Ccdc113/Ccdc96 complex, a novel regulator of ciliary beating that connects radial spoke 3 to dynein g and the nexin link

Ciliary beating requires the coordinated activity of numerous axonemal complexes. The protein composition and role of radial spokes (RS), nexin links (N-DRC) and dyneins (ODAs and IDAs) is well established. However, how information is transmitted from the central apparatus to the RS and across other ciliary structures remains unclear. Here, we identify a complex comprising the evolutionarily conserved proteins Ccdc96 and Ccdc113, positioned parallel to N-DRC and forming a connection between RS3, dynein g, and N-DRC. Although Ccdc96 and Ccdc113 can be transported to cilia independently, their stable docking and function requires the presence of both proteins. Deletion of either CCDC113 or CCDC96 alters cilia beating frequency, amplitude and waveform. We propose that the Ccdc113/Ccdc96 complex transmits signals from RS3 and N-DRC to dynein g and thus regulates its activity and the ciliary beat pattern.


Author summary
Motile cilia and flagella are evolutionarily conserved microtubule-based cell protrusions with the ability to actively beat. Their coordinated movement propels unicellular organisms, sperm cells and zoospores, or drives a flow of the extracellular fluids in multicellular organisms including humans. Lack or dysfunction of motile cilia cause respiratory diseases that can be accompanied by infertility and laterality defects. The emergence of cilia movement requires coordinated activity of the ciliary complexes, including inner and outer dynein arms and radial spokes. The nexin-dynein regulatory complex (N-DRC) is a main node that coordinates the activity of these ciliary complexes and mediates a transduction of the regulatory signals from radial spokes to axonemal dynein motors. Here we described the identification of a new ciliary complex composed of the two proteins, Ccdc96 and Ccdc113 that extends parallel to the N-DRC and connects radial spoke 3 with

Introduction
Motile cilia and flagella are microtubule-based cell protrusions that are indispensable for the development and physiology of eukaryotic organisms from a wide range of evolutionary lineages, including humans. The beating of cilia and flagella is regulated by the precise interplay of a large number of components that self-organize to form a complex machine. At the core of this machine is the axoneme, composed of nine outer microtubule doublets and two central microtubules that together serve both as a skeleton for the cilium and a docking site for numerous ciliary complexes [1][2][3][4][5][6][7][8][9][10][11][12][13][14][15], for review [16]. The main complexes of the outer doublets form a pattern that extends along the entire length of the axoneme with a characteristic 96-nm repeat, called the axonemal unit. In each axonemal unit, there are four outer dynein arms (ODAs) that are generally identical in their protein composition and function, seven inner dynein arms (IDAs) that differ in their architecture and protein subunits, three radial spokes (RSs), and one nexin-dynein regulatory complex (N-DRC). Cryo-electron microscopy analyses of the axoneme have shown that besides the main complexes, the axonemal unit also contains numerous minor structures, mostly of unknown protein composition and function [2,4,10,12,14,15]. The activity of these axonemal complexes has to be strictly coordinated locally within the axonemal unit and globally, both circumferentially and longitudinally along the axoneme, in order for it to be translated into normal ciliary beating [17]. How such coordination is achieved, however, remains largely unclear.
One factor known to play a key role in regulating the activity of ciliary complexes both within the axonemal unit and across the cilium circumference is the N-DRC [17]. This large complex is composed of eleven subunits [18,19] that organize into two well-defined regions: a part named a base plate docked onto the A-tubule of one microtubule doublet and a large linker extending from the base plate to the B-tubule of the adjacent outer doublet. Thus, the N-DRC, besides connecting and coordinating diverse complexes of the 96-nm axonemal unit, also functions as a linker between neighboring microtubule doublets and is fundamental for the integrity of the axoneme [17]. Deletion of different N-DRC subunits causes a range of motility defects, alters flagella waveform and impairs dynein coordination, depending on the targeted subunit [20][21][22][23][24]. Mutations that affect N-DRC subunits can restore some forms of flagella motility in Chlamydomonas central apparatus and radial spoke mutants, but do not rescue cell motility [20,21,25]. This suggests that, although the N-DRC complex can influence the activity of dynein arms, a signal from the central apparatus transmitted by radial spokes to dynein arms [26] is required to generate forces that enable cell swimming. How such a signal is transmitted from the central pair, through the radial spokes and the N-DRC to different inner and outer dynein arms is not fully understood.
Analyses in Chlamydomonas show that N-DRC makes at least 10 inter-and intra-molecular connections with other complexes within the axonemal unit, some of which are known to be involved in the regulation of the cilium beat [17]. However, which protein(s) build the connections mediating interactions between N-DRC and other axonemal complexes is largely unknown. Here, we describe the identification of a new structure that connects the N-DRC with IDA g and the Fap251-containing arch at the base of RS3 in the ciliate Tetrahymena thermophila. We provide evidence that this structure is composed of Ccdc96 and Ccdc113 proteins and is required for normal cilia beating. The existence of this structure further supports the signal transduction hypothesis [26] and sheds new light on the multi-level regulation and coordination of the complex molecular machine that powers the beating of the cilium.

Identification of two candidates to connect the N-DRC with other axonemal complexes in Tetrahymena
To provide novel insights into how the N-DRC contributes to the coordination of axonemal complexes activity, we searched for additional, as-yet unidentified N-DRC components or connectors. First, we used cryo-electron tomography (cryo-ET) followed by subtomogram averaging to obtain three dimensional reconstructions of the axonemal units from Tetrahymena wild-type cilia and established a more detailed 3D model of the N-DRC structure (Fig  1), which revealed several previously unknown connections with neighboring complexes. Compared to the Chlamydomonas N-DRC, for which only one connection to IDA g is known [17] (connection 12; Fig 1D), we observed four additional connections with the inner dynein arms in Tetrahymena, one with IDA c (connection 15; Fig 1D) and three with IDA e (connections 13, 16, and 18; Fig 1D). The Tetrahymena N-DRC also formed four connections with an as-yet uncharacterized complex that extends parallel to the N-DRC over the A-tubule surface (connections 2-4 and 6; Fig 1D). Only one of these connections (connection 4; Fig 1D) was described previously [17]. We identified new connections between the N-DRC and the radial spokes: besides the previously described connection with the base of RS2 (connection 11; Fig  1D) [17], the base plate protrusion of N-DRC connects to the stalk of RS3 (connection 5; Fig  1D). Connections of the N-DRC base with both tubule A (connections 7-9; Fig 1D) and tubule B (connection 10; Fig 1D) were in agreement with previous findings [17]. The N-DRC linker showed only one connection between the proximal lobe and the B-tubule of the neighboring microtubule doublet (connection 1; Fig 1D), whereas connections are present from both the distal and the proximal lobe in Chlamydomonas [17]. In addition to several connections with various IDAs, the distal lobe of the N-DRC also connected to two elongated structures: the first structure extended from the distal lobe to protofilament A5 (connection 14; Fig 1D) and the second structure, which we name as the "96-nm linker" (connection 17; Fig 1D and green dotted line; S1 Fig), ran parallel to the A-tubule from the heavy chain of IDA I1/f to IDA g and IDA d, and contacted the IC/LC of IDA I1/f, the N-DRC, and the MIA-like complex. Thus, this 96-nm linker connects all major components of the 96-nm repeat. Finally, in contrast to Chlamydomonas, where an outer-inner dynein (OID) linker connects the N-DRC to ODAs [17,27], we did not find any connection of N-DRC to the outer dynein arms in Tetrahymena. Thus, our cryo-ET analysis confirmed the structural similarities between the Chlamydomonas and Tetrahymena N-DRC, but it also revealed a number of previously undescribed structural differences and connections with other axonemal components. Ultrastructural differences between these two species may result in subtle differences in the N-DRC hub-activity and signal transduction between the axonemal complexes. Thus, structural differences may translate into differences in Tetrahymena and Chlamydomonas cilia/flagella waveforms and beating.
To identify proteins that connect the N-DRC complex with other axonemal structures, we performed a BioID assay. In the presence of biotin in the culture medium, the expression of a protein with a mutated BirA � ligase attached as a tag results in local (up to 10 nm from BirA � ) protein biotinylation [30]. We engineered Tetrahymena cells expressing Drc1 (TTHERM_013 45750), a subunit of the N-DRC complex, as a C-terminally HA-BirA � -tagged fusion protein  Table), the presence of two previously uncharacterized proteins, Ccdc113 and Ccdc96, among the ciliary proteins that were specifically biotinylated (S2E Fig) in cells expressing Drc1-HA-BirA � (50 peptides in total and 22 unique peptide sequences for Ccdc96, and 49/22 respectively for Ccdc113). In the same sample, we identified 36/17 peptides of Drc1. These data suggest that Ccdc113 and Ccdc96 proteins are positioned close to the N-DRC and, thus, could connect N-DRC with other axonemal complexes.

Lack of either Ccdc113 or Ccdc96 impairs cilia motility
Ccdc113 and Ccdc96 are evolutionarily conserved, coiled-coil domain-containing proteins (S3 Fig). So far, these proteins have only been reported to be present in cells assembling primary cilia [31]. In retinal pigment epithelial cells (RPE1), a CCDC96-GFP fusion protein localizes in centrosomes whereas CCDC113-GFP co-localizes with centriolar satellites, and knockdown of CCDC113 reduces the number of ciliated cells and the length of assembled primary cilia [31]. Survey of a publicly available database (see Material and Methods) revealed that in human and mouse, the CCDC96 and CCDC113 proteins are highly expressed in organs containing cells that assemble motile cilia, such as testis and trachea. Moreover, CCDC96 and CCDC113 are orthologous to the Chlamydomonas flagellar proteins FAP184 and FAP263, respectively [32], and our phylogenetic analyses indicate that homologous proteins are also present in other organisms that assemble only motile cilia (S3 Fig). Thus, a function of Ccdc96 and Ccdc113 in motile cilia as potential N-DRC connectors may be conserved in evolution.
Tetrahymena Ccdc113 (TTHERM_00312810) is a 43 kDa, likely post-translationally modified, protein (Figs 2A and S4A). These data are in agreement with studies in Chlamydomonas suggesting that FAP263 is a phosphoprotein [33]. Regretfully, we were unable to identify the nature of Ccdc113 modification(s). Ccdc96 (TTHERM_00529650) is a larger~94 kDa protein with poorly conserved N-terminal part but highly conserved C-terminal end (S3 Fig). When expressed under the control of a native promoter, Ccdc113-3HA localized along the entire cilium, with the exception of the ciliary tip. A limited amount was also observed near the basal bodies (Fig 2B-2G). Similar ciliary localization was observed when Ccdc96-2V5 was expressed under the control of its native promoter (Fig 2H-2K"). Domain analysis of Ccdc96 revealed that the less-conserved N-terminal fragment, HA-Ccdc96 M1-I431, accumulated within the cell body whereas the highly conserved C-terminal part, HA-Ccdc96 A370-Y794, was sufficient for ciliary localization (S5A- S5D Fig).
Cell proliferation and phagocytosis, two other processes which depend upon correct cilium function, were also slowed down in CCDC113-KO cells (S4F and S4G Fig). Beating of Tetrahymena cilia generates a rotational movement of the daughter cells which contributes to the breakage of a connecting bridge during the final stage of cytokinesis [34]. Consistent with this, the CCDC113-KO mutants divide slower (S4F Fig) and the calculated doubling time was approximately an hour longer than that of wild-type cells (wild-type-2.1 h, CCDC113-KO-3.0 h). The synchronous movement of cilia that surround the oral apparatus directs food particles to the oral cavity where food vacuoles are formed [35]. Addition of India ink to the culture medium allows tracking of food vacuoles and estimation of the rate of phagocytosis [36]. On average, during 10 min, CCDC113-KO cells form only 3.9 vacuoles (n = 300) whereas wild-   ; Fig 2A) restored normal swimming pattern and led to recovery of swimming velocity to~88% of wild-type (265 +/-4 μm/sec, n = 40) (Fig 3C and 3D). The rates of cell proliferation (2.4 h) and phagocytosis (5.7 vacuoles [n = 300] over 10 min) of the rescued cells were only slightly below those observed in wild type (S4F and S4G Fig). Thus, Ccdc113 appears to be required for normal cilia function.
Tetrahymena cilia beat synchronously, with distinguishable power and recovery strokes. Analysis of the ciliary beating pattern using high-speed video recording revealed that a lack of Ccdc113 resulted in an abnormal ciliary waveform and reduced amplitude and beating frequency (Figs 3I-3L and S6 and S1 and S2 Videos). In wild-type cells, the cilium marks an angle of nearly 180 o in a plane perpendicular to the cell surface during the power stroke (in red; Fig 3I) and returns to the initial position moving parallel to the cell surface during the recovery phase (in blue; Fig 3I). In CCDC113-KO cells, the angle marked by a beating cilium was smaller and the cilium frequently leaned toward the posterior end of the cell (Figs 3I, 3K and 3L and S6). Moreover, cilia of wild-type cells are straight during the power stroke, whereas they bend during the recovery stroke. The curvature shifts along the cilium length from its base to the tip as the cilium changes its position while moving parallel to the cell surface (Fig 3I and S1 Video). Cilia lacking Ccdc113 were also straight during the power stroke, but during the recovery stroke, the curvature shift was delayed and still in progress when the cilium came back to the upright position to start the next power stroke (positions 8, 9 and 1; Fig 3I and S2 Video). The ciliary beat pattern in CCDC96-coDel cells was similar to that of CCDC113-KO cells (Figs 3I-3L and S6 and S3 Video), although we observed some alterations in the waveform and amplitude in the subsequent beating cycles of the same cilium. Thus, Ccdc96 and Ccdc113 are required for normal ciliary beating.

Ccdc96 and Ccdc113 connect the base of RS3 to the tail of IDA g and the N-DRC
Traditional TEM analysis of fixed and resin-embedded cilia of CCDC96-coDel and CCDC113-KO mutants failed to reveal apparent alterations of the axonemal structure. Therefore, we used cryo-ET followed by subtomogram averaging to obtain three-dimensional reconstructions of the axonemal units from wild-type (Figs 4A, 4B, 4D, 4E, 4H, 4I, 4L and 4M and 5A, 5B, 5E, 5F, 5I, and 5J) and mutant cilia (Figs 4C, 4F, 4G, 4J, 4K, 4N, and 4O and 5C, 5D, 5G, 5H, 5K, and 5L), with similar resolutions (35 Å (WT), 35 Å (CCDC113-KO) and 34 Å Tetrahymena. Traced cilia were positioned at the middle dorsal region of the cell. (J) Graph representing cilia beating frequency of WT and mutant cells. T-test denotes P value < 0.001. (K) The analyses of the ciliary amplitude. The schematic representation of all recorded consecutive positions of the cilium during the power and recovery stroke. The amplitude was measured as the angle between two most angled positions of the cilium (calculated as the angle of the triangle marked by the intersection points of the cilium with the measuring line and cilium base. (L) Graph representing cilia beating amplitude of WT and mutant cells. T-test denotes P value < 0.001 for CCDC113-KO and DRC3-coDel, and 0.05 for CCDC96-coDel. Numerical data are in S10 Table. https://doi.org/10.1371/journal.pgen.1009388.g003
The interpretation of the cryo-ET data was supported by proteomic data showing that both Ccdc96 and Ccdc113 are either dramatically reduced or entirely missing in cilia assembled by CCDC96-coDel and CCDC113-KO mutants (S2 and S3 Tables). Besides Ccdc96 and Ccdc113 only one protein identified by a significant number of peptides in the wild-type ciliome, Cfap299, was substantially reduced or absent in the mutant ciliomes. However, this approximately 29 kDa protein was not identified among proteins located in the close vicinity of Ccdc96 and Ccdc113 (BioID assay, see below). Thus, most likely, the density missing in CCDC96-coDel and CCDC113-KO mutants is composed of only two proteins, Ccdc96 and Ccdc113.
In both CCDC113-KO and CCDC96-coDel mutants, a part of a Fap251-containing arch-like structure at the base of RS3 [38] was missing ( Fig 5G, 5H, 5K, and 5L). Thus, the most distal part of the arch, which is instead preserved in CCDC113-KO and CCDC96-coDel mutants, is composed of Fap251, whereas the proximal part of the arch, which connects to the N-DRC (NDRC4; Fig 5E, 5F, 5I, and 5J), is likely composed of Ccdc96 and/or Ccdc113. Alternatively, the fragment(s) of Ccdc96/Ccdc113 stabilize the docking of as-yet unidentified subunits of the arch. In agreement with this, the classification and averaging analyses of the axonemal units of Tetrahymena FAP251-KO mutants [38] showed that the entire arch is lost in only 49% of the analyzed axonemal repeats, while in the remaining 51% only part of this structure is missing. Accordingly, Fap251 is present in the ciliomes of CCDC113-KO and CCDC96-coDel mutants (Tables 1, S2 and S3). Thus, Fap251, Ccdc113, and Ccdc96 are likely to be positioned in close proximity to each other but Fap251 docking to the axoneme is Ccdc96 and Ccdc113-independent.
We observed that the Ccdc96/Ccdc113 complex forms in total twelve connections to the neighboring structures in wild-type cilia: six with the protofilaments of the A-tubule (MT1-MT6; Fig 5), four with the N-DRC (NDRC1-NDRC4; Fig 5), two with IDA g (IDAg1-IDAg2; Fig 5), and one connection to the base of RS3 (RS3; Fig 5). The absence of the Ccdc96/ Ccdc113 complex in the axonemes of the two investigated mutants uncovers a residual small density that, supported by protofilament A5 (MT1; Fig 5), bridges the N-DRC and the base of Each tomographic digital section is accompanied by a corresponding segmented copy (bottom panels) with the Ccdc96/Ccdc113 complex colored in orange (and orange arrowheads), the N-DRC in navy blue, the IDA in pink and the radial spokes in cyan. Labels: RS1 to RS3-radial spokes 1 to 3; IDA a, b, c, d, e, g-inner dynein arms a, b, c, d, e, g; IC/LC-intermediate chain/light chain of the inner dynein arm l1/f, fα, fβ-heavy chains α, β of inner dynein l1/f; N-DRC-nexin-dynein regulatory complex; ODA-outer dynein arms. IDAs are named according to [7]. https://doi.org/10.1371/journal.pgen.1009388.g004

PLOS GENETICS
Ccdc113/Ccdc96 complex links RS3, IDA g and N-DRC IDA g (NDRC2 and IDAg1, respectively; Fig 5). The identity of this residual density is not yet known: it could simply be formed by one of the N-DRC proteins or by an as-yet unidentified protein that would directly interacts with Ccdc96/Ccdc113.
The structure and position of the Ccdc96/Ccdc113 complex indicated that it may play a role in the transmission of signals that originated at the central apparatus and through the RS3 are passed to the dynein g and/or the N-DRC. Thus, we propose the Ccdc96/Ccdc113 complex as a key component for the local coordination of the activity of these three ciliary structures.
In Chlamydomonas the N-DRC is connected directly to the head of IDA g [17], and such a connection is formed by DRC3 protein [23]. To investigate if ciliary defects caused by the loss of Ccdc96/Ccdc113 and Drc3 are similar, we engineered a Tetrahymena DRC3-coDel mutant Table shows the numbers (X/Y) of all the identified peptides (X, in a Mascot program, all significant matches) and unique peptide sequences (Y, in a Mascot program, significant sequences). Data in two rows represent data from two independent experiments (complete data are presented in S2-S4 Tables). Note that Ccdc113 is missing in CCDC96-coDel mutants and Ccdc96 is missing in cilia isolated from CCDC113-KO cells and that both Ccdc96 and Ccdc113 are present in cilia of DRC3-coDel mutant. One of the control samples (WT) was common for CCDC96-coDel and DCR3-coDel mutants. The ortholog of Chlamydomonas FBB9 (flagellar/basal body protein 9) and mammalian Cfap299 was also reduced or missing in CCDC113-KO and CCDC96-coDel mutants but this protein was not identified among ciliary proteins co-immunoprecipitating with Ccdc113 or biotinylated in cells expressing Ccdc96-HA-BirA � or Ccdc113-HA-BirA � (see S5-S7 Tables). Radial spoke protein  (Tables 1 and S4). The swimming behavior ( Fig 3H) and ciliary beat in DRC3-coDel mutants were altered and the waveform and amplitude often varied during subsequent cycles of the analyzed cilium (Figs 3I-3L and S6 and S4 Video). Although the pattern of cilium beating of the DRC3-coDel mutant was often similar to that of CCDC113-KO and CCDC96-coDel mutants (Fig 3I, 3K and S6), the DRC3-coDel swimming paths were short but straight (Fig 3H) in contrast to the wavy trajectories of CCDC96 and CCDC113 mutants. Thus, likely both the Ccdc96/Ccdc113 complex and Drc3 contribute to the regulation of dynein g activity and consequently cilia beating in similar but not identical ways.

Axonemal assembly and functional activity of Ccdc96 and Ccdc113 is codependent
To verify if Ccdc96 and Ccdc113 are subunits of the same complex, first we searched for potential partner proteins using co-immunoprecipitation and BioID assays.   Table). Note that in Ccdc96-HA-BirA � expressing cells biotinylated proteins were identified by a significantly higher number of peptides compared to cells expressing Ccdc113-HA-BirA � . This could be due to: (i) the different distance between BirA � and neighboring proteins; (ii) a more or less exposed position of BirA � (the protein end can be hidden between other proteins); (iii) the conformation of the protein end to which the BirA � ligase is fused; we do not know yet where N-and C-termini of Ccdc96 and Ccdc113 are positioned; (iv) the level of the assortment of the transgene enabling expression of the BirA � tagged protein. https://doi.org/10.1371/journal.pgen.1009388.t002 (actin); and v) Fap251. Ccdc96 and actin were also biotinylated (Tables 2 and S5) when BirA � -HA tag was attached to the N-terminal end of Ccdc113 (S4M and S4N Fig).
Mass-spectrometry analyses of the biotinylated ciliary proteins isolated from cells expressing Ccdc96-HA-BirA � under the control of its promoter (S5K Fig) confirmed that Ccdc96 protein is positioned in close proximity to Ccdc113, Fap57 orthologs, actin and N-DRC subunits (Tables 2 and S6). When BirA � ligase was attached to the N-terminal end of Ccdc96, mass spectrometry failed to identify proteins other than Ccdc96 and, surprisingly, Dyh7, the motor protein of IDAf/I1 (S4N Fig and S6 Table), suggesting that a weakly conserved fragment of Ccdc96 could extend towards this two-headed dynein arm.
The putative interactions between Ccdc96 and Ccdc113 were further verified by a pulldown assay. GFP-Ccdc96 or GFP (control) were overexpressed in Tetrahymena cells and purified from the cytoplasmic fraction using anti-GFP conjugated resin. After incubation with the cytosolic fraction obtained from cells overexpressing Ccdc113-HA, the absorbed proteins were analyzed by Western blot. Only the sample containing GFP-Ccdc96 pulls down Ccdc113-HA protein (S8A and S8A' Fig). To summarize, Ccdc113 and Ccdc96 proteins are positioned in close proximity and likely form a novel ciliary complex.
Actin and Ccdc96 are the only proteins identified by high peptide number among proteins that also co-immunoprecipitate with Ccdc113-3HA expressed at the native level (S7 Table and [39] and Tetrahymena ACT1-KO cells swim extremely slowly [40]. The presence of actin indicates that Ccdc113 is likely located near one or more single-headed IDAs (which is in agreement with cryo-ET showing a connection between dynein g and a structure composed of Ccdc96/Ccdc113).
Fap251 was detected by the low number of peptides among proteins that are biotinylated in cilia of cells expressing Ccdc113-HA-BirA � (exp. 2; S5 Table) or Ccdc96-HA-BirA � (exp 2; S6 Table), but not among proteins that co-immunoprecipitated with Ccdc113-3HA (S7 Table). Thus, the biochemical data support the ultrastructural analysis showing that the Ccdc96/ Ccdc113 complex and part of the arch composed of Fap251 are positioned in close proximity but do not interact (or interact weakly) with each other.
Tetrahymena Fap57 orthologs A, B, and C are also biotinylated in cells expressing BirA �tagged subunits of the tether complex Fap43 and Fap44 [41]. The Chlamydomonas fap57 mutant lacks an elongated complex connecting different structures in the axonemal unit [42]. Because Fap57A, B and C proteins were identified near T/TH and Ccdc96/Ccdc113 complexes, we speculate that FAP57 orthologs can form or be a part of the elongated structure connecting all major complexes also in Tetrahymena cilia.
To investigate if Ccdc113 and Ccdc96 are required for ciliary localization of Fap57A, wildtype, CCDC113-KO, and CCDC96-coDel mutant cells were transformed with a construct enabling expression of 2V5-tagged Fap57A under the control of its native promoter. Immunofluorescence analysis showed that Fap57A-2V5 is targeted to cilia in both CCDC113 and CCDC96 knockout mutants as in wild-type cells (Fig 6A-6D). Moreover, Fap57A and Fap57C were found in the total ciliary proteome of both mutants, and the number of identified peptides was similar to the number of peptides in the proteome prepared from wild-type axoneme (Tables 1, S2 and S3). Thus, Fap57A protein and likely Fap57C do not require either Ccdc96 or Ccdc113 for their ciliary localization.
In contrast, Ccdc96-2V5 expressed under the control of its native promoter was undetectable in cilia of CCDC113-KO cells but was present in the cell body (Fig 6E and 6F). Mass-spectrometry analyses of the total cilia proteome confirmed that Ccdc96 was either missing or only present at a very low level in cilia isolated from CCDC113-KO cells (Tables 1 and S2). Expression of Ccdc113-3HA from the non-essential BTU2 locus in CCDC113-KO cells carrying the

PLOS GENETICS
Ccdc113/Ccdc96 complex links RS3, IDA g and N-DRC Ccdc96-2V5 transgene restored ciliary localization of both Ccdc96-2V5 and Ccdc113-3HA (Fig 6G and 6H), and led to restored cell motility. Similarly, a Ccdc113-2V5 fusion protein expressed under the control of the native promoter was not targeted to cilia in CCDC96-coDel cells (Fig 6I-6I3 and blot to the right, middle row; S8C Fig). Moreover, we observed some degradation of Ccdc113-2V5 (arrowhead; S8B Fig and blot to the left; S8C Fig), suggesting that Ccdc96 could stabilize Ccdc113 within the cell body. Accordingly, Ccdc113 was not found in the CCDC96-coDel ciliome (Tables 1 and S3). Expression of HA-Ccdc96 from the BTU1 locus restored ciliary localization of both fusion proteins, HA-Ccdc96 and Ccdc113-2V5 (Fig 6J-6J4 and blot to the right, first row; S8C Fig) and cell motility. Interestingly, expression of an HA-Ccdc96 A370-Y794 fragment but not an N-terminal Ccdc96 truncation was sufficient to restore ciliary localization of Ccdc113-2V5 in CCDC96-coDel mutant (S8D Fig). These data are in agreement with the cryo-ET analyses showing that both Ccdc96 and Ccdc113 are missing in cells with a single gene knockout.
Surprisingly, when Ccdc113-3HA was overexpressed in CCDC96-coDel mutants, it localized in cilia (Fig 6K), but the mutants retained slow motility. Similarly, overexpressed HAtagged Ccdc96 was present in cilia assembled by CCDC113-KO mutants. To determine whether overexpressed proteins present in cilia are stably incorporated into the axonemal structure, cells were permeabilized with 1% Triton-X-100 and then fixed with 2% PFA instead of simultaneous fixation and permeabilization [43]. Under such conditions, the ciliary HApositive signal was diminished in mutants (knockouts of the partner protein). In otherwise wild-type cells expressing the respective tagged protein under the control of the native promoter that were fixed side-by-side, the HA-positive signal was more prominent (S9A-S9B" Fig). The Western blot analyses of the ciliary fractions [44] showed that some of the overexpressed proteins remain bound to the axonemes (S9C- S9G Fig). Thus, in the absence of a partner protein, the other subunit of the Ccdc96/Ccdc113 complex can be effectively transported to cilia, probably docked (S9H Fig) but alone is insufficient to restore cilia motility.
Taken together, our results suggest that (i) Ccdc113 is unstable in CCDC96-coDel mutants (S8B and S8C Fig), but a part of the Ccdc113-3HA "escapes" degradation if Ccdc113 is overexpressed. Thus, under native conditions, either Ccdc96 stabilizes Ccdc113 in the cell body or their expression is co-regulated. (ii) When overexpressed, Ccdc113-3HA protein is present within the cell body in CCDC96-coDel mutants and can be transported to cilia. Thus, (iii) loading of Ccdc113 onto IFT particles and its transport to the cilium is independent of Ccdc96 (and likewise, transport of Ccdc96 to cilia is independent of Ccdc113). Based on genetic, microscopic, and mass-spectrometry studies, we therefore conclude that Ccdc96 and Ccdc113 are most likely transported to cilia independently but their functional activity in cilia requires the presence of both proteins.

Conclusions
Using genetic, biochemical, and cryo-ET analyses, we show that Ccdc113 and Ccdc96 jointly form a structure that bridges the base of RS3 with the tail of IDA g and several parts of the N-DRC and is indispensable for proper cilia beating. Moreover, at the base of RS3, Ccdc96/ Ccdc113 complex likely partly contributes to the formation of the arch-like structure known to contain Fap251. Since radial spokes are thought to transmit signals that come from the central pair towards the microtubule doublet [26], our data suggest that some of these signals are transmitted by RS3 and "received" by a Ccdc96/Ccdc113 complex to be then transferred to either N-DRC (and possibly other complexes) or dynein g (independently or via N-DRC). The additional connection between N-DRC and dynein g preserved in both CCDC96-coDel and CCDC113-KO mutants indicates the existence of more than one link between the axonemal structures and highlights the complexity of the mechanisms that regulate ciliary beating. Intriguingly, Ccdc96/Ccdc113 and N-DRC have at least four contact points. The existence of so many connections between these two complexes further points to the complexity of the system regulating cilia beating. In the future, identification of the amino acid residues that form these links would allow elimination of single connections and thus reveal their specific role in the regulation of cilia beating.
Our evidence also further supports the hypothesis that particular IDAs may contribute differently to the regulation of the ciliary waveform and amplitude, as lack of the Ccdc96/ Ccdc113 complex affects Tetrahymena cilia beating in a different way to that caused by the absence of the tether/tetherhead complex regulating IDAf/I1 or deletion of Dyh6 or Dyh 7, the IDAf/I1 motor domains, [41], or several IDAs [45].
Finally, a growing body of evidence shows that some proteins believed to be exclusively components of motile cilia also play a role in other structures such as the centrosome [46,47]. Consistent with this, our data show that Ccdc96 and Ccdc113, which were originally identified in cells assembling primary cilia as centrosome-associated proteins [31], are structural components of the motile cilia axoneme. In humans, mutations in genes encoding such proteins may lead not only to primary ciliary dyskinesia caused by motile cilia defects but also to more complex multi-symptom disorders.

Cells culture and phenotypic analysis
Wild-type CU428.2 and B2086.2 cells were obtained from the Tetrahymena Stock Center (Cornell University, Ithaca, NY, US). Cells of all types were grown in SPP medium [48]. The cellproliferation and phagocytosis rate were estimated as described previously [38,41]. To measure the length of swimming paths, cells were diluted to a density of 7500 cells/ml and a drop of cells was placed on the bottom of the Petri dish. Swimming cells were recorded at room temperature using a Zeiss Discovery V8 Stereo microscope (Zeiss, Germany) equipped with Zeiss Plans 10× FWD 81 mm objective and Axiocam 506 camera and ZEN2 (blue edition) software. The length of the paths was measured using ImageJ software. Colored lines parallel to the paths were drawn on a new layer using Adobe Photoshop CC software and slightly shifted to show the original paths.
To analyze cilia beating, a 15 μl drop of cells in a culture medium mixed 1:10 with 20% Ficoll in 10 mM Tris-HCl, pH 7.5 was placed on the glass slide between two strips of adhesive tape, covered with a coverslip, and cilia beating was recorded in at least 10 cells using a highspeed camera (Andor Zyla 5.5 sCMOS) mounted on a Leica DMI 6000 microscope (63x oil immersion lens, numerical aperture 1.4) with an Andor DsD2 unit. Video was recorded at 200 frames/s. To establish the position of the cilium during the power and recovery stroke, the subsequent frames were analyzed, and based on these frames, the position of the cilium was determined. The power and recovery stroke were established based on the direction and plane of the cilia movement (perpendicular versus parallel to the cell surface). To estimate the beat frequency, videos in which cilia could be clearly observed were selected and the number of frames showing complete full beat cycle(s) was scored.
To calculate the cilia beat amplitude, high-speed videos were recorded as ".tif" images and cilium position was marked in successive frames using ImageJ and Adobe Photoshop image editors. To measure the amplitude of the cilium at 1.5 μm above the cell surface, lines parallel to the cell surface were marked at the level of cell surface and 1.5 μm above. The ciliary amplitude was expressed as the angle with vertex at the point of the cilium base at the lower measuring line and arms crossing the points where the most external cilium traces intercepted the upper measuring line [49].

Genome modifications
The transgenes enabling expression of the tagged protein in its native locus, protein overexpression, or gene knockout were prepared using Phusion Hot Start II high-fidelity DNA polymerase (Thermo Fisher Scientific Baltics, Lithuania) and genomic DNA purified from CU428.2 strain as a template. All primers are listed in S8 Table. The transgene enabling germ-line knock-out of CCDC113 was constructed using the pNeo4 plasmid [50] and CCDC113 heterokaryons were generated as described previously [51,52]. Mutant strains with deleted CCDC96 or DRC3 were engineered using targeted ectopic DNA elimination in macronuclei (co-Deletion) and the pMcoDel plasmid, which was kindly provided by Dr. K. Mochizuki [37]. To express analyzed proteins under the control of the native promoter in the native locus, fragments of an open reading frame and 3'UTR were removed from pFAP44-3HA, pFAP44-2V5, or pFAP44-HA-BirA � plasmids [41] using MluI and BamHI in the case of the open reading frame and PstI and XhoI in the case of the 3'UTR. The fragments were replaced with fragments of the open reading frame and 3'UTR of the analyzed gene amplified by PCR (see S8 Table). To express Ccdc96 or Ccdc113 proteins with an N-terminal BirA � -HA tag in a native locus, we modified the MTT1-GFP overexpression plasmid by replacing the GFP coding region with the BirA � -HA coding region and 5'UTR of BTU1 by 5'UTR of the analyzed gene and inserting a neo2 cassette to enable selection of the positive transformants, and strengthen the phenotypic assortment.
To obtain a transgene enabling protein tagging and overexpression, the open reading frame of the analyzed gene was amplified by PCR and cloned into an MTT1-HA [53] or an MTT1-GFP [54] plasmid. Protein expression was controlled by the cadmium-inducible MTT1 promoter [55].
For rescue experiments, CCDC113-KO mutant cells were transformed with a transgene enabling expression of a C-terminally 3HA-tagged Ccdc113 under the control of the BTU2 promoter in the BTU2 locus. A transgene was constructed by insertion of the BTU2 promoter just before the CCDC113 open reading frame into a plasmid enabling expression of Ccdc113-3HA from its native locus. Next, the plasmid was digested with EcoRV and ApaI to remove the Neo4 cassette and the 3'UTR of CCDC113, and the fragment was replaced with a pPur4 cassette [56] followed by the 3'UTR of BTU2. To rescue CCDC113-KO cells with a transgene enabling expression of Ccdc113-HA-BirA � , the 3HA coding region in the CCDC113-3HA overexpression plasmid was replaced with the HA-BirA � coding region [41].
To rescue CCDC96-coDel mutants, cells were transformed with a~3 kb fragment amplified by PCR containing the 5'UTR (without ATG) and a coding region (with ATG) of CCDC96 separated by a coding region of the HA tag. Rescued cells were selected based on the restored motility.
After separation from the plasmid backbone (uncut plasmid in the case of the co-Deletion construct), DNA was precipitated onto DNAdel Gold Carrier Particles (Seashell Technology, La Jolla, CA) and Tetrahymena cells were transformed using the Biolistic PDS-1000/He Particle Delivery System (Bio-Rad, US). The transgenes were introduced into macro-or micronuclear genome via homologous recombination [36,51,52]. Transformed cells were selected using an appropriate drug and assorted as described previously [36,41].
For the BioID assay, we engineered Tetrahymena cells expressing one of the following fusion proteins: Ccdc96-HA-BirA � , Ccdc113-HA-BirA � or Drc1-HA-BirA � under the control of the respective native promoter or BirA � -HA-Ccdc96 and BirA � -HA-Ccdc113 under the control of the MTT1 promoter (cells were grown without cadmium to avoid protein overexpression). Cells at a density of 2 × 10 5 cells/ml were starved overnight in 10 mM Tris-HCl buffer, pH 7.5 and incubated in the same buffer supplied with 50 μM biotin for 4 h at 30˚C. After deciliation [58], cilia were resuspended in 0.5 ml of axoneme stabilization buffer (20 mM potassium acetate, 5 mM MgSO 4 , 20 mM HEPES, pH 7.5, 0.5 mM EDTA with protease inhibitors (Complete Ultra EDTA-free; Roche, Indianapolis, IN)) and incubated for 5 min on ice with addition of 0.2% NP-40 to remove the ciliary membrane. After spinning down, the collected axonemes (10 min at 21,100×g at 4˚C) were lysed for 1 hour (0.4% SDS, 50 mM Tris-HCl, pH 7.4, 500 mM NaCl, 1 mM DTT with protease inhibitors) at RT, centrifuged (8000×g at 4˚C) and the obtained supernatant was diluted with three volumes of 50 mM Tris-HCl buffer, pH 7.4 and incubated overnight with 100 μl of streptavidin-coupled Dynabeads (Dynabeads M-280 Streptavidin, Thermo Fisher Scientific, Waltham, MA, USA) at 4˚C. After washing (6 × 5 min with washing buffer: 15 mM Tris-HCl, pH 7.4, 150 mM NaCl, 0.1% SDS, 0.3 mM DTT) at 4˚C, the biotinylated, resin-bound proteins were analyzed by mass spectrometry (Laboratory of Mass Spectrometry, Institute of Biochemistry and Biophysics, PAS, Warsaw, Poland) and by Western blot with Pierce High Sensitivity-streptavidin-HRP (Thermo Scientific, Rockford, IL) diluted 1:40,000 in 3% BSA/TBST. The proteins detected by the high number of peptides are likely to be positioned in close proximity to the BirA � -tagged protein. Note that because the entire biotinylated proteins (not biotinylated peptides) were bound to the resin, the large but poorly biotinylated proteins (such as dynein heavy chains) can be detected by the similar number of peptides as small proteins but moderately biotinylated.
For immunoprecipitation, cilia from the Tetrahymena cells expressing Ccdc113p-3HA at the native level and wild-type cells (control) were harvested as described [58] and suspended in 10 mM Tris-HCl buffer, pH 7.5, with protease inhibitors (Complete Ultra EDTA-free; Roche, Indianapolis, IN), and combined with an equal volume of 2% NP-40 and 1 M NaCl in 80 mM Tris-HCl buffer, pH 7.5. After 20 min incubation on ice, the isolated axonemes were pelleted at 20,000 × g for 20 min and washed with 0.5 M NaCl, 40 mM Tris-HCl, pH 7.5, pelleted again for 30 min and extracted with 0.5 M KI, 30 mM NaCl, 5 mM MgSO 4 , 0.5 M EDTA, 1 mM dithiothreitol, and 10 mM HEPES, pH 7.5, also on ice [44]. After centrifugation at 16,000 × g for 15 min at 4˚C, the supernatant was diluted 250 x with 10 mM Tris-HCl, pH 7.5 to reduce KI concentration, and proteins were concentrated to 1.8 mg/ml by ultrafiltration in Vivaspin columns (Sartorius, Goettingen, Germany).
A concentrated supernatant was incubated with anti-HA-conjugated resin (Pierce HA Epitope Tag Antibody Agarose conjugated, Thermo Scientific, Rockford, IL) with rotation overnight at 4˚C according to the manufacturer's instructions. Resin was washed six times for 5 min with 1 ml of 10 mM Tris-HCl, pH 7.4, 150 mM NaCl, and 0.5 mM EDTA at 4˚C. Proteins were separated on 10% SDS-PAGE gel and silver stained, or were analyzed by mass spectrometry.
For pull-down assays, cells with introduced transgenes enabling overexpression of either Ccdc113-HA or GFP-Ccdc96, or GFP alone (control) were grown for 3 hrs in SPP medium supplemented with 2.5 μg/ml CdCl 2 to induce protein overexpression, washed with Tris-HCl pH 7.5 buffer, centrifuged, and resuspended in TBS buffer (20 mM Tris-HCl, pH 7.5, 300 mM NaCl with protease inhibitors). Next, cells were disrupted using a French press (French Pressure Cell Press Model FA-078, ThermoSpectronic) and centrifuged at 20,000g for 15 min at 4 o C to remove cell debris. The collected supernatant was ultracentrifuged at 100,000 × g for 1 h at 4 o C. Samples of the supernatant collected from GFP or GFP-Ccdc96-expressing cells and containing 1 mg of protein, were incubated for 1 h with anti-GFP-conjugated resin (GFP-Trap-Agasose, ChromoTek, Germany) on a shaker at 4 o C. After washing (5 x 5 min) with TBS buffer (10 mM Tris-HCl, pH 7.5, 150 NaCl, 0.5 mM EDTA with protease inhibitors), resin with bound GFP or GFP-Ccdc96 was incubated with supernatant obtained from Ccdc113-HA-expressing cells (1 mg of proteins) for 1 h at 4 o C with shaking. After washing (as above), precipitated proteins were separated from the resin with a loading buffer and analyzed by Western blot using polyclonal anti-GFP (1:2000, ab6556, Abcam) or monoclonal anti-HA (1:2000) antibodies.
For the entire cilia proteome, cilia were isolated as described [58]. Approximately equal amounts of ciliary proteins (0.2-0.25 mg) from wild-type and mutant cells were run on an SDS-PAGE gel and analyzed by mass spectrometry (Laboratory of Mass Spectrometry, Institute of Biochemistry and Biophysics, PAS, Warsaw, Poland).

Preparation of cryo-TEM grids
Isolated cilia in solution were applied on glow-discharged TEM grids with holey carbon support film (Quantifoil Micro Tools, R 3.5/1). Colloidal gold particles (10 nm) were added to the sample to be used as fiducial markers during tomogram reconstruction. The samples were blotted from the back with Whatman filter paper, rapidly frozen in liquid ethane at −182˚C using a Leica EM Grid Plunger, and stored in liquid nitrogen until image acquisition.

Cryo-electron tomography
Data were acquired with an FEI Titan Halo transmission electron microscope operated at 300 kV equipped with a field emission gun (FEG), a Gatan energy filter using a slit width of 20 eV, and a Gatan K2 direct detector. SerialEM software was used for the automated tomographic tilt series acquisition [60]. Full grid montages at low magnification were acquired to find suitable cilia connected to the cell body. The nominal image magnification was ×30,000, resulting in a calibrated pixel size of 2.36 Å in the super-resolution mode of the camera. Tilt series were recorded with 2˚increments with a bidirectional tilt scheme from -20˚to 64˚and -64˚(when possible). The defocus range was -4 to -5 μm, and the cumulative dose was 130-150 e per Å2 per tomogram. Images were acquired in the dose fractionation mode with frame times between 0.10 and 0.25 s.

Image processing and subtomogram averaging
Frames were aligned using the K2Align program (IMOD v.4.9.3), and tomogram reconstruction was performed using Etomo (IMOD v.4.9.3) [61] with weighted back projection. Contrast transfer function curves were estimated with CTFPLOTTER and corrected by phase-flipping with the software CTFPHASEFLIP, both implemented in IMOD [62]. The tomograms were binned 3 times, resulting in a pixel size of 0.7 nm. For particle picking and visualization of tomograms, a nonlinear anisotropic diffusion filter by IMOD [61] was applied to enhance the contrast of macromolecular structures.
Subtomogram averaging (S9 Table) was performed on the unfiltered tomograms with PEET v.1.11.0 from the IMOD package. Particles were picked exclusively from the round axonemes and from all the doublets in order to minimize the influence of the missing wedge in 96-nm distances. The center of the particles was on the base of radial spoke 3 [63].
Visualization of tomograms and average densities was performed in 3dmod from IMOD, and rendering of isosurfaces was performed using UCSF Chimera [64].
Confidence maps were generated with the false discovery rate control method according to reference [65] in order to validate our averaged maps and the identified connections between several protein complexes of the 96-nm repeat. The resolution of the three final 3D model (wild type = 35 Å, CCDCKO96 = 34 Å, CCDCKO113 = 35 Å) was estimated with Fourier Shell Correlation (FSC) and cutoff at 0.143 (S7 Fig).

Phylogenetic analyses
The Ccdc96 and Ccdc113 homologs were obtained from the NCBI protein database using Blastp search and either human or Tetrahymena proteins as bait. Protein amino acid sequences were aligned using ClustalX2 software [66] and edited using SeaView [67]. The identical and similar amino acid residues were shaded using GeneDoc [68]. The phylogenetic tree was calculated (www.phylogeny.fr) [69][70][71][72][73][74] and tree was drawn using iTOL (https://itol.embl.de) [75]. The expression profiles of CCDC96 and CCDC113 were checked in GEO Profiles (https://www.ncbi.nlm.nih.gov/geoprofiles) [76]. Ccdc96 and Ccdc113 homologs were obtained from the NCBI protein database using Blastp search and either human or Tetrahymena proteins as bait. Protein amino acid sequences were aligned using ClustalX2 software [66] and edited using SeaView [67]. The identical and similar amino acid residues were shaded using GeneDoc [68]. The phylogenetic tree was calculated (www.phylogeny.fr) [69][70][71][72][73][74] and the tree was drawn using iTOL (https:// itol.embl.de) [75]. The branch support values are shown as %. The coiled-coil domains (blue bars) were predicted using SMART (http://smart.embl-heidelberg.de/) [77] and COILS (https://embnet.vital-it.ch/software/COILS_form.html) [78]. Staining of CCDC96-coDel cell with anti-α-tubulin antibodies (12G10) revealed that mutant cells assemble cilia of a similar length (6.2 μm +/-0.54 (standard deviation), n = 50) to WT cells (6.36 μm +/-0.61, n = 60). (H) Deletion of CCDC96 reduces proliferation rate. (I) Graphical representation of the efficiency of the formation of food vacuoles. On average (data from three independent experiments), WT cells formed 5.9 vacuoles (n = 300 cells), CCDC96-coDel mutants 4.8 vacuoles (n = 300) and CCDC96-coDel rescued cells 6.2 vacuoles (n = 300, not shown). (J) PCR analysis of the DRC3 locus with primers annealing about 1kb upstream and 1 kb downstream of the gene fragment inserted into pMcoDel plasmid. Note that the PCR fragment amplified using WT genomic DNA as a template is significantly larger than PCR fragments amplified using genomic DNA isolated from mutants, indicating deletion of the fragment of the DRC3 gene. (K) Detection of the biotinylated proteins in cilia isolated from either WT cells (line to the left) or cells expressing Ccdc96-HA-BirA � at native level: input, unbound and beads bound fractions. Numerical data are in S10 Table. (TIF) (C, C') Note slightly lower level of the proteins in samples obtained from knockout cells overexpressing partner protein (lines 6 and 8) compared to the wild-type cells expressing the same partner protein (lines 5 and 7) suggesting less effective binding of the protein to the axonemes. (D-G) Western blot analyses of total ciliary proteins (T) and ciliary proteins present in the cilia fractions (supernatant (SN) and pellet (P)) obtained after 1% NP-40 treatment, extraction of the axonemes from NP-40 pellet with the 0.6 M NaCl and extraction of the axonemes collected after NaCl treatment, with 0.5M KI (according [44]). Proteins were visualized using Ponceau S and HA-tagged proteins were detected using anti-HA antibodies. (H) A scheme of the proposed model. In wild-type cells, both Ccdc96 and Ccdc113 are stable in the cell body, transported to cilia and docked to the axoneme. In CCDC96-coDel and CCDC113-KO mutants, a partner protein is degraded within the cell body and thus not transported to cilia. When partner proteins are overexpressed in the mutant background, some of them "escape" degradation and can be transported to cilia and docked to the axoneme, although it remains unknown (?) if the strength of the binding is the same as in wild-type cells where both proteins are present. (TIF) S1 Table. Mass-spectrometry analysis of proteins biotinylated in cells expressing Drc1-HA-BirA � . Proteins present in both control and experimental samples are highlighted in green. (XLSX) S2 Table. Comparison of the protein composition of cilia isolated from wild-type (WT) and CCDC113-KO cells using mass spectrometry. About 0.2 mg of ciliary proteins was analyzed in WT and CCDC113-KO, sample 1, and about 0.1 mg of proteins in the case of CCDC113-KO, sample 2. Proteins present in both control and experimental samples are highlighted in green (exp1) or blue (exp2). In previously published cilia and flagella proteomes [32,[79][80][81] the number of the identified proteins was e.g. 1134 proteins (482 identified by 1 peptide) in Chlamydomonas flagella [32] or 751 proteins in Trypanosoma flagella [81]. A higher number of the identified ciliary protein candidates in Tetrahymena ciliome could be due to: (i) differences in the cilia preparation method and analysis of whole cilia versus ciliary fractions, (ii) initial amount of ciliary proteins that were analyzed by mass spectrometry, (iii) presence of more than one ortholog of some ciliary proteins (e.g ciliary dyneins) in Tetrahymena; (iv) presence of ciliate-specific proteins as we determined by BlastP search using NCBI database. A similar number of putative ciliary proteins was obtained in our previous studies [41]. Paulina Urbanska for her technical help during the initial analyses of CCDC113-KO cell motility, Dr. Jacek Gaertig for providing a plasmid containing the coding region of Tetrahymena codon-optimized BirA � , Dr. Tokuko Haraguchi for providing plasmids with pPur4 cassette, and Dr. Kazufumi Mochizuki for providing the pMcoDel vector. Mass-spectrometry analyses were performed in the Mass Spectrometry Laboratory, Institute of Biochemistry and Biophysics, PAS, Warsaw, Poland (The equipment used was sponsored in part by the Centre for Preclinical Research and Technology (CePT), a project co-sponsored by European Regional Development Fund and Innovative Economy, The National Cohesion Strategy of Poland). The monoclonal anti-α-tubulin 12G10 antibody, developed by J. Frankel and E. M. Nelsen, was obtained from the Developmental Studies Hybridoma Bank developed under the auspices of the NICHD and maintained by The University of Iowa, Department of Biology, Iowa City, IA 52242. Cryo-EM data acquisition was performed at the Electron Microscopy Facility of the Max Planck Institute of Molecular Cell Biology and Genetics in Dresden, Germany. We thank I.K. Patten for fruitful discussions, comments and corrections to the manuscript. Both research groups contributed equally to this paper.