Previous physiological and pharmacological experiments have demonstrated that the Chlamydomonas flagellar axoneme contains a cAMP-dependent protein kinase (PKA) that regulates axonemal motility and dynein activity. However, the mechanism for anchoring PKA in the axoneme is unknown. Here we test the hypothesis that the axoneme contains an A-kinase anchoring protein (AKAP). By performing RII blot overlays on motility mutants defective for specific axonemal structures, two axonemal AKAPs have been identified: a 240-kD AKAP associated with the central pair apparatus, and a 97-kD AKAP located in the radial spoke stalk. Based on a detailed analysis, we have shown that AKAP97 is radial spoke protein 3 (RSP3). By expressing truncated forms of RSP3, we have localized the RII-binding domain to a region between amino acids 144–180. Amino acids 161–180 are homologous with the RII-binding domains of other AKAPs and are predicted to form an amphipathic helix. Amino acid substitution of the central residues of this region (L to P or VL to AA) results in the complete loss of RII binding. RSP3 is located near the inner arm dyneins, where an anchored PKA would be in direct position to modify dynein activity and regulate flagellar motility.
The goal of this study is to determine the mechanism for localization of the cAMP-dependent protein kinase (PKA) in ciliary and flagellar axonemes. In several experimental systems, cAMP or specific inhibitors of PKA alter motility of ATP-induced reactivated movement of ciliary and flagellar axonemes. For example, cAMP is required for activation of sperm tail axonemes from many species (for reviews see Brokaw 1987; Tash 1989; San Agustin and Witman 1994). Addition of cAMP increases beat frequency of reactivated ciliary axonemes from Paramecium (for example, Hamasaki et al. 1989), and inhibits reactivated motility of flagellar axonemes from Chlamydomonas (Hasagawa et al. 1987). Biochemical analysis in diverse cellular systems has demonstrated that several axonemal proteins, including dynein subunits, are phosphorylated in vitro in a cAMP-dependent manner (for example, Hamasaki et al. 1991; Inaba et al. 1999; Nomura et al. 2000). Moreover, selective inhibitors of PKA regulate dynein-driven microtubule sliding in axonemes isolated from Chlamydomonas flagella (Howard et al. 1994). Together the data demonstrate that PKA is a structural component of the 9 + 2 axoneme. However, the mechanism for localization and anchoring of PKA is not known.
We proposed that PKA is localized to the axoneme through association with A-kinase anchoring proteins (AKAPs). AKAPs are a family of proteins that target PKA to specific intracellular sites through interaction with type I (RI) or type II (RII) PKA regulatory subunits (for a review see Edwards and Scott 2000; and others). In most cases, the regulatory subunits bind to AKAPs through interaction with an amphipathic helix contained within the AKAP. AKAPs can often be identified by RII blot overlays in which radiolabeled RII is incubated with blots of protein (Westphal et al. 2000). By this approach, several AKAPs have been identified in flagella of sperm (Johnson et al. 1997; Moss et al. 1999; Vijayaraghavan et al. 1999; Reinton et al. 2000). However, to date none of the AKAPs have been localized to the axoneme.
To test the hypothesis that axonemal PKA is anchored by AKAPs, we used isolated axonemes from Chlamydomonas. Chlamydomonas offers several experimental advantages including the ease of axoneme isolation and the availability of mutants that are immotile and defective in specific axonemal structures. Based on pharmacological analysis, Chlamydomonas axonemal PKA regulates reactivated motility as well as dynein-driven microtubule sliding activity (Hasagawa et al. 1987; Howard et al. 1994). We performed RII overlays in the presence or absence of specific AKAP–RII binding inhibitors to identify two AKAPs in the Chlamydomonas axoneme. Through detailed analysis we have determined that one of the AKAPs is radial spoke protein 3 (RSP3). By testing truncated and point mutant forms of RSP3, we have mapped the RII-binding domain of RSP3. RSP3 is the first AKAP to be identified in axonemes and is the first AKAP to be identified from a unicellular organism. This finding provides a foundation for understanding the mechanism of kinase anchoring in the axoneme, and demonstrates that AKAPs may be evolutionarily conserved components of signal transduction in the cell.
Materials and Methods
Chlamydomonas Strains and Growth Conditions
Chlamydomonas reinhardtii strains used include 137c and cc124 (wild-type), pf14 (lacks radial spoke, paralyzed flagella), pf17 (lacks radial spoke head, paralyzed flagella), pf27 (radial spoke phosphorylation defect, paralyzed flagella), pf15 (lacks central pair apparatus, paralyzed flagella), pf16 (lacks C1 microtubule of central pair apparatus, paralyzed flagella), pf18 (lacks central pair apparatus, paralyzed flagella), pf19 (lacks central pair apparatus, paralyzed flagella), pf20 (unstable central pair apparatus, paralyzed flagella), pf6 (lacks projection on C1 microtubule of central pair apparatus, paralyzed flagella), pf28pf30 (lacks outer dynein arms and I1 inner dynein arms, paralyzed flagella), and pf2 (defective in the dynein regulatory complex, paralyzed flagella). All strains were obtained from the Chlamydomonas Genetics Center (Duke University, Durham, NC) with the exception of pf28pf30 which was generated by crossing pf28 with pf30 (strains obtained from the Chlamydomonas Genetics Center). The double mutant was then isolated from a nonparental ditype tetrad. Cells were grown in liquid modified medium I, with aeration and a 14-h/10-h light/dark cycle (Witman 1986).
Isolation of Axonemes
Axonemes were isolated as described previously (Howard et al. 1994). Where indicated, axonemes (at 5 μg/μl) were treated with 0.6 M NaCl (20 min on ice) in a buffer containing 10 mM Hepes, 5 mM MgSO4, 1 mM DTT, 0.5 mM EDTA, 0.1 M PMSF, and 0.6 TIU aprotinin, pH 7.4. Before fixation for SDS-PAGE, axonemes treated with 0.6 M NaCl were dialyzed to 30 mM NaCl in the same buffer. Axonemal protein samples were fixed for SDS-PAGE at a concentration of 5 μg/μl (50 μg total for each lane). Protein concentration was determined using the Bradford assay. Unless otherwise stated, all reagents were obtained from Sigma-Aldrich, and deionized H2O was used throughout.
RII overlays were performed according to the method of Hausken et al. 1998 with some modifications as described below. In brief, proteins were separated by SDS-PAGE and transferred to nitrocellulose. The nitrocellulose membrane was then incubated in TBS containing 5% nonfat milk (Carnation) and 0.1% BSA. Casein kinase II (New England Biolabs, Inc.) and [γ-32P]ATP (BLU502A; NEN Life Science Products) were used to radiolabel either biochemically purified RII (Promega) or recombinant RIIα (plasmid provided by Dr. J. Scott, Vollum Institute, Oregon Health Sciences University, Portland, OR). Radiolabeled RII was overlayed on the nitrocellulose blots. Blots were exposed on a phosphorimager screen, and data was recorded and analyzed using the Phosphorimager SI (Molecular Dynamics) and ImageQuant software (Molecular Dynamics). Ht31 (DLIEEAASRIVDAVIEQVKAAGAY) and Ht31-P (DLIEEAASRPVDAVIEQVKAAGAY) peptides were synthesized by the Emory University Microchemical Facility (peptide sequences from Vijayaraghavan et al. 1997).
Western Blot Analysis
Proteins were separated by SDS-PAGE and transferred onto nitrocellulose membranes. Membranes were treated with 3% nonfat milk in TBS to prevent nonspecific protein binding. Blots were incubated with primary antibodies: anti-RSP3 at 1:200 (Williams et al. 1989), and anti–glutathione-S-transferase (GST) at 1:500 (Amersham Pharmacia Biotech), for 1 h at 25°C. After washing in TBS, blots were incubated with HRP-conjugated secondary antibodies at 1:20,000 (goat anti–rabbit; Bio-Rad Laboratories; and rabbit anti–goat; ICN Biomedicals, respectively) for 1 h at 25°C. After a final wash, blots were developed using enhanced chemiluminescence (Amersham Pharmacia Biotech).
Identification of the RII-binding Site
RSP3 fragments were generated by PCR using the RSP3 cDNA (Diener et al. 1993) and primers that each contained one of three 5′ restriction sites: BamHI, EcoRI, or SphI. The PCR products were band purified, digested with the corresponding restriction endonucleases, and ligated into the pGEX-2T expression vector (Amersham Pharmacia Biotech) that had been digested with the same restriction endonucleases. In one instance (1–220), the RSP3 fragment was generated according to the method of Diener et al. 1993. After confirming the fidelity of the constructs by DNA sequence analysis, the constructs were transformed into either Escherichia coli DH5α (GIBCO BRL) or E. coli BL21(DE3) Gold cells (Stratagene) and expressed as GST fusion proteins. Expression was induced at mid-log phase by the addition of 0.1 mM IPTG for 1–2 h. After expression, the bacterial cells were pelleted, resuspended in SDS-PAGE sample buffer, and boiled in preparation for SDS-PAGE. Expression was monitored by Western blot analysis with an antibody to GST, and RII binding was tested by RII blot overlays.
Mutagenesis was performed using the RSP3 construct containing amino acids 104–180, generated as described above. Amino acid substitutions were made according to the protocol of the GeneEditor™ in vitro Site-Directed Mutagenesis System (Promega) with the modification of using 50 μl of GeneEditor™ antibiotic selection reagent. After mutagenesis, constructs were expressed and tested for RII binding as described above.
Results and Discussion
The Axoneme Contains Two AKAPs
To test the hypothesis that the Chlamydomonas axoneme contains an AKAP, we performed RII blot overlays. As illustrated in Fig. 1, the overlay resulted in the identification of several axonemal proteins that bind to RII (purified RII and recombinant RIIα yielded identical results). To determine which of these proteins bind to RII in a manner specific to AKAPs, the RII overlay was performed in the presence of the inhibitor peptide Ht31 or the control peptide Ht31-P. Ht31 is a synthetic peptide derived from the RII-binding domain of a known AKAP and contains an amphipathic helix (Carr et al. 1992). The peptide acts as a competitive inhibitor of AKAP-RII binding; thus, proteins that bind to RII in an AKAP-specific manner do not bind to RII in the presence of Ht31. Ht31-P, a control peptide, contains an isoleucine to proline amino acid substitution that disrupts the amphipathic helix and renders the peptide unable to bind to RII in an AKAP-specific manner (Scott and Faux 1998). Here we define AKAPs as proteins that bind to RII in the presence or absence of Ht31-P, but do not bind to RII in the presence of Ht31.
Ht31 inhibited RII binding to two particularly prominent axonemal proteins with masses of 240 kD (AKAP240) and 97 kD (AKAP97) (Fig. 1, compare panels). AKAP240 is partially extractable with 0.6 M NaCl, whereas AKAP97 is not extractable with 0.6 M NaCl, indicating that AKAP97 is very tightly associated with the axoneme (Fig. 1). In addition to AKAP240 and AKAP97, other proteins were identified in the RII overlay (Fig. 1). However, these proteins were either not consistently detected, or RII binding to the proteins was not inhibited by the Ht31 peptide. Although it is possible that these other RII-binding proteins may include additional AKAPs, they were not considered further in this study.
To localize AKAP240 and AKAP97 within the axoneme, we performed RII overlays on axonemal protein from four different classes of Chlamydomonas motility mutants that are defective for specific axonemal structures (Fig. 2 A). By doing so, we localized the AKAPs within the axoneme based on whether or not the AKAPs are present in the structural mutants. The analysis revealed that AKAP240 is absent in axonemes from a mutant defective for the central pair apparatus (pf18), and that AKAP97 is absent in axonemes from a mutant lacking the radial spokes (pf14) (Fig. 2 B). Both AKAPs are present in axonemes from a mutant lacking the outer dynein arms and the I1 inner dynein arm (pf28pf30) as well as in axonemes from a mutant defective for the dynein regulatory complex (pf2) (Fig. 2 B). Coomassie blue staining of a corresponding SDS-PAGE gel was performed to verify equivalent protein load. RII binding to the AKAPs was inhibited by addition of the Ht31 peptide (data not shown). The results of the mutant analysis suggest a model in which AKAP240 is associated with the central pair apparatus and AKAP97 is a component of the radial spoke.
AKAP240 Is Localized to a Substructure of the Central Pair Apparatus
To further define the location of AKAP240, RII overlays were performed on axonemes from mutants defective in various components of the central pair apparatus (for a review see Smith and Lefebvre 1997a). RII overlays revealed that, in addition to pf18, AKAP240 is also absent in axonemes from other mutants that are defective for the entire central pair apparatus, including pf15 and pf19 (Fig. 2 C). AKAP240 is reduced in axonemes from the mutant pf20, in which both microtubules of the central pair are unstable (Smith and Lefebvre 1997b). However, AKAP240 is present in axonemes from a mutant defective in only the C1 microtubule (pf16), as well as in axonemes from a mutant that is lacking a projection of the C1 microtubule (pf6; Dutcher et al. 1984). Based on this mutant analysis, we propose a model in which AKAP240 is associated with the C2 microtubule of the central pair apparatus. The localization of PKA on only the C2 microtubule of the central pair apparatus is consistent with the hypothesis that regulatory components are asymmetrically localized on the central pair apparatus, a feature which may be significant in the control of flagellar waveform (Smith and Lefebvre 1997a; Porter and Sale 2000).
AKAP97 Is RSP3
To further localize AKAP97 within the radial spoke, we analyzed axonemes from several different radial spoke mutants by performing RII overlays. The results indicated that although AKAP97 is absent in pf14 (lacking the entire radial spoke), AKAP97 is present in pf17, which is defective for the radial spoke head only (Fig. 3a and Fig. B). This suggests that AKAP97 is associated with the radial spoke stalk. We also examined pf27, a radial spoke mutant in which five radial spoke components that are normally phosphorylated are underphosphorylated in the mutant (Piperno et al. 1981). Three of the five proteins, RSPs 2, 3, and 13, are present in greatly reduced amounts (Huang et al. 1981). The overlay revealed that AKAP97 is greatly reduced in pf27, and also appears to be shifted slightly more positive on an SDS-PAGE gel, consistent with an underphosphorylated state.
Because RSP3 has been previously characterized to migrate at ∼97 kD on an SDS-PAGE gel, we hypothesized that AKAP97 may be identical to RSP3. To test this hypothesis, we performed an RII overlay of radial spoke mutant axonemes, and then probed the same blot with an antibody to RSP3. An identical staining pattern of AKAP97 was observed for both the RII overlay and analysis with anti-RSP3 (Fig. 3 C), indicating that AKAP97 corresponds to RSP3. Interestingly, although the predicted size of RSP3 is only 57 kD, RSP3 migrates on an SDS-PAGE gel to a position nearly two times its predicted molecular weight. This anomalous migration is a common feature among members of the AKAP family (Lester et al. 1996).
RSP3 Contains a Single RII-binding Site
To confirm that AKAP97 is RSP3, we expressed recombinant RSP3 as a GST fusion protein in E. coli and tested for RII binding using an RII blot overlay. The overlay demonstrated that recombinant RSP3 binds to RII (Fig. 4 A). RII binding was specifically inhibited by addition of the Ht31 peptide (data not shown). These results provided verification that RSP3 is an AKAP.
To map the location of the RII-binding domain within the amino acid sequence of RSP3, we expressed truncated forms of RSP3 as GST fusion proteins in E. coli. Expression of the protein fragments was monitored by Coomassie staining and Western blot analysis with an antibody to GST. Using RII blot overlays, we then tested the various RSP3 truncations for RII binding (results shown in Fig. 4 A). A COOH-terminal deletion of amino acids 221–516 had no effect on RII binding. However, an NH2-terminal deletion of amino acids 1–193 completely abolished RII binding, indicating that the RII-binding site is contained in the NH2-terminal portion of RSP3. By testing smaller protein fragments, we determined that RSP3 contains a single RII-binding domain located between amino acids 144–180. Curiously, RII binding to the 144–180 fragment is weaker than to larger fragments. One possible explanation for this observation is that the close proximity of the GST protein to the RII-binding site in the 144–180 fragment results in steric hindrance by GST. However, it cannot be ruled out that residues NH2-terminal of amino acid 144 enhance the binding of RII to RSP3.
RSP3 Contains an RII-binding Motif Similar to Other AKAPs
Recently the AKAP domain for RII binding has been refined, and conserved residues within the amphipathic helix have been proposed (Vijayaraghavan et al. 1999). Most of the conserved residues are hydrophobic, and in many cases only conservative substitutions appear to be allowed. We analyzed RSP3 amino acids 144–180 to determine whether they contain this proposed RII-binding motif. Comparative alignment of these amino acids with the RII-binding domains of other AKAPs revealed that amino acids 161–178 have sequence homology with other RII-binding domains. (Fig. 4 B). RSP3 161–178 contains 7 of 8 conserved amino acids when allowing for moderately conservative substitutions at amino acids 166 and 177 (valine for alanine/serine). Moreover, like the RII-binding domains of other AKAPs, amino acids 161–178 are predicted to form an amphipathic helix (Fig. 4 C).
To provide definitive evidence that amino acids 161–178 comprise the RII-binding domain of RSP3, we performed a point mutation at amino acid 170. Using the construct containing amino acids 104–180, the central, conserved residue, leucine 170, was replaced by a proline residue. We predicted that this substitution would disrupt the amphipathic helix and abolish RII binding. We expressed the construct as a GST fusion protein in bacteria, and tested its ability to bind to RII using an RII blot overlay. The analysis revealed that the leucine to proline substitution completely abolished RII binding to RSP3 (Fig. 4 D). To ensure that the loss of RII binding in the leucine to proline substitution did not occur as a result of nonspecific secondary structural alterations, we performed a more structurally conservative amino acid substitution in which valine 169 and leucine 170 were replaced by alanines. The substitutions again resulted in the complete loss of RII binding, indicating that these residues are essential for RII to bind to RSP3 (Fig. 4 D). Furthermore, the substitutions correspond to mutations in other AKAPs that disrupt RII binding (Glantz et al. 1993; Westphal et al. 2000), providing strong evidence that RSP3 is an AKAP with an RII-binding domain localized to amino acids 161–178.
The RII-binding Site in RSP3 Is Well Conserved in RSP3 Orthologs
Orthologs to RSP3 have been identified in human, mouse, and Drosophila databases. Alignment of the RII-binding site of RSP3 with the corresponding residues in these orthologs reveals that the RII-binding site is well conserved (Fig. 5). The locations of the orthologs in their respective organisms is not known, but since the axoneme is a highly conserved structure, they most likely will be found in the radial spokes of flagellar and perhaps ciliary axonemes. It is important to test whether any of the orthologs is also an AKAP. If so, it suggests that the AKAP is an important component of the axoneme, and therefore may also be important in the regulation of motility. There is some evidence that AKAPs indeed play an important role in the control of flagellar motility. For example, cell-permeable analogues of Ht31 have been shown to inhibit sperm motility (Vijayaraghavan et al. 1997).
The Role of RSP3 in Chlamydomonas Flagellar Axonemes
Genetic and biochemical experiments in Chlamydomonas have demonstrated that the radial spokes, and in particular RSP3, are important in the regulation of flagellar motility (for reviews see Brokaw et al. 1982; Curry and Rosenbaum 1993). A mutant defective in the gene encoding RSP3, pf14, lacks radial spokes and has paralyzed flagella. Based on our data, we hypothesize that in addition to being required for assembly of the radial spokes (Witman et al. 1978; Huang et al. 1981; Diener et al. 1993), RSP3 has an additional function of localizing PKA in a position to control dynein-driven flagellar motility. Consistent with this hypothesis, several experiments have shown that the radial spokes control dynein activity by regulating axonemal PKA (Smith and Sale 1992; Howard et al. 1994; Habermacher and Sale 1997). In addition, genetic studies of suppressor mutations have identified a dynein control system, which, in the absence of the radial spokes, inhibits dynein activity (Huang et al. 1982).
Genetic and molecular analyses have demonstrated that RSP3 is located at the base of the radial spoke stalk, adjacent to the outer doublet microtubules and near the inner dynein arms (Fig. 3 A). A basic region of RSP3, consisting of amino acids 42–80, is necessary and sufficient for RSP3 to be targeted to the outer doublet microtubules (Diener et al. 1993). This domain is analogous to the AKAP targeting domain. In addition to the RII-binding domain, each AKAP contains a unique targeting domain that localizes the AKAP and associated PKA to a specific intracellular site (Edwards and Scott 2000). Thus, amino acids 42–80, formerly designated as the axoneme-binding domain, can now also be considered the AKAP targeting domain. However, the axonemal protein(s) that binds to the targeting domain and facilitates the localization of RSP3 to the outer doublet microtubules remains unknown.
One disadvantage for our study of RSP3 is that, along with other proteins of the radial spoke, RSP3 is not easily solubilized and forms a strong association with the axoneme (Piperno et al. 1981). Previously thought to be insoluble, a novel extraction method has recently shown that RSPs can be solubilized by 0.5 M KI (Yang, P., D.R. Diener, J.L. Rosenbaum, and W.S. Sale, submitted for publication). The relative insolubility of RSP3 has precluded our efforts to show an association with either the endogenous regulatory or catalytic PKA subunits through traditional approaches such as cAMP-affinity purification or immunoprecipitation. The chaotropic conditions required for the solubilization of RSP3 likely also cause a dissociation of any associated enzymes. Equally as challenging, little information is known about Chlamydomonas PKA, and no identifying reagents are currently available. Further studies of the role of PKA and AKAPs in the axoneme will require a dedicated effort in the development of these reagents.
In summary, we have identified an AKAP that is associated with the central pair apparatus, and an AKAP that is a component of the radial spoke. Recent studies have shown that in addition to an AKAP, the radial spoke also contains a kinase, as well as calmodulin (Yang, P., D.R. Diener, J.L. Rosenbaum, and W.S. Sale, submitted for publication). We propose that RSP3 localizes PKA to a position that, in conjunction with other components of the radial spoke, regulates dynein activity and controls flagellar motility. Future experiments will test this prediction and examine the physiological role of the AKAP in the flagellar axoneme.
The authors would like to thank John Scott (Vollum Institute, Oregon Health Sciences University) for helpful discussion, experimental advice, and the generous gift of the RIIα plasmid and Ht31 peptide. We would like to gratefully acknowledge Katrina Waymire and Pinfen Yang for experimental advice, and Lynne Quarmby, Grant MacGregor, and Pinfen Yang for thoughtful discussion of the manuscript.
Abbreviations used in this paper: AKAP, A-kinase anchoring protein; GST, glutathione-S-transferase; PKA, cAMP-dependent protein kinase; RSP, radial spoke protein.