Cytoplasmic dynein is a multisubunit protein complex that functions as a minus-end–directed microtubule motor. Acting with another complex, dynactin, it powers movement and positioning of diverse cellular organelles in eukaryotic cells. Genetic screens in filamentous fungi Aspergillus nidulans and Neurospora crassa and in yeast Saccharomyces cerevisiae have identified many genes in the cytoplasmic dynein/dynactin pathway (Osmani et al., 1990 ; Plamann et al., 1994 ; Robb et al., 1995 ; Xiang et al., 1994 , 1995a ; Bruno et al., 1996 ; Tinsley et al., 1996 ; Geiser et al., 1997 ; Vierula and Mais, 1997 ; Beckwith et al., 1998 ; Minke et al., 1999 ; Xiang et al., 1999 ; Efimov and Morris, 2000 ; Lee et al., 2001 ; Zhang et al., 2002 ). In A. nidulans, mutations in dynein and dynactin genes impair distribution of nuclei along hyphae (filamentous fungal cells), which are multinucleated. Because of such phenotypes, the genes are called nud (nuclear distribution) genes. Dynein/dynactin null mutants are viable, but form abnormally compact colonies and fail to produce conidia (asexual spores). Cytoplasmic microtubules are less dynamic in nud mutants (Han et al., 2001 ), and their destabilization suppresses nuclear distribution defects (Willins et al., 1995 ; Alberti-Segui et al., 2001 ). Defects in vesicle trafficking and vacuole distribution are also likely, because they were observed in N. crassa dynein/dynactin mutants (Seiler et al., 1999 ; Lee et al., 2001 ) and in a dynactin mutant of Aspergillus oryzae (Maruyama et al., 2002 ). In addition to the subunits of dynein and dynactin, genetic screens in the above-mentioned fungi also identified several proteins that do not seem to be components of purified dynein or dynactin complexes, and thus whose relation to dynein/dynactin is not obvious. This work concerns two such proteins of A. nidulans, NUDF (Pac1p in S. cerevisiae) and NUDE (RO11 in N. crassa), encoded by the nudF and nudE genes, respectively.
According to genetic data, the nudF gene of A. nidulans and its S. cerevisiae homolog Pac1p function in the dynein/dynactin pathway (Xiang et al., 1995a ; Willins et al., 1997 ; Geiser et al., 1997 ). The Drosophila and Caenorhabditis elegans homologs of NUDF have also been linked to the dynein/dynactin function (Liu et al., 1999 ; Swan et al., 1999 ; Lei and Warrior, 2000 ; Liu et al., 2000 ; Dawe et al., 2001 ). The mammalian homolog of NUDF, LIS1, is the product of a gene whose mutations cause lissencephaly, a brain malformation characterized by a disorganization of neurons within the cerebral cortex and a reduction in brain surface convolutions (Dobyns et al., 1993 ; Reiner et al., 1993 ; Chong et al., 1997 ; Lo Nigro et al., 1997 ; Hirotsune et al., 1998 ). LIS1 coimmunoprecipitates with both dynein and dynactin, and colocalizes with dynein/dynactin (Faulkner et al., 2000 ; Niethammer et al., 2000 ; Sasaki et al., 2000 ; Smith et al., 2000 ; Tai et al., 2002 ). According to two-hybrid and coexpression/coimmunoprecipitation assays, LIS1 binds two regions of the cytoplasmic dynein heavy chain (CDHC): the first AAA repeat (P1 loop) implicated in motor activity, and the N-terminal domain implicated in cargo binding (Sasaki et al., 2000 ; Tai et al., 2002 ). NUDF also interacts with the first AAA repeat of the A. nidulans CDHC in the two-hybrid system (Sasaki et al., 2000 ) and in vitro (Hoffmann et al., 2001 ). That NUDF might affect CDHC was first suggested by the results of a genetic screen for extragenic suppressor of a nudF mutation in A. nidulans (Willins et al., 1997 ). Two such suppressors were mapped to the CDHC and turned out to be bypass suppressors. When observed in live A. nidulans cells, green fluorescent protein (GFP)-tagged CDHC and NUDF are seen at the ends of dynamic cytoplasmic microtubules as linearly moving, comet-like structures (Xiang et al., 2000 ; Han et al., 2001 ; Zhang et al., 2002 ). Although the physiological significance of this localization is unclear (e.g., how it influences nuclear distribution), such localization is characteristic of dynein/dynactin and several other microtubule-interacting proteins (reviewed by Schroer, 2001 ; Schuyler and Pellman, 2001 ; Dujardin and Vallee, 2002 ).
The nudF gene was isolated inadvertently as a multicopy suppressor of the temperature-sensitive (ts) nudC3 mutant of A. nidulans, in which the NUDF protein level is below normal at elevated temperatures (Osmani et al., 1990 ; Xiang et al., 1995a ). The mammalian NUDC homolog binds LIS1 (Morris et al., 1998 ) and colocalizes with cytoplasmic dynein in neurons (Aumais et al., 2001 ). However, it is likely that the role of NUDC is not restricted to the dynein/dynactin pathway, because the nudC null mutant of A. nidulans has different and more severe growth defects than the dynein/dynactin null mutants (Chiu et al., 1997 ).
The RO11 protein of N. crassa functions in the dynein/dynactin pathway (Minke et al., 1999 ). Its A. nidulans homolog, NUDE, was isolated in the screen for multicopy suppressors of a nudF ts mutation (Efimov and Morris, 2000 ). At least two mammalian homologs of RO11/NUDE exist, and both are known to bind LIS1 (Feng et al., 2000 ; Kitagawa et al., 2000 ; Niethammer et al., 2000 ; Sasaki et al., 2000 ; Sweeney et al., 2001 ). The conserved N-terminal coiled coil of NUDE is responsible for NUDF/LIS1 binding and is essential for the NUDE function, whereas its highly variable C-terminal domain is dispensable in A. nidulans (Efimov and Morris, 2000 ). Mammalian NUDE also coprecipitates and colocalizes with several dynein/dynactin subunits and centrosomal components (Feng et al., 2000 ; Niethammer et al., 2000 ; Sasaki et al., 2000 ). The exact place of the NUDE protein in the dynein/dynactin pathway and how it affects NUDF/LIS1 remain to be determined.
One of the findings presented in this article is an interaction between the nudF and apsA genes of A. nidulans. Similar to nud genes, the apsA and apsB genes (anucleate primary sterigmata) are involved in nuclear migration events in syncytial hyphae and during production of uninucleate conidia (Clutterbuck, 1994 ; Fischer and Timberlake, 1995 ; Suelmann et al., 1997 , 1998 ; Graïa et al., 2000 ). Although both nud and aps mutants display nuclear distribution defects, a possible connection between the aps genes and dynein/dynactin has not been previously investigated, probably because the aps mutants have much milder nuclear distribution and growth defects than the nud mutants. However, the S. cerevisiae homolog of APSA, Num1p (Kormanec et al., 1991 ), is required for dynein function in yeast (Geiser et al., 1997 ; Heil-Chapdelaine et al., 2000 ; Farkasovsky and Küntzel, 2001 ). The apsB gene encodes a 121-kDa coiled coil protein that does not have any obvious homologs in other organisms (Suelmann et al., 1998 ). APSA and Num1p are large proteins consisting of coiled coil segments at the N terminus, a variable number of short repeats in the middle, and a pleckstrin homology (PH) domain at the C terminus responsible for protein targeting to the cell cortex. Both Num1p and APSA are exclusively cortical proteins, which distinguishes them from any other dynein or dynactin subunit (Farkasovsky and Küntzel, 1995 ; Suelmann et al., 1997 ; Heil-Chapdelaine et al., 2000 ; Farkasovsky and Küntzel, 2001 ). Localization of Num1p to the yeast cortex is independent of dynein, dynactin, and microtubules. There is growing evidence for a cortically bound form of dynein/dynactin, but only in a few cases was it possible to visualize dynein/dynactin at the cell cortex (reviewed by Dujardin and Vallee, 2002 ).
To accurately compare growth rates of different A. nidulans strains and transformants, spores were point inoculated in the center of 10-cm Petri dishes with YAG or M-glucose solid medium plus required supplements and incubated at 37°C. Colony diameters were measured on the back of plates with a ruler every 24 h for up to 5 d. The increase in colony diameter from day 1 to day 5 was linear (coefficients of determination were typically >0.9995) and was used to calculate the colony radial growth rate. The SE of these measurements was <0.4 mm/d in each individual experiment, as estimated from the error of slope calculations and from the variation among duplicate plates or among independent transformants. Alternatively, colony diameters were calculated from the colony areas, which were measured after taking images of colonies. The latter method was used to compare growth rates of the SF2-9-9 (ΔnudE) strain transformed with pAid, pAid::nudE, and pAid::nudF.
Determination of the effect of multiple copies of different genes on different A. nidulans mutants was done routinely as follows. The mutants, each unable to grow without uridine and uracil due to the pyrG89 mutation, were transformed with the pAid-derived plasmids. At least four independent transformants were gridded together with control transformants on YAG and YAGK plates and incubated at 32, 37, and 43°C for 3 d. All ts mutants used in this work are somewhat suppressed by 0.6 M KCl, so that YAGK is a less restrictive condition than YAG at the same temperature. The colony sizes and conidiation of the transformants were compared with those of the controls. The main control was the same strain transformed with the empty vector pAid. As a wild-type control, transformants GR5[pAid], SF2-9-9[pAid::nudE], XX21[pAid::nudF], apsA5[pAid::apsA], and C3y-3[pAid::nudC], all of which grow at the same rate, were used. The SRF30 (ΔapsA) strain was transformed with the pAid2-14 clones (selection for arginine prototrophy) and transformants were analyzed on M-glucose plus pyridoxine and p-aminobenzoic acid. The colony radial growth rates of some transformants were also compared quantitatively as described above.
Transformation of A. nidulans was done using germinating conidia essentially as described previously (Osmani et al., 1987 ). A. nidulans genomic DNA was prepared according to Willins et al. (1995) with minor modifications. For 4,6-diamidino-2-phenylindole (DAPI) staining of nuclei in conidia, a suspension of conidia was spread on a cover glass, allowed to dry (~20 min at 55°C), and stained with DAPI according to a standard protocol (Willins et al., 1995 ).
pAid::nudF6 and pAid::nudF were isolated in screens for multicopy suppressors of the nudF7 mutation (Efimov and Morris, 2000 ) and the ΔnudE mutation (this work), respectively. The insert in pAid::nudF contains the nudF gene (1.3 kb, oriented toward AMA1), ~3 kb of upstream sequence, and ~5.5 kb of downstream sequence. To make pAid2::nudF, the ~7 kb AatII-BglII fragment from pAid::nudF was inserted at the AatII-BamHI sites of pAid2-14, resulting in an insert that is the same as in pAid::nudF, but carries ~2.6 kb less of the sequence downstream of nudF. The ~5-kb insert in pAid::nudF6 contains the nudF6 ts allele of the nudF gene and flanking regions.
pAid::nudC and pAid::nudCΔ were isolated in the screen for multicopy suppressors of the nudC3 mutation (this work). The insert in pAid::nudC is ~8 kb and contains the nudC gene and flanking regions. The insert in pAid::nudCΔ is ~6 kb, contains most of the nudC gene (oriented away from AMA1), and terminates inside the last intron of the nudC gene (the sequence of the nudC/vector junction is gcattgtgct/gatccccgggtacc… ).
The insert in Aid::apsA and pAid2::apsA is the 10.5-kb BamHI-BamHI fragment from pRF7 (Fischer and Timberlake, 1995 ) with the apsA gene and flanking regions.
pAid::nudE (Efimov and Morris, 2000 ), pAid::GFP::nudE, pAid::GFP::nudE-N, pAid::GFP::nudE-C were made by subcloning and are identical to each other except for the GFP gene or deletions within the nudE gene. To create GFP::nudE fusions, codons 3–238 of the adapted for plants GFP version GFP2-5 (Fernández-Ábalos et al., 1998 ) were amplified by polymerase chain reaction (PCR) from the plasmid pMCB4 (provided by John H. Doonan, John Innes Centre, Norwich, United Kingdom) and inserted after the third codon of the nudE gene by using PCR-mediated recombination. In-frame deletion of aa 45–214 in the NUDE-C variant was obtained by deleting the NruI-BglII fragment (579 base pairs after filling in). The C-terminal domain of NUDE (aa 216–586) was deleted in the NUDE-N variant by excising the BglII-MfeI fragment (539 base pairs after filling in). The latter deletion disrupts the nudE ORF, resulting in termination of the NUDE sequence after aa 215 and addition of 16 new aa. The cloning junctions and the regions amplified by PCR were verified by sequencing.
pAid clones with the GFP*::nudE fusions are identical to the plasmids with the GFP::nudE fusions described above except for a point mutation in the GFP gene introduced during PCR. The mutation in the GFP* gene changes Leu-42 of the GFP2-5 protein to His. The plasmid pSAL-1 was used to integrate the GFP*::nudE gene into the A. nidulans genome under the alcA promoter. It is pAL3 vector (Waring et al., 1989 ) carrying a 3.7-kb insert at the BamHI site with the GFP*::nudE fusion (oriented away from alcA), 0.27 kb of the sequence upstream of the nudE gene, and 0.9 kb of the sequence downstream of nudE.
The ts C3y-3 (nudC3) strain was transformed with genomic DNA fragments from the XX20 (nudF6) mutant (5–20-kb sucrose gradient fraction of Sau3AI partial digest) ligated to the pAid vector (cut with BamHI and dephosphorylated). Several growth conditions were tried to find the least restrictive condition with low conidiation level. The bulk of transformants was plated in YAGK at 37°C, overlaid with YAGK the next day, and shifted to 43°C after two more days at 37°C. Alternatively, the plates were overlaid with either YAG or YAGK and left at 37°C. The total number of transformants was >1.5 × 105. Suppressed transformants were identified as patches of yellow color brighter than the background. Approximately 160 clones were completely suppressed and were deemed to had been transformed with the nudC gene. The suppressing plasmids were recovered from five such clones and each was found to carry the full-length nudC gene. Plasmids from four strongly (but not completely) suppressed transformants were found to carry inserts with the 3′-truncated nudC gene, as evidenced by restriction mapping and PCR with the nudC-specific primers. The truncation site was determined in one such plasmid, pAid::nudCΔ, by sequencing the insert ends. Plasmids from three weakly suppressed clones were found to carry overlapping inserts with the same novel gene that was isolated in the screen for multicopy suppressors of the ΔnudE mutation. The suppressing plasmids could not be recovered from several transformants, including six clones phenotypically different from the clones described above.
GFP fusions were detected on Western blots using purified rabbit anti-GFP polyclonal antibody (Torrey Pines Biolabs, Houston, TX). An affinity-purified rabbit polyclonal antibody against the NUDF protein (Xiang et al., 1995a ) was a gift from Xin Xiang (Uniformed Services University of the Health Sciences, Bethesda, MD). An alkaline phosphatase conjugate was used as a secondary antibody. Detection was performed with BCIP/NBT (5-bromo-4-chloro-3-indolyl-phosphate/nitroblue tetrazolium) phosphatase substrate system (KPL, Gaithersburg, MD).
The microscope setup for observing GFP-tagged proteins in live A. nidulans hyphae was identical to that used by Xiang and colleagues (Xiang et al., 2000 ; Han et al., 2001 ; Zhang et al., 2002 ), except that the sample temperature was controlled by an air-heated chamber enclosing the microscope rather than by a heated stage. It was an Olympus IX70 inverted fluorescence microscope equipped with 5 MHz MicroMax cooled charge-coupled device camera (Princeton Instruments, Trenton, NJ), a shutter, and a controller unit connected to a Macintosh computer. A fluorescence filter cube for fluorescein isothiocyanate and 100× objectives were used. Unless stated otherwise, cells were grown and observed at 32°C on agarose pads after 23–26 h of incubation. At least four samples of each strain were examined. For each sample, time-lapse series were recorded for 8–15 individual hyphae first and then several hundreds of hyphae were examined by eye for the presence of fluorescent structures and more series were recorded if necessary. IPLab software (Scanalytics, Fairfax, VA) was used to acquire images and time-lapse series of GFP fluorescence. Images and series were acquired using identical microscope and camera settings. Images represented the first exposure of cells to the excitation light (the signal fades after prolonged exposure). Each time-lapse series was recorded shortly after taking the first image shown in the figure. For reproduction, images and series were converted to 8-bit format, and unless stated otherwise, used without modifying the intensities. Unless stated otherwise, the exposure time for images and series was 0.1 s, the time between exposures in series was 2 s, and the number of exposures was 30. The series were sped up fivefold during conversion into QuickTime videos. In Videos 5, A and B, 100 pixels equals 6.79 μm. In all other Videos, 100 pixels equals 6.69 μm.
In this work, the conditions for live imaging of the GFP::nudF and GFP::nudA strains have been modified as follows. First, cells were grown on the surface of solid media rather than in liquid media. This allowed observations of isolated hyphal tips at the colony margin, as well as of the internal hyphal segments closer to the center of the colony. Unless stated otherwise, the hyphal tips selected for figures and videos were from the periphery of the colony. Second, threonine was used as a carbon source instead of glycerol to overexpress the GFP fusions. The transcription of the GFP::nudF and GFP::nudA genes is controlled by the inducible alcA promoter, whose activity is repressed by glucose and induced by alcohols (Creaser et al., 1985 ; Waring et al., 1989 ). Glycerol neither represses nor induces the alcA promoter. In contrast, threonine is a potent inducer of the alcA promoter (Creaser et al., 1985 ). Thus, the induction levels of the GFP::nudF and GFP::nudA genes used in this work should be much higher than in the previous studies (Xiang et al., 2000 ; Han et al., 2001 ; Zhang et al., 2002 ). It was estimated from immunoblots that the level of the GFP::NUDF protein was similar to the wild-type NUDF level in cells grown on glycerol, but was 20–40 times higher in cells grown on threonine (our unpublished data). The intensity of GFP fluorescence varied between different experiments and between different hyphae within the same sample when glycerol was used. Such variability was not observed when threonine was used.
The experiments described further in this work reveal that the nudE deletion is completely suppressed by the overexpression of the NUDF protein. Consistent with this finding, the GFP::nudF; ΔnudE strain was identical to the GFP::nudF strain when grown on threonine (high GFP::nudF induction), but was inhibited compared with the GFP::nudF strain when grown on glycerol (low GFP::nudF induction). In contrast, the GFP::nudA; ΔnudE strain was inhibited compared with the GFP::nudA on both threonine and glycerol. It was not possible to compare GFP::NUDF behavior in the GFP::nudF and GFP::nudF; ΔnudE strains grown on glycerol, due to the mentioned high variability of the GFP signal among different hyphae. In addition, the background fluorescence was often higher in the GFP::nudF; ΔnudE strain when grown on glycerol, possibly due to a positive selection for a higher level of GFP::nudF induction. To bring down the GFP::NUDF fusion level, threonine (100 mM) was used in combination with the alcA repressor glucose (10 mM). On the threonine plus glucose medium, the GFP::nudF; ΔnudE strain was inhibited compared with the GFP::nudF strain, whereas the GFP::nudF strain seemed normal. Again, the comets were present in the GFP::nudF; ΔnudE strain (Figure 1D and Video 1D), and no differences were obvious between the GFP::nudF; ΔnudE and GFP::nudF strains. The background fluorescence was lower, giving comets more contrast. Also, cells grew more vigorously in the presence of glucose (notice that the tip visibly elongates in Video 1D). Unfortunately, some variability in the background fluorescence among different hyphae was present when threonine was used in combination with glucose, thus making thorough comparisons of the two strains problematic.
Previous studies of the GFP-tagged NUDF and dynein/dynactin subunits described the comets near hyphal tips (Xiang et al., 2000 ; Han et al., 2001 ; Zhang et al., 2002 ). Figure 1C and Video 1C show a hyphal segment ~200 μm away from the tip. Clearly, the GFP::NUDF comets are present there and move in all directions. However, the comets were typically the brightest and most easy to observe near the tips (e.g., Figure 1D and Video 1D).
To examine intracellular localization of the full-length NUDE protein and its N- and C-terminal domains, these were fused to the GFP and placed on the multicopy plasmid under the native nudE promoter (Figure 3A). The plasmids expressing the GFP::NUDE and GFP::NUDE-N fusions suppressed the ΔnudE and nudF7 mutations, whereas the GFP::NUDE-C fusion did not (Figure 3A). All three constructs were present at similar levels in total protein extracts (Figure 3B). When A. nidulans strains transformed with the above-mentioned plasmids, were examined for the GFP fluorescence, considerable variability in the fluorescence intensity was observed among different hyphae (Figure 4B shows a typical example). The most likely cause of this variability was a variation in the copy number of the A. nidulans autonomously replicating plasmid due to its mitotic instability (Gems et al., 1991 ). The fluorescence intensity was always the same along the length of each individual hypha, apparently because the septa that divide the hyphae into compartments are perforated and allow passage of cytoplasm. Due to the extreme variability of the levels of the GFP-tagged proteins among different hyphae, the conclusions about the localization of the GFP::NUDE constructs had to be qualitative. That is, it was possible to determine whether particular structures (e.g., comets) were present, but the abundance and intensity of structures were different in each hypha. One benefit of this variability was that a broad range of expression levels could be examined within the same sample.
The full-length GFP::NUDE localized to comet-like structures identical to those of the GFP::NUDA and GFP::NUDF (Figure 4A). Close to hyphal tips, the comets tended to move predominantly toward the tip and were typically the brightest at the tip. The comets were readily observed proximal to the tips and in the internal compartments, where they moved in both directions. Video 4A shows several comets moving in opposite directions. The specks were also observed in older hyphal regions, where they coexisted with comets. The specks of GFP::NUDE behaved like the specks of the GFP::NUDA (Video 2C) and are described in detail below.
The nonfunctional GFP::NUDE-C also localized to comets (Figure 4C and Video 4C). These comets resembled those of the GFP::NUDA in the ΔnudE background in that they were more disorganized compared with the GFP::NUDE comets. This was expected because the expression was done in a ΔnudE mutant, and GFP::NUDE-C does not complement it. The specks were never observed with the GFP::NUDE-C fusion. Instead, judging from the fact that the maximum background fluorescence inside hyphae with GFP::NUDE-C was higher than with GFP::NUDE, the excess of the GFP::NUDE-C fusion distributed uniformly.
The functional GFP::NUDE-N fusion was observed only as a uniform fluorescence throughout the cytoplasm (Figure 4B). Hyphae with different levels of fluorescence were examined, and neither comets nor specks could be detected either in still images or in time-lapse series. Occasionally, the GFP signal seemed to accumulate in nuclei, but that did not happen in every hypha.
Figure 5A and Video 5A show a typical example of specks and their movements. The brightness of specks varied significantly, but even the brightest specks were sharp and tiny. Very bright specks may show as large round objects due to image reproduction artifacts. The movements of specks were jerky and unpredictable. The net result of the movements was that the specks distributed uniformly along the hypha. Interestingly, specks often moved in pairs as if they were connected. A thin line of fluorescence was sometimes seen between adjacent specks. Destabilization of microtubules with benomyl (4 μg/ml, 2–5 h at 28°C) did not eliminate the specks, but completely stopped their movement (Figure 5B and Video 5B). In older hyphae, the specks were incorporated into bright cables (Figure 5C). No movements were seen in such cables.
A ΔnudE strain producing green conidia was transformed with its own genomic DNA fragments in the multicopy vector pAid. Suppressed colonies were identified on transformation plates as patches of green color resulting from enhanced conidiation in the background of brownish, poorly conidiating mycelium. Suppressing plasmids were recovered from several such transformants and were found to represent two different genes (see MATERIALS AND METHODS). One gene had properties of a transcription factor and will be described elsewhere. The second gene turned out to be the nudF gene. The suppressor plasmid pAid::nudF recovered from one of the transformants carried a ~10-kb genomic DNA fragment with the nudF gene (1.3 kb) and no evidence of the nudE sequence. A much smaller, ~2-kb genomic DNA fragment with the nudF gene also suppressed the nudE deletion when cotransformed with pAid. The plasmid pAid::nudF6, which carries genomic DNA fragment with a ts allele of the nudF gene, suppressed the nudE deletion only partially and only at 32°C (our unpublished data).
Remarkably, suppression of the nudE deletion by pAid::nudF was total (Figure 6A): the ΔnudE[pAid::nudF] transformants were indistinguishable from the ΔnudE[pAid::nudE] wild-type control transformants under all conditions tested (YAG and YAGK at 32–43°C, M-glucose at 37°C). The radial growth rates were 14.5 ± 0.2 mm/d for both transformants vs. 9.2 ± 0.2 mm/d for the ΔnudE[pAid] control (37°C, YAG). The defects in conidia production mentioned above were also corrected. The NUDF protein level seemed to be unaffected by the deletion of the nudE gene (Figure 6B). The NUDF protein level increased ~10-fold after transformation with pAid::nudF, consistent with the copy number of ~10 per haploid genome for the A. nidulans autonomously replicating vector (Gems et al., 1991 ).
The pAid::nudF plasmid suppressed the ts nudC3 mutation (Figure 7A). This was expected because the nudF gene was isolated as a multicopy suppressor of the nudC3 mutation during cloning of the nudC gene (Xiang et al., 1995a ). Suppression was not complete, even under the most permissive conditions for the nudC3 mutation. As reported previously (Efimov and Morris, 2000 ), multiple copies of the ts mutant allele of the nudF gene, nudF6, inhibited the nudC3 mutant (Figure 7A). In an attempt to identify other nudC-interacting genes, a screen for multicopy suppressors of the nudC3 mutations was conducted using genomic DNA fragments from the nudF6 mutant to prevent isolation of the nudF gene. The screen did not produce any new genes except for the transcription factor-like gene mentioned above (see MATERIALS AND METHODS). Interestingly, several plasmids with a truncated nudC gene were isolated. They suppressed the nudC3 mutation more strongly than pAid::nudF, so that the only difference from the wild-type was a slightly reduced conidiation under the most restrictive conditions (43°C, YAG). Sequencing of one such plasmid, pAid::nudCΔ, showed that the insert terminates within the last nudC intron, resulting in the loss of 10 aa from the NUDC's C terminus.
Considering the diversity of genetic interaction discovered through the multicopy suppressor screens so far, it seemed promising to examine the effects of multicopy plasmids with the nudF and other genes using direct transformations of different dynein-related mutants. This approach revealed several new interactions described below. Figure 8 summarizes the effects of multicopy plasmids described herein and in Efimov and Morris (2000) .
The third apsA mutant analyzed was a ΔapsA strain in which 96% of the apsA coding region had been deleted (Fischer and Timberlake, 1995 ). The genotype of this strain precluded the use of pAid-derived plasmids. The ΔapsA strain was transformed with the pAid2-derived multicopy plasmids, which carry the argB gene instead of the pyrG gene as a selective marker. For this reason, the transformants could be analyzed only on minimal media. Complicating the analysis, the ΔapsA[pAid2] transformants grew at a slower rate than the wild-type control transformants ΔapsA[pAid2::apsA] (13.1 vs. 14.7 mm/d; M-glucose, 37°C; SE <0.4 mm/d), whereas the untransformed ΔapsA strain grew at the wild-type rate both on complete and minimal media supplemented with arginine. This could indicate that the empty vector pAid2 has an inhibitory effect, or is less efficient in complementing the argB mutation in the ΔapsA strain than its clones. Such differences between the transformed and untransformed strains were not observed with the pAid vector: strains transformed with pAid, including apsA1 and apsA5 mutants, grew at the same or marginally higher rates than the untransformed strains. Nevertheless, the colony radial growth rates of the ΔapsA[pAid2::nudF] transformants were reduced by 10% compared with those of the ΔapsA[pAid2] controls (11.8 vs. 13.1 mm/d; average for eight transformants each; M-glucose, 37°C; SE <0.4 mm/d). Similar inhibition was observed at 32, 37, and 43°C.
Given the effects of multiple copies of the nudF gene on the apsA mutants, it was interesting to examine whether there were any mutants affected by multiple copies of the apsA gene. pAid::apsA had a slight inhibitory effect on the mutants nudF6 and nudK317 (our unpublished data) and a more pronounced inhibitory effect on the nudC3 mutant (Figure 7D). The inhibition of the nudC3 mutant occurred under conditions partially restrictive for the nudC3 mutation (37°C, YAG; 43°C, YAGK). No effect could be observed at 32°C when, judging from the smallness of the improvement conferred by pAid::nudF, the NUDF protein function was largely normal. In contrast, inhibition by pAid::nudF6 was noticeable under all conditions, including 32°C. Also, it is clear from the magnitude of the nudC3 suppression by pAid::nudF at 43°C (Figure 7A) that most of the growth defects seen in the nudC3 mutant are due to the NUDF protein defect. Thus, the inhibitory effect of pAid::apsA on the nudC3 mutant could be due to an apsA-nudF interaction rather than an apsA-nudC interaction.
Mutations in the apsA and apsB genes of A. nidulans result in similar phenotypes (Clutterbuck, 1994 ; Fischer and Timberlake, 1995 ; Suelmann et al., 1998 ). An apsB mutant was transformed with multicopy plasmids and was found to be unaffected by pAid::nudF or other plasmids.
To characterize how the defects seen in the apsA mutants are related to the nudE and nudF genes, the apsA5 mutant was crossed to the ΔnudE and ΔnudF mutants. The colony radial growth rates of relevant strains are compared in Table 2. The effects of the apsA5 and ΔnudE mutations were additive: the apsA5; ΔnudE double mutants formed smaller colonies than either of the parents, but still bigger than the ΔnudF mutant. Conidiation in the double mutant was also less efficient than in either of the parents. On the other hand, the apsA5; ΔnudF double mutants were indistinguishable from the ΔnudF mutant.
Certain features of the NUDE protein hint at how it might facilitate the function of NUDF. The N-terminal coiled coil domain of all NUDE homologs is slightly >161 aa, which corresponds to a 24-nm-long coiled coil structure. The LIS1 binding part has been mapped roughly to the internal one-third of this coiled coil in the mouse NUDE (Feng et al., 2000 ). The sequence of the NUDE coiled coil is evolutionarily conserved over its entire length, including the regions upstream and downstream of the LIS1 (and by extrapolation, NUDF) binding region. This suggests that these regions bind other proteins in addition to NUDF/LIS1. Thus, the NUDE coiled coil may serve as a scaffold that facilitates formation of a complex between NUDF and other proteins. It should be noted that mammalian NUDE interacts with many centrosome components and has been proposed to function in centrosome organization (Feng et al., 2000 ). Because A. nidulans NUDE, NUDF, and dynein are not observed at spindle pole bodies, it is not clear whether NUDE has a similar role in fungi. A possible function of LIS1/NUDF is promoting assembly of functional dynein and dynactin complexes, because LIS1 overexpression increases the size of dynein and dynactin complexes and stimulates their retrograde movement (Smith et al., 2000 ). LIS1 coimmunoprecipitates with both dynein and dynactin and interacts with CDHC and dynactin's subunit dynamitin (Faulkner et al., 2000 ; Niethammer et al., 2000 ; Sasaki et al., 2000 ; Smith et al., 2000 ; Tai et al., 2002 ), even though dynein and dynactin are observed as a complex in vitro only under special conditions (Kini and Collins, 2001 ; Kumar et al., 2001 ). It is possible that, acting as a scaffold, NUDE coiled coil stabilizes intermediate complexes between NUDF/LIS1 and dynein or dynactin, which ultimately assemble into a fully active motor complex. This could explain why increased NUDF concentration bypasses the requirement for NUDE. The dispensable C-terminal domain of NUDE, which is required for NUDE localization (see below), may have evolved to target the protein more precisely to the sites where dynein, dynactin, and NUDF are assembled, such as microtubule ends.
Localization to the ends of dynamic microtubules is common for dynein/dynactin and other microtubule-interacting proteins, but the mechanism and significance of this localization are still under investigation (reviewed by Schroer, 2001 ; Schuyler and Pellman, 2001 ; Dujardin and Vallee, 2002 ). The hierarchy of protein interactions at microtubule ends also remains to be determined. The A. nidulans CDHC has been observed in comets in the absence of NUDF (Zhang et al., 2002 ). This work shows that the CDHC and NUDF proteins do not need the NUDE protein to localize to comets (Figures 1 and 2 and Videos 1 and 2). However, because the nudE deletion is completely suppressed by NUDF overexpression, the possibility that NUDE facilitates NUDF localization to comets cannot be ruled out. Even though the functional and NUDF-binding NUDE-N is not seen in comets, it is still possible that it facilitates NUDF targeting to comets by transiently binding both NUDF and a comet's component. It is likely that the interaction between NUDF and NUDE-N is transient. Were it permanent, the NUDE-N would be seen in comets where NUDF is concentrated.
The changes in the behavior of GFP::NUDA comets in the A. nidulans ΔnudE mutant, observed in this work by using live imaging, mirror the changes in the CDHC and dynactin localization in the N. crassa Δro-11 mutant observed by indirect immunofluorescence (Minke et al., 1999 ). In both cases, the comets (streaks in Minke et al., 1999 ) were more prominent and oriented more randomly. In addition, a cloud of diffuse signal around the cluster of comets was frequently observed in this work. A possible cause of these changes is an altered dynamics of microtubules resulting from a compromised dynein/dynactin function (Han et al., 2001 ). It is also possible that the comets are an accumulation of inactive dynein/dynactin complexes (e.g., on vesicles), which periodically get activated and travel back (Seiler et al., 1999 ). A compromised dynein activity in the absence of RO11/NUDE would block the retrograde transport and result in a larger accumulation on dynein/dynactin complexes.
The strong inhibition of the nudA1 mutant, but not of three other ts nudA mutants, by multiple copies of the nudF gene is the first example of an allele-specific interaction between the nudF and CDHC genes. The existence of such an allele-specific interaction is consistent with a direct binding of NUDF/LIS1 to CDHC that has been observed in two-hybrid and coexpression/coimmunoprecipitation assays (Sasaki et al., 2000 ; Tai et al., 2002 ). An increase in NUDF concentration could result in a more robust formation of a complex between NUDF and the mutant CDHC encoded by the nudA1 allele. All ts nudA mutations used in this study reduce CDHC protein level at restrictive temperatures, and thus are likely to impair CDHC stability or folding (Xiang et al., 1995b ). Binding of NUDF to the nudA1-encoded mutant protein, especially premature binding, could further destabilize it or trap it in a wrong folding conformation.
The genetic interactions between the nudF and apsA genes are the first evidence connecting the APSA protein of A. nidulans to the cytoplasmic dynein pathway. That APSA might function in the dynein/dynactin pathway is not obvious from the phenotype of apsA mutants. Although they display mild nuclear distribution defects (Clutterbuck, 1994 ; Fischer and Timberlake, 1995 ), they are much healthier than dynein/dynactin mutants (Table 2). This observation implies that, unlike Num1p, which is required for all dynein functions in yeast (Geiser et al., 1997 ; Heil-Chapdelaine et al., 2000 ; Farkasovsky and Küntzel, 2001 ), APSA is only required for a subset of dynein functions in A. nidulans. Consistent with this, the apsA5; ΔnudF double mutant is identical to the ΔnudF mutant. The observed additivity of the apsA5 and ΔnudE mutations does not contradict APSA being in the dynein pathway, because the ΔnudE mutation is like a partial loss of the NUDF function, as evidenced by the complete suppression of the ΔnudE mutation by NUDF overexpression. Interestingly, the apsB14 mutation suppresses the ΔnudF and ΔnudA mutations (Efimov, unpublished data). This and the facts that APSA and APSB localize to different structures and do not coimmunoprecipitate (Suelmann et al., 1998 ) indicate that APSB functions independently of dynein and APSA, despite the similarity of the apsA and apsB mutant phenotypes. It is possible that genetic interactions between the nudF and apsA genes reflect physical interactions among NUDF, APSA, and dynein or dynactin. A biochemical evidence for an association between Num1p and cytoplasmic dynein in S. cerevisiae has recently been provided (Farkasovsky and Küntzel, 2001 ). Additional experiments are needed to identify the components of dynein or dynactin that bind APSA and the role of NUDF in these interactions. Given APSA localization at the plasma membrane and septa, such interactions would imply dynein presence at the cortex or septa of A. nidulans. Although such localization has not been observed so far, association with the ends of dynamic microtubules ideally positions dynein for probing the intracellular space for anchoring factors, perhaps in a manner similar to the capture of vesicles by dynactin associated with the ends of cytoplasmic microtubules in vertebrate cells (Vaughan et al., 2002 ). Dynein mediated sliding of astral microtubules along the cortex has been observed in S. cerevisiae and Schizosaccharomyces pombe, but detecting dynein and dynactin at the cortex has proved to be hard to accomplish (reviewed by Dujardin and Vallee, 2002 ). In S. pombe, accumulation of CDHC at sites of contact between the cortex and the ends of astral microtubules that pull the nucleus has been observed (Yamamoto et al., 2001 ). It will be interesting to see whether the S. pombe homolog of APSA/Num1p is the cortical dynein-anchoring factor postulated in the latter work, and whether similar proteins function in higher eukaryotes.
I thank Xin Xiang for providing anti-NUDF antibodies and A. nidulans strains, particularly the strains with the GFP-tagged nudF and nudA genes; Reinhard Fischer for providing apsA mutants and plasmids; and N. Ronald Morris for the use of laboratory space and equipment. This work was supported by a Scientist Development Grant from the American Heart Association.
The apsA1 and apsA5 strains grow slower than the apsA deletion strain (Table 2) because of the pyrG89 mutation.