Structures of SAS-6 Suggest Its Organization in Centrioles

See allHide authors and affiliations

Science  04 Mar 2011:
Vol. 331, Issue 6021, pp. 1196-1199
DOI: 10.1126/science.1199325


Centrioles are cylindrical, ninefold symmetrical structures with peripheral triplet microtubules strictly required to template cilia and flagella. The highly conserved protein SAS-6 constitutes the center of the cartwheel assembly that scaffolds centrioles early in their biogenesis. We determined the x-ray structure of the amino-terminal domain of SAS-6 from zebrafish, and we show that recombinant SAS-6 self-associates in vitro into assemblies that resemble cartwheel centers. Point mutations are consistent with the notion that centriole formation in vivo depends on the interactions that define the self-assemblies observed here. Thus, these interactions are probably essential to the structural organization of cartwheel centers.

Centrioles are ninefold symmetric, cylinder-shaped structures found in most animal cells. They template the nine microtubule doublets in cilia and flagella and constitute the core of the centrosome—the dominant microtubule organizing center in animal cells [reviewed in (1)]. Many diseases like ciliopathies are associated with functional abnormalities in these structures [reviewed in (2)]. In most organisms, immature centrioles (called procentrioles) consist of short microtubule triplets and a structure called the cartwheel, which is made up of a central ringlike hub and nine spokes radiating from the hub [reviewed in (1, 3, 4)] (Fig. 1A). Cartwheel formation occurs at a very early stage of centriole assembly, followed by formation of the peripheral microtubules (57). Thus, the cartwheel might serve as a scaffold that determines centriole diameter and symmetry through the radial arrangement of its nine spokes. This idea is supported by experiments with the two known cartwheel components SAS-6 and CEP135/Bld10. CEP135/Bld10 localizes to the distal part of the cartwheel spokes (Fig. 1A) (810) and is essential for centriole formation (811). Its truncation in Chlamydomonas results in aberrant centrioles with nine shortened cartwheel spokes and a decreased diameter that can accommodate only eight triplet microtubules (9). SAS-6, the other known cartwheel component, is also crucial for centriole assembly (1022), but localizes to the cartwheel hub (10, 12) and is required for hub formation (10, 12). Lack of SAS-6 in Chlamydomonas (12), Drosophila (14), and Paramecium (10) results in the formation of centrioles with an aberrant number of triplet microtubules, leading to the hypothesis that SAS-6 might self-assemble to form the cartwheel hub and thereby dictate centriole symmetry (3). The molecular structure of cartwheel hubs and SAS-6 and how they direct centriole symmetry is unknown.

Fig. 1

Structure and head-to-head dimerization of the N-terminal domain of SAS-6. (A) Overview of the in vivo localization of SAS-6. WT centriole scheme with peripheral microtubule (Mt.) triplets and central cartwheel structure (Cw.) with hub and spokes. The location of SAS-6 is shown in red (12). The localization of CEP135/Bld10 (8, 9) is indicated by an arrow. (B) Domain organization of D. rerio SAS-6. The used constructs and the type of experiment performed with them are indicated underneath. KEN box, recognition motif involved in the cell-cycle–dependent degradation of SAS-6 (21); MALS, multi-angular light scattering. (C) Structure of the N-terminal domain of SAS-6 shown in a ribbon diagram and rainbow-colored from N (blue) to C terminus (red). α helices (α), β sheets (β), and loops (L) are numbered sequentially. α3 corresponds to the start regions of the predicted coiled-coil domain in SAS-6. (D) Same as in (C), but as molecular surface with projected sequence conservation denoted by colors ranging from light blue (variable) to burgundy (conserved). The conservation pattern was calculated with ConSurf (35) using a Bayesian method for computing the conservation scores. The color scale represents the conservation scores. (E) Ribbon presentation of the observed head-to-head dimer of the N-terminal domain of SAS-6 with details of the dimerization interface. The hydrophobic F131 residue inserts into a hydrophobic pocket. Residues that make contact are labeled and shown in sticks for one of the two equivalent interface regions. Hydrogen bonds are indicated by yellow dashed lines.

SAS-6 consists of a conserved N-terminal domain, a central coiled-coil domain, and a less conserved C-terminal region (Fig. 1B). Because soluble recombinant full-length SAS-6 could not be obtained in sufficient quantities, we determined the three-dimensional structure of the N-terminal domain of Danio rerio (zebrafish, Dr) SAS-6 (N-SAS-61-156) at 1.92 Å by x-ray crystallography (Fig. 1, B and C, and tables S1 and S2) (23). The asymmetric unit comprises four N-SAS-61-156 molecules that superimpose well (fig. S1A). The N-terminal domain of SAS-6 adopts a seven-stranded β-barrel structure capped by two α helices (α1 and α2) at one end that form a helix-turn-helix–like (HTH) motif. The C-terminal helix (α3) that corresponds to the start region of the predicted coiled-coil domain of SAS-6 docks on the barrel wall and makes several polar and hydrophobic contacts with the barrel residues. Notably, despite the absence of sequence similarity, the structure of N-SAS-61-156 shows structural similarity to the N-terminal domains of XRCC4 and XLF, both implicated in DNA repair by nonhomologous end-joining (fig. S1B) (2427).

The N-terminal domain of SAS-6 contains two highly conserved regions (fig. S2) (28): (i) the PISA-motif (18) and (ii) a second motif (motif II) that forms the β hairpin β6-β7 that includes the protruding loop 8 (Fig. 1, C and D). The two motifs constitute a continuous highly conserved patch on the N-SAS-6 surface (Fig. 1D and fig. S1, C to E). In the crystal, two N-SAS-6 domains interact to form a head-to-head dimer through this patch (Fig. 1E; see fig. S1, F and G, for a stacking interaction between head-to-head dimers in the crystal). In this head-to-head dimer, F131 in loop 8 from each monomer is inserted into a hydrophobic pocket on its binding partner formed by the conserved L76, I78, F83, L90, and L137 residues (29) and the aliphatic side chain of K86 (Fig. 1E). The dimer is further stabilized by a hydrophobic contact between residues L134 from both subunits. Additionally, K132 and the main-chain carbonyl group of F131 form hydrogen bonds with the main-chain groups of Q74/T135 and L77, respectively. The area buried on dimer formation is ~1500 Å2, making dimerization a candidate for a biologically relevant interaction (30, 31).

Analytical gel filtrations with the recombinant N-terminal domains of D. rerio (N-SAS-61-156) and Homo sapiens (N-SAS-61-164) suggested possible dimerization of this domain in solution. These proteins eluted at lower molecular sizes when their hydrophobic F131 residue was replaced by aspartic acid, which should weaken head-to-head dimerization (fig. S3, A and B). Analytical ultracentrifugation with H. sapiens N-SAS-61-164 and D. rerio N-SAS-61-156 showed that the F131→D131 (F131D) mutants were monomeric, whereas the wild-type (WT) proteins were present in a monomer-dimer equilibrium with apparent dissociation constants of 50 μM (Homo sapiens) and 90 μM (Danio rerio) (Fig. 2A and fig. S3C). These results are consistent with a low-affinity, head-to-head dimerization in solution.

Fig. 2

Head-to-head dimerization of SAS-6 occurs in solution and is crucial for centriole formation. (A) Sedimentation equilibrium analysis of the molecular weight of human N-SAS-61-164 in solution (calculated molecular weight = 19.8 kD). (Top) Apparent molecular weight as a function of concentration. Replicates of the WT protein (blue and red traces) and the F131D mutant (black trace) are shown. The smooth traces were produced by a regularization algorithm implemented in the UltraSpin software. (Bottom) Residuals of the fit of the interference data to single- (for the F131D mutant) and double-component (for the wild type) models. (B) Projected fluorescence microscopy images of human U2OS cells expressing N-terminally green fluorescent protein–tagged full-length human SAS6 (wild type and F131D mutant) together with the RFP-tagged C-terminal PACT domain of pericentrin as a centrosomal marker (36). Constructs were transiently transfected into U2OS cells and analyzed by fluorescent microscopy 48 hours posttransfection. Maximum-intensity projected images of transfected cells are shown. Scale bar, 10 μm. (C) Functional analyses of mutated SAS-6 using a Chlamydomonas ΔSAS-6 mutant. The C-terminally haemagglutinin (HA) epitope–tagged WT and mutated SAS-6 genes were introduced into the bld12-1 mutant, and clones expressing each protein were isolated from the transformants. The table shows the percentages of those cells in each clone with zero, one, or two flagella. The percentages of flagellated cells in the bld-12 strain transformed with WT SAS-6 are lower than those in the WT Chlamydomonas strain. This may be due to the presence of the HA tag in these constructs. N, number of cells counted.

When transiently overexpressed in human U2OS cells, human full-length WT SAS-6 localizes to centrosomes (Fig. 2B) (18) and cytoplasmic foci (18). In contrast, the corresponding SAS-6 F131D mutant was found to be dispersed in the cytoplasm and localized less efficiently to centrosomes (Fig. 2B). This was also observed with the L76D or I78E (V78E in human) mutants that are located in a short loop region of the hydrophobic pocket, into which F131 inserts (fig. S4, B and C).

We tested the Chlamydomonas equivalent of the F131D mutant (F145D) (fig. S2) for its functional effect on centriole formation in Chlamydomonas reinhardtii. This organism has two flagella that are templated by the two matured centrioles found in each cell. Thus, problems in centriole formation are revealed by aberrant flagellar formation. Because a null mutant of SAS-6, bld12, has severe defects in centriole formation, ~90% of cells lack flagella (Fig. 2C) (12). When transformed with a cDNA construct coding for full-length WT Chlamydomonas SAS-6, the percentage of flagellated bld12 cells increased to 66%; in contrast, the F145D mutant of SAS-6 showed no recovery in flagellar formation (Fig. 2C), although SAS-6 expression levels were comparable (fig. S5).

The similarity between SAS-6, XLF, and XRCC4 suggests that SAS-6, like XLF (24, 25) and XRCC4 (26, 27), could form a parallel coiled-coil dimer. To avoid potential complications in crystal packing due to the presence of two dimerization interfaces, we used the F131D mutation to disrupt the head-to-head dimerization of SAS-6. We determined the crystal structure of Dr N-SAS-61-179 F131D at 1.98 Å (Fig. 3, A and B, and table S2). Two of the three molecules present in the asymmetric unit (fig. S6, A and B) formed a parallel coiled-coil dimer (Fig. 3A) through a canonical knob-into-holes packing of hydrophobic residues of the coiled-coil stalks that is further stabilized by a side-chain–to–side-chain H-bonding network between D146, T147, and K150 and a salt bridge between E163 and K164 (Fig. 3B and fig. S6C). Consistent with the structural model, light scattering measurements with Dr N-SAS-61-326 F131D, which contains ~56% of the predicted coiled-coil domain of SAS-6, suggest that SAS-6 stably dimerizes in solution through an elongated coiled-coil domain (fig. S7).

Fig. 3

SAS-6 forms a parallel coiled-coil dimer. (A) Structure of the N-SAS-61-179 F131D dimer drawn in ribbon presentation with subunits shown in green and red. (B) Detailed view of the dimer interface. The dimer is stabilized by a canonical hydrophobic packing (some of the involved residues are labeled and shown as sticks); a H-bonding network between D146, T147, and K150; and a salt bridge between E163 and K164 (dashed yellow lines).

Removal of ~90% of the predicted coiled-coil region of human SAS-6 (SAS-6 Δ167-466) led to its strong mislocalization in U2OS cells (fig. S4D). We also disturbed the coiled coil of SAS-6 and its packing against the head domains (fig. S6D) by introducing a bulky tryptophan residue at the L153 equivalent (L167W) in Chlamydomonas SAS-6 and assayed its effect on the ability to rescue the bld12 phenotype. As shown in Fig. 2C, expression of the mutant SAS-6 slightly increased the number of flagellated cells (21%), but the percentage was much less than that of the clone transformed with WT SAS-6, although SAS-6 expression levels were comparable (fig. S5). Circular dichroism spectroscopy results suggest that the L153W mutation does not disturb the fold of SAS-6 (fig. S8).

A SAS-6 construct containing both dimerization interfaces forms higher-order oligomers in solution, as judged by native mass spectrometry (fig. S9). Our structural models suggest that these oligomers may be curved (Fig. 4A). A purified construct containing both interfaces (Dr N-SAS-61-217) crystallized spontaneously in solution, giving rise to small, low-diffraction quality crystals. Cryo-electron microscopy (cryo-EM) images (Fig. 4B) of these crystals revealed notable ring-shaped (or, less likely, spiral-shaped) assemblies. These assemblies contain 16 distinct, slightly oval blobs that seem to stack on each other, resulting in long tubes connected by radially projecting spokes. The lengths of the spokes are consistent with the predicted coiled-coil length (~10.5 nm) of Dr N-SAS-61-217. The diameter of these assemblies is ~217 Å, as measured from the center of one oval blob to an equivalent position across the ring center.

Fig. 4

Model of SAS-6 ring assembly. (A) Ribbon presentation of a modeled SAS-6 tetramer based on the observed coiled-coil and head-to-head dimers. The distance between the base regions of the two coiled-coil domains is indicated. (B) Cryo-EM image of a face-on view of a thin crystal of N-SAS-61-217. Pixel size, 3.74 Å per pixel; scale bar, 60 Å. The image was Fourier-filtered and symmetry-averaged. The SAS-6 tetramer presented in (A) is shown as an overlay. The overlayed structure is based on N-SAS-61-179, which has a shorter coiled coil than N-SAS-61-217. (C) Models of SAS-6 rings with different symmetries. The approximate diameters of these rings (the double distance from head-domain center to ring center) are indicated above the modeled rings. The diameter of cartwheel hubs observed in procentrioles by cryo-electron tomography (32) is shown as dotted circles. To model these rings, we allowed for a change in the orientation of the head domains relative to the coiled-coil domain. To compare the required changes, we calculated the root mean square deviation (RMSD) between the N-SAS-61-179 structure and two equivalent head domains in the modeled ring. These values are indicated under the modeled rings.

The modeled N-SAS-61-179 tetramer shown in Fig. 4A overlays well with four of these blobs (Fig. 4B). Tube formation may be based on stacking interactions between the head-to-head dimers, similar to those that we observed in the N-SAS-61-156 crystal (fig. S1F). In our overlay in Fig. 4B, the coiled-coil domain in the crystal structure would need to be at an angle to better fit the EM density. This could be accommodated by some flexibility between the head domains and the coiled-coil domain of SAS-6. Allowing for this flexibility, we modeled ring assemblies containing 6 to 12 SAS-6 dimers (Fig. 4C). Compared with the N-SAS-61-179 F131D crystal structure, the eight- and ninefold symmetric rings required the smallest changes in the orientation of the head domains. With their radially projecting, stalklike coiled-coil domains, these ring assemblies resemble centriolar cartwheels. The inner diameter of the modeled ninefold symmetric ring was comparable to the diameter of the cartwheel hubs observed in procentrioles by cryo-electron tomography (200 Å) (32). Consistent with our ring models with projecting coiled-coil domains, the C-termini of SAS-6 in Chlamydomonas are found in the outer regions of the centriolar cartwheel spokes (fig. S10).

No cartwheel has been identified so far in Caenorhabditis elegans. Instead, a central centriolar tube was found, whose presence requires SAS-6 (33). This alternative assembly could be due to structural differences in SAS-6, as motif II in C. elegans and C. briggsae is very distinct from the other SAS-6 homologs [fig. S2 and (28)].

We suggest that SAS-6 self-assembly is a contributing factor in the structural organization of centriolar cartwheels cores. However, our data also suggest that other centriolar components are probably needed for a faithful and stable ninefold symmetric SAS-6 assembly in vivo. This notion is in agreement with the presence of alternative SAS-6 assemblies that are observed when SAS-6 is overexpressed in Drosophila (14, 34).

Supporting Online Material

Materials and Methods

Figs. S1 to S10

Tables S1 and S2


References and Notes

  1. Materials and methods are available as supporting material on Science Online.
  2. Single-letter abbreviations for the amino acid residues are as follows: A, Ala; C, Cys; D, Asp; E, Glu; F, Phe; G, Gly; H, His; I, Ile; K, Lys; L, Leu; M, Met; N, Asn; P, Pro; Q, Gln; R, Arg; S, Ser; T, Thr; V, Val; W, Trp; and Y, Tyr.
  3. The structure factors and coordinates of D. rerio N-SAS-61-156 and D. rerio N-SAS-61-179 F131D have been deposited at the Protein Data Bank Europe under wwPDB ID codes 2y3v and 2y3w. We gratefully acknowledge P. Evans (MRC-LMB, Cambridge, UK) for his modeling of SAS-6 ring assemblies and our beamline support team, M. Nanao at ID29 and H. Belrhali at BM14UK at the European Synchrotron Radiation Facility (Grenoble, France). We thank L. Rey (MRC-LMB) for the kind gift of U2OS cells and S. Munro (MRC-LMB) for the kind gift of a construct expressing the red fluorescent protein (RFP)–tagged C-terminal PACT domain of pericentrin. This work was supported by a MRC Career Development Fellowship (to M.v.B.); Grants-in-Aid for Scientific Research from the Ministry of Education, Culture, Science and Technology of Japan (21370088) (to M.H.); and a European Molecular Biology Organization long-term fellowship (to B.Z.). Material will be provided to academic and not-for-profit research organizations under the MRC’s standard academic Material Transfer Agreement. The UltraSpin software will be provided by MRC under an academic license.
View Abstract

Navigate This Article