Centrosome separation and satellite movement require Motif 1 of Cnn. Live imaging of cnn null embryos expressing GFP-Cnn^WT (cnn^WT, A–F) or GFP-Cnn^Δ1 (cnn^Δ1, G–L). Cropped stills of movies at metaphase (A and G), anaphase B (B and H), centrosome duplication/telophase (C and I), early centrosome separation (D and J), separation/prophase (E and K), and metaphase of the next cycle (F and L). The time code of these stills, taken from Supplementary Movie 1 (A–F) and Supplementary Movie 2 (G–L), are shown in the lower left in seconds. Small arrows point to centrosomes beginning at one metaphase spindle (A and G) and follows them through one cleavage cycle (Supplementary Movie 1 progresses through three cycles, and Supplementary Movie 2 through four cycles). Each of the two centrosomes indicated with arrows in the cnn^WT frames divides once (C) and separates (E), and the nascent pair contributes to a bipolar spindle in the ensuing metaphase (F). In contrast, the cnn^Δ1 centrosomes indicated with arrows in G are two pairs at each pole of a metaphase spindle that failed to separate in the previous cycle. These pairs divide, producing two clusters of four centrosomes (J) that fail to separate (K). One pair of centrosome pairs did separate, whereas the other remained as a cluster of four in the next metaphase (L). The asterisk (*) in G indicates two pairs of attached centrosomes that divide to produce a cluster of eight centrosomes that failed to separate (K). More centrosome clusters form in subsequent cycles (see Supplementary Movie 2). The arrowheads in E and K point to centrosomal satellites, which are abundant in cnn^WT embryos but scarce in cnn^Δ1 embryos. The large arrows in E, F, and K indicate the intracentrosomal fibers that connect separating centrosomes.

D-TACC and Msps are localized improperly at cnn^Δ1 centrosomes. Wild-type (A, E, I, and M), cnn null (B, F, J, and N), cnn^WT (C, G, and K) and cnn^Δ1 (D, H, L, and P) embryos at early syncytial blastoderm cleavage stage stained for Aurora A (A–D), D-TACC (E–H), Msps (I–L) or P-D-TACC (M–P). All samples were counterstained for microtubules (green) and DNA (blue). Below each row is displayed the channel for Aurora A (Aur-A), D-TACC, Msps and P-D-TACC (all are red in the merged images). Insets are shown to highlight the localization of Aurora A, D-TACC, and Msps at spindle poles. Arrows in (F, H, J, and L) point to centrioles/centrosomes that are free or displaced from the spindle pole. The arrows in P indicate centrosomes with P-D-TACC signal at the centrosome periphery. All embryos shown are in cycles 10–12 except for (Q and R), which appear to be at later cycles. In cnn null mutant embryos the cycle number is unclear after cycle 10 or 11 because of the high degree of nuclear loss, disorder, and aneuploidy. In early cortical cycles (cycles 10 and 11) centrioles are detected at cnn null mutant spindle poles. In later cycles (Q and R), where centrioles are lost from spindle poles, D-TACC accumulates in particles that localize near microtubule bundles in the proximity of mitotic chromosomes (arrows in Q). Bar, 10 μm.

Localization of CP60 and Nek2 to centrosomes does not require Cnn Motif1. Wild-type, cnn null, cnn^WT, and cnn^Δ1 embryos were stained for α-tubulin (green), DNA (blue), and CP60 (top row, red) or Nek2 (bottom row, red). Arrowheads in bottom row point to the dot of Nek2 signal at the centrioles. For cnn null embryos, the Nek2 signal is present in early syncytial blastoderm embryos shown here, but is lost at later cleavage stage embryos due to loss of centrioles (not shown). Bar, 10 μm.

Cleavage furrow assembly from cnn^Δ1 mutant centrosomes. Actin is organized aberrantly into pseudocleavage furrows in cnn^Δ1 mutant embryos. Wild-type (A), cnn null (B), and cnn null cleavage stage embryos with GFP-Cnn^WT (cnn^WT, C, E, and G) and GFP-Cnn^Δ1 (cnn^Δ1, D, F, and H) at metaphase stained for filamentous actin, Cnn, and DNA. The furrow organization is highly irregular in the cnn^Δ1 mutant, with variable integrity and size of furrows. Incomplete furrows are frequently formed (D). The depth of the furrow ingression is variable in the cnn^Δ1 mutant, but can achieve the same depth as the cnn^WT control (compare E and F). Arrows in E and F highlight the difference in cortical actin distribution in cnn^WT and cnn^Δ1 embryos at metaphase. The white arrowheads in F show nuclei that have dropped from the cortex. Embryos in G and H were stained for Cnn, microtubules, and DNA. Yellow arrows in D and H indicate mitotic figures that are rotated 90° from the normal parallel orientation to the embryo cortex. The yellow arrow in H indicates a misoriented half spindle, nearby to one that is oriented correctly (yellow arrowhead) and another that is splayed parallel and perpendicular to the cortex (blue arrow). Note that one of the spindles in G is at an oblique angle, such that the posterior half of the spindle is not included in the image stack. Bar, 10 μm.

Motif 1 of Cnn is essential. (A) Schematic of Cnn protein showing the two conserved motifs at the amino and carboxyl termini. An expanded view of Motif 1 is shown with the 13 amino acids that are deleted in Cnn^Δ1 underlined in red. The alignments include the putative Cnn homologues in Aedes egypti (mosquito, A.e.), CDK5RAP2 from Gallus gallus (chicken, G.g.) and human (H.s.), Myomegalin/PDE4DIP from human, and Mto1p from S. pombe (S.p.). (B) Western blot showing the expression of GFP-Cnn^WT and GFP-Cnn^Δ1 in a cnn null mutant (cnn^hk21/cnn^25cn1) background (labeled “cnn-” above the lanes). The samples were loaded onto two gels and blotted, and one gel was probed with anti-Cnn, anti-γ-Tub, and anti-α-Tub (loading control) antibodies as indicated. The second blot, shown below the solid line, was probed with anti-GFP antibody. Immunostaining of cnn null embryos expressing GFP-Cnn^WT (C) and GFP-Cnn^Δ1 (D) shows localization to centrosomes and mutant phenotype. Anti-GFP signal is green, anti-α-Tub signal is red, and DNA is blue. The erratic spacing of mitotic figures, linked spindles, and aneuploid nuclei in D are all characteristic of cnn mutant embryos, showing that GFP-Cnn^Δ1 did not rescue. Note the abundance of centrosome pairs or clusters in the cnn^Δ1 mutant embryo.

Cnn motif 1 recruits γ-tubulin to centrosomes. Wild-type (A), cnn null (B), and cnn null embryos with GFP-Cnn^WT (cnn^WT, C) and GFP-Cnn^Δ1 (cnn^Δ1, D) at early syncytial blastoderm cleavage stained for γ-Tub (separate channel shown below merged images). The centrosome signals in C and D are yellow compared with the signal for wild-type (A) because of the green signal from GFP-Cnn fusion proteins (red plus green overlap is yellow). The level of γ-Tub localized at cnn^Δ1 mutant centrosomes is reduced, yet similar to the levels seen at cnn null embryo spindle poles. Note the unseparated centrosomes in the cnn^Δ1 embryo on one spindle pole and none on the opposite spindle pole (arrows in D).

Cnn and D-TACC colocalization at centrosomes. cnn^WT (A) and cnn^Δ1 (B) embryos were stained for Cnn (green), D-TACC (red), and DNA (blue). Two close-up views of each are shown below, with the separate channels for Cnn and D-TACC signals. In the cnn^Δ1 mutant D-TACC localized at and near centrosomes, but with higher levels of accumulation at the immediate periphery of the centrosome (arrows in B and in close-up images). There is also a higher level of D-TACC associated with free centrosomes (arrowheads in B). The “Coloc” images show colocalized Cnn and D-TACC signals. The white signal in each image represents colocalized signals above a pixel intensity threshold of 150 (see Materials and Methods). Bar, 10 μm.

cnn^Δ1 is a dominant mutation. Overexpression of GFP-Cnn^Δ1 in a wild-type cnn background (B and D) has dominant phenotypes compared with GFP-Cnn^WT overexpression (A and C). Only mild cleavage failure defects were seen with GFP-Cnn^WT overexpression (arrows in A and C), which had little bearing on embryo viability (see Table 1). In wild-type embryos where GFP-Cnn^Δ1 was expressed, most centrosomes were in pairs at metaphase (arrowheads in B and D) due to centrosome separation failure. As was the case for expression in a cnn null mutant background, expression of GFP-Cnn^Δ1 in a wild-type background resulted in reduced recruitment of γ-Tub to centrosomes (compare γ-Tub channel in C to D).