|Home | About | Journals | Submit | Contact Us | Français|
Of the uncloned ODA genes required for outer dynein arm assembly in Chlamydomonas, ODA5 and ODA10 are of particular interest because they do not encode known subunits of the outer arm or the outer dynein arm-docking complex (ODA-DC), and because genetic studies suggest their products interact. Beginning with a tagged oda5 allele, we isolated genomic and cDNA clones of the wild-type gene. ODA5 predicts a novel, 66-kDa coiled-coil protein. Immunoblotting indicates Oda5p is an axonemal component that assembles onto the axoneme independently of the outer arm and ODA-DC and is uniquely missing in oda5 and oda10 axonemes. Oda5p is released from the axoneme by extraction with 0.6 M KCl, but the soluble Oda5p does not cosediment with the outer dynein arm/ODA-DC in sucrose gradients. Quantitative mass spectrometry by using isotope coded affinity tagging revealed that a previously unidentified adenylate kinase is reduced 35–50% in oda5 flagella. Direct enzymatic assays demonstrated a comparable reduction in adenylate kinase activity in oda5 flagella, and also in oda10 flagella, but not in flagella of other oda mutants. We propose that Oda5p is part of a novel axonemal complex that is required for outer arm assembly and anchors adenylate kinase in proximity to the arm.
Dyneins are large, multisubunit microtubule motors that are involved in many types of cellular movements, including vesicle transport, nuclear migration, spindle formation and orientation, chromosome movements, and beating of cilia and flagella. Eukaryotic flagella contain three major classes of dynein: dynein 1b/2, the retrograde motor for intraflagellar transport (Pazour et al., 1999a ; Porter et al., 1999 ); the heterogeneous inner arm dynein system, containing up to seven isoforms (Porter and Sale, 2000 ); and the outer arm dynein (Witman et al., 1994 ), of which there is only one known isoform. The outer dynein arms provide up to four-fifths of the power for flagellar movement (Brokaw, 1994 ) and are required for normal flagellar beat frequency and swimming speed. Failure to assemble the outer arms leads to poorly motile cilia and flagella in humans, and is a common cause of the inherited disease primary ciliary dyskinesia (Afzelius and Mossberg, 1995 ). Hence, the identification of outer dynein arm components and the mechanisms that mediate their assembly are of considerable interest.
The outer and inner dynein arms assemble independently and are located at specific sites within the axoneme. Inherent to this assembly process is the requirement for unique structural or biochemical landmarks that ensure the proper targeting of each dynein isoform to its correct site both around and along the length of the axoneme. The most extensively investigated paradigm for axonemal dynein assembly and targeting is the Chlamydomonas reinhardtii outer dynein arm (Pazour and Witman, 2000 ). This arm consists of at least 13 polypeptides, which include three heavy chains (HCs), two intermediate chains (ICs), and several light chains (LC1–8) (DiBella and King, 2001 ), and repeats at 24-nm intervals along the A-tubules of the flagellar outer doublet microtubules. The correct positioning of the outer arm is due at least in part to its association with the outer dynein arm-docking complex (ODA-DC), which is required for attachment of the outer arm onto the A-tubule. The ODA-DC, a heterotrimeric complex comprised of subunits DC1 (Koutoulis et al., 1997 ), DC2 (Takada et al., 2002 ), and DC3 (Casey et al., 2003 ), assembles onto the site normally occupied by the outer arm even in the absence of the latter structure.
Loss of the outer arm in Chlamydomonas results in a characteristic slow, jerky swimming phenotype, and this has been used to isolate “oda” mutants unable to assemble the outer dynein arm (Kamiya, 1988 ). To date, 16 ODA genes have been identified (Table 1), most of which have been cloned and determined to encode proteins of the outer dynein arm or ODA-DC. Of the uncloned ODA genes, ODA5, ODA8, and ODA10 are of particular interest for two reasons. First, although they are required for outer arm assembly, they do not encode known subunits of the outer arm or the ODA-DC (King, Pazour, and Witman, unpublished data). Second, complementation assays in temporary dikaryons suggest that the products of these three genes interact to form a complex (Kamiya, 1988 ). Thus, these genes may encode subunits of an unidentified component important for outer arm assembly.
Key steps to understanding such a complex will be to clone the ODA5, ODA8, and ODA10 genes and characterize their products. To that end, we used insertional mutagenesis to generate a tagged oda5 allele, which was then used to isolate the wild-type gene. ODA5 encodes a novel axonemal protein that assembles independently of the outer arm and ODA-DC and is missing in oda5 and oda10 axonemes but not in axonemes of other oda mutants. The absence of Oda5p is correlated with a reduction in the level of a previously uncharacterized flagellar adenylate kinase (AK). The results suggest that Oda5p is part of a complex that includes the flagellar AK. Therefore, the Oda5p-containing complex is not only essential for outer arm assembly but also probably anchors AK in proximity to the arm, ensuring that both high-energy phosphate bonds of ATP can be efficiently utilized at the axoneme's major site of power production.
Chlamydomonas reinhardtii strains used in this study include the following: CC-2454 (cw15, nit1-305, mt-), CC-48 (arg2, mt+), CC124 (nit1-137, nit2–137, mt-), CC-2229 (oda1, mt+), CC2233 (oda3-1, nit1, nit2, AC17, mt-), and 137C (nit1–137, nit2-137, mt+), all from the Chlamydomonas Genetics Center (Department of Biology (Duke University, Durham, NC); oda5+ (oda5-1, mt+), oda3+ (oda3-1, mt+), oda8+ (oda8, mt+), oda9+ (oda9, mt+), oda10+ (oda10-1, mt+) (Kamiya, 1988 ), V87.2 (oda10-2, nit1::NIT1, NIT2, agg1, mt-) (Koutoulis et al., 1997 ), oda9-V5 (oda9, mt+) (Wilkerson et al., 1995 ), and 45BO3 (oda5-2, cw15, nit1-305::NIT1, mt-), an insertional allele of oda5.
45BO3 was crossed to 137C to create strain 88b (oda5-2, NIT1, mt-). 88b was crossed to CC-48 to create strain 112b (oda5-2, arg2, mt-). 112b.76, 112b.150, 112b.219, and 112b.220 (oda5-2::ODA5, arg2, mt-) were created by transformation of the ODA5 gene into strain 112b. Strain 112b.221.4 (oda5-2::HA-ODA5, arg2, mt-) was created by transformation of a hemagglutinin (HA)-tagged ODA5 gene into strain 112b.
Chlamydomonas cells were grown in a 14:10 light:dark cycle in the following media: medium I of Sager and Granick (1953 ) modified to contain three times the original amount of phosphate (Witman, 1986 ); R-medium (medium I supplemented with 0.1% sodium acetate); R+Arg (R-medium supplemented with 50 μg/ml arginine); M-N (medium I without nitrogen); TAP (Harris, 1989 ); TAP+Arg (TAP-medium supplemented with 50 μg/ml arginine); and SGII/NO3 [medium II of Sager and Granick (1953 ) modified to contain 0.003 M KNO3 as the nitrogen source].
All transformations were done using the glass bead method as described previously (Kindle et al., 1989 ; Koutoulis et al., 1997 ). The insertional mutant 45BO3 was generated by transforming CC-2454 cells with plasmid pMN24 (Fernandez et al., 1989 ) containing the Chlamydomonas NIT1 gene. Transformants positive for NIT1 were selected on SGII/NO3 media. Motility mutants were identified by growing positive transformants in liquid culture and screening by light microscopy. Cotransformations were performed using ODA5 genomic constructs and pARG7.8 plasmid (Debuchy et al., 1989 ); transformants were selected on TAP plates.
Whole cells and isolated axonemes were processed as described previously (Hoops and Witman, 1983 ). Samples were embedded in a mixture of LX112/Araldite 502 epoxy resin and sectioned at 50–70 nm.
Chlamydomonas RNA was obtained before deflagellation and 30–45 min after deflagellation by pH shock (Witman et al., 1972 ). Total RNA was isolated by LiCl precipitation (Wilkerson et al., 1994 ) and polyA+ mRNA selected using Oligo dT cellulose (Ambion, Austin, TX). Polyadenylated RNA was separated on 1% formaldehyde agarose gels and transferred to Duralon-UV (Stratagene, La Jolla, CA). RNA was cross-linked to the membranes by using a Stratalinker (Stratagene). Chlamydomonas genomic DNA isolations were performed as described previously (Koutoulis et al., 1997 ). Genomic DNA was separated on 0.8% agarose gels. Transfer to Duralon-UV and cross-linking were the same as for RNA. Hybridization probes were generated by random prime labeling by using the Prime-It II kit (Stratagene).
Matings and temporary dikaryon, stable diploid, and tetrad analyses were performed according to standard procedures (Kamiya, 1988 ; Harris, 1989 ; Dutcher, 1995 ). The ODA5 allele was confirmed by crossing 45B03 with the oda5-1 strain. Progeny were scored for motility (Oda +/-) by light microscopy and for Arg+/- by comparing growth on TAP and TAP+Arg plates.
Swimming speed was determined as described in Kamiya (1988 ). The swimming velocity of 30 cells was used to determine the average swimming speed for each strain analyzed. Flagellar beat frequency was determined as described previously (Kamiya, 2000 ).
A probe, 36.1, to unknown sequence adjacent to the NIT1 insertion in oda5-2 was used to identify wild-type BAC clones (Clemson University Genome Institute, Clemson, SC) containing the unknown sequence. The BAC clones were tested for their ability to rescue the Oda5- phenotype by cotransforming strain 112b with the BAC clones and plasmid pARG7.8 (Debuchy et al., 1989 ). Transformants were scored for Oda+/- phenotype by light microscopy. The smallest rescuing fragment, 50.1, from one BAC clone was sequenced.
Primers to predicted exons were generated (Integrated DNA Technologies, Coralville, IA) and polymerase chain reaction (PCR) used to amplify fragments from a cDNA library constructed from polyA+ mRNA isolated 30′ postdeflagellation (Wilkerson et al., 1994 ) or a gametic cDNA library (gift of William Snell, University of Texas Southwestern Medical Center, Dallas, TX). cDNA clones were sequenced to confirm intron-exon boundaries. Primers designed within ODA5 introns amplified the exons from oda5-1 genomic DNA. Sequencing of three independent clones identified the oda5-1 mutation. All sequencing was performed by either the Iowa State DNA Sequencing Facility or the University of Massachusetts Medical School Nucleic Acid Facility. See online supplemental materials regarding the cloning of sequence flanking the oda5-2 insertion site and for construction of an HA-tagged ODA5 gene construct.
Sequence assemblies were performed using LaserGene SeqMan (DNAstar, Madison, WI). The LaserGene EditSeq module was used for translations and to obtain the theoretical isoelectric point and mass of the predicted Oda5 protein. Primer design was performed using PRIMER3 at www.genome.wi.mit.edu/cgi-bin/primer/primer3 (Rozen and Skaletsky, 2000 ). The BLAST server at www.ncbi.nlm.nih.gov/BLAST (Altschul et al., 1990 ) was used to search for homologous sequences. The COILS (www.ch.embnet.org/software/COILS_form.html) and PairCoil (http://paircoil.lcs.mit.edu/cgi-bin/paircoil) servers were used to predict coiled-coil regions in the Oda5 protein (Lupas et al., 1991 ; Berger et al., 1995 ). To predict regions of coding potential, the ODA5 and AK genomic sequences were analyzed using the GreenGenie gene prediction program (http://www.cse.ucsc.edu/%7Edkulp/cgi-bin/greenGenie) (Li et al., 2003 ). The AK protein was examined using the Joint Genome Institute (Walnut Creek, CA; JGI) Chlamydomonas version 2.0 genome database (http://shake.jgi-psf.org/chlre2/chlre2.home.html), and the PROSITE protein families and domains database (http://us.expasy.org/prosite/). The TreeTop-phylogenetic tree prediction program (http://www.genebee.msu.su/services/phtree_reduced.html) was used for comparisons of the C. reinhardtii flagellar AK versus Homo sapiens AK sequences.
A cDNA encoding an NH2-terminal fragment of Oda5p was subcloned into the pMAL vector (Invitrogen, Carlsbad, CA) to create a construct containing the first 154 amino acids of Oda5p fused to the maltose binding protein. This fusion protein was bacterially expressed, purified, and used to immunize rabbits for polyclonal antibody production (Invitrogen). Affinity purification was performed using an Oda5-GST fusion protein containing the same NH2-terminal fragment of Oda5.
Flagella were isolated by the method of Witman (1986 ) and extracted with 1% Nonidet P-40 (Calbiochem, La Jolla, CA) or 1% Tergitol Type NP-40 (Sigma-Aldrich, St. Louis, MO) in HMDEKP (30 mM HEPES, pH 7.4, 5 mM MgSO4, 1 mM dithiothreitol, 0.5 mM EGTA, 25 mM KCl, and 1 mM phenylmethylsulfonyl fluoride) as indicated. Demembranated axonemes were subsequently extracted with 0.6 M KCl in HMDEKP buffer. High-salt extracts were fractionated on 5–20% sucrose gradients under conditions that maintain the association of the outer arm and the ODA-DC (Takada et al., 2002 ). A mixture of bovine thyroglobulin, catalase, and bovine serum albumin (Sigma-Aldrich) was run in a parallel gradient for S-value determination.
SDS-PAGE and Western blots were performed according to standard methods. Protein extracts from intact cells were prepared by centrifuging cells and resuspending the cell pellets in sample buffer (10 mM Tris, pH 8.0, 32 mM dithiothreitol, 1 mM EDTA, 10% sucrose, and 1% SDS). Samples were heated for 10 min and sheared with a 22-gauge needle. Axonemal fractions were prepared as described above and dissolved in sample buffer. Proteins were separated by SDS-PAGE and transferred to Immobilon-P membranes (Millipore, Billerica, MA). Oda5p was localized using the anti-Oda5p antibody diluted 1:1000 in 5% horse serum in 1× Tris-buffered saline/0.5% Tween 20. IC2, the γDHC, DC2, and IC140 were revealed using monoclonal antibodies 1869A and 12γB (King et al., 1985 ), an anti-DC62 polyclonal antibody (Wakabayashi et al., 2001 ), and an anti-IC140 polyclonal antibody (Yang and Sale, 1998 ), respectively. Horseradish peroxidase-conjugated secondary antibodies (Pierce Chemical, Rockford, IL; Sigma-Aldrich) were used at 1:2000.
ICAT was performed at the University of Victoria GenomeBC Proteomics Centre (Vancouver, BC, Canada). Briefly, pellets of Tergitol-treated flagella from wild-type and oda5-1 were resuspended in 6 M urea, 0.1% SDS and then labeled using the ICAT reagent kit (Applied Biosystems, Foster City, CA). The samples were combined, digested with trypsin, and the resulting peptide mixture fractionated into four or 10 fractions. The peptides were then affinity purified using a streptavidin column. The results are the pooled analyses of both the four- and 10-fraction experiments. Analysis was performed on an Applied Biosystems/MDS QStar hybrid liquid chromatography/tandem mass spectrometry (MS/MS) quadripole time of flight system and quantitation performed using Applied Biosystems ProICAT software.
AK activity was determined by the method of Watanabe and Flavin (1976 ) with slight modifications. AK was assayed by coupling the formation of ATP from ADP to NADP+ reduction in the presence of hexokinase and glucose-6-phosphate dehydrogenase. The reaction mixture consisted of 55 mM Tris, pH 7.9, 40 mM glucose, 2 mM MgCl2, 1 mM ADP, 0.18 mM NADP+, 1 U each of hexokinase and glucose-6-phosphate dehydrogenase, and 1 mM sodium-(meta)vanadate to inhibit dynein ATPases. The reaction mixture was preincubated for 10 min to consume any ATP contaminating the ADP. Flagellar fractions were added to the reaction and the adenylate kinase activity measured by monitoring the change in absorbance at 340 nm that accompanied the production of NADPH. Data points were collected every 30 s for 10 min. Assays were performed in triplicate on three independently isolated flagellar samples.
To obtain insertional mutants with defects in motility, CC-2454 cells were transformed with plasmid pMN24 (Fernandez et al., 1989 ). Oda- phenotypes were distinguished from other motility mutants by their slow, jerky swimming motion that is characteristic of loss of the outer dynein arm. The new oda mutants were then identified by crossing them to known oda mutants. When transformant 45BO3 was crossed with the original oda5-1 mutant (Kamiya, 1988 ) and tetrads dissected, no progeny showed wild-type motility (PD: NPD:T, 37:0:0). Moreover, temporary dikaryons between oda5-1 and 45B03 did not undergo any increase in motility during 2 h of mating. Finally, oda5-1 and 45B03 did not complement in stable diploids. Thus, it is concluded that 45B03 is an insertional allele of the ODA5 gene. The 45B03 mutation hereafter will be referred to as oda5-2.
DNA flanking the integrated selectable marker in the oda5-2 strain (Figure S1) was cloned and used as a probe to identify four BACs containing that DNA. Restriction mapping indicated these BAC clones had overlaping inserts (our unpublished data). To test whether these BAC clones contained the ODA5 gene, we transformed oda5-2 with the BACs and screened for rescue of the Oda- phenotype. One BAC clone rescued the motility defect in oda5-2, suggesting it contained the ODA5 gene. This BAC contained an insert of ~40 kb. To delimit the ODA5 gene within this BAC, we subcloned smaller fragments from this insert and tested them for rescue. The smallest rescuing genomic fragment, 50.1, is 6.1 kb (Figure 1A) and rescued 22 of 72 colonies that had been cotransformed with the ARG7 selectable marker.
Motility assays of wild-type, oda5-2, and a strain rescued with the 50.1 genomic fragment revealed that both swimming speed and flagellar beat frequency were restored to near wild-type levels in the rescued strain (Figure 1B). The swimming speed of the rescued strain (133 ± 24 μm/s) was slightly slower than that of wild-type (172 ± 13 μm/s), but clearly rescued compared with the oda5-2 mutant (55 ± 9 μm/s). The flagellar beat frequency was restored completely (61 Hz for wild-type vs. 60 Hz for the rescued strain). To further verify the rescued strains, we performed Southern blot analysis using a probe to the rescuing fragment. In wild-type, the probe hybridized to a band of the expected size, whereas in the oda5-2 strain, the probe did not hybridize to any band, indicating this region is deleted in oda5-2 (Figure 1C). The probe also hybridized to bands in the rescued strains, demonstrating that the deleted sequences have been restored (Figure 1C). Probes to either end of the 9-kb rescuing fragment (Figure 1A) indicate this entire region is deleted in the mutant (our unpublished data).
Electron microscopy demonstrated that the outer arms are missing in oda5-2 and restored in the rescued strains. Figure 2 shows cross sections through the flagella of wild-type, the oda5-2 insertional mutant, and one of the rescued strains. Wild-type and the rescued strain show the full complement of outer arms (Figure 2, A and B). The oda5-2 strain specifically lacks these structures (Figure 2, C and D), although residual material was occasionally observed at the site normally occupied by the outer arm (Figure 2D). These data demonstrate that the 50.1 rescuing fragment contains the ODA5 gene.
The rescuing genomic fragment was completely sequenced. To identify potential coding regions, we analyzed the sequence using the GreenGenie gene prediction program (Li et al., 2003 ) (Figure 1A). PCR primers designed from the predicted exons were used to amplify the ODA5 cDNA from wild-type Chlamydomonas cDNA libraries. Analysis of the cDNA revealed a large open reading frame (Figure 3A). The cDNA sequence spans the entire rescuing fragment, and no other genes are found in this region (Figure 1A). Note that fragments 51.1, 52.2, 55.1, and 59.1, which contain only portions of the ODA5 gene, do not rescue oda5-2, indicating that the gene is disrupted in these smaller fragments (Figure 1A).
The ODA5 cDNA predicts a 652-amino acid protein (Figure 3A) with a mass of ~66-kDa and a pI of 5.2. Oda5p is predicted to contain several coiled-coil domains and a COOH-terminal noncoiled-coil region (Figure 3B). BLAST reports do not reveal significantly related sequences, suggesting that Oda5p is a novel protein.
To ascertain whether the oda5-1 mutant contains a defect in the ODA5 gene, the exons from oda5-1 genomic DNA were sequenced. A double-base pair replacement was identified 321 base pairs downstream of the initiating ATG. This mutation converts a GTA codon specifying a valine to the stop codon TAA (Figure 3A), which would result in translation of only 108 of the 652 amino acids encoded by the wild-type ODA5 gene. This result further verifies that the gene we have identified is the ODA5 gene and that the insertional mutant is an allele of oda5-1.
Transcription of genes encoding flagellar proteins is up-regulated in response to deflagellation (Silflow et al., 1982 ). To determine whether this is the case for ODA5, we performed Northern blot analysis on RNA isolated from wild-type cells either before deflagellation, or 30–45′ postdeflagellation. Using a probe to the antisense strand of the ODA5 cDNA, we found that the ODA5 transcript is a ~2.7-kb mRNA that is up-regulated during flagellar regeneration (Figure 4A). This result suggests that ODA5 encodes a flagellar protein.
To facilitate the in vivo localization of the Oda5 protein, we generated a polyclonal antibody to an Oda5 fusion protein. In wild-type whole cells, the Oda5p antibody detected a band having a Mr of 76,000 on SDS-PAGE (Figure 5A). This band was not detected in the null strain, confirming our antibody recognized the correct protein. Although the antibody slightly cross-reacted with other proteins, we were readily able to follow the distribution of Oda5p by comparing fractions from wild type and the oda5-2 null strain. When cells were deflagellated and cell bodies analyzed, we detected little to no Oda5p in the cell body fraction. Oda5p was highly enriched in isolated whole flagella, demonstrating that Oda5p is a bona fide flagellar protein (Figure 5A).
To further localize Oda5p within the flagellum, we isolated whole flagella and extracted them with Nonidet P-40 detergent followed by 0.6 M KCl (Figure 5B). Oda5p is not present in the detergent-soluble membrane + matrix fraction, but it remains associated with the demembranated axoneme, demonstrating that Oda5p is an axonemal component and not a soluble or membrane-associated protein. When we subsequently extracted the demembranated axonemes with 0.6 M KCl, Oda5p was released from the axoneme.
Previous data have shown that the outer arm and the ODA-DC also are released from the axoneme by 0.6 M KCl (Pfister et al., 1982 ; Takada et al., 1992 ; Takada et al., 2002 ). To determine whether Oda5p associates with the outer arm or the ODA-DC, we subjected the high-salt extract to sucrose gradient sedimentation, by using conditions designed to maintain the outer arm/ODA-DC association (Takada et al., 1992 ). Western blots of sucrose gradient fractions were probed with antibodies to IC2 and the γDHC, which confirmed the migration of the outer arm/ODA-DC at the expected position of ~23S (Figure 5C). In contrast, Oda5p migrated at ~5S in these gradients. These data demonstrate that under conditions that remove the outer arm and the ODA-DC as an intact complex, Oda5p is not associated with these two components.
Inasmuch as Oda5p behaved independently of the outer arm and ODA-DC in sucrose gradients, we investigated whether Oda5p can assemble onto the axoneme in the absence of the latter structures. Western blots (Figure 6A) revealed that Oda5p is present in axonemes of an oda9 mutant, which is defective in the IC1 gene and fails to assemble an outer arm (Wilkerson et al., 1995 ), and of oda1 and oda3 mutants, which are defective in the DC2 and DC1 components of the docking complex, respectively (Koutoulis et al., 1997 ; Takada et al., 2002 ) and fail to assemble the ODA-DC and the outer arm. These results show that Oda5p can assemble onto the axoneme independently of the outer arm and the ODA-DC.
Because Oda5p, Oda8p, and Oda10p have been proposed to interact, we investigated whether localization of Oda5p is disrupted in the oda8 and oda10 mutant strains. Western blots revealed that Oda5p does assemble onto oda8 axonemes. In contrast, Oda5p fails to assemble onto axonemes of the oda10 mutant (Figure 6A), demonstrating that a functional Oda10 protein is required for proper localization of Oda5p. This provides the first biochemical evidence for an interaction between Oda5p and Oda10p.
To determine whether the ODA-DC can assemble onto the axoneme in the absence of Oda5p, as well as in the absence of ODA8 and ODA10 gene products, we probed Western blots of isolated axonemes by using an antibody to DC2, the 62-kDa component of the ODA-DC (Figure 6B). As expected, axonemes from oda1 and oda3, which do not assemble the ODA-DC, lack DC2, whereas axonemes from oda9, which are missing only the outer dynein arm, contain DC2 as do wild-type axonemes. Oda5, oda8, and oda10 axonemes also contain DC2. Therefore, DC2 can assemble onto the axoneme in these three mutants, suggesting that the entire ODA-DC can assemble onto axonemes independently of Oda5, Oda8, and Oda10 proteins.
In an effort to identify proteins that interact with Oda5p, we used a quantitative mass spectrometry technique called ICAT (Han et al., 2001 ) to compare proteins present in wild-type versus oda5 flagellar fractions. ICAT allows one to determine the ratios of individual proteins in the two fractions being compared.
For the ICAT experiments, we wanted to analyze as comprehensive a set of flagellar proteins as possible. However, subassemblies of the outer dynein arm are known to remain in the cytoplasm of mutants unable to assemble the arm due to loss of an outer arm protein (Fowkes and Mitchell, 1998 ), and preassembled axonemal complexes are present in the flagellar matrix, even in nonregenerating flagella (Qin et al., 2004 ). Because the presence of any stable but unassembled Oda5-interacting proteins would compromise the ICAT analysis, we selectively removed the matrix proteins by treating the isolated flagella with the nonionic surfactant Tergitol Type NP-40, which, in contrast to Nonidet, disrupts but does not completely dissolve the flagellar membrane (Figure 7A). The axonemes and residual membrane were then collected by centrifugation, leaving the matrix in the supernatant. The resulting pellets were then washed, solubilized in SDS and urea, and labeled with the isotopically light (wild type) and isotopically heavy (oda5) forms of the ICAT reagent, which specifically labels cysteines. The labeled samples were combined, digested with trypsin, fractionated by cation exchange chromatography, and the labeled peptides purified by affinity chromatography. The purified, labeled peptides were analyzed by electrospray ionization MS/MS. The ratio of the isotopically heavy (oda5-1) to isotopically light (wild type) peptide in each peptide pair was determined from the signal intensities of the peaks as the pair eluted into the mass spectrometer. Peptide sequences were identified from the MS/MS spectra by searching the Chlamydomonas 20021010 expressed sequence tag (EST) database, or the BLAST nonredundant database with Chlamydomonas specified as the organism. Peptide and EST sequences were then used to search version 1.0 of the JGI Chlamydomonas genome database (http://genome.jgi-psf.org/chlre1/chlre1.home.html).
Although the ICAT analysis identified only some of the known axonemal proteins, the results for these were as expected (Table 2). For example, peptides representing the α, β, and γ heavy chains of the outer dynein arm were present in oda5-1 flagella in amounts ranging from 0 to 0.24 times their wild-type levels. In contrast, peptides representing proteins of the ODA-DC (DC1), inner dynein arm (p28), and central-microtubule-pair-complex (PF6) were not reduced in oda5-1 compared with wild type. These results validate the ICAT approach for the quantitative comparison of flagellar fractions.
Four other peptides, all derived from the same predicted gene, were present in oda5-1 in amounts ranging from 0.54 to 0.64 their wild-type levels. This gene is predicted to encode an AK with a mass of ~70 kDa and homology to AKs in other organisms. The cDNA sequence was derived from EST sequences, PCR-amplified cDNA clones, and predicted coding sequence in the JGI Chlamydomonas genome version 2.0 database (Figure 8B). Another peptide representing a previously unidentified WD-repeat protein was present in oda5-1 at about two-thirds its amount in wild type (Table 2); this protein (GreenGenie 492.9 in version 1.0 of the JGI Chlamydomonas genome) is represented in the EST database (20021010.5906.1).
AK activity has previously been observed within the Chlamydomonas flagellum (Watanabe and Flavin, 1976 ). To confirm the apparent reduction of this enzyme in oda5-1 flagella, we directly assayed AK enzymatic activity in Tergitol-extracted axonemes from wild-type, oda1, oda3, oda5-1, oda8, oda9, oda10-2, and oda5-2 rescued strains. Relative to its level in wild-type, AK activity was reduced by ~30–50% in oda5-1 and oda10-2 but not in oda1, oda3, oda8, or oda9 (Figure 7B). Corresponding reductions in activity also were observed in oda5-2 and oda10-1 (our unpublished data). A strain that was rescued for the Oda- motility defect with an HA-tagged ODA5 gene construct had wild-type levels of AK activity, demonstrating that it also is rescued for the AK- defect. The ~35–50% reduction in AK activity in oda5 axonemes correlates well with the 35–45% reduction in AK protein levels in oda5-1 as determined by ICAT. The similar reduction observed in oda10 provides additional evidence for a biochemical connection between Oda5p, Oda10p and AK. The loss of AK activity is not a general consequence of the failure to assemble the outer arm or ODA-DC, because AK activity is normal in oda1, oda3, and oda9 mutants lacking these structures.
Interestingly, we found that if wild-type flagella were extracted with Nonidet P-40 rather than Tergitol, the fraction of AK activity that is deficient in oda5-1 mutants is soluble. When wild-type and oda5 flagella were extracted with Tergitol, the AK specific activity in the resulting pellet was 35% less in oda5 than in wild type, whereas the AK specific activity in the soluble fraction was equivalent in wild type and oda5 (Figure 7C). However, when flagella were demembranated with Nonidet P-40, the amount of AK activity remaining in the axonemes was equivalent in wild type versus oda5, but now the relative amount of AK activity in the soluble fraction was reduced ~40% in oda5 relative to wild type (Figure 7C). The most likely explanation for these results is that the AK is associated with the Oda5p complex via a Nonidet-sensitive bond. Alternatively, because Nonidet, but not Tergitol, completely removes the flagellar membrane (Figure 7A), the Oda5p-associated AK also may be connected to the flagellar membrane.
Loss of the ODA5 gene causes loss of the outer dynein arm, resulting in a slow, jerky swimming phenotype (Kamiya, 1988 ), together with a previously undescribed phenotype – reduction in the level of a newly identified flagellar AK. We have isolated the ODA5 gene and find that outer dynein arm assembly and adenylate kinase activity are restored when the gene is transformed back into the oda5-2 mutant, demonstrating that the defect in ODA5 is responsible for both the Oda- and AK- (adenylate kinase) phenotypes. Oda5p localizes to the flagellum and remains associated with the axoneme after Nonidet P-40 demembranation, indicating that it is an axonemal protein. Oda5p assembly onto the axoneme is independent of the outer arm and the ODA-DC, but dependent upon the Oda10 gene product. These results indicate that Oda5p is part of a complex with Oda10p and the newly identified flagellar AK.
The ODA5 cDNA predicts a novel 66-kDa protein containing extensive α-helical domains. There are five regions with a high probability of forming coiled-coils. Coiled coils commonly mediate protein–protein interactions, and it is likely that Oda5p interacts with itself or with another protein via these regions. The region between coiled-coils C and D in Figure 3B is largely hydrophobic, with 63 of 126 amino acids being hydrophobic. Just after predicted coiled-coil domain E is a stretch of 10 glycines (G434–G444), which would break the coiled-coil structure and could serve as a flexible hinge between the coiled-coil domains and the COOH terminus. A very acidic region (E619–E634) is located near the COOH terminus and may mediate interactions between Oda5p and the outer dynein arm or other axonemal components; Oda5p is released from the axoneme by 0.6 M KCl, indicating its association with the axoneme is ionic in nature. These features of Oda5p are reminiscent of the DC1 and DC2 subunits of the ODA-DC, which also have predicted coiled-coil structures and COOH-terminal charged regions.
Although the predicted mass of Oda5p is 66 kDa, our antibody to Oda5p detected a protein with a Mr of 76,000 on SDS-polyacrylamide gels. An anomalously high Mr is not uncommon for coiled-coil proteins and similarly was observed for DC1 and DC2 (see Koutoulis et al., 1997 , for discussion).
Although Oda5p, the outer dynein arms, and the ODA-DC are all released from the axoneme by 0.6 M KCl, Oda5p sediments separately from the outer arms and the ODA-DC in sucrose density gradients. Inasmuch as Oda5p is required for assembly of the outer dynein arm, it is likely that it associates directly or indirectly with the arm in vivo, but that this association is disrupted by the high-salt extraction. However, Oda5p also assembles onto the axoneme independently of the outer arm and the ODA-DC, as evidenced by its presence on the axonemes of several oda mutants lacking the latter structures. Conversely, the ODA-DC can assemble onto the axoneme in the absence of Oda5p. Therefore, Oda5p is not simply a previously unidentified subunit of the outer arm or the ODA-DC, but a separate and independently assembling component of the axoneme that is necessary for outer arm attachment to the doublet microtubules.
The inability of oda5, oda8, and oda10 mutants to complement in temporary dikaryons suggested that the Oda5, Oda8, and Oda10 proteins interact in a complex (Kamiya, 1988 ). Our finding that Oda5p is missing from axonemes of the oda10 mutant demonstrates that a functional Oda10p is required for the localization of Oda5p and provides the first biochemical evidence for an interaction between these two proteins. In contrast, Oda5p was not missing from axonemes of the oda8 mutant. It is possible that Oda8p is associated with Oda5p and Oda10p, but not required for binding of Oda5p to the axoneme (Figure 9).
ICAT analysis revealed that a flagellar AK was reduced 35–45% in oda5-1 compared with wild type. AK catalyzes the reversible reaction: 2 ADP ↔ ATP + AMP. Direct enzymatic assays confirmed that flagellar AK activity was reduced in oda5 and indicated that the AK activity was similarly reduced in oda10 but not in the other oda mutants. Therefore, a portion of the newly identified AK is dependent upon Oda5p and Oda10p for its incorporation into the flagellar structure. Because the flagellar AK is not reduced in other mutants that lack the outer arms or the outer arms together with the ODA-DC, it is not simply a subunit of either of these structures. Rather, it seems to be specifically associated with the Oda5p/Oda10p complex. AK activity is not reduced in oda8, consistent with our finding that Oda5p is assembled onto oda8 axonemes.
A model illustrating the assembly of these complexes onto the axoneme is shown in Figure 9. Outer dynein arm assembly requires both the ODA-DC and the Oda5p/Oda8p/Oda10p complex, whereas assembly of the outer arm-associated AK requires only the Oda5p/Oda8p/Oda10p complex. Our biochemical results and the lack of cytoplasmic complementation between oda5, oda8, and oda10 can be accounted for as follows: If either the ODA5 or ODA10 gene is disrupted, then the remaining subunits of the Oda5p/Oda8p/Oda10p complex are unstable, and no partial complex is formed. In this case, neither outer arm assembly nor adenylate kinase assembly occurs. If the ODA8 gene is disrupted, then Oda5p (and possibly Oda10p) are stable, forming a partial complex, which is targeted to the axoneme. Assembly of the partial complex is sufficient for binding of adenylate kinase but not the outer arm.
The flagellar AK identified by ICAT analysis is encoded by a gene having five exons (Figure 8A). The NH2 terminus, predicted by exons 1 and 2, is unique, whereas the third, fourth, and fifth exons are each predicted to encode a nearly identical domain that closely resembles the conserved domain of other AKs. The conserved domain of AKs has two conserved motifs, the AK signature motif and the ATP-binding motif or P-loop. The AK signature motif constitutes part of the catalytic cleft and contains conserved arginine and aspartic acid residues. The P-loops of AKs deviate from the usual P-loop consensus sequence ([AG]-X4-G-K-[ST]) in that the last position is occupied by a glycine instead of a serine or threonine (Prosite PDOC00017). The flagellar AK identified here contains the conserved aspartic acid and arginine in the signature motif and the conserved glycine substitution within the P-loop motif in each of its three AK domains (Figure 8B). In a phylogenetic comparison with the known human AKs, the Chlamydomonas flagellar AK was most closely related to the cytosolic isoforms AK1 and AK5 (our unpublished results). Interestingly, although it is unusual for AKs to have repeats of the conserved AK domain, two isoforms of human AK5 (AK5 variant 1 and 2) both contain two AK domains (Figure 8C).
The flagellar AK is solubilized from the wild-type flagellum by treatment with Nonidet P-40 but not Tergitol. Because the AK is not predicted to contain transmembrane domains, and because its presence in the flagellum is dependent upon Oda5p, which is an axonemal protein, it is probable that the AK also is an axonemal component and that it is associated with the axoneme via interactions that survive the Tergitol treatment but are disrupted by Nonidet P-40.
AK activity previously has been reported in cilia and flagella from several organisms (Watanabe and Flavin, 1976 ; Schoff et al., 1989 ; Nakamura et al., 1999 ; Noguchi et al., 2001 ), although neither the specific AK nor its location in the flagellum were determined. The apparent association of the AK with Oda5p and Oda10p in the current study, and the fact that the latter are required for outer dynein arm assembly, suggest that the AK is held in proximity to the outer arm by the Oda5p/Oda10p complex (Figure 9). We propose that the outer dynein arm— one of the major ATP-hydrolyzing structures of the axoneme—is intimately associated with an ATP-regenerating system to achieve efficient conversion of ADP to ATP and AMP, thus ensuring that both high-energy phosphate bonds of ATP are readily accessible to this important force-generating machine.
That the flagellar AK is reduced but not absent in the oda5 and oda10 mutants indicates that the AK is located at additional sites within the axoneme. One likely site for the remaining flagellar AK is the inner dynein arm system. Such a localization would place the AK at two major sites of ATP utilization in the axoneme.
ICAT analysis also identified one peptide, from a predicted WD-repeat protein, that was reduced in oda5-1 flagella by about the same amount as the AK. It is tempting to speculate that this protein may be associated with the AK. The protein has homologues in other flagellated organisms, including H. sapiens, Mus musculus, Macaca fascicularis, Anopheles gambiae, and Drosophila melanogaster. Further characterization of the protein will be required to clarify its relationship to the flagellar AK.
Our ICAT analysis correctly reported the relative levels, in wild-type versus oda5 flagella, of all peptides that were identified as being derived from known axonemal proteins, as well as those from the previously undescribed flagellar AK. The main shortcoming of the ICAT approach was its inability to identify a more comprehensive set of flagellar proteins, as evidenced by its failure to identify peptides from a greater percentage of known axonemal proteins. This was most likely due to the long duty cycle (~14 s) of the mass spectrometer used, which prevented many peptides from being selected for fragmentation for MS/MS analysis. This problem should be greatly alleviated by newly available instruments, which have much faster data acquisition rates and thus can analyze a much larger number of peptides from a sample of the same complexity. With the use of such instruments, ICAT and similar quantitative proteomic approaches (Aebersold and Mann, 2003 ) are likely to become very valuable for identifying specific proteins whose levels are altered in mutant versus wild-type flagella.
We are grateful to Carolyn Silflow for providing the p3 × HA cassette, to Carol Dieckmann for arg- strains, to William Snell for the gametic cDNA library, to Susan Dutcher for help with the GreenGenie program, and to Toshiki Yagi for assistance with motion analysis. These studies were 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 the Japanese Ministry of Education, Culture, Sports, Science and Technology (to R.K.), and by CREST of Japan Science and Technology Corporation (to R.K.). We also acknowledge grant 5 P30 DK32520 from the National Institute of Diabetes and Digestive and Kidney Diseases for use of University of Massachusetts Diabetes and Endocrinology Research Center supported core facilities.
Article published online ahead of print. Mol. Biol. Cell 10.1091/mbc.E03-11-0820. Article and publication date are available at www.molbiolcell.org/cgi/doi/10.1091/mbc.E03-11-0820.
Abbreviations used: DHC, dynein heavy chain; IC, intermediate chain; LC, light chain; ODA-DC, outer dynein arm-docking complex; RFLP, restriction fragment length polymorphism; AK, adenylate kinase; ICAT, isotope-coded affinity tagging; JGI, Joint Genome Institute.
Online version of this article contains supporting material. Online version is available at www.molbiolcell.org.