Fa1p is a 171 kDa protein essential for axonemal microtubule severing in Chlamydomonas

The molecular mechanism and physiological functions of deciliation/deflagellation have received little scientific attention as biological problems. This is in spite of the fact that for nearly half a century, scientists have used deflagellation and deciliation to isolate flagella and cilia from Chlamydomonas and Tetrahymena (Lewin, 1953; Watson and Hopkins, 1962). Although the ecological/physiological roles of deciliation/ deflagellation are poorly understood, it is an evolutionarily conserved process. Sea urchin embryos, scallop gills, rabbit oviduct and porcine respiratory tissue, and rat cerebral ependymal cells have all been directly observed to deciliate in response to stress (Auclair and Siegel, 1966; Anderson, 1974; Stephens, 1975; Hastie et al., 1986; Mohammed et al., 1999). A hallmark feature of these processes is the severing of the nine outer-doublet axonemal microtubules at a specific site at the base of the flagellum/cilium, distal to the transition zone. The mechanism of deflagellation has been most thoroughly studied in Chlamydomonas (reviewed by Quarmby and Lohret, 1999).

Chlamydomonas is a well established system for tractable genetic, in vitro, and biochemical studies of the deflagellation behavior (Quarmby, 1996; Finst et al., 1998; Lohret et al., 1998). We have previously demonstrated that the Ca²⁺ sensing mechanism and the machinery of doublet microtubule severing purify with membrane-free preparations of flagellar-basal body complexes (FBBCs; Lohret et al., 1998). Furthermore, the established microtubule severing protein, katanin (Vale, 1991; McNally and Vale, 1993; Hartman et al., 1998), is a component of the FBBC suggesting that it may be a subunit of the microtubule severing machinery. In Chlamydomonas, the 60 kDa subunit of katanin is a single copy gene with high sequence identity to the human p60 ortholog (Lohret et al., 1999). Chlamydomonas p60 is found, among other places, on the inner face of the axoneme at the distal end of the flagellar transition zone where axonemal severing occurs during deflagellation (Lohret et al., 1999). We also showed that exogenously added katanin could sever the complex doublet microtubules of the axoneme. Although, katanin is localized at the proper position and has the necessary biochemical activity to sever axonemes during deflagellation, direct evidence of a role for katanin in deflagellation, as would be provided by mutant strains, is lacking.

In order to directly identify proteins involved in axonemal microtubule severing, we isolated deflagellation-defective Chlamydomonas mutants (Finst et al., 1998). From our analysis of non-essential mutations, we discovered three genes, ADF1, FA1, and FA2, that are necessary for deflagellation (Finst et al., 1998). ADF1 is essential for Ca²⁺ influx whereas FA1 and FA2 are required for shedding of the flagella in response to the Ca²⁺ signal. We did not isolate katanin mutants in this screen, suggesting that either it is not involved in deflagellation or it plays other, essential, roles in the cell. We favor the latter explanation (see Lohret et al., 1998) and hypothesize that the products of FA1 and FA2 may be important for the specific action of katanin during deflagellation (reviewed by Quarmby and Lohret, 1999).

We report here the cloning of FA1, the first gene required for deflagellation. We isolated a 10.7 kb genomic clone which rescued the deflagellation defect of fa1 mutants. Using RT-PCR we determined that the FA1 cDNA is ~5.7 kb and encodes a product of ~171 kDa with a predicted coiled-coil domain and consensus Ca²⁺/calmodulin (Ca²⁺/CaM) binding motifs. Northern analysis suggests that the FA1 message is upregulated following deflagellation. As anticipated, Fa1p is enriched in flagellar-basal body complexes, but is not detected in the flagella. FBBCs of fa1 mutants fail to undergo axonemal severing following addition of Ca²⁺. These results suggest that Fa1p plays a role in the transduction of a Ca²⁺ signal into activation of axonemal microtubule severing.

Chlamydomonas reinhardtii strains g1 and B214 were obtained from G. Pazour and G. Witman (Univ. of Massachusetts, Medical Center Worchester; Pazour et al., 1995) and mutant strains fa1-3 (Nit1+, mt+) and fa1-4 (nit1, mt-) were isolated in our lab as previously described (Finst et al., 1998). fa1-1 (previously known as fa-1; Lewin and Burrascano, 1983) and the arg7 strain, defective in the arginosuccinate lyase gene, were obtained from the Chlamydomonas Genetics Center (Durham, NC). A strain expressing hemagglutinin-epitope tagged α-tubulin was obtained from D. Diener and J. Rosenbaum (Yale University, New Haven, CT; Kozminski et al., 1993). Cells were maintained in liquid TAP medium or on plates (1.5% agar; Harris, 1989) at 22°C with constant illumination. All arg7 strains were maintained on medium supplemented with 0.02% arginine (Sigma, St Louis, MO). Nit1+ transformants (see below) were selected by growth on SGII (NO3) media (Sager and Granick, 1953; Fernandez et al., 1989).

Flagellar-basal body complexes were isolated from wild-type and fa1-4 strains; deflagellation was induced as previously described (Lohret et al., 1998).

Isolation of Chlamydomonas genomic DNA and Southern analysis was performed as previously described (Finst et al., 1998). Genomic DNA of strain fa1-3 was individually or doubly digested with endonucleases PstI, SacI, HindIII, BamHI, XhoI, SmaI, and XbaI and size fractionated by agarose gel electrophoresis. Following transfer to and immobilization on nitrocellulose membranes, DNA was hybridized with 32P-labeled probes (Prime-It II kit, Stratagene, La Jolla, CA) generated from either pUC119 or a 1.2 kb fragment from the 3¢ end of NIT1 (Finst et al., 1998). Using the known restriction map of the inserted DNA, pMN56 (NIT1 in pUC119; Fernandez et al., 1989), a partial restriction map of the DNA flanking the insertion was constructed. By measuring the size of the fragments containing portions of the insertional DNA, the distance from the ‘anchor’ site within the insert to the closest site in the flanking genomic DNA was determined.

To confirm the physical map, genomic DNA of wild-type, fa1-1, fa1-2, and fa1-3 strains was digested with SmaI and fragments separated by agarose gel electrophoresis. Following transfer to and immobilization on nitrocellulose membrane, DNA was hybridized to a 32P-labeled probe generated by random primer extension from a 2 kb fragment (MV2S/Sa) of bacteriophage MV2 isolated by restriction digestion with endonucleases SacI and SalI (provided by P. Ferris, Washington Univ., St Louis, MO).

To identify bacteriophage clones of wild-type Chlamydomonas DNA that could rescue the deflagellation defect, fa1-3;arg7 cells were cotransformed with a selectable marker, pARG7.8 (encoding arginosuccinate lyase; Debuchy et al., 1989), and genomic clones from the region of the NIT1 insertion (bacteriophage NJ1, W1, MS1, W2, HP3, and MS3; generously provided by P. Ferris and U. Goodenough, Washington Univ., St Louis, MO). Arg7+ transformants were selected by growth on medium lacking arginine. After identification of rescuing clones, fa1-4 cells were cotransformed with pMN56 (encoding nitrate reductase) and subclones of the relevant phage DNA. Medium supplemented with sodium nitrate was used to select Nit1+ transformants. Plasmid pW2SE was created by digesting bacteriophage W2 DNA with endonucleases SalI and EcoRI and ligating the 10.5 kb DNA fragment to pBluescriptII SK- (Stratagene, La Jolla, CA). Plasmid pW2BE was constructed by digesting pW2SE with BamHI and EcoRI and ligating the 8.8 kb fragment to pBluescriptII SK-. Nuclear transformation and determination of deflagellation phenotype was performed as described (Finst et al., 1998).

In preparation for DNA sequencing of the genomic clone, W2 bacteriophage DNA was digested with SmaI or EcoRI and KpnI and fragments ligated into pBluescript II SK-. Bacteriophage DNA of the MS1 clone was individually or doubly digested with EcoRI, SacI, HindIII, BamHI, KpnI or SalI and fragments ligated into pBluescriptII SK-. Subclones were sequenced by Genomis, Inc. (Augusta, GA) or the Emory University DNA Sequencing Facility (Atlanta, GA). Additional oligonucleotide sequencing primers were synthesized by Integrated DNA Technologies (Coralville, IA).

To identify the defect of the fa1-1 allele, genomic DNA isolated from strains B214 and fa1-1 was used as the template for PCR reactions (conditions described below). Using primers annealing near the 3¢ end of the FA1 gene, aberrant products were obtained with fa1-1 DNA as the template. Following DNA sequencing, we identified a rearrangement of this region. To create pW2SEInv, a genomic subclone in which the identified inversion of fa1-1 replaces the corresponding region of the wild-type clone, a series of subcloning steps were used. A 2.5 kb fragment of genomic DNA, isolated by digestion of pW2BE with KpnI and NotI, was ligated to pBluescriptII SK- to produce plasmid pW2KN. A 7.0 kb fragment of genomic DNA, isolated by digestion of MS1 phage DNA with SacI and EcoRI, was ligated to pBluescriptII SK- to create plasmid pMS1-7.0. The 1.6 kb DNA product obtained by PCR of DNA from fa1-1 with primers 8530(+) and 10112(-) (see Table 1) was digested with restriction endonucleases ApaI and NotI. The resulting ~1.3 kb product was ligated to pW2KN that had been similarly digested. This plasmid and pMS1-7.0 were digested with SfiI and NotI and the 1.6 kb and 8.9 kb fragments were ligated yielding plasmid pMS1-7.0Inv. To produce the final plasmid, pMS1-7.0Inv and pW2SE were digested with NdeI and EcoRI and the 4.6 kb and 9.0 kb products, respectively, were ligated yielding plasmid pW2SEInv. DNA sequencing of plasmid pW2SEInv confirmed that the inversion had been appropriately subcloned into the otherwise wild-type sequence.

Total RNA was isolated from vegetative cells following the methods of Wilkerson et al. (1994). Briefly, cells grown in liquid TAP medium with constant light were harvested by centrifugation and resuspended in lysis buffer (20 mM Tris, pH 8, 20 mM EDTA, pH 8, 5% SDS, 0.4 mg/ml Proteinase K) at room temperature. Following a 2 hour incubation, the lysis mixture was phenol/chloroform extracted in the presence of 0.23 M sodium acetate. Nucleic acid was precipitated with an equal volume of isopropanol, collected by centrifugation and washed with 80% ethanol. RNA was resuspended in water and an equal volume of 4 M LiCl, precipitated overnight at 4°C and collected by centrifugation. Polyadenylated RNA was isolated using oligo-(dT) cellulose (Life Technologies, Gaithersburg, MD) chromatography (Romero et al., 1998).

For RT-PCR, 500 ng of polyadenylated RNA was reverse transcribed using 200 U SuperScriptII reverse transcriptase (Life Technologies, Gaithersburg, MD) and 0.5 μg oligo (dT)(12-18), 2 pmol 7184(-) Junction, or 2 pmol 4058(-) Junction primer (see Table 1 for primer sequences). 1% of the reaction mixture was used as template for polymerase chain reaction (PCR). Initial PCR primers were designed to sequences predicted to be transcribed by the GeneMark algorithm (M. Borodovsky and A. V. Lukashin, unpublished; http://genemark.biology.gatech.edu/GeneMark/;Table 1). Subsequent primers were designed to bona fide exons and exon/exon junctions (Table 1). ‘Junction’ primers hybridize to the exon/exon junction of the FA1 mRNA, but not to any contaminating genomic DNA. Each PCR reaction mixture contained 0.5 μM of each primer, 0.5-2.0 mM MgCl2, 5% DMSO, 0.3 mM dNTP, 1´ Taq DNA polymerase buffer and 0.025 U/μl Taq DNA polymerase (Promega, Madison, WI). The PCR reaction mixture was denatured at 95°C for 5 minutes, followed by 40 cycles of 95°C for 1 minute, 58-60°C for 1 minute and 72°C for 3-5 minutes. The reaction mixture was then maintained at 72°C for an additional 8-10 minutes. To identify specific products, a second round of PCR was frequently performed using 0.2-2% of the first round product as template and primers predicted to anneal to sites internal to the first round primers. PCR products were separated by agarose gel electrophoresis, specific products were excised from the gel and purified using an Ultrafree-MC filter unit (Millipore, Bedford, MA), cloned into the pGEM-T Easy vector (Promega, Madison, WI) and sequenced.

For 3¢ RACE, we modified the SMART RACE cDNA Amplification protocol (Clontech, Palo Alto, CA) by designing primers with high melting temperatures. One-half microgram of dT Adaptor primer (Table 1) was used for reverse transcription. Each 3¢ RACE PCR reaction mixture contained 0.2 μM gene specific, 0.02 μM Long Universal and 0.1 μM Short Universal primers. The reaction mixture for a second round of 3¢ RACE PCR was identical except that 0.2 μM nested gene specific and 0.2 μM Nested Universal primers were used. The RACE reaction mixture was initially denatured at 95°C for 5 minutes followed by 3 cycles of 95°C for 1 minute, 70°C for 1 minute and 72°C for 5 minutes. Cycles were repeated decreasing the annealing temperature 2°C every 3 cycles to 58°C. Following 40 cycles at an annealing temperature of 58°C, the reaction mixture was maintained at 72°C for an additional 12 minutes.

For 5¢ RACE, the protocol of Eyal et al. (1999) was used with the following modifications. Reverse transcription was performed as described above using a 5¢-phosphorylated 4058(-) Junction primer. Following RNase treatment, the reaction mixture was purified with the QIAquick PCR Purification Kit (Qiagen, Valencia, CA) and eluted with 30 μl dH2O. After overnight ligation with T4 RNA ligase (New England Biolabs, Beverly, MA), 5% of the self-ligated cDNA served as template for PCR with primers 3702(+) and 3428(-) (Table 1). Products were separated by agarose gel electrophoresis and specific products were cloned and sequenced as described above.

For northern analysis, 10 μg of polyadenylated RNA was size fractionated on formaldahyde gels and transferred to Zeta Probe GT membranes (Bio-Rad, Hercules, CA) following established procedures (Sambrook et al., 1989). Three fragments of the FA1 cDNA, totaling 4.8 kb, were random primed with the High Prime DNA Labeling Kit (Boehringer-Mannheim, Indianapolis, IN) for 3 hours at room temperature using [32P]dCTP and [32P]dATP at 6000 mCi/mmol. Membranes were then hybridized at 65°C for 48 hours in a buffer containing 5´ SSC, 2´ Denhardt’s solution, 100 mg/ml dextran sulfate, 1% SDS and 0.4 mg/ml sonicated herring sperm DNA.

To identify genes related to FA1, searches of the SwissProt and GenBank (Flat File Release 115.0) databases were performed using BLAST and FASTA programs (Altschul et al., 1997). The program COILS (Lupas et al., 1991) was used to predict coiled-coil regions. Calmodulin binding domains were identified by eye using consensus sequences (Rhoads and Friedberg, 1997).

To facilitate detection of the FA1 product, a genomic construct encoding FA1 was engineered to contain three copies of the hemagglutinin (HA) epitope at the extreme carboxy terminus of the predicted product. First, wild-type genomic DNA was amplified with primers 11752(+) and 12476(-)-NruI and the PCR product subcloned into vector pGEM-T Easy. Following digestion with SpeI and NruI, the 700 bp fragment was ligated to vector p3´HA (a plasmid encoding 3 repeats of the 9 amino acid hemagglutinin epitope; generously provided by C. Silflow, Univ. of Minnesota, St Paul, MN) to create vector p3´HA/11752-12476. Genomic DNA was amplified with primers 12497(+)-NheI and 12926(-)-XbaI and the 500 bp PCR product subcloned into vector p3´HA/11752-12476 following digestion with SalI and NheI to create vector p3´HA/11752-12926. To generate the final construct, vector p3´HAFA1, vector p3´HA/11752-12926 was digested with EcoRI and XbaI and the 1.4 kb fragment ligated to pW2SE.

To determine whether the epitope tagged construct could rescue the fa1 defect, fa1-4 cells were cotransformed with pMN56 and p3´HAFA1, and Nit1+ transformants selected by growth on medium containing nitrate. Transformants that were wild-type for deflagellation were designated as 3´HAFA1. To confirm that these rescued mutants retained the nucleotide sequence encoding the epitope tag, genomic DNA was isolated and used as template for PCR with primers 12378(+) and 12777(-). A ~520 bp product demonstrated that the transformed strains retained the nucleotides encoding the epitope tag.

For western analysis, whole cell and flagellar-basal body complex protein was isolated as described (Lohret et al., 1998). Flagella and axonemes were isolated by the dibucaine method (Witman, 1986; King et al., 1986) and purified as described by Smith and Sale (1991). Protein concentration was determined using the Bio-Rad Protein Assay (Hercules, CA). 10 μg of protein was separated on a 5% SDS-PAGE gel and electroblotted to Immuno-Blot PVDF membrane (Bio-Rad, Hercules, CA). The membrane was stained with Ponceau S (Sigma, St Louis, MO) to confirm efficient transfer of all protein samples. The membrane was washed and incubated overnight at 4°C with an anti-HA rat monoclonal antibody (clone 3F10; 200 ng/ml; Boehringer-Mannheim, Indianapolis, IN) according to the manufacturer’s instructions. The membrane was then incubated with anti-rat Ig-POD (200 mU/ml; Boehringer-Mannheim, Indianapolis, IN) for 30 minutes at room temperature and immunoreactive protein identified with BM Chemiluminescence Blotting Substrate (Boehringer-Mannheim, Indianapolis, IN).

Detergent-permeablized cells deflagellate with addition of Ca²⁺ suggesting that the components of microtubule severing are localized to the transition zone and associated with the axoneme (Sanders and Salisbury, 1989). Unlike wild-type cells, permeabilized fa1 mutants fail to deflagellate indicating that the FA1 product is important for the formation of a microtubule-severing complex, possibly itself a subunit of the complex (Sanders and Salisbury, 1989). Consistent with this, flagellar-basal body complexes (FBBCs), composed of the two basal bodies, the distal striated fibers which connect the basal bodies and the two axonemes which derive from the basal bodies (Dutcher, 1995), isolated from wild-type cells exhibit Ca²⁺-dependent axonemal microtubule severing (Lohret et al., 1998; Fig. 1A,B). As shown in Fig. 1C and D, the axonemes of FBBCs purified from fa1 mutants are not severed following addition of Ca²⁺, suggesting that the fa1 mutants have lost the ability to sense Ca²⁺ or are unable to transduce the Ca²⁺ signal into the microtubule severing response.

The fa1 allele, fa1-3, is a tagged allele that was isolated in a screen for deflagellation mutants following NIT1 insertional mutagenesis (Finst et al., 1998). We created a physical map of the genomic DNA flanking the inserted plasmid in the fa1-3 allele. fa1-3 genomic DNA was digested with various restriction endonucleases and Southern analysis was performed using probes to the 5¢ and 3¢ ends of the transforming DNA. fa1-3 genomic DNA digested with SalI was hybridized with the pUC vector. As shown in Fig. 2A, the hybridizing product was 4.4 kb. Similarly, genomic DNA digested with BamHI and hybridized with a probe corresponding to the 3¢ end of NIT1 yielded a 7.5 kb band. Through this analysis, it was determined that ~2 kb of pUC was deleted, precluding the possibility of using plasmid rescue to clone the flanking DNA. Previous genetic analysis had placed FA1 within the mating-type locus (Ferris and Goodenough, 1994). The two restriction maps flanking the inserted DNA each matched only a single region in the 1.1 Mb published restriction map of the mating-type locus (Ferris and Goodenough, 1994). The identification of these two flanking sequences was confirmed by probing Southern blots of genomic DNA of wild-type, fa1-1, fa1-2, and fa1-3 with a fragment of DNA predicted to be deleted in fa1-3 (MV2S/Sa in Fig. 2A). A hybridizing band of identical molecular mass is present in the genome of fa1-1, fa1-2, and wild-type strains (Fig. 2B, lanes 1-3, respectively), but is absent in fa1-3 (Fig. 2B, lane 4). This result confirmed both the physical position of FA1 within the centromere-proximal region of the mating-type locus and the presence of a ~36 kb deletion in fa1-3.

A fa1-3;arg7 double mutant was cotransformed with wild-type DNA from the region of the deletion isolated from individual bacteriophage and ARG7 as a selectable marker. Transformants were selected by growth on medium lacking arginine and the deflagellation phenotype of individual transformants was determined. Transformation with DNA isolated from two phage, MS1 and W2, completely rescued the deflagellation defect (Fig. 3A). We deduced that the FA1 gene was located in the 14.7 kb region of overlap between these two phage. Other phage which extend into this region did not rescue the deflagellation defect, suggesting that the FA1 gene was large. To locate the FA1 gene within W2, the phage DNA was digested with restriction endonucleases and the resultant fragments were subcloned into plasmids. These plasmids were then cotransformed with a selectable marker, NIT1, into fa1-4 mutants. The deflagellation phenotype of transformants that were able to grow on medium containing nitrate as the sole nitrogen source was determined. Transformation with plasmid pW2SE rescued the deflagellation defect of fa1-4 (Fig. 3A). In contrast, plasmid pW2BE did not complement the fa1-4 defect. FA1 was, therefore, contained within the 10.5 kb genomic subclone of plasmid pW2SE.

In order to identify the FA1 gene, the DNA of the rescuing clone was sequenced. The GeneMark gene prediction algorithm identified 16 putative exons on one strand whereas the other strand was largely devoid of predicted exons (data not shown).

In order to identify the FA1 cDNA, we used reverse transcription coupled with polymerase chain reaction (RT-PCR). Polyadenylated RNA isolated from an asynchronous population of wild-type cells was reverse transcribed with a primer designed to anneal to the 3¢ poly(A) tail of mRNAs. PCR reactions were performed with the primers outlined in Table 1. Following DNA sequencing and subsequent alignment of the genomic sequence with the sequence of cDNA clones, multiple exons were identified, and all introns contained consensus 5¢ donor and 3¢ acceptor splice sites (LeDizit and Piperno, 1995).

To clone the 5¢ and 3¢ ends of the FA1 cDNA, Rapid Amplification of cDNA Ends (RACE) was used. For 3¢ RACE, mRNA was reverse transcribed with dT Adaptor, a primer containing unique 5¢ sequences for subsequent PCR with primary and nested primers. PCR with primers 10400(+) and Nested Universal resulted in a 1.7 kb product. By comparison of the cDNA and genomic DNA sequence, additional exons were identified as well as a polyadenylation signal (TGTAA) which occurs 19 nucleotides 5¢ of the poly(A) tail. To clone the 5¢ end of the FA1 transcript, mRNA was reverse transcribed with primer 4058(-) Junction. An inverse PCR approach employed primers 3428(-) and 3702(+) to yield a product corresponding to the 5¢ end of the FA1 transcript. The full-length FA1 cDNA, assembled from overlapping cDNA sequences, is 5753 bp (Fig. 3B, Accession Number AF246990).

For northern analysis, poly(A) RNA was hybridized with probes generated from three fragments totaling 4.8 kb and covering 84% of the FA1 cDNA. The hybridizing product was ~5.6 kb, consistent with the size of the cDNA obtained by RT-PCR (Fig. 4, lane 2). Lane 3 shows that the FA1 mRNA was not detected in a fa1 deletion allele, fa1-4 (see below). The level of expression of a number of flagellar genes is up-regulated following deflagellation (Lefebvre and Rosenbaum, 1986). Lane 1 of Fig. 4 shows that wild-type cells undergoing flagellar regeneration have increased FA1 mRNA expression.

By Southern analysis, the genomic DNA encoding FA1 in strains fa1-3 and fa1-4 is deleted (Fig. 2A; data not shown for fa1-4). To identify the defect of fa1-1, a panel of primers spanning the region containing the FA1 gene was used for PCR to ‘walk’ across the gene. Most primers produced products of wild-type size. In contrast, a reaction using primers 9226(+) and 10112(-) failed to produce any product. This suggested that fa1-1 carried a mutation near the predicted 3¢ end. The product of the PCR reaction with primers 8530(+) and 10112(-) was sequenced and a ~1.0 kb inversion was identified (Fig. 3B). Ferris and Goodenough (1994) found that the wild-type MT+ and MT- loci of Chlamydomonas are highly rearranged. These rearrangements, which are rare elsewhere in the genome, include translocations, inversions, duplications and deletions. In order test whether the inversion of fa1-1 was responsible for the deflagellation defect, we constructed a genomic subclone in which the inverted nucleotides of fa1-1 replaced the corresponding wild-type region. fa1-4 cells were cotransformed with the NIT1 gene, as a selectable marker, and a subclone of genomic DNA having either wild-type (pW2SE) or inverted sequences (pW2SEInv). The deflagellation phenotype was determined for both sets of cotransformants. As predicted, 23 of 284 transformants assayed were wild-type for deflagellation following transformation with pW2SE. In contrast, none of the 313 colonies assayed following transformation with pW2SEInv were rescued for the defect. This result demonstrated that the inversion found in fa1-1 is sufficient to impair the function of FA1.

The proposed translation initiation site is preceded by two in frame translation termination codons 45 and 114 nucleotides 5¢ of the initiation codon. The 3¢ translation termination codon (nucleotides 5490 to 5492) is 264 nucleotides 5¢ of the beginning of the poly(A) tail (Fig. 3B). The cDNA encodes 1787 amino acids with a predicted molecular mass of 171.6 kDa and isoelectric point of 7.2 (Fig. 5). Searches of public databases using the amino acid sequence did not identify any proteins with significant sequence similarity. A segment of ~225 amino acids near the amino terminus has a high probability of forming a coiled-coil domain (Lupas et al., 1991; underlined in Fig. 5). In addition, three segments of Fa1p contain consensus motifs for the 1-8-14 type B Ca²⁺-dependent/calmodulin (Ca²⁺/CaM) binding motif (boxed in Fig. 5; Rhoads and Friedberg, 1997). The consensus 1-8-14 type B Ca²⁺/CaM binding motif sequence is characterized by a 14 amino acid stretch having hydrophobic residues at positions 1, 8, and 14 and an overall net charge of +2 to +4.

Following isolation of the FA1 cDNA, it was determined that the smallest genomic construct that rescued the deflagellation defect lacked 905 bp of the 3¢ most exon and intron (pW2SE; Fig. 3A and B). Given that the carboxy-terminal 194 amino acids were apparently dispensable for function, we anticipated that introduction of an epitope tag at the carboxy terminus would not alter the activity of the product. We engineered three copies of the nine amino acid epitope of hemagglutinin (HA) at the carboxy terminus of a FA1 construct. Following cotransformation of fa1-4 cells with the nitrate reductase gene, as a selectable marker, and the epitope tagged FA1 construct, the deflagellation phenotype of selected colonies was determined. The frequency of rescue of the fa1-4 deflagellation defect with the HA-tagged construct (p3´HAFA1) was the same as that obtained with an untagged construct (pW2SE). Genomic DNA isolated from cells rescued for deflagellation was used as a template for PCR to confirm that the HA epitope was retained (data not shown). Monoclonal antiserum recognizing the HA epitope was used for western blots of whole cell, flagellar-basal body complex (FBBC), axonemal and flagellar protein from wild-type, fa1-4 and fa1-4 cells in which the mutant phenotype was rescued by the p3´HAFA1 construct (Fig. 6A). The sera recognized a protein of the expected size in FBBC protein from rescued fa1-4 cells (Fig. 6B, lane 6), but not in either the wild-type or fa1-4 mutant strains (Fig. 6B, lanes 1-2 and 3-4, respectively). The reactive species is not detected in axonemal or flagellar protein isolated from 3´HAFA1 cells (Fig. 6B, lanes 7 and 8). We estimate that ~0.2% of total cellular protein is contributed by the FBBC based on the mass of protein isolated following purification of FBBCs from whole cells. If Fa1p is a transition zone protein, as we predict, this likely explains our inability to detect Fa1p in whole cell protein samples of 3´HAFA1 cells (Fig. 6B, lane 5).

From previous work it was established that the components necessary for flagellar microtubule severing are tightly associated with the axoneme (Sanders and Salisbury, 1989; Quarmby and Hartzell, 1994; Finst et al., 1998; Lohret et al., 1998). As expected based on cellular phenotype, we report here that the axonemes of FBBCs isolated from fa1 mutants are not severed following treatment with Ca²⁺. Therefore, fa1 mutants have lost the ability to transduce a Ca²⁺ signal into axonemal microtubule severing.

To better understand the role of this gene in deflagellation, we have cloned FA1 and localized the gene product. The FA1 gene encodes a protein of 1787 amino acids with a predicted molecular mass of 171.6 kDa. Database searches using the deduced amino acid sequence did not identify any orthologs. The secondary structure of Fa1p is predicted to include a coiled-coil domain of ~225 amino acids near the amino terminus (Lupas et al., 1991). Coiled-coil domains are characterized by heptad repeats forming an amphipathic alpha helix and are present in a number of functionally diverse proteins. Hydrophobic interactions between coiled-coil domains result in stable protein-protein interactions. These interactions mediate assembly of both homo- and hetero-oligomeric complexes of varying subunit number (Shoeman and Traub, 1993; Beck and Brodsky, 1998; Heimburg et al., 1999). A candidate for hetero-oligomeric interactions with Fa1p is the product of FA2, a second gene required for microtubule severing (Finst et al., 1998). It should also be noted that other factors could interact with Fa1p because our screen for loss-of-function mutations affecting deflagellation would not have identified genes with functions essential for life. The presence of a coiled-coil domain is consistent with our hypothesis that Fa1p is a component of a stable axonemal microtubule severing complex.

As mentioned above, the Ca²⁺ sensor of deflagellation purifies with membrane-free preparations of flagellar-basal body complexes (Lohret et al., 1998). It is, therefore, of interest that Fa1p is predicted to have three 1-8-14 type B Ca²⁺/CaM binding domains. This class of calmodulin-binding motifs is found in a number of proteins including caldesmon and smooth muscle myosin light chain kinase (Rhoads and Friedberg, 1997). Gitelman and Witman (1980) determined that, in Chlamydomonas, calmodulin is present both as a detergent-soluble fraction and stably associated with the axoneme and is, therefore, a candidate for the Ca²⁺ sensor of deflagellation. Another candidate is the calmodulin-family member, centrin, which has been implicated in deflagellation (Sanders and Salisbury, 1989). Associated with the centrosome in most cells, centrin localizes to a variety of subcellular sites in Chlamydomonas including the distal end of the flagellar transition zone (Sanders and Salisbury, 1989). Ca²⁺-induced contraction of centrin-containing fibers in the transition zone is temporally and spacially correlated with the severing of outer doublet microtubules (Sanders and Salisbury, 1989).

Although Ca²⁺-induced contraction of centrin-containing fibers accompanies deflagellation in wild-type cells, the role of centrin in deflagellation is unclear. A mutant strain, vfl2-1, has a missense mutation in the first EF-hand Ca²⁺-binding domain of centrin (Taillon et al., 1992). In vfl2-1 cells, the stellate fibers of the flagellar transition zone are poorly organized likely resulting from the inability of centrin to be polymerized into filamentous structures (Taillon et al., 1992). Sanders and Salisbury (1994) reported that detergent-permeablized cell models of vfl2-1 treated with Ca²⁺ in the absence of any external shear forces (such as pipetting) do not deflagellate. However, under normal experimental conditions, vfl2-1 cells exhibit wild-type deflagellation behavior (Jarvik and Suhan, 1991; Quarmby and Lohret, 1999). In contrast, cells with the Fa^- phenotype do not shed their flagella under any circumstances. Ca²⁺-induced contraction of centrin-containing fibers and doublet microtubule severing are defective in detergent-extracted cell models of fa1-1 cells, yet centrin localization and transition zone ultrastructure in this strain are indistinguishable from wild-type (Sanders and Salisbury, 1989). The organization of the transition zone in the deletion allele, fa1-3, also shows no structural defects (P. Beech and R. Finst, unpublished observations). It is clear that even if centrin plays a role in deflagellation, additional factors are important.

To localize the FA1 product within the cell, an HA-epitope tagged product was created. This approach has been used previously to investigate the distribution of nonacetylatable α-tubulin in Chlamydomonas (Kozminski et al., 1993). The deflagellation defect of fa1-4 mutant cells was rescued following transformation with a genomic construct of FA1 having three copies of the HA epitope at the carboxy terminus. Therefore, the epitope tagged Fa1p was expressed, functional and at least partially targeted to the appropriate cellular compartment. Protein isolated from whole cells, FBBCs, axonemes and flagella of rescued mutants was used for immunoblots with a monoclonal anti-HA antibody. As predicted, Fa1p was enriched in FBBCs and was not detected in either axonemal or flagellar protein. This indicates that Fa1p is targeted to the basal body-transition zone region. Although Fa1p is present in FBBCs, it was not detected in whole cell protein consistent with our proposal that Fa1p is of low abundance. The low level of Fa1p, the restricted location of the microtubule severing activity to the transition zone of the FBBC, and the presence of Fa1p in the FBBC, but not in the axoneme, suggests that Fa1p is localized at the transition zone and may be the first protein identified that is specific to this region. Given the restricted location and precision of axonemal microtubule severing, it is reasonable to expect that only a few Fa1p molecules are required for this key event in deflagellation.

We warmly thank Drs P. Ferris and U. Goodenough (Washington University) for generously providing us with bacteriophage DNA, detailed physical maps of the mating type locus and much guidance. We thank Dr C. Silflow and M. LaVoie (Univ. of Minnesota) for the p3´HA plasmid and advice on epitope tagging. We are grateful to the members of the Quarmby lab and to Drs W. Sale, H. C. Hartzell, and M. Powers and members of their labs at Emory for helpful advice and stimulating discussion. This work was supported by National Science Foundation grant MCB-9603716 (L.M.Q.). R.J.F. is supported by a National Institutes of Health predoctoral fellowship (T32 GM-08367).