A Screen for Dynein Synthetic Lethals in Aspergillus nidulans Identifies Spindle Assembly Checkpoint Genes and Other Genes Involved in Mitosis

Corresponding author: N. Ronald Morris, Department of Pharmacology, UMDNJ-Robert Wood Johnson Medical School, 675 Hoes Lane, Piscataway, NJ 08854-5635, morrisnr{at}rwja.umdnj.edu (E-mail).

Communicating editor: R. H. DAVIS

	ABSTRACT

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Cytoplasmic dynein is a ubiquitously expressed microtubule motor involved in vesicle transport, mitosis, nuclear migration, and spindle orientation. In the filamentous fungus Aspergillus nidulans, inactivation of cytoplasmic dynein, although not lethal, severely impairs nuclear migration. The role of dynein in mitosis and vesicle transport in this organism is unclear. To investigate the complete range of dynein function in A. nidulans, we searched for synthetic lethal mutations that significantly reduced growth in the absence of dynein but had little effect on their own. We isolated 19 sld (synthetic lethality without dynein) mutations in nine different genes. Mutations in two genes exacerbate the nuclear migration defect seen in the absence of dynein. Mutations in six other genes, including sldA and sldB, show a strong synthetic lethal interaction with a mutation in the mitotic kinesin bimC and, thus, are likely to play a role in mitosis. Mutations in sldA and sldB also confer hypersensitivity to the microtubule-destabilizing drug benomyl. sldA and sldB were cloned by complementation of their mutant phenotypes using an A. nidulans autonomously replicating vector. Sequencing revealed homology to the spindle assembly checkpoint genes BUB1 and BUB3 from Saccharomyces cerevisiae. Genetic interaction between dynein and spindle assembly checkpoint genes, as well as other mitotic genes, indicates that A. nidulans dynein plays a role in mitosis. We suggest a model for dynein motor action in A. nidulans that can explain dynein involvement in both mitosis and nuclear distribution.

DYNEINS are multisubunit minus end-directed, microtubule-based motor proteins that mediate a wide variety of motile processes in eukaryotic cells. Multiple isoforms of axonemal dynein are responsible for ciliary and flagellar movements. A ubiquitous cytoplasmic dynein/dynactin complex is thought to power movement and correct positioning of a variety of intracellular organelles (reviewed by HOLZBAUER and VALLEE 1994 ). As a retrograde motor for endomembranes in mammalian cells, dynein supports retrograde axonal vesicle transport in neurons (PASCHAL and VALLEE 1987 ), perinuclear positioning of the Golgi complex (CORTHESY-THEULAZ et al. 1992 ) and of lysosomes (LIN et al. 1994 ), apical transport of vesicles in kidney cells (LAFONT et al. 1994 ), and transport of endocytic vesicles (ANIENTO et al. 1993 ; ODA et al. 1995 ). In Xenopus egg extracts, dynein drives formation of endoplasmic reticulum networks (ALLAN 1995 ). Dynein may also be involved in slow anterograde transport in axons (DILLMAN et al. 1996 ). A role for dynein in mitosis was suggested by its localization to kinetochores and spindle microtubules (PFARR et al. 1990 ; STEUER et al. 1990 ). In animal cells, injection of antidynein antibodies prevents centrosome separation during mitosis (VAISBERG et al. 1993 ) and dynein is required for mitotic spindle assembly in vitro (reviewed by MERDES and CLEVELAND 1997 ). At least some of the diversity of dynein functions in mammalian cells is probably achieved by different heavy chain isoforms (VAISBERG et al. 1996 ).

A number of genetic systems are available for studying cytoplasmic dynein function and regulation in vivo, including the fruit fly Drosophila melanogaster, the budding yeast Saccharomyces cerevisiae, and the filamentous fungi Aspergillus nidulans and Neurospora crassa. In Drosophila, the dynein heavy chain is essential for early development (GEPNER et al. 1996 ). The most conspicuous effects of mutations that affect cytoplasmic dynein function in the fungi are on nuclear migration. Dynein mutations in S. cerevisiae affect the movement of the daughter nucleus into the bud after mitosis, producing binucleate mother cells with anucleate buds (LI et al. 1993 ; ESHEL et al. 1993 ). Similar mutations in the filamentous fungi A. nidulans and N. crassa prevent nuclei from distributing evenly along the germ tube (XIANG et al. 1994 ; PLAMANN et al. 1994 ; TINSLEY et al. 1996 ; BRUNO et al. 1996 ). This causes nuclear clumping and long segments of the mycelium that are anucleate. The site and mode of dynein action that leads to nuclear movement and a remarkably even nuclear distribution in filamentous fungi are not known (discussed in MORRIS et al. 1995 ; AIST 1995 ). Because filamentous fungi grow by hyphal tip elongation, a complete block of nuclear movement would be lethal. However, a dynein-null mutant of A. nidulans is viable although it is slow growing (XIANG et al. 1995C ). Thus, in A. nidulans, a system must operate to provide for nuclear movement in the absence of dynein. Similarly, in N. crassa dynein/dynactin mutants, as the germlings continue to grow, some nuclei escape from the spore end of the germtube and distribute in clusters within the hyphae (PLAMANN et al. 1994 ; ROBB et al. 1995 ). Dynein mutants of both A. nidulans and N. crassa grow at a slower rate, but it is not known whether this growth inhibition is caused exclusively by a nuclear migration defect.

The existence of a dynein-independent system for nuclear migration in A. nidulans is supported by the finding of dynein bypass suppressors. In a search for dynein mutation suppressors, mutations in five genes that improved growth of a dynein mutant and suppressed a nuclear migration defect were uncovered (GOLDMAN and MORRIS 1995 ). These mutations turned out to be bypass suppressors because they were able to improve growth in the absence of dynein heavy chain (XIANG et al. 1995A ). Similarly, destabilization of cytoplasmic microtubules can suppress a dynein deletion mutant (WILLINS et al. 1995 ).

Because dynein is clearly involved in mitosis in mammalian cells, it was initially surprising to find that dynein mutations did not prevent nuclear division in either budding yeast or the filamentous fungi. The lack of an absolute mitotic requirement for cytoplasmic dynein in S. cerevisiae has been explained by the finding that there are other motor proteins, the Cin8p and Kip1p kinesins, that contribute to the anaphase movement of chromosomes (SAUNDERS et al. 1995 ; GEISER et al. 1997 ). Presumably, a similar redundancy accounts for the ability of nuclei to divide in the A. nidulans and N. crassa dy-nein mutants. In S. cerevisiae, dynein functions in nuclear migration and anaphase movement of chromosomes by interacting with astral microtubules (YEH et al. 1995 ; CARMINATI and STEARNS 1997 ).

There is also circumstantial evidence that a dynein complex may be involved in vesicle transport both in A. nidulans and N. crassa. Dynein is localized to the tips of growing germlings in A. nidulans, where diverse vesicles accumulate (XIANG et al. 1995C ). In N. crassa, inactivation of the dynein/dynactin complex suppresses a cot-1 mutation that affects tip elongation (PLAMANN et al. 1994 ; TINSLEY et al. 1996 ; BRUNO et al. 1996 ). It has been suggested that inactivation of the dynein motor system prevents retrograde transport (i.e., transport from the site of growth at the tip) of some components that are critical for tip growth and that are limiting in the cot-1 mutant (PLAMANN et al. 1994 ).

To identify A. nidulans genes overlapping in function with dynein, we searched for mutations that were lethal when dynein synthesis was downregulated but were tolerated in a wild-type background. We identified nine genes required for the growth in the absence of dynein. At least six of these genes, including sldA and sldB, are likely to play a role in mitosis, because they show a strong synthetic lethal interaction with a mutation in the mitotic kinesin bimC. sldA and sldB were cloned and found to be homologs of S. cerevisiae spindle assembly checkpoint genes BUB1 and BUB3. The genetic interaction between dynein and genes with mitotic functions reveals an involvement of A. nidulans cytoplasmic dynein in mitosis.

	MATERIALS AND METHODS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Strains and growth conditions:
A. nidulans strains used in this study are listed in Table 1. The standard rich medium used was YAG (2% glucose, 0.5% yeast extract, trace elements, 2% agar). 0.6 M KCl was included in the medium to improve the growth and sporulation of mutant strains (YAGK). The same media without agar were used for liquid cultures (YG or YGK). The minimal media used were M-glucose (2% glucose, 71 mM NaNO₃, 7 mM KCl, 11 mM KH₂PO₄, 2 mM MgSO₄, trace elements, 2% agar, pH 6.5), M-glycerol (10 ml/liter glycerol instead of glucose), and M-ethanol (20 ml/liter ethanol instead of glucose). For some experiments, 5% yeast extract with 10 ml/liter glycerol and M-glycerol with 5% yeast extract were used. Media were supplemented with 5 mM uridine and 10 mM uracil (UU) to support growth of pyrG89 strains. Trace elements, vitamins, and other supplements were used as described (KAFER 1977 ). When necessary, 0.08% Na-deoxycholate was added to solid media to reduce the size of colonies (MACKINTOSH and PRITCHARD 1963 ).

Mutagenesis and synthetic lethality screen:
A suspension of 10⁷ conidia from strain XX61 (alcA(p)::nudA::pyr4) in 10 ml sterile water was irradiated with UV light with gentle agitation at 0.8 W/cm². The survival rate was 9%. Mutagenized spores were spread onto M-glycerol with 0.08% Na-deoxycholate to obtain 200–400 colonies per 10-cm petri dish after incubation for 4 days at 32°. Colonies were replica plated onto YAG and M-glycerol plates containing 0.08% Na-deoxycholate. A pin replicator (ROBERTS 1959 ) was used for quick replication of A. nidulans colonies. Our homemade pin replicator contained ~2500 thin metal pins per 10-cm petri dish. After incubation for 2 days at 43°, colonies that grew on M-glycerol but not on YAG were gridded onto YAG and M-glycerol plates at 32° and 43°. Mutants that grew well on inducing M-glycerol medium at both temperatures but not on repressing YAG medium (at both temperatures or only at 43°) potentially carried mutations that were lethal in the absence of dynein. These mutant strains were designated DG# (dead on glucose), where # is a strain number. The responsible mutations were designated sld# (synthetic lethality without dynein).

To eliminate mutants unable to use glucose or having other metabolic defects, DG strains were tested for their ability to grow on repressing media of different compositions. Yeast extract plus glycerol (no glucose), M-glucose (no yeast extract), and M-glycerol plus yeast extract were used as alternative repressing media. These media all caused XX61 to grow with a nud-like phenotype, although repression of dynein was not as tight as on YAG. Also, mutants were tested for growth on inducing medium M-ethanol at all temperatures.

In total, ~100,000 mutagenized colonies were replicated and 419 DG strains were selected for further analysis. Thirty-eight DG strains failed to grow or produced minute colonies on all repressing media but were able to grow on M-glycerol and M-ethanol. Mutants defective in glucose utilization or inhibited by yeast extract were discarded. The 38 mutant strains selected were purified by streaking several times to single colonies, and then used for crosses. Some strains either could not be crossed or we were not able to isolate the respective sld mutation in the wild-type background. After such strains were discarded, 19 DG mutants were left. They were characterized genetically as described below.

Genetic characterization of sld mutations:
A. nidulans genetic techniques, including heterokaryon construction, sexual crosses, diploid construction, and haploidization, were as described previously (PONTECORVO et al. 1953 ; CLUTTERBUCK 1974 ; KAFER 1977 ). nudA5 and nudF7 are temperature-sensitive alleles of the nudA and nudF genes (XIANG et al. 1995B , XIANG et al. 1995C ).

Construction of sld single mutants: To produce strains carrying sld mutations without a dynein mutation (designated by D preceded by strain number) and strains that carried sld mutations together with nudF7 (designated by DF preceded by number), DG mutants (sld; alcA(p)::nudA::pyr4; pyrG89; pyroA4; wA2) were crossed with strain XX21 (nudF7; pyrG89; yA2). Progeny carrying ts^- nudA5 or nudF7 single mutations were easily identified by the fact that they formed wild-type colonies at 32° and typical small, tight nud colonies at 43°, whereas strains carrying single sld mutations were provisionally identified by their reduced conidiation and somewhat reduced size at 43° on YAG. sld; nudF7 double mutants were tentatively identified by their almost lethal phenotype at 43° on YAG. All D and DF strains were selected to carry pyrG89 and yA2 mutations. The fact that the selected D and DF strains were unable to grow without uracil and uridine ensured that they had a wild-type copy of the nudA gene from XX21 and not the alcA::nudA::pyr4 recombinant gene from the DG strains. Further crosses confirmed that D strains carried appropriate sld mutations segregating as single genes. Each D (sld; pyrG89; yA2) strain was crossed to XX61 (sld⁺; alcA(p)::nudA::pyr4). The progeny were plated on M-glycerol plates at 32° and then gridded onto YAG and M-glycerol plates at 43°. Because both parental strains carried the pyrG89 mutation, only alcA(p)::nudA::pyr4 segregants could grow without uridine and uracil on M-glycerol. One half of these failed to form a sizable colony at 43° on repressing YAG plates and, hence, carried the sld mutation. Other pyr4⁺ segregants were phenotypically identical to strain XX61 and formed small colonies on repressing YAG plates. Analysis of more than 210 pyr4⁺ segregants from each D to XX61 cross showed 1:1 segregation of all the sld genes. To confirm the genotypes of the sld; nudF7 double mutants, they were crossed to the wild-type R153 (sld⁺; nudF⁺) strain. Appearance of sld and nudF7 single mutants was observed among 25% of the progeny.

Construction of sld; nudA5 double mutants: sld; nudA5 double mutants were constructed by mating each of the D strains (sld; pyrG89; yA2) with strain A5.1 (nudA5; pabaA1; chaA1). At least 225 segregants from each cross were analyzed, and the four expected phenotypes were observed in a 1:1:1:1 ratio. sld; nudA5 segregants were identified by their nearly lethal phenotype at 43° and confirmed by backcrosses to the R153 (sld+; nudA⁺) wild-type strain. All selected sld; nudA5 double mutants (designated DA with numbers) also contained pyrG89 and yA2 mutations.

Assigning sld mutations to genes: To catalog the sld mutations, each of the original DG (sld; alcA(p)::nudA::pyr4) strains was crossed to each of the D (sld; nudA⁺) strains. The progeny were plated onto M-glycerol at 32°. As both parents contained pyrG89, only segregants having the alcA(p)::nudA::pyr4 gene could grow without uridine and uracil. These were gridded onto YAG and M-glycerol plates at 43°. If two sld mutations were unlinked, sld⁺; alcA(p)::nudA::pyr4 recombinants that formed small tight colonies on YAG plates at 43° were observed. In the absence of recombination between the two sld genes all pyr4⁺ segregants were sld; alcA(p)::nudA::pyr4, and they failed to form a colony at 43° on YAG. Crosses between some sld strains produced no cleistothecia or sterile cleistothecia. We assumed that these strains carried sld mutations in the same gene that impaired sexual development in a recessive way. In all such cases, either sterile or no cleistothecia was found in crosses between the D and the corresponding DG strain carrying the same sld mutation. As expected, when crosses between D and DG strains carrying the same sld mutation were successful, no sld⁺; alcA(p)::nudA::pyr4 recombinants were found. Crosses between self-sterile strains assigned to different loci were always successful. No significant linkage was observed between the different sld loci, except for two loci represented by sldC1518 and sldI1444. The recombination frequency between these two mutations was ~13%.

Secondary phenotype analysis: Some sld single mutants exhibit reduced conidiation or grow slower than wild-type strains at 32°. That these phenotypes were linked to sld mutations was suggested by crosses of D strains with A5.1, DF strains with R153, and DA strains with R153. All segregants lethal at 43°, i.e., double mutants sld; nudA5 or sld; nudF7, had these defects at 32°. All sld strains have conspicuous defects at 43°, the most common being reduced sporulation. That these defects were caused by the sld mutations was suggested by crosses between D (sld) strains and XX61 (sld⁺; alcA(p)::nudA::pyr4). All pyr4⁺ segregants that were lethal on YAG at 43°, i.e., sld; alcA(p)::nudA::pyr4 double mutants, were aconidial on M-glycerol at 43° and vice versa. Mutations sldA744, sldA828, sldB937, and sldB1449 were found to confer hypersensitivity to benomyl. Benomyl (0.4 µg/ml) completely inhibited colony formation by these mutants at all temperatures. Conditional ts^- mutant sldA1084 was sensitive to benomyl only at 43°. The same benomyl concentration only slightly inhibited wild type and had no effect on nudA5 at the restrictive temperature. To prove that sensitivity to benomyl was linked to both the sld mutations and the conidiation defects, we analyzed crosses between the sldA744 and sldB937 single mutants and R153. All sld; nudA5 double mutants (no growth at 43°) and sld single mutants (aconidial at 43°) were hypersensitive to benomyl.

DNA techniques:
Standard molecular biology techniques (SAMBROOK et al. 1989 ) were used for DNA manipulations. Vectors pBluescript KS⁺(Stratagene, La Jolla, CA), pGEM-5Zf⁺ and pGEM-7Zf⁺ (Promega, Madison, WI) were used for subcloning and sequencing. Vector pHELP1 (GEMS and CLUTTERBUCK 1993 ) was a kind gift from Dr. A. J. CLUTTERBUCK. Subcloning efficiency DH5 competent cells (GIBCO BRL, Gaithersburg, MD) were used for routine transformations. Plasmids were propagated in Escherichia coli DH5 or JM109 and usually purified using the Qiaprep Spin Miniprep Kit (Qiagen, Chatsworth, CA). A. nidulans genomic DNA was isolated as described (RAEDER and BRODA 1985 ). Partial digestion of genomic DNA with Sau3AI and size fractionation on sucrose gradient were done according to standard protocols (SAMBROOK et al. 1989 ). DIG DNA Labeling and Detection Kit (Boehringer Mannheim, Indianapolis, IN) was used for cDNA library screening. Sequencing of double-stranded plasmid DNA was performed by REGINA FELDER at the Robert Wood Johnson Medical School sequencing facility on a 4000L sequencer (LICOR, Lincoln, NE) using cycle sequencing with the enzyme Sequitherm-Epicentre DNA polymerase.

A. nidulans transformation:
sldA and sldB genes were cloned by complementation of hypersensitivity to benomyl of the mutants sldA744 and sldB937. For transformation, fresh spores from strains 744D and 937Dn were germinated at a density of 5 x 10⁶ spores/ml in YGK medium with UU at 30°. Protoplast preparation and transformation were as described previously (OSMANI et al. 1987 ). Before transformation, protoplasts were incubated on ice overnight. Transformed protoplasts were plated in YAGK, UU medium with 0.4 µg/ml of benomyl and incubated at 43° for 3–4 days. Only wild-type transformants—not mutants—produced colonies at that concentration of benomyl.

We performed transformation using integrating and replicating vectors. Integrating vectors were constructed by subcloning of complementing genes into an E. coli cloning vector. These vectors can replicate in A. nidulans only after stable integration into the chromosomal DNA. About 0.5 µg of supercoiled plasmid was used to transform the mutant strain. If any number of transformants resistant to 0.4 µg/ml of benomyl was obtained, the fragment was considered to be able to complement the mutation in the integrative mode. We usually obtained 1–50 transformants using protoplasts prepared from ~5 x 10⁷ spores. As expected, all transformants were mitotically stable.

Transformation efficiency is increased by several orders of magnitude by cotransformation with an A. nidulans replicating vector pHELP1 (GEMS and CLUTTERBUCK 1993 ). During cotransformation, the plasmids with a complementing gene recombine with pHELP1 and become able to replicate as extrachromosomal DNA. We transformed mutant strains with integrating vector alone and integrating vector plus pHELP1 (~0.5 µg of each plasmid). If the number of transformants increased at least 100-fold (usually to several thousands total) in the presence of pHELP1, the fragment was considered to be able to complement the mutation in a replicative mode. The majority of these transformants were mitotically unstable and reverted to the wild type in the absence of selection. Transformation efficiency was even higher when the complementing fragment was subcloned in the pHELP1 vector. Alternatively, the number of transformants was about the same with and without pHELP1, and all transformants were mitotically stable. This was interpreted as meaning that the fragment could complement the mutation only by correcting it through homologous integration because it contained the region of mutation but not the whole gene.

Cloning of the sldA gene:
Strain 744D (sldA744; pyrG89; yA2) was transformed with a mixture of 100 µg of BamHI-digested pHELP1 vector and 300 µg of linear genomic DNA from strain R153 partially digested with Sau3AI. DNA fragments had sizes from 5 to 25 kb with the peak in the 10–20-kb region. Transformed protoplasts were plated onto YAGK, UU with 0.4 µg/ml of benomyl at 43° to select for transformants with wild-type sensitivity to benomyl. In a single transformation using protoplasts from 6 x 10⁸ spores, two colonies that were wild type with respect to sporulation and benomyl sensitivity were obtained. Both were mitotically unstable and reverted to the sldA744 phenotype in the absence of benomyl selection. After streaking on YAG, UU at 43°, sectors of yellow (sporulating) and dark (aconidial) hyphae were produced. Only the yellow sectors could grow in the presence of 0.4 µg/ml benomyl. Transformants were purified by streaking several times on YAGK, UU with 0.4 µg/ml of benomyl at 43°. Spores were collected and grown in 100 ml of YAG, UU with 0.4 µg/ml of benomyl at 37° overnight. Total DNA was isolated by a standard method (RAEDER and BRODA 1985 ) and additionally purified by phenol/chloroform extraction and ethanol precipitation. About 20 µg of this DNA was used to transform 70 µl of JM109 competent cells (Stratagene; efficiency 10⁸ cfu/µg of pUC18 DNA). Thirty-three Amp^r bacterial clones were obtained when DNA from the first transformant was used. Twenty-two clones were analyzed and found to contain plasmids identical to pHELP1 in size. Sixteen Amp^r bacterial clones were obtained when DNA from the second transformant was used. Ten clones contained plasmids identical in size to pHELP1. The remaining six clones carried very large plasmids. Plasmids from four of these clones were isolated and subjected to restriction analysis. All four plasmids were identical and appeared to be a duplicated pHELP1 vector with a 24-kb insert. This plasmid transformed the sldA744 mutant to wild-type benomyl sensitivity. It did not complement the benomyl hypersensitivity of the mutant sldB937 or of the -tubulin mutant tubA22 (WILLINS et al. 1995 ).

The complementation activity was localized to the distal 5.1-kb HindIII fragment of the insert by subclonings and transformations. Sequencing showed that this fragment contained the 35-bp HindIII–BamHI portion from the pHELP1 polylinker and that the insert apparently was joined to the BamHI site through a Sau3AI site of the genomic DNA. The fragment was able to complement benomyl hypersensitivity of strain 744D after subcloning into a pGEM vector. The transformation efficiency increased at least 100-fold when pHELP1 was cotransformed with the pGEM subclone. This indicated that the subcloned fragment could complement benomyl hypersensitivity when replicating extrachromosomally and, hence, that it contained the whole gene. The ClaI and SphI sites were mapped inside the gene. Truncation at any of these sites preserved complementation activity but abolished ability to complement in a replicative mode as transformation efficiency was the same with and without the pHELP1 vector.

Sequencing in both directions from the ClaI and SphI sites revealed an open reading frame homologous to the Bub1p of S. cerevisiae (ROBERTS et al. 1994 ) and starting 223 bp downstream of ClaI. Restriction fragments from this sequence were used to screen an A. nidulans cDNA library in a gt10 vector (OSMANI et al. 1987 ). Sequences were obtained both from the genomic clone and several cDNA clones, and they were used to deduce the beginning and the end of the coding sequence and the positions of introns. Poly(A) tails were found in two clones that were sequenced at their 3' ends. In three of four cDNAs that were analyzed and that were long enough (two of them were full length), the first intron was apparently spliced incompletely, resulting in premature protein termination. The position of the first intron was deduced from the corresponding fragment of a single full-length cDNA clone. The GenBank accession number for the sldA sequence is AF032987.

Cloning of the sldB gene:
Strain 937Dn carrying sldB937 mutation was transformed with an A. nidulans genomic library constructed in pHELP1 as follows. Genomic DNA from the R153 strain was partially digested with Sau3AI and fractionated by sucrose gradient centrifugation. About 6 µg of the 10–15-kb fraction DNA was ligated to 2 µg of pHELP1 that had been digested with BamHI and dephosphorylated. The ligation mixture was ethanol precipitated, resuspended in water, and used to transform 937Dn protoplasts prepared from ~5 x 10⁸ spores. Transformants were plated at 43° in YAGK, UU in the presence of 0.4 µg/ml of benomyl to select for wild-type sensitivity to benomyl. Three mitotically unstable transformants were obtained in a single transformation. Plasmids were recovered from each transformant as described above, except that DH5 competent cells (GIBCO BRL; library efficiency, 10⁸ cfu/µg of pUC19 DNA) were used. Restriction analysis showed that the three recovered plasmids had overlapping inserts of ~10, 16, and 17 kb. All three plasmids rescued the benomyl sensitivity of the sldB937 mutant with the high efficiencies characteristic of the replicating pHELP1 vector. One transformant also contained a plasmid with an ~4-kb insert that was not related to the other three clones and that did not complement the mutation. A 4.2-kb XbaI fragment shared by all three complementing plasmids was subcloned into the pGEM-7Zf⁺ vector. The fragment apparently contained the whole sldB gene as the subclone complemented the benomyl hypersensitivity of sldB937 strain, and its transformation efficiency increased >100-fold in the presence of pHELP1 vector.

An SphI restriction site was mapped inside the gene following the same strategy that was used in the case of the sldA gene. The sequence around that site revealed the presence of an ORF homologous to the Bub3p of S. cerevisiae (HOYT et al. 1991 ). A restriction fragment from the putative gene was used to screen an A. nidulans cDNA library, as described above, for the sldA gene. Sequences were obtained from both the genomic clone and two cDNA clones. One cDNA clone was full length and contained a poly(A) tail. Two introns were identified that were spliced identically in both cDNA clones. The GenBank accession number for the sldB sequence is AF032988.

Construction of sldA and sldB null mutants:
sldA was disrupted by replacing the genomic sequence beginning 218 bp upstream of the sldA start and including nucleotides 1–2550 of the 3813-bp-long sldA structural gene (residues 1–795 of the encoded 1216 aa) with the selectable marker gene pyrG inserted in an orientation opposite to that of the sldA gene. A plasmid containing the sldA genomic region was digested with ClaI and BspEI, and the vector portion was isolated and ligated to the pyrG containing the XbaI–EcoRV fragment from pXX1 (XIANG et al. 1995C ) after filling in the sticky ends. The linear fragment with the disrupted sldA gene was cut from the resulting subclone, purified, and used to transform the (pyrG^-; sldA⁺) strain A5.3 to prototrophy.

sldB was disrupted by replacing nucleotides 288–690 (residues 79–190 of the encoded 375 aa) from the 1193-bp-long sldB structural gene with the selectable marker gene pyrG inserted in an orientation opposite to that of the sldB gene. A plasmid containing the sldB genomic region was digested with BglII and NarI, and the vector portion was isolated and ligated to the BamHI–ClaI fragment containing pyrG gene from pXX1 (XIANG et al. 1995C ). The linear fragment with the disrupted sldB gene was cut from the resulting subclone, purified, and used to transform (pyrG^-; sldB⁺) strain A5.3 to prototrophy.

Transformants with disrupted sldA and sldB genes were identified by their hypersensitivity to benomyl. Site-specific integration of the disrupted sldA and sldB sequences into the sldA and sldB loci (by double recombination between genomic DNA and the linear fragments) was confirmed by Southern blotting using the pyrG gene and the deleted fragments as probes. Strains with a single correct integration were outcrossed to GR5 strains to separate the deletion mutations from the nudA5 mutation present in the A5.3 strain used for disruption.

Microscopy:
Liquid YG, UU media were inoculated with fresh conidia at a density of 10⁴–10⁵ conidia/ml in petri dishes containing coverslips. At regular time points, the coverslips with adherent germlings were removed for DAPI staining (XIANG et al. 1995C ). Images were collected using an Axioplan microscope (Carl Zeiss, Thornwood, NY) with a 100x Neofluar objective, captured with a CCD camera, and assembled in Adobe Photoshop (Adobe Systems, Mountain View, CA).

	RESULTS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Identification of A. nidulans genes required for viability in the absence of dynein:
Mutations in the A. nidulans nudA gene, which encodes the cytoplasmic dynein heavy chain, impair nuclear migration and reduce the radial growth rate to ~20% of that of the wild type (XIANG et al. 1994 ). However, nudA is not an essential gene. The dynein-null mutant is viable and is similar to nudA ts^- mutants at restrictive temperature (XIANG et al. 1995C ). To identify additional functions that might require cytoplasmic dynein, we searched for A. nidulans genes required for survival in the absence of dynein. A synthetic lethality screen was designed to find mutations that significantly impair the growth of dynein mutants but have little or no effect of their own. The strategy of the screen is outlined in Figure 1. Strain XX61 has one copy of nudA under the control of the inducible alcohol dehydrogenase I gene promoter alcA(p) (XIANG et al. 1995C ). This strain grows like wild type on nonrepressing minimal medium with glycerol as the carbon source (M-glycerol), but it grows poorly on media with YAG because dynein expression is repressed. After UV mutagenesis of XX61 spores, 19 mutant DG strains were selected that were unable to grow on YAG but could grow on M-glycerol. These mutants presumably carried mutations lethal in the absence of dynein and tolerated when dynein was expressed. The phenotypes of nine representative DG strains are shown in Figure 2.

View larger version (36K): In this window In a new window Download PPT slide   Figure 1. —Isolation of sld mutants. Spores of the A. nidulans strain XX61 carrying a nudA dynein gene under the control of the inducible alcA promoter were subjected to UV mutagenesis and then screened for dynein-dependent growth by replica plating.

View larger version (79K): In this window In a new window Download PPT slide   Figure 2. —Phenotypes of the DG (alcA(p)::nudA; sld) strains under conditions of induced or repressed dynein expression. Spores were point-inoculated on M-glycerol and YAG media from fresh spore stocks adjusted to ~3 x 105 spores/ml and grown for 3 days at the indicated temperatures. The strains in positions 1–9 are different DG strains carrying mutations sldF445, sldA744, sldG898, sldB937, sldH1088, sldD1351, sldI1444, sldC1518, and sldE1540, respectively. Control strains in positions 10 and 11 are XX61 and R153, respectively.

The screen also identified mutants unable to use glucose or that were inhibited by yeast extract. This unwanted class of mutants was eliminated by testing strains for growth on repressing media of different composition (see MATERIALS AND METHODS). To prove unambiguously that the DG strains carried sld mutations, we crossed the sld mutations away from the alcA(p)::nudA::pyr4-inducible dynein gene to produce single sld mutants and sld; nudA5 double mutants (nudA5 is a ts^- allele of the dynein gene; Figure 3).

View larger version (84K): In this window In a new window Download PPT slide   Figure 3. —Phenotypes of single sld mutants and double sld; nudA5 mutants. Spores were point inoculated on YG plates from fresh spore stocks adjusted to ~3 x 105 spores/ml and grown for 3 days at 43° (restrictive for dynein mutant nudA5) and 32° (permissive for nudA5). All strains carry the yA2 mutation, which results in the yellow color of asexual spores. The intensity of the colony color is proportional to the number of spores produced. The strains from 3 to 11 are either D (sld) or DA (sld; nudA5) carrying the following sld mutations: sldF445, sldA744, sldG898, sldB937, sldH1088, sldD1351, sldI1444, sldC1518, and sldE1540. Control strains in positions 1, 2, and 12 are A5.4, XX21, and WT, respectively.

All 19 strains with single sld mutations were able to form sizable colonies on YAG plates at all temperatures. Crosses of these strains back to XX61 (alcA(p)::nudA::pyr4) produced segregants that were identical to the original DG strains. All sld mutations significantly reduced the growth of the nudA5 mutant at restrictive temperature (Figure 3). For some mutations the synthetic interaction with nudA5 is weaker than with the alcA(p)::nudA gene presumably because nudA5 is leaky and allows expression of some functional dynein at 43°. In all crosses, sld mutations segregated as single genes.

We also tested whether sld mutations were synthetically lethal with a mutation in another nuclear distribution gene, nudF (XIANG et al. 1995B ), whose biochemical function is not known. The double mutants sld; nudF7 (nudF7 is a ts^- allele of the nudF gene) were identical to the respective sld; nudA5 double mutants at restrictive temperature. This result is in accord with the suggestion that NUDF protein functions in the same pathway as NUDA and may by a dynein motor regulator (WILLINS et al. 1997 ; GEISER et al. 1997 ).

To determine if the sld mutations were in the same or in different genes, we crossed them pairwise to each other. Mutations were provisionally assigned to the same locus if no wild-type recombinants were obtained in the cross between them. These crosses defined at least nine different genes that are shown in Table 2.

The D (sld) strains have characteristic growth defects that are listed in Table 2 (see also Figure 2 and Figure 3), the most common being a reduction in the production of asexual spores (conidia). sld mutants with similar phenotypes turned out to be grouped within the same loci. Five mutants hypersensitive to benomyl were assigned to sldA and sldB loci (Table 2). The correctness of this assignment was confirmed by direct cloning (see below). The largest group is represented by six sldC mutants, all of which are temperature sensitive. Each of the six mutants manifest specific defects at 43° that are pronounced to different extents in different alleles. They fail to grow in the air or to produce aerial hyphae ("flat" phenotype). These flat colonies have a dark brown, almost black color ("dark" phenotype). Interestingly, although the sldC mutants are very inhibited on complete media (Figure 3), they are very similar to the wild type on minimal media except for a conidiation defect (Figure 2). The defects mentioned in Table 2 always cosegregated with the sld mutations and, thus, are caused by the same mutations that confer lethality in the absence of dynein. Diploids between FGSC154 (sld⁺) and nine D (sld) strains with sld mutations in different genes (underlined alleles in Table 2) were wild type in appearance, indicating that these sld mutations were recessive. sldF445 was mapped to chromosome I by means of parasexual genetics. sldC1518 and sldI1444 were linked to each other (recombination frequency ~13%), and both mutations were mapped to chromosome II.

It is important to recognize that not every mutation that slows growth inhibits A. nidulans in the absence of dynein as much as the sld mutations. No interaction was observed between mutations affecting septation and nudA mutation (HARRIS et al. 1994 ). We tested known A. nidulans genes for a synthetic lethal interaction with dynein by crossing mutant strains to the alcA(p)::nudA strain and analyzing double mutants in the downregulated nudA background (data not shown). apsA and apsB mutants defective in asexual reproduction (CLUTTERBUCK 1994 ) are superficially similar to some of our sld mutants in that they grow at a slightly reduced rate and do not produce asexual spores. Only the apsA5 mutation significantly reduced growth in the absence of dynein at 43° (but not at 32°) and could qualify as a sld mutation. On the contrary, apsB8 or apsB14 mutations had no effect at all on the dynein mutant. We also tested bimC4 and bimE7 cell cycle mutations in the absence of dynein at 37°. The above mutations are ts^- and block cell cycle progression at 43° (ENOS and MORRIS 1990 ; JAMES et al. 1995 ). At 37°, they impose a partial block in mitosis that results in a slower nuclear division and lower nuclear counts compared to wild type. Only bimC4 but not bimE7 strongly reduced growth in the absence of dynein. This implies that just slowing nuclear division is not enough to cause a synthetic lethal phenotype in the absence of dynein. It has been shown previously that mutations in the gene for the S. cerevisiae bimC homologue CIN8 are synthetically lethal with dynein mutations (SAUNDERS et al. 1995 ; GEISER et al. 1997 ). In a separate set of crosses, none of the sld genes was found to be linked to apsA or bimC.

Mutations in sldD and sldE genes cause an abnormal nuclear morphology and exacerbate the nuclear migration defect seen in the absence of dynein:
To characterize the nuclear migration phenotypes of the sld mutants, spores of sld and sld; nudA5 mutants were germinated in YG, UU at 43°, fixed, stained with DAPI to visualize nuclei, and examined microscopically (Figure 4). Germlings of the sldH mutant contained multiple dots and specks of DAPI staining of unknown origin. Germlings of mutants not shown in Figure 4 were similar to the sldA mutant and did not possess any characteristic features. The number of nuclei in sld mutants A, B, F, and J was about the same as in wild type (32) after 10 hr of growth. Lower nuclear counts (8–32) were observed in mutants C, D, E, G, and H. Nuclear migration appeared normal in all single sld mutants. No dramatic difference in nuclear numbers was observed between the sld; nudA5 double mutants and the corresponding sld single mutants. However, it was difficult to count nuclei accurately in the double mutants because of nuclear clumping.

View larger version (67K): In this window In a new window Download PPT slide   Figure 4. —DAPI staining of single sld mutants and double sld; nudA5 mutants. Fresh spores of D and DA strains were germinated in YG, UU at 43° for 10 hr, fixed, and stained.

During germination of nudA5 at restrictive temperature (43°), nuclei divide normally but fail to migrate from the spore end of the germtube and to distribute evenly along the cell length. However, the block of nuclear movement is incomplete, and in all nudA5 germlings, after 10 hr of growth, some nuclei that have migrated into the germtube can be found. To see if sld mutations exacerbated this defect in nuclear migration, we compared nuclear distributions in sld; nudA5 double mutants with those in the nudA5 single mutant. The nuclear migration defect was quantified by determining the percent of germlings with no nuclei in the germtubes. The nudA5 single mutant did not have any such germlings—at least one nucleus was always observed at some place in the germtube. Two sld mutations, sldD1351 and sldE1540, clearly exacerbated the nuclear migration defect. In ~50% of corresponding double mutants, sld; nudA5, all nuclei were restricted to the spore end of the germtube. When nuclei were observed inside the germtube, they often appeared abnormally intensely stained, enlarged, and elongated. This abnormal nuclear morphology was not a synthetic defect as similar abnormal nuclei were observed in single sld mutants. The remaining seven sld; nudA5 mutants had <10% of germlings with completely failed nuclear migration, and these were always short and contained a few nuclei.

Most of the sld mutations are synthetically lethal with a mutation in the mitotic kinesin bimC:
Recent studies demonstrated that mutations in the yeast cytoplasmic dynein are synthetically lethal with mutations in a number of mitotic kinesins involved in spindle assembly and chromosome separation (SAUNDERS et al. 1995 ; GEISER et al. 1997 ; COTTINGHAM and HOYT 1997 ; DEZWAAN et al. 1997 ). To see if the sld genes are involved in mitosis, we tested whether they interacted with the bimC4 mutation in the bimC mitotic kinesin (ENOS and MORRIS 1990 ). Because the ts^- mutation bimC4 is lethal at 43° we analyzed sld; bimC4 double mutants at 37°. bimC4 slows growth but is not lethal at 37°. Most of the sld mutations display a synthetic lethal interaction with the bimC4 mutation (Table 2). No interaction of bimC4 was observed with sldC, sldD, and sldE mutations. However, these sld mutants could not be tested reliably because all available alleles are ts^- and may be only partially inactive at 37°. For sldA and sldB mutants the synthetic lethal interaction with bimC4 is shown in Figure 5. For sldF–sldI, the synthetic lethal interaction is as strong as for sldA and sldB.

View larger version (60K): In this window In a new window Download PPT slide   Figure 5. —(A) Hypersensitivity of sldA and sldB mutants to microtubule destabilization. Spores were inoculated on YG plates with benomyl at 43°. Phenotypes of indicated strains on plates without benomyl are shown in Figure 3. (B) Synthetic lethal interactions between sldA and sldB mutations and the mitotic kinesin mutation bimC4.

Synthetic lethal interactions between some sld mutations and a mutation in the bimC mitotic kinesin suggest that these sld genes play a role in spindle assembly or chromosome separation. Indeed, cloning and sequencing showed that sldA and sldB genes are spindle assembly checkpoint genes (see next section). We examined the interaction of the other sld genes with the sldB937 mutation (Table 2). One mutant, sldI1444, clearly showed a strong synthetic lethal interaction with both sldB937 and bimC4. We were not able to recover sldF445 and sldG898 double mutants with sldB937 most likely because the corresponding double mutants were inviable at all temperatures (all these mutants are unconditional).

Cloning of sldA and sldB genes using an A. nidulans replicating plasmid:
Mutations in the sldA and sldB genes confer hypersensitivity to benomyl (Figure 5). This phenotype always cosegregated with the sldA and sldB mutations (see MATERIALS AND METHODS). We cloned sldA and sldB by complementation of benomyl hypersensitivity of their respective mutants. The sldA744 and sldB937 mutants were transformed with a wild-type genomic DNA library constructed in the replicating vector pHELP1 (GEMS and CLUTTERBUCK 1993 ) and plated in the presence of 0.4 µg/ml of benomyl to select for transformants with wild-type sensitivity to benomyl. pHELP1 carries an A. nidulans sequence, AMA1, that functions as a plasmid replicator and transformation enhancer and makes the plasmid capable of extrachromosomal maintenance in A. nidulans (GEMS et al. 1991 ; GEMS and CLUTTERBUCK 1993 ; ALEKSENKO and CLUTTERBUCK 1995 , ALEKSENKO and CLUTTERBUCK 1996 ; ALEKSENKO et al. 1996 ). AMA1-bearing vectors transform A. nidulans at frequencies up to 2000-fold higher than for typical integrating vectors. This greatly facilitates cloning of A. nidulans genes by complementation of mutant phenotypes (GEMS and CLUTTERBUCK 1993 ; GEMS et al. 1994 ; BOWYER et al. 1994 ).

The sldA gene was cloned by an "in vivo ligation" method (GEMS et al. 1994 ). The sldA744 mutant was transformed with a mixture of BamHI-digested vector pHELP1 and genomic DNA partially digested with Sau3AI. During cotransformation, the genomic DNA recombines with the vector and then replicates autonomously (GEMS and CLUTTERBUCK 1993 ; GEMS et al. 1994 ; BOWYER et al. 1994 ). A high concentration of genomic and vector DNA was used during cotransformation to promote end-to-end joining of the vector and genomic fragments (GEMS et al. 1994 ). Although we successfully cloned the sldA gene by this method, we found later that standard in vitro ligation is more convenient because it is more likely to result in correct joining of the insert and the vector and requires less DNA. In vitro ligation was used to clone sldB. Genomic DNA partially digested with Sau3AI was ligated to BamHI-digested and dephosphorylated pHELP1. The ligation mixture was transformed into the sldB937 mutant.

Complementing plasmids were recovered from the wild-type-like transformants resistant to 0.4 µg/ml benomyl and were amplified in E. coli. Inserts of genomic DNA were subcloned and tested for complementation of benomyl hypersensitivity of sldA744 and sldB937 (Figure 6). These subclones could complement mutations either by homologous recombination at the site of mutation or, if the fragment contained the whole gene, by integration at any site in the genomic DNA. On the other hand, complementation in the replicative mode, as part of the extrachromosomal DNA, was possible only if the fragment contained the whole gene with all regulatory sequences, including its promoter.

View larger version (15K): In this window In a new window Download PPT slide   Figure 6. —Identification of sldA and sldB genes. Restriction fragments from complementing genomic clones were subcloned, and the resulting plasmids tested by complementation. The fragment was scored positive in the integrative complementation mode if stable wild-type transformants were obtained. If the number of transformants increased >100-fold in the presence of helper plasmid pHELP1, the fragment was considered to be able to complement the mutation in the replicative mode as part of an extrachromosomal vector and scored as positive. If the number of transformants was about the same with or without helper plasmid (with all transformants mitotically stable), the fragment was scored as negative in the replicative mode of complementation. The protein-coding regions deduced from analysis of cDNA sequences are shown by thick arrows with gaps marking the introns. The thin arrow in B is a putative gene whose product is similar to the hypothetical yeast protein ORF SCYOR262w (GenBank accession no. Z75170). The box in A represents a fragment of the pHELP1 polylinker. Some ClaI and XbaI sites are not shown because they are protected by dam methylation. Restriction sites mapped inside complementing genes are marked by asterisks.

The pHELP1 vector's ability to increase the transformation efficiency during cotransformation was used to determine whether a sequence contained the whole gene and for rapid mapping of restriction sites within complementing genes (Figure 6). For example, the 5.1-kb HindIII fragment subcloned into a pGEM vector was able to complement the benomyl hypersensitivity of the sldA744 mutant. As expected for an integrative transformation, all transformants were mitotically stable. The number of transformants was at least 100-fold higher when the pHELP1 vector was cotransformed with the pGEM subclone. The majority of the latter transformants were mitotically unstable. This indicated that the fragment could complement benomyl hypersensitivity when replicating extrachromosomally and, hence, contained the whole sldA gene, including the promoter. Fragment truncation at the ClaI or SphI site preserved complementation activity by integration but abolished the ability to complement in a replicative mode, as transformation efficiency was the same with and without the pHELP1 vector and no mitotically unstable transformants were produced. This indicated that the ClaI and SphI restriction sites were located within the sldA gene. An SphI site was mapped within the sldB gene in the same manner.

Potential ORFs were revealed by sequencing the genomic DNA adjacent to restriction sites mapped within the sldA and sldB genes. Screening of a cDNA library with fragments from these putative ORFs identified several cDNAs. These cDNAs as well as the genomic DNA clones were sequenced to determine exon boundaries and to deduce the sequences of the encoded proteins. The ClaI site in the sldA gene was upstream of the sldA initiator methionine and very close to the 5' end of the corresponding cDNA. Truncation at this site destroyed complementation activity in the replicative mode (Figure 6) apparently by separating the promoter from the sldA gene.

Plasmids containing the sldA and sldB genes also complemented the benomyl hypersensitivity caused by other sldA and sldB mutations (Table 2). The sldA gene did not complement the sldB mutants and vice versa, nor did sldA complement the benomyl hypersensitivity of an -tubulin mutant tubA22 (WILLINS et al. 1995 ).

In addition to complementing benomyl hypersensitivity, the cloned sldA and sldB genes also corrected sporulation defects of sldA and sldB mutants (Figure 3). To prove that the sld phenotypes were also complemented, we transformed sldA744; nudA5 and sldB937; nudA5 double mutants with the sldA and sldB genes, respectively. The transformed protoplasts were plated at 32° (permissive temperature for nudA5 mutation) in the presence of 0.4 µg/ml of benomyl to select for wild-type benomyl sensitivity. The resulting transformants, when gridded at 43° in the absence of benomyl, produced small compact colonies characteristic of the nudA5 single mutation, indicating that the sld mutations were complemented. Thus, cloned genes complement all three phenotypes of the sldA and sldB mutants, the sld phenotype (strongly reduced growth in the absence of dynein), benomyl hypersensitivity, and defective sporulation. Together with genetic linkage analysis, this proves that all the defects described above are caused by single mutations.

The fact that plasmids containing truncated sldA and sldB genes were able to complement the mutant phenotypes by integration into the chromosome, but not as part of the extrachromosomal DNA (Figure 6), indicated that we have cloned genes mutated in sldA and sldB mutants rather then multicopy suppressors. To prove this and to produce null alleles of cloned genes, we constructed sldA and sldB deletion mutants by substituting the pyrG marker gene for part of each gene (see MATERIALS AND METHODS). The resulting null mutants were viable and identical to the original sldA and sldB mutants. They displayed the same benomyl hypersensitivity (Figure 5), cold-sensitive growth (which was slightly accentuated in the sldA deletion strain compared to the sldB deletion strain), reduced conidiation at 32° compared to wild type, and complete lack of conidiation at 43° (Figure 3).

sldA and sldB are spindle assembly checkpoint genes:
The sldA gene encodes a polypeptide of 1216 amino acids with 29% identity and 39% similarity to the product of the budding yeast spindle assembly checkpoint gene BUB1 (HOYT et al. 1991 ; ROBERTS et al. 1994 ). Other proteins in the database with similarity to SLDA are a mouse homologue of Bub1p (TAYLOR and MCKEON 1997 ) and a putative Bub1p homolog in Caenorhabditis elegans (GenBank accession number Z71266). Three domains of homology are clearly present (Figure 7), as noted previously by others (TAYLOR and MCKEON 1997 ). The C-terminal domain is the most conserved and corresponds to a serine/threonine protein kinase. The N-terminal domain of ~120 residues is also conserved and may be responsible for kinetochore binding (TAYLOR and MCKEON 1997 ). The middle region between these domains is less conserved.

View larger version (18K): In this window In a new window Download PPT slide   Figure 7. —Homology between SLDA protein, Bub1p of S. cerevisiae, and mouse Bub1p. Numbers above each domain show percents of identity/similarity to the corresponding domain of SLDA. Protein fragments delimited by residues indicated below were aligned with the GCG program GAP (endweighted). If alignments generated by simultaneous alignment of three sequences with the program PILEUP are used, the numbers do not change significantly, except that for the middle domain of the mouse Bub1p numbers decrease by ~3%. The GenBank accession number for the sldA sequence is AF032987.

The SLDA protein is almost 200 residues longer than yeast Bub1p (1021 aa), mouse Bub1p (1058 aa), or the putative C. elegans homolog (987 aa). Simultaneous alignment of these proteins using the endweighted Genetics Computer Group program PILEUP suggests that the difference results from multiple small insertions over the whole length of the SLDA protein rather than from one large insertion. In the C-terminal kinase domain, these insertions appear unambiguously at the boundaries between conserved kinase subdomains (TAYLOR and MCKEON 1997 ).

The protein encoded by sldB gene is 30% identical and 43% similar (Figure 8) to the Bub3p spindle assembly checkpoint protein from budding yeast (HOYT et al. 1991 ), which has been shown to interact with Bub1p (ROBERTS et al. 1994 ). Both Bub3p and SLDB show a similar degree of homology to the Rae1p involved in RNA export (BROWN et al. 1995 ; MURPHY et al. 1996 ). Taking into account that sldA and sldB mutants have identical phenotypes and that SLDA has similarity only to Bub1p, we conclude that sldB is a homolog of the spindle assembly checkpoint gene BUB3 rather than of RAE1.

View larger version (57K): In this window In a new window Download PPT slide   Figure 8. —Comparison of the amino acid sequence of the SLDB protein with the S. cerevisiae Bub3p generated with the GCG program GAP. The GenBank accession number for the sldB sequence is AF032988.

If sldA and sldB mutants are defective in the spindle assembly checkpoint, they should lose viability during division if spindle assembly is compromised (LI and MURRAY 1991 ; HOYT et al. 1991 ). sldA744 and sldB937 mutants and wild-type spores were germinated in the presence of benomyl (0.4 µg/ml), and aliquots were removed at different times and titered on plates without benomyl to determine viability. sldA and sldB mutants lost viability within the first 10 hr while the wild type did not. A similar experiment was done with the double mutants sldA744; bimC4 and sldB937; bimC4. Mutant spores were germinated at 37° on plates. At different times, the plates were shifted to 32° and incubated for an additional 3 days. All doubly mutant spores completely lost viability during the first 10 hr of incubation at 37° while ~30% of bimC4 mutant spores survived and formed colonies after several days of incubation at 37°. A different situation was observed with sldA744; nudA5 and sldB937; nudA5 double mutants. sldA and sldB mutants practically cannot grow in the absence of dynein (Figure 2). However, more than 50% of sldA744; nudA5 and sldB937; nudA5 doubly mutant spores were able to form colonies upon the shift to 32°, even after several days of incubation at 43°.

	DISCUSSION

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

A. nidulans genes required for survival in the absence of dynein:
Here, for the first time, we have used a synthetic lethality screen in a filamentous fungus to identify genes required for viability in the absence of dynein. The screen was made possible because deletion of dynein is not lethal in A. nidulans (XIANG et al. 1995C ). Most sld mutations do not appear to be lethal in the absence of dynein, but they strongly enhance the growth arrest (Figure 2). Further work will be required to determine the significance of genetic interactions between the sld genes and nudA gene. The fact that we recovered multiple alleles of the same genes (Table 2) illustrates that the screen strategy is meaningful and is biased towards a specific set of genes.

In filamentous fungi that grow by apical tip extension, long-distance movement of nuclei is necessary. Both in A. nidulans and N. crassa, dynein/dynactin mutants nuclei occasionally escape from the spore case and distribute in uneven clusters along the hyphae (XIANG et al. 1995C ; PLAMANN et al. 1994 ; ROBB et al. 1995 ). Therefore, among the sld mutations, we expected to find mutations that exacerbate the nuclear migration defect seen in the absence of dynein. Examination of the nuclear migration phenotypes of the sld muta-tions in the downregulated nudA background identified sldD1351 and sldE1540 as mutations that exacerbated the nudA nuclear migration defect. sldD and sldE genes are good candidates for components of a dynein-independent system of nuclear migration. Alternatively, mutations in sldD and sldE genes could cause alterations in the cytoskeleton such that nuclear movement is impeded, or they could activate a system that holds nuclei in place.

sldA and sldB are spindle assembly checkpoint genes; four other sld genes may also have mitotic functions:
Cytoplasmic dynein is known to be involved in mitosis in other organisms (VAISBERG et al. 1993 ; MERDES and CLEVELAND 1997 ), and mutations in yeast dynein are synthetically lethal with mutations in a number of mitotic motors (SAUNDERS et al. 1995 ; GEISER et al. 1997 ; COTTINGHAM and HOYT 1997 ; DEZWAAN et al. 1997 ). We searched for evidence of a role for sld genes in mitosis by asking if the sld genes interact with the bimC4 mutation in the bimC mitotic kinesin, which is required for spindle pole body separation and bipolar spindle formation (ENOS and MORRIS 1990 ). Mutations in six of the sld genes, including sldA and sldB, are synthetically lethal with the bimC4 mutation at semirestrictive temperature, indicating that these genes play a role in mitosis.

sldA and sldB mutants were the only ones recovered in our screen that were hypersensitive to the microtubule-destabilizing drug benomyl. Cloning and sequencing of sldA and sldB genes identified them as homologs of spindle assembly checkpoint genes BUB1 and BUB3, respectively, of S. cerevisiae (HOYT et al. 1991 ; ROBERTS et al. 1994 ). Spindle assembly checkpoint proteins prevent the onset of anaphase if the spindle does not assemble correctly with all chromosomes attached (ELLEDGE 1996 ; WELLS 1996 ). The properties of sldA and sldB mutants are consistent with a defect in the spindle assembly checkpoint pathway. Both mutants are hypersensitive to the microtubule-destabilizing compound benomyl and to impairment of the bimC mitotic kinesin (Figure 5). Similar phenotypes were observed in S. cerevisiae spindle assembly checkpoint mutants (LI and MURRAY 1991 ; HOYT et al. 1991 ; GEISER et al. 1997 ).

What might be the functions of the four other sld genes (sldF–sldI) that are hypersensitive to the loss of the bimC mitotic kinesin? Although at least six other spindle assembly checkpoint genes are known in yeast in addition to BUB1 and BUB3 (ELLEDGE 1996 ; WELLS 1996 ), the fact that sldF–sldI mutants are not hypersensitive to benomyl makes sldF–sldI unlikely candidates for spindle assembly checkpoint genes. However, sldF–sldI could be required for bimC function, as bimC is an essential gene and the combination of two mutations that partially inactivate the same process can completely block the process. Alternatively, these genes could perform mitotic functions that are normally redundant with those of dynein and the bimC kinesin. Examples of such genes are the mitotic kinesins KAR3, KIP1, and KIP3 of S. cerevisiae (SAUNDERS et al. 1995 ; COTTINGHAM and HOYT 1997 ; DEZWAAN et al. 1997 ).

The possible mitotic role of A. nidulans cytoplasmic dynein:
The fact that many sld genes interact with mitotic kinesin bimC suggests that the A. nidulans cytoplasmic dynein itself plays a role in mitosis. The discovery of spindle assembly checkpoint genes among sld genes is especially intriguing. The simplest explanation for the finding of checkpoint genes among the sld genes is that dynein participates in mitotic spindle assembly or function. However, spindle assembly and function appear to be normal in the dynein mutant (XIANG et al. 1994 , XIANG et al. 1995C ). It is possible that in the absence of dynein, spindle assembly may take slightly longer than usual. The spindle assembly checkpoint genes are required for normal timing of mitosis; their absence could lead to an earlier onset of anaphase (TAYLOR and MCKEON 1997 ). The combination of these two effects may lead to a premature anaphase and decrease the fidelity of chromosome segregation even if dynein mutation alone has little effect on spindle assembly. We also note that although strains doubly mutant in sldA or sldB and dynein gene cannot grow, they do not lose viability rapidly during germination and nuclear division. On the contrary, sldA and sldB mutant spores lose viability