Insights into the Structural Organization of the I1 Inner Arm Dynein from a Domain Analysis of the 1β Dynein Heavy Chain

Abstract

To identify domains in the dynein heavy chain (Dhc) required for the assembly of an inner arm dynein, we characterized a new motility mutant (ida2-6) obtained by insertional mutagenesis. ida2-6 axonemes lack the polypeptides associated with the I1 inner arm complex. Recovery of genomic DNA flanking the mutation indicates that the defects are caused by plasmid insertion into the Dhc10 transcription unit, which encodes the 1β Dhc of the I1 complex. Transformation with Dhc10 constructs encoding <20% of the Dhc can partially rescue the motility defects by reassembly of an I1 complex containing an N-terminal 1β Dhc fragment and a full-length 1α Dhc. Electron microscopic analysis reveals the location of the missing 1β Dhc motor domain within the axoneme structure. These observations, together with recent studies on the 1α Dhc, identify a Dhc domain required for complex assembly and further demonstrate that the intermediate and light chains are associated with the stem regions of the Dhcs in a distinct structural location. The positioning of these subunits within the I1 structure has significant implications for the pathways that target the assembly of the I1 complex into the axoneme and modify the activity of the I1 dynein during flagellar motility.

INTRODUCTION

The dyneins are a family of molecular motors that convert the chemical energy of ATP binding and hydrolysis into mechanical force, resulting in minus-end–directed movement along microtubules. These motors play important roles in a number of diverse cellular processes, including mitotic events, vesicle movement, and ciliary and flagellar motility (Mitchell, 1994; Porter, 1996; Hirokawa et al., 1998). All dynein isoforms characterized thus far are large, multisubunit complexes containing one to three dynein heavy chains (Dhcs) (400–500 kDa), variable numbers of intermediate chains (ICs) (45–140 kDa), and one or more light chains (LCs) (8–28 kDa). Dyneins can be separated into two different classes: cytoplasmic and axonemal. Axonemal dynein isoforms are much more diverse, e.g., in Chlamydomonas, as many as seven different inner dynein arm isoforms have been identified, along with one three-headed outer arm isoform (Goodenough et al., 1987; Kagami and Kamiya, 1992). Despite this diversity, several of the axonemal Dhcs, ICs, and LCs share considerable homology with their cytoplasmic counterparts (Mitchell and Brown, 1994; Wilkerson et al., 1994, 1995; King and Patel-King, 1995; Harrison et al., 1998; Yang and Sale, 1998).

Axonemal dyneins have been studied most extensively in Chlamydomonas because of its accessibility to combined genetic, biochemical, and structural analysis (Harris, 1989, Goodenough, 1992; Dutcher, 1995). Chlamydomonas is haploid, and so it is relatively easy to screen for mutations in motility genes and thereby evaluate the contribution of each dynein isoform to flagellar motility. The ability to reintroduce modified dynein genes by transformation also allows for the investigation of functional domains within dynein subunits (Perrone et al., 1998; Myster et al., 1999). Furthermore, because each dynein isoform is targeted to a specific location within the 96-nm axoneme repeat, wild-type and mutant axonemes can be compared to determine in situ the structural alterations that result from specific polypeptide defects (Piperno et al., 1990; Mastronarde et al., 1992; Gardner et al., 1994). Such studies have provided the experimental evidence that the conserved central and C-terminal portions of the Dhc form the globular head or motor domain, whereas the more divergent N-terminal region forms a stem domain that interacts with associated LCs and ICs (Sakakibara et al., 1993; Myster et al., 1999).

The I1 inner arm complex serves as an excellent model for dynein assembly and function. First, the I1 dynein is a relatively simple isoform that contains two distinct Dhcs, three ICs, and three LCs, and it shares many similarities with the major cytoplasmic dynein. Both are two-headed isoforms that share closely related, WD-repeat containing ICs (Yang and Sale, 1998) and two identical LC subunits (Harrison et al., 1998). One of these LCs (Tctex1) is related to a gene product of the t complex, a region of mouse chromosome 17 involved in transmission ratio distortion and male sterility (Lader et al., 1989; Harrison et al., 1998). These observations are consistent with studies in Chlamydomonas indicating that the I1 complex is an important target of the regulatory signals that control flagellar movement (Porter et al., 1992; Habermacher and Sale, 1996, 1997; King and Dutcher, 1997). Finally, mutations in four loci that affect the assembly of the I1 complex have been isolated (Kamiya et al., 1991; Porter et al., 1992; Perrone et al., 1998), and in two cases, the mutant gene products have been identified. The PF9/IDA1 locus corresponds to the Dhc1 gene, which encodes the 1α Dhc (Myster et al., 1997), whereas the IDA7 locus corresponds to the structural gene for the WD-repeat containing IC140 (Perrone et al., 1998). By introducing constructs of both genes into the appropriate mutant backgrounds, we have been able to identify regions within these polypeptides required for the reassembly of I1 subunits and the restoration of I1 motor activity (Perrone et al., 1998; Myster et al., 1999).

In this study, we have characterized the gene encoding the second Dhc of the complex, the 1β Dhc, to assess its role in the assembly and targeting of the I1 complex. Using an I1 mutant strain obtained by insertional mutagenesis, we have shown by complementation analysis that the mutation is an allele (ida2-6) at the IDA2 locus. The recovery of genomic DNA flanking the site of plasmid insertion has further demonstrated that the ida2-6 mutation is a defect in the Dhc10 gene. Transformation of ida2 with truncated constructs of the Dhc10 gene can partially rescue the mutant defects. The 1β Dhc fragment encoded by the truncated transgene represents <20% of the full-length Dhc, yet it still supports the assembly of other I1 complex subunits onto the axoneme. High-resolution structural analysis of wild-type and mutant axonemes has revealed the location of the missing 1β Dhc motor domain within the structure of the I1 complex. This work, together with our previous study of the Dhc1 gene (Myster et al., 1999), has allowed us to identify the location of the polypeptides that form the major structural domains of the I1 complex in situ and to further define the regions of the Dhc that are required for complex assembly and activity. Given the similarity of the 1β Dhc to other Dhc isoforms, our findings also have implications for the assembly of other dynein complexes.

MATERIALS AND METHODS

Cell Culture, Mutant Strains, and Genetic Analyses

The strains used in this study are listed in Table 1. All strains were maintained as vegetatively growing cultures (Myster et al., 1997, 1999; Perrone et al., 1998). The 27B3 strain (ida2-6) was isolated by David Mitchell (State University of New York Medical Center, Syracuse, NY) after transformation of a nit− strain (nit1-305) with the pMN24 plasmid containing the wild-type nitrate reductase gene (NIT1). 27B3 was identified as a potential I1 mutant as described by Perrone et al. (1998). J6H9 (ida2-7) was isolated by Gerald Rupp in our laboratory after transformation of a nit− strain (A54-e18) with a smaller NIT1 plasmid, pMN56.

Table 1.

Characteristics of motility mutants used in this study

Strain name	Dynein defect	Motility	Swimming velocity (μm/s, n > 40)	Ability to phototax	Reference
Wild type (137c)	None	Wild type	144.2  ± 17.1a	Yes	Harris, 1989
pf28	Outer arm	Slow, jerky swimming	51.5  ± 6.9b	Yes	Mitchell and Rosenbaum, 1985
pf9-2	I1 complex	Slow, smooth swimming	73.4  ± 12.4b	No	Porter et al, 1992
pf9-3	Il complex	Slow, smooth swimming	72.3  ± 14.1b	No	Myster et al., 1997
ida3-1	I1 complex	Slow, smooth swimming	77.3  ± 11.0c	No	Kamiya et al., 1991
ida7-1	I1 complex	Slow, smooth swimming	81.5  ± 14.0a	No	Perrone et al., 1998
ida2-1	I1 complex	Slow, smooth swimming	77.7  ± 15.2c	No	Kamiya et al., 1991
ida2-6 (27B3)	I1 complex	Slow, smooth swimming	77.6  ± 15.4a	No	Perrone et al., 1998; this study
ida2-6::λC (D11)	1β Dhc	Faster swimming	107.9  ± 15.3	Yes	This study
ida2-6::λH (9A)	1β Dhc	Faster swimming	NDd	Yes	This study
ida2-7 (J6H9)	I1 complex	Slow, smooth swimming	53.7  ± 10.7	No	This study
ida2-7::pCAP1 (B3)	1β Dhc	Faster swimming	91.8  ± 12.6	Yes	This study
ida2-7::pCAP3 (B6)	1β Dhc	Faster swimming	74.1  ± 9.9	Yes	This study

^a Velocity determined by Perrone et al. (1998). 

^b Velocity determined by Myster et al. (1999). 

^c Velocity determined by Kamiya et al. (1991). 

^d ND, not determined. 

To determine whether the motility defects in 27B3 were linked to the NIT1 plasmid used as the selectable marker, 27B3 was backcrossed to a nit− strain, and random progeny were analyzed for their ability to grow on selective medium and for their motility phenotypes. All 46 nit+ progeny had the same slow motility phenotype as the 27B3 strain, whereas all 52 nit− progeny had wild-type motility. These data suggested that the defect in 27B3 was the result of plasmid insertion into a motility gene.

To determine if the 27B3 mutation might be an allele at a previously identified I1 locus (i.e., PF9/IDA1, IDA2, IDA3, or IDA7), complementation tests were performed by constructing stable diploid cell lines with the auxotrophic markers arg2 and arg7, as described previously (Ebersold, 1967; Perrone et al., 1998).

Analysis of Motility

Motility phenotypes and swimming velocities were assessed using phase-contrast microscopy and video recordings of live cells (Porter et al., 1992; Perrone et al., 1998). The ability of wild-type and mutant strains to undergo phototaxis was determined using both a tube-based assay (King and Dutcher, 1997) and a microtiter dish–based assay (Myster et al., 1999). A strain was designated phototaxis positive if the majority of swimming cells were concentrated on the lighted side of the tube or microtiter well.

Southern Blot and Northern Blot Analyses

DNA and RNA isolation, restriction enzyme digests, agarose gels, Southern blots, and Northern blots were performed as described previously (Porter et al., 1996, 1999; Myster et al., 1997; Perrone et al., 1998) with minor modifications to our Northern blot protocol to improve sensitivity. Gels were blotted onto a Brightstar (Ambion, Austin, TX) membrane, and Northern blots were prehybridized and hybridized in an Ultrahybe solution (Ambion) containing 100 μg/ml salmon sperm DNA.

Restriction Fragment Length Polymorphism Mapping

To place Dhc10 on the genetic map, the ∼150-base pair (bp) PCR product (Porter et al., 1999) was first used as a probe on genomic Southern blots to identify a restriction fragment length polymorphism (RFLP) between two polymorphic Chlamydomonas reinhardtii strains, 137c and S1-D2. An RFLP was easily observed using an EcoRI–XhoI digest. The Dhc10 probe was then hybridized to a series of mapping filters containing genomic DNA isolated from tetrad progeny of crosses between multiply marked C. reinhardtii strains and S1-D2. The segregation of the Dhc10 RFLP in the tetrad progeny was analyzed with respect to the segregation of more than 42 genetic and molecular markers (Porter et al., 1996).

Electron Microscopy and Image Analysis

Axonemes for electron microscopy were prepared as described previously (Porter et al., 1992). Longitudinal images were digitized, averaged, and compared with the use of the methods described by Mastronarde et al. (1992) and O'Toole et al. (1995). The final average for each sample contained at least 70 of the 96-nm axoneme repeats.

Protein Purification, SDS-PAGE, and Western Blot Procedures

Large-scale (20–40 l) culture of vegetative cells, the isolation of purified axonemes, and sucrose density gradient centrifugation of dynein extracts were performed as described previously (Porter et al., 1992; Myster et al., 1997, 1999; Perrone et al., 1998). The Dhcs in whole axonemes were resolved on 3–5% polyacrylamide, 3–8 M urea gradient gels (Kamiya et al., 1991). Sucrose density gradient fractions were analyzed on 5–15% polyacrylamide, 0–0.25 M glycerol gradient gels. Dynein LCs were analyzed on 7.0% polyacrylamide gels. Gels were stained with silver (Wray et al., 1981) or transferred to polyvinylidene difluoride or nitrocellulose membranes.

Western blots were incubated as described previously (Myster et al., 1997, 1999; Perrone et al., 1998) with the following antibodies: a rabbit polyclonal antibody generated against an IC140 fusion protein (Yang and Sale, 1998), a rabbit polyclonal antibody raised against a specific peptide in the 1α Dhc sequence (Myster et al., 1997), or a rabbit polyclonal antibody (R5205) generated against a Tctex1 fusion protein derived from a human cDNA library (King et al., 1996). Blots were developed with the use of an alkaline phosphatase–conjugated secondary antibody and either colorimetric (Sigma Chemical, St. Louis, MO) or chemiluminescent detection (Tropix, Bedford, MA).

Recovery of Genomic DNA Flanking the Site of Plasmid Insertion in ida2-6

To recover genomic DNA flanking the site of plasmid insertion, genomic DNA from wild type and ida2-6 was digested with KpnI, which does not cut within the pMN24 plasmid, and ClaI, which digests the pMN24 plasmid at a single site near the 5′ end of the NIT1 gene. After size fractionation on a 0.8% agarose gel and Southern blotting, the samples were hybridized with an ∼11.5-kilobase (kb) fragment of the NIT1 gene to identify those fragments derived from the endogenous NIT1 gene present in both samples as well as additional bands corresponding to the inserted NIT1 plasmids in ida2-6. Analysis of the restriction patterns identified a unique 3.4-kb KpnI–ClaI fragment in ida2-6 that was likely to contain genomic DNA flanking the site of plasmid insertion. (Any unique band in ida2-6 that decreased in size in the double digest must represent a KpnI site present in flanking genomic DNA.) The 3.4-kb KpnI–ClaI fragment was cloned by screening a size-fractionated mini library with the NIT1 sequence, and one plasmid, p27B3, was selected for further analysis.

To identify the region containing only flanking genomic DNA, p27B3 was digested with several enzymes and rehybridized with the NIT1 gene. A 350-bp KpnI–Sau3A fragment that failed to hybridize with the NIT1 sequence was identified as potential flanking genomic DNA. To verify that this fragment was derived from the region flanking the site of plasmid insertion, the 350-bp fragment was rehybridized to a Southern blot of wild-type and ida2-6 genomic DNA.

Characterization of Genomic Clones in the IDA2/Dhc10 Region

To recover a wild-type copy of the IDA2 gene, the 350-bp KpnI–Sau3A fragment of p27B3 was used to screen a large-insert, wild-type genomic strain (21gr) library constructed in λFIXII (Schnell and Lefebvre, 1993), as described previously (Porter et al., 1996, 1999; Myster et al., 1997), and seven overlapping phage clones were identified. Four additional clones were obtained by screening the library with the 150-bp product of the Dhc10 gene. Additional flanking genomic DNA was obtained using a reverse transcriptase (RT)-PCR product derived from one end of the phage walk to screen a Chlamydomonas bacterial artificial chromosome (BAC) library (Genome Systems, St. Louis, MO) and recover five overlapping BAC clones, as described previously (Myster et al., 1999).

To test the ability of the phage clones to rescue the motility defect in ida2-6, an ida2-6 arg2 strain was cotransformed with 1–3 μl of phage DNA and 2 μg of the BamHI-linearized plasmid pARG7.8 (Debuchy et al., 1989) by means of the glass bead–mediated transformation protocol (Kindle, 1990; Nelson and Lefebvre, 1995; Perrone et al.,1998). Arg+ transformants were selected by plating on Tris-acetate-phosphate medium lacking arginine. After growth for 7–10 d, transformant lines were picked into liquid Tris-acetate-phosphate medium and screened for rescue of the ida2-6 motility defect on a dissecting microscope. A total of 150–300 transformants were screened per clone. Rescued strains were restreaked for single colonies and rescored by phase-contrast microscopy.

To identify the minimum region required to rescue the ida2 mutant phenotype, a 17.1-kb XbaI fragment from phage clone C was ligated into pBluescript KS II to obtain the subclone pCAP1. pCAP2 was obtained by isolating an ∼4.6-kb BglI fragment from pCAP1 and ligating into a HincII-digested plasmid. pCAP3 was obtained by isolating an ∼14-kb BglII fragment from pCAP1 and ligating into a BamHI-digested plasmid. pCAP1 encodes up to amino acid residue 989 of the Dhc10 sequence, followed by the addition of 8 novel amino acids (ESTPFSEG); pCAP2 encodes up to residue 508, followed by 4 amino acids (GRYR); and pCAP3 encodes up to residue 811, followed by 2 amino acids (IH).

Sequencing the Dhc10 Gene

Selected subclones were sequenced by primer walking at the DNA sequencing facility at Iowa State University (Ames, IA). The sequence data were assembled and analyzed using the Genetics Computer Group (GCG; Madison, WI) software, version 9.0, and the MacVector Sequence Analysis Software, version 6.0 (International Biotechnologies, Rochester, NY). Potential ORFs and splice sites were identified with the use of codon usage tables (Nakamura et al., 1997) and the consensus donor and acceptor sequences found in Chlamydomonas nuclear genes (Mitchell and Brown, 1994; LeDizet and Piperno, 1995; Zhang, 1996). All splice junctions were confirmed by sequencing RT-PCR products amplified with gene-specific primers designed to span intron–exon boundaries, as described previously (Myster et al., 1999; Porter et al., 1999). We were unable to sequence a small region of genomic DNA (∼700 bp) in the middle of the 4.7-kb SacI subclone, but we were able to sequence a 354-bp RT-PCR product spanning the gap. The sequence of the Dhc10 transcription unit (∼25 kb) is available under three linked accession numbers (AJ242523–AJ242525).

The predicted amino acid sequence of the Dhc10 gene product was analyzed using the GCG programs Bestfit, Compare, Pileup, Dotplot, and Motifs, as described previously (Myster et al., 1999; Porter et al., 1999). The COILS program, version 2.2 (Lupus et al., 1991; Lupus, 1996), was used to analyze regions of the amino acid sequence for their potential to form α-helical coiled coils.

Generation of a Specific Antibody for the Dhc10 Gene Product

Sequence alignment of Chlamydomonas axonemal Dhcs indicated that the Dhc10 gene product shared a high degree of sequence similarity with other Dhcs. However, small regions of sequence divergence could be identified in the N-terminal third of the Dhc. Two regions were chosen as sites for peptide synthesis and antibody production, residues 1–15 (MEPGDEGKGHQLTAD) and residues 945–959 (VALQLTDKQRRDMED). The peptides were coupled to keyhole limpet hemocyanin and injected separately into rabbits by Research Genetics (Huntsville, AL). A strong response against the second peptide was detected by ELISA. The antisera were pooled, affinity-purified on a peptide column, and then affinity-purified on Western blots of inner arm dynein extracts, as described previously (Myster et al., 1997, 1999; Perrone et al., 1998).

RESULTS

Identification of a New Axonemal Dhc Gene Linked to the IDA2 Locus

Recent studies have identified four new Dhc sequences in Chlamydomonas, and comparison with previously identified Dhc genes suggested that two of the genes (cDhc1a and cDhc1b) encode cytoplasmic Dhcs, whereas the other two sequences (Dhc10 and Dhc11) share homology with axonemal Dhc sequences (Porter et al., 1999). In particular, Dhc10 appears to be most closely related to Dhc1, which encodes the 1α Dhc of the inner dynein arm I1 complex (Figure 1A). This homology suggested that Dhc10 might encode the second Dhc of the I1 complex, known as the 1β Dhc. Mutations in Dhc10, therefore, might be expected to disrupt the assembly of the I1 complex and produce an I1 mutant phenotype.

Figure 1.

Dendrogram of Dhc sequences and mapping of the Dhc10 gene. (A) The deduced amino acid sequences surrounding the conserved ATP hydrolytic site (P-loop 1) were aligned with the GCG program Pileup. The subset of Dhc sequences most closely related to Dhc1 is shown here. The Chlamydomonas reinhardtii (Cr) Dhc1 (U61364, AF243806) and Dhc10 (AJ132799) sequences are marked by asterisks. The GenBank accession numbers for the other sequences are as follows: Tripneustes gratilla (Tg), U03976, U03973; Rattus norvegicus (Rn), D26502, D26493; Mus musculus (Mm), Z83812, Z83813; Paramecium tetraurelia (Pt) L18803; Drosophila melanogaster (Dm) L23200. (B) Genetic map location of the Dhc10 gene. The Dhc10 gene was mapped to linkage group XV based on linkage to the genetic marker ida2 (<2.6 cM, no recombinants in 18 tetrads) and the molecular marker Dhc7 (∼7.4 cM).

To determine whether the Dhc10 gene is linked to any previously identified I1 mutations, we used RFLP mapping procedures to place Dhc10 on the genetic map of Chlamydomonas. The RFLP data indicated that Dhc10 maps to linkage group XV, ∼7.4 cM from another Dhc locus, Dhc7, and less than 2.6 cM from the ida2 mutation (Figure 1B). Given recent estimates on the physical relationship between the molecular and genetic maps (Silflow, 1998), these results placed Dhc10 within ∼260 kb of the IDA2 locus.

Isolation of a Tagged ida2 Allele by Insertional Mutagenesis

Although the RFLP data indicated that Dhc10 and IDA2 are closely linked, they were not sufficient to identify them as the same locus. However, it is possible to isolate tagged motility mutations in Chlamydomonas with the use of insertional mutagenesis procedures (Tam and Lefebvre, 1993) and to screen genomic DNA for defects in specific genes. We have used this approach to identify mutations in the Dhc1, IC140, and cDhc1b genes (Myster et al., 1997; Perrone et al., 1998; Porter et al., 1999). During the course of these studies, we recovered a new strain, 27B3, with an I1-like motility phenotype by virtue of its slow swimming speed and its failure to phototax (Perrone et al., 1998). Cosegregation tests have since confirmed that the 27B3 motility phenotype is linked to the inserted NIT1 plasmid used as a selectable marker (see MATERIALS AND METHODS). Complementation tests have further demonstrated that 27B3 represents a new mutant allele at the IDA2 locus, now referred to as ida2-6. Diploid strains containing 27B3 and the pf9-2, ida3-1, or ida7-1 mutation had wild-type motility, whereas diploid strains containing both 27B3 and the ida2-1 mutation had the slow-motility phenotype characteristic of the parent strains. Therefore, we characterized the phenotype of the ida2-6 mutant to determine if the assembly of the I1 complex is disrupted and whether the mutant phenotype is a result of a defect in the Dhc10 gene.

ida2-6 Axonemes Lack the I1 Dynein Complex

To determine if the I1 complex is defective in ida2-6 axonemes, we isolated axonemes from both ida2-6 and wild type, fixed and embedded them for electron microscopy, and analyzed longitudinal sections with the use of image-averaging procedures (Mastronarde et al., 1992; O'Toole et al., 1995). Previous studies have shown that the I1 complex is a tri-lobed structure that occupies a specific position proximal to the first radial spoke and repeats every 96 nm along the length of the axoneme (Piperno et al., 1990; Mastronarde et al., 1992). Figure 2A shows an average from several wild-type axonemes. The relative positions of the radial spokes, the outer dynein arms, and the inner dynein arm structures are indicated in the model in Figure 2B. The I1 complex corresponds to lobes 1, 2, and 3; these structures are present in the wild-type axonemes (Figure 2A) but appear to be missing in the ida2-6 axonemes (Figure 2C). The difference plot (Figure 2D) confirms that ida2-6 axonemes lack the I1 structure and further demonstrates that this is the only significant difference between the ida2-6 and wild-type images.

Figure 2.

Inner arm structure in wild-type and mutant strains. Longitudinal images of the 96-nm axoneme repeat from wild type, ida2-6, and the rescued ida2-6 strain, D11, are shown here. The averages for wild type (A), ida2-6 (C), and the rescued ida2-6 strain (E) are based on 8, 10, and 9 individual axonemes and 74, 87, and 85 axoneme repeats, respectively. (B) Model of the 96-nm axoneme repeat with the major lobes of density in the inner arm region labeled 1–10. The outer arms (OA) are on top, the inner arm (IA) region is below, and the proximal and distal radial spokes are labeled S1 and S2, respectively. (D) Difference plot between wild type and ida2-6. (F) Difference plot between wild type and the rescued ida2-6 strain. The difference plots are derived from a pixel-by-pixel analysis of variance between averages, where statistically significant differences between the averages for two strains are indicated by the gray areas. Differences not significant at the 0.005 confidence level have been set to zero.

The I1 complex is composed of two Dhcs (1α and 1β), three ICs, and three LCs (Piperno et al., 1990; Porter et al., 1992; Harrison et al., 1998). To ascertain whether all of the I1 complex polypeptides are missing in ida2-6, axonemes were isolated from wild type, ida2-6, and pf9-2, a previously characterized I1 mutant (Porter et al., 1992), and analyzed by SDS-PAGE. In Figure 3A, the 1α and 1β Dhcs can be seen as two faint bands migrating between the outer arm β and γ Dhcs in wild-type axonemes. Both the 1α and 1β Dhcs appear to be missing in the pf9-2 and ida2-6 axonemes. To examine the dynein ICs, crude dynein extracts from wild type, ida2-6, pf9-2, and the outer arm mutant pf28 were fractionated by sucrose density gradient centrifugation and analyzed on 5–15% polyacrylamide gels. As shown in Figure 3B, the three ICs are clearly visible in the wild-type and pf28 samples but appear to be missing or reduced in the pf9-2 and ida2-6 extracts. To analyze the dynein LCs, Western blots of whole axonemes were probed with an antibody specific for the 14-kDa LC, Tctex1, which is also one of the t haplotype gene products (Harrison et al., 1998). Tctex1 was present in wild-type axonemes but appeared to be missing or reduced in the ida2-6 sample (Figure 3C). Therefore, most of the I1 dynein subunits are not assembled into the ida2-6 axonemes.

Figure 3.

Dynein polypeptide defects in ida2-6 axonemes. (A) The high-molecular-weight region of a 3–5% polyacrylamide, 3–8 M urea gradient gel that was loaded with whole axonemes from wild type, pf9-2, and ida2-6 is shown here. The 1α and 1β Dhcs of the I1 complex, which migrate between the outer arm β and γ Dhcs, are indicated by the asterisks on the right. Both pf9-2 and ida2-6 lack the 1α and 1β Dhcs. (B) Sucrose density gradient centrifugation of dynein extracts from pf28, wild type, ida2-6, and pf9-3. All sucrose gradient fractions were loaded on 5–15% polyacrylamide gels. The subunits of the I1 complex sediment in the 18–19S region, along with contaminating outer arm components (Piperno et al., 1990; Porter et al., 1992). Only the 18–19S peaks are shown here. IC140, IC138, and IC110 are visible in pf28 (which lacks the outer arm components) and wild type. (C) Western blot of wild-type and mutant axonemes probed with the antibody to the Tctex1 LC.

Cloning the IDA2 Locus

Because the ida2-6 mutation was generated by plasmid insertion, we expected to see a deletion or rearrangement of ida2-6 genomic DNA on Southern blots hybridized with the Dhc10 sequence. However, we did not detect any obvious polymorphisms in ida2-6 using the 150-bp PCR product as a probe. Therefore, to identify the gene that was disrupted in ida2-6, we recovered genomic DNA flanking the site of plasmid insertion. Southern blot analysis revealed that at least three copies of the NIT1 plasmid had integrated into ida2-6 genomic DNA, and all three copies cosegregated with the slow-swimming phenotype in the progeny (Figure 4A). Using the NIT1 plasmid as a probe, we then identified a 3.4-kb KpnI–ClaI restriction fragment in ida2-6 that was likely to contain both plasmid sequence and flanking genomic DNA (see MATERIALS AND METHODS and Figure 4B). The 3.4-kb fragment was subsequently cloned by screening a size-fractionated mini library with the NIT1 sequence and further characterized to identify a 350-bp KpnI–Sau3A fragment containing only ida2-6 genomic DNA. Southern blot analysis confirmed that this 350-bp fragment was derived from the region flanking the site of plasmid insertion in ida2-6 (Figure 4C).

Figure 4.

Recovery of genomic DNA flanking the site of the ida2-6 mutation. (A) Cosegregation of plasmid sequences with the ida2-6 motility defect. Shown here is an autoradiogram of a genomic Southern blot of SphI-digested DNA from a wild-type strain, ida2-6, and the slow-swimming progeny from four successive generations (F1, F2, F3, and F4) of ida2-6 × nit1 crosses. The blot was hybridized with the vector sequence from the NIT1 plasmid. Three copies of the plasmid are present in ida2-6 and all of the slow-swimming progeny. (B) Diagram of the plasmid insertion in ida2-6 and the strategy used to clone genomic DNA flanking the site of insertion. Genomic DNA from ida2-6 was digested with ClaI, which cuts once per plasmid, and KpnI, which does not cut within the plasmid sequences. Southern blots probed with the NIT1 plasmid identified a unique 3.4-kb KpnI–ClaI fragment in ida2-6 containing genomic DNA adjacent to the inserted plasmids. This fragment was cloned from a size-fractionated mini library to yield the plasmid p27B3 (see MATERIALS AND METHODS). p27B3 was then digested with KpnI–Sau3A to release a 350-bp fragment containing only genomic DNA. (C) Genomic Southern blot of wild-type and ida2-6 DNA probed with the 350-bp KpnI–Sau3A fragment. A RFLP is clearly visible between the two strains, confirming that the genomic DNA is close to the site of plasmid insertion in ida2-6. (D) Northern blot of wild-type and ida2-6 RNA hybridized with a 3.2-kb SacI fragment from the end of phage clone C. A single large (>13 kb) transcript is visible in wild type, and its expression is increased after deflagellation (compare 0 and 45 min). The Dhc10 transcript is significantly smaller in ida2-6 (∼4 kb), but it is still expressed at a high level after deflagellation. This transcript also hybridizes with a NIT8 probe (see text). To ensure that equal quantities of RNA were loaded in all lanes, Northern blots were also hybridized with the CRY1 gene encoding the ribosomal S14 subunit.

The 350-bp fragment was then used to screen a large-insert, wild-type genomic library, and seven overlapping phage clones spanning ∼21.5 kb of genomic DNA were recovered (Figure 5, A and B). The library was also screened with the 150-bp PCR fragment of Dhc10, and four phage clones spanning ∼27.8 kb of genomic DNA were recovered (Figure 5C). Comparison of the restriction maps and Southern blots probed with selected subclones confirmed that the two sets of clones overlapped.

Figure 5.

Recovery of the IDA2/Dhc10 transcription unit. (A) Shown on the top is a partial restriction map of the IDA2/Dhc10 genomic region in wild type. Shown below is a diagram of the genomic region in two different insertional mutants, ida2-6 and ida2-7. Vertical marks indicate SacI sites; the approximate sizes of the restriction fragments are also indicated. The shaded rectangle corresponds to the 350-bp KpnI–Sau3A probe recovered as the genomic fragment flanking the site of plasmid insertion in ida2-6; the small open rectangle corresponds to the 150-bp Dhc10 PCR product; and the rectangle with hash marks indicates the 3.2-kb SacI fragment isolated from the end of phage clone C. The approximate location of the Dhc10 transcription unit is illustrated by the arrow. The ida2-6 mutation was generated by plasmid insertion into the Dhc10 coding region. The plasmid insertion event in ida2-7 was associated with a large (>20 kb) deletion of the Dhc10 transcription unit. (B) Wild-type phage clones recovered using the 350-bp KpnI–Sau3A probe are shown with respect to the genomic region above. Clones C, B, and H were tested for their ability to rescue the motility defect of ida2-6 by cotransformation. Clone C encodes up to residue 989 of the 1β Dhc sequence, and clone H encodes up to residue 1695. The resulting N-terminal fragments are predicted to have molecular masses of ∼113 and 195 kDa, respectively, and completely lack the dynein motor domain. (C) Phage clones recovered using the 150-bp Dhc10 PCR fragment. (D) A BAC clone that was recovered using an RT-PCR product from the 3′ end of the Dhc10 gene. (E) Diagram of smaller subclones used in cotransformation experiments to rescue ida2-6 and ida2-7. These subclones are drawn to scale with respect to parts A–D. pCAP1 encodes up to residue 989 of the 1β Dhc; pCAP2 encodes up to residue 508; and pCAP3 encodes up to residue 811. (F) Diagram of the intron/exon structure of the Dhc10 gene as determined by RT-PCR. The exons are indicated by the open rectangles, and the introns are indicated by solid lines. A TATA box sequence located ∼140 bp upstream of the proposed translation start site is also shown. The identification of the translation start site was based on the recovery of an RT-PCR product using a forward primer downstream of the TATA box sequence and a reverse primer from exon 3. This RT-PCR product contained stop codons in all three frames preceding the proposed start codon. The region upstream of the TATA box sequence also contains six tub box sequences thought to enhance the expression of flagellar genes (Davies and Grossman, 1994). The 3′ end of the gene is contained within the 7.2-kb SacI subclone (see A). Two putative polyadenylation signals (TGTAA) were identified 116 and 445 bp downstream of the proposed stop site. Also shown are the sites that correspond to the junction between Dhc10 and the plasmid insertions in ida2-6, and the 3′ end of the Dhc10 sequence in phage clone C and pCAP1. The scale of this diagram is enlarged relative to parts A–E.

To determine how the Dhc10 gene was disrupted by the plasmid-insertion event in ida2-6, we used the phage clones to characterize both wild-type and mutant strains on Southern and Northern blots. Southern blots of wild-type and ida2-6 genomic DNA were probed with selected subclones and analyzed for deletions and/or rearrangements within the IDA2/Dhc10 region. The results are summarized in Figure 5A; all three copies of the NIT1 plasmid in ida2-6 were inserted into a region corresponding to the 4.7-kb SacI fragment in wild type, ∼6 kb upstream from the region encoding the first Dhc10 phosphate-binding motif (P-loop). Surprisingly, this plasmid-insertion event was not accompanied by a large deletion of genomic DNA in ida2-6, which explained why we previously failed to see an RFLP with the 150-bp P-loop probe for Dhc10.

To investigate how many transcripts are present in the IDA2/DHC10 region and how they might be altered in the ida2-6 mutant, we isolated total RNA from wild type and ida2-6 and analyzed the transcripts on Northern blots. Because flagellar transcripts are typically up-regulated in response to deflagellation, we isolated RNA both before and after deflagellation and then hybridized the Northern blots with subclones covering the cloned region. This analysis indicated that the Dhc10 transcription unit spans ∼25 kb of genomic DNA (Figure 5A) and encodes an ∼13-kb transcript in wild-type whose expression is enhanced by deflagellation (Figure 4D). Moreover, when we used a probe located close to the site of plasmid insertion in ida2-6, it became apparent that the Dhc10 transcript is significantly smaller (∼4 kb) in the ida2-6 mutant (Figure 4D). The plasmid insertion in ida2-6, therefore, disrupted the Dhc10 transcription unit upstream of the region encoding the first P-loop (Figure 5A).

Sequence Analysis of Dhc10

To further characterize the Dhc10 gene and its encoded product, we sequenced the Dhc10 transcription unit (EMBL accession numbers AJ242523–AJ242525). Putative exons were identified using codon preference programs and the conserved consensus sequences for Chlamydomonas splice donor and acceptor sites. All predicted splice sites were confirmed by sequence analysis of RT-PCR products (see MATERIALS AND METHODS). The predicted structure of the Dhc10 transcription unit is shown in Figure 5F. It includes ∼1 kb of sequence upstream from the proposed translation start site, ∼24 kb of coding region containing 53 exons, and ∼1.3 kb downstream from the proposed stop codon.

The deduced amino acid sequence of the Dhc10 gene product is shown in Figure 6. It contains 4513 residues and corresponds to a polypeptide of 510,628 Da. Analysis of the predicted amino acid sequence with the GCG program Motifs revealed the presence of three P-loops that conform to the consensus sequence GXXXXGKT/S for a nucleotide-binding site (Walker et al., 1982). A fourth P-loop that deviates from the consensus sequence (GVGGSGRK) was identified by alignment with other Dhc sequences. These four P-loops are spaced at ∼300-amino acid intervals within the central region of the Dhc (Figure 6), similar to those found in other Dhcs (reviewed by Gibbons, 1995). Recent sequence analysis has suggested that all Dhc sequences may contain two additional degenerate ATP-binding sites in the C-terminal region (Neuwald et al., 1999). Alignment with the cytoplasmic Dhc from Dictyostelium (Koonce et al., 1992) has confirmed that these sites are conserved in the Dhc10 gene product (Figure 6).

Figure 6.

Predicted amino acid sequence of the 1β Dhc. The deduced amino acid sequence (residues 1–4513) of the Dhc10 gene product is shown here. The four regions corresponding to P-loops 1–4 are underlined. The regions containing the degenerate P-loops identified by comparison with the cytoplasmic Dhc from Dictyostelium (Neuwald et al., 1999) are indicated by the dashed boxes. The peptide sequence (VALQTDKQRRDMED) used to generate a specific antibody (residues 945–959) and the C-terminal amino acid residues encoded by the different Dhc10 constructs used for cotransformation are also shown.

To determine whether the Dhc10 gene product contains the structural features seen in other Dhcs, we analyzed the deduced amino acid sequence using the COILS program (Lupus et al., 1991; Lupus, 1996). This analysis (Figure 7) identified two regions with a high probability of forming α-helical coiled coils on either side of the central catalytic domain (residues 1535–1570 and 3088–3178), consistent with the structural predictions reported for other Dhcs (Mitchell and Brown, 1994, 1997). By analogy with other Dhc sequences, the C-terminal coiled-coil domains probably correspond to the stalk structure identified as the B-link (Goodenough et al., 1987) and presumably contain the microtubule-binding site (Gee et al., 1997; Koonce, 1997). The Dhc10 gene product also contains a predicted coiled-coil domain close to the N terminus (residues 192–227); this domain appears to be unique to the Dhc10 sequence.

Figure 7.

Diagram depicting regions of high coiled-coil probability for the Dhc10 gene product. The plot was generated using the COILS program (Lupus et al., 1991; Lupus 1996) with a window size of 28 and the MTIDK matrix. The approximate positions of the four central P-loops are indicated by the arrowheads.

Comparison of the predicted amino acid sequence of the Dhc10 gene product with other axonemal Dhc sequences in Chlamydomonas indicates that they share significant sequence identity and similarity throughout their lengths. The greatest similarity is seen with the 1α Dhc of the I1 complex (34% identity and 62.6% similarity), and this homology extends into the N-terminal region (24% identity and 54.4% similarity within the first 1500 residues). Alignments with the programs Pileup and Clustal W have confirmed the presence of conserved domains within the N-terminal region but have also revealed several short stretches of sequence divergence. One of these regions, corresponding to residues 945–959 of the Dhc10 gene product, was used to design a specific peptide epitope for antibody production (see MATERIALS AND METHODS). This region was also chosen because we were previously successful in obtaining a Dhc-specific antibody to the analogous region in the 1α Dhc sequence (Myster et al., 1997).

Dhc10 Encodes the 1β Dhc of the I1 Complex

Characterization of the affinity-purified Dhc10 peptide antibody by Western blotting (Figure 8A) shows that Dhc10 encodes the 1β Dhc of the I1 complex. The Dhc10 antibody recognizes a Dhc present in wild-type and outer arm mutant axonemes but absent in I1 mutant axonemes (Figure 8B). To determine if the Dhc10 antibody is specific for one of the two I1 Dhcs, we analyzed purified I1 dynein complexes on 5% polyacrylamide gels to resolve the 1α and 1β Dhcs. Duplicate immunoblots of the I1 complex were then probed with affinity-purified Dhc1 and Dhc10 antibodies. As shown in Figure 8A, the Dhc1 antibody recognizes only with the 1α Dhc, as described previously (Myster et al., 1997), whereas the Dhc10 antibody recognizes only with the 1β Dhc. These results demonstrate that Dhc10 is the structural gene for the 1β Dhc and that defects in the 1β Dhc are the basis of the ida2 mutant phenotype.

Figure 8.

The Dhc10 antibody identifies the 1β Dhc as the gene product of the IDA2/Dhc10 locus. (A) Specificity of I1 Dhc antibodies. The I1 complex was purified by sucrose density gradient centrifugation and analyzed on 5% polyacrylamide gels to resolve the 1α and 1β Dhcs (left lane). Duplicate immunoblots were probed with an affinity-purified Dhc1 antibody (middle lane) and an affinity-purified Dhc10 antibody (right lane). (B) Reassembly of I1 complex subunits in Dhc10 transformants. Axonemes were isolated from wild-type and mutant strains, run on 5% polyacrylamide gels, and blotted to either nitrocellulose or polyvinylidene difluoride. One strip containing only the Dhc region was probed with the Dhc1 antibody to detect the 1α Dhc (top blot). A second strip containing the region above 85 kDa was probed with the Dhc10 antibody to detect the 1β Dhc and related N-terminal fragments (middle blot). A third strip containing the 140-kDa region was probed with the IC140 antibody (bottom blot). ida2-6 and ida2-7 are two different insertional mutants (see Figure 5A). Both strains were cotransformed with Dhc10 constructs encoding N-terminal 1β Dhc fragments of varying lengths (see Figure 5, B and E). Note that the 1β Dhc fragment encoded by the pCAP3 subclone does not contain the peptide epitope used to generate the Dhc10 antibody, so the ∼92-kDa fragment cannot be detected by Western blot analysis of axoneme polypeptides.

ida2-6 Encodes a Truncated 1β Dhc Fusion Protein That Inhibits I1 Complex Assembly

The Northern blot shown in Figure 4D revealed that the ida2-6 mutant generates a stable but truncated Dhc10 transcript. This observation raised the possibility that the 1β Dhc fragment in ida2-6 might lack a critical domain required for I1 complex assembly. To determine the nature of the truncated transcript in ida2-6, we sequenced the plasmid containing the junction between the Dhc10 gene and the inserted pMN24 sequences (Figure 4B). This sequence revealed that the Dhc10 gene is disrupted in intron 15 (Figure 5F) by the insertion of 63 bp of vector followed by a sequence that corresponds to the 3′ end of the NIT8 gene present in the pMN24 plasmid (Zhang, 1996). The junction between the vector and the 3′ end of the NIT8 gene forms a splice acceptor sequence, which could generate a transcript that fuses the 5′ end of Dhc10 in-frame with the 3′ end of NIT8. Hybridization of the Northern blot shown in Figure 4D with the NIT8 sequence has confirmed the presence of the hybrid transcript in ida2-6. The ida2-6 gene product, therefore, is a fusion protein containing the first 827 residues (∼94 kDa) of the 1β Dhc fused to the last 132 residues of the NIT8 gene product.

Previous work on the expression of other Dhc sequences has indicated that N-terminal fragments between 140 and 160 kDa are capable of complex assembly (Sakakibara et al., 1993; Koonce and Knecht, 1998; Iyadurai et al., 1999; Myster et al., 1999). Therefore, it was unclear whether the ida2-6 mutation prevents the assembly of the I1 complex into the axoneme because the mutant 1β Dhc fragment lacks a specific domain required for dynein complex formation or because the NIT8 sequence at the end of the 1β Dhc fragment destabilizes the complex. Comparison with another ida2 mutant has now indicated that the 1β Dhc in ida2-6 may retain some partial activity. Southern and Northern blot analysis of the ida2-7 mutant has shown that the Dhc10 transcription unit is completely deleted in this strain (Figure 5A), and interestingly, its motility defect is even greater than that observed in ida2-6 (Table 1). Moreover, when isolated axonemes from the two ida2 mutants are compared on Western blots with the use of chemiluminescent detection procedures, small but detectable amounts of I1 complex polypeptides can be seen in the ida2-6 axonemes but not in the ida2-7 axonemes (Figure 8B). These results suggest that the 1β Dhc fragment in ida2-6 can assemble into an I1 complex, but the mutant I1 complex is unstable and cannot assemble efficiently into the flagellar axoneme.

Partial Rescue of ida2 by Transformation with Dhc10 Constructs Encoding 1β Dhc Fragments

To identify the 1β Dhc domains required for I1 complex assembly, we crossed the ida2 strains into an arg7 mutant background and then cotransformed the double mutants with several of the clones shown in Figure 5 and the selectable marker ARG7. Arg+ transformants were then picked and screened for rescue of the ida2 motility defect by light microscopy.

With each phage clone tested (Figure 5B), we recovered several cotransformants that appeared to swim more quickly than the original ida2-6 mutant. The rescued cotransformants also regained the ability to phototax (Table 1). However, when we measured the swimming velocities of the rescued strains, it was clear that the motility phenotypes of the rescued cotransformants were not completely wild type. For example, the swimming velocity of the rescued ida2-6 strain D11, which was cotransformed with phage clone C, was significantly faster than that of ida2-6 (107.9 ± 15.3 μm/s versus 77.6 ± 15.4 μm/s) but still slower than that of wild type (144.2 ± 17.1 μm/s). This partial rescue of the motility defects would be consistent with the reassembly of a modified I1 complex lacking one of the two Dhc motor domains. Therefore, we sequenced the ends of the rescuing phage clones to determine the sizes of the predicted 1β Dhc fragments (Figure 5B). We also subcloned the rescuing DNA and tested progressively smaller Dhc10 constructs for their ability to rescue the ida2 motility defects (Figure 5E). Subclones pCAP1 and pCAP3, encoding 989 and 811 residues, respectively, could partially rescue the motility defects of the ida2 strains (Table 1), whereas subclone pCAP2, encoding only 508 residues, could not (Figure 5E). These results indicate that the region of the 1β Dhc between residues 508 and 811 is critical for I1 complex assembly and recovery of motility.

To verify that the Dhc10 constructs encode functional 1β Dhc fragments, axonemes were isolated from several of the rescued transformants and analyzed on Western blots probed with I1 complex antibodies (Figure 8B). Immunoblots probed with the Dhc10-specific antibody demonstrated that the axonemes from the rescued strains did contain 1β Dhc fragments of the expected sizes (see Figure 8 legend). Probing the immunoblots with antibodies to other I1 complex subunits also showed that other components of the complex had been restored. Given that none of the Dhc10 constructs tested encoded the 1β Dhc motor domain, we reasoned that the increased motility of the rescued strains must be due to the assembly of a modified I1 complex containing an N-terminal fragment of the 1β Dhc and a full-length 1α Dhc with an active motor domain.

To directly demonstrate the reassembly of the 1β Dhc fragment into an I1 complex, we prepared crude dynein extracts from wild type and one of the rescued ida2-6 strains and then fractionated the extracts by sucrose density gradient centrifugation. All of the gradient fractions were analyzed on 5–15% polyacrylamide gradient gels and duplicate immunoblots. Figure 9 shows the fractions that contain I1 complex subunits in both wild type and the rescued ida2-6 strain, D11. IC140, IC138, and IC110 are clearly visible at 18–19S in fractions 6 and 7 of the wild-type extracts (Figure 9A). However, in the rescued ida2-6 extracts, the three ICs cosediment at 16S, in fractions 8–10, indicating that the sedimentation behavior of the I1 complex has been altered (Figure 9B). In addition, a novel band migrating just above IC110 cosediments with the I1 subunits in the rescued ida2-6 extracts (Figure 9B). The size of this novel band is consistent with the predicted size of the truncated 1β Dhc fragment (∼113 kDa) seen in the rescued strain (Figure 8B). Duplicate immunoblots of these fractions probed with antibodies against the 1α Dhc, 1β Dhc, IC140, and Tctex1 demonstrate that all of the I1 subunits cosediment at 18–19S in wild-type extracts (Figure 9C) and at 16S in the rescued ida2-6 extracts (Figure 9D). Furthermore, the novel band seen at ∼113 kDa in the rescued strain is recognized by the Dhc10-specific antibody, confirming the presence of the N-terminal 1β Dhc fragment in the modified I1 complex.

Figure 9.

Formation of an I1 complex in the rescued ida2 strain. Dynein extracts from wild type and the rescued ida2-6 strain, D11, were subjected to sucrose density gradient centrifugation. The resulting fractions were run on duplicate 5–15% polyacrylamide gradient gels, and the gels were either silver stained or transferred to Western blots. (A) Gel of wild-type fractions sedimenting in the 19S region. The 110-, 138-, and 140-kDa ICs of the I1 complex (indicated by the solid circles) cosediment in fractions 6 and 7. (B) Gel of the rescued ida2-6 strain, D11. The I1 ICs cosediment at 16S in fractions 9 and 10. A novel polypeptide migrating at ∼113 kDa in fractions 9 and 10 is indicated by a diamond. This polypeptide is not present in the wild-type gradient. (C) Blots of the wild-type fractions shown in A were probed with antibodies against the 1α Dhc (top panel), IC140 (second panel), 1β Dhc (third panel), and Tctex1 LC (bottom panel). (D) Blots of the rescued ida2-6 fractions shown in B were probed with I1 complex antibodies as described in C. Note the presence of the 1β Dhc fragment at ∼113 kDa.

Localization of the 1β Motor Domain within the I1 Structure

To explore the effect of the truncated 1β Dhc on the structure of the I1 complex, we isolated axonemes from the rescued ida2-6 strain, fixed and embedded them for electron microscopy, and analyzed longitudinal thin sections using image-averaging procedures (O'Toole et al., 1995). As shown in Figure 2E, the average of the rescued ida2-6 axonemes indicates that densities of the I1 complex have been partially restored. The details can be seen more easily in the difference plot between wild-type and rescued ida2-6 axonemes (Figure 2F). The structures represented by lobes 2 and 3 have reappeared in the rescued strain, but lobe 1 is still missing. A small loss of density is also seen extending from lobe 1 to lobe 3. These results suggest that the central and C-terminal 75% of the 1β Dhc, which corresponds to the missing motor domain in the mutant, forms the globular head domain located in lobe 1. Furthermore, because lobe 2 has been identified as the site of the 1α Dhc motor domain (Myster et al., 1999), we conclude that lobe 3 is the site of the N-terminal regions of the Dhc and the associated IC and LC subunits of the I1 complex.

DISCUSSION

The IDA2 Locus Corresponds to a Dhc Gene Required for Flagellar Motility

In previous work, we identified a large family of Dhc genes in Chlamydomonas whose expression patterns were consistent with a role in axoneme assembly or motility (Porter et al., 1996, 1999). In this report, we now demonstrate that one of these genes, Dhc10, maps to the IDA2 locus (Figure 1) and plays an essential role in the assembly and activity of the I1 inner arm complex. Disruption of the Dhc10 gene by plasmid insertion (Figure 5) resulted in the formation of a truncated transcript (Figure 4) whose encoded gene product reduces the assembly of the I1 complex into a flagellar axoneme (Figures 2 and 3), leading to defects in the flagellar waveform, forward swimming velocity, and phototaxis (Table 1).

Sequence Analysis of the Dhc10 Transcription Unit

Dhc10 encodes a polypeptide that is very similar to other Chlamydomonas axonemal Dhcs (Mitchell and Brown, 1994; Wilkerson et al., 1994; Myster et al., 1999). For example, the central region of the polypeptide is bounded on both sides by regions predicted to form coiled-coil domains (Figure 7), and it also contains multiple P-loop sequences spaced at intervals similar to those observed in other Dhc sequences (Figure 6). The first P-loop is 100% identical to those found in other axonemal Dhcs, whereas the second and third P-loops are less well conserved. The fourth P-loop deviates from the consensus sequence found in other axonemal Dhcs (GVGGSGKQ) in that the terminal lysine and glutamine residues are replaced by an arginine and a lysine, respectively (GVGGSGRK). Two degenerate P-loop–like repeats were also found downstream from the C-terminal coiled-coil domain (Figure 6), as described previously for the cytoplasmic Dhc (Neuwald et al., 1999). Comparisons with Dhc genes identified in other organisms (Figure 1) have identified homologues of the Dhc10 sequence in sea urchin (TgDhc5C), rat (Dhc2), mouse (Dhc5), and Paramecium (Dhc5). The Dhc10-related transcripts appear to be most abundant in ciliated cells and tissues (Tanaka et al., 1995) and/or up-regulated in response to deciliation (Gibbons et al., 1994). Thus, it is likely that the Dhc10-related genes encode polypeptide sequences required for axonemal motility in these organisms as well.

The Dhc10 Gene Encodes the 1β Dhc of the I1 Complex

Several lines of evidence indicate that Dhc10 encodes the 1β Dhc of the I1 complex. First, the disruption of the Dhc10 gene in ida2-6 results in the failure to assemble the I1 complex (Figures 2, 3, and 5), which is composed of two Dhcs (1α and 1β), and associated ICs and LCs (Piperno et al., 1990; Porter et al., 1992; Harrison et al., 1998). Although defects in either the 1α or 1β Dhc might be predicted to result in an I1 mutant phenotype, we have shown previously that the 1α Dhc is encoded by a different Dhc locus (Myster et al., 1997). Second, amino acid sequence comparisons between axonemal Dhcs indicate that the Dhc10 gene product is most similar to the 1α Dhc (Myster et al., 1999), as might be expected for two Dhcs that coassemble into a heteromeric dynein complex. Finally, the generation of Dhc10-specific antibody that exclusively recognizes the 1β Dhc subunit in wild-type strains and related N-terminal fragments in the Dhc10 transformants (Figure 8) clearly establishes that the 1β Dhc is the Dhc10 gene product.

Rescue of ida2 Mutants with Truncated Dhc10 Genes: Implications for Assembly Domains within Dhcs

We have been able to partially rescue the motility defects in ida2 mutants by transformation with truncated Dhc10 genes. Our results indicate that the rescue is due to the reassembly of a modified I1 complex containing a full-length 1α Dhc and a truncated 1β Dhc that lacks the motor domain (Table 1, Figures 2, 9, and 10). Although the assembly of dynein complexes with N-terminal fragments has been described (Sakakibara et al., 1993; Iyadurai et al., 1999; Myster et al., 1999), all of those fragments were considerably larger (140–160 kDa) than those reported here. We have found that a 92-kDa N-terminal fragment containing 811 residues of the 1β Dhc, or <20% of the full-length Dhc, can still support the formation of the I1 complex, its transport to the flagellar compartment, and its assembly onto the flagellar axoneme, but fragments containing only 508 residues cannot (Figures 5 and 10A). These results indicate that the region between residues 508 and 811 contains a critical domain required for complex assembly. Interestingly, Clustal W alignment of Dhc sequences has shown that this region is moderately well conserved between both axonemal and cytoplasmic Dhc sequences. In addition, the assembly domain of the 1β Dhc overlaps with the region identified in the Dictyostelium cytoplasmic Dhc as critical for subunit association in vitro (Habura et al., 1999). Together, these observations suggest that the limited sequence homologies observed between the N-terminal regions of the cytoplasmic and axonemal Dhcs are related to their similar roles in the subunit interactions that lead to dynein complex assembly.

Figure 10.

(A) Scheme of 1β Dhc and N-terminal fragments analyzed in this study. Also indicated are the locations of the P-loops (P1, P2, P3, and P4) and the regions predicted to form α-helical coiled coils (curlicues). (B) Scheme illustrating the location of I1 complex components within the structure of the 96-nm axoneme repeat.

Contribution of the 1β Dhc Motor Domain to Motility and Phototaxis

Partial rescue of the ida2 motility defects was observed with Dhc10 constructs that completely lack the region encoding the 1β Dhc motor domain, and the swimming velocities of the rescued strains were intermediate between that of wild type and that of the particular ida2 mutant used as host (Table 1). A similar reduction in swimming velocity was also seen in strains that lack the 1α motor domain (Myster et al., 1999). We interpret the reduced swimming velocities relative to wild type as a direct effect of the loss of one of the two I1 Dhc motor domains. However, it is also interesting to compare the maximal speed of the Dhc10 rescued strains, which swim forward at ∼108 μm/s, with that of the Dhc1 transformants, which swim forward at ∼137 μm/s (Table 1) (Myster et al., 1999). These differences imply that the contribution of the 1β Dhc motor domain to forward swimming velocity is greater than that of the 1α Dhc. In addition, these results suggest that the two dynein heads can function as independent motor units and that the loss of one head does not inhibit the activity of the second motor domain. This differs from the results described previously for cytoplasmic dynein (Iyadurai et al., 1999).

Transformation of ida2 strains with truncated Dhc10 constructs not only resulted in an increased swimming velocity but also rescued the ability of the mutants to phototax. Given that strains lacking only the 1α head domain can also phototax in the presence of outer arms (Myster et al., 1999), it appears that only one I1 Dhc motor domain is necessary for a normal phototactic response. However, it is also possible that the rescued phenotype is not due directly to the dynein motor domain but instead is due to the reassembly of the associated IC/LC complex into the axoneme. For example, IC138 has been identified previously as an important component in the phototaxis response, because strains with altered IC138 phosphorylation states are unable to phototax (King and Dutcher, 1997). Clearly, additional work is needed to understand how the phosphorylation state of IC138 alters the activity of the I1 dynein motor domains during the phototaxis response.

Implications for the Structure, Assembly, and Regulation of the I1 Complex

Electron microscopic analysis of axonemes from the rescued strains has allowed us to determine the structural defects associated with the assembly of an I1 complex lacking the 1β motor domain. As shown in Figure 2, E and F, lobes 2 and 3 of the I1 complex are largely restored in axonemes obtained from the rescued strain. The major defect associated with the missing 1β motor domain is the loss of lobe 1, although some loss of density can also be seen extending into lobe 3. Structural analysis of strains lacking the 1α motor domain has identified similar defects in lobe 2 (Myster et al., 1999). Combining these data, we can now infer that lobe 3 of the I1 structure contains the N-terminal regions of the two Dhcs as well as the associated IC/LC complex (Figure 10B).

These observations have important implications for both the assembly and the regulation of the I1 complex. Previous work has indicated that both IC140 and IC110 play structural roles in the assembly of the I1 complex and its binding to the A-tubule of the flagellar axoneme (Perrone et al., 1998; Yang and Sale, 1998). The localization of IC140 and IC110 to lobe 3 confirms that lobe 3 is the site for the attachment of the I1 complex to the outer doublet microtubule. These findings are also consistent with views of this region in axoneme cross-sections (Mastronarde et al., 1992). In vitro reconstitution studies have further shown that both purified I1 complexes and expressed fragments of IC140 can rebind specifically to vacant sites in I1 mutant axonemes, but they will not bind to wild-type axonemes or purified tubulin (Smith and Sale, 1992; Yang and Sale, 1998). These results suggest the presence of accessory proteins that function as docking structures for the attachment of the I1 complex, similar to that observed in the outer dynein arm (Takada and Kamiya, 1994). Such docking proteins must be located in close proximity to lobe 3.

The association of IC138 with the base of the I1 complex is also significant in light of its proposed role in regulating dynein arm activity. As described above, mutations that affect the phosphorylation state of IC138 result in the loss of the ability to phototax (King and Dutcher, 1997). Furthermore, increases in microtubule sliding velocities in response to signals from the radial spokes are associated with changes in the phosphorylation state of IC138 (Habermacher and Sale, 1997), and both an axonemal phosphatase and an axonemal kinase appear to be involved (Yang and Sale, 1999; Yang et al., 2000). The localization of IC138 to lobe 3 places IC138 in relatively close proximity to the first radial spoke, which would be an ideal location for a component that mediates signals between the radial spokes, axonemal kinases or phosphatases, and the dynein motor domains. The axonemal kinases and phosphatases that act on IC138 also must be located near lobe 3.

The I1 complex LCs are also located in lobe 3, at the base of the I1 structure. Western blots of I1 complexes lacking one or the other Dhc motor domain reveal wild-type levels of the 14-kDa Tctex1 LC (Figure 9). Similar results have been seen with antibodies against the 8-kDa LC (our unpublished results). Together with other studies, these findings suggest that the 8-kDa LC plays multiple roles in the assembly and transport of the I1 complex. It may function to stabilize the I1 dynein and other dyneins containing multiple Dhcs during complex formation (Benashski et al., 1997; Harrison et al., 1998), and it may interact with the transport machinery involved in flagellar assembly (Cole et al., 1998). It is clearly required for the retrograde transport of components during flagellar assembly and maintenance, presumably as a subunit of a cytoplasmic dynein motor (Pazour et al., 1998).

The localization of Tctex1 to the base of the I1 complex is significant in light of its proposed role in the targeting of dynein complexes to specific cellular locations or cargoes (Harrison et al., 1998; King et al. 1998; Nagano et al., 1998; Tai et al., 1998, 1999). For instance, Tctex1 may serve to direct the I1 complex to its specific location in the 96-nm axoneme repeat (Figure 10B). Tctex1 is also known to bind directly to the fyn protein kinase in vitro (Campbell et al., 1998). These results and its position at the base of the I1 complex suggest that it may also facilitate interactions with the axonemal kinases that modify the activity of the I1 complex in situ (Habermacher and Sale, 1997; King and Dutcher, 1997).

The position of the Tctex1 LC may also be important with respect to its proposed function as one of the distorter gene products of the mouse t complex (Lader et al., 1989; Harrison et al., 1998). The t complex is a region of mouse chromosome 17 containing several genes that affect sperm motility and male fertility. Mutations in the t haplotype are usually inherited as a single group in a nonmendelian manner (reviewed by Silver, 1993; Olds-Clarke, 1997), and genetic analyses have indicated that this nonmendelian inheritance is the result of the interaction of several distorter gene products with a single responder gene product (Lyon, 1984, 1986). After the identification of Tctex1 and the outer arm LC, Tctex2, as two candidates for the distorter gene products, Patel-King et al. (1997) proposed a model in which the responder might function as a gatekeeper during flagellar assembly that interacts directly with the LCs and affects sperm motility. Mutations in the Tctex LCs would poison the wt-bearing sperm associated with a wild-type responder by the incorporation of defective dynein complexes that reduce sperm motility (Patel-King et al., 1997; Harrison et al., 1998). However, t-bearing sperm with a mutant responder would interact preferentially with wild-type dynein complexes, with the result that t-bearing sperm would be more motile than wt-bearing sperm (Patel-King et al., 1997; Harrison et al., 1998). The recent discovery that the t complex responder is a mutant form of a testis-specific, sperm motility kinase (smok) (Herrmann et al., 1999) suggests that defects in the phosphorylation of dynein subunits could be the basis of the t complex phenotype. Indeed, changes in the phosphorylation state of the Tctex LCs have been correlated with changes in sperm motility in other species (Inaba et al., 1999). Determining the subcellular distribution of smok kinase and its position relative to the Tctex1 and Tctex2 dynein LCs is clearly the next step needed to understand how these components interact to regulate flagellar motility.

Abbreviations used:

BAC

bacterial artificial chromosome

Dhc

dynein heavy chain

IC

intermediate chain

LC

light chain

P-loop

phosphate-binding motif

RFLP

restriction fragment length polymorphism

RT

reverse transcriptase