Identification of a Novel Site in the Tail of Dynein Heavy Chain Important for Dynein Function in Vivo*

Background: The dynein heavy chain tail is required for subunit interactions but not for in vitro motility. Results: A dynein tail mutation affects dynein function in vivo without affecting subunit interactions. Conclusion: A site upstream of the subunit interaction sites of the dynein tail is important in vivo. Significance: This study discovers a novel site of the dynein tail critical for motor function in vivo. The minus end-directed microtubule motor cytoplasmic dynein is responsible for the intracellular movements of many organelles, including nuclei and endosomes. The dynein heavy chain contains a C-terminal motor domain and an N-terminal tail domain. The tail binds other dynein subunits and the cargo-interacting dynactin complex but is dispensable for movement of single dynein molecules in vitro. Here, we identified a mutation in the Aspergillus nidulans heavy chain tail domain, nudAF208V, which causes obvious defects in dynein-mediated nuclear positioning and early endosome movement. Astonishingly, the nudAF208I mutation in the same position does not cause the same defects, suggesting that a subtle difference in the size of the amino acid side chain at this position has a significant consequence. Importantly, our biochemical analyses indicate that the nudAF208V mutation does not affect dynein subunit interactions and the mutant dynein is also able to bind dynactin and another dynein regulator, NUDF/LIS1. The mutant dynein is able to physically interact with the early endosome cargo, but dynein-mediated early endosome movement away from the hyphal tip occurs at a significantly reduced frequency. Within the small group of early endosomes that move away from the hyphal tip in the mutant, the average speed of movement is lower than that in the wild type. Given the dispensability of the dynein tail in dynein motility in vitro, our results support the notion that the structural integrity of the dynein tail is critical in vivo for the coordination of dynein force production and movement when the motor is heavily loaded.

neurodegenerative and neurodevelopmental disorders (8,9). The dynein heavy chains (HCs) 2 within the dynein complex form a dimer that contains motor activity responsible for ATPdependent movement of dynein along microtubules. Each HC contains a C-terminal motor domain that forms a ring containing six AAA (ATPase of various cellular activities) domains and a microtubule-binding stalk located in between AAA4 and AAA5 (10 -13). The N terminus of the HC is called the "tail" (or the "stem"), which is implicated in dynein HC dimerization and binding to other subunits, including intermediate chains (ICs), light intermediate chains (LICs), and light chains (14). The dynein IC contains binding sites for the light chains, dynactin, and another dynein regulator NudE that recruits LIS1 (14 -26). Although the linker region immediately before the AAA1 domain is involved in power stroke, the HC tail N-terminal to the subunit-interaction site is not important for movements of single dynein molecules in vitro (27)(28)(29)(30)(31). It has been proposed, however, that the tail plays a role in regulating dynein motor function (32). This hypothesis is supported by the results that the neurodegenerative mutation Loa (legs at odd angles, F580Y in mouse HC) in the HC tail affects dynein motor processivity (32) or velocity (33). However, because defects in dynein subunit interactions and/or interaction with dynactin have been detected in the Loa mutant (32,34), the contribution of other factors to the Loa phenotype is hard to be completely excluded.
The filamentous fungus Aspergillus nidulans is an excellent genetic system for studying intracellular trafficking and the regulation of cytoplasmic dynein (35)(36)(37)(38). In A. nidulans, components of the dynein-dynactin complexes and other regulators, including LIS1 and NudE/Nudel, are required for nud (nuclear distribution), and loss of function of these proteins produces a nud phenotype (19,36,39). In addition to its function in nuclear distribution, dynein in filamentous fungi is also critical for the retrograde movement of early endosomes in hyphae as first discovered in Ustilago maydis (35)(36)(37)(38)40). In A. nidulans, as well as in Saccharomyces cerevisiae, U. maydis, and mammalian cells, dynein forms comet-like structures, representing their accumulation at the microtubule plus ends (40 -44). In A. nidulans and U. maydis, this accumulation at the microtu-bule plus ends near the hyphal tip facilitates dynein-early endosome interaction, and thus is important for dynein-driven transport of early endosomes from the hyphal tip (37,40,(45)(46)(47). The plus-end accumulation of dynein depends on the integrity of the dynein-dynactin complexes and kinesin-1 (40, 48 -53). Loss of IC abolishes plus-end comet formation (48), presumably because dynein could no longer bind to p150 dynactin (15)(16)(17), a key factor in the plus-end localization of A. nidulans dynein (52,53). Loss of LIC also abolishes plus-end comet formation because HC-IC interaction is significantly weakened in the absence of LIC (50).
In this current study, we have identified a HC tail mutation that does not affect the accumulation of dynein at the microtubule plus ends but causes obvious defects in dynein-mediated early endosome transport and nuclear migration. This mutation, nudA F208V , is far upstream from the previously mapped IC-and LIC-binding site, and our experimental data demonstrate that it does not affect dynein complex formation or dynein-early endosome interaction. Interestingly, changing the valine residue to isoleucine completely rescues the mutant phenotype, suggesting that a subtle change in the size of the side chain of this residue is sufficient for causing a significant effect on HC function in vivo. Thus, beyond the known role of the HC tail in interacting with other dynein subunits implicated in cargo binding, a novel site upstream of the subunit interaction site plays a significant role in dynein motor function in vivo.

EXPERIMENTAL PROCEDURES
A. nidulans Strains, Media, and Mutagenesis-A. nidulans strains used in this study are listed in Table 1. For biochemical experiments, YG (yeast extract plus glucose) ϩ uridine ϩ uracil (or YUU) liquid medium was used. UV mutagenesis on spores of A. nidulans strains was done as described previously (54,55). For DAPI staining of nuclei, cells were incubated in YUU medium for 8 h at 37°C. For live cells imaging experiments, liquid minimal medium plus supplements was used, and cells were cultured at 32°C for overnight and observed at room temperature.
For observing early endosome movement in vivo, we used a minimal medium containing either 1% glycerol (v/v) or 0.9% (w/v) fructose and 6.25 mM threonine (56). For observing GFP-HC, we also used these two types of minimal media, and in addition, we used a minimal medium containing either 0.1% (w/v) glucose or 1% glucose.
Identification of the nudA F208V Mutation and Creation of the nudA F208I Mutation-The nudA sequence of the 4-63 mutant corresponding to the wild type nudA fragment that complemented the 4-63 mutant was sequenced using primers as described previously (57), which led to the identification of the nudA F208V mutation. To construct the mutant strain carrying the nudA F208I mutation, we first made the DNA fragment containing this mutation using fusion PCR. Specifically, two fragments were made using the primer pairs of NudA51 (AAAACTCTATCTGCCCGCGAA) and FIR (GAGTTC-CAGCTCCGCGATTTTCTTCTTAGTGCC), and of NudA33 (CAATGGAAATCTGATAACGCG) and FIF (GGCACTAA-GAAGAAAATCGCGGAGCTGGAACTC) respectively, followed by a fusion PCR using the primer pair of NudA51 and NudA33. This fragment was transformed into the RQ8 strain containing the nudA F208V mutation, and wild type-looking transformants were subjected to sequencing analysis.
Identification of Suppressors of the nudA F208V Mutation-We performed a UV mutagenesis on the RQ8 strain containing the nudA F208V mutation. UV mutagenesis on spores of the RQ8 strain was done as described previously (54,55). Colonies that looked healthier than that of the nudA F208V mutant were judged to be suppressors and selected from the plates. Two suppressors were selected and crossed with a wild type strain. We did not see any nudA F208V -like colonies among the progeny, suggesting that the suppressor mutations are linked to the original nudA F208V mutation. These suppressors were further analyzed by sequencing of the nudA gene. Primers for sequencing were as described previously (57).
Live Cell Image Analyses-Fluorescence microscopy of live A. nidulans cells was as described (47,52). Cells were grown at 32°C overnight using the Lab-Tek chambered cover-glass system. Images were captured at room temperature using an Olympus IX70 inverted fluorescence microscope (with a 100ϫ objective) linked to a PCO/Cooke Corp. Sensicam QE cooled CCD camera. A filter wheel system with GFP/mCherry-ET Sputtered series with high transmission (BioVision Technologies) was used. The IPLab software was used for image acquisition and analysis.
Analyses of Protein-protein Interactions, Sucrose Gradient Centrifugation, and Western Blotting-Dynein HC-IC interactions were examined using strains containing a functional S-tagged dynein IC as done previously (57). Alternatively, the MACS GFP isolation kit (Miltenyi Biotec) was used to determine whether GFP-tagged dynein HC pulls down IC. About 0.7 g of hyphal mass was harvested from overnight culture for each sample, and cell extract was prepared using a lysis buffer containing 25 mM Tris-HCl, pH 8.0, 0.4% Triton X-100, 0.3% (v/v) of a protease inhibitor mixture (Sigma-Aldrich), and 0.1% (v/v) of a phosphatase inhibitor mixture (Sigma-Aldrich). Cell extract was centrifuged at 16,000 ϫ g for 30 min at 4°C, and supernatant was used for the pulldown experiment. To pull down GFP-tagged protein, 50 l of anti-GFP MicroBeads were added into the cell extract and incubated on ice for 30 min. The MicroBeads/cell extract mixture was then applied to the column provided with the GFP isolation kit, followed by gentle wash with the lysis buffer. After washing four times with the lysis buffer, the column was finally washed one more time with the lysis buffer without any detergent. The preheated (95°C) SDS-PAGE sample buffer was used as an elution buffer. For analyzing dynein complex sedimentation using sucrose gradient centrifugation, the cell extract was centrifuged at 100,000 ϫ g, and the supernatant was loaded on the top of a 5-25% sucrose step gradient and centrifuged at 4°C overnight at 100,000 ϫ g using the SW41 rotor. Fractions were collected from the top and analyzed on protein gels. Thymoglobulin (Sigma-Aldrich) was used as a 19 S marker. Antibodies against dynein HC, dynein IC, p150 dynactin, and NUDF/LIS1 were described previously (39,51,58). Anti-GFP antibodies from Covance (monoclonal) and Clontech (polyclonal) were also used for Western analyses. Western analyses were performed using the alkaline phosphatase system, and blots were developed using the AP color development reagents from Bio-Rad.
Analyses of Dynein-early Endosome Interaction-For the biochemical analyses of dynein-early-endosome interaction, we used the strains containing GFP-labeled dynein HCs and mCherry-RabA-labeled early endosomes. The strains were grown overnight in the liquid minimal medium containing 0.9% (w/v) fructose and 50 mM threonine, which induces the expression of the mCherry-RabA fusion under the control of the alcA promoter. About 0.7 g of hyphal mass was harvested from overnight culture for each sample, and cell extract was prepared using a lysis buffer containing 25 mM Tris-HCl, pH 8.0, 0.01% Triton X-100, 0.3% (v/v) of a protease inhibitor mixture (Sigma-Aldrich), and 0.1% (v/v) of a phosphatase inhibitor mixture (Sigma-Aldrich). Cell extract was centrifuged at 4,000 ϫ g for 30 min at 4°C, and supernatant was used for the pulldown assay using the MACS GFP isolation kit (Miltenyi Biotec). As described earlier, 50 l of anti-GFP MicroBeads were added into the cell extract and incubated on ice for 30 min. The MicroBeads/cell extract mixture was then applied to the column provided with the GFP isolation kit, followed by gentle wash with the lysis buffer five times. Preheated (95°C) SDS-PAGE sample buffer was used as an elution buffer. The anti-mCherry antibody used on the Western blots to detect mCherry-RabA was purchased from BioVision Research Products. Western analyses were performed using the alkaline phosphatase system, and blots were developed using the AP color development reagents from Bio-Rad. Quantification of the protein band intensity was done using the IPLab software as described previously (52), with a minor modification. Specifi-cally, an area containing the whole band was selected as a region of interest, and the intensity sum within the region of interest was measured. Then, the region of interest box was dragged to the equivalent region of the negative control lane to take the background value, which was then subtracted from the intensity sum. The intensity ratio of the pulled down mCherry-RabA to GFP-HC was calculated. The ratios calculated from the wild type samples were set as 1, and the relative ratios of the mutant were calculated and presented.

RESULTS
The nudA F208V but Not nudA F208I Mutation Causes a Defect in Dynein Function-To study the regulation of dynein-mediated retrograde transport of early endosomes, we used UV mutagenesis to obtain mutants that are defective in early endosome transport. The strain we used for mutagenesis contains the GFP-nudA allele that replaces the endogenous nudA and reports HC localization (57) and the mCherry-RabA allele that reports the distribution of early endosomes (45,47). One mutant we selected was called 4-63, which formed a colony smaller than the wild type (Fig. 1A). In this mutant, GFP-HC formed normal comet-like structures near the hyphal tip (Fig.  1B, arrows), which represent dynein accumulation at the microtubule plus ends (41). However, a prominent buildup of mCherry-RabA-labeled early endosomes was observed in almost every hyphal tip, indicating a defect in dynein-mediated retrograde transport of early endosomes (Fig. 1B, arrows). The mutant also exhibited an obvious defect in nuclear distribution (Fig. 1C). Specifically, a cluster of multiple nuclei in the spore head was often observed (Fig. 1C, arrows). This phenotype differs from that of the null mutant of p25, a dynactin subunit required for dynein-early endosome interactions but not for nuclear positioning (59). Based on this phenotype, we concluded that the corresponding protein, unlike p25, is likely to be required for dynein activity rather than being specifically required for the physical interaction of dynein with early endosomes.
During an analysis of more than 50 progenies from a cross between the GFP-HC-containing mutant and a wild type strain without GFP-HC, we noticed that the mutant phenotypes of colony growth and early endosome accumulation were always linked with each other and also with the presence of plus-end comet signals formed by GFP-HC. This suggests that the colony phenotype and the defect in early endosome transport are caused by the same mutation located within or very close to the nudA gene encoding dynein HC.
To identify the mutation, we first prepared five different wild type genomic DNA fragments covering the entire 14-kb nudA gene by using high-fidelity PCR. Each nudA fragment was then co-transformed into the mutant with the auto-replicating pAid plasmid containing the selective marker gene pyrG (55). The 3-kb fragment containing sequences encoding the N-terminal part of HC was able to rescue the mutant phenotype via homologously replacing the equivalent sequence in the mutant genome. The corresponding region of the mutant was fully sequenced, and one single mutation, T622G (in the coding sequence), was found when compared with the wild type nudA sequence ( Fig. 2A). Sequencing of five more individual mutant progenies from a cross further confirmed the identity of this mutation. This mutation results in a change in the amino acid residue 208 from phenylalanine (Phe) to valine (Val).
The nudA F208V mutation is located in the tail of the dynein HC (Fig. 2B). An alignment of dynein HC sequences from various organisms showed that although this region is conserved to certain extent, this phenylalanine residue is changed to leucine (Leu) or isoleucine (Ile) in some species (Fig. 2C). Because the side chains of these amino acids are longer than that of valine, we hypothesized that a subtle difference in the size or hydrophobicity of the side chain may be responsible for the observed phenotype produced by the nudA F208V mutation. To test this, we mutated the F208 residue to isoleucine that also contains a branched side chain like valine but contains one extra ϪCH 3 group. We made a DNA fragment containing the codon change of nudA F208I and introduced it to the nudA F208V mutant. This fragment was able to rescue the mutant via homologous recombination and produce a colony essentially identical to wild type (Fig. 2D). The nudA F208I mutant also exhibited normal nuclear distribution and early endosome distribution (Fig. 2, E and F). The striking phenotypic difference between the nudA F208V mutant and the nudA F208I mutant indicates that a subtle change in the side chain of Phe-208 makes a significant difference in the function of dynein HC.
The nudA F208V Mutation in the Dynein Tail Does Not Affect Dynein Complex Assembly, but Affects the Cargo-transporting Capacity of Dynein-The nudA F208V mutation locates at a site far upstream of the linker region before AAA1 implicated in motor activity, and it is also more upstream of the subunit interaction sites within the HC tail. Previous in vitro studies mapped the sites of HC dimerization, HC-IC interaction, and HC-LIC interaction to the tail region of the HC. These studies were done using in vitro-expressed HC fragments of cytoplasmic dynein HCs from rat and Dictyostelium discoideum (60,61). The HC dimerization site on the rat HC was mapped to a big region containing amino acids 300 -1140 (60), which corresponds to a region containing amino acids 325-1180 of A. nidulans HC. However, the study using Dictyostelium HC fragments mapped the dimerization site to a region containing amino acids 627-780 (61), corresponding to amino acids 603-759 of A. nidulans HC. The IC binding site within the D. discoideum HC is in a region containing amino acids 629 -730 (61), corresponding to amino acids 605-690 of the A. nidulans HC. However, the IC binding site of the rat HC contains amino acids 446 -701 (60), corresponding to amino acids 470 -720 of A. nidulans HC. Within the rat HC, the LIC binding site contains amino acids 649 -800 (60), which correspond to amino acids 668 -816 of A. nidulans HC. All of these sites are downstream of the nudA F208V mutation that we identified.  (74). C, a sequence alignment showing that the Phe-208 residue is conserved in some species but changed to other amino acids such as isoleucine and leucine in the dynein HCs of other organisms. D, the nudA F208I mutant forms a wild type colony. E, the nudA F208I mutant exhibits normal nuclear distribution with no cluster of multiple nuclei in the spore head. The spore heads are indicated by arrows. F, the nudA F208I mutant exhibits normal early endosome distribution with no buildup of early endosomes at the hyphal tip. The hyphal tips are indicated by arrows. Bar, 5 m.
In addition to the positions of the deduced subunit binding sites in A. nidulans HC, several lines of experimental evidence also helped us to exclude the possibility that the nudA F208V mutation affects the HC-IC and/or HC-LIC interactions. First, although loss of IC or LIC in A. nidulans abolishes microtubule plus-end comets formed by the dynein HC (48, 50), GFP-NUDA F208V formed comet-like structures just like wild type HCs (Fig. 1B). Second, HC was pulled down with S-tagged IC from extract isolated from the nudA F208V mutant (Fig. 3A). Similarly, the amount of IC associated with GFP-NUDA F208V was normal as judged by pulldown experiments using a GFPantibody coupled to magnetic beads (Fig. 3B), and the same conclusion was obtained even when the pulldown experiments were performed using high-salt buffers (Fig. 3C). Because loss of dynein LIC is known to abolish HC-IC interaction (50), we concluded that neither HC-IC nor HC-LIC interaction is weakened by the nudA F208V mutation. This conclusion was further confirmed by a sucrose gradient sedimentation experiment showing normal sedimentation of the mutant dynein complex (Fig. 3D). Thus, we have identified a novel mutation in the dynein tail that clearly affects dynein function but not dynein complex assembly. In addition, our biochemical data also indicate that the nudA F208V mutation does not weaken the dyneindynactin interaction (Fig. 3C), which is expected because IC is the subunit that binds to dynactin (15)(16)(17) and because dynein forms normal plus-end comets (Fig. 1B), which is dynactin-dependent (49,(51)(52)(53). Moreover, the nudA F208V mutation does not affect the interaction between dynein and the LIS1 homolog NUDF (Fig. 3C).
To explore the mechanism of the nudA F208V mutation, we examined dynein-mediated movements of early endosomes in more detail in both the mutant and wild type control strains. We first used a minimal growth medium containing glycerol as a carbon source, which allows the expression of the early endosome marker mCherry-RabA under the control of the alcA promoter. Under this condition, early endosomes were clearly blocked at the hyphal tip in the mutant as shown in Fig. 1B. The frequency of early endosome movement away from the hyphal tip in the mutant was about seven times lower than that in wild type. Because the number of early endosomes that move away from the hyphal tip was very low, it was not easy to do a quantitative analysis on the speed of movement in the mutant. We next used another minimal medium containing fructose and threonine, which allows a robust hyphal growth and a relatively low level of expression of the alcA promoter. In this medium, the signal intensity of mCherry-RabA in the wild type control was quite low compared with that in the glycerol-containing medium, most likely reflecting the relatively low expression level of the alcA-driven mCherry-RabA fusion protein. Under this condition, dynein-mediated movements of early endosomes were still blocked in the mutant, as evidenced by the abnormal accumulation of mCherry-RabA signals at the hyphal tip (Fig. 4A, supplemental Movies 1 and 2). However, more retrograde movements were observed, allowing for a quantitative analysis of the movement speed. Measurements of kymographs (Fig. 4B) showed that although some early endosomes in the mutant moved as quickly as those in the wild type control, the frequency of the slower early endosomes was increased. Of 53 early endosomes we measured from the mutant, nine of them (ϳ17%) moved at a speed below 1 m/s, which was never observed in wild type (n ϭ 55) (Fig. 4C). Consequently, the average speed of dynein-mediated movements in the mutant was mildly reduced (Fig. 4D). The frequency of movements from the hyphal tip was about three times lower in the mutant than in the wild type (Fig. 4E). Sometimes, early endosomes in the mutant underwent short distance movements away from the tip but moved right back to the hyphal tip (supplemental Movie 3). Under these conditions, we also observed a cloud of GFP-HC signals at the hyphal tip region where early endosomes also accumulated (supplemental . This cloud was also observed in the mutant grown on glucose that shuts off the expression of mCherry-RabA, but not observed in the wild type strain grown under the same conditions (supplemental Movies 7 and 8). Because early endosomes also accumulate at the hyphal tip region, this localization pattern of dynein suggests that the mutant dynein is likely to interact with early endosomes. Such an idea is consistent with the result that the mutant HC is able to bind to IC, which binds dynactin (15-17), a com- The lane labeled as "negative control" contains the sample from a wild type strain whose IC is not tagged with the S-tag. B, the amount of IC pulled down by the GFP-antibody from the extract containing GFP-nudA F208V is normal compared with that from the extract containing GFP-nudA. The lane labeled as negative control contains the sample from a wild type strain whose HC is not tagged with GFP. The experiment was done using a protein extraction buffer containing detergent but no NaCl. C, the amount of dynein IC, p150 of dynactin or NUDF/LIS1 pulled down by the GFP-antibody from the extract containing GFP-nudA F208V is similar to that from the extract containing GFP-nudA. The experiment was done using a protein extraction buffer containing 0 mM, 50 mM, or 100 mM NaCl. D, the mutant HC is incorporated into the dynein complex as indicated by its normal sedimentation at ϳ19 S in a 5-25% sucrose gradient. The experiment was done using a protein extraction buffer containing 50 mM NaCl.
plex required for dynein-early endosome interaction (59,62). To confirm this idea, we used a biochemical pulldown assay to directly examine the physical interaction between the dynein HC labeled with GFP and the early endosome cargo labeled with mCherry-RabA. The GFP antibody-conjugated magnetic beads pulled down dynein-bound early endosomes from cell extract, which were detected on Western blots probed by an anti-mCherry antibody (Fig. 5A). A strain containing mCherry-RabA but not GFP-HC was used as a negative control for this assay, and the amount of mCherry-RabA pulled down from the extract of this negative control strain was negligible (Fig. 5A). The result of this assay showed that the mutant dynein is clearly able to bind early endosomes. In fact, the amount of mCherry-RabA pulled down from the mutant extract was even higher than that from the wild type extract (Fig. 5, A and B, p Ͻ 0.05, n ϭ 3), possibly because recycling of the mutant dynein from non-motile early endosomes is relatively inefficient. Together, our results suggest that rather than negatively affecting dyneincargo interaction, the nudA F208V mutation affects the ability of cargo-bound dynein in initiating or sustaining movements along microtubules in vivo.
Two Suppressor Mutations of the nudA F208V Mutant, nudA T1133S , and nudA N356K , Are also Located in the HC Tail-To gain further insight into the structural role of the nudA F208 residue, we performed a UV mutagenesis on the nudA F208V mutant. The spores were mutagenized as described previously and plated out on YUU plates. From ϳ8,000 colonies, two strains, RQ47 and RQ49, were selected. These two strains looked healthier than the original mutant on plates (Fig. 6A), but less healthy than the wild type, indicating only a partial recovery of dynein function. As expected, the defects in dyneinmediated early endosome transport and nuclear distribution were only partially but not fully suppressed (Fig. 6, B-E). To test whether the suppressor mutations were located in the nudA gene, we crossed the selected suppressor with a wild type strain. We reasoned that if the suppressor mutation is linked with nudA F208V , none of the progeny should look like the original mutant. This turned out to be the case for both suppressors. To identify the suppressor mutation in RQ47, we sequenced the entire nudA gene from the selected suppressor mutant. In addition to the original nudA F208V mutation as expected, we found only one additional mutation, nudA T1133S , which is also located in the dynein tail but downstream of the previously mapped . The average frequency of movements in the mutant within a time period of 16 s is relative to that of the wild type, which is set as 1. The frequency is significantly lower in the mutant than in wild type (p Ͻ 0.001).

FIGURE 5. A biochemical analysis showing that dynein in the nudA F208V mutant is able to interact with mCherry-RabA-labeled early endosomes.
A, Western blots showing a result of the pulldown assay for dynein-early endosome interaction using an anti-GFP antibody conjugated to magnetic beads. The amount of mCherry-RabA pulled down with the GFP-HC from the nudA F208V mutant extract is clearly not lower than that from the wild type extract. The lane labeled as negative control contains the pulldown sample from a strain containing mCherry-RabA but not GFP-HC. B, a quantitative analysis of the results of the pull-down assays for dynein-early endosome interaction. The intensity ratio of the mCherry-RabA band to that of the GFP-HC band is shown. The value of the mutant is relative to the wild type values, which is set at 1. Mean and S.D. values were calculated from three independent pulldown experiments.
subunit interaction sites. This Thr residue is not highly conserved (Fig. 6F), and how this subtle mutation (Thr to Ser) influences tail structure will need to be studied in the future. Similarly, the suppressor mutation in the RQ49 strain was identified by sequencing. In addition to the original nudA F208V mutation, one mutation, nudA N356K , was found. This Asn residue is conserved in most species examined, but it is changed to Lys in D. discoideum HC and to Asp in Schizosaccharomyces pombe HC (Fig. 6G). To verify that the improved colony phenotype of the original nudA F208V mutant was indeed caused by these mutations, we transformed the fragments containing either one of the suppressor mutations back to the original nudA F208V mutant. Both fragments generated transformants that looked like the originally selected suppressors, and sequencing analyses of two of these suppressor-like transformants confirmed the presence of the two additional mutations. Thus, the nudA T1133S or the nudA N356K mutation partially compensates for the structural alteration caused by the nudA F208V mutation.

DISCUSSION
In this study, we have identified a novel mutation, nudA F208V in the A. nidulans dynein HC tail that significantly weakens dynein function in nuclear migration and early endosome movement. Although the dynein tail has been implicated in dynein subunit interactions, this mutation is located upstream of the subunit interaction sites and does not apparently affect dynein complex assembly. In the nudA F208V mutant, early endosomes abnormally accumulate at the hyphal tip, which is indicative of a defect in dynein-mediated early endosome movement. The physical interaction between the dynein motor and the early endosome cargo is clearly not weakened. However, the frequency of dynein-mediated early endosome movement away from the hyphal tip is significantly reduced. Moreover, although some early endosomes that leave the hyphal tip move at a normal speed, some move at an abnormally low speed. Together, these results suggest that the nudA F208V mutation causes a defect of dynein in initiating and/or sustaining cargo movement along the microtubule track in vivo.
Although the HC tail is not needed for motility of single dynein molecules in vitro as long as they are dimerized (31), the function of the tail in regulating the dynein motor has been suggested by a recent study on the Loa mutation in the HC tail (32). Dynein from the mouse Loa mutant shows a weakened dynein-microtubule interaction and a decreased motor processivity along microtubules (32). Unlike the Loa mutation that affects HC-IC interaction (32), the nudA F208V mutation in A. nidulans does not affect HC-IC interaction and dynein complex assembly. Thus, our current result provides a strong support for the idea that besides its binding to other subunits, the tail is important for regulating dynein motor function in vivo. It is worth pointing out that in vivo, the dynein motor carrying a cargo not only needs to move in a cytoplasm with considerable viscosity (63,64) but also needs to compete with plus-end-directed motors bound to the same cargo (65)(66)(67). In filamentous fungi, kinesin-3 is able to bind to the dynein-bound early endosome and be carried to the minus end as a cargo of dynein (40,53,68). In U. maydis, binding to a single molecule of dynein is sufficient to cause a plus-end-directed early endosome to reverse its direction (68), suggesting that dynein turns off or overpowers kinesin-3. In A. nidulans, however, dynein on plusend-directed early endosomes was observed (53), suggesting that kinesin-3 is able to compete with dynein on the same cargo. In this scenario, dynein is likely to be detached from the microtubule or take backward steps as suggested by an in vitro study examining yeast dynein under a load that is bigger than its stall force (69). The abnormal accumulation at the hyphal tip in the nudA F208V mutant suggests a weakened ability of the mutant dynein to compete against kinesin-3. This may have resulted from a decreased microtubule affinity during the ATPase cycle causing the dynein motor to detach from the track or a decreased stall force causing the motor to reverse direction more frequently. Future in vitro work applying the optic trap technique will need to be done to address these possibilities.
Although several reports have described the structure of the HC motor domain (70 -76), the structure of the tail domain has not yet been reported. Thus, how the nudA F208V mutation affects tail structure remains unknown. However, identification of the nudA F208V mutation and its suppressor mutations in the dynein tail indicates that the fine structure of the tail is important for motor function in vivo. It is interesting to note that the nudA F208I mutation does not cause any obvious defect, suggesting that the proper size or hydrophobicity of the side chain in the Phe-208 residue is critical for function. Recently, a genetic study in another filamentous fungus Neurospora crassa has identified many mutations in the dynein HC that affect dynein function (77). Although most mutations are in the motor domain, two mutations, Y110S and W1308G, of the N. crassa HC tail have been identified (77). Interestingly, similar to the nudA F208V mutation, the Y110S mutation is also located far upstream of the subunit interaction site. Further insights into structural roles of these involved amino acids will be generated after the structure of the tail becomes available. Exactly how a tail mutation would affect dynein motor function is unclear at this point. It has been postulated that the budding yeast dynein tail may interact with the motor domain in a way that the tail would be "masked" in the presence of the motor domain (78,79). Interestingly, although the dynein regulator LIS1 binds to the motor domain at AAA3/AAA4 (26), removal of the tail appears to enhance dynein-LIS1 interaction (26,31,78). Given that the dynein HC is involved in multiple protein interactions (14 -18, 21-26), it remains to be tested whether the HC tail affects dynein function directly through regulating the motor domain or indirectly through coordinating the functions of other dynein regulators.