Eukaryotic flagella and cilia are complex structures consisting of more than six hundred different proteins. A current challenge to investigators of these organelles is to determine where each protein is located within the flagellum, and to identify its function there. Because only about 100 of these proteins have been characterized to this extent, there is a need for methods to rapidly and efficiently determine the flagellar locations of the uncharacterized proteins and obtain enough preliminary information to prioritize them for further study. The C. reinhardtii
flagellar proteome provided some guidance, as many proteins could be assigned to the membrane, matrix, or axonemal compartments. Approaches for learning more about the locations and functions of proteins on a large scale include transposon-based random epitope-tagging, and expression of proteins fused to a tag that can then be used to localize the proteins by fluorescence microscopy [for review see O'Rourke et al., 2005
]. For example, Andersen et al. 
and Kilburn et al. 
tagged candidate centrosomal and basal body components with GFP and determined their subcellular locations by immunofluorescence microscopy. Similarly, Van Damme et al. 
tagged selected Arabidopsis thaliana
proteins potentially involved in cytokinesis with GFP, and then observed their subcellular locations.
We have taken a similar approach with putative C. reinhardtii flagellar proteins, but used an HA tag rather than a GFP tag. The HA tag was chosen for several reasons. First, in our hands, an HA-tagged protein is much more likely to be expressed in C. reinhardtii than is a GFP-tagged protein, possibly because of the smaller size of the HA tag. Second, the smaller tag may be advantageous when the epitope-tagged protein is expressed in a wild-type background and has to compete with the endogenous protein for binding at the correct site, as is the case when a mutant for the target gene has not yet been identified. Third, although functional expression of proteins tagged with GFP has been accomplished for several soluble flagellar proteins, this has not been reported yet for axonemal components, where steric constraints may be high and interfere with the assembly of a protein fused to a large tag. Finally, a monoclonal antibody to HA is available that has a very high specificity for the HA tag when used in C. reinhardtii.
The HA-tagged versions of all ten proteins described here were expressed following transformation into wild-type C. reinhardtii
, and all were localized to the flagella by immunofluorescence microscopy. The epitope-tagged proteins generally showed a spotted distribution along the flagella. This pattern of distribution was not observed for tubulin (see ) and is unlikely to represent the native distribution for other axonemal proteins because the axoneme has uniform ultrastructure with a 96-nm periodicity over most of its length. Inasmuch as all of the tagged proteins initially were expressed in the presence of the endogenous wild-type protein, it is likely that the punctate distribution of the axonemal HA-tagged proteins is due to a mixture of tagged and wild-type proteins along the axoneme. Because there are only a limited number of binding sites for the proteins along the axoneme, competition between the endogenous and tagged proteins for these binding sites results in a speckled distribution of the tagged protein. In support of this hypothesis, strong and continuous labeling along the axoneme was observed when FAP134-HA was expressed in a mutant that does not express wild-type FAP134 . Similarly, the incorporation of PACRG-HA into the axoneme increased and became less punctate when the expression of both the endogenous and the HA-tagged PACRG were decreased by introduction of an RNAi inverted-repeat construct (; see Supplementary Fig. 1
). In contrast to the result in , where the PACRG-HA was localized most strongly to the distal part of the flagellum, the tagged protein was incorporated preferentially into the proximal and/or middle regions of the flagellum in the RNAi cells. The likely explanation for this is that the endogenous protein, when present in normal amounts, outcompetes the tagged protein for binding sites in the proximal part of the flagellum.
PACRG-HA Is Assembled on a Subset of Outer Doublets
The distribution of the PACRG-HA protein was further examined in disintegrated isolated axonemes. As expected from the work of Ikeda et al. 
, the protein was present exclusively on the outer doublet microtubules and not on the central pair, which was visualized following trypsin plus ATP-induced extrusion from the axoneme. Surprisingly, in splayed axonemes, only a subset of doublets was labeled, suggesting that PACRG-HA is asymmetrically distributed around the axoneme. This distribution may be related to previously described differences between the outer doublets. Hoops and Witman 
reported structural asymmetries in the axoneme, specifically beak-like projections in the lumens of the B-tubules of the number 1, 5, and 6 doublets, and a bridge between the number 1 and 2 doublets. Rupp et al. 
reported a biased loss of outer dynein arms on a subset of outer doublets in the sup-pf-2
mutant. It will be of interest to determine if PACRG provides the biochemical basis for any of these differences between individual outer doublets.
FAP73-HA Is Enriched at a Single Spot Between the Two Basal Bodies
In addition to its punctuate distribution along the flagella, FAP73-HA was localized to a single spot near the basal bodies. Double staining with anti-IFT46 showed that FAP73-HA is concentrated between the two flagella-bearing basal bodies of C. reinhardtii
. The region between the basal bodies is spanned by two proximal fibers and the massive distal connecting fiber, which contains the calcium-modulated protein centrin [Huang et al., 1988
]. FAP73 is predicted to have a high probability of forming an α-helical coiled-coil, a structural motif common to components of the flagellar basal apparatus. In future experiments, it will be interesting to determine the localization of FAP73 at an ultrastructural level.
FAP232 Is a Novel IFT Particle Protein and Is Renamed IFT25
FAP232 and FAP173 were found exclusively and primarily, respectively, in the detergent-soluble fraction of the C. reinhardtii flagellar proteome. This fraction is enriched in components of IFT, the cellular machinery that assembles and maintains flagella. IFT particles consist of at least 18 different proteins, and studies in various organisms have identified additional proteins that are required for, control, or are moved by IFT. We therefore were interested in whether FAP232 and FAP173 are involved in IFT. Both proteins exhibited a labeling pattern similar to that of known IFT proteins – i.e, a punctate distribution along the flagella with brighter foci at the basal bodies – and in both cases the flagellar labeling was lost upon treatment of the cells with nonionic detergents, as also is the case for IFT proteins. However, FAP173-HA did not co-localize with the complex B protein IFT46. Moreover, it did not accumulate at the tips of regenerating flagella, as IFT components do. These results suggest that it is not an IFT component and instead represents a different, currently unknown, compartment or process in the flagellum. In contrast, FAP232-HA exhibited a 100% co-localization with IFT46 in steady-state flagella. It also accumulated at the tips of growing flagella. The HA tag also allowed us to follow FAP232 in biochemical experiments: FAP232-HA co-sedimented with IFT46 in sucrose density gradients, and antibodies to the HA tag co-immunoprecipitated IFT46. Conversely, antibodies to IFT46 co-immunoprecipitated FAP232-HA. We conclude that FAP232 is a component of IFT complex B, and rename it IFT25, consistent with the nomenclature for other C. reinhardtii IFT-particle proteins.
IFT25 Exhibits an Unusual Pattern of Evolutionary Conservation
Similar to many axonemal proteins, most of the known components of IFT are highly conserved and widely distributed among organisms having cilia or flagella. Putative orthologues of IFT25 are present in many flagellated protists and metazoans, including both protostomes and deuterostomes, but appear to be absent from the genomes of diatoms, the moss Physcomitrella, apicomplexan protists, ciliates, D. melanogaster, C. elegans, and C. intestinalis. This distribution suggests that IFT25 was part of the ancestral IFT machinery but was repeatedly lost during evolution in unrelated groups. One possibility is that the role of IFT25 has been taken over by a different protein in some species. Another possibility is that IFT25 has a more specialized role than general flagellar assembly and that its function is required in some species but not in others. Interestingly, one feature common to those organisms lacking IFT25 is that, as far as we know, their cilia and flagella are not disassembled as part of the cell cycle. This also is of interest given the apparent connection between IFT25 and IFT27 (see below).
The Human Homologue of IFT25 Is HSPB11
The human homologue of IFT25, which shares 37% identity (58% similarity) with the N-terminal 130 residues of C. reinhardtii
IFT25, was originally identified as placental protein 25 (PP25) [Bohn and Winckler 1991
]. Recently, the protein was shown to have sequence similarity to and properties of small heat shock proteins and was renamed small heat shock protein beta-11 (HSPB11, also known as C1orf41 protein, HSPCO34, and HSP16.2) (Bellyei et al. 2007
]. X-ray crystallography and NMR spectroscopy have established the structure of HSPB11, which consists largely of short beta-strands (http://www.ncbi.nlm.nih.gov/Structure/mmdb/mmdbsrv.cgi?uid=30728
) [Ramelot et al. 2008
]. A similar structure has been observed for the galactose-binding domain of sialidase from Micromonospora viridifaciens
(PDB ID 1EUT), other cell surface-attached carbohydrate-binding domains, and the human anaphase-promoting complex subunit 10 (PDB ID 1JHJ). Using western blotting and immunocytochemistry, HSPB11 has been localized in the cytoplasm, the nucleus, and the mitochondria of cultured cells [Bellyei et al., 2007
]. However, it is not clear if the cells used in these studies were ciliated. The subcellular distribution of the FLAG-tagged mouse HSPB11/IFT25 homologue was examined in IMCD3 cells in the accompanying paper by Follit et al. 
, who observed that it was localized to the cilium and centrosome. Follit et al. also observed that an anti-FLAG antibody co-immunoprecipitated FLAG-IFT25 and complex B proteins IFT20, IFT27, and IFT88 but not the complex A protein IFT140. Thus, mouse HSPB11/IFT25, like C. reinhardtii
IFT25, appears to be an IFT complex-B component.
Several other chaperones and heat shock proteins have been identified in the flagella and cilia of C. reinhardtii
and other organisms [Bloch and Johnson, 1995
; Stephens and Lemieux, 1999
, and see references cited therein]. For example, HSP70A, which appears to be relatively abundant in the flagellum [Pazour et al., 2005
], has a punctate distribution along the flagellum that partially overlaps with that of the IFT motor kinesin-2 [Shapiro et al., 2005
]. Radial spoke proteins RSP12 and RSP16 are predicted to function as molecular chaperones [Yang et al., 2006
]; the latter is transported separately into the flagellum and then apparently joins with the radial spoke precursor complex during the latter's assembly onto the axoneme [Yang et al., 2005
]. Some Bardet Biedl Syndrome proteins, which are associated with IFT, show sequence homology to chaperonins [Blacque and Leroux, 2006
]. It is possible that IFT25 has a chaperone function during the loading and unloading of flagellar precursors from IFT particles, during the remodeling of IFT particles from anterograde to retrograde forms, or in the transport of cargo during IFT.
IFT25 and IFT27 Have Similar Phylogenetic Distributions and May Interact Physically and Functionally
Systematic mapping of human binary protein–protein interactions of 8,100 open reading frames using a stringent, high-throughput yeast two-hybrid system [Rual et al., 2005
] identified only one interaction involving an established IFT complex B protein, namely IFT27 [Qin et al., 2007
] as a putative binding partner of the human homologue of IFT25. As noted above, Follit et al. 
observed that the mouse homologue of IFT27 was co-immunoprecipitated with IFT25, although it was not demonstrated that this interaction is direct. Interestingly, the phylogenetic distribution of IFT27 is similar to that of IFT25. Like IFT25, IFT27 is present in many ciliated species including C. reinhardtii
and various vertebrates but not in C. elegans
and D. melanogaster
. Taken together, these findings suggest that IFT25 and IFT27 may cooperate in some function related to assembly, maintenance, or disassembly of cilia and flagella.