Sequence and Characterization of cDNA and Genomic DNA Encoding DC3
As the first step toward isolation of a cDNA encoding DC3, we isolated axonemes from the mutant oda
6, which lacks the outer dynein arms but retains the ODA-DC (Takada et al., 1994
), so that the latter can be obtained free of the former. The ODA-DC was then extracted from the axonemes and partially purified by sucrose density gradient centrifugation (Takada et al., 2002
). Proteins in the 7S fraction (containing the ODA-DC) were separated by SDS-PAGE and transferred to PVDF membrane, and the band corresponding to DC3 was excised and digested with trypsin. Peptides were separated by HPLC and selected peptides were sequenced. The following sequences were obtained: KIDAEELTELFLR (peptide 1), TYKAPPPPQKR (peptide 2), and AELDMQLEQVFGTADLNSGK (peptide 3). The sequences of peptide 1 and peptide 3 were then used to design the degenerate oligonucleotide primers 25E, 25J, and 25K (see MATERIALS AND METHODS).
The 25E/25J primer pair and PCR were used to amplify reverse-transcribed cDNA from cells regenerating their flagella (Wilkerson et al., 1994
). The reaction produced two major and several minor products. To amplify the product specific for DC3, the 25E/25J PCR products were reamplified with primer 25E and the nested primer 25K. This reaction produced just one prominent species of ~300-base–pairs, which was subcloned. Sequencing of one of the subclones revealed that it encoded amino-acid sequence present in peptide 3 but not used to design the primers, indicating that the correct cDNA had been amplified. This subclone was then used to isolate 6 cDNA clones from a wild-type cDNA library (Wilkerson et al., 1995
). Sequencing of the ends of these clones indicated that they all were derived from the same gene. To verify that the cDNAs encoded a flagellar protein, one of the clones was used to probe mRNA isolated from control and deflagellated wild-type cells. The cDNA hybridized to one prominent band of ~1.4 kb. As is characteristic of genes encoding flagellar proteins, DC3 gene transcription was upregulated in wild-type cells that had been deflagellated and were actively regrowing their flagella (, upper panel). On longer exposure, a very minor amount of a second message of ~1.8 kb was also detected in the deflagellated samples.
Figure 1. DC3 gene expression is induced by deflagellation. mRNA was isolated from wild-type cells that were either fully flagellated (non–deflagellated) or actively regrowing their flagella (30′ post deflagellation). mRNA was analyzed by northern (more ...)
The longest clone (p25K2–1) contained a single open reading frame with three in-frame stop codons located upstream of the first ATG (). With the putative translation initiation site at nucleotide position 148, the clone exactly predicts the sequence of the three peptides that were obtained by direct sequencing of tryptic fragments of DC3. Because peptide 2 was not used to obtain the cDNA clone, its presence in the predicted amino acid sequence unequivocally confirms that p25K2–1 encodes DC3. An in-frame stop codon (TGA) is located at nucleotide position 700. At nucleotide 1234 is a consensus polyadenylation signal and the poly A tail begins at nucleotide 1413. The clone predicts a protein product of 184 amino acids with a mass of 21,341 Da, in good agreement with the relative mobility of DC3 in SDS-polyacrylamide gels. We therefore conclude that the p25K2–1 cDNA clone contains the entire DC3 coding region.
Figure 2. DC3 sequence analysis. (A) Nucleic acid and deduced amino-acid sequence of DC3. The three in-frame STOP codons upstream of the start ATG are in bold. The peptides obtained by direct sequencing of DC3 tryptic fragments are underlined. An asterisk indicates (more ...)
In contrast to the other two components of the ODA-DC (Koutoulis et al., 1997
; Takada et al., 2002
), DC3 is not predicted by the COILS program to form coiled coils. We used BLASTP 2.2.4 (Altschul et al., 1997
) to search the database for related proteins. The best match (37% identity, 61% similarity) was to a protein predicted from the genomic sequence of both Plasmodium yoelii yoelii
and Plasmodium falciparum
. An alignment of DC3 and the protein from P. falciparum
, the human malaria parasite, is shown in . Although the parasite proteins have an N-terminal extension making them longer than the Chlamydomonas
protein, the identities extend throughout the sequences, suggesting they are true homologues. Secondary structure analysis using the PHDsec program (Rost and Sander, 1993
; Rost and Sander, 1994
; Rost, 1996
) predicted four helix-loop-helix motifs (double underlined in ). Analysis of the sequence at the Pfam protein families (Bateman et al., 2002
) site (http://pfam.wustl.edu
) confirmed that these are EF-hands and that DC3 is a novel member of the EF-hand superfamily of calcium-binding proteins (see DISCUSSION).
The DC3 cDNA was used to select a BAC clone containing the DC3 gene. A 5.6-kb subfragment of this BAC was sequenced and found to contain the entire DC3 gene. Unusual for a Chlamydomonas gene, the DC3 gene contains no introns.
Identification of Mutants with Defects in the DC3 Gene
, reverse genetics has been very useful for deducing the function of cloned genes encoding flagellar proteins (Pazour and Witman, 2000
). To find a mutant defective in DC3, a collection of insertional mutants having flagellar/cytoskeleton-related phenotypes (Pazour et al., 1995
; Koutoulis et al., 1997
; Pazour and Witman, 2000
) was screened by Southern blot using the DC3 cDNA as probe. The cDNA hybridized to a single band in wild-type genomic DNA digested with Pst
I (, arrow), Bam
dIII, and Xho
I. These data indicate that only one DC3 gene is present in the Chlamydomonas
genome. Two insertional mutants (F28 and V06) lacked hybridizing bands, indicating that the gene was completely deleted (). A third strain (V16) gave two hybridizing bands, suggesting that the gene had been disrupted by an insertion (). When the DC3 mutants were analyzed by light microscopy, they all had slow, jerky swimming, a hallmark of strains lacking the outer dynein arm (Kamiya, 1988
). One of the deletion mutants, V06 (hereafter referred to as the DC3-deletion strain), was selected for further studies. The DC3-deletion strain grew and divided normally, indicating that DC3 is not vital to cell survival. Conclusive evidence that deletion of the DC3 gene specifically is responsible for the mutant phenotype is presented later in this article.
Figure 3. Identification of insertional mutants with defects in the DC3 gene. Genomic DNA was isolated from insertional mutants having flagellar/cytoskeletal defects, cut with PstI, and analyzed by Southern blot using the DC3 cDNA as a probe. The probe hybridized (more ...)
DC3-null Axonemes Contain a Partial Complement of Outer Arms and ODA-DCs
To determine if axonemes of the DC3-deletion strain had an abnormal ultrastructure, we examined them by EM. In contrast to null mutants for DC1 (Koutoulis et al., 1997
) and DC2 (Takada et al., 2002
), which lack outer arms, some DC3-null outer doublets had outer dynein arms (, upper panel, center). Quantitative analysis of 59 DC3-null axonemal cross sections revealed an average of 3.1 outer dynein arms per axonemal cross section. Arms were more abundant in proximal than distal regions of the axoneme: an average of 4.73 outer arms was observed in 22 proximal axonemes (i.e., axonemes with beak structures; Hoops and Witman, 1983
), whereas an average of 2.14 outer arms was observed in 37 distal axonemes (i.e., axonemes without beak structures). The arms appeared to be attached to the correct location on the A-tubules; no ectopic location (e.g., on the B-tubules) of outer arms was observed. Although the outer arms were attached at the correct location, a gap was often seen between the outer arm and the A-tubule to which it was attached (, upper panel, center, arrowhead), suggesting the arms may not be as tightly associated with the outer doublet microtubule in the absence of DC3. This gap was not observed in wild-type axonemes prepared under identical conditions (, upper panel, left). In longitudinal sections, the outer arms appeared to be evenly spaced at 24-nm intervals along the DC3-null axoneme (, lower panel, arrowheads), indicating that, when present, they assemble in a cooperative manner.
Figure 4. Loss of DC3 results in partial loss of outer dynein arms. Top panel: Axonemal cross sections from a wild-type strain (left), the DC3-deletion strain (center), and the DC3-deletion strain rescued by transformation with the DC3 gene (right). DC3-null axonemes (more ...)
Because the ODA-DC is essential for the correct positioning of outer arms on the axoneme (Koutoulis et al., 1997
; Takada et al., 2002
), our observation that DC3-null axonemes contain some correctly positioned arms suggested a docking complex was present in these axonemes. If so, it would indicate that a docking complex lacking DC3 is competent to assemble on the axoneme and correctly position some outer dynein arms. To investigate this possibility, we compared axonemal cross sections from a strain lacking outer dynein arms but retaining the ODA-DC (oda
9), a strain lacking outer dynein arms and the ODA-DC (oda
3), and the DC3-deletion strain. The ODA-DC is visible on oda
9 axonemes as a small projection on the A-tubules at the sites where the outer dynein arms would normally attach (, left, arrowheads). These projections are missing from the A-tubules of oda
3 (, center, arrowheads). Because outer arms normally obstruct the docking complex from view, we searched for docking complexes on those A-tubules lacking arms in the DC3-deletion strain. A small projection (, right, arrowheads), positioned at sites where the outer arms would normally attach, was indeed present on some but not all doublets of the DC3-deletion strain. We refer to this structure as a “partial” ODA-DC, i.e., one missing DC3.
Figure 5. The DC3-deletion strain assembles a “partial” docking complex. Axonemal cross sections from a strain lacking outer arms but retaining the ODA-DC (oda9), a strain lacking outer arms and the ODA-DC (oda3), and the DC3-deletion strain (DC3Δ). (more ...)
A Complex of DC1 and DC2 Assembles onto DC3-null Axonemes
To confirm that DC1 and DC2 assemble onto axonemes of the DC3-deletion strain, we isolated and subfractionated DC3-null flagella to obtain membrane-free axonemes and the detergent-soluble membrane + matrix fraction. We then analyzed the fractions by western blot using antibodies specific for DC1 and DC2. Although there was no signal in the membrane + matrix fraction, DC3-null axonemes contained both DC1 (, upper panel) and DC2 (our unpublished results).
Figure 6. (A) DC1 and DC2 assemble on the axoneme in the absence of DC3, but not vice versa. Top panel: Axonemes from wild type, the DC3-deletion strain (DC3Δ), oda1, and oda3 were isolated and analyzed by western blotting. oda1 is null for DC2; oda3 is (more ...)
To determine whether DC1 and DC2 occur together as a complex and to investigate whether any additional polypeptides are contained in the ODA-DC when DC3 is missing, we immunoprecipitated the ODA-DC from biotinylated 0.6 M KCl extracts of DC3-null axonemes using an antibody specific for DC1. Immunoprecipitated proteins were resolved by SDS-PAGE, transferred to nitrocellulose, and detected using streptavidin-HRP. Two major bands corresponding to DC1 and DC2 were detected in the DC3-null immunoprecipitate (, lane 1). Except for bands present in the normal IgG control (, lane 2), no other bands were observed, indicating that no other protein substitutes for DC3 when DC3 is absent. These data confirm that a docking complex containing only DC1 and DC2 is competent to assemble on the axoneme in the complete absence of DC3. However, “partial” ODA-DCs apparently do not bind outer arms as well as wild-type ODA-DCs, because not all “partial” ODA-DCs are occupied by arms (, right panel).
As an independent test of our ultrastructural observation that the number of ODA-DCs is reduced in the DC3-deletion strain, we quantitatively analyzed gels such as that shown in (upper panel) to determine the relative amounts of DC1 and DC2 in wild-type vs. DC3-null axonemes. The data were normalized to those for the inner arm intermediate chain IC140 and the radial spoke protein RSP-1 (unpublished data), both of which are expected to be present in normal amounts in the mutant strain. The results indicated that the amounts of DC1 and DC2 in the mutant axonemes were ~83% and ~81% (average of two experiments) of their levels in the wild-type axonemes, confirming that the mutant does not assemble a full complement of ODA-DCs. Similar analyses revealed that DC3-null axonemes contain less than one-half the normal levels of the outer dynein arm subunits IC1 and IC2, confirming that the loss of outer arms is greater than the loss of ODA-DCs in the mutant strain.
DC3 Cannot Assemble in the Absence of DC1 and DC2
Our finding that a “partial” docking complex composed of only DC1 and DC2 could assemble in the DC3-deletion strain raised the question of whether DC3 could assemble onto the axoneme independently of DC1 and DC2. To investigate this, we probed western blots of DC1-null (oda3) and DC2-null (oda1) axonemes with our antibody to DC3 (, lower panel). The antibody detected one prominent axonemal protein of Mr ~25,000 in both wild type and oda9 (which lacks outer arms but has the ODA-DC). No protein was detected in axonemes from the DC3-deletion strain, indicating that the antibody is specific for DC3. The DC3 antibody did not react with any proteins in axonemes isolated from the DC1- or DC2-null strains. Therefore, although DC1 and DC2 can assemble in the absence of DC3, DC3 cannot assemble in the absence of the other two docking complex components. Moreover, because the only known structural difference between oda1, oda3, and oda9 is that oda9 has the ODA-DC, whereas oda1 and oda3 do not, these results provide additional evidence that DC3 is part of the ODA-DC.
DC3-null Outer Arms Can Generate Force
mutants completely lacking outer dynein arms typically swim in a jerky manner at about one-third the speed of wild type (Kamiya, 1988
). Although the DC3 mutants are slow, jerky swimmers, it was important to measure their swimming speed to see if the arms that were present were functional. If they were not functional, the swimming speed should be comparable to that of an oda
mutant completely lacking the outer arms; if they were functional, the swimming speed should be intermediate between that of such a mutant and wild type. Therefore, we compared the swimming speed of the DC3-deletion strain to that of oda
9 (completely lacking outer dynein arms) and g1 (wild type). As expected, the mean swimming speed of oda
9 (39 μm/s) was about one-third the mean swimming speed of wild type (116 μm/s). In contrast, the mean swimming speed of the DC3-deletion strain (61 μm/s) was about one-half that of wild type, indicating that the outer arms that assemble in the absence of DC3 are force-generating and functional.
DC3-null Mutants Have an Altered Photoshock Response
displays two distinct behaviors in response to light: phototaxis and photoshock (Witman, 1993
). Phototaxis is directed forward swimming toward or away from a light source. Photoshock, which occurs in response to an intense flash of light, is a sudden stop in forward swimming, brief backward swimming, and then resumed forward swimming. An essential feature of photoshock is a change in outer arm activity: although still able to phototax, outer armless strains have an altered photoshock response. They stop in response to a flash of bright light but either cannot generate the symmetrical waveform needed for backward swimming (Kamiya and Okamoto, 1985
) or produce small bend amplitudes with their flagella improperly aligned (Mitchell and Rosenbaum, 1985
). These data indicate that the outer arm is important for normal backward swimming during photoshock (Kamiya and Okamoto, 1985
We tested the DC3-deletion strain's ability to phototax and photoshock (see MATERIALS AND METHODS). Although phototaxis was normal, the photoshock response was altered; the cells stopped in response to an intense flash of light, but little or no backward swimming was evident. Thus, although the outer dynein arms in the DC3-deletion strain can generate force for forward swimming (see above), either the absence of DC3 or the incomplete number of arms precludes normal backward swimming during the photoshock response.
The DC3-null Phenotype Can Be Rescued by Transformation with DNA Containing DC3
To determine if the slow, jerky swimming was directly linked to the DC3 gene deletion, we crossed the DC3-deletion strain to a wild-type strain of opposite mating type. Offspring from the cross were scored for motility by light microscopy. DNA was isolated from one product each of 16 different tetrads, cut with PstI, blotted, and probed with the DC3 cDNA. As shown in (upper panel), all progeny lacking the DC3 gene had mutant motility, whereas those inheriting a DC3 gene had normal motility. These results strongly suggest that the DC3 gene deletion is responsible for the mutant phenotype.
Figure 7. Top panel: The slow-swimming motility phenotype and the DC3 gene deletion cosegregate. The DC3-deletion strain was crossed to a wild-type strain, tetrads were dissected, and progeny from 16 different tetrads (lanes 1–16) were scored for motility (more ...)
A frequently used procedure for determining whether a disruption or deletion of a particular gene is responsible for a mutant phenotype is to transform the mutant with a wild-type copy of the gene and see if it rescues the phenotype. Therefore, we cotransformed the DC3-deletion strain with a 5.6-kb fragment of genomic DNA (pDC3S-2) containing the DC3 gene and a separate DNA containing a selectable marker (see MATERIALS AND METHODS). Wild-type motility was restored in ~21% (53/253) of the cell lines that also took up the selectable marker. Genomic DNA was isolated from randomly picked transformants and probed for integration of the DC3 DNA by Southern blotting. Whenever a transformant had wild-type motility, it also had a DC3-positive genomic insertion (our unpublished results). Moreover, when one of the rescued strains was crossed to one of the DC3-deletion strains and the progeny from 15 different tetrads analyzed, wild-type motility always correlated with the presence of a DC3-positive insertion (, lower panel). These results conclusively demonstrate that the restored motility was due to the presence of the integrated DC3 DNA. The results also indicate that the DC3 DNA used for transformation contains all the 5′ and 3′ sequences necessary for DC3 expression.
We measured the swimming velocity of a DC3-transformant to determine if its speed had been restored to wild-type levels. The DC3-transformant swam at a speed of 92 μm/s, a number comparable to that of wild type (see above). We also examined the rescued strain by EM to determine whether all the outer arms had been restored. Axonemal cross sections from the rescued strain showed the complete restoration of outer dynein arms (, upper panel, right; one of the nine doublets normally does not have outer arms [Hoops and Witman, 1983
]). In addition, the rescued strain displayed normal backward swimming during the photoshock response. To see if DC3 had been restored to the ODA-DC, we immunoprecipitated the ODA-DC from biotinylated 0.6 M KCl extracts of DC3-transformant axonemes using an antibody specific for DC1. The DC3-transformant extract was treated exactly as the DC3-null extract (see above). Three major bands corresponding to DC1, DC2, and DC3 were detected in the DC3-transformant immunoprecipitate (, lane 3). These data indicate that transformation of the DC3-deletion strain with DNA containing DC3 completely rescues the mutant phenotype.
The DC3 Gene Is Responsible for the Rescued Phenotype
The above results demonstrated that DNA containing the DC3 gene could rescue the DC3-null phenotype. However, because the intronless DC3 gene is only ~1.4 kb, there was a possibility that the 5.6-kb rescuing clone contained one or more additional genes. If the insertional event that deleted DC3 also deleted a gene tightly linked to DC3 and if this gene were included in the 5.6-kb clone, it, rather than the DC3 gene, may have been responsible for rescuing the mutant phenotype. Therefore, to ascertain if the rescue was specifically due to DC3, we used site-directed mutagenesis to create a premature STOP codon in the DC3 gene and then transformed the mutagenized DNA into the DC3-deletion strain. Out of 495 transformants screened, none were rescued. To confirm that the modified gene's failure to rescue was due to the introduction of the STOP codon and not to some other mutation introduced during the mutagenesis procedure, we corrected the premature STOP by recreating the original sequence in a second round of mutagenesis. Transformation with the repaired DC3 gene restored wild-type swimming in 32 of 157 transformants screened, an efficiency of rescue comparable to that obtained with the wild-type gene (see above). These data demonstrate that DC3 is the only gene within the genomic clone that can rescue the DC3-deletion strain, thereby confirming that loss of DC3 is responsible for the DC3-null phenotype.
The DC3 Gene Is a Novel ODA Gene
To determine whether the DC3 gene was a novel gene or one of the known but still uncharacterized ODA
genes, we crossed 137c-derived lab strains carrying mutations in the latter genes to the wild-type Chlamydomonas
field isolate CC2290 (Gross et al., 1988
). The CC2290 and 137c strains are interfertile yet polymorphic such that when genomic DNA from the two strains is probed with the DC3 cDNA, an RFLP is detected (our unpublished results). The presence of a DC3 RFLP and the ability to score motility by light microscopy makes the identification of parental and nonparental offspring from these crosses straightforward. For instance, if the DC3 gene is either identical to or tightly linked to one of the mutant genes, all progeny with the mutant phenotype will have the 137c version of the DC3 RFLP (because the mutants are in the 137c background). Alternatively, if the DC3 gene is not linked to the previously identified genes, then the mutant phenotype and the 137c version of the DC3 RFLP will segregate independently. Using the cDNA insert of p25K2–1 as a probe, we screened the progeny of crosses between CC2290 and oda
13a, and pf
22 by Southern blot and scored motility by light microscopy. In all cases, the mutant phenotype and the 137c version of the DC3 RFLP segregated independently ().
Lastly, to determine whether the DC3 gene was identical to the uncloned ODA
10 gene, we screened an oda
10 insertional mutant (Koutoulis et al., 1997
) by Southern blot using the DC3 cDNA as a probe. No RFLPs were detected. Taken together, these data suggest that the DC3 gene is a previously unidentified gene that affects outer dynein arm assembly. We have named this new gene ODA
14; the alleles of the two deletion strains V06 and F28 are termed oda
14-1 and oda
14-2, respectively, and that of the V16 strain is termed oda
14-3. RFLP analysis (courtesy of Dr. C. Silflow, University of Minnesota) was used to map the ODA
14 locus to Chlamydomonas
Linkage group XV (http://www.biology.duke-.edu/chlamy_genome/nuclear_maps.html