The conserved Spd-2/CEP192 domain adopts a unique protein fold to promote centrosome scaffold assembly

Abstract

Centrosomes form when centrioles assemble pericentriolar material (PCM) around themselves. Spd-2/CEP192 proteins, defined by a conserved "Spd-2 domain" (SP2D) comprising two closely spaced AspM-Spd-2-Hydin (ASH) domains, play a critical role in centrosome assembly. Here, we show that the SP2D does not target Drosophila Spd-2 to centrosomes but rather promotes PCM scaffold assembly. Crystal structures of the human and honeybee SP2D reveal an unusual "extended cradle" structure mediated by a conserved interaction interface between the two ASH domains. Mutations predicted to perturb this interface, including a human mutation associated with male infertility and Mosaic Variegated Aneuploidy, disrupt PCM scaffold assembly in flies. The SP2D is monomeric in solution, but the Drosophila SP2D can form higher-order oligomers upon phosphorylation by PLK1 (Polo-like kinase 1). Crystal-packing interactions and AlphaFold predictions suggest how SP2Ds might self-assemble, and mutations associated with one such potential dimerization interface markedly perturb SP2D oligomerization in vitro and PCM scaffold assembly in vivo.

Fig. 1.. Domain organization of Spd-2/CEP192 proteins.

(A) Schematics illustrate the domain organization of H. sapiens (Hs) CEP192 and D. melanogaster (Dm) Spd-2 proteins. (B) Ribbon representation of structures determined for DmASH3 and HsASH7 by crystallography and DmASH1 by NMR spectroscopy.

Fig. 2.. The N-terminal region of DmSpd-2 is sufficient for centriole targeting, and the C-terminal region is required for PCM scaffold assembly.

(A) Schematics illustrate the Spd-2 truncation mutants tested here. (B) Western blot shows the expression levels of GFP fusions to WT or the truncated versions of Spd-2 in early embryos; actin is shown as a loading control. NT, N-terminal; CT, C-terminal. (C) Graphs quantify the hatching percentage of embryos laid by Spd-2^−/− females expressing a GFP fusion of either WT Spd-2 or the truncation mutants. n = 5 technical repeats; n ≥ 100 embryos per technical repeat. (D) (a) Confocal images illustrate, and the bar chart below quantifies, the centrosomal fluorescence intensity (means ± SD) of the WT and truncated Spd-2-GFP fusions in WT embryos expressing the centriole marker Asl-mCherry and injected with mRNA encoding each GFP fusion. Note that endogenous, unlabeled Spd-2 is also present in these embryos. Statistical significance was assessed using an unpaired t test in GraphPad Prism (****P < 0.0001). a.u., arbitrary units. (b) Graphs show the raw (left) or normalized (right) centrosomal fluorescence distribution profiles of the WT or truncated Spd-2-GFP fusions. (E) 3D-SIM images of centrosomes in living embryos expressing the mother centriole marker Asl-mCherry (magenta) injected with mRNAs encoding either Spd-2 WT-GFP or its truncated mutants (green). (F) (a) 3D-SIM images from a FRAP experiment show the dynamic behavior of WT Spd-2-GFP and Spd-2-NT-GFP at centrosomes. The time (seconds) after photobleaching is indicated. (b) Graphs show the normalized centrosomal fluorescence distribution profiles of the WT or truncated Spd-2-GFP fusions at successive time points after photobleaching. Note how the WT protein appears to spread outward, but Spd-2-NT-GFP remains tightly concentrated around the centriole. (G) 3D-SIM images of centrosomes in embryos coexpressing Spd-2-NT-GFP and full-length WT RFP-Spd-2. Scale bars in all images, 2 μm.

Fig. 3.. The SP2D, and both its constituent ASH domains, is required for DmSpd-2 scaffold assembly, but ASH3 is not.

(A) Schematics illustrate the deletion mutants of Spd-2 tested here. (B) Western blot shows the expression levels of GFP-fusions to the various deletion mutants shown in (A); actin is shown as a loading control. The expression levels of GFP fusions to the WT and N-terminal and C-terminal truncations (same blot as shown in Fig. 2B) are shown side by side here for ease of comparison. (C) Graphs quantify the hatching percentage of embryos laid by Spd-2^−/− females expressing GFP fusions of either WT Spd-2 or the various deletion mutants. n = 5 technical repeats; n ≥ 100 embryos per technical repeat. (D) (a) Confocal images illustrate, and the bar chart below quantifies, the centrosomal fluorescence levels (means ± SD) of the WT and Spd-2 deletion GFP fusions in WT embryos expressing the centriole marker Asl-mCherry injected with mRNA encoding each protein. Statistical significance was assessed using an unpaired t test in GraphPad Prism (**P < 0.01 and ****P < 0.0001). (b) Graphs show the raw (left) or normalized (right) centrosomal fluorescence distribution profiles of the WT or Spd-2 deletion GFP fusions. (E) 3D-SIM images from a FRAP experiment show the dynamic behavior of WT Spd-2-GFP and Spd-2 ΔASH3-GFP at centrosomes. The time (seconds) after photobleaching is indicated. Graphs show the normalized centrosomal fluorescence distribution profiles of WT Spd-2-GFP and Spd-2 ΔASH3-GFP at successive time points after photobleaching. Scale bars in all images, 2 μm.

Fig. 4.. Spd-2 can assemble an SP2D-dependent scaffold structure in the absence of Cnn.

(A) 3D-SIM images show the behavior of WT Spd-2-NG in cnn^−/− embryos in the presence or absence of colchicine (Colch), which depolymerizes the centrosomal MTs. Arrows highlight small “flares” of Spd-2-NG that rapidly flux away from the centriole on the centrosomal MTs in the cnn^−/− embryos; this outward flux is suppressed if the MTs are depolymerized, and Spd-2-NG forms a scaffold around the centriole in these embryos. (B) Confocal images illustrate, and the bar chart below quantifies, the centrosomal fluorescence levels (means ± SD) of WT and Spd-2 deletion GFP fusions in WT embryos (left panels) and cnn^−/− embryos (all remaining panels) expressing the centriole marker Asl-mCherry. Statistical significance was assessed using an unpaired t test in GraphPad Prism (****P < 0.0001 and ns, not significant). Scale bars in all images, 2 μm.

Fig. 5.. The overall structure of the SP2D is highly conserved.

(A) Schematics illustrate the domain organization of H. sapiens (Hs) CEP192 and A. dorsata (Ad) Spd-2. (B) Ribbon representation of two views of the x-ray crystal structures of HsSP2D (containing ASH4 and ASH5) (left panels, green in central overlay panels) and AdSP2D (containing ASH4 and ASH5) (right panels, yellow in central overlay panel). The central α1 helix of the “Bridge” structure—formed by an insertion between strands β-f and β-g in the β-sandwich of the second ASH domain—is highlighted. This structure forms extensive interactions with the first ASH domain that hold the SP2D in its characteristic extended cradle conformation.

Fig. 6.. The interactions that maintain the structure of the bridge and the extended cradle of the SP2D are highly conserved.

Panels show ribbon and stick representations of the x-ray crystal structures of Hs (left panels) and Ad (right panels) SP2Ds, highlighting some key interactions that maintain the extended cradle structure. (A) View of the interactions within the bridge element that packs between the two ASH domains of SP2D. The bridge is shown in orange, and labeled in red are the residues that form a hydrophobic core around one side of the α1 helix. Further stabilization is provided by a highly conserved network of electrostatic interactions (dotted lines) centered around R1993/R1195 and E1990/E1192 (human/honeybee) in motif 2 (which is highlighted in purple). (B) Views on the packing interactions centered around the conserved motifs 1 (pink) and 2 (purple). Interactions discussed in the main text are highlighted in different colors. Note that, although several conserved residues within these motifs contribute to the stability of the structure, the precise interactions they form partly vary between the HsSP2D and AdSP2D structures (see main text for details).

Fig. 7.. Mutations predicted to disrupt the Drosophila SP2D elongated cradle structure perturb Spd-2 scaffold assembly in vivo.

(A) Ribbon diagram of the AF2-predicted structure of the D. melanogaster (Dm) SP2D (rainbow color), overlaid with the Apis crystal structure (gray). The inset highlights the core interaction interface that involves hydrophobic (highlighted in blue labels) and electrostatic interactions (dotted lines) from motifs 1 and 2, which are similar to the stabilizing interactions observed in the Apis structure (Fig. 6B). (B) (a) Confocal images illustrate, and the bar chart below quantifies, the centrosomal fluorescence levels (means ± SD) of the WT and various Spd-2 mutant GFP fusions in WT embryos expressing the centriole marker Asl-mCherry and injected with mRNA encoding each protein. Statistical significance was assessed using an unpaired t test in GraphPad Prism (***P < 0.001 and ****P < 0.0001; ns, not significant). (b) Graphs show the raw (left) or normalized (right) centrosomal fluorescence distribution profiles of the WT or Spd-2-3A mutant (D920A/N923A/R926A) GFP fusions. Scale bar, 2 μm.

Fig. 8.. A human disease substitution (N1917S) located near the interaction interface between the two ASH domains of SP2D mildly perturbs Drosophila Spd-2 scaffold assembly in vivo.

(A) Views of the interaction network (dotted lines) made by the side chain of the conserved Asn1917/Asn1118/Asn844 (labeled in magenta) in humans, Apis, and Drosophila, respectively, that is substituted for a Ser in human patients. (B) CD analysis of DmSP2D WT and N844S shows that this substitution appears to slightly alter the protein fold. (C) (a) Confocal images illustrate, and the bar chart quantifies, the centrosomal fluorescence levels (means ± SD) of WT and Spd-2-N844S mutant GFP fusions in WT embryos expressing the centriole marker Asl-mCherry and injected with mRNA encoding each protein. Statistical significance was assessed using an unpaired t test in GraphPad Prism (****P < 0.0001). (b) Graphs show the raw (left) or normalized (right) centrosomal fluorescence distribution profiles of the WT and Spd-2-N844S GFP fusions. Scale bar, 2 μm.

Fig. 9.. A hydrophobic interaction interface may allow insect SP2Ds to form dimers.

(A) (a) SEC-MALS analysis of Sumo-tagged SP2D WT, P817D, or F822D performed at a range of protein concentrations (as indicated). Note how the WT, but not mutant, protein(s) has/have a tendency to form higher MW species at higher concentrations. (b) SEC-MALS analysis of Sumo-tagged Spd-2-CT (697 to 1146 amino acids) WT, P817D, or F822D that has been phosphorylated in vitro with recombinant human PLK1 kinase. The orange horizontal dotted lines indicate the theoretical mass of a monomer or dimer. Note how the WT, but not mutant, protein(s) can form large species when phosphorylated by PLK1. (B) Ribbon diagram showing the packing interactions in the top-ranked AF3 prediction of a DmSP2D dimer (purple) overlaid with the AdSP2D dimer (cyan) observed in crystallo; a similar hydrophobic interface observed in both structures is highlighted in orange and is shown in detail in the inset. (C) CD analysis of SP2D WT, P817D, or F822D shows that each single-point mutation does not strongly affect the protein fold. (D) (a) Confocal images illustrate, and the bar chart below quantifies, the centrosomal fluorescence levels (means ± SD) of WT and Spd-2-P817D or F822D mutant GFP fusions in WT embryos expressing the centriole marker Asl-mCherry and injected with mRNA encoding each protein. Statistical significance was assessed using an unpaired t test in GraphPad Prism (****P < 0.0001). (b) Graphs show the raw (top) or normalized (bottom) centrosomal fluorescence distribution profiles of the WT and Spd-2-P817D or F822D GFP fusions. Scale bar, 2 μm.