 hr of RBL2 overexpression, cin1 (two independent trials), pac2 (four independent trials), and wild-type cells were processed for tubulin immunofluorescence using an anti-ß-tubulin antibody and with DAPI to stain nuclei.

Synthetic interactions of cin1 and pac2:
Previous work established that RBL2 is essential in a small subset of specific -tubulin mutants (ARCHER et al. 1995 ). To test if cin1 and pac2 have similar interactions, we constructed strains bearing null alleles of each of those genes and a plasmid expressing wild-type -tubulin (marked with URA3) as their major source of -tubulin. We transformed those strains with each of 12 different tub1 mutants and then used the drug 5-FOA to identify those mutants that could not lose the plasmid carrying wild-type -tubulin. We found that 5 of the 12 -tubulin mutations tested do not support growth of pac2 and cin1 cells (Table 2). Significantly, 4 of these 5 mutants are also synthetically lethal with rbl2; among these is tub1-724, which encodes an -tubulin defective in ß-tubulin binding (VEGA et al. 1998 ). In contrast, several other -tubulin alleles do not interact with any of these 3 deletion mutants. The results suggest that PAC2, CIN1, and RBL2 affect related functions.

As noted above, none of these three genes is essential. In addition, cells containing each of the pairwise combinations of the three mutations are viable. FEIERBACH et al. 1999 reported that they could not recover cin1, rbl2 cells. Since Cin1p/cofactor D and Rbl2p/cofactor A perform partially redundant functions in the in vitro assay, together they might define an essential function in vivo. However, as shown in Fig 4, haploid strains deleted for both cin1 and rbl2 and containing a plasmid with a genomic copy of either CIN1 or RBL2 marked with the URA3 gene grow in the presence of the drug 5-FOA, which selects for loss of the covering plasmid. We can also recover double mutants without a plasmid containing either wild-type gene from sporulated cells (data not shown).

View larger version (39K): In this window In a new window Download PPT slide   Figure 4. Cells lacking both cin1 and rbl2 are viable. cin1 rbl2 haploid cells bearing URA3-CEN plasmids that express either CIN1 (JFY232) or RBL2 (JFY234) were plated to medium containing 5-FOA. Only cells that can lose the URA3-CEN plasmid can grow on this medium, which selects for cells that can lose the plasmid. Both double-mutant strains grew as well on this medium as the control, a wild-type strain (FSY183) containing the YCp50 CEN-URA3 plasmid.

The effect of CIN gene overexpression on tub1-724 mutant phenotypes:
We previously demonstrated that the conditional lethality in strains expressing tub1-724 is a consequence of dissociation of the unstable tubulin heterodimer to produce free ß-tubulin (VEGA et al. 1998 ; see DISCUSSION). Consistent with that conclusion, high levels of either Rbl2p/cofactor A or Pac2p/cofactor E kill tub1-724 haploid cells (ARCHER et al. 1995 ; VEGA et al. 1998 ).

Overexpression of CIN genes in tub1-724 cells yields a dramatically different result. We introduced a plasmid containing CIN1 under control of the GAL promoter into the tub1-724 mutant strain and monitored cell growth under various conditions. Excess Cin1p suppresses the growth defect of tub1-724 cells at 25°, a semipermissive temperature for this mutant (Fig 5A). In addition, excess Cin1p suppresses the lethality of PAC2 overexpression in these cells (Fig 5B). This effect of extra Cin1p is specific for the tub1-724 mutant; it has no effect on the cold-sensitive growth of the other -tubulin alleles listed in Table 2.

View larger version (30K): In this window In a new window Download PPT slide   Figure 5. Overexpression of CIN1 is able to rescue conditional phenotypes of the tub1-724 mutant. (A) Stationary phase cultures of tub1-724 mutant cells (FSY157) containing either GAL-CIN1 or a YCpGAL control plasmid were serially diluted (one-sixth dilutions for the 25° plate, one-fourth for the benomyl plate), spotted to selective galactose plates with or without 20 µg/ml benomyl, and incubated at 25°. (B) Stationary phase cultures of tub1-724 mutant cells (FSY157), each containing two plasmids, were serially diluted (one-fourth dilutions) and spotted to selective galactose plates. The plasmids were as follows: GAL-CIN1 (pJF17) and GAL-PAC2 (pJF16) (left column), GAL-CIN1 and pRS317 (middle column), and GAL-PAC2 and YCpGAL (right column). (C) Stationary phase cultures of tub1-724 mutant strains containing GAL-CIN1 plus the pRS313 control (first row), GAL-CIN2 or GAL-CIN4 plus the YCpGAL control plasmid (second and third rows), or control plasmid alone (fourth row) were serially diluted (one-fourth dilutions) and plated to selective galactose plates at 25°. (D) Stationary phase cultures of tub1-724 mutant strains containing GAL-CIN1 plus GAL-CIN2 or GAL-CIN4, or each of the individual plasmids, were serially diluted (one-fourth dilutions) and plated to selective galactose plates at 23°.

In principle, a possible explanation for this suppression is that Cin1p sequesters free ß-tubulin and so interferes with its toxicity. Indeed, Cin1p/cofactor D does bind to ß-tubulin in vitro (TIAN et al. 1996 ) and in vivo (see below). Contrary to that hypothesis, however, overexpression of CIN1 does not rescue cells from excess ß-tubulin in two other contexts. First, overexpression of Cin1p, unlike overexpression of either Rbl2p or -tubulin, does not rescue excess ß-tubulin lethality. For example, the plating efficiency on galactose of cells containing a single GAL-TUB2 gene is 0.01% in the presence or absence of a GAL-CIN1 plasmid (data not shown), but increases to 70% when either GAL-RBL2 or GAL-TUB1 are present (ARCHER et al. 1995 ). Second, overexpression of CIN1 does not rescue the benomyl supersensitivity of a tub3 strain (HOYT et al. 1997 and J. FLEMING, unpublished results), which has a modest constitutive excess of ß-tubulin. Significantly, the benomyl and cold sensitivities of tub3 cells are markedly less severe than those of tub1-724 cells, suggesting that the levels of excess ß-tubulin are higher in the latter strain. Therefore, the differential suppression by Cin1p overproduction in these two mutants is not a consequence of the relative levels of free ß-tubulin. Instead, the distinction between the one situation in which excess Cin1p does suppress and the two in which it does not can be rationalized on the basis of levels of -tubulin. tub1-724 cells contain stoichiometric - and ß-tubulin, while both TUB2 overexpressers and tub3 cells contain an excess of ß-tubulin.

Since CIN2 and CIN4 have properties similar to CIN1 in other assays, we tested them for interaction with tub1-724. Overexpression of CIN1, CIN2 but not CIN4 partially rescues the tub1-724 growth phenotype at 25° (Fig 5C, lanes 2 and 3). At 23°, none of the overexpressed CIN genes has a significant effect on tub1-724 growth (Fig 5D). However, overexpression of CIN1 in combination with either CIN2 or CIN4 both significantly enhance cell growth. The results suggest that, under restrictive conditions, the expression levels of each of the CIN genes can be limiting for growth.

Physical interactions of Cin1p:
To assay yeast cell extracts for the protein complexes suggested by the in vivo results above and the in vitro data, we made a HA-His₆-tagged version of Cin1p behind control of the GAL promoter (pJF15). Fig 6A shows that ß-tubulin but not -tubulin specifically copurifies with the tagged Cin1p (five independent trials). In contrast, there is no detectable enrichment of -tubulin among the proteins eluted with Cin1p. Formation of the Cin1p-ß-tubulin-containing complex is independent of Pac2p (data not shown). This result suggests that Cin1p can bind directly or indirectly to ß-tubulin but not -tubulin in vivo, similar to Rbl2p. The significance of this complex is analyzed in DISCUSSION. To assay for a complex between Cin1p and Pac2p, we overexpressed HA-tagged versions of both proteins. Using affinity chromatography, we show that Pac2p copurifies with the His₆-tagged Cin1p and, conversely, that Cin1p specifically copurifies with a His₆-tagged version of Pac2p (Fig 6B).

View larger version (16K): In this window In a new window Download PPT slide   Figure 6. Interactions among Cin1p, Pac2p, and tubulin. (A) Extracts from wild-type strains containing either GAL-CIN1-HA-His6 or YCpGAL control plasmid were obtained after 4 hr of growth in inducing media. The whole cell extract (WCE) was probed directly for the tubulin polypeptides and Cin1p, or applied to Ni-NTA columns (see MATERIALS AND METHODS) to purify the tagged Cin1p and proteins bound to it. The immunoblots were probed with antibodies against -tubulin, ß-tubulin, and the HA epitope tag (for HA-Cin1p). The data demonstrate that the -tubulin and ß-tubulin signals are comparable in the whole cell extracts, but that ß-tubulin binds preferentially to Cin1p. (B) Cin1p and Pac2p bind in vivo. Cells containing the plasmids indicated were grown on galactose for 4 hr, and extracts were incubated with Ni-NTA beads, eluted, and assayed by immunoblotting. Cin1p and Pac2p coelute from beads when either of the proteins bears a His-6 tag. The results shown are representative of nine independent trials using His-tagged Pac2p and two independent trials using His-tagged Cin1p.

	DISCUSSION

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

The data presented here identify genes that affect tubulin dimer formation in vivo. Previous work demonstrated that Rbl2p can compete with -tubulin for binding to ß-tubulin and so disrupt the heterodimer. Therefore, an -tubulin mutant protein (Tub1-724p) that has reduced affinity for ß-tubulin is supersensitive to Rbl2p overexpression. Here we show that mutations in four nontubulin genes—CIN1, 2, 4, and PAC2—similarly enhance the lethality of excess Rbl2p. Their function is identified by their interaction with the mutant -tubulin protein, Tub1-724p. The conditional lethality of that mutant is due to the dissociation of tubulin heterodimer to produce free toxic ß-tubulin (VEGA et al. 1998 and see below). Both Pac2p and Cin1p are required in tub1-724 cells, and overexpression of the CIN genes suppresses tub1-724 phenotypes. The results provide evidence that the products of these four genes participate in tubulin heterodimerization in vivo. Below, we discuss the activities of these different genes in vivo and how they correlate with the properties of their mammalian homologues in in vitro assays. We consider the evidence supporting their roles in de novo formation of heterodimer, and we propose an alternative, salvage pathway that acts on the products of dissociated heterodimer to protect the cell against accumulation of toxic free ß-tubulin.

Consequences of Rbl2p overproduction in cin1 and pac2 nulls:
We report three consequences of overexpressing RBL2 in cin1 and pac2 mutant cells: loss of cell viability, loss of assembled microtubules, and enhanced formation of Rbl2-ß-tubulin complex. These three phenotypes also occur when RBL2 is overexpressed in tub1-724 cells (ARCHER et al. 1995 ; VEGA et al. 1998 ). In the latter instance, the phenotypes are all readily interpreted given that the mutant -tubulin forms a heterodimer that is less stable than wild type. However, as Cin proteins and Pac2p are not stoichiometric components of the tubulin heterodimer, they obviously do not directly stabilize the heterodimer. An alternative explanation is that these proteins act to oppose the destabilization of heterodimer by facilitating heterodimer formation. Although not normally essential, they become essential (STEARNS et al. 1990 ; HOYT et al. 1997 ; Fig 5 above) when microtubule assembly is inhibited by depolymerizing drugs or growth at low temperature and the level of heterodimer is correspondingly increased.

Effects of Cin1p and Pac2p levels in tubulin mutants:
Null alleles of cin1, pac2, and rbl2 are lethal in combination with the same four cold-sensitive -tubulin mutants, suggesting that the mutants all affect a related function. Although those four mutants all arrest with no microtubules (SCHATZ et al. 1988 ), several other mutants with the same phenotype are not lethal in combination with cin1, pac2, or rbl2.

The molecular defect of one of the mutants is well understood: the -tubulin in tub1-724 cells forms a less stable heterodimer (VEGA et al. 1998 ). Significantly, tub1-724 is semidominant: TUB1/tub1-724 heterozygotes show enhanced sensitivity to low temperature and microtubule depolymerizing drugs. Although overexpression of the ß-tubulin binding protein Rbl2p is lethal in tub1-724 cells, it suppresses the phenotypes of the heterozygote. This result demonstrates that the phenotypes conferred by the Tub1-724p mutant -tubulin in the heterozygote can all be ascribed to the free ß-tubulin released by dissociation of the mutant heterodimer. This finding is consistent with, but does not prove, the hypothesis that the conditional defect in tub1-724 haploid cells also is a consequence of heterodimer dissociation and the formation of free ß-tubulin. This mutation may confer other defects in microtubule morphogenesis, but since the release of free ß-tubulin is sufficient for cell death, the suppression by excess Cin1p of tub1-724 haploids means that the excess Cin1p must at least suppress the formation of free ß-tubulin or its downstream effects.

This property of tub1-724 provides insight into the activities of PAC2 and CIN1. The most informative phenotypes are those of tub1-724 cells overexpressing either Pac2p or Cin1p. Excess Pac2p is lethal. We previously showed that Pac2p is an -tubulin binding protein (VEGA et al. 1998 ); its overexpression in the presence of a weakened heterodimer would be expected to release toxic free ß-tubulin.

Similarly, the fact that Cin1p binds ß-tubulin (Fig 6A) leads to the expectation that its overexpression would be lethal in tub1-724 cells, similar to overexpressed Rbl2p. However, we found that excess Cin1p actually suppresses the conditional phenotypes of this mutant -tubulin. In addition, excess Cin1p does not rescue the phenotypes associated with excess ß-tubulin in tub3 (HOYT et al. 1997 ) or ß-tubulin overproducing strains (data not shown). Taken together, these observations suggest that excess Cin1p does not act like Rbl2p to protect cells from excess ß-tubulin. Moreover, the phenotypes of the tub1-724 mutation are more severe than those of a strain deleted for tub3, suggesting that tub1-724 cells have more free ß-tubulin. A distinct difference between these two mutations is that the tub1-724 cells contain a pool of undimerized -tubulin stoichiometric with the ß-tubulin, while the tub3 and GAL-TUB2 strains do not. This distinction may explain why overexpressed Rbl2p but not Cin1p rescues high-level overexpression of ß-tubulin. Thus, Cin1p suppressing activity is not explained by its ß-tubulin binding activity alone.

Binding partners of Cin1p:
The results above also demonstrate physical interactions that are consistent with the functional data and with the in vitro data. Fractionation experiments performed using Cin1p and Pac2p have allowed us to characterize their possible interactions in vivo. We show that, as for the mammalian cofactor D, ß-tubulin but not -tubulin copurifies with Cin1p when it is overexpressed in wild-type cells. That complex is detected in the presence or absence of Pac2p. In addition, we have also shown that Cin1p and Pac2p associate in vivo, an interaction not detected using the two-hybrid assay (FEIERBACH et al. 1999 ).

Comparison of in vivo and in vitro properties of tubulin binding proteins:
Cowan and colleagues identified purified protein factors that can help catalyze incorporation of ß-tubulin and -tubulin into exogenous heterodimer (GAO et al. 1992 , GAO et al. 1993 ; TIAN et al. 1996 , TIAN et al. 1997 ). In that assay system, ß-tubulin released from the chaperone is bound by either cofactor A/Rbl2p or cofactor D/Cin1p, and -tubulin is bound by either cofactor B/Alf1p or cofactor E/Pac2p. The ß-tubulin released from cofactor A and the -tubulin released by cofactor B must bind cofactors D and E, respectively, before they can be incorporated into heterodimer. Cofactor D-ß-tubulin and cofactor E--tubulin form a quaternary complex. Cofactor C is believed to mediate the release of -ß tubulin heterodimer. It is not known whether these protein cofactors, required under the conditions of this assay, are essential in vivo or whether they account for all the activity found in the original extracts.

Four of these mammalian cofactors are homologous to yeast genes: Cofactor D shows 21% identity with Cin1p (HOYT et al. 1997 ); cofactor E is 30% identical to Pac2p (HOYT et al. 1997 ); cofactor A is structurally and functionally homologous to Rbl2p (ARCHER et al. 1995 ); and cofactor B is 32% identical to Alf1p (TIAN et al. 1996 ). In the in vitro assay, cofactors D and E are essential. However, none of the homologous yeast genes are essential. Several groups also have demonstrated that even several pairwise deletions—CIN1 and PAC2 (HOYT et al. 1997 and our unpublished results), RBL2 and PAC2 (our unpublished results), and ALF1 in pairwise combinations with each of the other three genes (FEIERBACH et al. 1999 )—do not define an essential function. Also, in contrast to a previous report (FEIERBACH et al. 1999 ), we find that cells deleted for both cin1 and rbl2 are viable (see Fig 4 above). These results argue strongly that this cofactor-dependent pathway is not essential for making tubulin heterodimers in vivo. There may be redundant functions in yeast specified by genes as yet undetected.

Several of the functional interactions we detect in vivo among these proteins are consistent with this in vitro model. However, other results demonstrate significant differences between the in vivo and in vitro situations. Most important, suppression by excess Cin1p/cofactor D of a mutant -tubulin with lowered heterodimer stability directly contradicts the in vitro model: Cofactor D in vitro disrupts the heterodimer to form a cofactor D-ß-tubulin complex (TIAN et al. 1997 ). Also, Cin1p activity in vivo does not require stoichiometric Pac2p: Overexpression of Cin1p alone is sufficient to suppress tub1-724 and can rescue the benomyl supersensitive phenotype of pac2 strains (HOYT et al. 1997 ). These in vivo results suggest that Cin1p action is not confined to bringing ß-tubulin into a quaternary complex containing Pac2p and -tubulin, as hypothesized for the in vitro action of cofactor D. Finally, unlike the in vivo situation, to date no role for Cin2p or Cin4p vertebrate homologues has been reported for the in vitro assay.

Significantly, our results demonstrate that the interactions of Rbl2p and Cin1p with ß-tubulin in vivo must be quite different; excess of the former kills tub1-724 cells, while the latter suppresses the same mutation. This distinction contrasts with the in vitro results suggesting that Rbl2p and Cin1p can bind equivalent forms of ß-tubulin.

The in vivo data clearly demonstrate that the Cin1p-ß-tubulin and Rbl2p-ß-tubulin complexes are functionally quite different. For example, excess Rbl2p suppresses the lethality of free ß-tubulin, but excess Cin1p does not. Conversely, Cin1p interacts with ß-tubulin to promote heterodimer assembly, but Rbl2p does not.

A pathway for rescuing dissociated tubulin heterodimers:
The experiments presented here demonstrate that proteins that interact with individual tubulin polypeptides can influence the formation of heterodimer in vivo. Especially in the case of Cin1p, the relationship between phenotype and expression levels suggests that this protein acts to promote -ß-tubulin complex formation and not merely by binding reversibly to free ß-tubulin. Such an activity will require coupling to a highly exergonic step to make the reaction act as if it were unidirectional. A candidate for that coupling factor is Cin4p: It has a predicted GTP binding motif (HOYT et al. 1997 ) and it enhances the ability of Cin1p to rescue tub1-724 cells (Fig 5D).

Since neither CIN1 nor PAC2 is essential in otherwise wild-type cells, it is unlikely that they are important in the major pathway of de novo heterodimer formation. An alternative possibility is suggested by the fact that they become essential under conditions that would be expected to increase levels of free tubulin polypeptides. First, both are essential in tub1-724 cells, in which the mutant -tubulin destabilizes the heterodimer and favors its dissociation. Second, cells deleted for cin1, cin2, cin4, or pac2 are supersensitive to low temperature and benomyl, two treatments that cause microtubule depolymerization; the increased level of heterodimer produced by microtubule disassembly will in turn generate an increased steady state level of free tubulin polypeptides as a result of dissociation. Third, both cin1 and pac2 mutant cells show enhanced formation of Rbl2p-ß-tubulin complex upon Rbl2p overexpression. Finally, pac2 and cin1 are synthetically lethal with pac10, which encodes a nonessential tubulin folding factor (ALVAREZ et al. 1998 ). Improper folding of - and ß-tubulin could also lead to higher levels of undimerized tubulin polypeptides.

A possible function for these genes is that they facilitate reassociation of tubulin polypeptides to reform heterodimer, thus protecting the cell against free tubulin polypeptides and maintaining required levels of the heterodimer itself. In contrast to the in vitro assay for tubulin heterodimer formation from denatured tubulin, in which the mammalian homologues of Cin1p and Pac2p are essential, facilitated heterodimer formation may be only conditionally required in vivo. The roles of Pac2p/cofactor E and Cin1p/cofactor D in such a mechanism are consistent with many of their activities in the in vitro assay. The data also predict that Cin2p and Cin4p would enhance Cin1p-mediated heterodimer formation in an in vitro assay. We cannot test directly whether levels of tubulin heterodimer in tub1-724 cells change as a function of Cin1p expression, since the mutant heterodimer dissociates under the conditions required for its isolation (VEGA et al. 1998 ). However, detailed analysis of other -tubulin mutants that interact with CIN1 levels may provide more direct tests of this model.

	FOOTNOTES

¹ Present address: Department of Molecular Biology, Princeton University, Princeton, NJ 08544-1014.

	ACKNOWLEDGMENTS

We thank D. Botstein (Stanford), M. A. Hoyt (Johns Hopkins), and T. Stearns (Stanford) for plasmids and strains. We thank M. Magendantz, A. Rushforth, and other members of our laboratory; the members of MIT M&M; and A. Grossman (MIT) and S. Sanders (MIT) for valuable contributions. J.A.F. was supported in part by a training grant from National Institute of General Medical Science (NIGMS) to the Department of Biology, MIT. L.R.V. was supported in part by a predoctoral fellowship from Howard Hughes Medical Institute. This work was supported by a grant to F.S. from NIGMS.

Manuscript received November 4, 1999; Accepted for publication May 22, 2000.

	LITERATURE CITED

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

ALANI, E., L. CAO, and N. KLECKNER, 1987  A method for gene disruption that allows repeated use of URA3 selection in the construction of multiply disrupted yeast strains. Genetics 116:541-545[Abstract/Free Full Text].

ALVAREZ, P., A. SMITH, J. FLEMING, and F. SOLOMON, 1998  Modulation of tubulin polypeptide ratios by the yeast protein Pac10p. Genetics 149:857-864[Abstract/Free Full Text].

ARCHER, J., L. R. VEGA, and F. SOLOMON, 1995  Rbl2p, a yeast protein that binds to ß-tubulin and participates in microtubule function in vivo.. Cell 82:425-434[Medline].

ARCHER, J., M. MAGENDANTZ, L. VEGA, and F. SOLOMON, 1998  Formation and function of the Rbl2p-ß-tubulin complex. Mol. Cell. Biol. 18:1757-1762[Abstract/Free Full Text].

FEIERBACH, B., E. NOGALES, K. H. DOWNING, and T. STEARNS, 1999  Alf1p, a CLIP-170 domain-containing protein, is functionally and physically associated with alpha-tubulin. J. Cell Biol. 144:113-124[Abstract/Free Full Text].

GAO, Y., J. THOMAS, R. CHOW, G.-H. LEE, and N. COWAN, 1992  A cytoplasmic chaperonin that catalyzed ß-actin folding. Cell 69:1043-1050[Medline].

GAO, Y., I. VAINBERG, R. CHOW, and N. COWAN, 1993  Two cofactors and cytoplasmic chaperonin are required for the folding of - and ß-tubulin. Mol. Cell. Biol. 13:2478-2485[Abstract/Free Full Text].

GAO, Y., R. MELKI, P. WALDEN, S. LEWIS, and C. AMPE et al., 1994  A novel cochaperonin that modulates the ATPase activity of cytoplasmic chaperonin. Cell 125:989-996.

GEISER, J. R., E. J. SCHOTT, T. J. KINGSBURY, N. B. COLE, and L. J. TOTIS et al., 1997  Saccharomyces cerevisiae genes required in the absence of the CIN8-encoded spindle motor act in functionally diverse mitotic pathways. Mol. Biol. Cell 8:1035-1050[Abstract].

GEISSLER, S., K. SIEGERS, and E. SCHIEBEL, 1998  A novel protein complex promoting formation of functional - and -tubulin. EMBO J. 17:952-966[Medline].

HIRATA, D., H. MASUDA, M. EDDISON, and T. TODA, 1998  Essential role of tubulin-folding cofactor D in microtubule assembly and its association with microtubules in fission yeast. EMBO J. 17:658-666[Medline].

HOYT, M., T. STEARNS, and D. BOTSTEIN, 1990  Chromosome instability mutants of Saccharomyces cerevisiae that are defective in microtubule-mediated processes. Mol. Cell. Biol. 10:223-234[Abstract/Free Full Text].

HOYT, M. A., J. P. MACKE, B. T. ROBERTS, and J. R. GEISER, 1997  Saccharomyces cerevisiae PAC2 functions with CIN1, 2 and 4 in a pathway leading to normal microtubule stability. Genetics 146:849-857[Abstract].

LIU, H., J. KRIZEK, and A. BRETSCHER, 1992  Construction of a GAL1-regulated yeast cDNA expression library and its application to the identification of genes whose overexpression causes lethality in yeast. Genetics 132:665-673[Abstract].

MAGENDANTZ, M., M. HENRY, A. LANDER, and F. SOLOMON, 1995  Inter-domain interactions of radixin in vitro.. J. Biol. Chem. 270:25324-25327[Abstract/Free Full Text].

MELKI, R., H. ROMMELAERE, R. LEGUY, J. VANDEKERCKHOVE, and C. AMPE, 1996  Cofactor A is a molecular chaperone required for ß-tubulin folding: functional and structural characterization. Biochemistry 35:10422-10435[Medline].

PASQUALONE, D. and T. HUFFAKER, 1994  STU1, a suppressor of a ß-tubulin mutation, encodes a novel and essential component of the yeast mitotic spindle. J. Cell Biol. 127:1973-1984[Abstract/Free Full Text].

SCHATZ, P. J., F. SOLOMON, and D. BOTSTEIN, 1986  Genetically essential and nonessential -tubulin genes specify functionally interchangeable proteins. Mol. Cell. Biol. 6:3722-3733[Abstract/Free Full Text].

SCHATZ, P., F. SOLOMON, and D. BOTSTEIN, 1988  Isolation and characterization of conditional-lethal mutation in the TUB1 a-tubulin gene of the yeast Saccharomyces cerevisiae. Genetics 120:681-695[Abstract/Free Full Text].

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].

SOLOMON, F., L. CONNELL, D. KIRKPATRICK, V. PRAITIS and B. WEINSTEIN, 1992 Methods for studying the yeast cytoskeleton, pp. 197–222 in The Cytoskeleton, edited by K. CARRAWAY and C. CARRAWAY. Oxford University Press, Oxford.

STEARNS, T., M. HOYT, and D. BOTSTEIN, 1990  Yeast mutants sensitive to antimicrotubule drugs define three genes that affect microtubule function. Genetics 124:251-262[Abstract].

TIAN, G., Y. HUANG, H. ROMMELAERE, J. VANDEKERCKHOVE, and C. AMPE et al., 1996  Pathway leading to correctly folded ß-tubulin. Cell 86:287-296[Medline].

TIAN, G., S. LEWIS, B. FEIERBACH, T. STEARNS, and H. ROMMELAERE et al., 1997  Tubulin subunits exist in an activated conformational state generated and maintained by protein cofactors. J. Cell Biol. 138:821-832[Abstract/Free Full Text].

VEGA, L. R., J. FLEMING, and F. SOLOMON, 1998  An -tubulin mutant destabilizes the heterodimer: phenotypic consequences and interactions with tubulin binding proteins. Mol. Biol. Cell 9:2349-2360[Abstract/Free Full Text].

VELCULESCU, V., L. ZHANG, W. ZHOU, J. VOGELSTEIN, and M. BASRAI et al., 1997  Characterization of the yeast transcriptome. Cell 88:243-251[Medline].

WEINSTEIN, B. and F. SOLOMON, 1990  Phenotypic consequences of tubulin overproduction in Saccharomyces cerevisiae: differences between -tubulin and ß-tubulin. Mol. Cell. Biol. 10:5295-5304[Abstract/Free Full Text].

This article has been cited by other articles:

				A. R. Paredez, S. Persson, D. W. Ehrhardt, and C. R. Somerville Genetic Evidence That Cellulose Synthase Activity Influences Microtubule Cortical Array Organization Plant Physiology, August 1, 2008; 147(4): 1723 - 1734. [Abstract] [Full Text] [PDF]