Ana1 helps recruit Polo to centrioles to promote mitotic PCM assembly and centriole elongation

Abstract

Polo kinase (PLK1 in mammals) is a master cell cycle regulator that is recruited to various subcellular structures, often by its polo-box domain (PBD), which binds to phosphorylated S-pS/pT motifs. Polo/PLK1 kinases have multiple functions at centrioles and centrosomes, and we have previously shown that in Drosophila phosphorylated Sas-4 initiates Polo recruitment to newly formed centrioles, while phosphorylated Spd-2 recruits Polo to the pericentriolar material (PCM) that assembles around mother centrioles in mitosis. Here, we show that Ana1 (Cep295 in humans) also helps to recruit Polo to mother centrioles in Drosophila. If Ana1-dependent Polo recruitment is impaired, mother centrioles can still duplicate, disengage from their daughters and form functional cilia, but they can no longer efficiently assemble mitotic PCM or elongate during G2. We conclude that Ana1 helps recruit Polo to mother centrioles to specifically promote mitotic centrosome assembly and centriole elongation in G2, but not centriole duplication, centriole disengagement or cilia assembly. This article has an associated First Person interview with the first author of the paper.

Keywords: Ana1; Centriole; Centrosome; Cep295; PCM; PLK1; Pericentriolar material; Polo.

Fig. 1.

An mRNA injection-based screen for proteins that help recruit Polo to centrosomes. (A) Table shows the number of potential PBD-binding sites (S-S/T motifs) in several centriole proteins (aa, amino acids). (B) Schematic illustrates the mRNA injection assay used to test the effect on Polo recruitment of mutating all the potential PBD-binding sites in a candidate protein. Green circles represent centrosomes recruiting Polo–GFP. (C) Micrographs of embryos expressing Polo–GFP (green in merged channel, top) injected with mRNA encoding WT (left panels) or SnT mutant (right panels) forms of each of the candidate proteins (Sas-4, Asl, Cep135 or Ana1) tagged with mKate2 (mKate2–FP; magenta in merged channel, top). (D) Magnified view highlighting a pair of centrosomes for each condition, as described in C. Arrowheads indicate centrosomes that contain Ana1-S34T–mKate2 and that recruit Polo–GFP; arrows indicate centrosomes that contain Ana1-S34T–mKate2 but do not detectably recruit Polo–GFP. A total of 5–14 embryos were injected and analysed for each mRNA. Note that Ana1 is normally significantly brighter at OM centrioles than at NM centrioles (Saurya et al., 2016) (Fig. S1B,C), making it hard to visually infer the relative amount of fluorescent fusion protein at OM and NM centrioles in these mRNA injection experiments. (E,F) Graphs quantify (E) the centrosomal levels of either the WT or mutant candidate–mKate2 fusions and (F) the corresponding centrosomal levels of Polo–GFP in each condition in S phase (a.u. arbitrary units). A total of 8–10 pairs of centrosomes were analysed per embryo (n=276, 280, 100, 100, 140, 198, 180 and 140 for WT and SnT Sas-4, Asl, Cep135 and Ana1, respectively). Note that the distribution of WT Asl is bimodal as Asl is usually much brighter at OM centrioles than NM centrioles (Novak et al., 2014); this effect is less pronounced in the Asl-S6T mutant for unknown reasons. Error bars represent s.d. P-values were calculated using an unpaired two-tailed t-test with Welch's correction.

Fig. 2.

The Ana1-S34T protein appears to be largely functional. (A) Graphs show the quantification of negative gravitaxis climbing assays. Each small shape shows the distance climbed by one of 15 individual ana1^−/− flies expressing either WT or S34T-mutant versions of Ana1 tagged with either GFP or mCherry (mCh) after being tapped to the bottom of a cylinder. The larger shapes show the average distance climbed by all flies in four technical repeats. Note that ana1^−/− flies without any transgene were not scored in this assay, as all of the mutant flies were severely uncoordinated due to the lack of cilia and so did not climb at all. Nevertheless, we show this bar as zero – marked with not applicable (N/A) – to better illustrate the level of rescue for each transgene. (B) Micrographs show the cilia membrane (green arrow) in a sensory neuron marked with mCD8–GFP (green) extending into a sensory bristle in an antenna from an adult ana1^−/− mutant fly expressing either WT-Ana1–mCherry or Ana1-S34T–mCherry (magenta), which both localise to the cilium basal body (magenta arrow). We examined >20 bristles in seven antennae from four different females (>100 in total) and detected no obvious morphological differences between the WT and mutant conditions. (C) Quantification of the percentage of mitotic neuroblasts with one or two centrosomes in ana1^−/− larval brains co-expressing Spd-2–GFP and either WT-Ana1–mCherry or Ana1-S34T–mCherry. Live neuroblasts were analysed blind, with centrosomes being identified by the colocalisation of both markers. (D) Electron micrographs show longitudinal (top) and cross-section (bottom) views of typical centrioles in either WT or ana1^−/− mutant third-instar larval wing-disc cells expressing either WT-Ana1–GFP or Ana1-S34T–GFP. (E) Graph quantifies the average longitudinal length of the centrioles in each condition (scored blind). As shown previously (Saurya et al., 2016), centrioles are slightly elongated when WT-Ana1–GFP is overexpressed, but this was not the case when Ana1-S34T–GFP was overexpressed. Small shapes indicate individual centriole lengths (126, 142 and 172, respectively), large shapes indicate the average centriole length in a whole wing disc (n=5, 7 and 8, respectively). Error bars represent s.d. P-values in A and E were calculated using an unpaired two-tailed t-test with Welch's correction and the ordinary one-way ANOVA with Tukey's multiple comparison test, respectively.

Fig. 3.

Centrosomal Polo recruitment is severely perturbed in Ana1-S34T embryos. (A,B) Examples of conventional spinning disk confocal images from living (A) WT-Ana1–mCherry and (B) Ana1-S34T–mCherry (magenta) embryos co-expressing Polo–GFP (green). Polo–GFP localised strongly to centrosomes at all stages of development and of the nuclear division cycle in WT-Ana1–mCherry embryos (A), also strongly staining the kinetochores during mitosis (arrows). Although embryos expressing Polo–GFP and Ana1-S34T–mCherry were very sick (and so difficult to accurately stage), Polo–GFP was usually barely detectable at the centrosomes (B, arrowheads), but was still strongly recruited to kinetochores (B, arrows) in the embryos that were in mitosis. (C,D) Graphs show the mean centrosomal (C) Ana1–mCherry and (D) Polo–GFP intensities in WT-Ana1–mCherry (black) or Ana1-S34T–mCherry embryos (red and green, respectively; a.u., arbitrary units). In total, n=64 centrosomes from ten different embryos co-expressing Polo–GFP and WT-Ana1–mCherry and n=23 centrosomes from six different embryos co-expressing Polo–GFP and Ana1-S34T–mCherry. Error bars indicate s.d. P-values were calculated using an unpaired two-tailed t-test with Welch's correction.

Fig. 4.

The centriole recruitment of Cep135, Asl and Sas-4 is largely unperturbed in Ana1-S34T embryos. (A) Graph shows the mean centrosomal GFP–Cep135 intensity in WT-Ana1–mCherry (+WT; black and grey triangles) or Ana1-S34T–mCherry (+S34T; green and light green circles) embryos (11 and seven embryos, respectively) in S phase. Multiple centrosome pairs were analysed per embryo (n=82 and 37 pairs in total for WT-Ana1 and Ana1-S34T embryos, respectively), for each pair the centrosomes were classified as OM or NM based on their Ana1–mCherry levels (data not shown). (B) Graph shows mean Asl–mCherry intensity at OM and NM centrosomes in WT-Ana1–GFP embryos (+WT; black and grey triangles) or Ana1-S34T–GFP embryos (+S34T; red and pink circles) in S phase. Five embryos were analysed per genotype, and ten pairs of centrosomes were analysed per embryo (n=50 centrosome pairs each). For each pair, the centrosome with the highest mean Asl–mCherry intensity was classified as the OM (Novak et al., 2014). (C) Graph shows mean centrosomal Sas-4–GFP intensity in WT-Ana1–mCherry embryos (+WT; black and grey triangles) or Ana1-S34T–mCherry embryos (+S34T; green and light green circles) at the beginning of S-phase. Eight embryos were analysed per genotype, and multiple centrosome pairs were analysed per embryo (n=74 and 70 pairs in total). For each pair the centrosomes were classified as OM or NM based on their Ana1–mCherry levels (data not shown). Error bars represent s.d. P-values were calculated using the ordinary one-way ANOVA with Tukey's multiple comparison test. a.u., arbitrary units.

Fig. 5.

Mitotic PCM expansion is impaired in Ana1-S34T embryos. (A,B) 3D-SIM images from fixed (A) WT-Ana1–mCherry and (B) Ana1-S34T–mCherry embryos. The embryos were stained with a general Cnn antibody (blue) and an antibody that recognises a specific Polo-dependent phospho-epitope in Cnn (Cnn-pS567, green; Feng et al., 2017). Cnn phosphorylated at S567 was detected at high levels within the PCM in WT-Ana1–mCherry embryos (A), indicating that Polo is present within the PCM. In Ana1-S34T–mCherry embryos (B), Ser567 phosphorylated Cnn was present in some centrosomes (arrowheads), but not in others (arrows), even though unphosphorylated Cnn was present at the centriole wall in all cases. The lack of Cnn-pS567 was correlated with a lack of mitotic PCM expansion, suggesting that these centrioles lacked sufficient Polo to phosphorylate Cnn and to drive PCM expansion. (C,D) Micrographs show 3D-SIM images of individual centrosomes in living (C) WT-Ana1–mCherry or (D) Ana1-S34T–mCherry (magenta in merged images) embryos expressing Spd-2–GFP (green in merged images). While Ana1–mCherry images are shown here for reference, reliably reconstructing the Ana1–mCherry signal was challenging, due to its low levels and the fast bleaching of the fluorophore. Thus, images were selected for analysis based only on whether the Spd-2–GFP reconstructed image was deemed of sufficient quality by SIMcheck (Ball et al., 2015). All centrosomes were imaged in approximately mid-S-phase when the centrosomal levels of Spd-2 are maximal. All the centrosomes in WT-Ana1–mCherry embryos organised Spd-2–GFP PCM scaffolds, but this was true only in a minority of Ana1-S34T–mCherry embryos (arrowheads), where many centrosomes recruited Spd-2–GFP only to the centriole wall (arrows). (E) Pie charts quantify the percentage of centrosomes that qualitatively showed a strong (dark green), weak (light green) or no (white) pericentriolar scaffold (n=21 reconstructed centrioles for each genotype, scored blind).

Fig. 6.

Some OM centrioles can partially bypass the requirement for Ana1 to help recruit Polo to centrioles, and so recruit some of the mitotic scaffold protein Cnn. (A,B) Schematic illustrates (A) how a sequential phosphorylation cascade comprising Sas-4, Ana1 and Spd-2 might drive increasing levels of Polo recruitment to the mother centriole and then to the expanding mitotic PCM in WT embryos, and (B) how this process might be perturbed in embryos expressing a form of Ana1 (Ana1-S34T) that cannot efficiently recruit Polo. Proteins recruiting Polo are indicated in shades of blue; proteins not recruiting Polo are indicated in black. Black arrows indicate how the phosphorylation of one protein can recruit Polo and so lead to the phosphorylation of the next protein in the putative cascade. See main text for details. (C) Schematic illustrates the genealogy of the centrosomes analysed for their ability to recruit GFP–Cnn from one cycle to the next. In the first division cycle, the centrosome with the OM is associated with a larger GFP–Cnn scaffold than the centrosome with the new mother centriole (NM1) (Conduit et al., 2010). When these centrosomes divide, the OM and NM1 centrosomes each generate a new centrosome containing a younger mother centriole (that is again smaller than the centrosome containing the original mother centriole) – NM2 and NM3, respectively. (D,E) Examples of OM1 and NM1 centrosomes generated at the start of the first cycle, and the NM2 and NM3 centrosomes they generated at the end of the second cycle in (D) WT-Ana1–mCherry or (E) Ana1-S34T–mCherry (magenta) embryos expressing GFP–Cnn (green). In Ana1-S34T–mCherry embryos the centrosome pairs (particularly NM1 and NM3) sometimes failed to separate properly. (F) Graph shows the mean GFP–Cnn intensity at each centrosome type in WT-Ana1–mCherry (black and grey triangles) and Ana1-S34T–mCherry (green circles) embryos. N=5 and 8 embryos analysed for WT-Ana1 and Ana1-S34T genotypes, respectively; three pairs of centrosomes in the first cycle were analysed per embryo, so a total of n=15 and 24 centrosome pairs for the WT-Ana1 and Ana1-S34T genotype, respectively (note that for the Ana1-S34T genotype, only 23 OM–NM2 pairs and 19 NM1–NM3 pairs could be analysed, due to the lack of centrosome separation at the beginning of the second cycle). To facilitate visualisation, only the P-values corresponding to the most informative statistical comparisons are shown, coloured by the type of centrosome being compared against others: WT NM1 in the second cycle (navy blue), S34T OM in the second cycle (magenta), and S34T NM1 in the second cycle (gold). (G) Graph shows the same data as in F, but expressed as the average sum of GFP–Cnn levels for OM and NM1 centrosomes in the first cycle (dark grey for WT-Ana1–mCherry embryos, dark green for Ana1-S34T–mCherry embryos), and the average sum of GFP–Cnn levels for OM, NM1, NM2 and NM3 centrosomes in the second cycle (light grey for WT-Ana1–mCherry embryos, light green for Ana1-S34T–mCherry embryos). Error bars represent s.d. P-values were calculated using an ordinary one-way ANOVA with Tukey's multiple comparison test. a.u., arbitrary units.

Fig. 7.

The C-terminal region of Ana1 helps recruit Polo to centrosomes in vivo and can interact directly with the Polo PBD in vitro. (A) Schematic representation of the protein sequence of Drosophila melanogaster Ana1 indicating S-S/T motifs that are either conserved (present in at least 11/12 Drosophila species analysed, red lines) or not conserved (blue lines). The boundaries of the Ana1 fragments we analysed are indicated, with numbers indicating amino acid positions. (B) Micrographs of embryos expressing Polo–GFP (green in the merged images) injected with constructs encoding full-length Ana1–mKate2 (WT), or Ana1–mKate2 constructs in which either the whole protein (S34T), or only the various Ana1 sub-regions – as indicated in A – have had their S-S/T motifs replaced with T-S/T (magenta in the merged images). Arrows highlight some of the centrosomes that do not recruit Polo–GFP. Embryos were scored positive if they showed loss of Polo–GFP (scored blind) from at least one centrosome; the results for each injected construct are indicated numerically as affected embryos/total injected embryos analysed. Note that the CTa, CTb, CTb1 and CTb2 injections were performed at a later date on a different microscope system, so the images look different to the others presented in this paper. (C) Western blots of an in vitro assay in which purified recombinant MBP fusion proteins to either WT or mutant (S10T) CTa or CTb were incubated with or without PLK1 and then assessed for their ability to bind to recombinant human GST–PBD. Only CTb exhibited specific binding (i.e. binding was enhanced by PLK1 phosphorylation, and this depended on the S-S/T motifs). (D) Same as C, except in this experiment we tested the ability of WT or mutant MBP–CTb to bind to GST–PBD after phosphorylation by either Aurora A or PLK1. Only WT CTb phosphorylated by PLK1 exhibited specific binding. Asterisk indicates a smaller band, presumably a partial degradation product of the fusion protein. (E) Graphs quantify the level of GST–PBD binding to the different MBP–CTb fusion proteins in two (Aurora A) or four (PLK1) technical repeats of these in vitro binding assays, and one technical repeat using CTa and PLK1 (a.u., arbitrary units). Although these assays are somewhat variable, the WT CTb fragment consistently exhibits elevated levels of binding to GST–PBD when it is phosphorylated by PLK1.

Fig. 8.

Ana1 helps to recruit Polo to the centriole distal end to promote centriole elongation. (A) Graph quantifies centriole length in ana1^−/− mutant testes expressing either WT-Ana1–GFP or Ana1-S34T–GFP. Each data point represents an individual testis and shows the average centriole length calculated from >10 centrioles. Error bars indicate the s.d. (B) Micrographs show typical centriole pairs in fixed ana1^−/− mutant spermatocytes expressing Polo–GFP (green) and either WT-Ana1–mCherry (top panels) or Ana1-S34T–mCherry (bottom panels) (Ana1–mCh; red) stained to reveal the distribution of the centriole protein Asl (white). The centriole pairs in spermatocytes are engaged and arranged in a characteristic V-shape, and they grow to a much longer length than the centrioles in most other Drosophila cell types during an extended G2 period. In the Ana1-S34T–mCherry spermatocytes, the centriole pairs are duplicated and arranged in the typical V-shape, but they are much shorter than normal. In the WT centrioles the Asl, Ana1 and Polo extend along the entire length of the extended centrioles, whereas in the S34T centrioles Polo does not extend outwards as far as the centriole distal end, suggesting that Ana1 is normally required to recruit Polo to the centriole distal end. (C) Graph quantifies the length that the Asl, Ana1 and Polo signals spread outwards along the centrioles in the experiment described in B. Each data point represents an individual testis and shows the average spread of each protein calculated from >10 centrioles. Error bars indicate the s.d. Note how in WT centrioles Asl, Ana1 and Polo spread along the entire length of the centriole to the same extent, while in the S34T centrioles Polo specifically does not extend to the distal end. P-values in A and C were calculated using an unpaired two-tailed t-test with Welch correction and the ordinary one-way ANOVA with Tukey's multiple comparison test, respectively.