The nudF gene of the filamentous fungus Aspergillus nidulans acts in the cytoplasmic dynein/dynactin pathway and is required for distribution of nuclei. NUDF protein, the product of the nudF gene, displays 42% sequence identity with the human protein LIS1 required for neuronal migration. Haploinsufficiency of the LIS1 gene causes a malformation of the human brain known as lissencephaly. We screened for multicopy suppressors of a mutation in the nudF gene. The product of the nudE gene isolated in the screen, NUDE, is a homologue of the nuclear distribution protein RO11 of Neurospora crassa. The highly conserved NH2-terminal coiled-coil domain of the NUDE protein suffices for protein function when overexpressed. A similar coiled-coil domain is present in several putative human proteins and in the mitotic phosphoprotein 43 (MP43) of X. laevis. NUDF protein interacts with the Aspergillus NUDE coiled-coil in a yeast two-hybrid system, while human LIS1 interacts with the human homologue of the NUDE/RO11 coiled-coil and also the Xenopus MP43 coiled-coil. In addition, NUDF coprecipitates with an epitope-tagged NUDE. The fact that NUDF and LIS1 interact with the same protein domain strengthens the notion that these two proteins are functionally related.
During growth of the filamentous fungi Aspergillus nidulans and Neurospora crassa, nuclei migrate into the germtube and distribute evenly along the cell length (reviewed in Fischer 1999; Xiang and Morris 1999). Analysis of mutations that affect nuclear distribution in these two fungi and in S. cerevisiae has identified microtubules and cytoplasmic dynein and dynactin complexes as the main components of the nuclear distribution machinery (Oakley and Morris 1980; Plamann et al. 1994; Xiang et al. 1994; Tinsley et al. 1996; Geiser et al. 1997). Other proteins whose functions are less well understood are also required for nuclear migration. One such protein, NUDF of A. nidulans (Xiang et al. 1995a), is particularly interesting because it is 42% identical to LIS1, a human protein whose haploinsufficiency causes a devastating brain malformation known as lissencephaly (smooth brain; Reiner et al. 1993; Hattori et al. 1994). Lissencephaly is characterized by a smooth cerebral cortex, which is thought to be the result of impaired neuronal migration to the cortex during brain development (Hirotsune et al. 1998; reviewed in Walsh 1999; Walsh and Goffinet 2000).
Genetic analysis indicates that nudF and its S. cerevisiae homologue PAC1 function in the dynein/dynactin pathway (Xiang et al. 1995a; Geiser et al. 1997; Willins et al. 1997). Particularly intriguing is that mutations in the cytoplasmic dynein heavy chain (CDHC) can suppress nudF mutations (Willins et al. 1997). However, none of the proteins identified so far as components of the purified dynein or dynactin complexes from animal sources appear to be homologues of NUDF/Pac1p or LIS1, and the role of NUDF/Pac1p protein in dynein/dynactin function is unclear. Recent studies of a LIS1 homologue in Drosophila (Liu et al. 1999; Swan et al. 1999) revealed that LIS1 acts as a cortical anchor for dynein during Drosophila oogenesis (Swan et al. 1999).
Additional evidence that the NUDF and LIS1 proteins are functionally related comes from their interaction with NUDC. The nudF gene was cloned fortuitously as an extra copy suppressor of the nudC3 mutation in the nudC gene (Xiang et al. 1995a). Remarkably, a mammalian homologue of the NUDC protein, which is known to be functional in A. nidulans (Morris et al. 1997), binds LIS1 protein (Morris et al. 1998b). Unfortunately, besides the detrimental effect of the A. nidulans nudC3 mutation on the NUDF protein level, little is known about the function of the nudC gene. In Aspergillus, the nudC null mutant is phenotypically very different from nudF or CDHC null mutants (Chiu et al. 1997), and thus nudC does not necessarily function only in the dynein/dynactin pathway. Other proteins that have been reported to bind LIS1 include platelet-activating factor acetylhydrolase (Hattori et al. 1994), pleckstrin homology domains (Wang et al. 1995), p72syk-like tyrosine kinase (Brunati et al. 1996), and tubulin and microtubules (Sapir et al. 1997). It is not known if fungal homologues of LIS1 are involved in similar interactions.
Here we identify nudE, the A. nidulans homologue of the N. crassa nuclear distribution gene ro-11, as a multicopy suppressor of a mutation in the nudF gene. According to genetic data, the RO11 protein functions in the cytoplasmic dynein pathway, but its precise role is not clear (Minke et al. 1999). We show that A. nidulans NUDE and NUDF proteins interact in a two-hybrid system and in A. nidulans protein extracts and demonstrate the generality of this interaction by showing that a similar interaction occurs between the human NUDE/RO11 homologue and human LIS1.
Materials and Methods
Strains and Plasmids
A. nidulans strains used were 20.3.10 (argB2; pyrG89; pabaA1; fwA1), GR5 (pyrG89; pyroA4; wA2) (made by G. May), AO1 (nudC3; pyrG89; pabaA1; nicA2; wA2) (Osmani et al. 1990), SF2-9 (ΔnudE::argB; argB2; pyrG89; pabaA1; fwA1) (this work), XX3 (nudA1; pyrG89; chaA1) (Xiang et al. 1994), XX20 (nudF6; pyrG89; yA2), XX21 (nudF7; pyrG89; yA2) (Xiang et al. 1995a), XX21x6-17 (nudE::(VSV-G)6::pyrG; nudF7; pyrG89; yA2) (this work), ΔF54 (ΔnudF::pyr4; pyrG89; wA2; pyroA4) (Willins et al. 1995). Construction of the pAid vector has been described by Xiang et al. 1999. pMS12 was obtained from the Fungal Genetics Stock Center. X. laevis MP3 cDNA clone (data are available from GenBank/EMBL/DDBJ under accession number U95097) was kindly provided by Todd Stukenberg (Harvard Medical School, Boston, MA). A human LIS1 cDNA clone (accession number L13385) was obtained from Takeo Kubota (NHGRI, NIH, Bethesda, MD). A human EST clone with the accession number AA424443 was purchased from Genome Systems, Inc.
Cloning of the nudE Gene
The screen for nudF7 multicopy suppressors was based on the method previously used by us to clone A. nidulans genes (Efimov and Morris 1998). Genomic DNA fragments (5–20 kb) from the XX20 strain were ligated to the pAid vector, which is the A. nidulans replicating vector pHELP1 (Gems and Clutterbuck 1993) carrying the pyrG gene as a selective marker. XX21 protoplasts were transformed with the ligation mixture and grown at 43°C on twenty-eight 10-cm plates in YAG (0.5% yeast extract, 2% glucose, 2% agar, and trace elements) with 0.6 M KCl. The total number of pyrG+ transformants was >3 × 104. 79 putative suppressor clones were identified as patches of yellow color in the gray background of unconidiating mycelium. In 71 of those, conidiation was mitotically unstable and was lost in the absence of selection for the pyrG gene. Plasmids from nine unstable transformants were rescued in E. coli. Seven plasmids contained different but overlapping inserts with the nudF6 gene. Two plasmids contained overlapping inserts with a gene that displayed homology to the N. crassa gene ro-11, which we named nudE. A 4.8-kb fragment of A. nidulans genomic DNA from the smallest suppressor plasmid (∼7-kb insert) was subcloned and sequenced. The position of the nudE gene was predicted using homology between its ORF and the RO11 protein. The presence of a single intron, as well as the positions of the initiator and terminator codons, were confirmed by PCR analysis and sequencing of an A. nidulans cDNA library. The nudE sequence has been deposited in the GenBank (accession no. AF085679).
Disruption and Epitope-tagging of the nudE Gene
To disrupt the nudE gene, the A. nidulans argB gene was inserted as a 1.7-kb EcoRI-BamHI fragment from pMS12 between the unique MfeI and BamHI sites of the nudE gene on a plasmid. This resulted in deletion of the first 1,233 bp of the nudE gene and 265 bp upstream of the nudE start. The disrupted gene was introduced into the wild-type nudE locus of the 20.3.10 strain by homologous double recombination with a linear DNA fragment. The correct integration was confirmed by Southern hybridization for two independent argB+ transformants, which were designated SF2-3 and SF2-9.
Tagging of the NUDE protein with six copies of the epitope YTDIEMNRLGK, which is recognized by the monoclonal antibody P5D4, was done as follows. First, the plasmid pP5D4x6-3 was constructed. This carries the epitope-encoding sequence GTAGATCT [TAC ACC GAC ATC GAG ATG AAC CGC CTC GGC AAG (GGATCT)]5 (TAC ACC GAC ATC GAG ATG AAC CGC CTC GGC AAG) GGATCCA flanked by NcoI sites in the pGEM-5Z vector. Second, plasmid pSH-1 was made by subcloning the NruI-AccI A. nidulans genomic fragment (2,621 bp), which contains the truncated nudE gene (residues 45–586 of the NUDE protein), at the EcoRV-ClaI sites of the pXX1 vector (Xiang et al. 1995b). The epitope-encoding NcoI-NcoI fragment from pP5D4x6-3 was inserted at the unique NcoI site of the pSH-1 in frame with the nudE gene to produce plasmid pSHx6-3. The XX21 strain was transformed with the circular pSHx6-3 plasmid and pyrG+ transformants were analyzed by PCR. Four transformants, designated XX21x6-#8, 12, 15, 17, were found that had undergone a single recombination event at the nudE locus resulting in the insertion of the epitope sequence in the intact nudE gene.
Domain Analysis of the NUDE Protein
The constructs used for the experiments shown in Fig. 1 B all carry the same 4.8-kb region of the A. nidulans genomic DNA at the SmaI site of pAid. The nudE gene is approximately in the middle of the insert and is oriented toward the AMA1 insert. Changes within the nudE gene were made as follows. The COOH-terminal part of NUDE (residues 216–586) was deleted by excising the BglII-MfeI fragment (539 bp after filling in). This disrupts the ORF after residue 215 of NUDE and adds 16 new amino acid residues followed by a stop codon. Hybrids between the nudE gene and human and mouse cDNAs were made using recombination by PCR. The region of the nudE gene encoding residues 27–182 was replaced with the fragments of human cDNA clone (accession number AA424443) or X. laevis MP43 cDNA clone (accession number U95097). The replacing fragments encode 152-residue-long domains similar to the NUDE coiled-coil. The sequence of the used portion of the AA424443 clone was identical to the consensus derived from overlapping ESTs AA424443, R55738, AA442918, and H24090.
The affinity-purified polyclonal antibody against the NUDF protein was made by Xiang et al. 1995a. P5D4 mouse monoclonal antibody and protein G–coupled Sepharose 4B were from Sigma-Aldrich. 200 μl of protein G–Sepharose (packed volume) and 60 μl of P5D4 antibody solution (6.5 mg/ml of IgG) were coupled with dimethylpimelimidate as described (Harlow and Lane 1988). Antigens were detected on immunoblots using an appropriate alkaline phosphatase conjugate and BCIP/NBT as a chromogenic substrate.
To make total protein extracts, A. nidulans cells from 18 h cultures were disrupted by grinding in liquid nitrogen and resuspended in 50 mM Tris-HCl (pH 8.0), 0.1% Triton X-100, 0.5% SDS. 1 ml of extraction buffer was added per gram of dry-pressed mycelium. The extracts were cleared by centrifugation for 10 min at 16,000 g and adjusted to the same protein concentration using the Bradford assay. Alternatively, the ground mycelium was resuspended in the SDS-PAGE loading buffer, boiled and cleared by centrifugation.
For immunoprecipitations, XX21 and XX21x6-17 strains were grown at 32°C for 18 h. 0.45 g of dry-pressed mycelium was disrupted by grinding in liquid nitrogen followed by addition of 0.45 ml of PBS containing protease inhibitor cocktail for fungal extracts (Sigma-Aldrich) and vortexing with glass beads. All subsequent steps were performed on ice or at 4°C. Extracts were cleared by centrifugation for 5 min at 10,000 g and incubated for 2 h with 40 μl (packed volume) of protein G–Sepharose beads coupled to P5D4 monoclonal antibody in a total volume of 0.6 ml on a rocker. Beads were washed five times with 1 ml of PBS, boiled in 60 μl of SDS-PAGE loading buffer, and removed by centrifugation. 10-μl samples were analyzed by Western blotting and silver staining.
A Gal4p-based yeast two-hybrid system Matchmaker 3, which includes S. cerevisiae strain AH109 and plasmids pGBKT7 and pGADT7, was purchased from CLONTECH Laboratories and used according to supplied instructions.
DAPI staining of nuclei and quantitation of nuclear migration was done essentially as described (Willins et al. 1995). The spores of the nudF7 mutant were transformed with different multicopy plasmids, germinated under the most restrictive conditions (43°C, no KCl), and stained with DAPI when the majority of cells had 8–16 nuclei. Nuclear migration was scored for germlings with 8–16 nuclei. If a spore had several germtubes, only the one with the most nearly normal nuclear distribution was considered. Nuclear migration was scored as failed if fewer than three nuclei had entered the germtube. If three or more nuclei were observed in the germtube, but nuclei were not evenly distributed or an abnormally large cluster was present in the spore remnant, nuclear migration was scored as intermediate.
Extra Copies of the nudE Gene Suppress the nudF7 and Other Conditional nud Mutants
We employed an A. nidulans autonomously replicating vector (Gems and Clutterbuck 1993; Aleksenko and Clutterbuck 1997) to screen for multicopy suppressors of the temperature sensitive nudF7 mutation in A. nidulans. This vector maintains itself extrachromosomally in A. nidulans at ∼10 copies per nucleus (Gems et al. 1991). The screen was based on partial complementation of the XX21 (nudF7; pyrG89; yA2) conidiation defect by genomic DNA fragments ligated to the A. nidulans replicating vector pAid. The XX21 strain is temperature sensitive due to the presence of the nudF7 mutation. It grows like wild-type and produces yellow conidia (asexual spores) at the permissive temperature of 32°C, but forms smaller colonies and fails to produce conidia at the restrictive temperature of 43°C. Under semi-restrictive conditions, such as intermediate temperature (37°C) or at 43°C on high salt medium (0.6 M KCL), this strain grows almost to the same size as wild-type, but fails to conidiate and hence has the brownish color of unconidiating mycelium. We expected that a multicopy suppressor that improved the growth of the XX21 would also improve its conidiation. This would render the mycelium with the suppressor plasmid a yellow color that should be noticeable in the background of brownish, unconidiating mycelium.
Two genes were repeatedly identified in the screen. One was the nudF6 allele of the nudF gene (the nudF6 mutant was used as a source of DNA for the screen). Another gene was similar to the N. crassa nuclear migration gene ro-11 (Minke et al. 1999). The ro-11-like gene was named nudE. Note, that nud mutations assigned to loci B, D, and E in the original screen (Morris 1976) were subsequently mapped to the nudA gene (Morris, N.R., unpublished data). Multicopy plasmids with the nudE gene improve the growth, conidiation and nuclear distribution of the nudF7 mutant (Fig. 1 A; Table). Suppression of the nudF6 mutation, which is tighter than nudF7 (Xiang et al. 1995a), was very weak. Extra copies of the nudE gene also suppressed the conidiation defect of the nudA1 CDHC mutant grown under mildly restrictive conditions (32°C, no KCl, Fig. 1 A), but suppression of the nudA1 mutant and of the nudC3 mutant was barely detectable under the most restrictive conditions (43°C, no KCl). The conidiation defect of the nudC3 mutant grown under mildly restrictive conditions was not suppressed, nor was the conidiation defect of another mutation that affects nuclear migration and conidiation, apsA5 (Fischer and Timberlake 1995; data not shown).
The amount of NUDF protein is reduced at the restrictive temperature in nudF6 and nudF7 mutants (Xiang et al. 1995a). However, suppression by nudE was not due to the affect on NUDF concentration. Extra copies of the nudE gene did not increase the level of NUDF protein in the nudF7 mutant (Fig. 2). Predictably, extra copies of the nudF6 gene increase the amount of the mutant NUDF protein (Fig. 2). Interestingly, we consistently observed inhibition of the nudC3 mutant by extra copies of the nudF6 allele (Fig. 1 A). This is opposite to the effect of the wild-type nudF gene, which is a multicopy suppressor of the nudC3 mutation (Xiang et al. 1995a). A possible interpretation is that the overexpressed mutant NUDF protein sequesters mutant NUDC.
The nudE null mutant was viable, but displayed impaired nuclear migration, reduced colony growth, and a conidiation defect. All these defects were less severe than in the ΔnudF or ΔnudA mutants (Table and Fig. 1 B), which have identical phenotypes (Willins et al. 1995). The double mutants ΔnudE;ΔnudF and ΔnudE;ΔnudA were similar to the ΔnudF and ΔnudA single mutants (data not shown), indicating that nudE affects the same pathway as nudF and CDHC. To make sure that differences between the ΔnudE and ΔnudF strains were not caused by background mutations, we crossed the ΔnudE strain to GR5, the wild-type parent of the ΔnudF strain. All progeny obtained from the cross were either wild-type or ΔnudE-like.
Only the NH2-terminal Coiled-Coil Domain of the NUDE Protein Is Absolutely Required for Its Function
The NUDE protein closely resembles the RO11 protein of N. crassa (Minke et al. 1999). The similarity is strongest within the first 190 residues, which are predicted to form a continuous coiled-coil structure (Fig. 3 A, see also Minke et al. 1999). The COOH-terminal region that follows the coiled-coil is poorly conserved between NUDE and RO11. In both proteins, however, it has an excess of positively charged residues (calculated pI = 12) and a high serine content (>18%). Other similar features of the COOH-terminal domain include a short central coiled-coil or an α-helix and the sequence DLGET(F,Y) at the very end.
A search of the public database for proteins similar to NUDE/RO11 turned up multiple sequences that had extensive homology with the coiled-coil in the NH2-terminal part of the NUDE/RO11 but little or no homology with the rest of the molecule (Fig. 3 B). One was the mitotic phosphoprotein 43 (MP43) of Xenopus laevis (Stukenberg et al. 1997). Multiple human and mouse ESTs, apparently coming from at least two different genes, were also identified. The translation product of the human EST AA424443 that we chose to work with is almost identical (94% identity for the sequence shown in Fig. 3 B) to the rabbit endooligopeptidase A–related protein (Hayashi et al. 2000). A human cDNA encoding a similar coiled-coil (72% identity with AA424443) was recently annotated as HOM-TES-87 tumor antigen mRNA (accession number AF124431).
We constructed two chimeric nudE genes by substituting the sequence encoding the most conserved region of the NUDE coiled-coil (the region between arrowheads in Fig. 3 B) for the corresponding sequences from the human EST AA424443 and from the frog MP43 cDNA. Neither of the constructs was functional in A. nidulans (Fig. 1 B). A NUDE deletion variant with the in-frame deletion of residues 45–214 was also not functional (data not shown). The latter construct, when expressed as a fusion with green fluorescent protein, could be readily observed in live cells (Efimov, V.P., and N.R. Morris, manuscript in preparation), ruling out the possibility that the deletion variant was not expressed. Surprisingly, the NUDE construct containing the coiled-coil region alone complemented the nudE deletion and suppressed the nudF7 mutation (Fig. 1 B). Thus, the biological function of the NUDE protein is carried out by its NH2-terminal coiled-coil, while the function of the COOH-terminal domain appears to be dispensable.
The NUDE Coiled-Coil Domain Interacts with the NUDF Protein and this Interaction Is Evolutionary Conserved
Multicopy suppressors often identify physically interacting proteins. We detected a NUDE-NUDF interaction in a Gal4p-based yeast two-hybrid system (Fig. 4). The same results were obtained with the full-length NUDE and its NH2-terminal coiled-coil fragment (data not shown). The interaction appears to be weak as it can be detected only with the most sensitive reporter gene HIS3. The use of animal homologues of the NUDE and NUDF proteins provides a rigorous control for the specificity of this interaction. No interaction can be detected between A. nidulans NUDE and human LIS1 or between chimeras of NUDE carrying the coiled-coil regions from human or frog proteins and A. nidulans NUDF. On the other hand, the frog and human NUDE chimeras interact with human LIS1. The latter interactions are readily observed with both the HIS3 and ADE2 reporter genes. As expected, all coiled-coil constructs interact with themselves. Interestingly, the Aspergillus NUDE coiled-coil does not interact with the Xenopus or human coiled-coils, while the Xenopus and human coiled-coils interact with each other. These results demonstrate that the region delimited by amino acid residues 25–183 of the A. nidulans NUDE coiled-coil specifically binds NUDF protein, while the corresponding region of the human and frog NUDE/RO11 homologues binds human LIS1.
Finally, we tested whether NUDE and NUDF proteins interacted with each other physically in A. nidulans protein extracts. To facilitate detection of the NUDE protein, we tagged the nudE gene in the A. nidulans genome with six copies of the Vesicular Stomatitis Virus Glycoprotein (VSV-G) epitope sequence, which is recognized by the monoclonal antibody P5D4 (Soldati and Perriard 1991). The tagged NUDE protein differs from the native protein by the extra sequence RS[YTDIEMNRLGK (GS)]6TMG following glycine 569. Four independently obtained tagged strains were indistinguishable from the parent strain, indicating that the tagged gene was functional. The tagged NUDE::(VSV-G)6 protein was readily detected in total protein extracts by Western blotting with the P5D4 antibody (Fig. 5 A). The protein is likely to be modified as it often migrates as a closely spaced doublet and can be further separated into at least four distinct bands (Fig. 5 B). Immunoprecipitations were performed with the P5D4 antibody using protein extracts from the tagged nudE::(VSV-G)6 strain and, as control, the parent nudE+ strain. The NUDE::(VSV-G)6 and NUDF proteins were found to coprecipitate (Fig. 5 C), thus corroborating the results of the two-hybrid assay.
We conducted a multicopy suppressor screen in the filamentous fungus A. nidulans to identify genes interacting with the nudF gene, a putative homologue of the human gene LIS1, which is implicated in brain development. The nudE gene, an apparent homologue of N. crassa ro-11, was obtained from the screen. Multicopy suppressors often correspond to physically interacting proteins. We observed a specific interaction between NUDE and NUDF in a yeast two-hybrid system (Fig. 4). In addition, an epitope-tagged NUDE coprecipitated with NUDF in A. nidulans protein extracts (Fig. 5 C). Thus, the simplest explanation for the suppression of conditional nud mutants by extra copies of the nudE gene is that increasing the level of the NUDE protein results in a more efficient formation of the NUDE-NUDF complex. NUDE-NUDF association is consistent with our observations that in live A. nidulans cells, full-length NUDE- and NUDF-GFP fusions localize to similar comet-like structures that move toward the hyphal tips (Xiang, X., V.P. Efimov, and N.R. Morris, manuscript in preparation). Similar structures are also observed with the CDHC-GFP fusion (Xiang et al. 2000). Given the size of the NUDE coiled-coil that binds NUDF (∼170 residues or 25 nm in length) and its conservation between fungal, frog and human proteins (Fig. 3 B), it is likely to bind other proteins in addition to NUDF.
The closest match to the NUDE protein is the N. crassa nuclear migration protein RO11. Deletion of the ro-11 gene in N. crassa causes the same nuclear distribution defect that is characteristic of dynein/dynactin mutants, without affecting cytoplasmic microtubules, dynein/dynactin accumulation at hyphal tips or the levels of CDHC (RO1) and p150Glued (RO3) (Minke et al. 1999). Deletion of the A. nidulans nudE gene similarly produced a viable strain with a nuclear distribution defect. The sequence similarity between NUDE and RO11 proteins and the involvement of both proteins in nuclear distribution indicate that NUDE and RO11 are functional homologues. Both proteins consist of two distinct domains: a highly conserved NH2-terminal coiled-coil and a poorly conserved COOH-terminal tail. The importance of the NH2-terminal coiled-coil domain is underscored by the fact that it is sufficient for the biological activity of NUDE protein when expressed from a multicopy plasmid. However, it is possible that the extra copies of the gene make up for the lack of the COOH-terminal domain.
We selected the coiled-coil regions encoded by the X. laevis protein MP43 (Stukenberg et al. 1997) and by a human EST (accession number AA424443) as potential homologues of the NUDE/RO11 NH2-terminal domain (Fig. 3 B). However, they were not functional in A. nidulans (Fig. 1 B). Also, we were not able to complement the A. nidulans nudF mutations with the human LIS1 gene (Xiang, X., V.P. Efimov, and N.R. Morris, unpublished data), despite the 42% sequence identity between the two proteins. As an alternative, we used a yeast two-hybrid system to analyze interactions between Aspergillus, human, and Xenopus proteins (Fig. 4). The A. nidulans NUDE coiled-coil interacts with NUDF and itself, while NUDE chimeras carrying the coiled-coil regions from human or frog homologues interact with human LIS1 and themselves. The above interactions are extremely specific as A. nidulans NUDE does not interact with human LIS1 or human and frog NUDE homologues, nor does NUDF interact with the human or frog NUDE homologues.
While this work was in progress, we learned that a mouse NUDE/RO11-like protein was identified by Dr. S. Hirotsune's lab in a two-hybrid screen with the mouse LIS1 protein (Sasaki, S., A. Shionoya, M. Ishida, Y. Sugimoto, and S. Hirotsune, manuscript submitted for publication). At the amino acid level, the human EST fragment used by us (sequence in Fig. 3 B) has 99.5% identity with the mouse sequence. A slightly different sequence has been identified in the two-hybrid screen with the human LIS1 protein by Dr. C. Walsh's lab (Feng, Y., and C. Walsh, personal communication).
The region of the MP43 protein that follows the coiled-coil is much shorter than the equivalent region in the NUDE protein (160 residues vs. 394) and shows little sequence homology to NUDE. The salient features of this region in NUDE and RO11, an excess of positively charged residues and serines and the presence of a short central α-helix, are recognizably conserved in the MP43 COOH-terminal domain. It has a pI of 9.8, a serine content of 15%, and a predicted central α-helical region of ∼50 residues. Considering all the facts, it is very likely that MP43 is a homologue of the NUDE/RO11 protein that functions in the cytoplasmic dynein pathway and interacts with the X. laevis homologue of LIS1 protein. As the MP43 protein is phosphorylated specifically during mitosis (Stukenberg et al. 1997), an exciting possibility is that its modification may be related to dynein/dynactin regulation. Cytoplasmic dynein drives the movement of membranous networks in Xenopus interphase extracts, and this movement ceases during mitosis, presumably because dynein dissociates from membranes (Niclas et al. 1996).
The sequence similarity between NUDF and LIS1, and the fact that both NUDF and LIS1 interact with the same domain of NUDE and its human homologue, respectively, support the idea that NUDF and LIS1 are functionally related. The documented involvement of fungal NUDF/Pac1p and NUDE/RO11 proteins in the cytoplasmic dynein pathway suggests that LIS1 exerts its effect through cytoplasmic dynein. Several hypothetical roles for cytoplasmic dynein in lissencephaly have been discussed (Sapir et al. 1997; Morris et al. 1998a,Morris et al. 1998b; Vallee et al. 2000). They include a role in the regulation of microtubule dynamics, in mitosis, in nuclear translocation, and in retrograde transport of signaling molecules. Of particular interest is the recent finding that in axons cytoplasmic dynein generates a force counterbalancing contractile forces generated by myosin (Ahmad et al. 2000).
We thank Drs. Shinji Hirotsune, Yuanyi Feng, and Christopher Walsh for communicating their unpublished results, Todd Stukenberg for sending the MP43 cDNA, and Xin Xiang for numerous suggestions and reagents.
Supported by National Institutes of Health grant no. 5R01GM52309-03 to N.R. Morris.
Abbreviations used in this paper: CDHC, cytoplasmic dynein heavy chain, GFP, green fluorescent protein; nud, nuclear distribution defective; ts, temperature sensitive; VSV-G, Vesicular Stomatitis Virus Glycoprotein.