Structural organization of the C1a-e-c supercomplex within the ciliary central apparatus

Cilia and flagella are highly conserved organelles in eukaryotes. They have roles in cell motility, generating fluid flow, and sensing extracellular cues. Defects in ciliary assembly or function cause a wide range of human diseases, collectively termed ciliopathies (Afzelius, 2004; Fliegauf et al., 2007). The “9+2” axonemal core structure of motile cilia consists of nine outer doublet microtubules (DMTs) surrounding two singlet microtubules (C1 and C2) that form the central apparatus (CA) or central pair complex (CP). Attached to this axonemal microtubule scaffold are hundreds of proteins (Pazour et al., 2005), including the inner and outer arm dynein motors, and regulatory complexes forming part of the signal transduction pathways that coordinate dynein activity to generate ciliary motility (Summers and Gibbons, 1971; Sale and Satir, 1977; Lin and Nicastro, 2018; Witman et al., 1978; Smith and Sale, 1992; Piperno et al., 1994; Smith and Lefebvre, 1997a; Porter and Sale, 2000; Smith, 2002; Mitchell, 2004; Nicastro et al., 2006; Dymek and Smith, 2007; Wirschell et al., 2007; Bower et al., 2009; Heuser et al., 2009, 2012a,b; Yamamoto et al., 2013; Loreng and Smith, 2017; Fu et al., 2018; Kubo et al., 2018).

The CA is the largest known ciliary regulatory complex. Early structural analyses described the CA as an asymmetric assembly with seven C1 and C2 projections, but our previous cryo-electron tomography (cryo-ET) study of the WT Chlamydomonas reinhardtii CA revealed at least 11 projections that have 16–32-nm periodicities along the ciliary length and form connections between C1 and C2, as well as to the radial spoke (RS) heads (Witman et al., 1978; Dutcher et al., 1984; Mitchell and Sale, 1999; Mitchell, 2003; Mitchell and Smith, 2009; Carbajal-González et al., 2013; Loreng and Smith, 2017). Mutations of CA components often result in impaired or paralyzed cilia (Witman et al., 1978; Dutcher et al., 1984; Smith and Lefebvre, 1996, 1997b; Smith and Yang, 2004). Deficiency of CA proteins can cause mammalian ciliopathies, including primary ciliary dyskinesia (PCD; Teves et al., 2016; Horani and Ferkol, 2018). Mice deficient in either Cfap54 or Pcdp1, which encode conserved C1d proteins, show typical PCD symptoms, including ineffective mucus clearance, male infertility, and hydrocephalus (Lee et al., 2008; DiPetrillo and Smith, 2010; McKenzie et al., 2015). Similarly, mutations of the C2b protein Hydin result in hydrocephalus caused by the loss of cilia-generated fluid flow in the brain ventricles (Lechtreck et al., 2008).

Despite many biochemical and structural studies of the CA (Witman et al., 1978; Dutcher et al., 1984; Smith and Lefebvre, 1997a; Mitchell, 2003; Mitchell and Smith, 2009; Loreng and Smith, 2017), the protein composition, 3D organization, and functional mechanisms of the CA in ciliary motility are not fully understood. Our recent mass spectrometry (MS) study compared the proteomes of Chlamydomonas WT and mutant axonemes and identified 44 new candidate CA proteins assigned to the C1 or C2 microtubule (Zhao et al., 2019). However, questions about the organization, assembly, and function of the CA and its projections remain, making the CA the structurally and functionally least understood axonemal complex to date.

Here we combined biochemical, genetic, and structural analyses to investigate the protein composition and molecular organization of a group of interconnected CA projections, here termed the C1a-e-c supercomplex, in WT and CA mutants of Chlamydomonas. Sucrose gradient sedimentation and MS revealed that several CA proteins identified in this study, flagellar associated protein 76 (FAP76), FAP81, FAP92, and FAP216, are associated with the protein PF16, previously assigned to the C1 microtubule but not to a specific projection. Chlamydomonas mutants that lacked any of these proteins showed impaired motility. Structural comparisons of flagella from WT, these mutants, and tagged rescue strains revealed the precise locations of PF16, FAP76, FAP81, FAP92, and FAP216 within the C1a-e-c supercomplex. Our data show that stable assembly of this supercomplex and its interaction with the neighboring C1d projection are required for the proper regulation of ciliary motility.

Cilia were isolated from Chlamydomonas cells, demembranated, and frozen rapidly for cryo-ET imaging and subtomogram averaging of the DMT and CA repeats. Our previous cryo-ET study of the WT CA structure of Chlamydomonas flagella achieved 3.5-nm resolution (Fourier shell correlation [FSC] 0.5 criterion; Fig. 1, A and C–E; Carbajal-González et al., 2013). Here, we improved the resolution of the CA structure to 2.3 nm (Fig. 1, B, C, F, and G) by applying advanced hardware and software. For example, tilt series were recorded with multiple frames per image (to correct for beam-induced sample motion; Brilot et al., 2012) on a direct electron detector (Cheng et al., 2015), using a Volta-Phase-Plate (to improve image contrast close to focus; Danev et al., 2014) and a dose-symmetric tilting scheme (to reduce radiation damage of the sample in images recorded at low tilt angles; Hagen et al., 2017). This improves the resolution of the subtomogram averages of axonemes, as shown by averages of DMT-associated structures with resolution as high as 1.8 nm (FSC 0.5 criterion; Lin et al., 2019).

The improved 3D structure of the Chlamydomonas WT CA reveals details of the CA microtubules, the bridge between them, and the various CA projections (Fig. 1, F–H; and Video 1). Previously unresolved details include (a) three filamentous structures, with a <2-nm diameter, between protofilaments 9 and 12 of the C2 microtubule (compare Fig. 1, D and F); (b) the microtubule inner protein, MIP-C2a, which forms an arch-shaped structure on the inside of the C2 wall (Carbajal-González et al., 2013; Fig. 1 E), consists of two distinct substructures that are bound to alternating tubulin subunits of protofilament 5 (Fig. 1 G); (c) the C2a projection, previously shown to have an 8-nm periodicity (Carbajal-González et al., 2013), instead has a 16-nm periodicity and adopts two conformational states (Fig. S1, A–E); (d) two connections between the C1a and C1e projections; and (e) a peripheral density at the C1c projection (Fig. S1, F and G).

Our recent CA proteomics study identified several candidate C1-proteins that are reduced or absent from isolated axonemes of the Chlamydomonas “paralyzed flagella” mutant pf16 (Zhao et al., 2019); these axonemes lack the entire C1 microtubule and its associated projections (Dutcher et al., 1984; Smith and Lefebvre, 1996). To identify proteins that are closely associated with the PF16 subunit, and to visualize their locations within distinct C1 projections using cryo-ET, we narrowed down the list of C1-candidate proteins to those that cosediment with PF16. Rather than comparing the cosedimentation profiles of the full axonemal proteomes between WT and pf16, we limited the proteome of interest by comparing the pf28 mutant, which lacks the large outer dynein arm complexes but contains the CA (Mitchell and Rosenbaum, 1985; Kamiya, 1988), with the double mutant pf28;pf16. The isolated axonemes were extracted using high potassium iodide (KI) concentration, which eliminates the DMTs and proteins strongly associated with the DMTs from the sample. The KI extracts were separated on sucrose gradients under native conditions.

We probed the sucrose gradient fractions of the pf28 extract with anti-PF16 antibody, analyzed the positive fractions with MS, and subtracted those proteins that were also found in comparable fractions of the C1-less double mutant pf28;pf16 (Table S1). Cross-referencing the proteins of this difference set with previous studies identified 16 potential PF16-associated proteins (Table 1): five proteins (PF6, FAP101, FAP114, FAP119, and FAP227) were previously assigned to the C1a projection and two (FAP221 and FAP54) to the C1d projection, whereas PP1c (Yang et al., 2000) and eight recently identified candidate C1 proteins (Zhao et al., 2019) had not yet been assigned to specific C1 projections. Of these nine unassigned proteins, FAP76, FAP81, FAP92, and FAP216 were large enough to be suitable targets for cryo-ET imaging. Chlamydomonas mutants corresponding to FAP76 (fap76-1 and fap76-2), FAP81, FAP92, and FAP216 were purchased from the Chlamydomonas Library Project (CLiP; https://www.chlamylibrary.org; Li et al., 2016). Mutations were verified by PCR (Fig. S2 A) and MS (Table 1), and rescue of the mutant phenotypes (motility and structure), by transformation with the corresponding WT genes. We also generated a fap76-1;fap81 double mutant.

A previous conventional EM study showed that in the pf16 mutant, the C1 microtubule and its projections were unstable upon isolation of axonemes, i.e., only 8% of pf16 axonemes contained the C1 and C2 microtubules (9+2 axonemes), whereas most of the intact pf16 flagella had both CA microtubules, as did flagella and isolated axonemes from WT Chlamydomonas (Dutcher et al., 1984; Mitchell and Sale, 1999). However, the resolution in previous studies was insufficient to visualize the extent of missing C1 projections in pf16 axonemes containing the C1 microtubule. Therefore, to characterize the CA structure that requires PF16 protein for its stable assembly, we performed cryo-ET and subtomogram averaging of those pf16 CAs that still contained a C1 microtubule (9+2) and compared the averages to the WT CA structure. Of 39 cryo-tomograms of pf16 axonemes, only 10% had both microtubules (9+2), whereas the remaining axonemes lacked either the C1 microtubule (9+1, 51%) or the entire CA (9+0, 39%), consistent with previous studies (Dutcher et al., 1984; Mitchell and Sale, 1999). The subtomogram averages revealed that in the pf16 9+2 axonemes, the C1d, C1f, and C1b projections remained structurally intact, and only the C1a, C1e, and C1c projections, with an estimated total mass of ∼2.0 MD, were absent (Fig. 2, E–H; for consistency, we follow the previous nomenclature for CA projections [Carbajal-González et al., 2013]). Thus, the C1a-e-c projections require PF16 for stable assembly onto the C1 microtubule, and they form a protein interaction network suggestive of a “C1a-e-c supercomplex” within the ciliary CA.

Although the structural defect in the pf16 mutant was informative about the extent of the PF16-dependent protein network, it did not reveal the location of the PF16 protein within the C1a-e-c supercomplex. To determine the location of PF16, we used the clonable biotin carboxyl carrier protein (BCCP) tag, developed for visualizing gene products by cryo-ET, and subtomogram averaging (Oda and Kikkawa, 2013; Song et al., 2015). We rescued the pf16 mutant with WT PF16 C-terminally tagged with BCCP. Axonemes of the rescued strain were isolated, and the biotinylated tag was enhanced with streptavidin-nanogold, which is visible as additional cryo-EM density in comparison with the WT structure (Song et al., 2015; Fu et al., 2018). Tomograms of the pf16;PF16::BCCP rescue showed only 9+2 axonemes, and all CA projections were restored to the WT architecture with and without (control) addition of streptavidin-gold (Fig. 2, I–P), confirming that stable assembly of the C1 microtubule and C1a-e-c supercomplex requires PF16 protein. In contrast to the WT (Fig. 2, A–D) and control axonemes without streptavidin-gold (Fig. 2, I–L), the CA average of pf16;PF16::BCCP with streptavidin-gold revealed an additional cryo-EM density in the periphery of the C1a projection (blue arrowheads in Fig. 2, M–P; and Video 2), indicating the location of the PF16 C-terminus.

We were unable to rescue the pf16 mutant with N-terminally tagged PF16, possibly because the PF16 N-terminus is required for protein–protein interactions and assembly of the C1a-e-c supercomplex, or the N-terminal BCCP tag (9 kD) disrupts PF16 folding or its transport into the axoneme.

CA mutants often have paralyzed flagella (see pf16 in Fig. 3 A). In contrast, the fap76-1, fap81, fap92, and fap216 cells swim, but with slower velocity and a curving swimming path, caused by lack of synchronization between the two flagella (Video 3), rather than the straight or loose helical paths of WT cells (Fig. 3, A and B; and Video 3). Among the four mutants, fap92 has the mildest motility impairment (70% WT swimming speed and only a slight curving path). Cells from fap76-1, fap81, fap216, and the double mutant fap76-1;fap81 have similarly severe motility defects, with slower velocity (∼50% of WT) and clearly curving swimming paths (Fig. 3, A and B). Flagella asynchrony has also been reported for a Chlamydomonas fap74 RNAi mutant that lacks the C1d projection (DiPetrillo and Smith, 2011).

Previous studies have shown that C1d-less mutants lack a proper photoshock response (DiPetrillo and Smith, 2011; Brown et al., 2012). Our results show that the photoshock response is also impaired in the C1a-e-c supercomplex CLiP mutants, i.e., only 35%, 6%, 7%, and 5% of fap76-1, fap81, fap216, and double mutant fap76-1;fap81 cells, respectively, exhibited photoshock, whereas fap92 had a milder photoshock defect (81%; Fig. 3 C). The WT swimming speed and straight swimming path were completely restored in the rescue strains fap76-1;BCCP::FAP76 and fap216;BCCP::FAP216 (Fig. 3, A and B); however, the photoshock response of fap76-1;BCCP::FAP76 was still slightly impaired (72%; Fig. 3 C). The similar phenotypes of the C1a-e-c and previously reported C1d mutants suggest that these CA projections might functionally interact to regulate ciliary motility.

The impaired cell motility and the perturbed proteome of the C1a-e-c supercomplex mutants suggest defects in their flagellar structure. Therefore, we used cryo-ET and subtomogram averaging to compare the CLiP mutant CA and control DMT structures with WT axonemes. None of the mutant averages showed DMT-associated defects (Fig. S3, A–G), but all of them revealed structural defects in the CA, with a loss of mass ranging from 80 kD to 1.2 MD.

Compared with WT, the fap92 CA is missing a sheet-like C1a density (compare Fig. 3, D–G with H–K; and Video 2). In the WT CA, this sheet-like structure has a 16-nm periodicity along the axonemal length and contributes to the connection between neighboring C1a projections in the periphery of the CA (Fig. 3 G). The MS analyses identified similar numbers of unique peptides throughout the predicted amino acid sequence of FAP92 in both WT and fap92 axonemes, suggesting that FAP92 is present in isolated fap92 axonemes (Table 1). However, sequencing of the genomic DNA showed that the cassette was inserted into the 13th exon of FAP92 in the mutant genome (Fig. S2 A), which was accompanied by the deletion of three nucleotides (ACC) immediately upstream of the insertion site in exon 13 (unpublished data). The remainder of exon 13 was intact downstream of the cassette. RT-PCR confirmed that the coding region of FAP92 was disrupted, with a fragment corresponding to exons 12–15 being amplified from WT but not from fap92 cDNA (Fig. S2 B).

The mutant FAP92 protein structure is likely altered, causing it to be partially disordered or positionally flexible and thus not visible in the axonemal averages, and/or precluded from forming stable interactions with FAP413, which was significantly reduced in fap92 axonemes (Table S2). The remaining C1a structures are not destabilized in the fap92 mutant (e.g., the classification analyses did not reveal structural flexibility of the C1a projections), and neighboring C1a projections are still connected through additional linking structures (Fig. 3 K), even though fap92 cells swim slowly, indicative of impaired flagellar motility. Thus, FAP92 probably has a role in regulating ciliary beating, as opposed to having a major scaffolding function.

Cryo-ET and subtomogram averaging of the fap76-1 mutant revealed that the outer part of the C1c projection was missing in the mutant CA compared with WT (compare Fig. 4, A–D with E–H). The FAP76-dependent density (magenta in Fig. 4 I) has a triskelion-like shape with 32-nm periodicity along the C1 microtubule. MS analysis of fap76-1 axonemes revealed that FAP76 is significantly reduced (Table 1), and the estimated size of the triskelion-shaped density of ∼190 kD suggests it contains a single copy of FAP76 (170 kD). FAP76 forms at least three connections with neighboring structures involving three different CA projections: #1 is the interaction between FAP76 and other C1c densities near the interface between the C1c and C1e projections (Fig. 4 I and Video 2); #2 is the attachment of FAP76 to a rod-shaped C1d density; and #3 links two C1d projections (Fig. 4 J and Video 2).

The FAP76 density consists of a 32-nm-long filamentous part and a thicker “branch” extending from the middle of the filament to connection #1 (Fig. 4 I). To analyze the FAP76 structure, we rescued fap76-1 with N-terminally BCCP-tagged FAP76. Rescue cells swim (mostly) with WT motility (Fig. 3, A–C). Using immunofluorescence microscopy and SDS-PAGE, we confirmed that BCCP-FAP76 assembled into the axonemes of the rescued cells (Fig. 5, A and B). In the subtomogram averages of fap76-1;BCCP::FAP76 axonemes labeled with streptavidin-nanogold, we found an additional density where the short “branch” of FAP76 connects to the C1e projection (Fig. 5, J–L, blue arrowheads), indicating the location of the FAP76 N-terminus. This additional density was not visible in WT axonemes or fap76-1;BCCP::FAP76 without streptavidin-nanogold (Fig. 5, D–I, white arrowheads). Although the overall CA structure of fap76-1;BCCP::FAP76 resembles that of WT, one of the two globular-shaped structures at the C1c-e junction (C1c peripheral subunit 1 [psu1] and black arrow in Fig. 5, E and F) appears to be missing with and without incubation with streptavidin-Au (white arrow in Fig. 5, H, I, K, and L). A classification analysis focused on C1c psu1 revealed that ∼40% of WT and ∼38% of fap76-1 CA repeats, but <10% of CA repeats in the fap76-1;BCCP::FAP76 rescued strain, contained the C1c psu1 structure (Fig. 5). The close proximity of the FAP76 N-terminus to this subunit suggests that the BCCP tag in the rescue strain might disrupt assembly of the globular subunit. This defect might also explain the partial rescue of the photoshock response in fap76-1;BCCP::FAP76 (Fig. 3 C). Comparisons of the proteomes of WT, fap76-1 (Table 1), and fap76-1;BCCP::FAP76 (not depicted) did not reveal missing proteins that might correspond to C1c psu1, suggesting that C1c psu1’s apparent absence in the tomogram averages of the latter strain is due to positional flexibility that occurs when FAP76 is replaced with FAP76-BCCP.

We also analyzed a second CLiP mutant, fap76-2, in which the FAP76 gene was disrupted by insertion of the cassette at a location closer to the 3′ end of the gene than in fap76-1 (Fig. S4 A). The phenotype of fap76-2 is very similar to that of fap76-1 with regard to both motility (Fig. S4, B–D) and structural defects (Fig. S4, F and G). In contrast to fap76-1, MS analysis of fap76-2 showed that a truncated FAP76 protein is assembled into the fap76-2 axoneme, and that levels of two more proteins, DPY30 and FAP380, are reduced compared with fap76-1 (Fig. S4, E–G). Competition with this truncated FAP76 protein could explain our observation that FAP76-HA expression in fap76-2 results in only a partial rescue of motility (unpublished data), whereas fap76-1 is completely rescued (Zhao et al., 2019).

The fap81 CA had the most severe structural defects of the four CLiP mutants. The WT versus mutant comparison revealed that the C1e and C1c projections were missing in the averaged fap81 CA repeat (compare Fig. 6, A–C with D–F; and Video 2), indicating that FAP81 is required for stable assembly of the C1e-c subcomplex. The estimated size of the WT C1e-c subcomplex is 1.2 MD. However, MS and Western blot analyses of fap81 axonemes revealed that, of the 16 proteins associated with PF16, only FAP81 (172 kD) and FAP216 (79 kD) were substantially reduced (Table 1 and Fig. S2 D). This suggests that some of the C1e-c proteins might still bind to the fap81 axoneme but cannot be visualized in the averaged fap81 CA due to partial reduction and/or positional flexibility. This is supported by the observation that in the region where FAP76 connects to the C1d projection (#3 connection, Fig. 4 I), a weak FAP76 density remained visible in the averaged fap81 CA, whereas this density was missing in fap76-1 and fap76-1;fap81 (compare Fig. S5, A and C with B and D). A classification analysis focused on this region could clearly visualize the part of FAP76 that connects to the C1d projections in more than half of the fap81 CA repeats (compare Fig. S5, E and F), which is consistent with MS data showing that a considerable number of FAP76 peptides were detected in the fap81 axonemes (Table 1). Nonetheless, large parts of FAP76, including its N-terminal region, remained undetectable even in the class averages of fap81. This suggests that FAP81 interacts with the N-terminal part of FAP76, and without this interaction FAP76 becomes positionally flexible. Despite the partially remaining FAP76 protein in fap81, the similar phenotypes of fap81 and fap76-1;fap81 indicate that the positionally flexible FAP76 did not function properly in fap81 axonemes.

Comparison of the WT and fap81 CAs allowed a more precise characterization and delineation of the C1a and C1e densities. In WT, the C1a projection, which repeats every 16 nm, has multiple interaction sites with the C1e projection, which repeats every 32 nm (Fig. 6 G). Our data clearly showed that alternating C1a projections interact with different parts of each C1e projection (Fig. 6 G). The results for fap81 also provided insight into the interactions between FAP81 and C1d. Since FAP81 is not significantly reduced in fap76-1 axonemes (Table 1), the remaining C1c densities in fap76-1 are likely composed of FAP81 and FAP216. Comparison of fap81 with fap76-1 reveals that FAP81 interacts with a rod-shaped C1d density that is anchored to C1 protofilament 4 (Fig. 6 H).

Our MS analyses revealed that, in addition to FAP81, FAP216 was significantly reduced in fap81 axonemes, whereas FAP81 was present at WT level in fap216 (Table 1). This suggests that the assembly of FAP216 into the axoneme requires FAP81 but not vice versa. Structural analysis of fap216 showed that two small densities were missing or reduced in the C1c and C1e projections (Fig. 7, E–H, light orange and white arrowheads). In the WT CA, the affected C1c structure links the more peripheral C1c projection density to the C1 microtubule between protofilaments 2 and 3 (Fig. 7, A–D, yellow arrowheads). The estimated molecular mass of the affected C1c density is ∼90 kD, suggesting that this density is composed of a single copy of FAP216 (79 kD). The affected C1e density, which forms a bridge between the C1c and C1a projections in WT (orange arrowhead in Fig. 7, A and C), is only partially reduced in fap216 (light orange arrowhead in Fig. 7, E and G), and classification analysis revealed that this C1e density was missing in only 57% of the fap216 CA repeats (Fig. S5, G–J). This, together with the results of the MS analyses, which did not detect any FAP216 peptides in fap216 axonemes, indicates that the partial reduction of this density is likely a secondary defect. The proteins corresponding to the reduced C1e density could not be identified by our MS comparisons between WT and fap216 mutant axonemes.

Although the structural CA defects in fap216 are considerably less severe than those of fap81, both mutants display equally severe swimming defects. We also showed that transformation of fap216 with the WT gene for FAP216 is sufficient to restore the WT phenotype (Fig. 3, A–C). Classification analyses of the fap216 CA focused on the areas neighboring the structural defects did not reveal any positional flexibilities of the remaining densities, i.e., FAP216 does not appear to play a critical role in the stable assembly of the remaining C1a-e-c protein network. Instead, these results suggest that FAP216 is an essential component in a signal transduction cascade that regulates ciliary motility in Chlamydomonas.

Our WT versus mutant comparisons have revealed a 2-MD PF16-associated protein network, the C1a-e-c supercomplex, within the ciliary CA. We show that PF16 is potentially associated with ≥16 proteins (Table 1) with a combined molecular mass of 1.6 MD, suggesting that multiple protein copies and/or additional proteins form the supercomplex. Our MS analyses also identified two proteins, FAP54 and FAP221, previously assigned to C1d, among the PF16-associated proteins, suggesting that they have stable attachments to C1a-e-c supercomplex proteins. Such strong interaction between the C1c and C1d projections is consistent with the observed structural connections.

We characterized mutant phenotypes and the molecular organization for five C1a-e-c supercomplex proteins. PF16 and FAP92 are C1a projection proteins, whereas FAP76, FAP81, and FAP216 are the first identified C1e and C1c proteins (Table 1, Fig. 8, and Video 2). The CA structural defects in the mutants suggest hierarchical assembly of the C1a-e-c supercomplex. The complex was disrupted in pf16 axonemes, suggesting that the C1a protein PF16 is a crucial scaffold protein required for stable assembly of the entire supercomplex (Figs. 2 and 8). Loss of PF16, and the supercomplex, destabilizes the entire C1 microtubule and associated projections, as only ∼10% of the pf16 axonemes had 9+2 axonemes.

The second largest supercomplex defect was observed in fap81 axonemes, suggesting that FAP81 is essential for stable assembly of the C1e-c subcomplex that consists of at least FAP76, FAP81, and FAP216. In contrast, the single loss of FAP76, FAP92, or FAP216 had minimal effects on the assembly of other C1a-e-c supercomplex components. Despite at least two FAP76 contact sites and one FAP81 contact site to C1d projection densities (Figs. 4 and 6), none of the mutants showed defects in C1d projection assembly. Likewise, classic thin-section transmission EM (TEM) of two C1d-defective mutants, fap46 and FAP74 RNAi, which lack the entire C1d projection, showed no obvious effects in structures corresponding to the C1a-e-c supercomplex (DiPetrillo and Smith, 2010; Brown et al., 2012). Our precise localization of FAP76, FAP81, FAP92, and FAP216 within the C1a-e-c supercomplex agrees well with the recent assignment for these proteins to the C1 microtubule (Zhao et al., 2019). In our previous study (Zhao et al., 2019), FAP81 was found to immunoprecipitate with known C1a/e proteins, but it could not be definitively assigned to either of these projections. This highlights that the integration of biochemical and structural data are critical for a comprehensive understanding of the molecular architecture of the CA.

The improved resolution of our cryo-ET data revealed that the C1a-e-c supercomplex makes three connections with the neighboring C1d projection and multiple interactions within the complex, such as the interface between the C1a and C1e projections, where the structural periodicity transitions from 16 nm (C1a) to 32 nm (C1e; Fig. 6 G). Alternating C1a projections, which physically interact with each other, have different connections to each C1e projection, suggesting that the alternate C1a projections could have distinct functions.

The C1a projection is anchored to C1 protofilament 1, with 16-nm periodicity (Figs. 2 C and 6 E), which might involve the N-terminal domain of PF16 (Fig. 8). Every 16 nm, the C1e-c subcomplex has two adjacent connections to C1 protofilament 2. The smaller connecting density consists of FAP216 (yellow arrowheads in Fig. 7, A–D) and did not appear essential for complex docking, whereas the larger distal connection is a protein interaction hub involving multiple structures, including C1e, the C1c psu1 and psu2, FAP81, and the N-terminal domain of FAP76 (#1 connection, Fig. 4 I). In addition to being part of the latter interaction hub, the triskelion-shaped FAP76 formed at least three more interfaces with neighboring structures: connections to each other in the longitudinal direction (Fig. 4 I) and two connections to the neighboring C1d projections (Fig. 4 J). The C1d projections themselves were bound every 32 nm to C1 protofilaments 4 and 6 (Fig. 4, G and H; Fig. 6, E and F; and Fig. 8). Given the motility defects observed in all C1a-e-c mutants, including fap216 with its very small structural defect, these subunits are likely involved in the same signal transduction pathway that regulates ciliary motility. Their connectivity with neighboring structures may provide the structural basis for signal transmission.

Most previously identified CA protein mutations, including the C1a-e-c mutant pf16, result in paralyzed or twitching-only cilia (Dutcher et al., 1984; Smith and Lefebvre 1996; Rupp et al., 2001). The four other C1a-e-c supercomplex mutants identified here, fap76-1, fap81, fap92, and fap216, have slow swimming phenotypes. However, loss of FAP92 did not result in major motility or structural defects, likely because the major C1a components remained largely unaffected. In contrast, cells of the remaining three mutants, fap76-1, fap81, and fap216, frequently changed swim orientations, resulting in curving swimming paths (Fig. 3 B). A straight swimming orientation, or sharp turns, seen in Chlamydomonas cells, are coupled to synchronous or asynchronous ciliary beating, respectively (Polin et al., 2009). Consistent with this, the period of synchronous beating between the two flagella of the mutant cells was shorter than that of WT (Video 3), causing frequent reorientation during swimming. If the overall shape of the flagellar waveform is not changed, as is the case here (Video 3), then the loss of synchrony between the flagella could be caused by random delays in switching of dynein activity from one side of the axoneme to the other in one of the two flagella. Our findings indicate that the C1e-c subcomplex, which is largely composed of FAP76, FAP81, and FAP216, may play a critical role in coordinating the oscillatory switching of dynein activity, and thus in maintaining synchrony between the two beating cilia.

The C1e-c subcomplex may act in the same regulatory pathway as the adjacent C1d projection, because previous studies reported similar motility defects for Chlamydomonas C1d mutants, i.e., slow swimming, uncoordinated ciliary beating, and a deficient photoshock response (DiPetrillo and Smith, 2010, 2011; Brown et al., 2012). In our model, the C1e-c subcomplex is part of the protein network that requires PF16 for stable assembly, but it has a 32-nm periodicity, identical to that of the C1d projection and twice that of the C1a projection. Our cryo-ET data revealed that the C1c projection interacts directly with the C1d projection through FAP76 and FAP81 (Fig. 8 B), and our combined sucrose gradient and MS data provided evidence that the PF16-associated protein network is physically attached to at least two proteins previously assigned to C1d. Thus, the C1e-c subcomplex may be physically and functionally distinct from the C1a projection, but closely related to C1d. Given the more severe motility defects observed in the C1d mutants (DiPetrillo and Smith, 2010, 2011; Brown et al., 2012), regulatory signals might be transmitted from the C1d projection to the C1e-c subcomplex through their direct connections. From the C1e-c subcomplex, the signal could be sent downstream either to the C1a projection or to RS heads to ultimately modulate dynein activities. One such signal could be initiated by changes in intraflagellar calcium concentration, because the C1d protein FAP221 binds calmodulin in a Ca²⁺-sensitive manner and thus could mediate Ca²⁺-induced changes in flagellar waveform (Witman, 1993; DiPetrillo and Smith, 2010; Brown et al., 2012). Although FAP76 could still bind to the C1 microtubule and connect to the neighboring C1d projection in the fap81 mutant (Fig. S5, A–F), the positional flexibility of FAP76 caused by loss of its N-terminal binding partner FAP81 could be sufficient to disrupt the proposed signal-transmission cascade between C1d and C1c-e.

How the CA signals are transmitted to DMT dyneins through the interaction with RSs is poorly understood. In Chlamydomonas, the CA is slightly twisted around its longitudinal axis (90° twist over 2-µm CA length; Carbajal-González et al., 2013), and the CA has been shown to rotate around the ciliary axis during motility (Kamiya, 1982). Therefore, the CA projections in Chlamydomonas flagella are not expected to have a fixed interaction with RSs from a specific DMT. However, previous in vitro microtubule sliding assays and classic thin-section EM showed that the C1 microtubule was predominantly oriented toward the set of actively sliding DMTs, which switches between DMTs 2–4 and 6–8 in the two halves of the axoneme (Wargo and Smith, 2003). In addition, the orientation of the C1 microtubule seems to be correlated with the flagellar bending direction, as observed in axonemes prepared by freeze-etching (Mitchell, 2003). Switching of ciliary bending direction, which is critical to generate the undulating motion typical for cilia, correlates with the switching of dynein activity that drives interdoublet sliding between either DMTs 2–4 or 6–8 (Wais-Steider and Satir, 1979; Tamm and Tamm, 1984; Sale, 1986; Hayashi and Shingyoji, 2008; Lin and Nicastro, 2018). Therefore, we examined the spatial association between the CA projections and DMTs in WT and mutant axonemes. We found that the C1a-e-c supercomplex predominantly faced DMTs 6–8 in isolated and inactive WT and C1e-c mutant axonemes (Fig. S3 H). The preferential orientation of the C1a-e-c supercomplex toward DMTs 6–8 is consistent with the CA projections playing a role in controlling ciliary bending direction. However, the axonemes we isolated were inactive (no ATP added in the buffer); therefore, all dyneins were in their post–power stroke conformation. Future studies of the CA in actively beating cilia could reveal if the relative location of the C1a-e-c supercomplex switches between DMTs 2–4 and 6–8 in a bend direction–specific manner.

A previous study reported that the motility defects of the pf6 mutant, which lacks most of the C1a-e projections (not depicted), could be restored by adding exogenous tags to the RS heads (Oda et al., 2014). This result led to a working model that the CP–RS communication is, at least in part, mediated by nonspecific collision-based mechano-signaling between the CA projections and RS heads without requiring specific physiological protein–protein interactions at the CP–RS interface (Oda et al., 2014). The severe motility defects of the fap216 mutant suggest that a mechanically intact CA–RS interface alone is not sufficient for proper CA–RS communication and regulation of dynein activity. Despite missing only a small density in the center of the CA, 18 nm away from the interface with the RS heads (Figs. 7G and 8), the motility defect in fap216 is as severe as in fap81, which lacks the entire C1e-c subcomplex. This raises the possibility that FAP216 contributes to a chemical signal transduction pathway between the CA and RSs to regulate ciliary motility.

In humans, many axonemal gene mutations have been shown to cause ciliopathies such as PCD (Horani and Ferkol, 2018). Historically, conventional TEM analysis of patient ciliary samples is considered a standard clinical diagnostic tool for PCD (Kott et al., 2013; El Khouri et al., 2016; Edelbusch et al., 2017). However, the resolution of conventional TEM is greatly limited, hindering the diagnosis of PCD types caused by minor structural alterations. Therefore, future comparisons of normal and potential ciliopathy CA structures by cryo-ET may facilitate recognition of previously undiagnosed or misinterpreted defects in human patients (Lin et al., 2014). Overall, our study provides a partial “molecular blueprint” for the C1a-e-c supercomplex that will be the foundation for future studies into detailed protein–protein interactions and molecular mechanisms by which CA signals are transmitted to the RS heads to ultimately regulate dynein activity and thus ciliary beating.

Chlamydomonas WT strains used were g1 (nit1, agg1, mt+, a cross of nit1-305 to CC-124 [Pazour et al., 1995]; Chlamydomonas Resource Center, https://www.chlamycollection.org, CC-5415), and A54-e18 (CC-2929). The mutant pf28 (CC-3661) was obtained from Chlamydomonas Resource Center, and pf16A and pf16 (D2, the pf16 insertional allele) were generated as previously described (Dutcher et al., 1984; Smith and Lefebvre, 1996). The pf16 (D2) strain was used for MS and cryo-ET analyses. The pf28 and pf16 (D2) strains were mated to generate pf28;pf16 double mutant (DiBella et al., 2004). Strains from the CLiP (Li et al., 2016) included the parental strain CC-5325 and the insertional mutants fap76-1 (CLiP ID: LMJ.RY0402.089534), fap76-2 (LMJ.RY0402.088713), fap81 (LMJ.RY0402.092632), fap92 (LMJ.RY0402.204383), and fap216 (LMJ.RY0402.218389), all of which were obtained from the Chlamydomonas Resource Center. For MS analysis of axonemal proteins, fap76-1 was crossed to g1 (Zhao et al., 2019), which served as the control. As previously described (Fu et al., 2018), Chlamydomonas cells were maintained in solid Tris-acetate-phosphate (TAP) plates (supplied with 7.5 µg/ml paromomycin for CLiP mutants) and cultured in liquid TAP medium or modified M medium (Witman, 1986) under a 12:12-h light:dark cycle at 23°C with filtered air bubbling into the growing culture. Insertion sites of CLiP mutants were confirmed by PCR (Fig. S2 A) using the primers indicated in Table S4.

To generate the BCCP-tagged pf16;PF16::BCCP strain, the BCCP domain of Chlamydomonas acetyl-coenzyme A carboxylase gene (Cre17.g715250, Phytozome 12; amino acids 141–228 were used for tagging) was amplified by RT-PCR from genomic DNA from WT Chlamydomonas (strain A54-e18, CC-2929) as previously described (Oda and Kikkawa, 2013). A BCCP-tagged PF16 construct was generated by ligating the BCCP domain into the C-terminal MluI site of the pB6D2 plasmid that contains the PF16 gene (a 4.5-kb genomic fragment in pBluescript that rescues the mutant phenotype, Smith and Lefebvre, 1996). The BCCP-pB6D2 plasmid was cotransformed into pf16 (D2), along with the APHVIII gene, using the glass bead method (Kindle, 1990). Transformed cells were plated onto TAP plates with 20 µg/ml paromomycin. Cells were screened first for rescued motility, then by Western blot of axonemes to confirm the presence of PF16 along with the BCCP tag (rabbit anti-PF16 antibodies; 1:1,000, affinity purified; Smith and Lefebvre, 1996) and a streptavidin-HRP probe (1:5,000, GE Healthcare, GERPN1231).

For the FAP76-BCCP rescue construct, sequence encoding the BCCP tag was amplified from pIC2L-BCCPC-3xHA-Hyg (kindly provided by Professor Toshiyuki Oda, University of Yamanashi, Kofu, Japan) with primers F7/R7 (Table S4) and then was inserted into the MauB1 site of the plasmid pBC8, which contains a hygromycin cassette and the complete FAP76 gene (Cre09.g387689, Phytozome 12), to yield pBC25.

For the FAP216-BCCP rescue construct, portions of the FAP216 gene (Cre12.g497200, Phytozome 12) were amplified from genomic DNA with primers F8/R8, F9/R9, and F10/R10 (Table S4). Plasmid pBH (Zhao et al., 2019) was linearized by digestion with NdeI and SbfI, and the FAP216 fragments were assembled into it using NEBuilder HiFi DNA assembly master mix (NEB), yielding pBC26. To introduce the BCCP tag, the sequence encoding the tag was amplified from pIC2L-BCCPC-3xHA-Hyg with primers F11/R11 (Table S4) and then cloned into pBC26 at the MauB1 site, yielding pBC27.

All constructs were verified by sequencing. Chlamydomonas cells were transformed by the glass bead method (Kindle, 1990). After transformation, cells were grown on TAP agar supplemented with 10 µg/ml hygromycin (Sigma-Aldrich). Cells were screened for rescued motility; incorporation of the construct was then confirmed by Western blotting with a streptavidin-HRP probe (1:3,000, Molecular Probes). West Dura Extended Duration Substrate (Thermo Fisher Scientific) was used for HRP detection.

Total RNA was isolated from WT and fap92 cells using RNeasy Plus Mini Kit (Qiagen). cDNA synthesis was performed using SuperScript IV reverse transcription (Invitrogen) following the manufacturer’s protocol, and both Oligo(dT)₂₀ and random hexamers were used as primers. FAP92 (Cre13.g562250, Phytozome 12) fragments were then amplified using primers F12/R12 (Table S4) and sequenced. The gene encoding G protein β subunit (Cre06.g278222, Phytozome 12; Schloss, 1990) was used as the control and amplified with primers F13/R13 (Table S4).

Isolated axonemes were extracted in 0.5 M KI in NaLow (10 mM Hepes, pH 7.5, 5 mM MgSO₄, 1 mM DTT, 0.5 mM EDTA, and 30 mM NaCl) at 6 mg/ml for 30 min on ice. After centrifugation, the extracted proteins were dialyzed into NaLow and then loaded onto a 5–20% sucrose gradient made in NaLow and centrifuged at 35,000 rpm for 16 h in a SW41Ti rotor (Beckman Coulter). Gradients were fractionated (0.5 ml fractions) from bottom (fraction 1) to top (fraction 24).

For Western blot analyses, the sucrose gradient fractions were separated using SDS-PAGE and transferred to polyvinylidene difluoride membrane. For detection of PF16 protein, membranes were probed with anti-PF16 antibodies (1:1,000, affinity purified). The ECL Prime Kit (GE Healthcare) was used for HRP detection.

Immunofluorescence microscopy was performed as described previously (Zhao et al., 2019). Briefly, the slides with attached intact axonemes were treated overnight at 4°C with blocking buffer containing the diluted primary antibody (mouse anti-acetylated tubulin, Sigma-Aldrich, clone 6-11B-1, 1:1,000). On the next day, the slides were washed four times over 1 h with blocking buffer, and then treated for 1 h with blocking buffer containing the secondary antibody (F(ab′)2-goat anti-mouse IgG (H+L) cross-adsorbed secondary antibody, Alexa Fluor 568, Invitrogen, A11019, 1:1,000) and Streptavidin (Alexa Fluor 488, Invitrogen, A32354, 1:200). The samples were mounted and examined as before (Zhao et al., 2019) using structured illumination microscopy performed on a DeltaVision OMX system (GE Healthcare) with a 1.42 NA 60× Plan-Apochromat objective lens (Olympus) and immersion oil with a refraction index of 1.512. Structured illumination microscopy images were reconstructed with softWoRx 6.1.3 (GE Healthcare). Capture times and adjustments were the same for images with the same antibody. Image brightness and contrast were adjusted using ImageJ (National Institutes of Health).

All observations and recordings except for high-speed video were performed on cells grown in liquid M medium at room temperature. The swimming speed and swimming paths of cells were analyzed as described previously (Zhao et al., 2019). To analyze swimming speed, 50 µl of cell culture were transferred to a plastic chamber (0.127-mm-deep Fisherbrand UriSystem DeciSlide; Thermo Fisher Scientific). Cells were imaged with nonactinic (deep-red) light using a Zeiss inverted microscope equipped with a 16×/0.35 NA Plan objective and a Kopp #2408 red long-pass filter (Kopp Glass). Movies were recorded at 30 images/s with a digital charge-coupled device (CCD) camera (UP-610; UNIQ Vision) and Video Savant 3.0 software (IO Industries). Swimming speeds were determined using ImageJ as previously described (Awata et al., 2015). To record swimming paths, 1-s exposures were acquired using white light and phase-contrast optics on a Zeiss Axioskop 2 Plus microscope equipped with a 20× Plan-NeoFluar 0.5 NA Ph2 objective, a digital CCD camera (AxioCam MRm), and AxioVision 3.1 software (Zeiss).

For high-speed video recording of swimming cells, a PCO 1200HS camera with Camware software (Cooke Corp.) was used (500 frames/s) on an Axioskop microscope (Zeiss), and videos were taken with a 40×/0.65 NA objective lens under phase-contrast mode. Cells were in TAP medium at room temperature. A red filter placed on the light source was removed to induce waveform switching due to photoshock. Videos (saved at 20 frames/s) were created in ImageJ (Schneider et al., 2012), and the videos of different swimming cells were combined in Photoshop to generate Video 3. To evaluate photoshock ability, at least 30 cells were analyzed.

Axonemes of Chlamydomonas cells were purified by the pH shock method as previously described (Witman, 1986; Song et al., 2015). Briefly, cells were cultured in liquid TAP medium, collected by centrifugation (2,200 rpm for 5 min), and washed twice with fresh M-N/5 minimal medium (Iomini et al., 2009). The cell pellet was resuspended in pH shock buffer (10 mM Hepes, 1 mM SrCl₂, 4% sucrose, and 1 mM DTT, pH 7.4), and 0.5 M acetic acid was added to the buffer to reduce the pH to 4.3. After 80 s, 1 M KOH was added to increase the pH to 7.2; the pH shock treatment was performed on ice. After pH shock, 5 mM MgSO₄, 1 mM EGTA, 0.1 mM EDTA, and 100 µl protease inhibitor (Sigma-Aldrich) were added to the solution. The solution was centrifuged (1800 g for 10 min, 4°C) to separate the detached flagella from the cell bodies. To further purify the flagella, the flagella-containing supernatant was centrifuged twice with a 20% sucrose cushion (2,400 g for 10 min, 4°C). After centrifugation, 1% IGEPAL CA-630 (Sigma-Aldrich) was added to the supernatant to demembranate the flagella for 20 min at 4°C. Axonemes were collected by centrifugation (10,000 g for 10 min, 4°C), and the freshly isolated axonemes were resuspended in HMEEK buffer (30 mM Hepes, 25 mM KCl, 5 mM MgSO₄, 0.1 mM EDTA, and 1 mM EGTA, pH 7.2). Axonemal samples were either plunge-frozen for cryo-ET or stored at −80°C for biochemical assays.

For gold labeling of axonemes from pf16;PF16::BCCP and fap76-1;BCCP::FAP76, 5 µl of 80 µg/ml 1.4-nm-sized streptavidin nanogold particles (Nanoprobes) was added to 200 µl of the freshly prepared axonemal solution and incubated for 4 h at 4°C. Meanwhile, 200 µl of samples without added streptavidin gold particles were also prepared as control. The axonemes were washed by adding 800 µl HMEEK, collected by centrifugation (10,000 g for 1 min, 4°C), and resuspended in HMEEK. Axonemal samples treated with and without streptavidin gold were plunge-frozen for cryo-ET analysis. In addition, 40 µg of the axonemes from the fap76-1;BCCP::FAP76 sample (with and without gold particles) were separated on a 4–12% gradient SDS-polyacrylamide gel. The gel was stained with a silver enhancement kit (Nanoprobes) for 45 min at room temperature, and the bands were imaged using a ChemiDoc Touch Imaging System (Bio-Rad). The gel was stained with Coomassie brilliant blue for 2 h and destained until the background was clear.

Axonemal protein (40 µg) of the WT, fap76-2, fap81, and fap216 strains were separated on a 4–12% gradient SDS-polyacrylamide gel. After the proteins had entered the gel 3.0–3.5 cm, the gel was stained with Coomassie brilliant blue for 30 min and destained until the background was clear. Each gel lane was cut into four slices, and each slice was further excised into 1-mm cubic pieces. In-gel trypsin digestion and peptide identification were conducted by the proteomics core facility at the University of Texas Southwestern Medical Center. Proteins that had >10 unique peptides identified in the WT sample were selected for further analyses. Quantification of the mutant/WT ratio for each axonemal protein was estimated by the MIC sin (matched isotope pattern chromatogram spectral index) value (Trudgian et al., 2011). Data for pf16 are from Zhao et al., (2019) (repeat 2, Table S4). Some data for fap76-1 are from Zhao et al. (2019) (repeat 2, Table S6). Intensity-based absolute quantification (IBAQ; Schwanhäusser et al., 2011) and Top3 precursor quantification methods (Top3; Silva et al., 2006) were used to estimate the abundance of each protein.

For MS studies of sucrose gradient fractions from pf16(D2) or pf28;pf16(D2) axonemal extracts, proteins were run 0.5 cm into a 10% polyacrylamide resolving gel and stained with Coomassie brilliant blue. The protein-containing segments were excised and analyzed at the University of Massachusetts Medical School by matrix-assisted laser desorption/ionization time-of-flight tandem MS with postsource decay.

Freshly prepared axonemal samples (30 µl) were gently mixed with 10 µl of 10-fold-concentrated, BSA-coated 10-nm gold solution (Iancu et al., 2006). 4 µl of the solution was applied to a glow-discharged (30 s at −35 mA) copper R2/2 holey carbon grid (Quantifoil Micro Tools). After removing excess liquid by blotting the grid from the back side with Whatman filter paper no. 1 for 2 s, the grid was plunge-frozen into liquid ethane using a homemade plunge-freezer. Grids were then stored in liquid nitrogen until use.

For most strains, grids were mounted in Autogrids (Thermo Fisher Scientific) and loaded into a Titan Krios transmission electron microscope (Thermo Fisher Scientific) operated at 300 kV. The microscope control software SerialEM (Mastronarde, 2005) was used to acquire tilt series images in low-dose mode from −60° to 60° with 1.5°–2.5° increments using a dose-symmetric tilting scheme (Hagen et al., 2017). The images were recorded with a 4,000 × 4,000 K2 direct electron detection camera (Gatan) in counting mode (15 frames, 0.4-s exposure time per frame, dose rate of 8 electrons/pixel/s for each tilt image). The postcolumn energy filter (Gatan) was operated in zero-loss mode (20-eV slit width), and a Volta-Phase-Plate (Danev et al., 2014) was used with −0.5-µm defocus. The magnification was set to 26,000 with an effective pixel size of 5.5 Å. The total electron dose per tilt series was limited to ∼100 e/Ų.

Grids of the pf16 and pf16;PF16::BCCP samples were imaged using a Tecnai F30 transmission electron microscope (Thermo Fisher Scientific) operated at 300 kV. A 2,000 × 2,000 CCD camera (Gatan) was used for recording tilt series images with SerialEM (Mastronarde, 2005) as described above, except neither dose-symmetric tilting nor Volta-Phase-Plate were used, and the defocus was set to –8 µm. The magnification was 13,500, with an effective pixel size of 1.077 nm.

Frame alignment for data collected on the K2 camera, alignment of tilt serial images, and tomogram reconstruction were performed as previously described (Fu et al., 2018). In brief, motion correction of the frames was done with a script extracted from IMOD software (Kremer et al., 1996). IMOD software was also used for aligning the tilt serial images using the 10-nm gold particles as fiducial markers and for reconstructing the tomograms by weighted back-projection. For subtomogram averaging, either CA or DMT repeats were picked from the raw tomograms, and alignment and missing-wedge compensated averaging were performed using PEET software (Nicastro et al., 2006), which is integrated in the IMOD software package. The axoneme and CA orientation (proximal to distal) was determined for each tomogram based on both DMT orientation in the axoneme cross section (e.g., “clockwise” orientation as shown in Fig. 1 H represents “viewed from distal”) and initial averages that included only the CA repeats within each tomogram. After the CA polarity and the same center of the repeat were determined for each tomogram, a second alignment was performed combining the CA repeats from all tomograms in the correct orientation and periodicity register. After global alignment of all CA repeats, the alignment of each individual CA microtubule was refined by local alignment of each microtubule and its associated projections separately, while the other microtubule was masked as described previously (Carbajal-González et al., 2013). Visualization of the 3D structures of the averaged CA and DMT repeats was done with the UCSF Chimera package software (Pettersen et al., 2004). For generating the isosurface rendering images, the same isosurface threshold was applied to the WT and mutant averages. Classification analyses used a principal component analysis clustering method incorporated in the PEET software (Heumann et al., 2011). The number of tomograms and CA repeats, as well as the estimated resolution of the averages (using FSC 0.5 criterion), are summarized for each strain in Table S3.

The 3D averaged structures of the CA from different strains have been deposited in the Electron Microscopy Data Bank under accession codes EMD-20160 (WT), EMD-20161 (fap92), EMD-20162 (fap76-1), EMD-20163 (fap76-2), EMD-20164 (fap81), EMD-20165 (fap216), and EMD-20166 (fap76-1;fap81).

Fig. S1 shows different conformational states of the C2a projection and refined structural details within the C1a-e-c supercomplex. Fig. S2 shows the genotyping analysis by PCR of the fap76-1, fap76-2, fap81, fap92, and fap216 CLiP mutants to confirm disruption of the corresponding genes; and immunoblot analysis of PF16 in pf16 rescue strain and seven of the strains studied. Fig. S3 shows the isosurface renderings of averaged 96-nm axonemal DMT repeats from WT and the CLiP mutants, confirming that the mutants have no DMT-associated defects; the orientation of the C1a-e-c supercomplex relative to specific DMTs is also shown. Fig. S4 depicts insertion sites for the fap76 mutant alleles and shows the phenotype analyses and structural defects of the fap76-2 mutant. Fig. S5 provides classification analyses of the fap81 and fap216 mutant showing that half of the fap81 CA particles contain FAP76 density, and that the C1e density was reduced in the fap216 mutant. Table S1 lists all the proteins that cosedimented with PF16 in our sucrose gradient analysis and that were significantly reduced in pf28;pf16 axonemes as revealed by sucrose gradient and MS analyses. Table S2 provides unique peptide numbers and mutant/WT ratios of additional C1 proteins identified by Zhao et al. (2019), but not by our sucrose gradient analyses. Table S3 summarizes the image processing information (number of tomograms, CA particles, and resolution) for the strains used in this study. Table S4 lists the primers used in this study. Video 1 shows the 3D isosurface rendering of the averaged Chlamydomonas WT CA repeat. Video 2 summarizes the location and structural organization for FAP76, FAP81, FAP92, and FAP216 within the ciliary CA structure. Video 3 is a combination of videos showing swimming cells of Chlamydomonas WT and the mutant strains studied here.

We thank T. Loreng of Dartmouth College for technical assistance and K. Cai and L. Gui of University of Texas Southwestern Medical Center (UTSW) for discussions and critical reading of the manuscript. We thank A. Lemoff and the proteomics core facility at UTSW and M. Dubuke of the MS facility at the University of Massachusetts Medical School for the liquid chromatography-MS/MS analyses.

We are grateful to D. Stoddard for management of the UTSW cryo-electron microscope facility, which is funded in part by a Cancer Prevention and Research Institute of Texas Core Facility Award (RP170644). This study was supported by National Institutes of Health grants R01 GM083122 to D. Nicastro, R35 GM122574 to G.B. Witman, and R01 GM112050 to E.F. Smith, and by Cancer Prevention and Research Institute of Texas grant RR140082 to D. Nicastro.

The authors declare no competing financial interests.

Author contributions: G. Fu and D. Nicastro conceived the study and designed experiments. G. Fu, K. Song, and Z. Shang performed cryo-ET data collection and analysis. L. Zhao, E. Dymek, and Y. Hou analyzed the phenotypes of mutants and generated rescued strains. G. Fu and L. Zhao analyzed the liquid chromatography-MS/MS data. E. Dymek and L. Zhao performed sucrose gradient and Western blot analyses. N. Phan and L. Zhao performed genotyping analysis of the CLiP mutants. G. Fu and D. Nicastro wrote the manuscript, and G. Fu, D. Nicastro, L. Zhao, G.B. Witman, and E.F. Smith edited the manuscript.