Cytoplasmic dynein performs multiple cellular tasks but its regulation remains unclear. The dynein heavy chain has a N-terminal stem that binds to other subunits and a C-terminal motor unit that contains six AAA (ATPase associated with cellular activities) domains and a microtubule-binding site located between AAA4 and AAA5. In Aspergillus nidulans, NUDF (a LIS1 homolog) functions in the dynein pathway, and two nudF6 partial suppressors were mapped to the nudA dynein heavy chain locus. Here we identified these two mutations. The nudAL1098F mutation resides in the stem region, and nudAR3086C is in the end of AAA4. These mutations partially suppress the phenotype of nudF deletion but do not suppress the phenotype exhibited by mutants of dynein intermediate chain and Arp1. Surprisingly, the stronger ΔnudF suppressor, nudAR3086C, causes an obvious decrease in the basal level of dynein's ATPase activity and an increase in dynein's distribution along microtubules. Thus, suppression of the ΔnudF phenotype may result from mechanisms other than simply the enhancement of dynein's ATPase activity. The fact that a mutation in the end of AAA4 negatively regulates dynein's ATPase activity but partially compensates for NUDF loss indicates the importance of the AAA4 domain in dynein regulation in vivo.
CYTOPLASMIC dynein is a microtubule motor that plays multiple roles in mitosis and organelle distribution and in transport of vesicles, proteins/mRNAs, and viruses (reviewed by Vale 2003; Greber and Way 2006; Levy and Holzbaur 2006; Vallee and Hook 2006). However, it is not yet clear how cytoplasmic dynein is targeted to various cellular locations and how its motor activity is regulated. Cytoplasmic dynein in higher eukaryotic organisms has been purified as a multi-subunit complex with a molecular mass >1 mDa. It consists of heavy chains (HCs) (∼500 kDa), intermediate chains (ICs) (∼74 kDa), light intermediate chains (50–60 kDa), and light chains (8, 14, and 22 kDa) (reviewed by Pfister et al. 2006). Many proteins involved in cytoplasmic dynein function have been discovered, including proteins in its accessory complex, dynactin (reviewed by Schroer 2004), and also the LIS1 protein (reviewed by Gupta et al. 2002; Hatten 2005). Lis1 was initially identified as a causal gene for human lissencephaly, a disease associated with abnormal brain development (Reiner et al. 1993). The connection between LIS1 homologs and cytoplasmic dynein was first made in the filamentous fungus Aspergillus nidulans and the budding yeast Saccharomyces cerevisiae by genetic studies (Xiang et al. 1995a; Geiser et al. 1997). Further studies have demonstrated that LIS1 and its homologs in higher eukaryotic systems are also involved in cytoplasmic dynein function, and physical interactions between LIS1 and dynein/dynactin have been shown (Liu et al. 2000; Sasaki et al. 2000; Dawe et al. 2001; Tai et al. 2002; reviewed by Wynshaw-Boris and Gambello 2001; Gupta et al. 2002; Tsai and Gleeson 2005; Vallee and Tsai 2006). Recently, purified LIS1 has been shown to enhance the microtubule-stimulated ATPase activity of the dynein motor (Mesngon et al. 2006).
Cytoplasmic dynein heavy chain, which contains the ATPase and the microtubule-binding domains, is responsible for motility. The N-terminal one-third of the protein forms the stem region, which includes the sites for heavy chain dimerization and interaction sites between heavy chain and intermediate chain, light intermediate chain, and LIS1 (Habura et al. 1999; Tynan et al. 2000; Tai et al. 2002). The stem region is followed by the motor head containing six AAA domains that form a ring-like structure (Samso et al. 1998; Burgess et al. 2003; Samso and Koonce 2004). AAA1 has been shown to be the major ATP hydrolysis site on the basis of the UV-vanadate-mediated photocleavage assay (Gibbons et al. 1987). But mutational analyses suggest that ATP binding and hydrolysis at AAA3 are also critical for dynein function and allow dynein to be released from microtubules (Silvanovich et al. 2003; Kon et al. 2004; Reck-Peterson and Vale 2004). ATP hydrolysis at AAA2 or AAA4 does not seem to be essential (Reck-Peterson and Vale 2004), but ATP binding at AAA2 or AAA4 may enhance the microtubule-binding or the ATPase activity of dynein (Kon et al. 2004; Reck-Peterson and Vale 2004). Between the fourth and the fifth AAA domains, there is a microtubule-binding stalk of ∼10–15 nm in length (Goodenough and Heuser 1984; Gee et al. 1997; Samso et al. 1998; Burgess et al. 2003). It is not entirely clear how these domains affect each other during the ATPase cycle to produce mechanical force. Electron microscopic analyses of a flagella dynein have suggested that the stem connects to the motor head through a linker ∼10 nm long, and a change in the orientation of the linker correlates with dynein power stroke that results in an ∼15-nm displacement of the tip of the microtubule-binding stalk (Burgess et al. 2003). This linker may correspond to a region, ∼600 amino acids before the first AAA domain, which may affect the ATPase cycle of the AAA1 domain (Gee et al. 1997; Vallee and Hook 2006). Taken together, the dynein motor is organized in a unique fashion different from other motors such as kinesins and myosins (King 2000; Asai and Koonce 2001; Burgess and Knight 2004; Koonce and Samso 2004; Vallee and Hook 2006). The structure of the dynein motor is complex and also recent studies on dynein behaviors have added another layer of complexity onto this motor (Mallik et al. 2004; Reck-Peterson et al. 2006; Ross et al. 2006; Toba et al. 2006). For example, unlike conventional kinesins that walk in one direction toward the plus end, the cytoplasmic dynein–dynactin complex exhibits energy-dependent bidirectional movements in single-molecule motility assays (Ross et al. 2006).
In live cells, proteins in cytoplasmic dynein and/or dynactin complex form comet-like structures representing their accumulation at the microtubule plus end, a site implicated in microtubule–cortex interaction and in dynein cargo loading (Han et al. 2001; Vaughan et al. 2002; Lee et al. 2003; Sheeman et al. 2003; Lenz et al. 2006). In A. nidulans, the plus-end accumulation of cytoplasmic dynein depends on a conventional kinesin–KINA and dynactin, but deletion of NUDF (LIS1 homolog) makes the comets longer, thereby increasing the sum of comet intensity (Zhang et al. 2003). These results are very similar to the result from the dimorphic fungus Ustilago maydis in which the microtubule plus-end localization of dynein is implicated in transporting endosomes from the microtubule plus end toward the minus end in hyphae (Lenz et al. 2006). Consistently, dynein comets are more prominent in the absence of NUDF-interacting protein NUDE/RO11 in A. nidulans and Neurospora crassa (Minke et al. 1999; Efimov 2003). Thus, while LIS1 homologs in filamentous fungi may be targeted to the microtubule plus end via mechanisms different from that of dynein's plus-end targeting (Efimov et al. 2006), they are not required for dynein's plus-end localization, but instead may facilitate dynein's departure from the microtubule plus end. It is important to note, however, that in S. cerevisiae, the LIS1 homolog Pac1p is required for dynein's microtubule plus-end accumulation (Lee et al. 2003; Sheeman et al. 2003). These results suggest that LIS1 homologs may regulate dynein in multiple modes.
In a previous genetic study in A. nidulans, Willins et al. (1997) isolated extragenic suppressors of the nudF6 mutation (snf) that partially suppress the nudF6 phenotype, and two of the mutations in the snfC locus were shown to locate within or extremely close to the dynein heavy chain gene (Willins et al. 1997). To gain better insights from these initial findings, we have sequenced the heavy chain genes of the suppressor strains and found the positions of these mutations. These two mutations are located in two different domains of the dynein heavy chain: one, nudAL1098F, is in the stem region, and the other, nudAR3086C, is in the end of the fourth AAA domain (AAA4), but not in the ATP-binding (Walker A) or the hydrolysis (Walker B) site. These mutations partially suppress the ΔnudF mutant phenotype but not the phenotype of dynein intermediate chain and Arp1 mutants. The nudAR3086C mutation located in the end of AAA4 is a relatively stronger suppressor of nudF mutants. Interestingly, while dynein containing this mutation interacts with NUDF just like wild-type dynein, its basal-level ATPase activity is obviously decreased. This is also correlated with an increase in dynein's distribution along cytoplasmic microtubules. We suggest that these mutations may specifically alter dynein to partially compensate for NUDF loss, but the alterations may not necessarily increase dynein's ATPase activity.
MATERIALS AND METHODS
Strains and Aspergillus techniques:
A. nidulans strains are listed in Table 1. Media, growth conditions, genetic crosses, A. nidulans transformations, and genomic DNA isolation were as described previously (Xiang et al. 1995a,b; Willins et al. 1995, 1997). In many experiments described in this article, we transformed A. nidulans strains in which the A. nidulans Ku70 homolog was deleted (ΔnkuA) (Nayak et al. 2006), with individual genomic fragments plus the plasmid pAid containing the pyrG selective marker (Xiang et al. 1999). The ΔnkuA strains made most of the described experiments feasible due to an increased frequency of homologous integration of a genomic fragment into the A. nidulans genome (Nayak et al. 2006).
Molecular biology techniques:
Overlapping regions of the nudA genomic DNA were amplified with multiple sets of primers in the nudA coding sequence. The PCR products were sequenced and the sequence data were analyzed using the DNA Star program. Oligonucleotides used for PCR amplification of the nudA genomic DNA and for sequencing analyses are listed in supplemental Table 1 at http://www.genetics.org/supplemental/.
Construction of a GFP-nudA strain in which the GFP-nudA fusion is under the control of its own promoter:
We have previously constructed a strain in which GFP is inserted in the nudA coding region between the eighth and the ninth aa and the GFP-nudA fusion is under the control of the alcA promoter (Xiang et al. 2000). In medium containing glucose, the expression of the GFP-nudA fusion is repressed, and our previous observations of GFP-NUDA were all done using medium containing glycerol as the carbon source. In this study, we constructed a new strain in which the GFP-nudA fusion is under the control of the endogenous promoter of nudA, and the GFP-NUDA fusion protein in such a strain can be observed using regular medium containing glucose. To make this new strain, a 1.5-kb fragment upstream of the nudA coding sequence was amplified from genomic DNA of the wild-type strain GR5 by using the following primers: NudAupper (5′-AGTAAGCCAAGTTTCTACCGAACG-3′) and NudAlower (5′-TTCGCGGGCAGATAGAGTTTTCCCGCGACGCTGGTCCGTAAC-3′). A 2.3-kb SmaI and BamHI fragment was obtained from the original alcA-GFP-nudA plasmid (Xiang et al. 2000). Subsequently, a fusion PCR was performed to fuse the 1.5- and 2.3-kb fragments with the following primers: NudAupper (5′-AGTAAGCCAAGTTTCTACCGAACG-3′) and NudAend (5′-CTTCATTAAGGTGCGGAATTCGCG-3′). This resulted in a 3.8-kb product that contains the nudA upstream region followed by the nudA coding sequence with GFP inserted. This fragment was cotransformed with the auto-replicating plasmid pAid that contains the selective marker pyrG (Xiang et al. 1999) to the TNO2A3 strain with ΔnkuA (Nayak et al. 2006). Transformants were screened under a fluorescence microscope for the presence of the comet-like structures near the hyphal tip, which represent GFP-NUDA accumulated at the dynamic microtubule plus ends. Several strains with positive GFP signals were selected. A strain (LZ12) that subsequently lost the auto-replicating plasmid pAid after growing on the nonselective medium Y + UU + FOA was used for further experiments.
The GFP-nudA strain was crossed to nudF mutant strains, and the strains carrying both GFP-nudA and the nudF mutations were obtained. To screen these strains for the presence of ΔnkuA, we sequenced the nKuA locus using the primers ktF (5′-cgtcgtacaggtaccaggactttc-3′) and ktR (5′-ctgcaattattgcatgcgttcatc-3′) (Nayak et al. 2006). The ΔnkuA-positive strains were then transformed with fragments to confirm that the identified nudA dynein heavy chain gene mutations cause suppression of the nudF mutant phenotype.
Construction of the S-IC strain carrying an S-tagged dynein intermediate chain (nudI) gene:
We constructed a plasmid containing an N-terminal truncated nudI (dynein intermediate chain) gene tagged at its C-terminal coding region with an S-tag (a peptide in RNase A). Specifically, a DNA sequence encoding a 15-amino-acid S-tag peptide (Novagen) was added to the 3′-end of the nudI coding region immediately before the stop codon. The nudI coding region and its 3′-untranslated region were used as templates in PCR reactions using the following primer sets: S-tag1 (5′-AAA GAA ACC GCT GCT GCT AAA TTC GAA CGC CAG CAC ATG GAC AGC TAA TTG ATT GGA TCT AC-3′) and zh5 (5′-CCC CAC CGC GGT GGC AAT AAT GAA TAG GGA C-3′) and S-tag2 (5′-GCT GTC CAT GTG CTG GCG TTC GAA TTT AGC AGC AGC GGT TTC TTT GCT CAT TCG GTC CTT ATC C-3′) and zh1 (5′-AAG CGG CCG CAG ATG GCC TTT CAA G-3′) (the nucleotides corresponding to the S-tag are underlined). S-tag1 and zh5 were used to generate a PCR product containing the S-tag, the stop codon, and the nudI sequence downstream of the stop codon. S-tag2 and zh1 were used to generate a PCR product containing a 5′ truncated nudI coding region and the S-tag. The two PCR products were annealed and used as a template in a second PCR reaction with zh1 and zh5 as primers. The fusion PCR product was cloned into the NotI–BstXI sites of a vector pXX1 (containing the selective marker pyrG) (Xiang et al. 1995b) to generate a plasmid named “pS-tag.” The in-frame insertion of the S-tag immediately before the stop codon was confirmed by DNA sequencing. This plasmid was transformed into the nudI416 temperature-sensitive (ts) mutant to determine if S-tagged-nudI is functional. ts+ transformants were obtained, indicating that pS-tag integrated into the nudI locus and rescued the nudI416 phenotype. The presence of the S-tag in the genome was confirmed by PCR reactions using chromosomal DNA as a template and with the primers S-tag3 (5′-GCTGGCGTTCGAATTTAGC-3′) and zh1. Using FOA selection, we obtained a strain in which the pyrG selective marker was lost, and this strain, JZ11 or S-IC, was used for genetic crosses to obtain strains with the suppressor mutations in the background of S-tagged nudI.
A.nidulans dynein purification and ATPase assay:
A. nidulans protein extract was obtained from an overnight culture of 1 liter using the liquid nitrogen grinding method for breaking the hyphae, which was similar to what has been described previously (Zhang et al. 2002), except that the protein isolation buffer contains 25 mm Tris (pH 8.0), 0.4% Triton X-100, 10 μm ATP, 1 mm DTT, and a protease inhibitor cocktail (Sigma, St. Louis). The S-tag is a 15-amino-acid peptide that binds to the S-protein (the larger fragment of RNase A produced by subtilisin digestion) attached to agarose beads (Novagen), and the S-tag-based purification method has been previously used to purify NUDF in A. nidulans (Ahn and Morris 2001). For purification of A. nidulans dynein from strains with the S-IC background, ∼30 ml of a protein extract (∼10 mg/ml) was incubated for half an hour at room temperature with 0.5 ml of S-protein beads (Novagen). The beads were repeatedly washed with the same buffer used for protein isolation except that no detergent was added. Finally, the S-tagged protein was eluted with 5 mg/ml S-peptide in 0.2 ml of a buffer containing 25 mm Tris–HCl (pH 8.0), 10 μm ATP, and 1 mm DTT. About 30 μl of the eluate was loaded onto a 4–15% SDS–PAGE gradient gel (Bio-Rad, Hercules, CA). Dynein heavy chain, intermediate chain, and NUDF were detected by Western blot analyses using an antibody against NUDA (Xiang et al. 1995b), S-protein-conjugated alkaline phosphatase (Novagen), and an anti-NUDF antibody (Xiang et al. 1995a). For silver staining of protein gels, the Silver Stain Plus kit from Bio-Rad was used.
ATPase assays were done by incubating a mixture of 30 μl containing the S-peptide-eluted fraction, 25 mm Tris–HCl (pH 8.0), 10 mm MgCl2. and 1 mm ATP at 37° for 1 hr. Twenty microliters of the mixture was measured for the level of free phosphate using the PhosFree phosphate assay biochem kit (Cytoskeleton). A multiple-channel pipette was used to simultaneously add reagents to eight individual samples contained in separate wells of a 96-well plate. The MRX Microplate Reader with Revelation Software (Version 4.06) (Dynex Technologies) was used for reading the OD values at 630 nm. Concentrations of the dynein heavy chain in different samples were estimated by comparing the intensity of the heavy chain band to differently diluted BSA samples on the same silver-stained gel. The image of the gel was acquired using LAS-1000 (Fujifilm) with the 1R LAS-1000 Life software, and quantification of the bands was done using the Image Gauge software (version 4.01).
Fluorescence microscopy of live A. nidulans hyphae was as described (Li et al. 2005). Images were captured using an Olympus IX70 inverted fluorescence microscope (with a ×100 objective) linked to a cooled CCD camera. The IPLab software (BioVision Technologies) was used for image acquisition and analysis. Cells were incubated at 32° in ΔTC3 culture dishes (Bioptechs, Butler, PA) overnight before observation at 32°. Liquid minimal medium containing glucose and supplements was used. A Bioptechs heating stage and heating objective system was used. Single GFP images were obtained by using a 0.1-sec exposure time. For imaging cells with both CFP-tubA and GFP-nudA, Ludl electronic products dual individual excitation and emission motorized filter wheels were used. Chroma 8600 filters for cyan fluorescent protein (CFP) (430-nm peak excitation with a bandwidth of 25 nm and 470-nm peak emission with a bandwidth of 30 nm) were used for observing the CFP microtubule, and filters for yellow fluorescent protein (500-nm peak excitation with a bandwidth of 20 nm and 535-nm peak emission with a bandwidth of 30 nm) were used for observing GFP-NUDA. A 0.2- and a 0.6-sec exposure time were used for capturing CFP microtubule and GFP-NUDA images, respectively.
The two snfC mutations reside in the N-terminal stem region and the end of the fourth AAA domain, respectively:
We sequenced the entire coding region of the A. nidulans cytoplasmic dynein heavy chain gene (nudA) in the two snfC strains. We found that each snfC mutant contains one single base-pair mutation. The nudA gene in the snfC1524 mutant contains a change in nucleotide 3294 (starting from ATG) from A to T, corresponding to a change in amino acid 1098 from leucine to phenyalanine. This change is located in the dynein heavy chain's stem region (Figure 1A). Interestingly, the snfC1232 mutation is located at the end of the fourth AAA domain, near the microtubule-binding stalk that contains coil-coiled motifs, which is far away from the snfC1524 mutation in the primary sequence (Figure 1A). The snfC1232 mutation changes nucleotide 9258 from C to T, and the corresponding amino acid 3086 is changed from arginine to cysteine. We did a sequence alignment with dynein heavy chains from several species and found that both L1098 and R3086 in the A. nidulans cytoplasmic dynein heavy chain are conserved from fungi to humans, suggesting that they are important residues for dynein function (Figure 1B).
Because we sequenced only the coding regions of nudA in these two snfC strains, we were concerned about the possibility that other mutations upstream or downstream of the nudA coding sequence could be causally related to the partial suppression. To verify that the single mutations that we found are the only ones that cause the partial suppression, we did the following experiments. We first amplified two fragments within the nudA coding region and, in each of them, the single mutation is located near the middle. The following two sets of primers—(1) NudA52 (5′-CAACTTCTGGTTATCCATGGA-3′) and NudA36 (5′-TTGGATCTACCAGCATAGCCA-3′) and (2) NudA57 (5′-TCCAAGCTATGGGTCGTATCT-3′) and NudA312 (5′-AGAGCATCGACCTTGTAGCTT-3′)—were used for PCR reactions to amplify a 5.2-kb nudA fragment containing the L1098F mutation and a 6.2-kb fragment containing the R3086C change, respectively. Since the previous genetic study suggested that these mutations represent bypass suppressors (Willins et al. 1997), we used different nudF alleles to test for suppression. To verify that the L1098F mutation in the stem region caused nudF suppression, the 5.2-kb fragment was transformed into a strain (LZ35) that carries the nudF7 mutation and the GFP-tagged nudA dynein heavy chain controlled by its own promoter. A transformant that formed a bigger, nudF7-suppressor-like colony was selected for sequencing analysis, and a codon change corresponding to the L1098F mutation was found. To rule out the possibility that other mutations in the 5.2-kb fragment introduced during the PCR procedure may contribute to the partial suppression, we sequenced the entire corresponding 5.2-kb genomic sequence of this transformant and confirmed that the L1098F change was the only amino acid change in this region. This result is consistent with the idea that the L1098F mutation partially suppressed the nudF7 mutant phenotype. To further demonstrate that the observed suppression is indeed linked to the mutation in nudA, we crossed this transformant to the ΔnudF mutant and analyzed the progeny. Because the nudF7 mutant forms only nud-like colonies at 42° but the ΔnudF mutant forms nud-like colonies at any temperature, including 32°, we were able to select the ΔnudF colonies and ΔnudF-suppressor-like colonies at 32°. If the suppressor mutation introduced into the nudF7 strain (with GFP-nudA in the same genome) is linked to nudA, then the ΔnudF-suppressor-like phenotype should be linked to the GFP-NUDA signal. Indeed, five randomly picked progeny with the ΔnudF-suppressor-like phenotype all showed GFP-NUDA signal under the microscope. These results strongly suggested that the L1098F mutation caused the nudF partial suppression phenotype observed in the snfC1524 mutant and confirmed that the L1098F change represents a bypass suppressor rather than an allele-specific suppressor of nudF (Figure 2).
To verify that the R3086C change caused the partial suppression phenotype observed in the original snfC1232/nudF6 strain, we used a slightly different strategy. Because snfC1232 by itself exhibited a conidiation phenotype (Figure 2) (Willins et al. 1997), we transformed the 6.2-kb fragment with the R3086C mutation to a wild-type strain containing GFP-tagged nudA under the control of its own promoter. A transformant that grew like a snfC1232 single mutant was selected and sequence analysis of the entire corresponding 6.2-kb region in the genomic DNA confirmed the presence of only the R3086C amino acid change. This strain was then crossed to the ΔnudF mutant, and both ΔnudF and ΔnudF-suppressor-like colonies were analyzed. None of the five randomly selected ΔnudF-like progeny had GFP signals while five suppressor-like progeny all had GFP-NUDA signals. Because the GFP-NUDA signal is linked to the R3086C change in the original transformant, our result strongly supports the conclusion that the R3086C change in the end of the fourth AAA domain partially suppresses the phenotype caused by loss of nudF.
The nudAR3086C mutation acts as a stronger ΔnudF suppressor than nudAL1098F and exhibits a mild growth defect:
The previous genetic study has suggested that both snfC alleles act as nudF bypass suppressors, but the extent of suppression was not shown (Willins et al. 1997). In this study, by using sequencing analyses and/or the linkage to the GFP-NUDA signal, we isolated strains containing each suppressor mutation in the ΔnudF background. Compared to nudAL1098F, nudAR3086C acts as a stronger suppressor of the ΔnudF phenotype both at 32° and at 42° (Figure 2). The level of suppression by nudAR3086C on the nudF6 phenotype is also clearly higher than that by nudAL1098F (Figure 2), which is consistent with data from the previous study (Willins et al. 1997). It has been shown previously that nudF mutants exhibited a conidiation (asexual spore formation) defect (Xiang et al. 1995a), and the suppressor mutations did not obviously restore conidiation of the nudF mutants. DAPI staining showed that the nuclear distribution defect in the vegetative hyphae of the ΔnudF mutant was clearly suppressed partially by the nudAR3086C mutation, but not obviously suppressed by the nudAL1098F mutation (Figure 3). The nudAR3086C mutation also acted as a stronger suppressor for the nuclear distribution phenotype of nudF6 (Figure 3), which is consistent with the previous report (Willins et al. 1997).
The previous genetic study has suggested that the snfC1232 (nudAR3086C) single mutant has a conidiation defect but the snfC1524 (nudAL1098F) mutant has no growth phenotype on its own (Willins et al. 1997). Our current results obtained with a genetic cross and sequencing analyses of selected progeny agreed well with this conclusion but revealed that the nudAR3086C single mutant formed a colony with a slightly reduced size (Figure 2). When the strain carrying nudF6/nudAR3086C was crossed to a wild-type strain, progeny of four different sizes appeared when plated at 42°: (1) the wild-type progeny that formed big and healthy colonies, (2) the nudF6 mutant that formed small colonies, (3) the nudF6/nudAR3086C-like colonies that were bigger than the nudF6 mutant but still significantly smaller than the wild-type ones, and (4) the nudAR3086C-like colonies that were slightly smaller than the wild type and had a clear conidiation defect (Figure 2).
However, when the strain carrying nudF6/nudAL1098F was crossed to a wild-type strain, progeny of three different sizes appeared: the nudF6-like and the nudF6/nudAL1098F-like colonies and the wild-type-like colonies that included both wild type and the nudAL1098F single mutant whose genotypes were confirmed by sequencing analyses. Thus, while the stronger suppressor, nudAR3086C, causes a clear conidiation defect and a mild colony growth defect, the nudAL1098F mutant has no obvious colony phenotype (Figure 2). It should be pointed out that although the colony phenotype exhibited by the nudAR3086C single mutant was easily detectable, the defect in nuclear distribution pattern was not severe (Figure 3).
Neither nudAR3086C nor nudAL1098F suppresses the nud phenotype exhibited by the nudI (dynein intermediate chain) and nudK (Arp1 of the dynactin complex) mutants:
To determine whether these two suppressor mutations specifically compensate for the loss of NUDF or act as general nud suppressors by somehow enhancing the overall function of dynein, we examined whether these two mutations suppress the phenotype of nudI (dynein intermediate chain) and nudK (Arp1 of the dynactin complex) mutants.
Dynein intermediate chain and the dynactin complex are implicated in linking dynein to its membranous cargoes and in increasing dynein motor processivity (Schroer 2004). If NUDF's function in dynein regulation differs from that of the dynein intermediate chain and dynactin, then nudF-specific suppressor mutations should not compensate for the loss of functions associated with these proteins. We crossed strains carrying GFP-nudA (as a control), GFP-nudAR3086C, and GFP-nudAL1098F, respectively, to the nudI416 and nudK317 ts mutants (Xiang et al. 1999; Zhang et al. 2002). The progeny of the crosses were plated out at the restrictive temperature of 42°. For the cross between the strain carrying GFP-nudAL1098F and the nudI416 or the nudK317 mutant, only the nud-like progeny and the wild-type progeny were found, similar to the crosses using the control strain carrying GFP-nudA. Similar results were obtained with crosses using the GFP-nudAR3086C strain, except that progeny with even smaller colony size than that formed by the original nudK317 ts mutant were observed, which was likely due to the additive effect of the GFP-nudAR3086C and nudK317 mutations. To confirm this result, we picked eight nud-like progeny from each cross and analyzed the presence of GFP-NUDA signals at the permissive temperature. As expected, for the cross between the GFP-nudAR3086C and nudK317 strains, GFP-NUDA signals were observed only in progeny that formed smaller colonies than the original nudK317 mutant. For the other crosses, some nud-like progeny exhibited GFP-NUDA signals while other nud-like progeny did not. Since the presence of the GFP-NUDA signal is linked to the suppressor mutations, and the colonies with GFP-NUDA are not bigger than colonies without GFP-NUDA (Figure 2), these analyses confirmed the notion that neither suppressor mutation suppresses the nud colony phenotype exhibited by the nudI416 and nudK317 mutants. We further examined the nuclear distribution phenotype of progeny containing one of the suppressor mutations (GFP-NUDA positive) and either the nudI416 or the nudK317 mutation, and as expected, there was no suppression of the nuclear distribution phenotype (Figure 3).
The nudAR3086C mutation enhances GFP-dynein's distribution along cytoplasmic microtubules:
We next examined the effect of these two nudF suppressor mutations on dynein's microtubule-plus-end localization. In the nudF7/GFP-nudAL1098F strain, GFP-NUDA formed comet-like structures similar to that in the wild-type GFP-nudA strain and in the nudF7/GFP-nudA strain grown at 32° (the nudF7 mutant grows like a wild type at the permissive temperature of 32°). Thus, the nudAL1098F mutation did not change dynein's plus-end accumulation significantly (supplemental movies 1–3 at http://www.genetics.org/supplemental/). In contrast, the nudF7/GFP-nudAR3086C or the GFP-nudAR3086C strain exhibited an obvious alteration in dynein localization. The GFP-NUDA proteins formed small punctates along filament-like structures that most likely represented microtubules (Figure 4; supplemental movies 4 and 5 at http://www.genetics.org/supplemental/). To verify that the signals indeed occurred along microtubules, we crossed the CFP-tubA strain (Li et al. 2005) with the GFP-nudAR3086C strain and found colocalization of GFP-NUDA with CFP microtubules in hyphal tips (Figure 4). It should be pointed out that GFP-NUDA in wild-type cells also formed some small dots along microtubules, but the extent of decoration along the microtubule was much less dramatic compared to that caused by GFP-nudAR3086C. In the ΔnudF strain, GFP-NUDA comets seemed longer and more prominent (Figure 4), which is similar to what we have shown previously (Zhang et al. 2003). In the ΔnudF/GFP-nudAR3086C strain, the comet-like structures could still be seen, but localization along microtubule-like filaments was also found (Figure 4), although in most cells, the signals were less intense compared to that in the GFP-nudAR3086C strain.
The nudAR3086C mutation causes a decrease in the basal level of dynein's ATPase activity:
To purify A. nidulans dynein for biochemical analyses, we have constructed a strain, S-IC, in which the dynein intermediate chain (NUDI) is tagged with the S-tag at its C terminus (materials and methods). By genetic crosses, we have made strains carrying GFP-nudAR3086C and GFP-nudAL1098F mutations in the S-IC background and these strains are referred to as GFP-nudAR3086C/S-IC and GFP-nudAL1098F/S-IC, respectively. For biochemical analyses, the positive control that we used was GFP-nudA/S-IC, and the negative control contained only GFP-nudA and not the S-IC. After affinity purification using the S-tag-based method (see materials and methods), dynein heavy chain could be clearly detected on a silver-stained protein gel (Figure 5A) and a Western blot (Figure 5D).
We examined the ATPase activities of dynein isolated from the GFP-nudAR3086C/S-IC, GFP-nudAL1098F/S-IC, GFP-nudA/S-IC, and GFP-nudA strains. As expected, the S-peptide-eluted fraction from the GFP-nudA/S-IC (positive control) but not that from the GFP-nudA strain (negative control) contained significant ATPase activity. In every experiment (n = 5), dynein with the nudAR3086C mutation exhibited a lower basal level of ATPase activity compared to the positive control (Figure 5B). The mean value of dynein's ATPase activity in the nudAR3086C mutant was about half of the wild-type value. However, data from the nudAL1098F mutant were not consistent with a decreased dynein's ATPase activity, and the difference between this mutant and wild type is statistically not significant at the P-value of 0.05. Thus, we concluded that the nudAR3086C mutation but not the nudAL1098F mutation negatively affects dynein's ATPase cycle.
Under our purification conditions, the S-protein that interacted with the S-tag in the dynein intermediate chain not only was able to pull down the dynein heavy chain, but also was able to pull down the NUDF protein (Figure 5D). Neither nudAR3086C nor nudAL1098F apparently affected the association between HC and intermediate chains (IC) of dynein and that between NUDF and the dynein complex (Figure 5D). Since IC contains all the light-chain-binding sites (reviewed by Pfister et al. 2006), we suggest that dynein complex formation is not drastically altered by these mutations.
LIS1 and its homologs function in the cytoplasmic dynein pathway in various experimental systems, but how they affect the dynein motors mechanistically in different systems is a question that remains to be further explored. In A. nidulans, deletion of nudF (the lis1 homolog) produces the same growth and nuclear distribution defect as produced by the dynein heavy chain loss-of-function mutants (Xiang et al. 1995b; Willins et al. 1995, 1997). A previous genetic study has suggested that two nudF suppressor mutations are located either within or very close to the dynein heavy chain gene, nudA (Willins et al. 1997). In this study, we have identified these two point mutations in the dynein heavy chain that partially compensate for the loss of NUDF/LIS1. These two mutations, nudAL1098F and nudAR3086C, occur in amino acids that are conserved from fungi to humans. Thus, further characterization of these mutant forms of dynein in different systems may shed light on how LIS1 and its homologs may regulate the dynein motor.
LIS1 binds to two different sites on the dynein heavy chain, and in addition, it also binds to the intermediate chain of dynein, as well as the dynamitin subunit of the dynactin complex (Sasaki et al. 2000; Tai et al. 2002). Since it binds to both dynein and dynactin, it is possible that LIS1 may mediate dynactin function by facilitating the physical interaction between these two complexes (Tai et al. 2002). In this study, we have shown that the two nudF suppressor mutations on the heavy chain do not suppress the phenotypes exhibited by the loss-of-function mutants of dynein intermediate chain and Arp1 of the dynactin complex. Thus, at least some functions of NUDF/LIS1 on dynein regulation are distinct from those of dynein intermediate chain or dynactin. The LIS1-binding sites on the dynein heavy chain were determined by the yeast two-hybrid assay: one is in the stem region between amino acids 649 and 907, and the other is the AAA1 domain (Tai et al. 2002). Whether and how LIS1 affects dynein's ATPase cycle in vivo is still not clear, but an enhancement on dynein's microtubule-stimulated ATPase activity by purified LIS1 has been detected in vitro (Mesngon et al. 2006). This enhancement on dynein's motor activity is consistent with LIS1's positive role in dynein function. However, given our current result that the nudAR3086C mutation decreases dynein's ATPase activity but partially suppresses the phenotype caused by NUDF loss, it is unlikely that the mechanism of NUDF/LIS1 action on dynein is related only to an enhancement of dynein's ATPase activity.
The nudAR3086C mutation resides at the end of the AAA4 that is close to the microtubule-binding stalk, but not in the walker A or walker B motif that is implicated in ATP binding or hydrolysis. However, the nudAR3086C mutation clearly decreases dynein's ATPase activity, suggesting that the conformational change caused by this mutation in turn may cause changes in other domains that directly influence dynein's ATPase cycle. It is not known whether the decrease in the ATPase activity per se is causally related to the suppression. LIS1's binding site on the heavy chain stem region is close to the sites for heavy chain–intermediate chain and heavy chain–light intermediate chain interactions implicated in cargo-motor binding (Tai et al. 2002), and thus LIS1 may help to coordinate cargo binding with motor activation (Tai et al. 2002). One may speculate that a decreased ATPase activity may benefit such coordination in the absence of NUDF. Alternatively, a decrease in dynein's ATPase activity is just a secondary phenotype associated with this suppressor mutation. Further genetic analyses of more nudF suppressors are needed to address these issues.
Unlike the nudAR3086C mutation that acts as a stronger suppressor of the nudF deletion phenotype, the nudAL1098F mutation, which acts as a weaker suppressor, does not seem to cause any obvious decrease in the ATPase activity of dynein. Although its suppression of nudF mutants' nuclear distribution defect is hardly detectable at 42°, it clearly enhances colony growth of the nudF mutants to some extent and weakly suppresses the nuclear distribution phenotype of ΔnudF at 32°. Thus, this mutation may weakly compensate for the absence of NUDF. Given the physical closeness between the nudAL1098F mutation and the deduced LIS1-binding site in the stem, one can speculate that NUDF/LIS1 binding causes a subtle conformational change of the stem region and that the L1098F mutation may partially mimic such a change. On the basis of deletion analyses and the analysis of trypsin digestion sites, the LIS1-binding site does not seem to be part of the linker domain that connects directly to the motor head domain (Gee et al. 1997; Tai et al. 2002; Hook et al. 2005; Vallee and Hook 2006). Similarly, the A. nidulans nudAL1098F mutation, which corresponds to the L1060 residue in human dynein heavy chain, should also reside in the first part of the stem region rather than in the linker domain. A recent electron microscopic study has suggested that the stem-to-ring connection is flexible, and the authors speculated that potential regulators such as LIS1 may stiffen the connection to help dynein and its cargo to move forward after a power stroke (Meng et al. 2006). This idea would need further tests. In S. cerevisiae, an artificially constructed dynein heavy chain dimer in which both heavy chains miss the N-terminal regions, including the deduced LIS1-binding sites in the stems as well as the sites for binding other dynein subunits, seems to show an even more robust physical association with Pac1/LIS1 (supplemental data in Reck-Peterson et al. 2006), suggesting that Pac1/LIS1's binding to other site(s) may be negatively regulated by part of the stem region. Moreover, this artificial dynein dimer isolated from either the wild-type background or the Pac1 deletion background moves on microtubules processively in single-molecule assays, indicating that Pac1/LIS1 is not required for dynein's processive movement (Reck-Peterson et al. 2006). It remains to be tested, however, whether LIS1 and their homologs have any effect on the motility of native dynein from various organisms. It is worthwhile to point out the recent evidence supporting the idea that at least part of the stem region close to AAA1 is important for power stroke (Reck-Peterson et al. 2006; Shima et al. 2006).
In contrast to the nudAL1098F mutation that does not seem to cause any apparent change in dynein's microtubule-plus-end accumulation, the nudAR3086C mutation causes an obvious change in dynein's localization pattern. In wild-type background, GFP-dynein forms comet-like structures representing its localization at the microtubule plus end (Xiang et al. 2000; Han et al. 2001). In some hyphae, small speckles along microtubules can also be found, but it is hard to track them because of the low intensity and the transient nature of these signals. Interestingly, GFP-dynein with the nudAR3086C mutation forms many more speckles along the microtubule (Figure 4). Most speckles seem relatively nonmotile, although a time lapse with longer intervals did reveal movements of some speckles (supplemental movie 5 at http://www.genetics.org/supplemental/). Given that the ATPase activity of the nudAR3086C mutant is lower than that in the wild-type control, one possibility is that dynein molecules in this mutant may have difficulty moving toward the minus end of the microtubule. In the ΔnudF background, the nudAR3086C mutation also causes dynein speckles to locate along the microtubule, but in most cells, the decoration seems dimmer compared to that in the nudAR3086C single mutant. In many cells, the plus-end comet intensity is not apparently lower than that in the ΔnudF mutant. Thus, although we speculated that the dynein heavy chain fails to interact with the microtubule as a minus-end-directed motor in the absence of NUDF, and that the nudAR3086C mutation provokes dynein to leave the plus end by mimicking a conformational change caused by NUDF binding, we have not yet obtained evidence to support this idea. Future studies will be needed to address the mechanism(s) of NUDF/LIS1 action as well as the mechanism(s) of suppression by dynein heavy chain mutations.
In summary, we have identified two nudF suppression mutations in the A. nidulans dynein heavy chain, nudAL1098F and nudAR3086C. Because they are located in different domains and affect dynein differently, we suspect that they may bypass NUDF/LIS1 by different mechanisms. Because these changed residues are conserved from fungi to human, further analyses of dynein with these mutations in several different experimental systems should help to understand how dynein is regulated in the cell. Regarding the application of these data to studies of diploid systems, one problem is that we do not yet know whether these mutations are recessive or dominant in terms of nudF suppression since our numerous attempts to make the desired diploids have failed (this study; Willins et al. 1997). On the other hand, we do know that the nudAR3086C mutation that lowers dynein's ATPase activity by itself is recessive to the wild-type allele, which makes it harder to study its effect in a diploid system. However, the success in analyzing the ATPase cycle of recombinant dynein in vitro makes it possible to study these recessive mutations in higher systems (Hook et al. 2005; Vallee and Hook 2006), and such studies are likely to generate new insights into the structural–functional relationship of the dynein heavy chain domains.
We thank Berl Oakley for sending us nKuA deletion strains prior to publication, which made this work possible. We thank Erika Holzbaur, Steve Osmani, and Teresa Dunn for discussions and suggestions and Shihe Li, Liqin Wang, and Young Lee for technical help. This work was supported by a National Institutes of the Health grant (GM069527-01) and a Uniformed Services University of the Health Sciences intramural grant (R071GO).
This article is dedicated to the memory of Bo Li.
↵1 These authors contributed equally to this work.
Communicating editor: S. Dutcher
- Received November 29, 2006.
- Accepted January 4, 2007.
- Copyright © 2007 by the Genetics Society of America