Chlamydomonas FAP133 is a dynein intermediate chain associated with the retrograde intraflagellar transport motor

In eukaryotic cilia and flagella, intraflagellar transport (IFT) is the bi-directional movement of large protein complexes along the length of axonemal outer doublet microtubules (Rosenbaum and Witman, 2002; Scholey, 2003). IFT was first observed in Chlamydomonas reinhardtii (Kozminski et al., 1993), and it was later shown that IFT is a highly conserved process, necessary for assembling and maintaining cilia in evolutionary distant organisms such as Caenorhabditis elegans and humans (Cole et al., 1998; Rosenbaum and Witman, 2002). Cilia and flagella are assembled and continuously turnover at their distal tip (Johnson and Rosenbaum, 1992; Marshall and Rosenbaum, 2001; Witman, 1975), and it is thought that IFT mediates the transport of flagellar components from the basal body region to their assembly site as well as the removal of turnover products from the flagellar tip (Qin et al., 2004). Additionally, specific receptors and other membrane components that preferentially localize to the cilium may be transported there by IFT (Christensen et al., 2007; Pan et al., 2005).

IFT particles have been isolated from Chlamydomonas flagella and found to consist of two biochemically distinct complexes, A and B, which collectively comprise at least 17 different IFT particle proteins (Cole et al., 1998; Piperno and Mead, 1997; Cole, 2003). Most complex A and complex B proteins are highly conserved among different organisms, and mutant analyses in Chlamydomonas and C. elegans have shown that many IFT proteins are essential for flagellar and ciliary assembly (Cole, 2003), although specific functions have been ascribed to only a few. For example, IFT172 interacts with the microtubule plus-end-binding protein EB1 and is believed to regulate the anterograde-retrograde IFT transition at the flagellar tip (Pedersen et al., 2005). Additionally, IFT20 may regulate the transport of membrane bound proteins from the Golgi complex to the cilium (Follit et al., 2006), and another complex B protein, IFT27, was recently shown to be involved in cell cycle control (Qin et al., 2007). Moreover, analysis of a partial suppressor mutation in IFT46 suggests that the IFT46 protein is required for transporting outer dynein arms (Hou et al., 2007).

IFT particles together with cargo proteins are ferried to the ciliary tip by a plus-end-directed heterotrimeric kinesin-2 that has been well characterized in several model systems (Cole et al., 1993; Cole et al., 1998; Pan et al., 2006; Yamazaki et al., 1995); in some organisms, such as C. elegans, an additional homodimeric kinesin-2 motor OSM-3 cooperates with heterotrimeric kinesin-2 during anterograde IFT (Snow et al., 2004). The molecular motor required for retrograde IFT is cytoplasmic dynein 1b (also termed cytoplasmic dynein 2) about which comparatively little is known (Pazour et al., 1999; Porter et al., 1999; Signor et al., 1999).

Dyneins drive microtubule minus-end-directed motion in cells and are multi-protein complexes that consist of one or more heavy chains (HCs) belonging to the AAA⁺ family of ATPases and a number of accessory proteins required for structural integrity, regulation of motor activity and cargo specificity (reviewed in King, 2002). The major isoform of cytoplasmic dynein (cytoplasmic dynein 1 or 1a) is involved in a wide range of intracellular activities including vesicular transport, chromosome segregation and organization of the cytoplasmic microtubule network (Vallee et al., 2004). This dynein is a HC homodimer and contains two intermediate chains (ICs), two light intermediate chains (LICs) as well as at least three different light chain (LC) dimers of the LC8 (DYNLL), Roadblock/LC7 (DYNLRB) and Tctex1 (DYNLT) families (Pfister et al., 2006). By contrast, to date only three proteins – an isoform-specific HC (DHC1b) and LIC (D1bLIC), and possibly LC8 – have been identified as components of the cytoplasmic dynein that drives retrograde IFT. In Chlamydomonas, null mutations of DHC1b result in impaired ability to grow flagella of normal length. Instead, these cells assemble stumpy flagella with large accumulations of IFT particles at their tip, suggesting a defect in retrograde IFT (Pazour et al., 1999). In C. elegans, this dynein is required for retrograde transport in chemosensory cilia (Signor et al., 1999), and the mammalian ortholog is enriched in ciliated epithelia and localizes at the connecting cilia in the retinal epithelium and in primary cilia of cultured cells (Mikami et al., 2002), indicating a conserved function in ciliary assembly.

A LIC (D1bLIC) was identified first in mammals and subsequently in Chlamydomonas and C. elegans, as an integral component of the retrograde IFT dynein complex (Grissom et al., 2002; Hou et al., 2004; Perrone et al., 2003; Schafer et al., 2003). In Chlamydomonas a null mutation in D1bLIC also results in stumpy flagella with large accumulations of IFT particles (Hou et al., 2004). Moreover, LC8 – which is a highly conserved component of both cytoplasmic and axonemal dyneins as well as other enzyme complexes – is also thought to play a role in retrograde IFT, because the Chlamydomonas LC8-null mutant strain (fla14) exhibits defects in this process (Pazour et al., 1998). However, to date there is no direct evidence that LC8 is part of the same complex as DHC1b and D1bLIC.

Although cytoplasmic and axonemal dyneins containing more than one HC associate with two ICs that belong to the WD-repeat protein family, no such IC has yet been identified for the retrograde IFT motor. In an effort to fully understand the dynein motor complex that powers retrograde IFT, we have characterized the product of the Chlamydomonas gene FAP133, which encodes a protein similar to other dynein ICs and has conserved homologs in vertebrates and many other ciliated organisms. We show, using genetic, biochemical and immunofluorescence approaches, that FAP133 is part of the IFT machinery in Chlamydomonas. Furthermore, we provide biochemical evidence that FAP133 associates specifically with DHC1b-D1bLIC and that LC8 is also part of this complex. Together, these results suggest that FAP133 is a component of the retrograde dynein motor and plays a conserved role in retrograde IFT.

As many dyneins associate with WD-repeat intermediate chains (ICs), to identify a putative IC associated with the IFT dynein motor complex we searched the Chlamydomonas genomic and EST databases using the protein sequence of the rat cytoplasmic dynein IC (DYNC1I2; accession Q62871) as a query. We identified a previously uncharacterized gene consisting of 12 exons that has subsequently been annotated as FAP133 (JGI version 2.0, scaffold 2:870454-873524) and obtained a full-length cDNA clone (BP095745) from the Kazusa DNA Research Institute, Chiba, Japan. Southern blot analysis of Chlamydomonas genomic DNA using the FAP133 cDNA as a probe, confirmed that there is a single gene for FAP133 (Fig. 1A). Moreover, expression of FAP133 results in a ∼2.5 kb message that is upregulated upon deflagellation (Fig. 1B), which is a common property of mRNAs that encode flagellar proteins (Stolc et al., 2005).

The FAP133 cDNA is predicted to encode a 558-residue protein with a mass of 61,160 Da and pI of 5.34. Database searches and sequence alignment revealed that FAP133 is most closely related to the uncharacterized vertebrate protein WD34 [P_(n)=2×e^–68 vs the zebrafish (Danio) homolog; accession BC133909; Fig. 1C]. Furthermore, domain analysis identified six WD-repeat domains within the C-terminal region of FAP133 (Fig. 1D), similar to other dynein ICs (Ogawa et al., 1995; Wilkerson et al., 1995); the region between the third and fourth domains (residues 295-392) may contain a seventh degenerate WD-repeat. FAP133 also contains two putative degenerate LC8 binding sites (Lo et al., 2001) near the N-terminus, VETQT (residues 46-50) and QGTQT (residues 56-60) (Fig. 1D).

In the recently published Chlamydomonas flagellar proteome (Pazour et al., 2005), FAP133 peptides were primarily found in the detergent-soluble flagellar membrane and matrix fraction, similar to DHC1b and other IFT proteins. To further characterize the molecular properties of FAP133, we raised a specific antiserum (CT248) against a recombinant FAP133 peptide (residues 89-558), expressed as a C-terminal fusion with maltose-binding protein. Immunoblot analysis of isolated wild-type Chlamydomonas flagella revealed that the affinity-purified CT248 antibody specifically recognized a single band migrating at M_r ∼66,000 (Fig. 2A). To identify the specific flagellar compartment(s) in which FAP133 is located, we extracted flagellar matrix proteins by subjecting isolated flagella to repeated freeze-thaw cycles in order to mechanically disrupt the membrane, or treated flagella with a non-ionic detergent to solubilize both membrane and matrix proteins. In both cases the majority of FAP133 was found in the soluble fraction with lesser amounts remaining associated with the axoneme (Fig. 2B,C). We next tested whether ATP would induce the release of the FAP133 fraction that remained associated with the axoneme after detergent extraction, as is the case with the heterotrimeric kinesin-2 and DHC1b polypeptides (Henson et al., 1997; Pazour et al., 1999). ATP treatment, indeed, caused a portion of FAP133 to detach from the axonemes; a similar amount of D1bLIC was also released. However, subsequent ATP additions did not solubilize more protein. The remaining axoneme-bound FAP133 was extracted with 0.6 M NaCl as was an additional fraction of D1bLIC and most of the outer arm dynein component LC2 (Fig. 2C).

To further define the flagellar location of FAP133 in wild-type Chlamydomonas cells, we performed immunofluorescence microscopy with the CT248 antibody. The analysis revealed that the great majority of FAP133 localized to a bi-lobed area near the basal bodies (Fig. 2D). In addition, FAP133 was found in numerous puncta that were randomly distributed along the entire length of both flagella. This localization is remarkably similar to that observed previously for IFT particle subunits and motors (e.g. Hou et al., 2004; Pazour et al., 1999).

To further investigate the possible association of FAP133 with the IFT machinery, we employed a Chlamydomonas strain with a temperature-sensitive mutation in the FLA10 gene, which encodes the 90 kDa motor unit of the heterotrimeric kinesin-2 that powers anterograde IFT (Huang et al., 1977; Kozminski et al., 1995; Walther et al., 1994). If grown at 22°C, fla10 cells can form flagella of normal length that are virtually indistinguishable from wild-type flagella. When the temperature is raised to 32°C, fla10 flagella become depleted of IFT particle proteins within the first 1-2 hours as a result of stalled anterograde IFT but still active retrograde IFT (Cole et al., 1998); DHC1b and D1bLIC also become depleted from fla10 flagella under these conditions (Iomini et al., 2001; Pedersen et al., 2006). We found that the protein levels of FAP133 in fla10 flagella were dramatically reduced after only 1 hour at the restrictive temperature, in contrast to other axonemal proteins, such as IC1, which was not affected by the temperature shift (Fig. 3A). This finding was confirmed by immunofluorescence microscopy, which also revealed that almost all FAP133 protein in fla10 cells incubated at 32°C was located at the peri-basal body region; by contrast, FAP133 was readily detected in the flagella of wild-type cells at this temperature (Fig. 3B). These results indicate that FAP133 depends on FLA10 kinesin-2 for flagellar localization as do IFT particle proteins and IFT dynein motor subunits suggesting that FAP133 associates specifically with IFT. To further test this, we compared FAP133 incorporation into wild-type flagella and mutants lacking outer arms alone (oda6), outer arms and the docking complex (oda3), outer arms and the Oda5p/adenylate kinase complex (oda5), inner arm I1/f (ida1), inner arms a, c and d (ida4), radial spokes (pf14) and the central pair microtubule complex (pf18). Immunoblot analysis revealed that FAP133 was present at essentially wild-type levels in all mutant strains (Fig. 4).

The flagellar localization pattern and fla10 analysis described above strongly suggest that FAP133 is associated with the IFT machinery. To elucidate the specific IFT complex proteins with which FAP133 associates, we examined the hydrodynamic properties of FAP133 and other IFT components from the flagella matrix. Specifically, freshly prepared freeze-thaw flagella extracts were fractionated by sucrose-density-gradient centrifugation using a buffer of relatively low ionic strength (25 mM KCl). Under these conditions, and in agreement with previous reports (Cole et al., 1998), based on immunoblots and analysis of Coomassie-Blue-stained gels, we found IFT complex A and complex B polypeptides sedimenting at ∼16 S and the heterotrimeric kinesin-2 motor (FLA10) at ∼10 S (Fig. 5A). We also found that the DHC1b-D1bLIC complex sedimented in two distinct peaks; the majority of this dynein complex was at ∼18 S with a smaller amount at ∼12 S. Interestingly, FAP133 was also detected in two distinct peaks: a minor peak occurring at ∼18 S that followed the sedimentation profile of DHC1b and D1bLIC in that region, as well as a major FAP133 peak at ∼10 S. LC8 co-purified with both pools of FAP133, suggesting that these two proteins interact (and see below). These results suggest either that there are two distinct pools of FAP133 in the flagellum, or that FAP133 binds weakly to the DHC1b-D1bLIC complex and is readily dissociated from this motor complex under our experimental conditions. To distinguish between these two possibilities, we pooled the fractions corresponding to the ∼18 S peak and fractionated them again in a second gradient. Again, DHC1b was present at ∼18 S but FAP133 was not and instead was found only at ∼10 S (Fig. 5B), suggesting that the FAP133 pool at ∼10 S results from dissociation of the ∼18 S complex. Interestingly, we also found that conditions which altered the sedimentation behavior of DHC1b also changed that of FAP133. Specifically, the use of 1% Tergitol to extract flagellar membrane and matrix proteins caused DHC1b to sediment as a single peak at ∼12 S. Under these conditions, all FAP133 was present in a single peak at ∼10 S (Fig. 5C).

As FAP133 readily dissociated from the ∼18 S IFT dynein complex in sucrose gradients (Fig. 5A,B), possibly as a result of high hydrostatic pressure, we sought a more gentle fractionation method. For this, we extracted flagella matrix components by freeze-thaw in the absence of a detergent using a relatively low-ionic-strength extraction buffer. To separate the different complexes, we employed a Superose-6 gel-filtration column, which has an exclusion limit of ∼40 MDa, equilibrated with the same low-ionic-strength buffer that was used previously for the extraction of flagella matrix proteins. Essentially all DHC1b and D1bLIC were eluted as a single peak very early from the column (fractions 2 and 3; Fig. 6A). At least 50% of total FAP133 and the FLA10 kinesin-2 subunit were also detected in these same fractions. Furthermore, the elution profile of LC8 showed three distinct peaks, of which the first two coincided with those of FAP133. Similarly, we detected two distinct peaks of the complex A protein IFT139; ∼10% co-eluted with DHC1b, FAP133 and FLA10. By contrast, we obtained only a single peak of the complex B protein IFT81 that did not co-fractionate with IFT139, DHC1b or FAP133. These results confirm that FAP133 is associated with the retrograde IFT dynein complex (DHC1b, D1bLIC, LC8), and also suggest that this complex is associated with IFT complex A and kinesin-2. A similar elution profile was obtained for a flagellar extract from the mutant pf28pf30ssh1 that lacks both outer arms and inner arm I1/f (not shown).

To further test whether these proteins are part of the same complex we immunoprecipitated FAP133 and IFT139 from freshly prepared flagella-matrix extracts, using CT248 and a monoclonal IFT139-specific antibody (Cole et al., 1998), respectively. Both pellets contained FAP133, DHC1b-D1bLIC, IFT139 and FLA10 polypeptides: control proteins (IC1 and EB1) were not detected in the precipitates; D1bLIC co-migrates with the FAP133 polyclonal antibody heavy chain band and, thus, any potential signal in that precipitate was obscured (Fig. 6B,C). Moreover, in the FAP133 pellet we detected a considerable amount of LC8, suggesting a strong interaction between these two proteins (see below) and also some IFT81 (Fig. 6B). When we immunoprecipitated IFT172 from complex B, we found IFT81 and a very small amount of IFT139 but no dynein or kinesin components in the pellet (Fig. 6D). These data are somewhat different to those reported by (Pedersen et al., 2006) and probably stem from differences in extract preparation or other experimental conditions.

Previous work has shown that the level of D1bLIC is dramatically reduced in dhc1b-null mutant cells and that DHC1b levels are reduced in the d1blic-null strain (Hou et al., 2004). Moreover, DHC1b is required for localization of D1bLIC at the basal bodies, but D1bLIC is not required for basal-body localization of DHC1b (Hou et al., 2004; Perrone et al., 2003). To determine whether the protein levels and subcellular localization of FAP133 are affected by the absence of specific subunits of the retrograde IFT dynein motor, we employed three Chlamydomonas null mutant strains; dhc1b, d1blic and fla14, which lack the DHC1b, D1bLIC and LC8 genes, respectively (Hou et al., 2004; Pazour et al., 1999; Pazour et al., 1998). All three of these mutants are defective in retrograde IFT and have small and stumpy flagella that accumulate IFT particle components. We prepared whole-cell extracts from these mutant strains and analyzed them by immunoblotting. FAP133 levels were similar in wild-type and in the dhc1b and d1blic mutant cells, but showed a modest decrease in fla14 (Fig. 7A). By contrast, DHC1b was essentially undetectable in both d1blic and fla14 strains, whereas levels of D1bLIC were reduced in both dhc1b and fla14. Finally, the amount of LC8 present in wild-type as well as dhc1b and d1blic mutant cells did not vary significantly (Fig. 7A), consistent with the participation of this protein in many functionally distinct complexes.

We next used immunofluorescence microscopy to examine the subcellular localization of FAP133 in these mutant strains. We detected a striking difference in the distribution of FAP133 between wild-type and the dhc1b or d1blic mutant cells. Specifically, in the absence of DHC1b or D1bLIC, FAP133 was almost completely redistributed from the peri-basal body region, was present throughout the cytoplasm and most prominently occurred in a zone near the cell center where the Golgi usually resides (Fig. 7B). By contrast, the absence of LC8 did not alter the normal localization of FAP133 at the basal bodies, but we could not detect this protein in the fla14 flagella stubs. The redistribution of FAP133 in the dhc1b and d1blic mutants was specific because in all strains used here, the FLA10 kinesin-2 subunit, and complex A and complex B proteins were located either at the basal bodies, or accumulated in the flagella stubs (data not shown). In summary, these results suggest that DHC1b and D1bLIC are required for targeting and/or tethering of FAP133 near the basal bodies, potentially as a result of incorporation into the dynein holoenzyme, but they do not affect the stability of this protein in cytoplasm. By contrast, LC8 is not required for localizing FAP133 to the basal bodies but might affect the stability of FAP133 and its transport into the flagella.

LC8 has been assumed to be part of the retrograde IFT motor complex because the absence of LC8 in Chlamydomonas results in defects in this process (Pazour et al., 1998); however, it was not observed to copurify with the DHC1b-D1bLIC complex (Perrone et al., 2003). We found that LC8 co-purified with FAP133 in sucrose gradients (Fig. 5A) and following gel filtration (Fig. 6A). In addition, LC8 was particularly enriched in the pellet of flagella-matrix proteins that were immunoprecipitated with an antibody against FAP133 (Fig. 6B). To further examine the possible interaction of LC8 with FAP133, we fractionated flagella matrix extracts using anion-exchange chromatography. Immunoblot analysis of the eluted fractions revealed two distinct peaks of FAP133 (Fig. 8). We separately pooled the fractions that contained these two peaks and layered them over two sucrose density gradients to further purify the complexes that contained FAP133. We found that FAP133 and LC8 co-sedimented at ∼10 S in both gradients (Fig. 8), strongly suggesting that these proteins are part of the same complexes.

In this study, we have characterized the Chlamydomonas FAP133 gene and have shown that it encodes a novel IC that associates with the retrograde IFT dynein motor. We have also provided evidence that FAP133 interacts strongly with LC8, suggesting that this LC is also an integral component of this motor complex. Furthermore, we have found that conditions that help stabilize the interaction between FAP133/LC8 and DHC1b-D1bLIC also permit other IFT proteins, such as the FLA10 kinesin-2 subunit and IFT139, to associate with the same complex, thus providing insight into the mechanism(s) by which IFT motors assemble with the IFT particle scaffold.

FAP133 is encoded by a single gene in Chlamydomonas that is upregulated upon deflagellation, similar to other genes that code for flagellar proteins (Stolc et al., 2005). Our findings that FAP133 is located primarily at the peri-basal body region of the cell and in punctate structures along the flagella, and that it is readily extracted from the flagella when the membrane is disrupted, suggested a role for FAP133 in IFT. This hypothesis was further supported by our observation that FAP133 is specifically depleted, together with other IFT components, from the flagella of fla10 cells (temperature-sensitive mutant for the anterograde IFT FLA10 kinesin-2 subunit), shortly after being transferred to the restrictive temperature. In addition, we found that FAP133 protein levels were similar to wild type in the flagella of various Chlamydomonas mutant strains that lack specific axonemal components, including the outer arm and inner arm dynein complexes, the radial spokes or the central pair microtubules. Taken together, these observations indicate that FAP133 is a novel component of the IFT system.

In Chlamydomonas a null mutation in the DHC1b gene results in very short, stumpy flagella that accumulate large amounts of IFT particles at their tips, suggesting that retrograde IFT requires this cytoplasmic dynein HC isoform (Pazour et al., 1999). In later studies, genetic and biochemical evidence indicated that D1bLIC is also an integral component of this retrograde IFT dynein (Hou et al., 2004; Perrone et al., 2003). However, in contrast to other dynein motor complexes that contain multiple HCs, no IC had yet been identified associated with this motor. Structural analysis of the 558-residue FAP133 polypeptide revealed a domain arrangement similar to that of other dynein ICs (Ogawa et al., 1995; Wilkerson et al., 1995), with six WD-repeats (and potentially a seventh at residues 295-391) in the C-terminal region. Furthermore, sequence alignment revealed that FAP133 is closely related to vertebrate WD34 (an uncharacterized protein and a possible ortholog of FAP133) and also to the IC of rat cytoplasmic dynein. On the basis of these findings, the proteomic analysis of Chlamydomonas flagella (Pazour et al., 2005), and the observation that FAP133 is involved in IFT, we predicted that FAP133 may associate with the retrograde IFT dynein motor. Our finding that FAP133 does not localize at the peri-basal body region in the absence of DHC1b or D1bLIC provided strong genetic evidence to support this hypothesis.

IFT proteins present in flagella matrix extracts can be separated into at least four biochemically distinct complexes on sucrose-density gradients, including IFT complex A and complex B, the heterotrimeric kinesin-2 motor and the DHC1b-D1bLIC dynein complex (Cole et al., 1998; Perrone et al., 2003). Based on the requirement for IFT dynein to localize FAP133 at the basal body region, the behavior of ICs in other dynein complexes and consistent with our initial hypothesis, we expected FAP133 to co-migrate with DHC1b in a sucrose-density gradient. Instead, we found almost 90% of FAP133 migrating at ∼10 S and only a small peak of this protein was detected in fractions containing the ∼18 S form of DHC1b. Additional sedimentation experiments indicated that certain conditions, such as increased ionic-strength, non-ionic detergents and high hydrostatic pressure, induce the depletion of FAP133 from the ∼18 S fractions owing to dissociation from the DHC1b-D1bLIC complex; some of these treatments also cause the further breakdown of the ∼18 S IFT dynein motor into smaller subunits. Based on sedimentation analysis of both axonemal and cytoplasmic dyneins, the ∼18 S form of the IFT dynein probably contains two HCs, whereas the ∼12 S particle represents a HC monomer (for a review, see King, 2002). However, using a gel filtration column, we were able to purify DHC1b-D1bLIC-containing complexes that had a mass greater than that of the outer arm dynein. In these same fractions we also detected about 50% of the total FAP133 that was present in the flagella matrix extract. In addition, we also found that DHC1b co-immunoprecipitated with FAP133 from flagella matrix extracts. These results indicate that, whereas FAP133 is an integral part of the IFT dynein complex, the interaction between FAP133 and DHC1b-D1bLIC is relatively weak compared with other intra-dynein interactions, raising the possibility that it is subject to regulation in vivo.

Chlamydomonas fla14 cells, which lack LC8, have very short flagella that accumulate IFT particles, indicating a defect in retrograde IFT (Pazour et al., 1998). Since LC8 is found in both cytoplasmic as well as outer arm and the I1/f inner arm axonemal dyneins, it may also associate directly with the retrograde IFT dynein motor. However, in previous studies LC8 did not co-purify with DHC1b or D1bLIC (Perrone et al., 2003) and it has been unclear why lack of LC8 results in defective retrograde IFT. Using various fractionation methods to separate flagella matrix components, we found that LC8 consistently co-purified with FAP133. Furthermore, immunoprecipitation using an antiserum against FAP133 pulled down a significant proportion of the LC8 that was present in the flagella matrix. These data indicate that LC8 interacts strongly with FAP133 and, therefore, may also be an integral component of the retrograde IFT dynein complex. Interestingly, analysis of the FAP133 sequence identified two potential degenerate LC8 binding sites in the region N-terminal of the WD repeats (see Fig. 1D), suggesting that these proteins interact directly as occurs in the major cytoplasmic dynein isoform. We propose that FAP133-LC8 represents an IC-LC complex that is readily dissociated from DHC1b-D1bLIC. It remains unclear whether this dynein sub-complex contains additional LC components as do the analogous IC-LC structures in other dyneins (see King, 2002).

It has previously been shown that DHC1b and D1bLIC are normally located around the basal bodies of fla14 cells but do not enter the flagella stubs, in contrast to the rest of the IFT proteins, which appear to accumulate at the tips (Pazour et al., 1998; Perrone et al., 2003). Similarly, we found that the majority of FAP133 was located primarily at the peri-basal body region in fla14 cells but we did not detect any protein inside the flagella stubs. These observations suggest that DHC1b, D1bLIC and FAP133 associate within the cell body and are targeted to the basal bodies, even in the absence of LC8. However, LC8 appears to play a crucial role in the transition of the retrograde IFT dynein motor from the basal body region into the flagellum.

Dynein ICs are known to play an essential role in maintaining the stability of the motor complex and in regulating the attachment of cargo. For example, the ICs that associate with cytoplasmic dynein serve as scaffolds that mediate interactions between the HCs, the three LC dimers as well as the p150 subunit of dynactin and other protein cargoes (Pfister et al., 2006; Vaughan and Vallee, 1995). Similarly, in Chlamydomonas axonemal outer arm dynein, both ICs (IC1 and IC2) associate with the HCs and LCs at the base of the motor complex and are required for the assembly and docking of the motor onto the A tubules of the outer doublets (King et al., 1991; Mitchell and Kang, 1991; Wilkerson et al., 1995). We found that the retrograde IFT dynein dissociates into two distinct sub-complexes; one that contains the HC and LIC and a second that contains FAP133 and LC8. Sedimentation of the FAP133-LC8 subunit at ∼10 S in sucrose-density gradients indicates a particle of size similar to other IC-LC complexes. Thus FAP133 and LC8 both probably exist as dimers and potentially may associate with other components, such as Roadblock-LC7 and/or Tctex1 family LCs.

On the basis of our results and previously published data, we propose a model for the structural architecture of the retrograde IFT dynein motor (Fig. 9). We suggest that a FAP133-subunit dimer binds directly to the N-terminal region of DHC1b. LC8 might not be required for the association of FAP133 and DHC1b but is essential for the loading of the retrograde IFT dynein complex onto the anterograde transport system or onto other IFT cargoes, either by stabilizing the FAP133-LC8 sub-complex or by interacting directly with the IFT scaffold and/or IFT cargo components. In agreement with this model, we were able to purify by gel filtration particles that were larger than the ∼2 MDa outer-arm dynein complex, and contained almost all flagellar DHC1b and D1bLIC, a portion of the soluble flagellar pool of IFT complex A IFT139 as well as at least 50% of the FAP133 and FLA10 kinesin-2. Although IFT81 from complex B was immunoprecipitated by the FAP133 antibody, only a very small amount of this protein was present in the same gel-filtration fractions as dynein, FLA10 kinesin-2 and IFT139. The finding that a fraction of FAP133 always purified separately from DHC1b could indicate that the FAP133-LC8 sub-complex serves as an adaptor that readily dissociates from the DHC1b-D1bLIC complex when the motor needs to exchange cargoes. Alternatively, some fraction of FAP133 may be involved in a separate and currently unidentified process during IFT. It will be important to follow the movement of tagged FAP133 in vivo to determine whether or not FAP133 is indeed transported bi-directionally as predicted.

The model proposed here for the role of FAP133 in the retrograde IFT dynein motor predicts that retrograde IFT would be greatly compromised in the absence of FAP133. Indeed, in a recent study of conserved genes that are involved in flagellar activity in trypanosomes, it has been found that reduction of FAP133 protein levels by RNA interference resulted in cell clustering owing to flagellar dysfunction (Baron et al., 2007). The presence of FAP133 homologues in the genomes of vertebrates, sea urchins, insects and protists suggests a conserved function for this protein amongst ciliated organisms. However, the observation that a FAP133 homolog cannot be identified in C. elegans, which contains non-motile sensory cilia, raises the possibility that FAP133 functions only in motile cilia.

Wild-type Chlamydomonas reinhardtii (CC124), and the fla10 (CC1919), ida1 (CC2664), ida4 (CC2670) oda3 (CC2232), oda5 (CC2236), oda6 (CC2239), pf14 (CC1032) and pf18 (CC1036) mutant strains were obtained from the Chlamydomonas Center (Duke University, Durham NC). The d1blic (YH43), dhc1b (3088.4) and fla14 (V64) mutant strains were provided by Gregory Pazour (University of Massachusetts Medical School, Worcester, MA). The pf28pf30ssh1 strain (also referred to as WS4 isolate) was provided by Winfield Sale (Emory University School of Medicine, Atlanta, GA). Cells were grown in Tris-acetate-phosphate (TAP) (Gorman and Levine, 1965) or R medium at 22°C, aerated with 5% CO₂ and 95% air on a light-dark cycle (15 hours light 9 hours darkness).

Searches of the Chlamydomonas genomic and expressed sequence tag databases were performed using BLAST at http://genome.jgi-psf.org/Chlre3/Chlre3.home.html and http://est.kazusa.or.jp/en/plant/chlamy/EST/index.html, respectively. FAP133 homologues were identified using a TBLASTN search of the non-redundant database at NCBI (http://www.ncbi.nlm.nih.gov/BLAST/). Sequence alignment using CLUSTALW and generation of the neighbor-joining phylogenetic tree were performed at http://align.genome.jp. Domain analysis of FAP133 was achieved using SMART (http://dylan.embl-heidelberg.de/).

A cDNA (BP095745) encoding full-length FAP133 was obtained from the Kazusa DNA Research Institute. Northern and Southern blot analyses of Chlamydomonas RNA and genomic DNA samples, respectively, were performed using our standard methods (King and Patel-King, 1995).

Chlamydomonas flagella were isolated using the dibucaine method as described previously (King, 1995). Cells were grown in 2-8 l cultures and harvested by centrifugation. Purified flagella were resuspended in HME buffer (30 mM HEPES pH 7.4, 5 mM MgSO₄, 0.5 mM EGTA, 1 mM DTT) containing 25 mM potassium chloride (HMEK) or potassium acetate (HMEA) and supplemented with a protease inhibitor cocktail (Sigma, Catalog No. P9599). To study the effects of impaired anterograde IFT, wild-type and fla10 cells were grown in TAP at 22°C in single 2-liter cultures. When the cultures reached log-phase, they were divided equally into two, one of which was transferred to a 32°C water-bath while the other was kept at 22°C; all cultures were placed under the same light conditions. After 1 hour, a small aliquot was removed from all cultures for immunofluorescence, whereas the rest of the cultures were used for isolating flagella as described above.

To obtain flagellar matrix proteins, purified flagella were resuspended in HMEK buffer, quick-frozen on dry ice and subsequently left to thaw at room temperature to facilitate disruption of the membrane. After three consecutive freeze-thaw cycles, flagella were pelleted by centrifugation at 10,000 rpm in a Hermle Z233 M-2 microfuge for 10 minutes and the supernatant removed and kept on ice until use. Alternatively, isolated flagella were demembranated with 1% Tergitol NP-40 (Sigma) or 0.1% Igepal CA-630 in HMEK buffer. Detergent-extracted axonemes were resuspended in HMEK buffer containing 10 mM ATP. After removing ATP-extracted components by centrifugation or following detergent extraction of whole flagella, axonemes were further treated with HME buffer containing 0.6 M NaCl to disrupt ionic interactions. Equivalent amounts of all fractions were loaded on 5-15% gradient SDS polyacrylamide gels and either stained with Coomassie Blue or blotted to nitrocellulose membrane for immunodetection.

For the production of antibody against FAP133, a 1.4 kb fragment of the Chlamydomonas FAP133 cDNA (encoding residues 89-558) was amplified. The PCR product was inserted into the pMAL-c2 vector (New England Biolabs) across the XmnI-XbaI restriction sites. This vector was used to transform Escherichia coli BL21 cells for the expression of a recombinant FAP133 polypeptide fused at the N-terminus with maltose-binding protein (MBP). The purified fusion protein was dialyzed against PBS pH 7.2 and was used as the immunogen to raise rabbit polyclonal antibody (CT248) (Covance, Denver, PA). The resulting antibody was affinity-purified against recombinant FAP133 after removing the MBP tag, essentially as described previously (Wakabayashi et al., 2007).

Other primary antibodies used in this study include: R4058 (vs LC8) (King and Patel-King, 1995), R5391 (vs LC2) (Patel-King et al., 1997), 1878A (vs IC1) (King et al., 1986) and anti-EB1 (Pedersen et al., 2003). Rabbit polyclonal antibodies against DHC1b and D1bLIC were generously provided by George Witman (University of Massachusetts Medical School, Worcester, MA). Mouse monoclonal antibodies against IFT81, IFT139 and IFT172 and a rabbit polyclonal antibody against FLA10 were generously provided by Douglas Cole (University of Idaho, Moscow, ID).

For sedimentation analysis, 200 μl of freshly prepared flagellar matrix extracts were layered onto 5 ml 5-20% sucrose-density gradients and centrifuged in a SW55Ti rotor for 10 hours at 30,000 rpm. Gel filtration was performed using a Superose-6 HR10/30 column (Amersham Biosciences), equilibrated with HMEK buffer supplemented with a protein inhibitor cocktail. Flagella from 8-liter cultures were extracted in the same buffer by freeze-thaw in a total volume of 5 ml. The flagellar matrix extract was then concentrated to 250 μl using an Amicon Ultra-4 filter unit and injected onto the column. The column was eluted with a 0.4 ml/minute flow rate and a total of 30 (0.75-ml) fractions were collected. Anion-exchange chromatography was performed on a Mono-Q HR5/5 column (Amersham Biosciences). Briefly, freshly prepared flagellar matrix extract was loaded onto the column that was previously equilibrated with HMEK buffer. The column was then washed with the same buffer for 10 minutes using a flow rate of 1 ml/minute. Elution was performed with a 0.5 ml/minute flow rate using a linear gradient of 0.025-0.6 M KCl with a total volume of 30 ml. Sixty fractions of 0.5 ml each were collected and analyzed. For subsequent sedimentation analysis, FAP133-containing fractions were pooled and concentrated using an Amicon Ultra-4 filter unit before layering onto a 5-20% sucrose-density gradient. Gel filtration and anion-exchange chromatography were performed using a Biologic II FPLC system (Bio-Rad).

For immunoprecipitation of FAP133, protein G-agarose beads (Pierce) were washed three times with TBS pH 7.2 supplemented with 3% BSA and then incubated with the anti-FAP133 antibody in the same buffer for 1 hour at 22°C. Beads were then washed three times with HMEK buffer and incubated with freshly prepared flagella matrix extract, for 1 hour at 22°C. The immunoprecipitated pellet was washed three times with HMEK buffer and finally resuspended in SDS gel sample buffer for further analysis. As a control, protein-G–agarose beads were treated only with 3% BSA before being added to the flagella matrix extract. Immunoprecipitation of IFT172 from flagellar extracts was performed similarly; immunoprecipitation with the IFT139 monoclonal antibody was as described previously (Pedersen et al., 2005).

Isolated Chlamydomonas flagella, flagellar fractions, and various protein samples were routinely solubilized in SDS gel sample buffer, separated by 5-15% gradient PAGE and stained with Coomassie Blue or blotted to nitrocellulose membranes for immunodetection using standard protocols.

Chlamydomonas cells for immunofluorescence analysis were grown in TAP medium, harvested in mid-log phase by low-speed centrifugation at 90 g for 5 minutes and re-suspended in 10 mM Hepes pH 7.4. Microscope slides were treated before use with 1% poly-L-lysine (Sigma) in distilled water and air dried for 1 hour at 60°C. Cells in suspension were placed on treated slides and left to adhere for 5 minutes before fixing with methanol at –20°C for 10 minutes and air dried at room temperature for an additional 10 minutes. Subsequently, slides were placed in humidity chambers and re-hydrated with 1% Igepal CA-630 (Sigma) in PBS pH 7.2 for 10 min. After briefly rinsing with PBS, cells were treated for 1 hour at room temperature with blocking buffer containing 3% normal goat serum, 1% BSA, 1% cold-water-fish gelatin, 0.1% Igepal CA-630 and 0.05% Tween-20 in PBS. All antibodies used for immunofluorescence microscopy were diluted in PBS buffer containing 1% BSA, 0.1% cold-water-fish gelatin and 0.05% Tween-20. Cells were incubated with primary antibodies for 1-2 hours at 22°C or alternatively for 16 hours at 4°C, washed four times for 5 minutes with PBS and incubated with either Alexa Fluor 488-conjugated anti-mouse or Alexa Fluor 568-conjugated anti-rabbit secondary antibodies (Invitrogen) for 1 hour. Slides were washed four times for 5 minutes with PBS, dehydrated in a series of 30, 70 and 100% ethanol baths, air dried and mounted with coverslips using a glycerol-based mounting medium containing DABCO (Sigma) as an anti-fade agent. Stained cells were viewed on an Olympus BX51 epifluorescence microscope (Olympus America Inc.), equipped with PlanApo 60×/1,4 and 100×/1,35 oil immersion lenses, and images were acquired with a Magnafire cool-CCD digital camera (Optronics).

We thank Oksana Gorbatyuk (University of Connecticut Health Center, Farmington, CT) for assistance with preliminary biochemical analysis of FAP133, Gregory Pazour and Winfield Sale for Chlamydomonas mutant strains, and Douglas Cole and George Witman for generously providing antibodies and for helpful discussions. This work was supported by grant GM51293 from the National Institutes of Health and an investigator award from the Patrick and Catherine Donaghue Medical Research Foundation to S.M.K., and by grants from the Danish Natural Science Research Council (no. 272-05-0411) and the Novo Nordisk Foundation to L.B.P.