Regulated Targeting of a Protein Kinase into an Intact Flagellum

In the green alga Chlamydomonas reinhardtii flagellar adhesion between gametes of opposite mating types leads to rapid cellular changes, events collectively termed gamete activation, that prepare the gametes for cell-cell fusion. As is true for gametes of most organisms, the cellular and molecular mechanisms that underlie gamete activation are poorly understood. Here we report on the regulated movement of a newly identified protein kinase, Chlamydomonas aurora/Ipl1p-like protein kinase (CALK), from the cell body to the flagella during gamete activation. CALK encodes a protein of 769 amino acids and is the newest member of the aurora/Ipl1p protein kinase family. Immunoblotting with an anti-CALK antibody showed that CALK was present as a 78/80-kDa doublet in vegetative cells and unactivated gametes of both mating types and was localized primarily in cell bodies. In cells undergoing fertilization, the 78-kDa CALK was rapidly targeted to the flagella, and within 5 min after mixing gametes of opposite mating types, the level of CALK in the flagella began to approach levels normally found in the cell body. Protein synthesis was not required for targeting, indicating that the translocated CALK and the cellular molecules required for its movement are present in unactivated gametes. CALK was also translocated to the flagella during flagellar adhesion of nonfusing mutant gametes, demonstrating that cell fusion was not required for movement. Finally, the requirement for flagellar adhesion could be bypassed; incubation of cells of a single mating type in dibutyryl cAMP led to CALK translocation to flagella in gametes but not vegetative cells. These experiments document a new event in gamete activation in Chlamydomonas and reveal the existence of a mechanism for regulated translocation of molecules into an intact flagellum.

In the green alga Chlamydomonas reinhardtii flagellar adhesion between gametes of opposite mating types leads to rapid cellular changes, events collectively termed gamete activation, that prepare the gametes for cell-cell fusion. As is true for gametes of most organisms, the cellular and molecular mechanisms that underlie gamete activation are poorly understood. Here we report on the regulated movement of a newly identified protein kinase, Chlamydomonas aurora/Ipl1p-like protein kinase (CALK), from the cell body to the flagella during gamete activation. CALK encodes a protein of 769 amino acids and is the newest member of the aurora/ Ipl1p protein kinase family. Immunoblotting with an anti-CALK antibody showed that CALK was present as a 78/80-kDa doublet in vegetative cells and unactivated gametes of both mating types and was localized primarily in cell bodies. In cells undergoing fertilization, the 78-kDa CALK was rapidly targeted to the flagella, and within 5 min after mixing gametes of opposite mating types, the level of CALK in the flagella began to approach levels normally found in the cell body. Protein synthesis was not required for targeting, indicating that the translocated CALK and the cellular molecules required for its movement are present in unactivated gametes. CALK was also translocated to the flagella during flagellar adhesion of nonfusing mutant gametes, demonstrating that cell fusion was not required for movement. Finally, the requirement for flagellar adhesion could be bypassed; incubation of cells of a single mating type in dibutyryl cAMP led to CALK translocation to flagella in gametes but not vegetative cells. These experiments document a new event in gamete activation in Chlamydomonas and reveal the existence of a mechanism for regulated translocation of molecules into an intact flagellum.
Cell-cell interactions leading to fusion between gametes of opposite sexes during fertilization are complex processes that involve dramatic changes in each of the interacting gametes. In most multicellular organisms, interactions between adhesion molecules/receptors on the sperm plasma membrane and ligands on the egg surface activate poorly understood signaling pathways that bring about transformation of the sperm into a fusion competent, activated gamete (1). For example, the sperm surface is remodeled as a consequence of the acrosome reaction, an event that accompanies gamete activation, and previously existing adhesion molecules are mobilized to new sites on the sperm (2)(3)(4). The molecular mechanisms that underlie and regulate signal transduction and movement of molecules between different gamete compartments during gamete activation largely are unknown (2).
As in multicellular organisms, gamete activation and fertilization in the unicellular green alga Chlamydomonas (5) depend on adhesion-induced signaling pathways and intercompartmental communication. At the completion of gametogenesis, during which asexually growing mtϩ and mtϪ vegetative cells differentiate into sexually competent cells, the resulting mtϩ and mtϪ gametes bear mating type-specific adhesion molecules, agglutinins, on their flagellar membranes and on the plasma membrane of the cell body. The agglutinins on the flagella are in an active state and are capable of binding to flagellar agglutinins on gametes of the opposite mating type, whereas agglutinin molecules on the contiguous plasma membrane of the cell body are inactive (6).
When mtϩ and mtϪ gametes are mixed together, random contacts between their highly motile flagella bring the mtϩ and mtϪ flagellar agglutinins together. In addition to binding the flagella of gametes of opposite mating types to each other, these receptor/ligand-like interactions also initiate a complex series of events termed gamete activation. One of the earliest documented steps in gamete activation triggered by agglutinin interactions is activation of a gamete-specific flagellar adenylyl cyclase and generation of cAMP (7)(8)(9)(10)(11). As part of an intricate feedback mechanism required to maintain and enhance flagellar adhesiveness and to keep the cells bound to each other until cell-cell fusion occurs, cAMP levels increase in the cell bodies of the gametes of both mating types during flagellar adhesion and lead to increased flagellar adhesiveness (6,9,12,13).
The increased cellular levels of cAMP also induce additional events in the cell body, including regulated secretion of molecules required for release and degradation of the extracellular matrix (cell wall), activation of fusion organelles called mating structures on the apical ends of the interacting cells between their sets of flagella, and phosphorylation of an mtϩ gametespecific homeodomain protein, GSP1 (14,15). Adhesion and fusion of the mating structures on the mtϩ and mtϪ gametes are followed by complete merging of the two cell bodies. Fusion itself also induces the flow of signals from the cell body to the flagella, triggering inactivation and loss of the flagellar agglutinins (16,17), flagellar resorption, and zygote maturation (reviewed in Refs. 18 and 19).
Studies on the flagellar adenylyl cyclase, a key regulatory enzyme in gamete activation, have shown that, like the adenylyl cyclases of gametes in multicellular organisms (20), the Chlamydomonas enzyme appears not to be regulated by G proteins. Instead, we have found that the flagellar adenylyl cyclase is regulated by protein phosphorylation and dephosphorylation. A flagellar membrane-associated protein kinase activity in nonactivated gametes inhibits the adenylyl cyclase (10). During gamete activation, agglutinin interactions relieve this inhibition and also stimulate a second protein kinase whose activity is required to activate adenylyl cyclase (11). Concomitantly, flagellar adhesion leads to the inhibition of a third protein kinase, whose substrate itself is yet another protein kinase (GenBank TM accession number U36196) (11,21). In addition to these protein kinase activities involved in the very early stages of gamete activation upstream of generation of cAMP, several protein kinases act downstream of cAMP. For example, the recently discovered homeodomain protein GSP1 is phosphorylated within minutes after gametes of opposite mating types are mixed together. GSP1 phosphorylation can be induced by incubation of mtϩ gametes in dibutyryl cAMP, thus bypassing flagellar adhesion in the gamete activation pathway (15).
To learn more about gamete activation and intercompartmental communication in Chlamydomonas, we have begun to focus on a new protein kinase that undergoes activation-dependent changes during fertilization. Here we report that a novel member of the aurora/Ipl1p family of protein kinases is translocated to the flagella during gamete activation, within minutes after gametes adhere to each other. Translocation of this Chlamydomonas aurora/Ipl1p-like protein kinase, CALK, 1 is induced by flagellar adhesion in the absence of cell fusion and upon incubation of gametes of a single mating type in dibutyryl cAMP. Moreover, the regulated translocation of CALK is specific to gametes.

MATERIALS AND METHODS
Cells and Cell Culture-Chlamydomonas reinhardtii strains 21gr (mtϩ) (CC-1690), 6145C (mtϪ) (CC-1691), and imp1-15 (mtϩ) (CC-462), available from the Chlamydomonas Genetic Center, Duke University, were cultured either with medium I or medium II of Sager and Granick (22) at 23°C on a 13:11 h light/dark cycle as described previously (23). Vegetative cells were induced to become gametes by incubation in medium without nitrogen (N-free medium) followed by culturing in continuous light at room temperature (23).
Treatment of Cells with Dibutyryl cAMP-For experiments with dibutyryl cAMP, gametes in N-free medium and vegetative cells in medium II were incubated in 15 mM dibutyryl cAMP and 0.15 mM papaverine for 30 min with aeration. Cell wall loss, which is a measure of gamete activation, was assessed by determining whether cells became sensitive to disruption by incubation in 0.075% Triton 100-X, 0.5 mM EDTA, 10 mM Tris, pH 8.0, as described earlier (24,25).
Flagellar Isolation-Flagella were isolated essentially as described in Zhang et al. (10). Typically, 3-4 liters of cells were concentrated to 30 ml by centrifugation at 3500 ϫ g for 5 min at 4°C, and ice-cold 25% sucrose in 10 mM Tris, pH 7.2, was added to yield a final concentration of 7% sucrose. While stirring the suspension, its pH was rapidly decreased to 4.5 by the addition of 0.5 M acetic acid; after the flagella were detached (which typically required about 20 s) the pH was raised to 7.2 with 0.5 M KOH. All subsequent steps were carried out at 4°C. The suspension of cells and flagella was underlayed with 25% sucrose in 10 mM Tris, pH 7.2, and centrifuged for 10 min at 2500 ϫ g. The upper phase that contained flagella and a few remaining cell bodies was underlayed with 25% sucrose, 10 mM Tris, pH 7.2, and centrifuged again as above. The upper phase containing purified flagella was carefully removed and centrifuged at 9000 ϫ g for 8 min. The sedimented flagella were resuspended in buffer A (20 mM HEPES, pH 7.2, 5 mM MgCl 2 , 1 mM EDTA, 1 mM dithiothreitol) containing a 1/100 dilution of the Sigma protease inhibitor mixture for plant cells (Sigma catalogue number P9599) and were flash frozen in liquid nitrogen.
Protein Determination-Protein concentration was determined by use of a Bio-Rad protein assay kit with bovine serum albumin (Albumin Standard from Pierce) as a standard.
Cloning of the calk cDNA-A 426-base pair calk genomic fragment was first cloned from a polymerase chain reaction product obtained by amplification of Chlamydomonas genomic DNA with degenerate primers (sense primer, (A/G)T(C/G/T)-(A/G)TI-TT(C/T)-(A/G)CI-GA(C/T)-AT; antisense primer, (A/G)TI-(C/G)GN-CC(A/G)-TA(A/G)-TA(A/G)-TC) originally designed for amplification of adenylyl cyclase. A probe for screening a Lambda II genomic library (kindly provided by Paul Lefebrve, University of Minnesota) was prepared from the cloned polymerase chain reaction amplicon by random labeling using a rapid labeling kit from Roche Molecular Biochemicals. The same probe also was used to screen a ZapII cDNA library prepared from activated mtϩ gametes (26,27). The screen of the cDNA library yielded a single positive clone from 30,000 plaques. The plaque was picked and rescreened by polymerase chain reaction using unique primers designed from the sequence of the first cloned polymerase chain reaction product. The calk phagemid clone was in vitro excised as recommended by the manufacturer (Stratagene, San Diego, CA), yielding a recombinant pBluescriptII plasmid containing calk cDNA. The cDNA clone contained a 3.2-kilobase insert, which was sequenced in both directions by automated sequencing methods. Portions of 1 of 3 genomic clones obtained were used to confirm the sequence of the cDNA clone. The nucleotide sequence of the cDNA, which is termed CALK (see "Results"), is available from the GenBank TM (accession number AF199021). The cDNA contained a 769-amino acid open reading frame. The presence of two stop codons upstream was confirmed by comparison to that region of the genomic clone. For routine propagation of cloned DNA, plasmid constructs were transformed into Escherichia coli DH5␣ cells.
Sequence Analysis-The amino acid sequences of CALK and related proteins were aligned with the ALLALL sequence alignment server, and the alignment shown in Fig. 2 was further refined by hand according to known protein kinase subdomains (28). The phosphorylation sites were predicted by analysis with PhosphoBase software (29).
For Southern blot analysis, 10 g of DNA was digested with BamHI, HindIII, and KpnI, and the fragments were separated on 1% agarose gels and blotted onto nylon membranes (Schleicher & Schuell) using standard methods. The blot was probed with the random primer-labeled (rapid labeling kit, Roche Molecular Biochemicals) genomic fragment of calk in Bright-star hybridization solution (Ambion) at 55°C. The blot was washed in 0.1% SDS, 15 mM NaCl, 1.5 mM sodium citrate (0.1 ϫ SSC) followed by washing in buffer containing 0.5% SDS, 0.1 ϫ SSC at 65°C for 30 min. The blot was then exposed to x-ray film and developed.
Recombinant Protein Expression and Purification-To prepare a Histagged, truncated, recombinant form of the protein, the SphI/PstI fragment of the cDNA corresponding to amino acids 12-343, which includes the entire protein kinase domain, was cloned into expression vector pQE30 (Qiagen) in frame with the coding sequence for six consecutive His residues (6 ϫ His) under the control of an isopropyl ␤-D-thiogalactoside-inducible lac promoter. A 1-liter culture of M15 bacteria (Qiagen) harboring the recombinant plasmid was grown at 37°C with vigorous shaking until an A 600 of 0.6 was reached. The temperature of the incubation was switched to 30°C, and 1 h later isopropyl ␤-D-thiogalactoside was added (final concentration of 0.5 mM) and incubation was continued for 2 h. The His-tagged recombinant protein was purified by use of nickel-nitrilotriacetic acid-agarose according to the instructions from the manufacturer (Qiagen). A nearly full-length recombinant, His-tagged CALK lacking the N-terminal 11 amino acids was also expressed and purified as described above.
In Vitro Protein Kinase Assay-For in vitro protein kinase assays, nearly full-length, His-tagged, recombinant CALK (100 ng in 20 mM HEPES, 1 mM EGTA, 1 mM dithiothreitol, 10 mM benzamidine, 0.2 mM phenylmethylsulfonyl fluoride, 10% glycerol) was added to 30 l of protein kinase assay buffer (20 mM Tris, pH 8.0, 10 mM MgCl 2 , 0.1 mM dithiothreitol, 100 M ATP, 2 Ci [␥-32 P]ATP (6000 mCi/mmol, Amersham Pharmacia Biotech)) and incubated for 30 min at 30°C. Some samples also contained 5 g of dephosphorylated bovine casein (Sigma) or bovine myelin basic protein (Sigma). The reactions were stopped by the addition of SDS-PAGE sample buffer and boiling, as described below, and analyzed by SDS-PAGE and autoradiography.
SDS-PAGE-Samples for SDS-PAGE were mixed with 1/3 volume of 4ϫ SDS sample buffer (0.25 M Tris, pH 6.8, 40% glycerol, 16% SDS, 0.4 mM dithiothreitol, 0.1% bromphenol blue) and boiled for 5 min. In some experiments sample buffer was used at a final concentration of 2ϫ. The samples were subjected to electrophoresis in 9% acrylamide (15,30) minislab gels at 30 mA in buffer containing 25 mM Tris, 192 mM glycine, 0.1% SDS and then transferred for immunoblot analysis (see below). Typically 15-30 g of protein was loaded in each lane.
Immunoblot Analysis-For immunoblot analysis of CALK, sedimented whole cells or isolated flagella were subjected to electrophoresis. After SDS-PAGE, proteins were transferred to a polyvinylidene difluoride membrane (Immobilon P, Millipore, Bedford, MA) in buffer containing 25 mM Tris, 192 mM glycine, 20% methanol at 100 V for 1 h or at 35 V overnight at 4°C. The membrane was blocked with 5% Carnation dry milk (Nestles) in 20 mM Tris, pH 7.6, 137 mM NaCl, 0.05% Tween-20 (TBST) for 1 h and then incubated either with preimmune serum or with anti-CALK serum (obtained from a rabbit immunized with the truncated, His-tagged CALK) in 3% Carnation dry milk in TBST for 1 h. The membrane was washed three times for 5 min each with TBST followed by incubation for 1 h with a horseradish peroxidase-conjugated goat anti-rabbit IgG antibody (Bio-Rad) diluted 1/10,000 in TBST containing 3% Carnation dry milk. The membrane was washed as before and incubated in ECL immunoblotting reagents (Amersham Pharmacia Biotech) for 1 min as described by the manufacturer, exposed to Hyperfilm ECL (Amersham), and developed in an automatic film processor. In some experiments, after the immunoblot membrane was exposed, total proteins were visualized by staining with 0.1% Coomassie Brilliant Blue R-250 in 40% methanol, 10% acetic acid, followed by destaining with 40% methanol, 10% acetic acid.

Molecular
Cloning of a cDNA Encoding a Protein Kinase-To investigate regulatory molecules in gamete activation, we characterized a newly obtained 3.2-kilobase cDNA that encodes a protein kinase (see "Materials and Methods"). The cDNA has an open reading frame of 2307 base pairs beginning with the ATG codon at nucleotide position 99 through the termination codon at position 2406. The ATG codon at position 99 was chosen because of two in-frame stop codons immediately upstream from it and because of its association with a conventional initiation sequence (31). The cDNA ends with a long poly(A) tail and contains a putative polyadenylation signal (AATGTA) 19 nucleotides upstream of the poly(A) sequence (see nucleotide sequence in the GenBank TM ), similar to those found in other Chlamydomonas transcripts (32,33). The cDNA predicted a polypeptide containing 769 amino acids with a molecular mass of 80.6 kDa and a pI of 9.76 (Fig. 1A). Analysis of the sequence showed that the polypeptide contains the 12 subdomains known to be conserved in the protein kinase superfamily (Fig. 1B, shaded). The presence of the sequence DIK-PEN in subdomain VIb indicates that the molecule belongs to the serine-threonine family of protein kinases (28). The sequence analysis also showed that the protein does not contain a putative signal peptide or a predicted transmembrane domain, suggesting that it is a cytoplasmic protein.
Sequence Analysis Shows That CALK Is a Member of the Aurora/Ipl1p Protein Kinase Family-A BLAST (34) search of the GenBank TM data bases indicated that the molecule showed highest similarity to members of the aurora/Ipl1p-like protein kinase family, many of which are regulators of chromosome segregation and cytokinesis (reviewed in Refs. 40 and 41). The first protein on the list of similar proteins found in the BLAST search was the human protein kinase, aurora2 (accession number AF011468). Within the protein kinase catalytic domain, the Chlamydomonas polypeptide exhibited 36% identity and 56% similarity over 279 amino acids with aurora2. Aurora2 (also The asterisks below subdomain VII indicate amino acids suggested to be a consensus sequence that is a signature for aurora/Ipl1p family members (40). named STK15, BTAK, and ARK1) is amplified and overexpressed in multiple tumor cell types, localizes to centrosomes, and when expressed ectopically in cells in culture, induces centrosomal duplication/distribution abnormalities and transformation (35)(36)(37)(38)(39). Additional members of this family (40,41) that were among the highest scoring protein kinase sequences in the BLAST search included other mammalian protein kinases (37,42,43), pEg2 in Xenopus (44), AIR1 in Caenorhabditis elegans (45,46), aurora in Drosophila (47), and Ipl1p in Saccharomyces cerevisiae (48).
Because of the similarity of this new Chlamydomonas protein kinase to members of the aurora/Ipl1p family, we have named it Chlamydomonas aurora/Ipl1p-like protein kinase, CALK. A sequence alignment of the protein kinase subdomains of CALK with aurora2 and with the two original members of this emerging family, aurora from Drosophila and Ipl1p from S. cerevisiae, is shown in Fig. 1B. The protein kinase domains of CALK, several aurora/Ipl1p family members, and several other types of protein kinases were used to make the phylogenetic tree shown in Fig. 2A, which is consistent with the BLAST search results indicating that CALK is more related to aurora/Ipl1p protein kinases than it is to other members of the protein kinase superfamily.
The BLAST search and visual examination indicated that the similarity between CALK and other members of the aurora/ Ipl1p protein kinase family was restricted to the protein kinase domain as depicted in Fig. 2B. For most of the members of this family, the protein kinase catalytic domain represents the majority of the polypeptide. Those family members that contain significant numbers of noncatalytic domain amino acids usually have them in an N-terminal extension, and some family members share limited sequence similarity in these N-terminal domains (41). Interestingly, CALK has few amino acids Nterminal to the catalytic domain; but, it has an extensive noncatalytic, C-terminal domain that makes it nearly twice as large as all other known aurora/Ipl1p family members.
The large C-terminal domain of CALK has several features that are notable. Two regions are enriched in basic amino acids. The sequence from 491-578 has a pI of 12.5 and is rich in Gly (19/88 amino acids; 22% Gly). A second basic region (amino acids 579 -677) immediately C-terminal to the first has a pI of 12.8 and shows similarity to the Ser, Pro-rich microtubulebinding domain of the microtubule-associated protein MAP4 (Fig. 1A, single underline) (49 -51). The very C-terminal portion of CALK contains an acidic region that includes a PEST sequence (amino acids 738 -755) with a PEST score of 27 (Fig.  1A, double underline) (52). Several consensus phosphorylation sites also are present throughout the CALK sequence, including sites for cAMP-dependent protein kinase, protein kinase C, calcium/calmodulin-dependent protein kinase II, and p34cdc2 (not shown).
Southern blot hybridizations of Chlamydomonas genomic DNA with a nucleotide probe derived from calk genomic DNA (Fig. 3) indicated that CALK is encoded by a single copy gene. By use of RFLP mapping, the calk gene maps near the Gs2 and Lc3 loci on linkage group XII. 2 Recombinant CALK Has Protein Kinase Activity-To determine if CALK had protein kinase activity, we evaluated the ability of the nearly full-length (lacking the N-terminal 11 2 P. Kathir, C. Silflow, and P. Lefebvre, personal communication.

FIG. 2. Comparison of CALK to other aurora/Ipl1p protein kinases.
A, the protein kinase catalytic domains from the sequences indicated in the tree were aligned using Clustal W (92). The distance matrix was calculated from the alignment using the PROTDIST program in the PHYLIP package, and the tree was built by the neighbor-joining method using the NEIGHBOR program in the PHYLIP package (93). The tree was viewed with TreeViewPPC.1 (94). amino acids) His-tagged, bacterially expressed form of CALK to undergo autophosphorylation and to phosphorylate casein and myelin basic protein in in vitro protein kinase assays. As shown in the autoradiograph in Fig. 4, CALK underwent autophosphorylation and also phosphorylated myelin basic protein. Unlike the human aurora2 protein (53), CALK showed low protein kinase activity toward casein (Fig. 4).

CALK Exists in Two Forms in Vegetative Cells and Gametes of Both Mating Types and Is Present Primarily in Cell Bodies-
To learn more about the cellular properties of CALK, we investigated its expression in vegetative cells and gametes. A rabbit polyclonal antiserum raised against a His-tagged, recombinant, truncated CALK containing the protein kinase domain was used in immunoblot analysis of whole cell lysates (see "Materials and Methods"). Fig. 5A shows that the anti-CALK antibody reacted with a protein doublet of 78/80 kDa in synchronously growing vegetative cells of both mating types and in gametes of both mating types. Antibody reactivity with the doublet was blocked when the primary antibody was absorbed with recombinant CALK protein, indicating the specificity of the antibody (Fig. 5B, bottom panel). As expected, the apparent molecular mass of the proteins in the doublet was similar to the mass of CALK (80 kDa) predicted from the amino acid composition.
With few exceptions (see "Discussion"), most of the aurora/ Ipl1p family members are expressed only in dividing cells and are associated with microtubule-containing structures such as centrosomes and the mitotic spindle. On the other hand, even though Chlamydomonas gametes are nondividing cells (unless returned to N-containing medium), they expressed CALK. To determine if CALK was associated with one of the most prominent microtubule-containing structures in Chlamydomonas, the flagella, we analyzed the cellular distribution of CALK. Our results indicated that it was present primarily in cell bodies, with very low amounts detectable in flagella (Fig. 6).
CALK Moves to the Flagella during Gamete Activation-Previous studies from our laboratory have demonstrated that protein kinases play key roles in Chlamydomonas fertilization (10,11,15,54,55). To determine if CALK underwent changes during cell-cell interactions, we investigated the levels of this protein kinase in flagella of mtϩ and mtϪ gametes undergoing fertilization. To do this, mtϩ and mtϪ gametes were mixed together and at various times after mixing, we harvested the cells and isolated their flagella. Surprisingly, as shown in Fig.  7 (immunoblot, upper panel), gamete adhesion led to a rapid and striking accumulation of CALK in the flagella. Moreover, only the 78-kDa form of CALK appeared in the flagella (Fig. 7). The lower panel in Fig. 7 shows the immunoblot after staining with Coomassie Blue.
In this experiment with mtϩ and mtϪ wild type gametes, phase contrast examination showed that nearly 70% of the cells had fused within 30 min after mixing. To test if CALK appearance in the flagella required cell fusion or if flagellar adhesion and gamete activation were sufficient to induce translocation, we examined flagellar levels of CALK during cell-cell adhesion of imp1-15 mtϩ gametes. These impotent cells undergo flagellar adhesion and gamete activation after being mixed with mtϪ gametes but are unable to fuse (56). When we mixed the gametes together, they underwent normal flagella adhesion (not shown), and immunoblotting showed that CALK translocated to the flagella (Fig. 8A). These results indicated that cell-cell fusion was not required for appearance of CALK in the flagella.
In earlier studies we showed that flagellar adhesion and gamete activation induce new protein synthesis (57). To determine if the appearance of increased CALK in the flagella during activation required synthesis of new proteins, we mixed mtϩ and mtϪ gametes together in the presence of the protein synthesis inhibitor, cycloheximide, and then analyzed CALK by immunoblotting. The results shown in Fig. 8B demonstrated that cycloheximide treatment did not block the appearance of CALK in the flagella. Thus, the increased CALK in flagella during flagellar adhesion resulted from movement of pre-existing CALK and did not require synthesis of new CALK. Moreover, the results indicated that all of the cellular machinery required for CALK translocation was present in the gametes before they were mixed together.
CALK Translocation Can Be Induced by Dibutyryl cAMP-One of the important molecules in gamete activation is the cyclic nucleotide, cAMP, which undergoes 10-fold increases in cellular levels within minutes after gametes are mixed together. Because many of the cellular events that comprise gamete activation can be induced by incubation of gametes of a single mating type in dibutyryl cAMP, we investigated the influence of this molecule on CALK. To do this, mtϩ gametes were incubated with dibutyryl cAMP and the phosphodiesterase inhibitor, papaverine; after 30 min, the cells were harvested and their flagella were isolated and immunoblotted for CALK. An assay for cell wall loss, one cellular response to  cAMP and an indicator of gamete activation, indicated that the dibutyryl cAMP/papaverine indeed induced gamete activation (data not shown). Moreover, as shown in Fig. 8C (upper panel), dibutyryl cAMP also induced movement of CALK to the fla-gella. These results indicated that flagellar adhesion per se was not required for translocation and also showed that CALK translocation is a part of gamete activation.
Both vegetative cells and gametes contain adenylyl cyclase activity (7,8,54,58) and use cAMP for regulation of cellular activities (59,60). When we examined the behavior of CALK in vegetative cells incubated in dibutyryl cAMP, however, we found that movement of CALK to the flagella was not induced (Fig. 8C, lower panel). These results indicated that cAMPinduced CALK translocation is a unique property of gametes. DISCUSSION The current studies have uncovered a novel cellular mechanism in gamete activation, regulated translocation of a protein kinase into intact flagella. Whereas only small amounts of CALK were detectable in flagella isolated from nonadhering gametes, flagella of adhering gametes acquired significantly increased amounts of the aurora/Ipl1p-like protein kinase, amounts that approximated levels normally found in the cell body. Results from experiments with nonfusing mutant gametes and with gametes of a single mating type incubated with dibutyryl cAMP support the conclusions that CALK appearance in the flagella does not require cell fusion or even flagellar adhesion per se, that CALK translocation is a part of gamete activation, and that CALK appearance in the flagella is because of movement of pre-existing CALK to the organelles.
The initial experiments with adhering wild type mtϩ and mtϪ gametes showed that CALK appeared in the flagella very soon after the cells were mixed, becoming detectable within 1 min after mixing, and appearing in significant quantities by 5 min after mixing. That only the 78-kDa form of CALK was translocated remains enigmatic. It is possible that both forms were translocated but that the 80-kDa form underwent a posttranslational modification that rendered it unable to bind to the anti-CALK antibodies. This seems unlikely, though, because the anti-CALK antibody is a polyclonal antibody raised against the entire protein kinase domain of the protein. Another possibility is that both forms were translocated, and the 80-kDa form was rapidly converted to the 78-kDa form in the FIG. 5. CALK expression. A, equal amounts of protein (30 g) from vegetative (V) cells and gametes (G) were prepared for SDS-PAGE and immunoblot analysis. B, the CALK antibody is specific. Anti-CALK antiserum (1.5 l) was incubated for 30 min at room temperature with 20 l (ϳ3 g of protein) of the nearly full-length, affinity-purified, recombinant CALK, the sample was diluted up to 7.5 ml with 3% Carnation milk in TBST and used for immunoblotting as described under "Materials and Methods." The nonabsorbed sample was incubated with anti-CALK antisera at the same dilution as the absorbed antisera. flagella, or the 80-kDa form may not be able to interact with the cellular machinery that is responsible for translocation. Future experiments examining the mechanisms of translocation should help to distinguish among these and other possible explanations for the apparent selectivity of the translocation process.
Studies with an impotent, nonfusing mutant demonstrated that cell fusion was not required for CALK movement. When imp1-15 mtϩ gametes were mixed with wild type mtϪ gametes, the appearance of CALK in the flagella was indistinguishable from that observed in wild type cells (Fig. 8A). These results indicating that CALK translocation was part of gamete activation and did not require cell fusion were confirmed in related experiments. Incubation of mtϩ gametes in dibutyryl cAMP for 30 min also led to movement of CALK to the flagella (Fig. 8C). Moreover, the effects of dibutyryl cAMP were unique to gametes. Incubation of vegetative cells in this cyclic nucleotide did not induce CALK translocation to the flagella (Fig. 8C).
Because Chlamydomonas gametes are transcriptionally active, it was possible that the appearance of CALK in the flagella was because of targeting of newly synthesized protein to these organelles. This possibility was ruled out by the experiments showing that CALK appeared in the flagella of gametes that were mixed together in the presence of the protein synthesis inhibitor, cycloheximide (Fig. 8B). Thus, CALK appearance in the flagella was because of movement of pre-existing cell body CALK into the flagella. Moreover, the results showed that the cellular machinery required for translocation also was present in unactivated gametes.
CALK Is an Aurora/Ipl1p-like Protein Kinase-Database searches and analysis by protein alignment methods indicated that CALK is a member of the aurora/Ipl1p family of serinethreonine protein kinases (Figs. 1 and 2). Aurora/Ipl1p-related protein kinases (or AIRKs, as suggested by Giet and Prigent (40)) are present in S. cerevisiae, C. elegans, Drosophila melanogaster, Xenopus laevis, mouse, rat, and humans; AIRK sequences also are present in Arabidopsis (accession numbers AC003680.1 and AC0053951) and Schizosaccharomyces pombe (accession number AL022245.2). In those cells in which AIRKs have been studied, these protein kinases are localized in centrosomes, the midbody, and at the poles of the bipolar spindle in mitotic cells and are necessary for completion of many mitotic events. The overexpression of AIRKs in several tumor cell types and the observations that ectopic expression of AIRKs in rat cells and human cells produces a transformed phenotype also indicate that AIRKs play key roles in cell division. Experiments with the Xenopus AIRK pEg2 have shown that the protein binds to spindle microtubules in the cell in anaphase and to taxol-stabilized microtubules in vitro. Moreover, dominant negative constructs of the AIRK pEg2 provoke inhibition of mitotic spindle assembly in Xenopus egg extracts. To date, the only known substrate for any AIRK is the Xenopus oocyte kinesin-related protein, XIEg5 (40), a motor protein also known to be involved in assembly of a bipolar spindle (44,61,62).
The absence of a putative signal sequence or transmembrane domain in the CALK sequence was consistent with the idea that, like all the other aurora/Ipl1p protein kinases studied, CALK is a cytoplasmic protein. Most of the AIRKs have been found only in cells in mitosis, although aurora2 (also known as STK15 and BTAK) (53) localizes to centrosomes in interphase cells in addition to being found at each spindle pole during mitosis. Because our experiments were carried out with synchronized cells that were not in mitosis, we do not know whether Chlamydomonas cells express CALK during cell division. The vegetative cells used in our studies were harvested around the middle of the light period of the light/dark cycle and were in the G 1 phase of the cell cycle (63). Such synchronously growing cells undergo mitosis only during the dark period. Moreover, gametes are in the G 0 phase of their life cycle and remain in an undividing state for weeks unless a nitrogen source is added back to their medium (64). Thus, CALK has been recruited by Chlamydomonas for uses not previously described for AIRKs.
The molecular mass of CALK and several features of its sequence also distinguish it from other AIRKs. For example, the CALK sequence is identical in only 11 of 16 (69%) of the amino acids between subdomains VII and VIII of the protein kinase catalytic domain that Giet and Prigent (40) identified as a consensus sequence signature domain for AIRKs (marked by asterisks in Fig. 1B). Even the Arabidopsis and S. pombe sequences have substitutions in at most two of the signature sites. This particular divergence of the CALK protein kinase domain from that of the other AIRKs is one example of the overall divergence reflected in the tree shown in Fig. 2A. B, incubation of gametes in the protein synthesis inhibitor cycloheximide does not block gamete activation-induced appearance of CALK in the flagella. mtϩ and mtϪ gametes were mixed together in the presence or absence of 10 g/ml cycloheximide for 30 min, and the cell bodies and flagellar were analyzed as above. At this concentration of cycloheximide, flagellar regeneration was blocked in similarly treated samples (data not shown), indicating that the protein synthesis inhibitor was effective (67,95). C, mtϩ gametes and vegetative cells were incubated with dibutyryl cAMP as described under "Materials and Methods," and cell bodies and flagella were analyzed as above.
The nearly 2-fold larger molecular mass of CALK compared with other AIRKs suggests that the nonprotein kinase region, mostly in the C terminus (Fig. 2B), contains domains that are important for its nonmitotic functions. For example, the putative microtubule-binding domain might be important for the gamete activation-associated localization to gametic flagella. In addition, whereas CALK does not have the putative D-box (degradation box) set of consensus amino acids found in the protein kinase domain of other AIRKs (40), the presence of a putative PEST sequence (Fig. 2B) at the C terminus of CALK suggests that proteolysis might be an important aspect of its regulation. It is also possible that the C terminus is involved in regulation of the protein kinase activity of CALK. The recombinant, nearly full-length CALK that was used for the in vitro protein kinase assays showed a much greater preference for myelin basic protein as a substrate than casein (Fig. 4). Assays using the truncated form of CALK should indicate if the C terminus plays a role in this specificity.
The immunoblot result showing ϳ78and ϳ80-kDa forms of CALK in cells (Fig. 5) was unexpected, and the presence of the two antigens remains unexplained. It is unlikely that the two forms of CALK are different gene products, because Southern blot analysis was consistent with a single calk gene in Chlamydomonas (Fig. 3). The two proteins might arise from differential splicing, an internal ATG start site, or posttranslational modifications such as proteolysis, phosphorylation, ubiquitination, or addition of lipid moieties.
CALK and Gamete Activation during Fertilization-It is likely that the function of CALK in flagella is more related to the sensory and signaling properties of Chlamydomonas flagella than to their motile functions (65,66). For example, it could be one of the protein kinase activities implicated in known steps in gamete activation and fertilization. As discussed in the Introduction, our laboratory has shown that the coupling of mtϩ and mtϪ agglutinin interactions to activation of adenylyl cyclase during flagellar adhesion depends on the activity of several protein kinase activities. In nonadhering gametes, the gamete-specific, non-G protein-dependent adenylyl cyclase is inhibited by a membrane-bound protein kinase and agglutinin interactions relieve this inhibition (10,11). A second protein kinase activity is required for adhesion-induced activation of adenylyl cyclase (11), and a third protein kinase activity regulates the phosphorylation of a fourth, soluble protein kinase, SksC (21,55). In other experiments Pan et al. (67) have shown that protein phosphorylation is involved in the light-dependent regulation of agglutinin activity.
In addition to those just described, other gamete-specific events could require CALK. For example, CALK may be involved in agglutinin mobilization during gamete activation. Indirect evidence has been presented from several groups that the agglutinin undergoes gamete activation-induced increases on the flagella. Greater than 90% of the total cellular agglutinin molecules are present in an inactive or cryptic form associated with the cell bodies of unactivated gametes. During gamete activation, levels of cell body agglutinins fall and flagellar adhesiveness increases, providing indirect evidence that the agglutinins are moving from the cell body to the flagella (6,9,12,68). The flagellar tips of gametes, which are normally tapered, become bulbous as amorphous material accumulates just underneath the flagellar membrane during gamete activation, and the A microtubules of the outer doublets lengthen (69). This process, termed flagellar tip activation, also coincides with accumulation of flagellar adhesion sites at the flagellar tips (68). It is possible that CALK participates in flagellar tip activation and becomes localized at the flagellar tips in association with agglutinin or even with adenylyl cyclase.
Regulated Translocation of a Protein into an Intact Flagellum-Whereas cilia and flagella are widely studied organelles (60, 70 -72), to our knowledge this is the first direct evidence for regulated translocation of a protein into an intact, nonregenerating cilium/flagellum in any eucaryotic cell. It will be interesting to learn if regulated translocation of proteins to flagella is a common cellular mechanism that may be important in cells bearing nonmotile as well as motile cilia/flagella. Much is known about the protein components of these organelles and how their structure and composition is related to motility. Motifs required for targeting proteins to the flagella also are beginning to be identified (73). In addition, it has been possible to investigate mechanisms of assembly of cilia and flagella by studying regeneration of the organelles after their experimental detachment. Recently, a new form of motility, intraflagellar transport, has been shown to be intimately involved in flagellar assembly (65, 74 -79). Submembranous intraflagellar transport particles traveling between the cell body and the flagella via a kinesin-related protein (anterograde movement-FLA10 (kinesin II)) and a component of cytoplasmic dynein (retrograde movement) (80) ferry flagellar components between these two compartments. There is compelling evidence that intraflagellar transport is responsible for carrying flagellar precursors into newly forming flagella and for returning flagellar components to the cytoplasm during flagellar resorption. It is likely that this system also is involved in the constitutive turnover of flagellar components in intact cilia/flagella (78,(81)(82)(83)(84), although other motility mechanisms also may play a role (6,13,85).
The translocation of CALK to flagella is distinct, however, from these previously described movements of molecules to flagella/cilia during assembly and resorption and also is distinct from constitutive turnover of flagellar components. Unlike molecules such as tubulin, for example, CALK normally is present in extremely small amounts in flagella; translocation occurs in the apparent absence of flagellar assembly or disassembly; CALK itself is a molecule with signaling potential; and, finally, CALK translocation is regulated by cAMP and only in gametes. It will be interesting, however, to determine if, despite these several differences, CALK translocation also requires intraflagellar transport.
The discovery of the regulated translocation of CALK also provides a potential mechanism for cyclic nucleotide-dependent regulation of sperm motility. The sperm of many species are relatively or completely immotile until they undergo activation via poorly characterized, cAMP-dependent pathways (86 -89). Whereas cyclic nucleotide-dependent protein kinases pre-existing in the sperm flagella might be obvious candidates for regulating initiation of motility (90), a recent study demonstrated that nonaxonemal components can be required for motility (91). Thus, sperm motility may depend on cAMP-dependent translocation of new regulatory molecules into the flagella. Future experiments should be instructive about the role of CALK in gamete activation and fertilization in Chlamydomonas and whether CALK homologues or related signaling molecules undergo regulated targeting to cilia or flagella in other cell types.