|Home | About | Journals | Submit | Contact Us | Français|
To learn more about how dyneins are targeted to specific sites in the flagellum, we have investigated a factor necessary for binding of outer arm dynein to the axonemal microtubules of Chlamydomonas. This factor, termed the outer dynein arm-docking complex (ODA-DC), previously was shown to be missing from axonemes of the outer dynein armless mutants oda1 and oda3. We have now partially purified the ODA-DC, determined that it contains equimolar amounts of Mr ~105,000 and ~70,000 proteins plus a third protein of Mr ~25,000, and found that it is associated with the isolated outer arm in a 1:1 molar ratio. We have cloned a full-length cDNA encoding the Mr ~70,000 protein; the sequence predicts a 62.5-kDa protein with potential homologs in higher ciliated organisms, including humans. Sequencing of corresponding cDNA from strain oda1 revealed it has a mutation resulting in a stop codon just downstream of the initiator ATG; thus, it is unable to make the full-length Mr ~70,000 protein. These results demonstrate that the ODA1 gene encodes the Mr ~70,000 protein, and that the protein is essential for assembly of the ODA-DC and the outer dynein arm onto the doublet microtubule.
Dyneins are large, multisubunit molecular motors that generate force against microtubules. Cilia and flagella contain three major classes of dyneins: cytoplasmic dynein, of which cytoplasmic dynein 1b/2 is the retrograde motor for intraflagellar transport (Pazour et al., 1998 , 1999 ; Porter et al. 1999 ); axonemal inner arm dyneins, of which there may be as many as seven different forms (Porter and Sale, 2000 ); and axonemal outer arm dynein, of which only one form is currently known (Witman et al., 1994 ). Each of these different dyneins binds with high fidelity to sites that are specific for that particular dynein. Thus, these sites must be structurally or biochemically unique in a way that ensures that the correct dynein is targeted to them. The molecular basis for this specific binding is of great interest because it literally provides the foundation for the proper functioning of the flagellum. Moreover, knowledge of how these dyneins are targeted to their correct binding sites may provide a paradigm for how other axonemal components, e.g., the radial spokes and the projections of the central pair microtubules, are correctly positioned to form one of the most complex and highly ordered macromolecular structures in the cell.
To learn more about specific targeting of dyneins, we have been studying the Chlamydomonas reinhardtii outer arm dynein, which is the most well characterized of axonemal dyneins (Witman et al., 1994 ; Pazour and Witman, 2000 ). This dynein, which produces as much as four-fifths of the force for flagellar movement (Brokaw, 1994 ), is attached to specific sites on the A-tubules of the flagellar doublet microtubules and repeats at 24-nm intervals along the length of the doublet. It contains three dynein heavy chains (DHCs) (termed α, β, and γ), two intermediate chains (ICs) (termed IC78 and IC69), and several light chains (LCs) (Figure (Figure6B).6B). The two ICs and most of the LCs are associated in an IC/LC complex located at the base of the dynein (King and Witman, 1990 ). One of the ICs, IC78, is in direct contact with tubulin in vivo (King et al., 1991 ) and is a microtubule-binding protein in vitro (King et al., 1995 ), so it is believed to be at least one of the dynein components that anchors the outer arm to the A-tubule.
Studies of Chlamydomonas mutants lacking the outer dynein arm (oda mutants) indicate that an additional factor is necessary for efficient assembly of the outer arm onto flagellar doublet microtubules. When the Chlamydomonas outer arm dynein is removed from the wild-type axoneme by extraction with 0.6 M KCl in the absence of Mg2+, it dissociates into a single-headed γ subunit (containing the γ DHC and two LCs) that sediments as a 12S particle, and a two-headed αβ subunit (containing the α and β DHCs, the two ICs, and all the remaining LCs) that sediments as a 21S particle (Piperno and Luck, 1979 ; Pfister et al., 1982 ; King and Witman, 1989 ). When these two particles were purified and added back to axonemes of the outer armless mutants oda2, oda4, oda5, or oda6, outer arms were reconstituted in their correct positions on the doublet microtubules (Takada and Kamiya, 1994 ). In contrast, if the subunits were added back to axonemes of the outer armless mutants oda1 or oda3, very few arms rebound. Efficient rebinding of arms to oda1 or oda3 axonemes required a factor that was present in the 0.6 M KCl extract of wild-type axonemes and sedimented at 7S. This factor apparently was missing in axonemes of oda1 and oda3, but present in axonemes of oda2, oda4, oda5, and oda6. The only discernible structural difference between axonemes of these two groups of mutants was that the latter had a small projection at the site where the outer arm normally would attach. These findings strongly suggested that the projection facilitates attachment of the dynein to the doublet microtubule, and that the 7S factor is the solubilized form of the projection. Partial purification of the factor and comparison of extracts from oda mutant axonemes showed that the factor was correlated with a polypeptide of Mr ~70,000 (Takada and Kamiya, 1994 ). Because the factor can assemble onto the doublet microtubules in the absence of the outer dynein arms in vivo, and is necessary for binding of the arms to the microtubules, it has been termed the outer dynein arm-docking complex (ODA-DC).
In this report, we show that the ODA-DC contains equimolar amounts of proteins of Mr ~105,000 and ~70,000 plus a third protein of Mr ~25,000, that it is present in a 1:1 stoichiometry with the outer dynein arm polypeptides, and that it remains associated with the outer dynein arm subunits when these subunits are isolated under conditions that keep them together as a three-headed αβγ complex (Takada et al., 1992 ). We have cloned and sequenced a full-length cDNA encoding the Mr ~70,000 protein; the sequence predicts a novel 62-kDa polypeptide with three long coiled-coil domains. Sequencing of the corresponding DNA from the outer armless mutant oda1 reveals that it has a mutation resulting in a stop codon just downstream from the translation initiation site; hence, it is unable to make the Mr ~70,000 ODA-DC protein. These results indicate that the Mr ~70,000 polypeptide is the ODA1 gene product and is essential for assembly of both the outer dynein arm and the ODA-DC onto the doublet microtubule. We previously showed that the ODA3 gene product corresponds to the Mr ~105,000 ODA-DC polypeptide and is predicted also to have three long coiled-coil regions (Koutoulis et al., 1997 ). Herein, we use a zero-length protein cross-linker to show that the ODA1 gene product is in direct contact with the ODA3 gene product in the soluble ODA-DC. Therefore, the ODA1 and ODA3 gene products interact with each other, possibly via their coiled-coil domains, and with an Mr ~25,000 subunit to form a structure that targets the outer dynein arm to its correct attachment site on the doublet microtubule. Potential homologs of the ODA1 gene are expressed in higher organisms, including humans, suggesting that the ODA-DC is widespread in ciliated animals.
C. reinhardtii strains used were wild type (137c), and outer armless mutants oda1-1, oda3-1, and oda6-1, all of which were derived from strain 137c (Kamiya, 1988 ).1 Cells were grown in liquid culture as described by Witman (1986) .
Flagellar axonemes from oda6 cells were isolated by the method of Witman (1986) , washed with 0.5 M potassium acetate in 30 mM HEPES pH 7.5, 5 mM MgSO4, 1 mM dithiothreitol, 0.5 mM EGTA (Nakamura et al., 1997 ), and then extracted with 0.6 M KCl in HMDEK (30 mM HEPES pH 7.5, 5 mM MgSO4, 1 mM dithiothreitol, 0.5 mM EGTA, 25 mM potassium acetate) to solubilize the ODA-DC. Axonemes from oda3 cells, which lack the ODA-DC, were treated identically. The extracts were fractionated by 5–20% sucrose density gradient centrifugation under Mg2+-free conditions (Piperno and Luck, 1979 ; Pfister et al., 1982 ). Fractions were collected from the bottom of the tube and their proteins analyzed by SDS-PAGE.
Axonemes from wild-type cells were isolated, washed, and extracted as described above. The dynein extract was then fractionated by centrifugation (5-ml tube, Beckman SW50.1 rotor, 39,000 rpm for 7 h) through a 5–20% sucrose density gradient containing 5 mM Mg2+ (Takada et al., 1992 ).
Proteins from sucrose density gradient fractions were separated by electrophoresis in 5–20% polyacrylamide-SDS gels (King et al., 1986 ). The gels were stained with Coomassie blue and photographed on 35-mm Tech Pan film. Negatives were scanned with a Nikon Coolscan II film scanner. The resulting image files were analyzed and bands of interest integrated using one-dimensional gel analysis software (Quantity One; PDI, Huntington Station, NY).
Axonemes from wild-type cells were isolated, washed, and extracted as described above. The extract was dialyzed against TEDKS (30 mM Tris pH 7.5, 0.5 mM EDTA, 1 mM dithiothreitol, 25 mM KCl, 10% sucrose) containing 1 mM phenylmethylsulfonyl fluoride overnight at 4°C to dissociate outer arm dynein from the ODA-DC, and then dialyzed against phosphate-buffered saline (PBS) (6.4 mM Na2HPO4, 0.15 mM KH2PO4 pH 7.5, 137 mM NaCl, 2 mM KCl) at 4°C for 5 h. To biotinylate the proteins in the extract, sulfo-N-hydroxysuccinimide-biotin (Pierce Chemical, Rockford, IL) was added to an ~50-fold molar excess over protein and the extract incubated at 24°C for 30 min. To remove unreacted biotin, the extract was dialyzed against PBS at 4°C overnight. The biotinylated extract was divided into two tubes (48 μg of protein/tube) and diluted to 200 μl with IP buffer 1 (3% bovine serum albumin [BSA], 1% Nonidet P-40, 10 mM Tris pH 8.0, 150 mM NaCl). Rabbit polyclonal IgG (4 μg) specific for the Mr ~105,000 ODA-DC polypeptide (anti-DC105 antibody) (Wakabayashi et al., 2001 ) was added to one of the tubes and incubated on ice for 2 h. As a control, 4 μg of rabbit normal IgG was added to the other tube. Affi-Prep protein A beads (20 μl ; Bio-Rad, Hercules, CA) in 100 μl of IP buffer 1 were added and incubated at 4°C for 1 h. The beads were recovered by centrifugation and washed with IP buffer 1 at 4°C for 5 min, three times. A final wash was performed with IP buffer 1 without BSA at 4°C for 10 min. Then 50 μl of 1× SDS-PAGE sample buffer was added to the beads and the sample boiled at 100°C for 5 min. Proteins in the resulting supernatants were separated by electrophoresis in 10% SDS-polyacrylamide gels, transferred to a nitrocellulose membrane, probed with streptavidin-horseradish peroxidase (HRP) (Molecular Probes, Eugene, OR), and detected using a chemiluminescent substrate. Simultaneously, an extract from oda1 axonemes was prepared and processed identically.
Axonemes from wild-type cells were isolated and extracted with 0.6 M KCl in HMDEK as described above except that the 0.5 M potassium acetate wash was omitted. The extract was dialyzed against HMDEK for 8 h. Immunoprecipitation was performed by the method of Fowkes and Mitchell (1998) with modifications. A 0.5-ml aliquot of the 0.6 M KCl extract (2.8 mg/ml) was added to IP buffer 2 (HMDEK, 75 mM NaCl, 3% BSA, 0.1% Triton X-100, pH 7.4). This mixture was precleared by incubation with protein A-agarose (Roche Diagnostics, Tokyo, Japan) for 30 min at 4°C, followed by centrifugation. The supernatant was then incubated with the anti-DC105 antibody or normal rabbit IgG for 4 h followed by incubation with protein A-agarose for 1 h. The precipitated agarose beads were washed twice with IP washing buffer (IP buffer 2 with 0.05% Triton X-100) and twice with IP washing buffer without BSA. The immune complex was eluted by addition of an equal volume of 2× SDS-sample buffer and incubation at 100°C for 3 min. The eluted proteins were separated by electrophoresis in 6.5% SDS-polyacrylamide gels and transferred to Immobilon membranes (Millipore, Bedford, MA). Immunoreactive bands were detected using HRP-conjugated anti-rabbit IgG or anti-mouse IgG secondary antibodies and a chemiluminescent substrate.
Proteins in 5–20% polyacrylamide gels of sucrose gradient fractions containing the ODA-DC from oda6 cells were transferred electrophoretically to polyvinylidene difluoride membrane (Immobilon-PSQ; Millipore) and stained with Ponceau S. A band at Mr ~70,000, present in the 7S fraction from oda6 but absent in comparable fractions from oda3, was excised and digested with trypsin. The resulting peptides were separated by reverse phase high-performance liquid chromatography, and the amino-terminal sequences of three of the peptides (peptides 1, 2, and 3; Table Table1)1) were determined directly using an amino acid sequencer (model 477A; Applied Biosystems, Foster City, CA).
Wild-type cells were deflagellated by pH shock (Witman et al., 1972 ) in modified Sager and Granick Medium I (Witman, 1986 ), and allowed to regenerate new flagella. Total RNA was isolated from the cells ~30 min after deflagellation (Wilkerson et al., 1994 ). First strand cDNA was made from the RNA by using Superscript II reverse transcriptase (Invitrogen, Carlsbad, CA) and oligo-dT as primer. cDNA fragments were amplified using specific primers (Table (Table1)1) designed from the peptide sequences of the Mr ~70,000 protein. Polymerase chain reaction (PCR) products were subcloned between the EcoRI and BamHI sites of pBluescript II KS (−) (Stratagene, La Jolla, CA), and their ends were sequenced using Sequenase version 2.0 and the 7-deaza-dGTP sequencing kit (U.S. Biochemical, Cleveland, OH). A 750-base pair insert (pcST8-1) amplified by primers A and D was found to be correct, having additional sequences contained in peptides 1 and 2 but not used to design the primers. This cDNA was labeled with [γ-32P]dCTP by using the Stratagene random primer labeling kit.
A λZAPII cDNA library of a C. reinhardtii wild-type strain (Wilkerson et al., 1995 ) was screened with the DNA probe pcST8-1. Positive phage plaques were subjected to a second round of screening. A phagemid pBluescript II SK (−) having a 2.3-kb insert (pcST737) was rescued by in vivo excision. pcST737 was digested with PstI, BamHI, SacII, and XhoI, and four fragments were subcloned. The ends of the fragments were sequenced and the data used to design internal sequencing primers. Single-stranded DNA from both strands of clone pcST737 was isolated and completely sequenced using Sequenase version 2.0 and the 7-deaza-dGTP sequencing kit (U.S. Biochemical).
oda1-1 cells were deflagellated by pH shock and allowed to regenerate flagella for 40 min under illumination, at which time total RNA was isolated from these cells and from oda1-1 and wild-type cells that had not been deflagellated. First-strand cDNA was made from the oda1-1 and wild-type RNA by using reverse transcriptase and oligo-dT primers. The PCR was then carried out with Elongase Enzyme Mix (Invitrogen), which contains Taq DNA polymerase and the proofreading Pyrococcus sp. GB-D polymerase for high fidelity, and by using two primers (Figure (Figure2A,2A, double underlines) designed to amplify the complete open reading frame encoding the Mr ~70,000 ODA-DC protein. Products were cloned between the EcoRI and KpnI sites of pBluescript II KS (−) and subjected to in vitro transcription by using T7 RNA polymerase; the transcripts were then translated in vitro by using a reticulocyte lysate system (Promega, Madison, WI) containing [35S]methionine. The translation products were separated in 7.5 or 10% polyacrylamide gels and autoradiographed. Four clones obtained from oda1-1 were sequenced at their 5′ ends.
The GCG suite of programs (Devereux et al., 1984 ) was used for sequence assembly and protein structure predictions. The program COILS (Lupas et al., 1991 ; Lupas, 1996a ) was used to predict regions of coiled-coil structure. The PSI-BLAST program (Altschul et al., 1997 ) was used to search databases for related sequences. The program CLUSTAL W (Thompson et al., 1994 ) was used to align homologs to the Chlamydomonas sequence. The PROSITE database was used to determine possible sites for post-translational modifications (Bairoch et al., 1995 ).
Preparation of 0.6 M KCl extracts of wild-type axonemes was as described under “Characterization of the ODA-DC” but without the 0.5 M potassium acetate wash. The zero-length cross-linker 1-ethyl-3-(3-dimethylaminopropyl) carbodiimide (EDC) (Sigma Chemical, St. Louis, MO) was added to aliquots of the extract to various final concentrations and the mixtures incubated at 24°C for 1 h. The cross-linking reactions were stopped by addition of equal volumes of 2× SDS-PAGE sample buffer. The cross-linked products were separated by electrophoresis on 6% SDS-polyacrylamide gels, transferred to nitrocellulose membranes, probed with a rabbit polyclonal antibody specific for the Mr ~70,000 ODA-DC polypeptide (anti-DC70 antibody) (Wakabayashi et al., 2001 ), and immunoreactive bands were detected with HRP-conjugated anti-rabbit IgG and a chemiluminescent substrate. The membrane was then stripped and reprobed with the anti-DC105 antibody.
The ODA-DC previously was shown to be absent from axonemes of the outer dynein armless mutants oda1 and oda3, but to be present and functional in axonemes of mutants oda2, oda4, oda5, and oda6 (Takada and Kamiya, 1994 ). The complex also was shown to be present in 0.6 M KCl extracts of oda6 axonemes, and to sediment at 7S in Mg2+-free sucrose density gradients. Therefore, to partially purify the complex and positively identify its polypeptide components, axonemes of oda3 and oda6 were isolated, pre-extracted with 0.5 M potassium acetate to remove proteins that otherwise would contaminate the 0.6 M KCl fraction (Nakamura et al., 1997 ), and then extracted with 0.6 M KCl. The 0.6 M KCl extracts were fractionated by sucrose density gradient centrifugation, and the polypeptide composition of the fractions examined by SDS-PAGE. Polypeptides of Mr ~105,000, ~70,000, and ~25,000 cosedimented at 7S in the oda6 fractions but were uniquely missing from the oda3 fractions (cf. Figure Figure1,1, A and B). The Mr ~70,000 polypeptide previously had been identified as a component of the ODA-DC (Takada and Kamiya, 1994 ). The present results provide evidence that the factor contains additional proteins of Mr ~105,000 and ~25,000.
To confirm that the Mr ~105,000, ~70,000, and ~25,000 polypeptides occur together as a complex, and to investigate whether the wild-type complex contains additional polypeptides not identified in the above-mentioned sucrose density gradient analyses, an antibody specific for the Mr ~105,000 polypeptide (Wakabayashi et al., 2001 ) was used to immunoprecipitate the ODA-DC from an 0.6 M KCl extract of wild-type axonemes. Before the immunoprecipitation, the extract was dialyzed against a Mg2+-free buffer to dissociate the ODA-DC from the outer arm dynein, and then the proteins in the extract were biotinylated. The immunoprecipitated proteins were separated by SDS-PAGE, transferred to a nitrocellulose membrane, and detected using streptavidin-HRP. Three major bands corresponding to the Mr ~105,000, ~70,000, and ~25,000 polypeptides were detected in the wild-type immunoprecipitate (Figure (Figure1C,1C, lane b). Except for two proteolytic fragments and bands present in the normal IgG control (Figure (Figure1C,1C, lane a), no additional bands were observed. These three polypeptides were specifically missing in immunoprecipitates prepared identically from 0.6 M KCl extracts of oda1 axonemes (Figure (Figure1C,1C, lane d), which lack the ODA-DC. Western blot analyses of the wild-type 0.6 M KCl extract verified that the biotinylated Mr ~105,000 and ~70,000 ODA-DC polypeptides comigrated with the two major high-molecular-weight proteins in the immunoprecipitate (Figure (Figure1C,1C, lanes e and f). These results confirm that the three putative ODA-DC polypeptides occur together in a complex in wild-type axonemes. Moreover, because a functional ODA-DC can be isolated from the 0.6 M KCl extract of wild-type axonemes (Takada and Kamiya, 1994 ), it is likely that these three polypeptides are sufficient for ODA-DC function and constitute the entire ODA-DC.
When the outer dynein arm is removed from the wild-type axoneme by extraction with 0.6 M KCl and then purified by sucrose density gradient centrifugation in the presence of Mg2+, the α, β, and γ subunits remain associated as a “three-headed” complex that sediments at 23S (Takada et al., 1992 ). In contrast to the “two-headed” αβ subunit and “single-headed” γ subunit isolated in the absence of Mg2+, this three-headed dynein is competent to bind to the correct sites on oda1 axonemes without addition of exogenous ODA-DC (Takada et al., 1992 ). This raised the question of whether the ODA-DC might remain associated with the soluble three-headed dynein under these conditions. To investigate this, the three-headed dynein was prepared by the method of Takada et al. (1992) , with the sucrose gradient centrifugation carried out at relatively low hydrostatic pressure to prevent dissociation of the outer arm dynein subunits (Nakamura et al., 1997 ). SDS-PAGE analysis of the resulting sucrose density gradient fractions revealed that the Mr ~105,000, ~70,000, and ~25,000 polypeptides now cosedimented with each other and with the outer arm polypeptides at 23S (Figure (Figure1D).1D). Therefore, in the presence of Mg2+ and low hydrostatic pressure, the ODA-DC remains associated with the α, β, and γ outer arm dynein subunits. This is the first demonstration that the ODA-DC and the outer dynein arm directly interact with each other. That all three ODA-DC polypeptides now sediment together at 23S provides additional evidence that they are all part of the same complex.
To obtain further evidence for the direct interaction of the ODA-DC and outer arm dynein in the presence of Mg2+, a 0.6 M KCl extract in HMDEK was prepared from wild-type axonemes. The ODA-DC was immunoprecipitated from the extract by using the antibody specific for the Mr ~105,000 ODA-DC polypeptide. Western blot analysis indicated that, as expected, the immunoprecipitate contained both the Mr ~105,000 and ~70,000 ODA-DC polypeptides (Figure (Figure1E,1E, lanes c and d). To determine whether dynein was coimmunoprecipitated with the ODA-DC, the blots were probed with mouse monoclonal antibodies 1869A and 1878A specific for the outer arm dynein intermediate chains IC69 and IC78, respectively (King et al., 1991 ). Both intermediate chains were detected (Figure (Figure1E,1E, lanes a and b). These results confirm that the ODA-DC is associated with outer arm dynein in the presence of Mg2+.
To determine the stoichiometries of the ODA-DC polypeptides relative to themselves and to the outer dynein arm, quantitative densitometry was carried out on gels such as that shown in Figure Figure1D.1D. The amount of Coomassie blue dye bound to a protein is approximately proportional to the number of positive charges on the protein (Tal et al., 1980 ). The combined mole percentages of arginine, lysine, and histidine in the Mr ~105,000 protein, the Mr ~70,000 protein, IC78, and IC69 are 18, 16, 15, and 13, respectively. Therefore, one would expect the values obtained by quantitative densitometry of a Coomassie blue-stained gel to accurately reflect the relative amounts of these four proteins. From the gel shown in Figure Figure1D,1D, the molar ratios of the Mr ~105,000 protein, IC69, and the Mr ~70,000 protein relative to IC78 were determined to be 0.92, 1.05, and 1.05, respectively (Table (Table2).2). Similar results were obtained with other gels. These results indicate that the Mr ~70,000 ODA-DC protein is present in an equimolar amount with the Mr ~105,000 ODA-DC protein. Moreover, because the outer dynein arm contains one copy of each IC (King and Witman, 1989 ), there appears to be one ODA-DC for each outer arm dynein.
Visual examination of gels indicated that the Mr ~25,000 ODA-DC protein is present in about the same amount as the Mr ~20,000 outer dynein arm LC, but in only about one-half the amount of the Mr ~22,000 outer dynein arm LC. The outer dynein arm is estimated to contain one copy of the Mr ~20,000 LC and two copies of the Mr ~22,000 LC (King and Witman, 1989 ). Therefore, it is likely that the three ODA-DC proteins are present in the complex in a 1:1:1 molar ratio.
The Mr ~70,000 ODA-DC polypeptide from gels such as that shown in Figure Figure1A1A was transferred to polyvinylidene difluoride membrane, stained with Ponceau S, excised, and digested with trypsin. The resulting fragments were isolated by high-performance liquid chromatography and their amino-terminal sequences determined directly (Table (Table1).1). From these sequences, PCR primers were designed (Table (Table1)1) for reverse transcription-PCR; total RNA from wild-type cells that were regenerating their flagella was used as template. Primer A designed from the sequence PGDPFAQ in peptide 1 and primer D designed from the sequence MAQNVA in peptide 2 yielded a 750-base pair product (pcST8-1). Sequencing the ends of this product revealed that it encoded the sequence PGDPFAQALINR—-AMAQNVA, which included sequence from both peptides that was not used in designing the primers. Insert pcST8-1 was then used to screen a wild-type cDNA library constructed in λZAPII (Wilkerson et al., 1995 ). A phagemid with a 2.3-kb insert (pcST737) was isolated and both strands sequenced.
The nucleotide sequence (Figure (Figure2A)2A) reveals that the clone contains a complete open reading frame that predicts a 552-amino acid (62,234 Da) polypeptide with a pI of 5.74. An in-frame stop codon is located at nucleotide 142; the presumed translation initiation site is located at nucleotide 148. A polyadenylation signal (TGTAA) is located just upstream of the poly(A) tail in the 3′-untranslated region. The sequence exactly predicts the three amino acid sequences obtained by direct sequencing of peptides 1, 2, and 3 (cf. Figure Figure2A,2A, single underlines, and Table Table1),1), confirming that it encodes the Mr ~70,000 ODA-DC protein.
The Mr ~70,000 ODA-DC protein is predicted to have a high α-helical content. The COILS program (Lupas et al., 1991 ; Lupas, 1996a ) indicates that three regions (amino acids 27–60, 120–215, and 323–370) have a high probability (1.0) of forming coiled-coil structures (Figure (Figure2B).2B). In addition, there is a region between amino acids 234 and 262 that is predicted to form an α-helix but does not contain a heptad repeat; 57% of the amino acids in this region are charged. This region is similar in structure and sequence to a portion of the Mr ~105,000 ODA-DC protein (see DISCUSSION). The amino-terminal 26 amino acids, of which eight are glycine, are predicted to form a random coil. The carboxyl-terminal 59 amino acids, of which 42% are charged, also are predicted to form a random coil; this region includes a glutamic acid-rich cluster (residues E494 to E507).
A PSI-BLAST search of the databases by using the entire predicted amino acid sequence of the Mr ~70,000 ODA-DC protein revealed one potential homolog in Leishmania major (accession no. CAB55364), two potential homologs in Drosophila melanogaster (AAF55345 and AAF56123), two potential human homologs (AK057357 and AK057488) that appear to be variants expressed from the same gene, and one mouse homolog (BC013491) that is very similar to the human homolog AK057488. An alignment of the Leishmania homolog, one of the Drosophila homologs, and the two human homologs to the Chlamydomonas protein is shown in Figure Figure3;3; BLAST E values and predicted masses and isoelectric points for the homologs are given in Table Table3.3. The proteins are similar in predicted size, and the identities extend throughout the proteins' predicted sequences, suggesting that they are true homologs. The human homologs are known from full-length (AK057357) or partial (AK057488) testis cDNAs; comparison of their predicted carboxyl-terminal amino acid sequences reveals that the last 11 residues of the former are replaced by 249 different residues in the latter, suggesting that the two transcripts are produced by alternative splicing. The gene encoding these proteins maps to chromosome position 19q13 (Human Genome Project Working Draft, University of California, Santa Cruz, CA). No close matches were found in yeast, Caenorhabditis elegans, or Arabidopsis, all of which lack motile cilia.
Axonemes of the mutants oda1 and oda3 lack the ODA-DC as well as the outer dynein arm (Takada and Kamiya, 1994 ), suggesting that the primary defect in these mutants involves the ODA-DC polypeptides. Indeed, the ODA3 gene previously was found to encode the Mr ~105,000 polypeptide (Koutoulis et al., 1997 ), which we herein show is a component of the ODA-DC (see above). Therefore, ODA1 was a strong candidate gene for the Mr ~70,000 protein.
As a rapid initial test to assess if the oda1-1 mutant had a defect in the gene for the Mr ~70,000 protein, we determined whether that protein could be synthesized in vitro beginning with oda1-1 RNA. From the wild-type nucleotide sequence, we designed a pair of PCR primers to amplify the complete coding region for the Mr ~70,000 protein. One primer used sequence >100 base pairs upstream from the initiator ATG in the 5′-untranslated region; the other used sequence next to the poly(A) tail in the 3′-untranslated region (Figure (Figure2A,2A, double underlines). RNA was isolated from oda1-1 cells that were actively regenerating flagella, as well as from nonregenerating wild-type and oda1-1 cells. First-strand cDNA was made using reverse transcriptase, and the PCR carried out with the above-mentioned primers. A large amount of a single product of 2.3 kb was obtained when the cDNA template was prepared from oda1-1 cells that were regenerating flagella (Figure (Figure4A);4A); a smaller amount of the 2.3-kb product was obtained with cDNA prepared from either wild-type cells or oda1-1 cells that were not regenerating flagella. The PCR products from wild-type and regenerating oda1-1 cells were ligated into pBluescript II KS (−) and cloned into Escherichia coli strain XL1Blue. Two wild-type clones and four oda1-1 clones were transcribed using T7 RNA polymerase, and the transcripts translated in a rabbit reticulocyte lysate containing [35S]methionine. Although all clones produced transcripts of the expected size (Figure (Figure4B),4B), only the wild-type transcripts yielded protein of Mr ~70,000 (Figure (Figure4C).4C). With oda1-1 transcripts, smaller protein bands were detected but none had an intensity comparable to that of the Mr ~70,000 band produced by wild-type transcripts. These results strongly suggested that strain oda1-1 has a mutation in the gene encoding the Mr ~70,000 protein and that the complete protein is not expressed in vitro.
This was confirmed by sequence analysis of the 5′ ends of the oda1-1 cDNA clones. In all four clones, the C at nucleotide position 283, 135-base pairs downstream from the initiator ATG, was changed to T (Figure (Figure2A).2A). This single-base replacement converts a CAG codon specifying glutamate to the stop codon TAG. This would result in the translation of only 45 of the 552 amino acids encoded by the wild-type ODA1 gene. It is likely that the minor bands <Mr ~60,000 produced by the oda1-1 transcripts (Figure (Figure4C)4C) are the result of translation initiation at internal sites downstream from this stop codon.
Both the Mr ~70,000 and ~105,000 ODA-DC polypeptides are predicted to have extended coiled-coil regions that are likely to function in dimerization or heterodimer formation (see DISCUSSION). This raised the question of whether the two proteins interact with each other to form a heterodimer that then binds the Mr ~25,000 protein to form the ODA-DC, or whether each polypeptide associates with another copy of itself to form a homodimer, with the two types of homodimers possibly coming together during assembly of the complete ODA-DC. To investigate this, we used the zero-length cross-linker EDC to examine interactions between the ODA-DC proteins in 0.6 M KCl extracts of wild-type axonemes. KCl extracts were chosen for these experiments because interactions between the ODA-DC and tubulin should be minimized in the high-salt environment, thus simplifying the number of cross-linked products obtained. As shown in Figure Figure5,5, EDC generated a series of cross-linked products that migrated between Mr ~175,000 and ~210,000 and contained both the Mr ~70,000 and ~105,000 ODA-DC polypeptides. The apparent size of the smallest of these conjugates is of the size expected for a simple heterodimer of the Mr ~70,000 and ~105,000 polypeptides, but it is not clear why multiple, more slowly migrating products appeared simultaneously. One possibility is that intermolecular cross-linking at different sites produced products with different relative mobilities, a phenomenon postulated to account for a band-spreading observed for cross-linked products of α and β spectrin (Huang and Richards, 1977 ). In any case, all of the products appear to contain both ODA-DC polypeptides, the amounts of the monomeric proteins decrease simultaneously and concomitantly with the appearance of the cross-linked products, and there is no evidence for cross-linking of homodimers of either protein. Because EDC generates covalent linkages solely between groups that interact directly, these results indicate that the Mr ~70,000 and ~105,000 ODA-DC polypeptides are intimately associated with each other, and support the hypothesis that the two proteins interact with one another to form a heterodimer.
To understand the mechanism by which the outer dynein arm is targeted to its correct binding site on flagellar microtubules, we have been characterizing the C. reinhardtii ODA-DC, which originally was described as a “7S factor” when released from axonemes by high-salt extraction (Takada and Kamiya, 1994 ). This factor assembles onto flagellar doublet microtubules in the absence of outer arms in vivo, but is necessary for the outer dynein arm to attach to the doublet microtubules (Takada and Kamiya, 1994 ). Hence, the ODA-DC must be an important intermediary in the binding of outer arm dynein to its unique attachment site within the flagellar axoneme.
Herein, we present three independent lines of evidence that the ODA-DC contains polypeptides of Mr ~105,000, ~75,000, and ~25,000: 1) Sucrose density gradient centrifugation was used to partially purify the ODA-DC after its release from the axoneme by a modified high-salt extraction procedure (Nakamura et al., 1997 ) that reduces the number of “background” polypeptides in the high-salt extract and in subsequent sucrose gradient fractions. Under these conditions, the 7S fraction from axonemes of the outer armless mutant oda6, which has the ODA-DC (Takada and Kamiya, 1994 ), contained polypeptides of Mr ~105,000, ~70,000, and ~25,000 that were missing in equivalent fractions from the mutant oda3, which lacks the ODA-DC. These results strongly suggested that the ODA-DC contains these three polypeptides. The Mr ~70,000 polypeptide previously was reported to be a component of the ODA-DC (Takada and Kamiya, 1994 ). That we were able to detect the presence or absence of the Mr ~105,000 and ~25,000 polypeptides in sucrose gradient fractions probably was due to the reduced number of background bands resulting from our modified extraction conditions. 2) An antibody specific for the Mr ~105,000 protein coimmunoprecipitated the Mr ~70,000 and ~25,000 proteins but no other proteins from a 0.6 M KCl extract of wild-type axonemes. These results confirmed that all three proteins were part of the ODA-DC and indicated that the three proteins are likely to constitute the complete ODA-DC in wild-type cells. 3) The three polypeptides shifted simultaneously from the 7S to the 23S fraction in sucrose density gradients centrifuged under conditions where the ODA-DC remained associated with the more rapidly sedimenting outer dynein arm (see below).
If the ODA-DC is directly responsible for binding the outer arm to the doublet microtubule, then it should interact with the outer arm. When the outer dynein arm was removed from the wild-type axoneme by extraction with high salt and subjected to sucrose density gradient centrifugation in the presence of Mg2+, the α, β, and γ DHCs remained associated with each other as a three-headed particle that sediments at 23S (Takada et al., 1992 ). We found that under these conditions the three polypeptides of the ODA-DC remained associated with the outer arm polypeptides and cosedimented with them at 23S. Therefore, the ODA-DC interacts directly with the outer arm. This interaction was confirmed by experiments in which the outer dynein arm intermediate chains were found to coimmunoprecipitate with the ODA-DC from high-salt extracts prepared in the presence of Mg2+. The association of the ODA-DC with the three-headed dynein undoubtedly explains the ability of this dynein to bind efficiently to oda1 and oda3 axonemes, which lack the ODA-DC (Takada et al., 1992 ). Because the entire outer arm dynein appears to be able to assemble in the cytoplasm of oda1 and oda3 mutants (Fowkes and Mitchell, 1998 ), the ODA-DC is not promoting outer arm assembly simply by binding the αβ and γ subunits together. More likely, it forms a direct link between the outer dynein arm and the outer doublet microtubule.
Quantitative densitometry indicated that the Mr ~70,000 and ~105,000 ODA-DC proteins are present in an equimolar ratio in a 23S particle consisting of the ODA-DC and the outer dynein arm; visual comparison of band intensities suggested that the Mr ~25,000 ODA-DC protein also is present in equimolar amounts with the two larger ODA-DC subunits. Therefore, it is likely that the three ODA-DC subunits occur in a 1:1:1 molar ratio. These data alone do not distinguish whether the ODA-DC is a heterotrimer or a higher order polymer, such as a hexamer. However, we also found that there was one copy of each ODA-DC polypeptide per soluble three-headed outer arm dynein. Therefore, the simplest model is that the ODA-DC is a heterotrimer, and that each outer arm is associated with a single ODA-DC.
Beginning with peptide sequence obtained directly from microsequencing of tryptic fragments from the Mr ~70,000 ODA-DC polypeptide, we cloned and sequenced a full-length cDNA encoding the protein. The sequence predicts a novel 62-kDa polypeptide with long regions that have a very high probability of forming coiled-coil structures, a distinctive α-helical region that probably does not form a coiled-coil, and a C terminus that is highly charged. These structural features are generally similar to those of the Mr ~105,000 ODA-DC protein (Koutoulis et al., 1997 ) with which the Mr ~70,000 protein is associated.
The Mr ~70,000 protein has three regions (~34, ~96, and ~48 amino acids in length) totaling 178 amino acids that are >99% likely to form a coiled-coil structure. The Mr ~105,000 ODA-DC protein likewise has three regions (~120, ~70, and ~36 amino acids) totaling ~236 amino acids that are >99% likely to form a coiled-coil. Coiled-coils commonly function in dimerization and heterodimer formation (Lupas, 1996b ). As discussed above, each ODA-DC is likely to contain just one copy of each of its component polypeptides. Moreover, experiments using the zero-length protein cross-linker EDC revealed that the Mr ~70,000 protein is in direct contact with the Mr ~105,000 protein, but provided no evidence that either protein interacts with itself. Therefore, we postulate that these two proteins interact via their coiled-coil domains to form a heterodimer, and that this heterodimer has an extended rod-like structure (Figure (Figure6A).6A). Assuming 1.5 Å/residue in a coiled-coil (Fraser and MacRae, 1973 ), a rod-like structure based on the presumptive coiled-coil domains of the Mr ~70,000 protein could be up to 26.7 nm in length. This is long enough that ODA-DCs could link up end-to-end, with some overlap, to form a filament with a 24-nm repeat structure (Figure (Figure6B;6B; see below).
Both the Mr ~70,000 and ~105,000 proteins also contain a highly charged domain, following the second coiled-coil region, that is predicted to form an α-helix but does not have a heptad repeat and thus is not likely to form a coiled-coil structure (cf. region C, Figure Figure2B2B and region C, Figure 6 of Koutoulis et al., 1997 ). A portion of this region is 42% identical in the two proteins:
In the Mr ~105,000 protein this sequence consists of an imperfect 11-amino acid tandem repeat (Koutoulis et al., 1997 ) that previously was found to be closely related to repeats in mammalian trichohyalin, a protein that interacts with intermediate filaments (Fietz et al., 1993 ; Lee et al., 1993 ). In trichohyalin these repeats are proposed to form a single-stranded α-helical rod that is stabilized by ionic interactions between successive turns of the helix (Lee et al., 1993 ); it has been proposed that these repeats interact with charged residues on intermediate filaments. Weak repeats with similar structure and composition also occur after a long coiled-coil region in troponin T and caldesmon (e.g., K512-Q527 in chicken caldesmon; Bryan et al., 1989 ); these regions are candidates for the tropomyosin-binding sites of these proteins (Bryan et al., 1989 ). Therefore, this part of the Mr ~70,000 protein may be involved in protein–protein interactions.
The carboxyl-terminal 59 amino acids of the Mr ~70,000 protein contain a short glutamic acid-rich region followed by a region with a high percentage of positively and negatively charged residues. Both tubulin and IC78 of outer arm dynein have highly charged basic and acidic domains, so it is possible that this portion of the Mr ~70,000 protein interacts with those proteins. The Mr ~105,000 protein likewise has a glutamic acid cluster near its carboxy terminus.
The size of the Mr ~70,000 protein estimated by SDS-PAGE is greater than the 62.5-kDa mass predicted from nucleotide sequence. Disparities in this direction and of this magnitude or more are commonly observed in proteins with similar structure (e.g., chicken caldesmon, Mr 120,000–150,000 estimated vs. 87.0 kDa predicted [Bryan et al., 1989 ]; C. reinhardtii Mr ~105,000 ODA-DC protein, Mr ~105,000 estimated vs. 83.4 kDa predicted [Koutoulis et al., 1997 ]).
Outer dynein armless mutants having defects at 15 different ODA loci (ODA1-ODA15) have been reported, of which 12 (oda1-oda10, oda12, and oda15) lack the complete outer arm and have a slow swimming phenotype (Kamiya, 1988 ; Koutoulis et al., 1997 ; King, 2000 ; Pazour and Witman, 2000 ). The genes ODA2, ODA4, ODA6, ODA9, ODA11, ODA12, ODA13, and ODA15 encode the γ and β DHCs, IC69, IC78, the α DHC, LC2, LC6, and LC7, respectively, of the outer arm dynein (Mitchell and Kang, 1991 ; Sakakibara et al., 1991 , 1993 ; Wilkerson et al., 1994 , 1995 ; Pazour and Witman, 2000 ). A sixth ODA gene (ODA3) has been cloned and found to encode the Mr ~105,000 ODA-DC protein (Koutoulis et al., 1997 ). We now report that the ODA1 gene encodes the Mr ~70,000 ODA-DC protein. Sequencing of mutant oda1-1 cDNA encoding the Mr ~70,000 protein revealed that the mutant DNA has a point mutation that converts codon 46, which in wild-type DNA specifies a glutamine, to a stop codon. This mutation thus precludes production of the full-length Mr ~70,000 protein in oda1-1 cells, a defect that undoubtedly is directly responsible for the loss of the ODA-DC in these cells (Takada and Kamiya, 1994 ). This mutant should be very useful for studies to investigate the detailed roles of the Mr ~70,000 protein in outer arm assembly and ODA-DC function.
The fact that a defect in the Mr ~70,000 protein leads to loss of the ODA-DC in oda1-1 indicates that the Mr ~70,000 protein is essential for assembly of the ODA-DC, and confirms that loss of the ODA-DC leads to an inability to target outer dynein arms to doublet microtubules. In contrast, loss of a major outer arm dynein structural protein in oda2, oda4, oda6, and oda9 results in failure of the outer arm to assemble, but does not lead to loss of the ODA-DC (Takada and Kamiya, 1994 ; Koutoulis et al., 1997 ). Therefore, the ODA-DC can assemble and bind to the correct position on flagellar doublet microtubules independently of the outer arm dynein.
How the ODA-DC itself assembles onto the correct site on the doublet microtubules is not known. One possibility is that the ODA-DC is a rod-like structure that links together with adjacent ODA-DCs to form a filament running longitudinally along the doublet (Figure (Figure6B);6B); assembly of this filament might be initiated on the correct microtubule protofilament by some structure at the base of the axoneme, or by some discontinuity in the tubulin lattice. This model is consistent with the coiled-coil nature of the Mr ~70,000 and ~105,000 proteins, with the direct interactions of these proteins as revealed by protein cross-linking experiments, and with the appearance of the ODA-DC in electron micrographs of cross sections of axonemes (Takada and Kamiya, 1994 ). Inasmuch as we observed one ODA-DC per outer arm dynein, the ODA-DCs would be expected to repeat at 24-nm intervals along the filament. Indeed, independent evidence for such a spacing has been provided by immunogold labeling of the Mr ~105,000 ODA-DC polypeptide on outer doublet microtubules of demembranated axonemes (Wakabayashi et al., 2001 ). Therefore, the ODA-DC could act as a ruler to determine the 24-nm longitudinal spacing of the outer arms, although it is not known whether such a ruler is necessary in Chlamydomonas. Chlamydomonas outer arm dynein has been shown to assemble onto brain microtubules with a 24-nm periodicity (Haimo and Fenton, 1988 ), but it is not clear if the ODA-DC was present in the dynein preparations used for those experiments. Further studies will be necessary to determine whether purified Chlamydomonas outer arm dynein requires the ODA-DC for periodic assembly onto microtubules.
Takada and Kamiya (1997) reported that the ODA-DC has an important role in regulating the differential beat frequency of the cis- versus the trans-axoneme in demembranated, reactivated cell models of Chlamydomonas. These results indicate that the ODA-DC might regulate the activity of outer arm dynein. A potential mechanism for regulating the activity of a molecular motor is protein phosphorylation. Luck and Piperno (1989) reported that an Mr ~73,000 phosphoprotein was missing from axonemes of oda1 and oda3 but not oda2, oda4, oda5, or oda6; this protein presumably is the same as the Mr ~70,000 ODA-DC polypeptide. The protein had multiple isoelectric variants in two-dimensional isoelectric focusing/SDS-PAGE, suggesting that it was phosphorylated at multiple sites. We have similarly observed that the Mr ~70,000 ODA-DC protein in axonemal extracts and purified three-headed dynein focuses in two-dimensional gels as multiple spots having different isoelectric points (Takada and Kamiya, unpublished data). Therefore, the Mr ~70,000 protein probably is phosphorylated at multiple sites in vivo. Consistent with this, the sequence reported here has numerous potential sites for protein phosphorylation.
Our current findings will greatly facilitate studies to investigate the role of ODA-DC phosphorylation in regulating axonemal beat frequency. For example, knowledge of the sequence of the Mr ~70,000 protein will now permit identification of those peptides and residues that are phosphorylated in vivo. Moreover, it will be possible to alter those sites by site-directed mutagenesis of the ODA1 gene, and then determine the effect of the modifications by transforming (Kindle, 1990 ) the altered genes back into strains containing the oda1-1 allele. These strains would be rescued for assembly of the ODA-DC and the outer dynein arm but would be defective in phosphorylation of the targeted site.
A BLAST search of gene and protein databases revealed potential homologs of the Chlamydomonas Mr ~70,000 ODA-DC polypeptide in the protozoan Leishmania, in Drosophila, and in mice and humans, but not in organisms such as yeast, C. elegans, and Arabidopsis that lack motile cilia. This is the first evidence that the ODA-DC occurs in higher organisms. In humans, primary ciliary dyskinesia (PCD), an inherited disorder in which ciliary and flagellar movement is impaired, seems most frequently to be caused by loss of the outer dynein arms (Afzelius and Mossberg, 1995 ). As a result of this defect, PCD patients develop bronchiectasis and chronic sinusitis; male patients are infertile. Inasmuch as the Mr ~70,000 ODA-DC gene is necessary for outer arm assembly in Chlamydomonas, its potential human homolog AK057357 is now a candidate gene for those cases of PCD in which the outer dynein arms are missing. AK057357 maps to chromosome 19q13. Interestingly, in some PCD patients lacking the outer dynein arm, the defective locus has been mapped to this same region (Meeks et al., 2000 ). Further studies are warranted to determine whether a defect in AK057357 causes PCD.
We are grateful to Dr. John Leszyk of the UMMS Protein Microsequencing Facility for peptide sequencing, and to Kazuaki Homma for expert assistance with the protein cross-linking experiments. This work was supported by a National Institutes of Health grant (GM-30626) (to G.W.), by the Robert W. Booth Fund at the Greater Worcester Community Foundation (to G.W.), by a fellowship from the Japan Society for the Promotion of Science (to S.T.), and by a grant from the Ministry of Education, Science and Culture of Japan (to R.K.).
Article published online ahead of print. Mol. Biol. Cell 10.1091/mbc.01–04–0201. Article and publication date are at www.molbiolcell.org/cgi/doi/10.1091/mbc.01–04–0201.
1The three oda strains previously were referred to as oda1 strain 38, oda3 strain 73, and oda6 strain 95, respectively (Kamiya, 1988 ).