Scaffold subunits support associated subunit assembly in the Chlamydomonas ciliary nexin–dynein regulatory complex

Significance

Cilia and flagella are small hair-like appendages in eukaryotic cells that play essential roles in cell sensing, signaling, and motility. The highly conserved nexin–dynein regulatory complex (N-DRC) is one of the key regulators for ciliary motility. At least 11 proteins (DRC1–11) have been assigned to the N-DRC, but their precise arrangement within the large N-DRC structure is not yet known. Here, using cryo-electron tomography combined with genetic approaches, we have localized DRC7, the subcomplex DRC8/DRC11, the N terminus of DRC4, and the C terminus of DRC5. Our results provide insights into N-DRC structure, its function in the regulation of dynein activity, and the mechanism by which n-drc mutations can lead to defects in ciliary motility that cause disease.

Abstract

The nexin–dynein regulatory complex (N-DRC) in motile cilia and flagella functions as a linker between neighboring doublet microtubules, acts to stabilize the axonemal core structure, and serves as a central hub for the regulation of ciliary motility. Although the N-DRC has been studied extensively using genetic, biochemical, and structural approaches, the precise arrangement of the 11 (or more) N-DRC subunits remains unknown. Here, using cryo-electron tomography, we have compared the structure of Chlamydomonas wild-type flagella to that of strains with specific DRC subunit deletions or rescued strains with tagged DRC subunits. Our results show that DRC7 is a central linker subunit that helps connect the N-DRC to the outer dynein arms. DRC11 is required for the assembly of DRC8, and DRC8/11 form a subcomplex in the proximal lobe of the linker domain that is required to form stable contacts to the neighboring B-tubule. Gold labeling of tagged subunits determines the precise locations of the previously ambiguous N terminus of DRC4 and C terminus of DRC5. DRC4 is now shown to contribute to the core scaffold of the N-DRC. Our results reveal the overall architecture of N-DRC, with the 3 subunits DRC1/2/4 forming a core complex that serves as the scaffold for the assembly of the “functional subunits,” namely DRC3/5–8/11. These findings shed light on N-DRC assembly and its role in regulating flagellar beating.

Cilia and flagella are dynamic microtubule (MT)-based organelles that emanate from the surface of many eukaryotic cells and are involved in sensory functions, motility, and signaling. Defects in cilia assembly or function have been associated with multiple human disorders collectively known as ciliopathies, such as polycystic kidney disease, Bardet–Biedl syndrome, infertility, hydrocephalus, and primary ciliary dyskinesia (1, 2).

The MT-based axoneme forms the core structure of motile cilia and is highly conserved, from the green algae Chlamydomonas reinhardtii to differentiated cells in the human body. The “9 + 2” axoneme is composed of 9 outer doublet MTs (DMTs) and a central-pair complex (CPC) composed of 2 singlet MTs and associated projections (Fig. 1A). Each DMT is built from many copies of a 96-nm-long unit that repeats along the length of the axoneme and consists of an A-tubule and B-tubule. The outer and inner dynein arms (ODAs and IDAs, respectively) are arranged in 2 distinct rows along the length of the A-tubule of each DMT and function as the motors that drive ciliary motility. The dynein motors “walk” on the B-tubule of the neighboring DMT and thereby generate sliding forces between the doublets (3). The nexin link between DMTs is thought to restrict this interdoublet sliding and thus transform sliding into axonemal bending (4–6). Our previous ultrastructural study demonstrated that the nexin link is an integral part of the dynein regulatory complex, referred to as the nexin–dynein regulatory complex (N-DRC) (7). Formation of ciliary bending in different directions and beating patterns require precise regulation and coordination of the activities of thousands of axonemal dyneins (8–10). Genetic, biochemical, and structural studies suggest that several complexes are involved in the transduction of signals that ultimately control the activity of the dyneins. These regulatory complexes include the CPC, the radial spokes, the calmodulin- and radial spoke-associated complex (CSC), the I1 inner dynein (or dynein f) on the proximal end of the 96-nm axonemal repeat, and the N-DRC on the distal end of the repeat (11).

Fig. 1.

Structure of the N-DRC and organization of known N-DRC subunits in Chlamydomonas flagella. (A–C) Schematic of a Chlamydomonas flagellum in cross-sectional view (A, Bottom), seen from the flagellar tip toward the base. The outer and inner dynein arms (ODA, IDA) and the nexin–dynein regulatory complex (N-DRC) connect the neighboring doublet microtubules (DMTs), whereas the radial spokes (RSs) connect to the central-pair complex (CPC). (A, Top, B, and C) Isosurface rendering of the 3D structure of the 96-nm axonemal repeat in Chlamydomonas WT flagella reconstructed by cryo-electron tomography (cryo-ET), shown in cross-sectional (A, Top), longitudinal front (B), and bottom views (C). (A, Top and B) Overviews showing the location of the N-DRC (blue) within the axonemal repeat, and connections between the N-DRC and other axonemal structures, including the outer dynein/inner dynein (OID) linkers (red arrowheads). The N-DRC linker region has 3 protrusions: the L1 and L2 protrusions, and the OID linker. (C) Three-dimensional maps of the N-DRC in longitudinal views from the bottom. The known locations of DRC1–4 (1–4) are colored; N and C correspond to the N and C termini of the proteins, respectively. The dashed line indicates ambiguity in localization of the N-terminal region of DRC4. Note that DRC5 and DRC6 are thought to be located at the L2 projection, which is in the dorsal side of the linker and thus cannot be visualized in C. Other labels: A_t, A-tubule (protofilaments 1 to 4 are labeled in A, Top, and A4 in C); B_t, B-tubule; BP, base plate; dL, distal lobe; I1α/I1β/ICLC, α- and β-heavy chain, and intermediate/light-chain complex of I1 dynein; IJ, inner junction; a–e and g, inner dynein arm isoforms; pL, proximal lobe.

The N-DRC is a large 1.5-MDa structure consisting of 2 main domains, the base plate and the linker (Fig. 1C). The base plate attaches to the DMT, beginning at the inner junction between the A- and B-tubule and ending at protofilament A4. The linker domain extends across the interdoublet space to the neighboring B-tubule (7). Genetic and biochemical studies have demonstrated that the N-DRC is composed of at least 11 subunits, but the precise arrangement of these subunits in the N-DRC structure remains elusive (12, 13). Recent work with tagged subunits has revealed the approximate locations of DRC1–6. The C termini of DRC1, DRC2(FAP250), and DRC4 are located close to each other at the end of the base plate near the inner junction, and their N termini extend through the linker toward the neighboring B-tubule. DRC3(FAP134), DRC5(FAP155), and DRC6(FAP169) are located in the linker region (7, 14–16). However, little is yet known about the locations of DRC7-11 or the specific interactions within the structure of the N-DRC. The identification of DRC7-11 mutants will be important for the dissection of the N-DRC and its function as regulatory hub.

Here, we have integrated genetic and biochemical approaches with cryo-electron tomography (cryo-ET) to localize DRC7 (FAP50) and the subcomplex DRC8/11 (FAP200/FAP82) within the N-DRC. DRC7 is found in the central region of the N-DRC linker, including the outer dynein/inner dynein (OID) linker between the N-DRC and the row of outer dynein arms (Fig. 1B). The drc11 mutant lacks 2 N-DRC subunits: DRC11 and DRC8, which localize to the proximal lobe of the N-DRC linker domain. We also use cryo-ET of SNAP-tagged DRCs to precisely locate the C terminus of DRC5 in the middle region of the linker, and the N terminus of DRC4 to the proximal lobe of the linker domain.

Results

Identification of drc Mutants for DRC7 and DRC11.

The N-DRC contains at least 11 DRC subunits with distinct functional domains, but drc mutations have only been characterized in 5 genes (drc1-drc5) in Chlamydomonas. To extend the repertoire of drc mutants, we analyzed a collection of mutants generated by insertional mutagenesis the Chlamydomonas Library Project (CLiP) (17, 18) (SI Appendix, SI Materials and Methods). PCR confirmed the sites of plasmid insertion in 2 drc7 strains and 2 drc11 strains; however, the impact of plasmid insertion was highly variable (SI Appendix, Table S1 and Fig. S1).

None of the drc7 strains examined displayed an obvious motility defect by phase contrast light microscopy (Fig. 2B and SI Appendix, Table S1), and so axonemes were isolated from 2 strains and analyzed on Western blots. No DRC7 band was detected in the 197909 strain (from now on referred to as drc7) (Fig. 2C), whereas strain 080526 showed 2 DRC7-related bands (SI Appendix, Fig. S1D) that are likely the result of alternative splicing. To determine whether the absence of DRC7 affected the assembly of other axonemal proteins, the drc7 axonemes were labeled with isobaric tags for relative and absolute quantitation (iTRAQ) and analyzed by tandem mass spectrometry (MS/MS). Between about 500 and 650 proteins were identified by at least 5 peptides in 2 independent iTRAQ experiments. However, only one protein, DRC7, was significantly reduced (P < 0.05) below 50% in both experiments (SI Appendix, Table S2). All of the other DRC subunits were present at wild-type (WT) levels (SI Appendix, Table S2), as confirmed by Western blots (Fig. 2C).

Fig. 2.

Characterization of recently isolated mutations in DRC7 and DRC11, and their impact on flagellar motility and assembly of the N-DRC. (A) The DRC7 and DRC11 protein sequences are drawn to scale, with the location of predicted polypeptide domains indicated: AAA, ATPase associated with diverse cellular activities domain; CC, coiled-coil domain; IQ, calmodulin binding domain; Kelch, Kelch domain; LRR, leucine-rich repeat; SNAP, C-terminal SNAP tag; TGL, transglutaminase-like domain. Also shown are the approximate locations of the plasmid insertion sites in the drc7 and drc11 mutants. (B) The forward swimming velocities of drc7 and drc11 as measured by phase contrast microscopy are shown relative to the background strain (cw15). No significant difference was detected between drc7 and cw15. The drc11 strain was slightly slower than cw15, and transformation with a tagged WT gene, DRC11-SNAP, increased swimming velocities to near cw15 levels. *P < 0.05; ***P < 0.001; n.s., not significant (P > 0.05). (C and D) Western blots of axonemes isolated from the “WT” background (cw15), the drc mutants, and the DRC11-SNAP rescued strain were probed with antibodies against several DRC subunits. Note the presence of a band detected by both the DRC11 and SNAP antibodies that migrated at the size predicted for a SNAP-tagged DRC11 subunit in D. Note that the DRC11 blot in D was reprobed with the DRC1 antibody, so that the blot shown for DRC1 (immediately below DRC11) shows not only the DRC1 bands in all lanes, but also the DRC11 (for cw15) and the DRC11-SNAP (for the rescue) bands. Antibodies against the IC2 subunit of the outer dynein arm served as a loading control, and antibodies against AOX1 served as a marker for cell body contamination.

Although both drc11 candidates had confirmed plasmid insertions in introns (SI Appendix, Fig. S1 E–G), they displayed different motility phenotypes. The 130937 strain displayed an obvious slow swimming phenotype (SI Appendix, Fig. S2B) and failed to assemble both DRC11 and DRC8 (SI Appendix, Fig. S2C). However, transformation with a BAC clone (39g16) containing the DRC11 gene failed to rescue the motility defect (0 rescues out of 538 transformants). This observation suggested the possibility of a second motility mutation in this strain. Further analysis by iTRAQ labeling, MS/MS, Western blots, and cryo-ET revealed that the drc11-937 strain contained an unmapped mutation in a gene that disrupted the assembly of the outer dynein arms (SI Appendix, Fig. S2 C–I and Table S3).

The second drc11 strain, 068819, displayed a slight but significant decrease in its forward swimming velocity (Fig. 2B). Western blots of axonemes confirmed that both DRC11 and DRC8 were missing, but several other DRC subunits and the outer dynein arms (as detected by an IC2/IC69 antibody) were present at WT levels (Fig. 2D). To determine whether the loss of DRC8 and DRC11 affected the assembly of other polypeptides, the 068819 axonemes were analyzed by iTRAQ labeling and MS/MS. Between about 530 and 630 proteins were identified by at least 5 peptides in 2 independent experiments. However, only 3 proteins were significantly reduced below 30% in both iTRAQ experiments (SI Appendix, Table S4): DRC8, DRC11, and Cre12.g513650. Little is known about Cre12.g513650; it is a small protein (146 amino acids) found only in green alga.

To verify that the defects observed in the 068819 strain were caused by the drc11 mutation, we transformed the mutant with a WT DRC11 cDNA containing a C-terminal SNAP tag (Fig. 2A). A SNAP-positive strain was recovered that increased forward swimming velocities to near WT levels (Fig. 2B) and restored the assembly of both DRC11 and DRC8, as shown by Western blots of axonemes (Fig. 2D). Furthermore, the DRC11-C-SNAP polypeptide migrated at the size expected for the SNAP-tagged protein (∼115 kDa). These results demonstrate that the 068819 strain is a bona fide drc11 mutation and that DRC8 depends on the presence of DRC11 for its assembly into the axoneme.

DRC7 Localizes to the N-DRC Linker Including the OID Linker and Distal Lobe.

Three-dimensional reconstruction of the axoneme by cryo-ET and subtomogram averaging revealed that the N-DRC linker begins approximately at protofilament A4 and projects into space between neighboring doublets (Fig. 1 A and C) (7, 19). The linker domain contains several protrusions: L1 that connects to IDA g, L2 at the branch point between the proximal and distal lobes, and the OID linker that connects the N-DRC to the ODAs (Fig. 1A). The proximal lobe is at least twice as the size of the distal lobe, but they both connect to the neighboring B-tubule (Fig. 1C).

To determine the location of DRC7 subunit, we compared the 3D structure of WT and drc7 axonemes by cryo-ET and subtomogram averaging (Fig. 3). We found missing densities only within the N-DRC linker domain and not in the base plate. Specifically, the drc7 N-DRC lacks the OID linker, a density on the distal side of the linker, and the most distal portion of the distal lobe (Fig. 3 D–F and J–L and Movie S1), which reduces the size of the distal lobe interface with the neighboring B-tubule. The total mass of the densities missing in the drc7 N-DRC is estimated to be ∼200 kDa, suggesting that the density contains a single copy of DRC7, which has a molecular weight of 177 kDa.

Fig. 3.

Comparison between WT and drc7 axonemes reveals the location of DRC7 near the OID linker and in part of the distal lobe of the N-DRC. (A–F) Tomographic slices of the averaged 96-nm-long repeats from Chlamydomonas WT (A–C) and drc7 axonemes (D–F) viewed in cross-sectional (A and D) and longitudinal (B, C, E, and F) orientations. The white lines indicate the locations of the slices in the respective panels. Cryo-ET densities corresponding to DRC7 in the nexin–dynein regulatory complex (N-DRC) of WT (red arrowheads in A–C) were absent from drc7 axonemes (white arrowheads in D–F). (G–L) Isosurface renderings show the 3D structures of the averaged axonemal repeats of WT (G–I) and drc7 (J–L) in cross-sectional (G and J), longitudinal (H and K), and enlarged longitudinal bottom views (I and L; looking from the axoneme center outward). The structural difference between the WT and drc7 is manually colored red in G–I and includes the OID linker, a portion of the distal lobe (dL), and the connection between the distal lobe and the L1 protrusion (named connection 7 in ref. 7). Other labels: a–e and g, inner dynein arm isoforms; A_t, A-tubule; B_t, B-tubule; I1α/I1β/ICLC, α- and β-heavy chain, and intermediate/light-chain complex of I1 dynein; IDA, inner dynein arm; ODA, outer dynein arm; pL, proximal lobe; RS, radial spoke. (Scale bar: 10 nm [in E, valid also for A, B, and D]; 10 nm [in F, valid also for C].)

DRC8/11 Localize to the Proximal Lobe of the N-DRC Linker.

Our biochemical assays have indicated that the drc11 strain lacks 2 polypeptides, DRC8 and DRC11. To determine the location of the DRC8/11 subcomplex, we analyzed drc11 axonemes by cryo-ET and subtomogram averaging (Fig. 4). Compared with wild type, the averaged 96-nm repeats of drc11 axonemes showed deficiencies in the proximal lobe of the N-DRC linker region (Fig. 4 D–F and J–L and Movie S2). The structural defect greatly reduces the interface between the proximal lobe and the neighboring B-tubule. The missing density is estimated to be ∼250 kDa. The morphology of the other N-DRC domains, including the distal lobe and the base plate, resembled the WT structure. Although we found that strain drc11-937 was a double mutant with additional ODA defects, we were still able to detect the same N-DRC structural defects in this double mutant (SI Appendix, Fig. S2 C–I) as described for drc11.

Fig. 4.

Three-dimensional localization of the DRC8/11 subcomplex at the proximal lobe of the N-DRC linker using cryo-ET. (A–F) Tomographic slices of the averaged 96-nm repeats of Chlamydomonas WT (A–C) and drc11 axonemes (D–F) viewed in cross-sectional (A and D) and longitudinal (B, C, E, and F) orientations. The white lines indicate the locations of the slices in the respective panels. Cryo-ET densities corresponding to DRC8/11 in the nexin–dynein regulatory complex (N-DRC) of WT (orange arrowheads in A–C) were missing from the drc11 axonemes (white arrowheads in D–F). (G–L) Isosurface renderings show the 3D structures of the averaged axonemal repeats of WT (G–I) and drc11 (J–L) in cross-sectional (G and J), longitudinal (H and K), and enlarged longitudinal bottom views (I and L; looking from the axoneme center outward). The structural difference between the WT and drc11 is manually colored orange in G–I and includes a major portion of the proximal lobe (pL). Other labels: a–e and g, inner dynein arm isoforms; A_t, A-tubule; B_t, B-tubule; dL, distal lobe; I1α/I1β/ICLC, α- and β-heavy chain, and intermediate/light-chain complex of I1 dynein; IDA, inner dynein arm; ODA, outer dynein arm; RS, radial spoke. (Scale bar: 10 nm [in E, valid also for A, B, and D], 10 nm [in F, valid also for C].)

The tip region (close to the neighboring DMT) and the upper part of the proximal lobe (facing the ODA) were completely missing in drc11 axonemes. In addition, we observed a remaining, yet weakened and blurred cryo-ET density that appears to be the lower portion of the proximal lobe that faces the IDAs (Fig. 5 B and G). This blurring effect could be due to reduced occupancy or positional flexibility of the structures that comprise this region. Therefore, we applied automatic image classification analyses with different masks that focused the analyses on specific structures of interest (20). Classification of WT repeats showed one structurally homogenous state across all of the axonemal repeats (Fig. 5 A and F). In contrast, we found that the lower proximal lobe structure in drc11 fell into 4 conformational classes, which differed mainly in its position (“height”) between the ODA and IDA rows (Fig. 5 and Movie S3). More specifically, 23% of the repeats showed the remaining proximal lobe density in a low position close to the IDAs (class 1, Fig. 5 C and H); 25% were positioned in the middle (class 2, Fig. 5 D and I); and 27% were located in a higher position closer to the ODAs (class 3, Fig. 5 E and J). The remaining 25% could not be categorized because of blurred densities and anisotropic density distribution caused by the missing wedge-artifact typical for raw single-axis tomographic reconstructions (21). Thus, the proximal lobe of the N-DRC is composed of 2 parts: The DRC8/11 subcomplex comprises the tip and upper (ODA-facing) region of the proximal lobe, whereas the lower (IDA-facing) part of the proximal lobe likely contains the N-terminal regions of the “backbone subunits” DRC2 and DRC4.

Fig. 5.

Classification analyses reveal that the DRC8/11 subcomplex is important for the stability/rigidity of the proximal lobe scaffold (DRC2/4). (A–J) Tomographic slices show the averaged axonemal repeats from WT (A and F) and drc11 (B–E and G–J) in cross-sectional (A–E) and longitudinal front (F–J) views. The first 2 columns show averages of all repeats (100%), whereas columns 3 to 5 show 3 classes of the drc11 nexin–dynein regulatory complex (N-DRC) that vary in the positions of the proximal lobe (arrowheads). Note the relatively weak and blurry cryo-ET densities of the proximal lobe of the N-DRC in the average-all of drc11 (light orange arrowheads in B and G). Classification analysis of this weak density in drc11 revealed different positions occupied by the proximal lobe (percentages of repeats are indicated for each class); blue dashed lines at the “middle height” serve as reference for the cross-sectional (C–E) and longitudinal front (H–J) views. Other labels: A_t, A-tubule; B_t, B-tubule; IDA, inner dynein arm; ODA, outer dynein arm. (Scale bar: 10 nm.)

Rescue of drc4 and drc5 Mutants with SNAP-Tagged Constructs Clarifies the Location of DRC4 and DRC5 in the N-DRC Linker.

Previous studies focusing on DRC subunits DRC1–6 showed that DRC1, DRC2, and DRC4 are coiled-coil proteins that extend from the inner junction between the A- and B-tubule through the N-DRC base plate and linker to the neighboring DMT (7, 14, 16) (Fig. 1C). Mutations in these proteins disrupt the assembly of multiple DRC subunits (12, 13, 22–25). In contrast, DRC3, DRC5, and DRC6 are located in the N-DRC linker region, and defects in DRC3, DRC5, or DRC6 have only small effects on the assembly of other DRC subunits or the overall structure of the N-DRC (7, 12, 13, 15, 16). Rescues of various drc mutants with tagged WT DRC genes allowed the localization of the N- and/or C-termini of DRC1 (N/C), DRC2 (N/C), DRC3 (N/C), DRC4 (C), and DRC5 (C) (Fig. 1C, Right). For example, the SNAP-tagged N and C termini of DRC3 have been unambiguously located in the L1 projection of the N-DRC linker that connects to IDA g (16). However, tagging the N terminus of DRC4 with BCCP resulted in variable labeling throughout the proximal lobe. A C-terminal biotin carboxyl carrier protein (BCCP) tag of DRC5 appears to localize to the upper (ODA-facing) region of an unbranched part of the N-DRC linker, and the cryo-ET density corresponding to the tag at the C terminus of DRC5 was diffuse and blurred (7, 14). Moreover, an N-terminally BCCP-tagged DRC5 construct was poorly expressed and could not be localized (14).

To address the uncertainty about the locations of the N-termini of DRC4 and DRC5, and the C terminus of DRC5, we generated different SNAP- or BCCP-tagged constructs of DRC4 and DRC5 and used them to rescue the corresponding mutants drc4 (pf2) and drc5 (sup-pf4) (Fig. 6 and SI Appendix, Figs. S3–S5). The forwarding swimming speeds of the rescued strains, i.e., pf2-4;N-SNAP-DRC4, sup-pf4;DRC5-C-SNAP, and sup-pf4;N-BCCP-DRC5, were significantly faster than the mutants and increased to near WT levels (Fig. 6B and SI Appendix, Fig. S5B). Western blots of purified axonemes demonstrated that the tagged proteins were expressed at endogenous levels and migrated at the sizes predicted for SNAP- or BCCP-tagged DRC subunits (Fig. 6 C and D and SI Appendix, Fig. S5C). The SNAP-DRC4 rescue also restored the assembly of other DRC subunits (Fig. 6C) that were previously shown to be missing in pf2 mutant axonemes (13).

Fig. 6.

Rescue of drc4 (pf2) and drc5 (sup-pf4) mutants by transformation with SNAP-tagged DRC subunits. (A) The DRC4 and DRC5 protein sequences are drawn to scale, with the location of predicted polypeptide domains indicated: CC, coiled-coil domain; LRR, leucine-rich repeat; SNAP, SNAP tags introduced near the N terminus of DRC4 and the C terminus of DRC5. (B) The forward swimming velocities of drc mutants as measured by phase contrast microscopy are shown relative to those of wild-type (WT) and rescued strains generated by transformation with the indicated SNAP-tagged DRC gene. As indicated by the asterisks (*P < 0.05; ***P < 0.001), both drc mutants were significantly slower than WT and the SNAP-tagged rescue strains, but the latter were also slightly slower than WT. (C and D) Western blots of axonemes isolated from WT, drc mutants, and SNAP-tagged rescued strains were probed with antibodies against several DRC subunits. Note the presence of bands that migrated at the sizes predicted for SNAP-tagged DRC subunits. An antibody against the IC2 subunit of the outer dynein arms served as a loading control.

Analyses of axonemes from the rescued strains by cryo-ET and subtomogram averaging revealed that the defects in N-DRC structure were completely rescued to WT structure and that the small SNAP and BCCP tags did not interfere with the proper assembly of the N-DRC or other axonemal structures (e.g., compare SI Appendix, Fig. S6 D–F with SI Appendix, Fig. S6 J–L). After enhancing the visibility of the cloned tags in the rescued axonemes with streptavidin-nanogold labels, we were able to detect additional densities in the averages of the SNAP-DRC4 and DRC5-SNAP axonemes (Fig. 7, SI Appendix, Figs. S3 and S4, and Movies S4 and S5), but not in the BCCP-DRC5 averages (SI Appendix, Fig. S5 J–L). The failure to visualize the N terminus of DRC5 is similar to previously reported results (14). However, a clear density corresponding to the labeled N terminus of DRC4 was found on the proximal side of the proximal lobe close to the neighboring B-tubule in SNAP-DRC4 axonemes (Fig. 7 D–F, SI Appendix, Fig. S3, and Movie S4). Taken together with the previously determined location of the C terminus of DRC4, our data demonstrate that the DRC4 stretches along the entire length of the N-DRC, i.e., from its attachment site on the A-tubule near the inner DMT junction through the base plate to the proximal lobe. This is a similar structural arrangement to DRC1 and DRC2. The results reveal that DRC4 likely serves both as an anchor that attaches the N-DRC to the A-tubule and as a scaffold for the assembly of the other N-DRC subunits, similar to DRC1 and DRC2. Furthermore, averages of gold-labeled, DRC5-SNAP axonemes clearly showed the additional density of the gold nanoparticles that revealed the location of the DRC5 C terminus at the center of the linker close to where the proximal lobe emerges from the linker (Fig. 7 G–I, SI Appendix, Fig. S4, and Movie S5). Together with previous studies, our data provide a better understanding of the molecular organization of the N-DRC.

Fig. 7.

Precise localization of the N terminus of DRC4, and the C terminus of DRC5 by SNAP tag labeling and cryo-ET. Three-dimensional isosurface renderings show the structures of the 96-nm repeats in WT (A–C), labeled SNAP-DRC4 (D–F), and labeled DRC5-SNAP (G–I) axonemes after cryo-ET and subtomogram averaging. Purified axonemes with SNAP-tagged DRC4 (D–F) or DRC5 (G–I) were labeled with streptavidin-Au (1.4-nm nanogold). Cross-sectional (A, D, and G), longitudinal front (B, E, and H), and enlarged longitudinal bottom views (C, F, and I; looking from the axoneme center outward) of the nexin–dynein regulatory complex (N-DRC, blue) show the additional density of the SNAP-biotin-strep-Au label (manually colored gold): at the proximal lobe (pL) of the N-DRC linker in the N-terminally labeled DRC4 axonemes (D–F), and at the center of the N-DRC linker in the C-terminally labeled DRC5 axonemes (G–I). Other labels: a–e and g, inner dynein arm isoforms; A_t, A-tubule; B_t, B-tubule; dL, distal lobe; I1α/I1β/ICLC, α- and β-heavy chain, and intermediate/light-chain complex of I1 dynein; IDA, inner dynein arm; ODA, outer dynein arm; RS, radial spoke.

Discussion

Analysis by 2D gel electrophoresis, immunoprecipitation, iTRAQ labeling, and mass spectrometry has identified at least 11 N-DRC subunits (12, 13). In this study, by comparing the missing densities in the subtomogram averages of 2 recently isolated n-drc mutants, we have revealed the location of 3 DRC subunits: DRC7 in the linker region including the OID linker, and DRC8/11 in the upper portion of the proximal lobe. We also used gold labeling to determine the precise locations of the N terminus of DRC4 and the C terminus of DRC5 (Table 1). Combined with previous studies (7, 14, 16), we propose a revised model for the overall architecture of the N-DRC: The subunits DRC1, DRC2, and DRC4 form the core scaffold of the N-DRC and a platform for the assembly of functional subunits, including DRC3, DRC5, DRC6, DRC7, DRC8, and DRC11, with this backbone (Fig. 8 and Movie S6).

Table 1.

Summary of polypeptides and structures present or absent in the strains used in this study

Components	Strains
	WT	drc7	drc11	pf2	SNAP-DRC4	sup-pf-4	DRC5-SNAP	BCCP-DRC5
Polypeptides (kDa)
DRC1 (79)	+	+	+	+	+	+	+	+
DRC2 (65)	+	+	+	+	+	+	+	+
DRC3 (60)	+	+	+	−	+	+	+	+
DRC4 (55)	+	+	+	−	+	+	+	+
DRC5 (43)	+	+	+	−	+	−	+	+
DRC6 (28)	+	+	+	−	+	−	+	+
DRC7 (177)	+	−	+	−	+	+	+	+
DRC8 (19)	+	+	−	−	+	+	+	+
DRC9 (46)	+	+	+	−	+	+	+	+
DRC10 (41)	+	+	+	−	+	+	+	+
DRC11 (95)	+	+	−	−	+	+	+	+
Structure (kDa)
Base plate (∼300)	+	+	+	+/−	+	+	+	+
Linker
Proximal lobe (∼350)	+	+	−	−	+	+	+	+
Distal lobe (∼100)	+	+/−	+	−	+	+/−	+	+
L2 protrusion (∼200)	+	+	+	−	+	−	+	+
OID linker (∼200)	+	−	+	−	+	+	+	+

+ indicates that the component was present at WT levels; – indicates that the component was absent; +/− indicates that the component was present at reduced levels. Molecular mass of the N-DRC structure is estimated from the isosurface rendering and indicated in parentheses (in kilodaltons).

Fig. 8.

Updated model of the DRC subunit organization and N-DRC interactions with neighboring structures. The locations of the DRC subunits are summarized and colored in the WT nexin–dynein regulatory complex (N-DRC) structure, which is shown in cross-sectional (A), longitudinal front (B), and enlarged bottom (C) views. The structural comparison between the averaged WT axonemal repeat and deletion mutants of DRC subunits or mutant rescues by tagged DRC subunits revealed the architecture of the N-DRC: The core scaffold extends along the entire length of the N-DRC and consists of DRC1 (yellow), DRC2 (dark green), and DRC4 (magenta); with this scaffold associate several functional subunits, i.e., DRC3 (light green), DRC5/6 subcomplex (cyan), DRC7 (red), and DRC8/11 subcomplex (orange). The composition of a few areas within the N-DRC remain unknown (dark blue). Other labels: a–e and g, inner dynein arm isoforms; A_t, A-tubule; B_t, B-tubule; I1α/I1β/ICLC, α- and β-heavy chain, and intermediate/light-chain complex of I1 dynein; IDA, inner dynein arm; ODA, outer dynein arm; RS, radial spoke.

DRC1, DRC2, and DRC4 Form the Backbone of the N-DRC.

Cryo-ET and subtomogram averages of the axonemal repeat from Chlamydomonas null mutants of drc1 (pf3), drc2 (ida6), and drc4 (pf2) previously revealed large-scale defects in the structure of the N-DRC, i.e., missing or reduced densities in both the base plate and the linker domains (7, 16, 23). Transformation with WT or tagged versions of DRC1, DRC2, or DRC4 fully restored the missing polypeptides and structures (7, 14, 16). The localization of the termini of DRC1, DRC2, and DRC4 in the previous and current studies suggests that all 3 subunits stretch along the entire N-DRC, i.e., from the inner junction to the neighboring DMT (14, 16). In addition, the use of advanced microscope hardware, such as a direct electron detector camera and Volta-Phase-Plate, has increased the resolution of our subtomogram averages, allowing us to visualize 3 filamentous structures in the base plate that are slightly twisted around each other like a rope (Fig. 8C). This is consistent with the high-resolution reconstruction of the WT Tetrahymena N-DRC using our recently developed TYGRESS software (26). These findings suggest that DRC1, DRC2, and DRC4 function as scaffold and play a pivotal role in the assembly of the entire N-DRC structure. Among the 3 backbone proteins, the subcomplex of DRC1 and DRC2 is the most critical as its absence in the pf3 (drc1) and ida6 (drc2) mutants caused reduced densities for the entire N-DRC base plate, including DRC4 (16, 23). Thus, DRC1 and DRC2 form likely the core filament at the very bottom of the N-DRC base plate, which directly associates with the A-tubule (Fig. 8C, yellow and dark green). DRC4 does not appear to be necessary for assembly of DRC1 and DRC2 into the base plate, as the pf2 mutant has substantial remaining base-plate density, suggesting that DRC4 likely forms the filamentous structure on top of the DRC1/DRC2 subcomplex in the N-DRC base plate (Fig. 8C, purple) (16, 23).

All scaffold components (DRC1, DRC2, and DRC4) are highly conserved in motile cilia and flagella, from unicellular algae Chlamydomonas to human. Genetic studies of the orthologs in different species have revealed that DRC1, DRC2, and DRC4 play essential and conserved roles in ciliary and flagellar motility. For example, an insertion into the DRC4 locus in Chlamydomonas resulted in an altered ciliary waveform and slow swimming cells of the pf2 strain (22). RNA interference-mediated knockdown of the DRC4 ortholog in trypanosomes revealed that DRC4 is required for flagellar motility and is essential for the viability of the bloodstream form of the parasite (27). The vertebrate orthologs of DRC4 are GAS11 in human and GAS8 in mice and zebrafish, and mutations in GAS11 cause primary ciliary dyskinesia (28).

Previous cryo-ET studies failed to unambiguously identify the location of the N terminus of DRC4. Oda et al. (14) reported that the labeling density of the N terminus of DRC4 appeared to be present in 2 locations, with one density at the proximal lobe and the other at the distal lobe. In contrast, in our studies, the SNAP-tagged density clearly localized the N terminus of DRC4 at the proximal lobe (Fig. 7 D–F). The 96-nm averaged repeats of the drc11 mutant showed a density that remains at the bottom (IDA-facing side) of the proximal lobe (Fig. 5). These results suggest that the N-terminal parts of DRC2 and DRC4 form the structural scaffold at the bottom of the proximal lobe, which can bind to the DRC8/11 subcomplex.

DRC3, 5, 6, 7, 8, and 11 Are Attached to the N-DRC Backbone and Interact with Other Axonemal Complexes.

Loss of DRC3 (drc3), DRC5/DRC6 (sup-pf4), DRC7 (drc7), and DRC8/DRC11 (drc11) cause defects in certain regions of the N-DRC but have minimal effects on the assembly of other DRC subunits, unlike loss of the backbone subunits (7, 15, 16).

DRC5 is a leucine-rich repeat (LRR) protein that appears to be closely associated with DRC6 based on the absence of DRC6 in the DRC5 mutant sup-pf4 (12, 29). Cryo-ET and subtomogram averaging revealed that the L2 protrusion of N-DRC linker, which is located in front of the OID linker, is missing in sup-pf4 axonemes (7) (SI Appendix, Fig. S6 G and I). The motility defects observed in the Chlamydomonas sup-pf4 mutant are subtle. The sup-pf4 mutant and the tubulin polyglutamylation (tpg1) mutant can both undergo reactivated motility in the presence of ATP, like WT cells, whereas the sup-pf4;tpg1 double mutant failed to reactivate motility in the presence of ATP. In addition, sup-pf4;tpg1 axonemes were more prone to splaying, i.e., the 9 DMTs separated more easily from each other, suggesting that DRC5/6 and B-tubule glutamylation of the neighboring DMT work together to maintain axonemal integrity and ciliary motility (30). However, it is not known which DRC subunit(s) make direct contact with the neighboring B-tubule. Here, cryo-ET localization of a tagged rescue strain suggested that the C terminus of DRC5 is located in the center of the N-DRC linker domain, rather than close to the neighboring DMT. The N-terminal tag of BCCP-DRC5 could not be visualized (SI Appendix, Fig. S5 J–L), even though Western blots of the rescued strain clearly showed that BCCP-DRC5 was expressed at WT levels (SI Appendix, Fig. S5C), indicating that the addition of the BCCP tag does not interfere with the incorporation of DRC5 into N-DRC. Two possible reasons for failing to detect the location of the N terminus of DRC5 might be that it is buried so deep in the structure of the N-DRC that the streptavidin-gold could not bind to the tag or that the N-terminal domain may be so flexible in position that the gold density would be averaged out in the subtomogram averages.

DRC7 is the largest N-DRC subunit (177 kDa) identified in Chlamydomonas. It contains a highly conserved transglutaminase-like (TGL) peptidase domain, which is predicted to bind tightly to glutamylated proteins, suggesting DRC7 may interact with glutamylated residues on the neighboring B-tubule (13). This idea is supported by our cryo-ET and subtomogram averaging images that show that a part of the distal lobe is missing in the drc7 axonemes. However, the major defect in the drc7 axonemes was seen in the middle segment of the N-DRC linker, including a missing OID linker that usually connects the N-DRC to the ODA row. The OID linker is proposed to be part of the signal transduction pathway that regulates ciliary beating: Signals are transferred from the CPC to the RS, then to the N-DRC and I1 dynein, finally reaching the ODAs via 2 routes; one OID linker from the N-DRC and a second OID linker from the I1 intermediate/light-chain complex, respectively (31–33). However, even though the defects in N-DRC structure seem to disrupt 2 important interfaces with neighboring structures, drc7 mutant did not show significant defects in forward swimming velocity under standard culture conditions in Chlamydomonas. A possible explanation for the lack of a strong motility phenotype in Chlamydomonas drc7 mutants might be that there are other—(partially) redundant—pathways for signal transduction. The N-DRC connects to the neighboring B-tubule via both the proximal and distal lobes, and the loss of the distal lobe in the drc7 mutant might be compensated for by the remaining proximal lobe that still interacts with the neighboring B-tubule. In addition to the OID linker from the N-DRC, also the I1 ICLC connects between the RS/IDA row to the ODAs, and these 2 signal transduction pathways between the IDA and ODA rows might be redundant for the activity control of the ODAs. Future study of double mutants in Chlamydomonas may reveal a more profound effect of the drc7 mutation. Another possibility is that DRC7 might be critical for motility under more stringent conditions. In Drosophila, loss of the DRC7 ortholog CG34110 (lost boys, lobo) causes defects in sperm storage and fertility (34). Future studies will have to clarify the reasons for this more severe phenotype in flies, but a recent survey of motility genes expressed in Drosophila sperm has revealed that not all DRC subunits are expressed in the testis (35). Thus, it is also possible that loss of DRC7 may cause a more significant functional defects in other species.

DRC8 contains an EF-hand domain (a calcium-binding helix–loop–helix structural motif) that could bind Ca²⁺, and DRC11 contains a calmodulin-binding motif, suggesting these 2 subunits may contribute to the calcium regulation of flagellar motility. DRC11 also contains an “ATPases associated with diverse cellular activities” (AAA) domain that could mediate nucleotide-sensitive conformational changes. Our studies of the drc11 mutant showed that DRC11 is required for the assembly of DRC8, and DRC8/11 probably form a subcomplex localized to the proximal lobe of the linker domain that contacts the neighboring B-tubule (Fig. 4). Thus, we propose that the DRC8/11 subunits regulate cilia motility by modulating contacts with the neighboring B-tubule, which could be important for transforming interdoublet sliding to ciliary bending, and/or provide signal feedback. The ortholog CMF22 in Trypanosoma brucei is required for normal flagellar beating patterns and propulsive cell motility (36). However, in Chlamydomonas, we only observed small decreases in the swimming velocity of the drc11 mutant compared to WT cells (Fig. 2B). One possible explanation may be that the N-DRC connects to the neighboring B-tubule via both the proximal lobe and the distal lobe. The proximal lobe is missing in the drc11 mutant, but the remaining distal lobe could interact with the B-tubule and transform the interdoublet sliding to bending.

Orthologs of N-DRC subunits have been found in motile cilia, both “9 + 2” and “9 + 0” of many species (13, 37, 38), but not in nonmotile primary 9 + 0 cilia (39, 40). The fact that the N-DRC seems to be present in all motile cilia, even in motile 9 + 0 cilia that can lack almost any other major axonemal complexes, such as the CPC, RSs, and IDAs or ODAs (6, 37), suggests an essential and central role for the N-DRC in regulating ciliary motility. Our results provide insights into several important aspects of N-DRC assembly and function, from revealing the precise location of several DRC subunits, the overall organization of the N-DRC with a 3-subunit core-complex (DRC1/2/4) that serves as the scaffold for the assembly of the “functional subunits” (DRC3/5–8/11), shedding light on the molecular mechanism by which the N-DRC regulates dynein activity and thus flagellar beating.

Methods

Culture Conditions, Strain Construction, and Identification of Recently Isolated Mutations.

Strains, including WT Chlamydomonas reinhardtii and CLiP mutants (SI Appendix, Table S1), and mutant strains rescued with tagged DRC subunits were cultured as previously described (13) (see SI Appendix for more details on the identification and characterization of drc mutants and the construction of epitope-tagged N-DRC strains). The Chlamydomonas strain cw15, which was used as the background strain for the CLiP library, is cell wall-less but assembles flagella resembling those of WT (cc-125).

Phase Contrast, Fluorescence Microscopy, and Measurements of Swimming Velocity.

Transformants and candidate drc mutants were screened by phase contrast microscopy using a 20× or 40× objective, a halogen light source, and a red filter on a Zeiss Axioskop. Forward swimming velocities were measured using a Rolera-MGi EM-CCD camera (Q Imaging) and the Metamorph software package, version 7.6.5.0 (Molecular Devices), as previously described (13). The forward swimming velocities of the CLiP mutants and CLiP-mutant rescues were compared with the “WT”-like CLiP background strain cw15, whereas pf2, suppf4, SNAP-DRC4, DRC5-SNAP, and BCCP-SNAP were compared to the WT strain cc-125. Some transformants were screened for the presence of the SNAP or BCCP tag by immunofluorescence microscopy (41).

Preparation of Purified Axonemes for SDS/PAGE, Western Blotting, iTRAQ Labeling, and Mass Spectrometry.

Chlamydomonas isolated axonemes were prepared as previously described (13, 42). Briefly, Chlamydomonas cells were collected by centrifugation and resuspended in pH shock buffer containing 10 mM Hepes, pH 7.4, 1 mM SrCl₂, 4% sucrose, and 1 mM DTT. Flagella were detached from cells by adding 0.5 M acetic acid to reduce the pH to 4.3. After 80 s, the pH was increased to 7.4 by adding 1 M KOH. The flagella pellets were then demembranated by adding 1% Igepal CA 630 (Sigma-Aldrich) to the solution with gentle rotation for 20 min at 4 °C. Axonemes were collected by centrifugation at 10,000 × g for 10 min and resuspended in HMEEN (10 mM Hepes, 5 mM MgSO₄, 1 mM EGTA, 0.1 mM EDTA, 30 mM NaCl, pH 7.4) plus 1 mM DTT and 0.1 μg/mL protease inhibitors (leupeptin, aprotinin, and pepstatin). Samples were separated on 5 to 15% polyacrylamide gradient gels, transferred to Immobilon P, and probed with the different antibodies listed in SI Appendix, Table S8, as described by ref. 13.

For iTRAQ labeling and mass spectrometry, isolated axonemes were washed with 10 mM Hepes, pH 7.4, to remove salt, DTT, and protease inhibitors, and then resuspended in 0.5 M triethylammonium bicarbonate, pH 8.5, and processed for trypsin digestion and iTRAQ labeling as described in refs. 13 and 41. Samples were combined and fractionated off-line using high pH, C18 reversed-phase chromatography. Column fractions were vacuumed dried, resuspended in solvent (98:2:0.01, water/acetonitrile/formic acid), and loaded in 1- to 1.5-µg aliquots for capillary LC using a C18 column at low pH. The C18 column was mounted in a nanospray source directly in line with a Velos Orbitrap mass spectrometer (Thermo Fisher Scientific). On-line capillary LC, MS/MS, database searching, and protein identification were performed as previously described (13, 43) using ProteinPilot software, version 5.0 (AB Sciex), and the most recent version (version 5.5) of the Chlamydomonas database (https://phytozome.jgi.doe.gov/pz/portal.html). The bias factors for all samples were normalized to α- and β-tubulin.

WT (cw15) and drc axonemes were each labeled with 2 different iTRAQ reagents in a 4-plex experiment (technical replicates), and then each 4-plex experiment was repeated with new samples (biological replicates). For the drc7 (197909) mutant, 1,166 proteins were detected with high confidence at a false-discovery rate of 5% in the first 4-plex experiment, and 701 proteins were detected in the second experiment. For the drc11 (068819) mutant, 949 proteins were detected in the first 4-plex experiment, and 768 proteins were detected in the second experiment. The datasets were further filtered to identify those proteins that were identified by at least 5 peptides and whose mutant/WT ratios were significantly different (P < 0.05) from WT/WT in all replicates.

Axoneme Preparation and SNAP-Gold Labeling for Cryo-ET.

Chlamydomonas axonemes for cryo-ET were prepared as mentioned above. The only difference is that the final axoneme pellets were resuspended in HMEEK buffer (30 mM Hepes, 5 mM MgSO₄, 1 mM EGTA, 0.1 mM EDTA, and 25 mM KCl, pH 7.4). A volume of 3 μL of the isolated axonemes was added to glow-discharged R2/2 holey carbon-coated grids and gently mixed with 1 μL of 10-times-concentrated, BSA-coated 10-nm gold solution (44). After ∼2 s of backside blotting with filter paper, the grid was rapidly plunge frozen in liquid ethane with a homemade plunge freezer. Frozen grids were stored in liquid nitrogen until used.

For SNAP-gold labeling samples, the SNAP tag was labeled with gold as previously described (16). Briefly, 1 µL of 1 mM BG-biotin or BG-(PEG)₁₂-biotin (New England Biolabs; PEG linker available on request) was added to 200 µL of the purified axonemes in HMEEK buffer and incubated overnight at 4 °C. Unbound BG substrate was removed by centrifugation at 10,000 × g for 10 min at 4 °C and 3 wash steps with HMEEK buffer. The axoneme pellet was resuspended in 200 µL of HMEEK buffer. After adding 5 µL of 80 µg/mL 1.4-nm-sized streptavidin nanogold particles (strep-Au; Nanoprobes), the suspension was incubated at 4 °C for 4 h. After adding 800 µL of HMEEK buffer, the labeled axonemes were pelleted by centrifugation at 10,000 × g for 10 min at 4 °C and resuspended in 200 µL of HMEEK buffer.

Cryo-ET.

Vitrified grids were imaged using a Titan Krios transmission electron microscope (used for all samples unless otherwise noted) or Tecnai F30 (used for WT, SNAP-DRC4, DRC5-SNAP, and BCCP-DRC5 axonemes) both operated at 300 kV. Images were captured using a 4k × 4k K2 direct detection camera (Gatan) on the Titan Krios at a magnification of 26,000× (yielding a scaling of 5.5 Å per pixel) or a 2k × 2k charge-coupled device camera (Gatan) on the Tecnai F30 at a magnification of 13,500× (yielding a scaling of 10.0 Å per pixel). Tilt series were collected from 60° to −60° in 2° steps using dose-symmetric method on the Titan Krios. Counting mode of the K2 camera was used and for each tilt image 15 frames (0.4-s exposure time for each frame) were recorded. Both cameras were placed behind a postcolumn energy filter (Gatan) that was operated in zero-loss mode (20-eV slit width). Data acquisition was performed using the microscope control software SerialEM (45) in low-dose mode, and the total electron dose per tilt series was limited to ∼100 e/Ų.

Image Processing.

The frames of each tilt series image collected on a K2 camera were aligned and then merged using the script extracted from the IMOD software package (46) to generate the final tilt serial dataset. Tomograms were reconstructed using fiducial alignment and the backprojection method in the IMOD package (46). Subtomogram averaging was performed using the PEET software (19, 20) to average the 96-nm axonemal repeat units from 3D reconstructed axonemes. The resolution of the DMT averages was estimated using the 0.5 criterion of the Fourier shell correlation method (SI Appendix, Table S5). The Chimera package (47) was used for 3D visualization and surface rendering. Mass estimations of the DRC complex and subvolumes were calculated using the average density of 1.43 g/cm³ for proteins (48) and after normalizing the isosurface-rendering threshold to the mass of MTs.

For the classification of the lower proximal lobe of the N-DRC linker, a cylindrical mask (size, 12 × 12 × 16 pixels) was placed over the area containing the lower proximal lobe to “hide” everything except this area of interest. Focusing on this area, we then performed a principal-component analysis of the aligned subtomogram volumes (axonemal repeats) from WT and drc11 (20); 15 features were selected for clustering, generating 12 subclasses that were then partially combined into 3 distinct classes based on the position of the remaining proximal lobe between the rows of the inner and outer dynein arms.

Data Availability.

The averaged structures reported in this paper have been deposited in the Electron Microscopy Data Bank under accession codes EMD-20338 (49), EMD-20339 (50), EMD-20340 (51), EMD-20341 (52), and EMD-20821 (53).