Caenorhabditis elegans centriolar protein SAS-6 forms a spiral that is consistent with imparting a ninefold symmetry

Abstract

Centrioles are evolutionary conserved organelles that give rise to cilia and flagella as well as centrosomes. Centrioles display a characteristic ninefold symmetry imposed by the spindle assembly abnormal protein 6 (SAS-6) family. SAS-6 from Chlamydomonas reinhardtii and Danio rerio was shown to form ninefold symmetric, ring-shaped oligomers in vitro that were similar to the cartwheels observed in vivo during early steps of centriole assembly in most species. Here, we report crystallographic and EM analyses showing that, instead, Caenorhabotis elegans SAS-6 self-assembles into a spiral arrangement. Remarkably, we find that this spiral arrangement is also consistent with ninefold symmetry, suggesting that two distinct SAS-6 oligomerization architectures can direct the same output symmetry. Sequence analysis suggests that SAS-6 spirals are restricted to specific nematodes. This oligomeric arrangement may provide a structural basis for the presence of a central tube instead of a cartwheel during centriole assembly in these species.

Centrioles are essential for organizing cilia and flagella in most eukaryotes as well as forming the centrosome of animal cells (1). As such, they play a crucial role in cell signaling, cell motility, embryonic development, and genome stability. Aberrations in centriole structure and function are linked to a number of diseases, such as ciliopathies, male sterility, primary microcephaly, and cancer (2). Caenorhabotis elegans has proven instrumental for the discovery of five components essential for centriole formation (3, 4): the coiled coil-containing proteins spindle defective protein 2 (5, 6), SAS-4 (7, 8), SAS-5 (9), and SAS-6 (10, 11) as well as the kinase zygote defective protein 1 (12, 13). Related proteins have since been found across the eukaryotic domain (14, 15), indicating a common evolutionary route for centriole formation.

In most species, centrioles and the related basal bodies are organized around a cartwheel composed of a circular central hub with outward facing spokes that are connected to microtubules (1). Centrioles are characterized by a universal ninefold symmetric arrangement of microtubules, and proteins of the SAS-6 family, one of the earliest components recruited during centriole formation (11, 16–18), are critical for establishing this arrangement (16). SAS-6 proteins from C. elegans, Chlamydomonas reinhardtii, and Danio rerio (ceSAS-6, crSAS-6, and drSAS-6, respectively) were shown to oligomerize through two interaction interfaces: a strong one mediated by a long coiled coil (the C-C interface), leading to homodimerization, as well as a 100-fold weaker one across a globular N-terminal domain (the N-N interface), leading to higher-order oligomerization between homodimers (19, 20). Crystal structures and EM of crSAS-6 and drSAS-6 oligomers showed that these proteins can assemble into ninefold symmetric rings, reminiscent of the cartwheels observed in centrioles (19, 20). It was, therefore, proposed that SAS-6 functions by forming rings in vivo, through which it establishes the ninefold symmetry of centrioles (19, 20). This model is consistent with the recent demonstration of cartwheels in Trichonympha sp. basal bodies by electron tomography (21).

The most significant difference to this consensus model of centriole assembly is found in C. elegans, the very system in which SAS-6 was originally discovered. Although they also exhibit the signature ninefold radial symmetric arrangement of microtubules, C. elegans centrioles are smaller compared with other metazoans (4), and many of their components show a significant level of sequence divergence (22), raising the question of whether they assemble in the same manner. Intriguingly, instead of cartwheels, C. elegans centrioles have been reported to be organized around a central tube (4), which is initially ∼60 nm in length and then increases to ∼110 nm as the centriole grows. Whereas ceSAS-6 exhibits the same domain composition as other SAS-6 family members (19) and is recruited to centrioles concomitant with elongation of the central tube (4, 23), it remains unclear how such a SAS-6–containing central tube could establish a ninefold symmetrical arrangement. We sought to address this question by studying the oligomerization properties of ceSAS-6 and comparing it with SAS-6 variants from C. reinhardtii and D. rerio (19, 20). Strikingly, we find that ceSAS-6 differs from other members of this protein family, because it self-assembles into filamentous spiral oligomers instead of rings. Our findings lead us to propose that the function of SAS-6 proteins in determining centriolar ninefold symmetry is conserved across evolution but that the structural details of how this symmetry is achieved may fundamentally differ in certain species.

Results

Crystal Structure of the Coiled Coil-Mediated ceSAS-6 Dimer.

To assess the oligomeric conformation of ceSAS-6, the structure of a homodimer mediated by the C-C interface was necessary. For C. reinhardtii and D. rerio, this structure was obtained by crystallizing SAS-6 constructs that included a dimerization-deficient N-terminal domain and a small stretch of the coiled coil (19, 20). However, analogous fragments of ceSAS-6 evaded crystallization. Given that the long α2-β5 loop of the ceSAS-6 N-terminal domain (residues 1–168) was not observed in the crystal structure of this construct (19) (ceN[S123E]), we hypothesized that disorder in this loop would adversely affect crystal packing. Removing this loop by excising residues N103 to P130 did not alter the structure of the remainder of the N-terminal domain, which was shown by the 1.8-Å resolution structure of ceN[Δ103–130] (Fig. S1 and Table S1). Thus, we designed ceSAS-6 constructs lacking this loop and were able to crystallize two variants that included the N-terminal domain and the first 47 residues of the coiled coil. These variants harbored a single amino acid substitution (ceN-CC[Δ103–130,I154G]) or a short deletion (ceN-CC[Δ103–130/Δ151–156]) to disrupt dimerization through the N-N interface, whereas coiled coil-mediated dimerization was maintained as shown by multiangle light scattering experiments (Fig. S2).

The structures were solved by molecular replacement and refined to 2.85- and 3.55-Å resolution for ceN-CC[Δ103–130/I154G] and ceN-CC[Δ103–130/Δ151–156], respectively (Table S1). In total, we obtained three ceN-CC dimer models (Fig. 1A) with an average pairwise C_α rmsd of 0.55 Å; analysis henceforth focuses on the highest resolution structure. The N-terminal domain structure is not affected by inclusion of the coiled coil, with a C_α rmsd of just 0.42 Å compared with the previously resolved ceN[S123E] variant (19). The overall topology of the ceSAS-6 dimer is also comparable with the topology of crSAS-6 and drSAS-6 (19, 20), with the coiled coil yielding a parallel in-register dimer and the N-N dimerization interfaces facing in opposite directions (Fig. 1A).

Fig. 1.

Crystal structure of the coiled coil dimer of ceSAS-6. (A) Superposition of the three ceSAS-6 dimer structures: Shown in blue is ceN-CC[Δ102–130/I154G], which was solved with one dimer per asymmetric unit. In orange and magenta are the two dimer copies per the asymmetric unit of ceN-CC[Δ102–130/Δ151–156]. The first three heptad repeats of the coiled coil were resolved in all structures (residues 1–191), except for ceN-CC[Δ103–130/I154G], where additional segments of the coiled coil are stabilized through contacts with a symmetry-related subunit. The N-N dimerization interfaces of the two subunits face in opposite directions as indicated, thereby allowing formation of large oligomers. (B and C) Six ceN-CC dimers were associated such that their N-terminal domains interact as observed in the N-N dimer structure (19). B presents a side-on view of the spiral. The two dimer interfaces (N-N and C-C) used for this model are indicated. C shows a top-down view that emphasizes the approximately fivefold symmetry of the model. The individual ceN-CC dimers are numbered. The outer diameter of the spiral was calculated using the ceN-CC[Δ102–130/I154G] model, where almost the entire coiled coil length is visible. (D) Superposition of the N-terminal domains of ceN-CC (blue) and the equivalent structure of crSAS-6 (green), showing the displacement of the coiled coil helix-α3.

ceSAS-6 Models Reveal a Spiral Conformation That Is Consistent with Ninefold Symmetry.

Surprisingly, ceSAS-6 oligomers modeled using the coiled-coil dimer and the previously determined N-N interface dimer (19) display a steep right-handed spiral with ∼33-nm pitch and ∼5 dimeric units per helix turn (Fig. 1 B and C). This arrangement stands in marked contrast to the rings of crSAS-6 (19) and drSAS-6 (20). The spiral is not driven by either the N-N interface, which is nearly identical between the three SAS-6 variants (19, 20), or the conformation of the coiled coil itself, the first three heptad repeats of which can be superimposed between ceSAS-6 and crSAS-6 with just 0.97-Å C_α rmsd. Rather, it is the relative orientation of the coiled-coil helix-α3 with respect to the N-terminal domain that differs between ceSAS-6 and crSAS-6 by ∼38° (Fig. 1D). This displacement is responsible for moving successive ceSAS-6 dimers away from the plane of a SAS-6 ring (Fig. S3); a similar displacement can be seen when superimposing the ceSAS-6 and drSAS-6 structures.

This intriguing ceSAS-6 spiral model compelled us to consider whether it is compatible with the ninefold symmetry of centrioles. Interestingly, we were able to model a ceSAS-6 spiral that features 4.5 dimeric units per turn by simply adjusting the ψ-angle of V167 at the pivot between the N-terminal domain and the coiled-coil helix from the observed 124° to 128° (Fig. S4A). This minor change creates no significant steric clashes. The resulting spiral is ninefold symmetric every two turns and displays a ∼31-nm pitch (Fig. S4 C and D). In contrast, modeling a ninefold symmetric ceSAS-6 ring would necessitate larger changes in the backbone dihedrals of the pivot residues, most significantly ∼30° adjustments of the H165 ψ- and R168 ϕ-angles, as well as the adoption of an α-helical conformation by G169 (Fig. S4B). Moreover, the α3 backbone would clash substantially with the side chain of V167 (Fig. S4E).

To better explore the limits of ceSAS-6 conformations, we performed atomistic molecular dynamic simulations starting from the ceN-CC[Δ103–130/Δ151–156] structure (Fig. S5). Oligomeric models produced during a 20-ns simulation show spirals with an average periodicity of 4.9 ± 0.6 ceSAS-6 dimers per turn and a pitch of 31 ± 3 nm. A significant degree of protein flexibility is evident in these simulations, because the oligomeric models explore symmetries in the range of ∼3.5–6.5 dimers per turn and pitches in the range of 25–39 nm. We conclude that crystallographic and modeling data strongly suggest a ceSAS-6 spiral oligomer as opposed to a ring and that ninefold symmetry is easily within the conformational reach of a spiral arrangement.

Stabilization of the ceSAS-6 N-N Interface.

We set out to directly visualize the ceSAS-6 oligomers by EM. However, neither negative stain EM nor rotary metal shadowing preparations showed the presence of large oligomers (Fig. S6 A and B); in contrast, crSAS-6 displayed clear rings under similar conditions (19). We hypothesized that an open-ended spiral assembly may not be stably populated under the relatively dilute (∼20–40 μM) protein concentrations suitable for EM, because dimerization between N-terminal domains is weak (K_D ∼ 100 μM) (19).

The N-N interface of ceSAS-6 critically depends on an isoleucine at the tip of the β6-β7 loop (I154) (19), which buries in a hydrophobic cavity upon N-N dimer formation (Fig. 2E). Substitution at this position by a charged residue (I154E) disrupted dimerization (19), whereas substitution with tryptophan improves N-N dimerization affinity ∼20-fold compared with WT (I154W, K_D = 4.7 ± 0.2 μM) (Fig. 2A). Two different crystal structures of ceN[S123E/I154W] at 2.65- and 2.8-Å resolutions (Table S1) showed only small changes in the N-terminal domain compared with WT, with an average C_α rmsd of 1.55 Å against ceN[S123E] (Fig. 2 B and C). The structural changes primarily involve repositioning of α1 and partial closure of the hydrophobic cavity originally occupied by I154 (Fig. 2 C and D). Crucially, the relative orientation and position of domains in the N-N dimer remain unaltered, and ceSAS-6 oligomers modeled using the ceN[S123E/I154W] structure show a spiral arrangement similar to the WT.

Fig. 2.

I154W increases the ceSAS-6 N-N dimerization affinity. (A) Fluorescence polarization experiments of WT- (purple squares), I154E- (orange crosses), or I154W-substituted (green triangles) ceN constructs to determine the N-N dimerization affinity. The I154E substitution disrupts N-N dimerization (19), and the K_D of WT ceN matches the value estimated from isothermal titration calorimetry experiments (110 ± 30 μM) (19). The I154W substitution leads to a 20-fold increase in affinity compared with the WT. Error bars derive from five independent measurements. (B) Superposition of the four N-N dimer models from the two crystal structures of ceN[S123E/I154W]. The average rmsd for all C_α atoms is 0.51 Å. The N-N dimer interface is essentially unchanged compared with the interface of ceN[S123E], with the β6-β7 loop from one subunit contacting a hydrophobic patch between α1 and α2 of the second molecule. (C) Superposition of one copy of ceN[S123E/I154W] (gold) onto ceN[S123E] (red). The main difference in the two structures is the slight displacement of α1 in ceN[S123E/I154W]. Note that the relative position and orientation of the head domains are unchanged. (D and E) Close-up view of the area around residue 154 reveals the reason for α1 displacement. Whereas I154 in ceN[S123E] fits tightly into a cavity formed between primarily hydrophobic residues (E, green), the aromatic ring of W154 sits on top of this hydrophobic patch, leading to a partial collapse of the cavity (D).

To evaluate the potential biological effect of the I154W substitution, we generated transgenic worms expressing GFP fused to SAS-6[I154W] that were engineered to be resistant to RNAi directed against endogenous SAS-6 (GFP-SAS-6RR[I154W]) (19). Whereas most WT embryos assemble a monopolar spindle in each blastomere on sas-6(RNAi) at the end of the second cell cycle as a result of defective centriole formation (Fig. 3A and Movie S1), 75% of embryos expressing GFP-SAS-6RR[I154W] undergo bipolar spindle assembly in each blastomere under the same RNAi conditions (Fig. 3 C and D and Movie S3). This level of rescue is comparable with the level observed in embryos expressing WT GFP-SAS-6RR (Fig. 3 B and D and Movie S2), despite the relatively low levels of GFP-SAS-6RR[I154W] expression (Fig. 3 B, D, and E). It is likely that the increased N-N affinity of GFP-SAS-6RR[I154W] compensates for the reduced expression level of this construct. We conclude that I154W represents a functionally competent substitution, thereby supporting the crystallographic data that I154W does not alter ceSAS-6 structure in a significant manner.

Fig. 3.

Functional analysis of ceSAS-6[I154W]. (A–C) Still images at the end of the second cell cycle from representative differential interference contrast recordings of embryos treated with sas-6(RNAi) and expressing (B) GFP-SAS-6RR or (C) GFP-SAS-6RR[I154W]; complete cytokinesis movies are provided (Movies S1, S2, and S3). Elapsed time after the onset of cytokinesis is shown in minutes and seconds. In every panel, anterior is to the left; arrowheads indicate cells. (Scale bar: 10 μm.) (D) Quantification of experiments illustrated in A–C. (E) Western blot analysis of GFP-SAS-6RR and GFP-SAS-6RR[I154W] worm lysates probed with SAS-6 antibodies to reveal both endogenous protein and the GFP fusion variant. Note that the substituted construct expresses at lower levels but functions as well as the WT, likely because of the enhanced N-N affinity.

EM of ceSAS-6 Spiral Assemblies.

Analytical ultracentrifugation (AUC) sedimentation velocity experiments of ceN-CC[S123E/I154W] confirmed that this construct, with two strong dimerization interfaces, forms larger oligomers compared with the WT (Fig. 4A). Whereas ceN-CC[S123E/I154E] migrated as a dimer and ceN-CC[S123E] migrated as a mixture of dimeric, tetrameric, and hexameric species at concentrations of 50 μM, ceN-CC[S123E/I154W] formed 8–10mer assemblies already at 25 μM (1 mg/mL). However, even these assemblies, if elongated, would be hard to visualize by EM. To increase the probability of observing large oligomers, we prepared EM grids using highly concentrated (5–7.5 mg/mL) ceN-CC[S123E/I154W] samples, which were first cleared of pelletable material by centrifugation, rapidly diluted to 1 mg/mL, placed on grids and stained with uranyl acetate. Under these conditions, ceN-CC[S123E/I154W] showed spirals of variable length (Fig. 4B and Fig. S6 D–F), albeit in the presence of significant protein background. Consistent with the molecular dynamics simulations that indicated protein flexibility, the spiral diameters and pitch are variable. However, their average values (diameter = 15 ± 3 nm, n = 112; pitch = 26 ± 6 nm, n = 124) (Fig. 4 D–F) correspond well to the outer diameter (∼17 nm) and pitch (∼33 nm) of a ceN-CC oligomer derived from the crystallographic model (Fig. 2 D and E) and fit even better to the equivalent dimensions of a hypothetical 4.5-fold symmetric ceN-CC spiral (Fig. S4 C and D).

Fig. 4.

AUC analysis and visualization of ceSAS-6 oligomers. (A) AUC velocity analysis of ceN-CC[S123E/I154E] (blue) at 50 μM concentration shows a narrow sedimentation distribution corresponding to the molecular mass of a dimer. ceN-CC[S123E], which can oligomerize through its coiled coil and the low-affinity N-N interface, shows a distribution of small oligomers at 50 μM concentration (red). In contrast, ceN-CC[S123E/I154W], which exhibits high N-N dimerization affinity, forms larger oligomers at 25 μM concentration (green). (B and C) Electron micrographs of negatively stained ceN-CC[S123E/I154W] (B) after or (C) before sample centrifugation. The stabilized ceN-CC forms spirals at higher concentrations, which in turn, give rise to micrometer-long filaments that show evidence of spiral intertwining. The white arrowhead in C indicates the point where the two strands separate. (D) Close-up view of a representative ceN-CC[S123E/I154W] micrograph from B compared with our crystallographic model of ceN-CC spirals. (Scale bar: 10 nm.) (E and F) Histograms of spiral diameter and pitch, respectively. The values are consistent with the dimensions of our crystallographic model of ceN-CC spirals (black solid lines; 17-nm diameter, 33-nm pitch) but fit better to the dimensions of a 4.5-fold symmetric spiral (red lines; 15-nm diameter, 31-nm pitch). Experimental mean values and SDs are shown by dotted lines. (G) Histogram of filament diameter from C and similar migrographs (Fig. S7B).

Attempts to further stabilize the spirals and reduce electron micrograph background by mild chemical cross-linking did not yield substantially better results; however, we did notice the formation of micrometer-long filamentous structures that appeared to be composed of two spirals (Fig. S7A). Reexamination of the native (noncross-linked) ceN-CC[S123E/I154W] samples prepared without an initial centrifugation step revealed similar filamentous structures on ammonium molybdate staining (Fig. 4C and Fig. S7B). The presence of two distinct spirals that open and recombine can be frequently observed in these filaments. The filament diameter, measured after two spirals can be seen to combine, is similar to the diameter of a single spiral (13 ± 3 nm, n = 1,047), which is consistent with the two spirals intertwining in a relatively tight arrangement (Fig. 5C).

Fig. 5.

A ceSAS-6 spiral can support ninefold symmetry. (A and B) Orthogonal views of a single ninefold symmetric ceSAS-6 spiral modeled with the full-length coiled coil. The individual ceSAS-6 dimers are numbered. The relative position of SAS-5 interaction sites (purple patches) is indicated. (C) Side view of two ninefold symmetric intertwined spirals in red and cyan with the ceSAS-6 dimers of the cyan spiral numbered. Note that the top view of two intertwined spirals would not differ from the view of a single spiral as shown in B. (D) Simplified representation of a C. elegans centriole showing our proposed model, where four turns of an intertwined ceSAS-6 double spiral account for the central tube (4).

In conclusion, we find no evidence of ring-like structures in either negative stain or rotary metal shadowing preparations of ceSAS-6 (Fig. S6), which is in contrast to similar constructs of crSAS-6 where cartwheel rings were readily observed (19). Instead, ceSAS-6 self-assembles into shorter isolated spirals or long filamentous structures that show evidence of spiral intertwining. We postulate that the filamentous form may be self-stabilized and preferentially populated after the spirals reach a certain length.

Discussion

Our crystallographic and EM studies established that ceSAS-6 self-assembles into a spiral arrangement. This spiral model stands in marked contrast to the SAS-6 rings that form cartwheels in C. reinhardtii (19), D. rerio (20), and Trichonympha sp. (21) centrioles. Our analysis suggests that the spiral conformation is compatible with a ninefold symmetric output; however, oligomer flexibility yields a broad distribution of symmetry outputs by ceSAS-6 alone. Similar flexibility was observed in ring-forming SAS-6 variants, which yield a variety of symmetry levels (8- to 11-fold) in vitro (19, 20). Although the role of SAS-6 in dictating centriolar ninefold symmetry is not disputed (16), we, therefore, surmise that additional components may assist in robustly enforcing this symmetry level in vivo. Thus, from a mechanistic perspective, centriolar architecture could be built upon either a SAS-6 spiral or ring foundation.

In contrast to 4.5-fold symmetric ceSAS-6 spirals, which could be modeled from the crystallographic data with minimal adjustments, modeling ninefold symmetric ceSAS-6 rings required substantial distortions of the ceSAS-6 structure (Fig. S4 B and E). Given the near-identity of the N-N interfaces of ceSAS-6, crSAS-6, and drSAS-6 (19, 20) as well as the high similarity of the C-C interfaces seen in this study, ceSAS-6 rings could be created only by reversing the displacement of the coiled coil helix-α3 (Fig. 1D). This reversal requires both large backbone changes at the pivot residues between the ceSAS-6 N-terminal domain and the coiled coil and overcoming strong steric clashes from residue V167. This residue is highly interesting, because V167 is conserved in most nematode species but substituted by small amino acids (G or A) in other SAS-6 variants (Fig. S8A). In silico substitutions of the corresponding residue in crSAS-6 (G158V) or drSAS-6 (G144V) yielded clashes between the valine methyl groups and the α3 main chain (Fig. S8C) similar to those clashes observed for a ceSAS-6 ring model (Fig. S4E). Thus, we consider V167 and the equivalent residues in other nematode variants as strong indicators for SAS-6 self-assembly into spiral conformations. Although such conformations could be overcome in vivo by, for example, strong protein–protein interactions involving the SAS-6 pivot region, no such SAS-6 binding proteins are known at this time.

Therefore, the intriguing possibility emerges that a SAS-6 spiral may be critical for building the central tube that is characteristic of C. elegans centrioles. Indeed, we presented a ceSAS-6 spiral model that displays ninefold symmetry every two turns, with an approximate length of 62 nm over this period (Fig. 5 A and B). This size is comparable with the initial length of the C. elegans central tube (∼60 nm) (4), whereas its full length observed on centriole maturation (∼110 nm) fits well to four turns of the ceSAS-6 spiral. Furthermore, the diameter of the spiral arrangement including idealized ceSAS-6 coiled coils is ∼72 nm, which is similar to the diameter observed for the central tube in vivo (4). Thus, we propose that C. elegans centrioles feature at their center a SAS-6 spiral (single or intertwined vide infra) that, in analogy to cartwheels in other species, is necessary for determining the conserved ninefold symmetry (Fig. 5D). Compared with cartwheel rings observed elsewhere, a ceSAS-6 spiral allows establishment of ninefold symmetry over considerable length using relatively few protein molecules. This scarcity of ceSAS-6 molecules may explain the seemingly hollow nature of the central tube observed in electron tomography (4); furthermore, a filament of ceSAS-6 N-terminal domains would be difficult to visualize by current in situ microscopy methods.

How widespread across evolution could a SAS-6 spiral be? The V167 pivot residue noted earlier is only present in certain nematode sequences (Fig. S8A) but absent from the SAS-6 of Trichinella spiralis, a worm that diverged early in the nematode evolutionary tree (24). This finding suggests that the SAS-6 spiral conformation is a relatively recent adaptation, restricted to parts of the nematode phylum. It is intriguing to speculate whether the aflagellate nature of the C. elegans sperm (25) and the lack of cilia in all C. elegans cells except sensory neurons (26) may have released a fundamental constraint in centriole assembly, which then allowed alternative SAS-6 conformations to arise.

What other differences would a spiral imply for centriolar architecture? Compared with cartwheel rings, a spiral may require different components to cap it at a predetermined length; likewise, the mechanical links between a spiral and microtubules will differ from the mechanical links of cartwheels. We note, for example, that a protein equivalent to C. reinhardtii Bld10p (27) or Homo sapiens Cep135 (28), which are thought to connect the cartwheel spokes with microtubule walls, has not been identified in nematodes. Furthermore, a SAS-6 spiral model would predict relatively few physical connections between the microtubule walls and the central tube, with direct implications on the mechanical stability of nematode centrioles. Importantly, by its very nature, an open-ended spiral oligomer of over 100 nm in length is inherently less stable than a closed ring arrangement. Intertwining of SAS-6 spirals (Fig. 5 C and D) as observed in our electron micrographs (Fig. 4C and Fig. S7) could intrinsically stabilize these oligomeric arrangements. At the same time, we postulate that other centriolar components [for example, SAS-5 (9)] may stabilize the overall conformation. Indeed, the SAS-5 interaction site was identified very recently at approximately the middle of the ceSAS-6 coiled coil by Qiao et al. (29) (Fig. 5 A and B) and independently by us (Fig. S9). SAS-5 includes a predicted coiled coil dimerization interface (9), and it has been shown to oligomerize in yeast two-hybrid experiments (30). It is, thus, possible that SAS-5 can bridge two adjacent binding sites on ceSAS-6, thereby stabilizing the formation of a ceSAS-6 spiral; a similar function has been proposed for the Drosophila melanogaster functional homolog of SAS-5 (31).

In summary, our findings show how small structural switches can drastically transform the architecture of SAS-6 oligomers, while retaining the same ninefold output symmetry. We believe that the fundamental role of SAS-6 in imposing centriolar ninefold symmetry is conserved in C. elegans like in other species. The emerging and intriguing question now is whether ceSAS-6 forms spirals in vivo.

Materials and Methods

An extended description of all methods is provided in SI Materials and Methods; briefly, protein fragments harboring N-terminal His₆-tags were expressed in Escherichia coli and purified by metal affinity and size exclusion chromatography. Protein identity was confirmed by mass spectrometry.

Size exclusion chromatography/multiangle light scattering experiments were performed on 2–4 mg/mL protein samples in PBS using the same setup as described previously (19). AUC velocity experiments were performed on 25–100 μM protein samples using an Optima XL-I analytical ultracentrifuge (Beckman). Sedimentation velocities were recorded by measuring either absorbance at 282 nm, with 200 scans every 4 min at 10 °C and 35,000 rpm, or fluorescence emission (λ_ex = 488 nm, λ_em = 505–565 nm) with the Fluorescence Detection System (Aviv) at 10 °C and 45,000 rpm. Data were processed using SEDFIT. The protein partial specific volume was calculated from the amino acid sequence. Fluorescence polarization measurements were either recorded with a 5-((2-((iodoacetyl)amino)ethyl)amino)naphthalene-1-sulfonic acid label using a PHERASTAR FS fluorimeter (λ_ex = 340 nm, λ_em = 520 nm; BMG Labtech) or for measurements of ceN[I154W] and ceN[S123E/I154W], the intrinsic fluorescence of the single tryptophan residue using an M5 Spectramax fluorimeter (λ_ex = 280 nm, λ_em = 340 nm; Molecular Devices).

Molecular dynamics of the ceSAS-6 coiled coil dimer were simulated for 20 ns at 310 K in explicit solvent with spherical boundary conditions using NAMD and the CHARMM27 force field. Protein crystals were obtained using the sitting drop vapor diffusion technique, data were recorded at Diamond Light Source or the European Synchrotron Radiation Facility, and structures were solved by molecular replacement using the ceN[S123E] crystal structure (Protein Data Bank ID code 3PYI) as the search model. Crystallographic data processing and refinement statistics are provided in Table S1.

For EM, protein samples diluted to 0.1–1 mg/mL were transferred to freshly glow-discharged homemade carbon-coated copper grids and stained by uranyl acetate or ammonium molybdate solutions. To avoid disassembly of ceSAS-6 oligomers, all steps were carried out swiftly.

Generation of mutant nematode strains and analysis of resistance to SAS-6 RNAi were carried out as described previously (19).