1. Cell Biology
Download icon

Evidence that a positive feedback loop drives centrosome maturation in fly embryos

Research Article
  • Cited 0
  • Views 481
  • Annotations
Cite this article as: eLife 2019;8:e50130 doi: 10.7554/eLife.50130

Abstract

Centrosomes are formed when mother centrioles recruit pericentriolar material (PCM) around themselves. The PCM expands dramatically as cells prepare to enter mitosis (a process termed centrosome maturation), but it is unclear how this expansion is achieved. In flies, Spd-2 and Cnn are thought to form a scaffold around the mother centriole that recruits other components of the mitotic PCM, and the Polo-dependent phosphorylation of Cnn at the centrosome is crucial for scaffold assembly. Here, we show that, like Cnn, Spd-2 is specifically phosphorylated at centrosomes. This phosphorylation appears to create multiple phosphorylated S-S/T(p) motifs that allow Spd-2 to recruit Polo to the expanding scaffold. If the ability of Spd-2 to recruit Polo is impaired, the scaffold is initially assembled around the mother centriole, but it cannot expand outwards, and centrosome maturation fails. Our findings suggest that interactions between Spd-2, Polo and Cnn form a positive feedback loop that drives the dramatic expansion of the mitotic PCM in fly embryos.

https://doi.org/10.7554/eLife.50130.001

Introduction

Centrosomes play an important part in many aspects of cell organisation, and they form when a mother centriole recruits pericentriolar material (PCM) around itself (Conduit et al., 2015). The PCM contains several hundred proteins (Alves-Cruzeiro et al., 2014), allowing the centrosome to function as a major microtubule (MT) organising centre, and also as an important coordination centre and signalling hub (Chavali et al., 2014; Vertii et al., 2016). Centrosome dysfunction has been linked to several human diseases and developmental disorders, including cancer, microcephaly and dwarfism (Bettencourt-Dias et al., 2011; Nigg and Raff, 2009).

During interphase, the mother centriole recruits a small amount of PCM that is highly organised (Lawo et al., 2012; Mennella et al., 2012; Sonnen et al., 2012; Fu and Glover, 2012). As cells prepare to enter mitosis, however, the PCM expands dramatically around the mother centriole in a process termed centrosome maturation (Palazzo et al., 2000). Electron microscopy (EM) studies suggest that centrioles organise an extensive ‘scaffold’ structure during mitosis that surrounds the mother centriole and recruits other PCM components such as the γ-tubulin ring complex (γ-TuRC) (Moritz et al., 1995; Moritz et al., 1998; Schnackenberg et al., 1998).

In the fruit fly Drosophila melanogaster and the nematode Caenorhabditis elegans, a relatively simple pathway seems to govern the assembly of this mitotic PCM scaffold. The conserved centriole/centrosome protein Spd-2/SPD-2 (fly/worm nomenclature) cooperates with a large, predominantly predicted-coiled-coil, protein (Cnn in flies, SPD-5 in worms) to form a scaffold whose assembly is stimulated by the phosphorylation of Cnn/SPD-5 by the mitotic protein kinase Polo/PLK-1 (Conduit et al., 2014a; Conduit et al., 2014b; Feng et al., 2017; Woodruff et al., 2017; Woodruff et al., 2015; Wueseke et al., 2016). Mitotic centrosome maturation is abolished in the absence of this pathway, and some aspects of Cnn and SPD-5 scaffold assembly have recently been reconstituted in vitro (Feng et al., 2017; Woodruff et al., 2015; Woodruff et al., 2017). Vertebrate homologues of Spd-2 (Cep192) (Gomez-Ferreria et al., 2007; Zhu et al., 2008), Cnn (Cdk5Rap2/Cep215) (Barr et al., 2010; Choi et al., 2010; Fong et al., 2008; Kim and Rhee, 2014; Lizarraga et al., 2010) and Polo (Plk1) (Haren et al., 2009; Lane and Nigg, 1996; Lee and Rhee, 2011) also have important roles in mitotic centrosome assembly, indicating that elements of this pathway are likely to be conserved in higher metazoans. In vertebrate cells another centriole and PCM protein, Pericentrin, also has an important role in mitotic centrosome assembly that is dependent upon its phosphorylation by Plk1 (Haren et al., 2009; Lee and Rhee, 2011). Pericentrin can interact with Cep215/Cnn (Buchman et al., 2010; Kim and Rhee, 2014; Lerit et al., 2015), but in flies the Pericentrin-like-protein (Plp) has a clear, but relatively minor, role in mitotic PCM assembly when compared to Spd-2 and Cnn (Lerit et al., 2015; Martinez-Campos et al., 2004; Richens et al., 2015).

Although most of the main players in mitotic centrosome-scaffold assembly appear to have been identified, several fundamental aspects of the assembly process remain mysterious. Cells entering mitosis, for example, contain two mother centrioles that assemble two mitotic centrosomes of equal size. It is unclear how this is achieved, as even a slight difference in the initial size of the two growing centrosomes would be expected to lead to asymmetric centrosome growth—as the larger centrosome would more efficiently compete for scaffolding subunits (Conduit et al., 2015; Zwicker et al., 2014). The centrioles in fly embryos appear to overcome this problem by constructing the PCM scaffold from the ‘inside-out’: Spd-2 and Cnn are only incorporated into the scaffold close to the mother centriole, and they then flux outwards to form an expanded scaffold around the mother centriole (Conduit et al., 2014b; Conduit et al., 2010). In this way, the growing PCM scaffold could ultimately attain a consistent steady-state size—where incorporation around the mother centriole is balanced by loss of the scaffold at the centrosome periphery—irrespective of any initial size difference in the PCM prior to mitosis (Conduit et al., 2015; Raff, 2019).

A potential problem with this ‘inside-out’ mode of assembly is that the rate of centrosome growth is limited by the very small size of the centriole. Mathematical modelling indicates that the incorporation of a crucial PCM scaffolding component only around the mother centriole cannot easily account for the high rates of mitotic centrosome growth observed experimentally (Zwicker et al., 2014). To overcome this problem, it has been proposed that centrosome growth is ‘autocatalytic’, with the centriole initially recruiting a key scaffolding component that can subsequently promote its own recruitment (Woodruff et al., 2014; Zwicker et al., 2014). It has been proposed that Spd-2 and Cnn could form a positive feedback loop that might serve such an autocatalytic function: Spd-2 helps recruit Cnn into the scaffold, and Cnn then helps to maintain Spd-2 within the scaffold, thus allowing higher levels of Spd-2 to accumulate around the mother centriole, which in turn drives higher rates of Cnn incorporation (Conduit et al., 2014b; Conduit et al., 2015; Raff, 2019).

In worms (Decker et al., 2011) and vertebrates (Joukov et al., 2010; Joukov et al., 2014; Meng et al., 2015), SPD-2/Cep192 can help recruit PLK1/Plk1 to centrosomes and Cep192 also activates Plk1 in vertebrates, in part through recruiting and activating Aurora A, another mitotic protein kinase implicated in centrosome maturation. We suspected, therefore, that in flies Spd-2 might recruit Polo into the centrosome-scaffold to phosphorylate Cnn and so help to generate a positive feedback loop that drives the expansion of the mitotic PCM. In flies, however, no interaction between Polo and Spd-2 has been reported. Indeed, an extensive Y2H screen for interactions between key centriole and centrosome proteins identified interactions between Spd-2 and the mitotic kinases Aurora A and Nek2, and between Polo and the centriole proteins Sas-4, Ana1 and Ana2, but not between Polo and Spd-2 (Galletta et al., 2016). A possible explanation for this result is that Polo/Plk1 is believed to be largely recruited to its many different locations in the cell, including centrosomes, through its Polo-Box-Domain (PBD), which binds to phosphorylated S-S/T(p) motifs (Elia et al., 2003a; Elia et al., 2003b; Lee et al., 1998; Liu et al., 2004; Reynolds and Ohkura, 2003; Seong et al., 2002; Song et al., 2000). Perhaps any such Polo binding sites in fly Spd-2 were simply not phosphorylated in the Y2H experiments. In support of this possibility, phosphorylated S-S/T(p) motifs in SPD-2/Cep192 have previously been shown to help recruit PLK1/Plk1 to centrosomes in worms (Decker et al., 2011), frogs (Joukov et al., 2010) and humans (Meng et al., 2015).

Here, we examine the potential role of Spd-2 in recruiting Polo to centrosomes in Drosophila embryos. We find that, like Cnn, Spd-2 is largely unphosphorylated in the cytosol, but is highly phosphorylated at centrosomes, where Spd-2 and Polo extensively co-localise within the pericentriolar scaffold. We show that a Spd-2 fragment containing 19 S-S/T motifs exhibits enhanced binding to the PBD in vitro when it has been phosphorylated by Plk1, but no enhancement is seen if these S-S/T motifs are mutated to T-S/T—a mutation that strongly perturbs PBD binding (Elia et al., 2003b). We express forms of Spd-2 in vivo in which either all 34 S-S/T motifs, or the 16 most conserved S-S/T motifs, have been mutated to T-S/T to perturb PBD-binding. These mutant Spd-2 proteins are still recruited to mother centrioles, as are Polo and Cnn, and these proteins assemble a PCM scaffold around the mother centriole. Strikingly, however, this PCM scaffold can no longer expand outwards, and centrosome maturation fails. These observations provide strong support for the hypothesis that Spd-2, Polo and Cnn cooperate to form a positive feedback loop that is required to drive the rapid expansion of the mitotic PCM in fly embryos.

Results

Spd-2 is phosphorylated specifically at centrosomes

We showed previously that Cnn is specifically phosphorylated at centrosomes (Conduit et al., 2014a), so we wondered if this was also the case for Spd-2. We partially purified centrosomes from embryo extracts by sucrose step-gradient centrifugation and compared the electrophoretic mobility of Spd-2 on western blots of the gradient fractions (Figure 1A). As was the case for Cnn, we observed a prominent slower migrating form of Spd-2 in the heavier centrosomal fractions that was largely absent in the lighter cytosolic fractions. However, unlike Cnn, a faster migrating form of Spd-2 was also present in the centrosomal fractions. Treatment of the centrosomal fractions with phosphatase revealed that the reduced mobility of Spd-2 in the centrosomal fractions could be attributed to phosphorylation (Figure 1B). Thus, like Cnn, Spd-2 is specifically phosphorylated at centrosomes, although not all the Spd-2 at the centrosome appears to be phosphorylated.

Figure 1 with 1 supplement see all
Spd-2 is phosphorylated at centrosomes and can bind the PBD in vitro in a manner that is enhanced by phoshorylation.

(A) Western blot of a sucrose step-gradient purification of centrosomes from embryo extracts probed with anti-Cnn, Spd-2, γ-tubulin and Actin antibodies, as indicated. The gradient fractions are labelled 1 (heaviest) to 12 (lightest), and the cytosolic and centrosomal-peak fractions are indicated. (B) A western blot of centrosomal fractions from the step gradient treated with phosphatase (with or without phosphatase inhibitor), and probed with the indicated antibodies; the cytosolic fraction is also shown. The Cnn and Actin blots were presented previously (Conduit et al., 2014a) (reproduced here under a CC-BY 3.0 licence https://creativecommons.org/licences/by/3.0/), and were performed contemporaneously with the Spd-2 blot shown here. The blots shown are representative of two technical replicates for each of two biological repeats. (C) Western blot of an experiment in which recombinant WT MBP-Spd-2(aa352-758) or mutant MBP-Spd-2(aa352-758)−19T were bound to MBP-Antibody-beads then either phosphorylated or not phosphorylated by human recombinant Plk1 (+/-) before mixing with human recombinant GST-PBD (Input). The beads were washed and any proteins still bound to the beads were eluted (IP). The Input and IP fractions were probed with either anti-Spd-2 antibodies (top panels) or anti-GST antibodies (bottom panels). (D) Graph shows the quantification of the amount of GST-PBD bound to the indicated beads (normalised to the amount of Spd-2 in each fraction) in three independent repeats; + /- indicates treatment with Plk1. To facilitate comparisons between repeats, the data is shown normalised to the signal from each 19T(-) sample. Enhanced binding of GST-PBD to the phosphorylated WT protein was observed in all three experiments, but was somewhat variable. Due to the variation and the small sample size (n = 3), we were unable to determine whether the data was normally distributed, so we could not apply parametric tests for statistical significance. Non-parametric tests did not indicate a significant increase in binding to the phosphorylated protein (p<0.05), but these tests do not work well with such a low number of repeats.

https://doi.org/10.7554/eLife.50130.002

Mutating multiple centrosomal phosphorylation sites in Spd-2 only mildly perturbs Spd-2 function in vivo

Using Mass Spectroscopy, we identified seven Spd-2 peptides that were phosphorylated consistently and with high confidence in the centrosomal fractions, but not in the cytosolic fractions (Table 1). One of these peptides contained a phosphorylated S-S(p) motif which could potentially help recruit Polo to centrosomes via its PBD. We generated transgenic lines expressing WT Spd-2-GFP and a mutant form of Spd-2-GFP in which all seven of the centrosomally phosphorylated residues—together with an additional 4 Ser/Thr resides that could potentially be phosphorylated in these peptides (Table 1)—were mutated to Ala (Spd-2-11A-GFP). Interestingly, although Spd-2-11A-GFP was expressed at substantially lower levels than WT Spd-2-GFP (Figure 1—figure supplement 1A), both Spd-2-GFP and Spd-2-11A-GFP rescued the female sterility phenotype of Spd-2 mutant flies, and in Spd-2 mutant embryos the mutant protein localised to centrosomes nearly as well as the WT protein (Figure 1—figure supplement 1B–D). Thus, although present at lower levels, Spd-2-11A-GFP is recruited to centrosomes nearly as well as the WT protein and preventing the phosphorylation of at least seven sites in Spd-2 that are specifically phosphorylated at centrosomes appears to only mildly perturb Spd-2 function in vivo.

Table 1
Identification of Spd-2 sites phosphorylated at the centrosome.

The Table lists amino acids in Spd-2 that were identified as being phosphorylated in the centrosomal fractions of embryo extracts, but not the cytosolic fractions, by Mass Spectroscopy. The Peptide score is the Mascot Ion Score (Koenig et al., 2008)—scores > 29 indicate identity or extensive homology (p<0.05). Phosphorylated amino acids are marked in red. These sites were mutated together with an additional 4 Ser/Thr resides that could potentially have been phosphorylated in these peptides (blue) to generate Spd-2-11A-GFP.

https://doi.org/10.7554/eLife.50130.006
Peptide sequencePeptide scoreHigh-scoring phosphorylated siteAdditional sites mutated
GTNISFEPAEITGR53.03S121-
TNQPLLEPESNVTLDSVGEK65.26T329-
RPPSSSEILSLSAIDK38.85S397-
KPLSPLADHPQITISR34.55S484-
RVSIATMGLIPR29.93S569-
NLSPLSSPR42.33S614S617, S618
GLGTSSVAVPR64.8S673T671, S672

Generating a mutant form of Spd-2 with reduced binding to the Polo PBD

The Spd-2 phosphorylation sites we identified here were also identified previously in at least one of several phospho-proteomic screens in D. melanogaster (Bodenmiller et al., 2007; Habermann et al., 2012; Hu et al., 2019; Zhai et al., 2008). These screens, however, also identified many additional phosphorylated peptides in Spd-2, including 16 peptides that contained at least one phosphorylated S-S/T(p) motif that could potentially bind the PBD (Table 2). We speculated, therefore, that Drosophila Spd-2 might utilise multiple phosphorylated S-S/T(p) motifs to help recruit Polo to centrosomes.

Table 2
The Table lists several previously identified Spd-2 peptides that include the potential PBD binding motif S-S/T(p).

Definitively identified phosphorylated sites within PBD binding motifs are shown in red. Sites listed in brackets have been identified as phosphorylated but the scores were low (shown in blue); or have not been identified in Drosophila melanogaster, but have been definitively identified in other closely related species (shown in purple). Other phosphorylated sites which are on the same reported peptide, but are not part of a PBD binding motif, are shown in bold. The phospho-proteomic screens in which these peptides were identified are listed.

https://doi.org/10.7554/eLife.50130.007
Peptide sequencePhospho sitesOther phospho sitesRef.Also
VFGDLSSFSKGRRS34
S35
(S37)Zhai et al., 2008-
ALETLEKPRPSRSSQAKS76S73Bodenmiller et al., 2007-
EKPSLSVAEILKSSFVEK(S156)S146
S148
Bodenmiller et al., 2007Zhai et al., 2008
SSSS(S185)Hu et al., 2019-
SENIWNIVSNSSPNRSRS310
S311
(S308)
S315
Bodenmiller et al., 2007Zhai et al., 2008;
Hu et al., 2019
RPPSSSEILSLSAIDK(S389)
(S390)
S391
(S395)
S397
Bodenmiller et al., 2007Zhai et al., 2008
DIDLNSDTSTVEVVNHLWEHGR(S413)
(T414)
S410
(T412)
Zhai et al., 2008-
ADTDPVETEAEADIDEWPSTPVKEPSRR(S515)
(T516)
T499
T504
S522
AASPSSSDGVRPLTCTEDENDEEDEDKTPVNKKS538
S539
(S540)
S536
(T547)
(T549)
T561
KASSLSSTRLDGCDVAVASSTERS581
S582
(S584)
NLSPLSSPRS617
S618
S614Bodenmiller et al., 2007Zhai et al., 2008
Hu et al., 2019
SCLSSPLLDSTTSSDRRS624
S625
(S630)
(T631)
Zhai et al., 2008Hu et al., 2019
SCLSSPLLDSTTSSDRRS634Habermann et al., 2012-
ANSSPAGSEASSTSGFTASGRS650S654Bodenmiller et al., 2007-
KANSSPAGSEASSTSGFTASGR(S658)S654Zhai et al., 2008-
RGLGTSSVAVPRS673
S674
(T672)Zhai et al., 2008This study;
Habermann et al., 2012

As a first test of this hypothesis, we examined whether phosphorylated S-S/T motifs in Spd-2 could bind the PBD in vitro. Full-length Spd-2 fusion proteins were unstable when expressed in E. coli, so we purified an MBP-fusion containing approximately the middle 1/3 of Spd-2 (aa352-758, MBP-Spd-2-WT), a fusion that we had previously purified and raised antibodies against (Dix and Raff, 2007). This fusion protein contained 19 S-S/T motifs and, when phosphorylated by recombinant human Plk1 in vitro, it exhibited an enhanced ability to co-immunoprecipitate (IP) with recombinant GST-PBD (Figure 1C). Importantly, this phosphorylation-enhanced binding to GST-PBD was abolished when these 19 S-S/T motifs were mutated to T-S/T (MBP-Spd-2–19T)—a conservative substitution that has nevertheless been shown to perturb PBD-binding to these motifs (Elia et al., 2003b). Thus, a fragment of Spd-2 can bind directly to the PBD in a manner that is enhanced when the fragment is phosphorylated by Plk1, and this enhanced binding is prevented when the S-S/T motifs are mutated to T-S/T. Interestingly, this suggests that, in vitro at least, Plk1 can phosphorylate Spd-2 to ‘prime’ its own binding to Spd-2.

To test the potential role of Spd-2 in recruiting Polo to the mitotic PCM in vivo, we generated transgenic lines expressing a Spd-2-GFP fusion in which all 34 S-S/T motifs in D. melanogaster Spd-2 were mutated to T-S/T (Spd-2-ALL-GFP) (blue and red lines, Figure 2A). We reasoned that the conservative substitution of Thr for Ser might not disturb the overall folding of the protein, but that mutating all these sites should prevent them from binding the PBD efficiently (Elia et al., 2003b). In addition, 16 of the 34 S-S/T motifs were highly conserved in Drosophila species (Figure 2A, red lines), so we generated transgenic lines expressing a form of Spd-2-GFP in which only these 16 conserved motifs were mutated (Spd-2-CONS-GFP).

Figure 2 with 1 supplement see all
Generating forms of Spd-2 that should be unable to bind the PBD.

(A) A schematic representation of Drosophila melanogaster Spd-2, indicating the position of S-S/T motifs that are either highly conserved (present in at least 11/12 Drosophila species analysed—red lines), or not highly conserved (blue lines). (B) Western blot of WT embryos, or Spd-2 mutant embryos expressing either WT Spd-2-GFP, Spd-2-CONS-GFP or Spd-2-ALL-GFP, as indicated, probed with either anti-Spd-2 antibodies or anti-GAGA transcription factor antibodies (Raff et al., 1994) (as a loading control). The blot shown is representative of three technical replicates. (C) Bar charts quantify the percentage of Spd-2 mutant embryos that had initiated development after expression of either WT Spd-2-GFP, Spd-2-CONS-GFP or Spd-2-ALL-GFP, as indicated. The chart shows the data from two independent biological repeats in which 3 lots of >50 embryos were collected and scored independently; error bars represent the standard deviation (SD).

https://doi.org/10.7554/eLife.50130.008

Western blotting revealed that the mutant Spd-2-GFP-fusions were expressed at slightly lower levels than WT Spd-2-GFP in embryos (Figure 2B). Importantly, however, both mutant fusion proteins rescued the defect in pronuclear fusion in Spd-2 mutant embryos (although to a slightly lesser extent than the WT fusion protein), allowing these embryos to start to develop (Figure 2C). This demonstrates that the mutant proteins are not simply misfolded (see also below). Unlike Spd-2-11A-GFP, however, Spd-2 mutant embryos expressing either Spd-2-CONS-GFP (hereafter Spd-2-CONS-GFP embryos) or Spd-2-ALL-GFP (hereafter Spd-2-ALL-GFP embryos) died early in embryonic development and almost never hatched as larvae (<1/500 embryos hatching; n > 2000 embryos scored).

Spd-2-CONS-GFP and Spd-2-ALL-GFP embryos die early in development

To investigate why Spd-2-CONS-GFP and Spd-2-ALL-GFP embryos almost never hatched as larvae, we expressed a Jupiter-mCherry transgene in these embryos to follow the behaviour of centrosomes and MTs. Syncytial Drosophila embryos rapidly cycle between S- and M-phases without any Gap phases (Foe and Alberts, 1983). When the embryos exit mitosis they immediately enter S-phase and the centrosomes start to mature in preparation for the next round of mitosis; thus, the centrosomes in these syncytial embryos organise a relatively robust, mitotic-like, PCM at all stages. WT Spd-2-GFP localised strongly to centrosomes, and the centrosomes organised robust astral and spindle MT arrays (Figure 3A, left panels). The mutant proteins were less abundant at centrosomes (Figure 3A,B) and the astral MT arrays organised by the centrosomes were less robust (Figure 3A,C) (see ‘Analysis of centrosome and MT fluorescence intensities’ section in the Materials and methods for a full explanation of quantification methods); as a result, centrosomes were often detached from the spindle poles (arrowheads, Figure 3A). Note that the reduced centrosomal levels of the Spd-2 mutant proteins are unlikely to be due to their lower expression levels (Figure 2B), as the Spd-2-GFP-11A mutant protein was also present at reduced levels, but localised to centrosomes nearly normally (Figure 1—figure supplement 1). Embryos expressing the mutant proteins exhibited progressively more severe mitotic defects as they developed, and these defects were qualitatively, but reproducibly, more pronounced in Spd-2-ALL-GFP embryos. We conclude that the mutant Spd-2-GFP fusions allow the assembly of a centrosome that can support pronuclear fusion, but these centrosomes cannot properly support the rapid syncytial divisions—and so the embryos accumulate mitotic defects and die during early development.

Centrosomes in Spd-2-CONS and Spd-2-ALL embryos recruit less Spd-2 and organise fewer MTs.

(A) Micrographs show stills of living Spd-2 mutant embryos expressing the MT-marker Jupiter-mCherry (red) and either WT Spd-2-GFP, Spd-2-CONS-GFP or Spd-2-ALL-GFP (green, as indicated). Time (in seconds) as the embryos progress from early S-phase (t = 0 s) to Anaphase (t = 720 s) is indicated. Embryos expressing the mutant proteins exhibited a range of mitotic defects, such as detached spindle poles (white arrowheads), that were more severe in Spd-2-ALL-GFP embryos. The Spd-2-CONS-GFP embryo shown here did not yet have any defects making it easier to compare to the WT-Spd-2-GFP embryo. (B,C) Graphs quantify the centrosomal intensity of the various Spd-2-GFP fusion proteins (B) or the centrosomal MT intensity (C) during late S-phase (t = 270 s). Each dot represents the average intensity of the five brightest centrosomes in a single embryo (n = 10, 9 and 8 embryos for WT, CONS and ALL embryos, respectively); error bars indicate the mean ± SD of each population of embryos scored. The D’Agostino–Pearson omnibus normality test was used to test for the Gaussian distribution of data. One-Way ANOVA with Tukey's multiple comparisons test was used when data passed the normality test (Jupiter-mCherry datasets); Kruskal-Wallis test with Dunn's multiple comparisons test was used otherwise (**, p<0.01; ***, p<0.001).

https://doi.org/10.7554/eLife.50130.016

Spd-2-CONS-GFP and Spd-2-ALL-GFP are recruited to centrioles but cannot efficiently form a scaffold that spreads out from the centrioles

Spd-2 is localised to centrioles and to the mitotic PCM, so we wondered whether the reduction in the centrosomal levels of Spd-2-CONS-GFP and Spd-2-ALL-GFP was a result of a failure to properly localise these proteins to centrioles, to the PCM, or to both. Live-cell 3D-structured illumination super-resolution microscopy (3D-SIM) revealed that, as shown previously (Conduit et al., 2014b), WT Spd-2-GFP localised to the mother centriole, and also to a fibrous scaffold-like structure that extended outwards around the mother centriole (Figure 4, left panels). Strikingly, both mutant proteins still localised strongly to the mother centriole, but the extended scaffold-like structure appeared to be greatly reduced (Figure 4)—suggesting either that the mutant proteins were unable to efficiently incorporate into the scaffold, or that very little scaffold was assembled in these embryos.

Spd-2-CONS and Spd-2-ALL do not efficiently assemble a PCM scaffold.

Micrographs show 3D-SIM images of individual centrosomes from Spd-2 mutant embryos expressing WT Spd-2-GFP, Spd-2-CONS-GFP or Spd-2-ALL-GFP (as indicated). Pie charts quantify the percentage of centrosomes that were scored qualitatively as having a strong (dark green), weak (light green) or no (white) pericentriolar scaffold (n = 36, for each genotype, respectively). In this, and all other SIM experiments, images were only included in the analysis if the reconstruction was deemed of sufficient quality by SIM-Check (Ball et al., 2015) (see Materials and methods for a full explanation of image quality control). All centrosomes were imaged in mid-late S-phase when the centrosomal levels of Spd-2 are maximal (see Figure 8). All scorings were performed blind by researchers not involved in the data acquisition.

https://doi.org/10.7554/eLife.50130.018

To distinguish between these possibilities we tested whether the mutant Spd-2 proteins could co-assemble into a PCM scaffold formed by WT Spd-2. In Spd-2 mutant embryos, we expressed one copy of WT Spd-2-mCherry and one copy of either WT Spd-2-GFP, Spd-2-CONS-GFP or Spd-2-ALL-GFP. Both mutant GFP-fusions co-assembled into a scaffold with WT Spd-2-mCherry in a manner that was very similar to the WT Spd-2-GFP (Figure 5). Thus, the mutant proteins can assemble into a Spd-2 scaffold if it is present, suggesting that the failure to detect a scaffold in Spd-2-CONS-GFP and Spd-2-ALL-GFP embryos (Figure 4) is likely to be due to the absence of the scaffold. Importantly, these observations also indicate that the mutant Spd-2 proteins are not misfolded, as they can clearly interact with the proteins that normally recruit and maintain Spd-2 at centrioles and within the mitotic PCM. A potential caveat to this interpretation is that Spd-2 proteins could form homo-oligomers, potentially allowing a largely misfolded mutant protein to oligomerise with a WT partner that then localises the mutant protein correctly. However, it has recently been shown using Fluorescence Correlation Spectroscopy (FCS) that SPD-2 is monomeric in worm embryos (Wueseke et al., 2016), and we found that this was also the case for Spd-2 in fly embryos (Figure 5—figure supplement 1). Thus, the mutant Spd-2 proteins are likely to be recruited to centrioles and centrosomes as monomers.

Figure 5 with 1 supplement see all
Spd-2-CONS and Spd-2-ALL can efficiently assemble into a PCM scaffold formed by WT Spd-2.

(A-C) Micrographs show 3D-SIM images of individual centrosomes from Spd-2 mutant embryos expressing WT Spd-2-mCherry and one copy of either WT Spd-2-GFP (A), Spd-2-CONS-GFP (B) or Spd-2-ALL-GFP (C). Pie charts quantify the percentage of centrosomes that were scored qualitatively as having a strong (dark green), weak (light green) or no (white) pericentriolar scaffold (n = 16 individual centrosomes, two images (channels) per centrosome, for each genotype, respectively). All centrosomes were imaged in mid-late S-phase when the centrosomal levels of Spd-2 are maximal (see Figure 8). All scorings were performed blind by researchers not involved in the data acquisition.

https://doi.org/10.7554/eLife.50130.020

A mitotic PCM scaffold is assembled in Spd-2-CONS-GFP and Spd-2-ALL-GFP embryos but it cannot expand efficiently around the centriole

We have previously hypothesised that Spd-2 cooperates with Polo and Cnn to form the mitotic PCM scaffold in flies (Conduit et al., 2014b; Conduit et al., 2015). We therefore tested whether Polo or Cnn could form a scaffold even when the Spd-2-ALL and Spd-2-CONS proteins cannot. We first examined the distribution of Polo-GFP in living Spd-2 mutant embryos expressing either WT Spd-2-mCherry, Spd-2-CONS-mCherry or Spd-2-ALL-mCherry. The co-expression of Polo-GFP with the mutant Spd-2-mCherry proteins led to mitotic defects in embryos that were more severe than those observed when mutant GFP- or mCherry-fusions were expressed in the absence of Polo-GFP, suggesting that the GFP-tagged Polo sensitises the embryos to the expression of the mutant Spd-2-fusions (Figure 6—figure supplement 1). These defects were so severe in embryos expressing Spd-2-ALL-mCherry that we could not reliably stage them, so they were excluded from this analysis. WT Spd-2-mCherry and Polo-GFP extensively co-localised at the mother centriole and spread outwards together into the scaffold—supporting the idea that Spd-2 normally helps recruit Polo to the scaffold (Figure 6A). In contrast, although the Spd-2-CONS-mCherry and Polo-GFP proteins still co-localised around the mother centriole, neither protein formed a robust scaffold (Figure 6B). These observations suggest that phosphorylated S-S/T(p) motifs in Spd-2 are not required to recruit Polo to mother centrioles—and it is known that Polo can be recruited to centrioles by phosphorylated S-S/T motifs in at least one other centriole protein, Sas-4 (Novak et al., 2016)—but are required to recruit Polo to the PCM scaffold that expands around the mother centriole.

Figure 6 with 1 supplement see all
Polo is recruited to centrioles but cannot assemble into a PCM scaffold in Spd-2-CONS and Spd-2-ALL embryos.

(A,B) Micrographs show 3D-SIM images of individual centrosomes from Spd-2 mutant embryos expressing Polo-GFP (green in merged images) and either WT Spd-2-mCherry (A) or Spd-2-CONS-mCherry (B) (red in merged images). Pie charts quantify the percentage of centrosomes that were scored qualitatively as having a strong (dark green), weak (light green) or no (white) pericentriolar scaffold (n = 15 individual centrosomes, two images (channels) per centrosome, for each genotype, respectively). All centrosomes were imaged in mid-late S-phase when the centrosomal levels of Spd-2 are maximal (see Figure 8). All scorings were performed blind by researchers not involved in the data acquisition. The defect in scaffold assembly is stronger in the Spd-2 mutant embryos expressing Polo-GFP and Spd-2-CONS-mCherry when compared to Spd-2 mutant embryos expressing just Spd-2-CONS-GFP (Figure 4). This appears to be due to a genetic interaction between Polo-GFP and Spd-2-CONS-mCherry, as mutant embryos expressing just Spd-2-CONS-mCherry had a similar phenotype to embryos expressing just Spd-2-CONS-GFP (data not shown) (see main text).

https://doi.org/10.7554/eLife.50130.024

We next used 3D-SIM to examine the distribution of RFP-Cnn in living Spd-2 mutant embryos expressing WT Spd-2-GFP or Spd-2-CONS-GFP. In WT Spd-2-GFP embryos, RFP-Cnn spread outwards along the centrosomal MTs forming a robust scaffold that extended beyond the Spd-2-GFP scaffold (Figure 7A), as reported previously (Conduit et al., 2014b). In contrast, in Spd-2-CONS-GFP embryos only an occasional protrusion of RFP-Cnn and Spd-2-CONS-GFP could be detected extending outwards from the centriole (arrowheads, Figure 7B). These relative distributions of RFP-Cnn were confirmed and quantified by ‘radial-profiling’ using data obtained on a standard spinning disk confocal system (Conduit et al., 2014b) (Figure 7C). Strikingly, radial-profiling also revealed how the RFP-Cnn scaffold (red lines, Figure 7D) normally extends beyond the Spd-2-GFP scaffold (green lines in Figure 7D) in WT embryos (Figure 7Di), but these distributions essentially overlap in Spd-2-CONS-GFP embryos (Figure 7Dii). Thus, if Spd-2 cannot efficiently recruit Polo via the PBD, the Cnn scaffold cannot efficiently expand outwards beyond the Spd-2 scaffold.

Figure 7 with 1 supplement see all
Cnn is recruited to, and phosphorylated at, centrioles in Spd-2-CONS embryos, but it does not efficiently assemble into an extended scaffold.

(A,B) Micrographs show 3D-SIM images of individual centrosomes from living Spd-2 mutant embryos expressing RFP-Cnn (red in merged images) and either WT Spd-2-GFP (A) or Spd-2-CONS-GFP (B) (green in merged images). Arrowheads indicate examples of occasional protrusions of RFP-Cnn and Spd-2-CONS-GFP. (C,D) Graphs compare the radial distributions of RFP-Cnn around the mother centriole in WT Spd-2-GFP and Spd-2-CONS-GFP embryos (C), or the radial distribution of RFP-Cnn and Spd-2-GFP in either WT Spd-2-GFP (D[i]) or Spd-2-CONS-GFP embryos (D[ii]). Data for these graphs was obtained from living embryos examined on a spinning disk confocal system; five centrosomes per embryo were analysed: n = 8 and 7 embryos for WT and CONS embryos, respectively. (E,F) Micrographs show 3D-SIM images of individual centrosomes from Spd-2 mutant embryos expressing either WT Spd-2-GFP (E) or Spd-2-CONS-GFP (F) that were fixed and stained with antibodies against GFP, phospho-Cnn, or total-Cnn (green, red and cyan in merged images, respectively).

https://doi.org/10.7554/eLife.50130.027

Cnn is normally phosphorylated at centrosomes in a Polo-dependent manner, and this allows Cnn to assemble into a scaffold (Conduit et al., 2014a; Feng et al., 2017). We tested whether Cnn could still be phosphorylated by Polo at centrosomes in Spd-2-CONS-GFP embryos using an antibody that specifically recognises a phospho-epitope in Cnn that is phosphorylated by Polo (Feng et al., 2017) (Figure 7E,F). Phosphorylated Cnn was still strongly detected around the mother centriole in Spd-2-CONS-GFP embryos. This phosphorylation is presumably dependent upon the Polo that is still recruited to the centrioles in Spd-2-CONS-GFP embryos (Figure 6B). We conclude that a Polo/Spd-2/Cnn ‘mini-scaffold’ assembles around the mother centriole even if Spd-2 cannot recruit Polo via the PBD; this scaffold, however, is unable to efficiently expand outwards around the mother centriole.

As the Polo/Spd-2/Cnn scaffold is essential for mitotic centrosome assembly in flies (Conduit et al., 2014b; Dobbelaere et al., 2008; Feng et al., 2017), the inability of these proteins to form an expanded scaffold in Spd-2-CONS embryos should lead to a failure to recruit any other PCM proteins to an expanded mitotic PCM scaffold. This appeared to be the case, as the centrosomal recruitment of the PCM components Aurora A-GFP (Figure 7—figure supplement 1A–D) and γ-tubulin (Figure 7—figure supplement 1E–H) was greatly reduced in Spd-2-CONS-GFP embryos. Thus, the failure to form an expanded pericentriolar scaffold in these embryos appears to lead to a general failure in centrosome maturation.

Centrosome maturation fails in Spd-2-CONS-GFP embryos

To directly examine the kinetics of centrosome maturation in Spd-2-CONS-GFP embryos we quantified the centrosomal levels of WT Spd-2-GFP and Spd-2-CONS-GFP through an entire embryonic cell-cycle. As described above, in these rapidly dividing syncytial embryos the two centrosomes separate at the start of S-phase and immediately start to mature in preparation for the next M-phase. WT Spd-2-GFP started to accumulate at the maturing centrosomes in early S-phase, reached maximal levels just before nuclear envelope breakdown, and then started to decline as the embryos entered mitosis (blue line, Figure 8; Figure 8—figure supplement 1). In contrast, centrosomal levels of Spd-2-CONS-GFP remained at a constant low level throughout the cycle (red line, Figure 8; Figure 8—figure supplement 2). These results suggest that centrosome maturation fails in Spd-2-CONS and Spd-2-ALL mutant embryos.

Figure 8 with 3 supplements see all
Centrosome maturation dynamics in Spd-2-CONS embryos.

Graph compares the mean (± SD) centrosomal Spd-2-GFP intensity (in arbitrary units [a.u.]) through an entire embryonic cell cycle (nuclear cycle 12) in Spd-2 mutant embryos expressing either WT Spd-2-GFP (blue) or Spd-2-CONS-GFP (red); n = 155 and 75 centrosomes analysed, respectively. Time in seconds is indicated, and the time when centrosomes first separate at the start of S-phase is set as t = 0; the time of mitotic entry—scored as the time of nuclear envelope breakdown (NEB)—is indicated by the dotted vertical lines. Because the length of S-phase varies in individual embryos—384s ± 46s or 369s ± 30s (mean ± SD) for WT Spd-2-GFP and Spd-2-GFP-CONS, respectively—we cannot simply average the data at each time point from multiple embryos, so representative embryos are shown here. The analysis of all 14 WT Spd-2-GFP (blue) or Spd-2-CONS-GFP (red) embryos that were monitored in this way is shown in Figure 8—figure supplement 1 and Figure 8—figure supplement 2, respectively.

https://doi.org/10.7554/eLife.50130.032

Multiple regions of Spd-2 appear to help recruit Polo to the mitotic PCM scaffold

Studies in worms and frogs have concluded that a single S-S/T(p) motif in SPD-2/Cep192 is required to recruit PLK-1/Plk1 to centrosomes (Decker et al., 2011; Joukov et al., 2010; Joukov et al., 2014) while in human cells a second S-S/T(p) motif also plays a part (Meng et al., 2015). Our data, however, raise the possibility that multiple S-S/T(p) motifs in Drosophila Spd-2 may help recruit Polo to the mitotic PCM. To examine whether Polo recruitment by fly Spd-2 could be linked to any of the previous S-S/T(p) motifs identified, we used a previously established assay in which mRNAs encoding mKate2-tagged Spd-2-fusion proteins are injected into embryos expressing Polo-GFP; the mRNAs are gradually translated so the fusion proteins eventually out-compete the endogenous (unlabelled) WT Spd-2 and their effect on Polo-GFP recruitment can be assessed (Novak et al., 2016) (Figure 9A).

Figure 9 with 2 supplements see all
No single S-S/T motif in Spd-2 is essential for recruiting Polo to centrosomes.

(A) Schematic illustration of the mRNA injection assay used to analyse the effect of various Spd-2-mKate2 fusion proteins on Polo-GFP recruitment. (B) Graph compares the radial distribution of Polo-GFP around the mother centriole in living WT embryos expressing Polo-GFP and injected with mRNAs encoding either WT Spd-2-mKate2, Spd-2-CONS-mKate2 or Spd-2-ALL-mKate2, as indicated; five centrosomes per embryo were analysed: n = 7, 6 and 9 embryos, respectively. Insets show examples of typical spinning disk confocal images used for this analysis. (C) Graph shows the same data as shown in (B), but normalised so that the peak intensity of all genotypes = 1. This emphasises how even if the centrosomal Polo-GFP signal is normalised for fluorescence intensity, Polo-GFP spreads out around the mother centriole to a lesser extent in the embryos expressing Spd-2-CONS-mKate2 or Spd-2-ALL-mKate2 than WT Spd-2-mKate2. These observations recapitulate our findings from transgenic lines expressing these Spd-2-GFP-fusions in a Spd-2 mutant background. (D,E) Same analysis as presented in panels (B,C), but analysing the distribution of Polo-GFP in embryos expressing either Spd-2-ALL-mKate2, or one of two versions of Spd-2 in which the potential Polo binding sites in either the N-terminal or C-terminal region of Spd-2 have been mutated, as indicated; five centrosomes per embryo were analysed: n = 13, 6, 13 and 10 embryos, respectively.

https://doi.org/10.7554/eLife.50130.038

We first confirmed that the centrosomal recruitment of Polo-GFP was severely compromised by the injection of mRNAs encoding Spd-2-CONS-mKate2 or Spd-2-ALL-mKate2 (Figure 9B,C), and then assessed the contribution of various S-S/T motifs to Polo recruitment. The single S-S(p) motif in SPD-2 that recruits PLK-1 to centrosomes in C. elegans is a potential CDK1 substrate that may be conserved in Drosophila (T516, blue box, Figure 9—figure supplement 1). We mutated this site, together with the only other conserved S-S/T motif that is a potential CDK1 substrate in Drosophila Spd-2 (S625, yellow box, Figure 9—figure supplement 1) to Ala. This construct (Spd-2-AA-mKate2) did not detectably perturb the centrosomal distribution of Polo-GFP, suggesting that Spd-2 can recruit Polo to the PCM without any requirement that it first be primed to do so by Cdk1 phosphorylation (Figure 9—figure supplement 2). Interestingly, all the Plk1-recruiting S-S/T motifs identified in worms and vertebrates are restricted to the N-terminal half of the protein (Figure 9—figure supplement 1). We therefore independently mutated the 17 S-S/T motifs in the N-terminal and C-terminal regions of Spd-2 to T-S/T (Spd-2-NT-mKate2 and Spd-2-CT-mKate2, respectively) (Figure 2A). Both constructs led to a reduction in Polo-GFP levels at the centrosome (Figure 9D,E). Spd-2-NT-mKate2 had the biggest effect, but this was still mild compared to Spd-2-ALL-mKate2 (Figure 9D,E). We conclude that there is no single S-S/T(p) motif in Spd-2 that is essential to recruit Polo to the mitotic PCM, and motifs in both the N- and C-terminal regions can contribute to this process.

Spd-2-ALL and Spd-2-CONS cannot efficiently recruit Polo to the PCM, even when a PCM scaffold is present

An important caveat in interpreting our data is that we cannot be certain that the mutants prevent efficient mitotic PCM expansion because they cannot recruit Polo efficiently. Perhaps these mutations prevent PCM expansion for some other reason, and so Polo cannot be recruited to the expanded mitotic PCM because it simply does not exist in these embryos. Our mRNA injection assay potentially allowed us to address this issue. We looked for embryos injected with mRNA encoding Spd-2-CONS-mKate2 where the loss of Polo-GFP from the centrosome was just becoming apparent. We then used 3D-SIM to compare the centrosomal recruitment of mutant Spd-2-mKate2 and Polo-GFP to similarly staged controls injected with WT Spd-2-mKate2 mRNA (Figure 10). We found that Spd-2-CONS-mKate2 could often still be detected in an expanded PCM even when the amount of Polo recruited to the scaffold was severely reduced. Thus, in these centrosomes at least, an expanded PCM is still present (presumably because some unlabelled endogenous Spd-2 is still present), but the recruitment of Polo is severely reduced (presumably because the Spd-2-CONS-mKate2 in the scaffold cannot recruit Polo efficiently). These data strongly support our conclusion that the mutant Spd-2 proteins cannot efficiently recruit Polo to the expanded PCM.

The recruitment of Polo-GFP to the PCM is perturbed in embryos expressing Spd-2-CONS-mKate2, even if a PCM scaffold is still detectable.

(A,B) Micrographs show 3D-SIM images of individual centrosomes from embryos expressing Polo-GFP (green in merged images) injected with mRNA encoding either WT Spd-2-mKate2 (A) or Spd-2-CONS-mKate2 (B) (red in merged images). Pie charts quantify the percentage of centrosomes that were scored qualitatively as having a strong (dark green), weak (light green) or no (white) pericentriolar scaffold (n = 10 and 11 individual centrosomes, two images (channels) per centrosome, for WT and CONS injections, respectively). All centrosomes were imaged in mid-late S-phase when the centrosomal levels of Spd-2 are maximal (see Figure 8). All scorings were performed blind by researchers not involved in the data acquisition. Note that mKate2 is relatively slow folding, and the fusion proteins are just expressed from the injected mRNA, so the signal-to-noise ratio is low and the 3D-SIM images reconstruct relatively poorly. Nevertheless the presence of a Spd-2-CONS-mKate2 scaffold is clear, even when the Polo-GFP signal in the scaffold is very weak (B).

https://doi.org/10.7554/eLife.50130.048

Discussion

Centrosome maturation appears to be a near-universal feature of the metazoan cell cycle. Although many of the key proteins required for centrosome maturation have been identified, how these proteins drive this process is unclear. We have previously proposed that three proteins—Spd-2, Polo and Cnn—together form a scaffold that expands around the mother centriole to recruit other PCM components to the mitotic centrosome (Conduit et al., 2014b; Conduit et al., 2015). The data presented here suggests that these three proteins cooperate to form a positive feedback loop that drives the dramatic expansion of the mitotic PCM scaffold in fly embryos (Figure 11A).

Spd-2, Polo and Cnn appear to form a positive feedback loop that drives the expansion of the mitotic PCM scaffold.

(A) A schematic summary of the proposed positive feedback loop that drives the expansion of the mitotic PCM scaffold in Drosophila embryos. Solid black lines indicate recruitment, solid orange line indicates phosphorylation, dashed line indicates that Cnn does not recruit Spd-2, but rather helps to stabilise the Spd-2 scaffold that has been recruited to the mother centriole and is fluxing outwards. Polo may phosphorylate Spd-2 to create additional S-S/T(p) motifs that can then recruit more Polo, but this is not depicted here. This circuit is a classical positive feedback loop; ultimately, however, it relies on the mother centriole as a source of Spd-2, because Cnn itself cannot recruit more Spd-2 or Polo into the scaffold. (B,C) Schematics illustrate the process of centrosome maturation in a WT cell (B), and a cell in which Spd-2 cannot recruit Polo (C). During interphase (i), Spd-2, Polo and Cnn are all recruited to the mother centriole. Polo is inactive, Spd-2 and Cnn are not phosphorylated, so no scaffold forms. As cells prepare to enter mitosis (ii), Polo is activated, and the centrosomal Spd-2 and Cnn are phosphorylated, allowing them to initially assemble into a ‘mini-scaffold’ around the mother centriole. The phosphorylated Spd-2 scaffold then starts to flux away from the mother centriole (red arrows). In normal cells (B[iii]), the expanding Spd-2 scaffold recruits more Cnn and more Polo, allowing more Cnn scaffold to assemble. The Cnn scaffold cannot recruit more Spd-2 or Polo, but it stabilises the expanding Spd-2 scaffold; this allows the Spd-2 scaffold to accumulate around the mother centriole. This creates a positive feedback loop that drives an increasing rate of expansion of the Spd-2 and Cnn scaffolds around the mother centriole. If the expanding Spd-2 scaffold cannot recruit Polo (C[iii]), the Cnn recruited to the expanding Spd-2 scaffold is too far away from the centriole to get phosphorylated by Polo so it cannot form a scaffold and rapidly dissipates into the cytosol (orange arrows). The positive feedback loop is broken, and centrosome maturation fails.

https://doi.org/10.7554/eLife.50130.050

We propose the following model (Figure 11B). In interphase cells, Spd-2, Polo and Cnn are recruited around the surface of the mother centriole (Fu and Glover, 2012), but Polo is inactive and Spd-2 and Cnn are not phosphorylated—so no scaffold is assembled (Figure 11Bi). As cells prepare to enter mitosis (Figure 11Bii), centrosomal Spd-2 becomes phosphorylated. Our in vitro data suggests that Polo is involved in this phosphorylation (via a ‘self-priming and binding’ mechanism), but other mitotic kinases may also be involved. Phosphorylation allows Spd-2 to form a scaffold (red, Figure 11Bii) that fluxes outwards (red arrows, Figure 11Bii) and that can recruit both Polo (via phosphorylated S-S/T(p) motifs) and Cnn (Conduit et al., 2014b). The active Polo (blue dots, Figure 11Bii) phosphorylates Cnn (orange arrow, Figure 11A), allowing it to also form a scaffold (green, Figure 11Bii) (Conduit et al., 2014a; Feng et al., 2017). The Spd-2 scaffold is inherently unstable (Conduit et al., 2014b), so it can only accumulate around the mother centriole if it is stabilised by the Cnn scaffold (dotted arrow, Figure 11A). The Cnn scaffold therefore allows the Spd-2 scaffold to expand outward, increasing Spd-2 levels within the PCM scaffold and allowing Spd-2 to recruit more Cnn and more Polo into the scaffold. This is a classical positive feedback loop in which the Output (the PCM scaffold in toto) directly increases the Input (the Spd-2 scaffold).

If Spd-2 cannot efficiently recruit Polo, as appears to be the case with the Spd-2-ALL and Spd-2-CONS mutants, it can still recruit Cnn, and this is, at least initially, phosphorylated by the pool of Polo that is still present around the mother centriole (Figure 11C). Our data suggests that this centriolar pool of Polo is not recruited by Spd-2 (at least not via the PBD), and we suspect that S-S/T(p) motifs in other centriole proteins, such as Sas-4 (Novak et al., 2016), normally recruit Polo to centrioles. As a result, mutant Spd-2 proteins can still support the assembly of a ‘mini-scaffold’ around the mother centriole, and this can recruit some PCM and organise some MTs (Figure 11Cii). The mutant Spd-2 scaffold that fluxes outwards from the mother centriole, however, cannot recruit Polo. Therefore the Cnn recruited by the expanding Spd-2 network cannot be phosphorylated, and it cannot form a scaffold to support the expanding Spd-2 network. As a result, the expanding mitotic PCM scaffold rapidly dissipates into the cytosol (Figure 11Ciii).

Although this mechanism is autocatalytic—as the expanding Spd-2 scaffold allows Polo and Cnn to be recruited into the PCM at an increasing rate—crucially, the mother centriole remains the only source of Spd-2 (Figure 11A). This potentially explains the conundrum of how mitotic PCM growth is autocatalytic (Zwicker et al., 2014), but at the same time requires the mother centriole (Basto et al., 2006; Cabral et al., 2019; Kirkham et al., 2003). This requirement for centrioles can also potentially explain how two spatially separated centrosomes usually grow their mitotic PCM to the same size (Conduit et al., 2015; Raff, 2019), as PCM size may ultimately be determined by how much Spd-2 can be provided by the centrioles, rather than how much PCM was present in the centrosome when maturation was initiated.

A key feature of this proposed mechanism is that Cnn cannot recruit itself or Spd-2 or Polo into the scaffold (although it helps to maintain the Spd-2 scaffold recruited by the centriole; Figure 11A). If it could do so, mitotic PCM growth would no longer be constrained by the centriole as Cnn could catalyse its own recruitment. Interestingly, although Spd-2 and Cnn are of similar size in flies (1146aa and 1148aa, respectively) Spd-2 has >5X more conserved potential PBD-binding S-S/T motifs than Cnn (Figure 2—figure supplement 1). Moreover, a similar ratio of conserved sites is found when comparing human Cep192 (1941aa) to human Cep215/Cdk5Rap2 (1893aa) (Figure 2—figure supplement 1), even though the human and fly homologues of both proteins share only limited amino acid identity. Perhaps, these two protein families have evolved to ensure that phosphorylated Spd-2/Cep192 can efficiently recruit Polo/Plk1, whereas phosphorylated Cnn/Cep215 cannot.

Our data indicates that multiple S-S/T(p) motifs in Spd-2 may be involved in Polo recruitment to the PCM. When only the most conserved motifs are mutated, other motifs in Spd-2 appear to be able to help recruit Polo, as evidenced by the additive effect of the Spd-2-ALL mutant compared to the Spd-2-CONS mutant. This mechanism of multi-site phosphorylation and recruitment could help amplify the maturation process (as the additional Polo recruited would allow Cnn to be phosphorylated at a higher rate) and so contribute to the establishment of the positive feedback loop.

Another important feature of this proposed mechanism is that Spd-2 is incorporated into the mitotic PCM at the centriole surface and then fluxes outwards (Conduit et al., 2014b). This Spd-2-flux has so far only been observed in Drosophila embryos and mitotic brain cells (Conduit et al., 2015; Conduit et al., 2010; Conduit et al., 2014b). In fly embryos, Cnn also fluxes outwards but, unlike Spd-2, this flux requires MTs and is only observed in embryos (Conduit and Raff, 2015). In C. elegans embryos, SPD-5 behaves like Cnn in somatic cells: it does not flux outwards and is incorporated isotropically throughout the volume of the PCM (Laos et al., 2015). Moreover, a very recent study found no evidence for an outward centrosomal flux of SPD-2 in worm embryos (Cabral et al., 2019). Clearly, it will be important to determine whether Spd-2/Cep192 homologues flux outwards in other species and, if so, whether this flux provides the primary mechanism by which the mother centriole influences the growth of the expanding mitotic PCM.

In vertebrates, Cep192 serves as a scaffold for Plk1 and also Aurora A (Joukov et al., 2010; Joukov et al., 2014; Meng et al., 2015)—another mitotic protein kinase that plays an important part in centrosome maturation in many species (Barr and Gergely, 2007). There appears to be a complex interplay between Cep192, Plk-1 and Aurora A in vertebrates, with Cep192 acting as a scaffold that allows these two important regulators of mitosis to influence each other’s activity and centrosomal localisation. Spd-2 clearly plays an important part in recruiting Aurora A to centrosomes in fly cells (Conduit et al., 2014b; Dobbelaere et al., 2008)—although it is unclear if this is direct, as fly and worm Spd-2/SPD-2 both lack the N-terminal region in vertebrate Cep192 that recruits Aurora A (Meng et al., 2015). How Aurora A might influence the assembly of the Spd-2, Polo/PLK-1 and Cnn/SPD-5 scaffold remains to be determined, although in worms AIR-1 (the Aurora A homologue) is required to initiate centrosome maturation, but is not required for subsequent PCM growth (Cabral et al., 2019).

Finally, there has been great interest recently in the idea that many non-membrane bound organelles like the centrosome may assemble as ‘condensates’ formed by liquid-liquid phase separation (Banani et al., 2017; Boeynaems et al., 2018; Raff, 2019). In support of this possibility for the centrosome, purified recombinant SPD-5 can assemble into condensates in vitro that have transient liquid-like properties, although they rapidly harden into a more viscous gel- or solid-like phase (Woodruff et al., 2017). Moreover, a mathematical model that describes centrosome maturation in the early worm embryo treats the centrosome as a liquid, and it is from this model that the importance of autocatalysis was first recognised (Zwicker et al., 2014). In vivo, however, the Cnn and SPD-5 scaffolds do not appear to be very liquid-like (Conduit et al., 2010; Conduit et al., 2014b; Laos et al., 2015) and fragments of Cnn can assemble into micron-scale assemblies in vitro that are clearly solid- or very viscous-gel-like (Feng et al., 2017). Our data suggests that the incorporation of Spd-2 into the PCM only at the surface of the centriole, coupled to an amplifying Spd-2/Polo/Cnn positive feedback loop, could provide an ‘autocatalytic’ mechanism that functions within the conceptual framework of a non-liquid-like scaffold that emanates from the mother centriole.

Materials and methods

Key resources table

Request a detailed protocol

See Supplementary file 1.

Fly husbandry, stocks and handling

Request a detailed protocol

Flies were kept at 25°C or 18°C on Drosophila culture medium (0.77% agar, 6.9% maize, 0.8% soya, 1.4% yeast, 6.9% malt, 1.9% molasses, 0.5% propionic acid, 0.03% ortho-phosphoric acid and 0.3% nipagin). Stocks were kept in 8 cm x 2.5 cm plastic vials or 0.25-pint plastic bottles. Embryos were collected on cranberry-raspberry juice plates (25% cranberry-raspberry juice, 2% sucrose and 1.8% agar) supplemented with fresh yeast. Standard fly handling techniques were employed (Roberts, 1998). In vivo studies were performed using 1.5–2 hr-old syncytial blastoderm stage embryos. After 0–1 hr collections at 25°C, embryos were aged at 25°C for 30–60 min. When injecting mRNA, embryos were collected for 20 min, injected, and imaged after 120–150 min at 21°C (but always within the syncytial blastoderm stage of development). Prior to injection or imaging, embryos were dechorionated on double-sided tape and mounted on a strip of glue onto a 35-mm glass-bottom petri dish with a 14 mm micro-well (MatTek). After desiccation for 1 min (non-injection experiments) or 3 min (pre-mRNA injection) at 25°C, embryos were covered in Voltalef oil (ARKEMA). Live imaging was performed using either the spinning disk confocal or the 3D-SIM systems described below.

Transgenic Drosophila lines

Request a detailed protocol

Potential Polo binding sites in the amino acid sequence of Drosophila melanogaster Spd-2 were identified by searching for the consensus Polo binding motif S-S/T. Site conservation was assessed using FlyBase BLAST (selecting the genus Drosophila) and Jalview for protein alignment. The ALL and CONS constructs were designed in silico and synthesised externally by GENEWIZ Co. Ltd. (Suzhou, China); the WT spd-2 cDNA was obtained from Geneservice Ltd (UK). The WT, ALL and CONS Spd-2 cDNAs were cloned into a pDONR-Zeo vector and then introduced in Ubq-GFPCT and Ubq-mCherryCT destination vectors via Gateway cloning as indicated (Key Resources Table). The Ubq-Spd-2-11A-GFP plasmid was derived via site-directed mutagenesis on pDONRSpd-2 vector using QuikChange Multi Site-Directed mutagenesis followed by Gateway cloning into the Ubq-GFPCT vector. The plasmids for monomeric and dimeric NeonGreen expressed from the Sas-6 or Plk4 promoter were generated using the NEBuilder HiFi DNA Assembly (New England Biolabs). The fluorophores of dimeric NeonGreen were linked with a five amino acid-long peptide to minimise energy transfer between them. The 2 kb upstream of the start codon and 1 kb downstream of the stop codon of Sas-6/Plk4 were amplified from Oregon-R genomic DNA. The amplified fragments were cloned into the pDONR-Zeo vector. The plasmids were sent to BestGene Inc (Chino Hills, California) or the University of Cambridge Genetics Fly Facility (UK) for generation of the transgenic lines via random P-element insertion in to a w1118 background. Other GFP, RFP, and mCherry lines have been described previously (see Key Resources Table).

For the Spd-2 mutant embryo analyses we used embryos laid by spd-2Z35711/spd-2 Df(3L)st-j7 or spd-2Z35711/spd-2G20143 transheterozygotes expressing two copies of the Spd-2-GFP fusions, or one copy of a Spd-2 fusion and one copy of another fusion protein. Drosophila melanogaster Oregon-R and w67 were used as a WT stock where indicated. Balancer chromosomes and markers used were described previously (FlyBase, USA).

Centrosome purification

Request a detailed protocol

Whole centrosomes were isolated from extracts of early Drosophila embryos (0–4 hr) using a modified version of a centrosome isolation protocol (Lehmann et al., 2006). Embryo extract containing 50% sucrose was layered on top of a sucrose cushion comprising 55% and 70% sucrose. The tubes were spun at 27,000 rpm, causing the centrosomes in the extract to move through the 55% layer and into the 70% sucrose layer. A ‘Cytosolic’ fraction was collected from the top of the tube, and fractions were then collected from the bottom of the tube. Western blotting was performed to identify the ‘Centrosome’ fractions that contained the greatest enrichment of centrosomal proteins. Phosphatase treatment was carried out on the centrosome fractions using alkaline phosphatase (Roche) for 4.5 hr at 37°C with or without phosphatase inhibitor cocktails 2 and 3 (Sigma).

Centrosome immunoprecipitation and mass spectrometry

Request a detailed protocol

Centrosomes were immunoprecipitated from the centrosomal fractions using anti-Cnn antibodies coupled to protein A conjugated magnetic Dynabeads (Life Technologies). Cytoplasmic Spd-2 was immunoprecipitated from the cytoplasmic fractions using anti-Spd-2 antibodies coupled to Dynabeads. The dynabead/antibody suspensions were rotated at 4°C overnight. The antibodies were cross-linked to the beads using the BS3 crosslinker (Thermo Fisher). Centrosomal and cytoplasmic fractions were diluted 1:1, added to the antibody-crosslinked beads and rotated at 4°C for 2 hr. Beads were washed, boiled in sample buffer (SB) and separated on a polyacrylamide gel, and the band containing Spd-2 was cut out. Samples were prepared for mass spectrometry and enriched for phosphopeptides as described previously (Conduit et al., 2014a). Liquid chromatography-MS/MS analysis was performed using a LTQ Orbitrap Mass Spectrometer (Thermo Scientific) coupled to an UltiMate 3000 Nano LC system (Thermo Scientific). The mass spectrometry data were searched against the FlyBase sequence database (http://flybase.bio.indiana.edu/) using Mascot software (Matrix Science). The following settings were used for the searches: enzyme: trypsin; fixed modification: carbamidomethylation; variable modifications: methionine oxidation, glutamine/asparagine deamidation; serine/threonine/tyrosine phosphorylation; error tolerance for the precursor ions, 20 ppm; mass error tolerance for the fragment ions, 0.6 Da; number of missed cleavage sites, 3. The MS/MS spectra for identified phosphopeptides were manually inspected in Mascot.

Recombinant protein expression and purification

Request a detailed protocol

The cDNA sequences encoding Drosophila Spd-2352-758 (WT and ALL mutant) were subcloned into a pETM44 (EMBL) vector encoding an N-terminal His6-MBP tag. Proteins were expressed in Escherichia coli (E. coli) B21 strains in LB, and purified using a pre-poured amylose column containing 4 mL amylose resin (New England Biolabs) followed by size exclusion chromatography (protein buffer: 20 mM Tris pH 8.0, 150 mM NaCl, 0.5 mM TCEP) using an AKTA pure chromatography system with a HiLoad-Superdex 200 16/600 column attached (GE Healthcare).

In vitro interaction assays

Request a detailed protocol

Anti-MBP antibody was coupled to magnetic beads (7.5 μg of antibody per 1 mg of beads) using the Dynabeads Antibody Coupling Kit (Thermo Fisher), following manufacturer’s instructions. Each sample (100 μL of resuspended beads) was incubated with 32.2 μg of the appropriate protein in protein buffer (see above) for 30 min rotating at RT. The beads were rinsed twice with kinase buffer (CST) and resuspended in 60 μL of kinase buffer containing 200 μM of ATP (CST) and either kinase storage buffer (50 mM HEPES pH 7.6, 100 mM NaCl, 5 mM DTT, 20% glycerol, 15 mM reduced glutathione), for non-phosphorylated ‘blank’ controls; or 8.8 ng/μL of commercial PLK1 kinase (ProQinase) for phosphorylated samples. The samples were rinsed 3X with binding buffer (50 mM Tris pH 8.0, 200 mM NaCl, 1 mM DTT, 0.1% Tween-20, 10 mg/mL BSA, 1X phosphatase inhibitor cocktails 2 and 3, 1X SIGMAFAST EDTA-free protease inhibitor cocktail (Sigma)). The beads were then resuspended in 0.3 mL of 0.2 µM GST-Plk1-PBD (Sigma) in binding buffer, and incubated rotating for 3 hr at 4 °C. The beads were then rinsed 3X with 500 µL of bead wash buffer B (50 mM Tris pH 8.0, 200 mM NaCl, 1 mM DTT, 0.1% Tween-20, 1 mg/mL BSA, 1X phosphatase inhibitor cocktails 2 and 3, 1X SIGMAFAST EDTA-free protease inhibitor cocktail), transferred to a clean tube, and rinsed once with bead wash buffer HA (same as wash buffer B, but without BSA) before protein elution with 1X SB.

Western blot analysis

Request a detailed protocol

Western blotting to estimate embryonic protein levels was performed as described previously (Novak et al., 2014). The following primary antibodies were used for western blot analysis (see Key Resources Table): rabbit anti-Spd-2 (1:500), rabbit anti-Cnn (1:1000), mouse anti-γ-tubulin (1:500), mouse anti-Actin (1:2000), mouse anti-GST (1:500) and rabbit anti-GAGA factor (1:500). HRP conjugated secondary antibodies used (all at 1:3000): swine anti-rabbit (Dako), or ECL anti-mouse and ECL anti-rabbit (GE Healthcare).

RNA synthesis and microinjection

Request a detailed protocol

The mRNA injection assay and the modified pRNA destination vector with the C-terminal mKate2 tag used here have been described previously (Novak et al., 2014; Novak et al., 2016). Spd-2-ALL and Spd-2-CONS cDNA were introduced into the vector via Gateway cloning. Two point mutations were introduced into WT Spd-2-mKate2 using QuikChange mutagenesis (Agilent) to generate Spd-2-AA-mKate2. Spd-2-NT-mKate2 and Spd-2-CT-mKate2 partial mutants were derived from PCR-amplified fragments of WT Spd-2-mKate2 and Spd-2-ALL-mKate2 via NEBuilder HiFi assembly (New England Biolabs). The last potential binding site mutated in the N-terminus group was S538-S540, and the first potential binding site mutated in the C-terminus group was S581-S582, so that each group would include 17 potential sites. In vitro RNA synthesis was performed using a T3 mMESSAGE mMACHINE kit (Thermo Fisher) and RNA was purified using an RNeasy MinElute kit (Qiagen). All RNA constructs were injected at a concentration of 2 mg/mL.

Immunofluorescence

Embryos were collected for 0–1 hr, aged for 45–60 min, and processed as described (Stevens et al., 2010). Samples were mounted onto microscopy slides with high-precision glass coverslips (CellPath). Specifics for each experiment as follows:

Quantification of successful completion of pronuclear fusion

Request a detailed protocol

Embryos were stained using a mouse anti-α-tubulin (1:1000), followed by Alexa 594 nm anti-mouse and GFP-Booster Atto488 (1:500 dilution). Samples were mounted in Vectashield medium with DAPI. Embryos were counted using a Zeiss Axioskop two microscope (Zeiss International) with a 10x/0.30-NA and a 40x/0.75-NA objectives. Embryos were counted as developing beyond pronuclear fusion if they had clearly reached syncytial/gastrulation stages. For each of the four conditions (non-rescue, WT-rescue, CONS-rescue and ALL-rescue), we performed two biological replicates (embryos from separate sets of mothers), each with three technical replicates (embryos collected and processed independently);>50 embryos were counted per sample.

Phospho-Cnn staining

Request a detailed protocol

Embryos were stained using a guinea pig anti-Cnn antibody (1:1000) and a rabbit anti-Cnn pSer567 antibody (1:500); followed by Alexa 594 nm anti-rabbit, CF405S anti-guinea pig, and GFP-Booster Atto488 (1:500 dilution). Samples were mounted in Vectashield medium without DAPI.

γ-tubulin staining

Request a detailed protocol

Embryos were stained using a mouse anti-γ-tubulin antibody (1:500), followed by Alexa 594 nm anti-mouse and GFP-Booster Atto488 (1:500 dilution). Samples were mounted in Vectashield medium with DAPI.

Imaging

Spinning disk confocal microscopy

Request a detailed protocol

Embryos were imaged at 21°C on a Perkin Elmer ERS spinning disk (Volocity software) mounted on a Zeiss Axiovert 200M microscope using a 63X/1.4-NA oil immersion objective and an Orca ER CCD camera (Hamamatsu Photonics, Japan). 488- and 561 nm lasers were used to excite GFP and RFP/mCherry, respectively. Confocal sections of 13 slices with 0.5-μm-thick intervals were collected every 30 s (17 slices for the analysis of protein dynamics throughout the cell cycle). Focus was occasionally manually readjusted in between intervals.

3D-SIM

Request a detailed protocol

3D-SIM microscopy was performed and analysed as described (Conduit et al., 2014a) on an OMX V3 Blaze microscope (GE Healthcare, UK) with a 60x/1.42-NA oil UPlanSApo objective (Olympus); 405-, 488- and 593 nm diode lasers, and Edge 5.5 sCMOS cameras (PCO). The raw acquisition was reconstructed using softWoRx 6.1 (GE Healthcare) with a Wiener filter setting of 0.006 and channel- specific optical transfer function. Living embryos were imaged at 21°C, acquiring stacks of 6 z-slices (0.125 μm intervals). Stacks of 13 z-slices (0.125 μm intervals) were acquired from fixed samples (phospho-Cnn staining). The images shown are maximum intensity projections. For multi-colour 3D-SIM, images from the different colour channels were registered with alignment parameters obtained from calibration measurements using 1 μm to 0.2 μm TetraSpeck Microspheres (Thermo Fisher) using OMX Editor and Chromagnon alignment software. The SIM-Check plug-in in ImageJ (NIH) was used to assess the quality of the SIM reconstructions (Ball et al., 2015).

Airyscan

Request a detailed protocol

Fixed samples (γ-tubulin staining) were imaged using an inverted Zeiss 880 microscope fitted with an Airyscan detector. The system was equipped with Plan-Apochromat 63x/1.4-NA oil lens. The laser excitation lines used were 405 nm diode, 488 nm argon and 561 nm diode laser. Stacks of 25 slices with 0.14-μm-thick intervals were collected with pixel size (xy) of 0.035 μm, using a piezo-driven z-positioner stage. Images were Airy-processed in 3D with a strength value of ‘auto’ (∼6). The software used to acquire images and process the images taken in super-resolution Airyscan mode was ZEN (black edition, Zeiss).

Image and statistical analysis

Blind analysis of 3D-SIM images

Request a detailed protocol

Centrosome images were selected based on quality of the reconstruction as assessed by the SIM-Check plug-in and the presence of a visible, well-formed ring corresponding to the presence of protein at the mother centriole wall. Each individual centrosome image was saved as a separate file, renamed and randomised post acquisition. The entire dataset for each experiment were scored blind by researchers not involved in any aspect of the data acquisition. The Spd-2-GFP in mutant background dataset was scored by one person. It included three different conditions (WT, CONS and ALL) with 36 centrosomes per condition. The Spd-2-GFP in WT background and Spd-2-mCherry/Polo-GFP datasets included three and two conditions (WT, CONS and ALL; or WT and CONS), respectively. The former included 32 images per condition, and the latter included 30 images per condition (16 and 15 individual centrosomes, two different channels, respectively). They were scored independently by three different people, and an average score was calculated. The Spd-2-mKate2 injection into PoloGFP dataset included two conditions (WT and CONS) with 20 and 22 images each (10 and 11 individual centrosomes, two different channels) and it was scored by one person.

Analysis of centrosome and MT fluorescent intensities

Request a detailed protocol

We used ImageJ to calculate the maximum intensity projection of z-stacks of movies taken from the PE spinning disk system. The time frame chosen for analysis corresponded to 1 min before nuclear envelope breakdown. The five brightest centrosomes per embryo were identified via manual thresholding and analysed; the number of embryos analysed is indicated in each Figure. For both the green and red channels we measured the mean intensity within a square of fixed size (5.04 μm x 5.04 μm) centred manually on each individual centrosome. Similarly, we measured the mean intensity of the background near each centrosome. We calculated the average centrosome intensity and subtracted the average background intensity per embryo. The values for all the embryos were plotted on Prism 7 (GraphPad Software). Prism was also used to check column statistics and Gaussian distribution of the data. For the Jupiter-mCherry/Spd-2-GFP data we used the D’Agostino–Pearson omnibus normality test. For the statistical analysis, we used ordinary one-way ANOVA with Tukey's multiple comparisons test if data passed the normality test, or the Kruskal-Wallis test with Dunn's multiple comparisons test otherwise. For the dataset comparing Spd-2-GFP and Spd-2-11A-GFP we used the Shapiro-Wilk normality test followed by the unpaired t test with Welch’s correction. Significance in statistical tests was defined by p<0.05.

Radial profiling of centrosomes

Request a detailed protocol

We used ImageJ to calculate an average ‘radial profile’ of the distribution of the different PCM proteins around the mother centriole (Conduit et al., 2014b). For embryos imaged live, the five brightest centrosomes in each embryo were analysed (the number of embryos analysed for each genotype is indicated in the individual Figures). For the analysis of fixed embryos, we analysed 1 pair of centrosomes per embryo, five embryos per technical replicate (embryos collected and processed independently), and three technical replicates in total per condition (so total centrosomes analysed = 15).

For each individual centrosome we found its center of mass by thresholding the image and running the ‘analyse particles’ (centre of mass) macro on the most central Z plane of the centrosome, as described (Conduit et al., 2014b). We then centred concentric rings spaced at 0.021 μm and spanning across 2.09 μm on this centre (0.007 μm and spanning across 1.41 μm for the fixed γ-tubulin images) and measured the average fluorescence in each ring, and subtracted the average cytosolic signal. Each individual centrosome profile was then normalised to the average peak intensity for all the centrosomes of the control condition (WT Spd-2-GFP embryos). Each profile was then mirrored to produce a full centrosome profile. The final radial profiles shown are an average of all the full centrosomal profiles per condition. In some graphs, we show a ‘normalised’ profile, where each individual centrosome profile was normalised to the average peak intensity of its corresponding condition (rather than to the WT Spd-2-GFP embryo control). The resulting radial profile peaks for all conditions were then normalised to 1; this allows the distribution of different proteins around the centriole to be compared, independently of differences in centrosomal protein levels.

Analysis and regression modelling for the dynamics of Spd-2-GFP

Request a detailed protocol

Spd-2-GFP dynamics throughout the cell cycle were analysed as described (Aydogan et al., 2018). Briefly, we used ImageJ to calculate the maximum intensity projection of z-stacks of movies taken from the PE spinning disk system. The backgrounds were subtracted using the subtract background function with a rolling ball radius of 10 pixels. Spd-2-GFP foci (centrosomes) were tracked using TrackMate plug-in (Tinevez et al., 2017) with the following analysis settings: track spot diameter size of 2.1 μm, initial threshold of 0, and quality of >0.03. Regression analysis on the centrosome growth curves were carried out using Prism, and the mathematical modelling was done using the nonlinear regression (curve fit) analysis function, excluding the last three points (90 s) of the cell cycle (as the data was very variable towards the end of mitosis).

The data for WT Spd-2-GFP (Figure 8, blue line) was fitted against four different functions to assess the most suitable model (Figure 8—figure supplement 3A): (1) linear growth followed by linear decrease; (2) linear growth followed by plateau followed by linear decrease; (3) Gaussian function; (4) Lorentzian function. Functions (1) and (2) are bespoke algorithms, with the following equations for X amount of time:

(1) Y1 = intercept1 + slope1X YX0 = slope1X0 + intercept1Y2 = Yx0+ slope2(XX0)Y = IF(X<X0,Y1,Y2)
(2) Y1 = intercept1 + slope1X  YX0 = slope1X0 + intercept1Y2 = YX0+ slope2(XX0)slope2=0YX1 = YX0+ slope2(X1X0)Y3 = YX1+ slope3(XX1)Y = IF(X<X0,Y1,IF (X<X1,Y2, Y3))

The only constraints applied to these equations were the requirements for slope1 and inflection points (X0, X1) to be greater than 0, and slope3 to be less than 0. Centrosomes that come from a single embryo were treated as internal replicates, and thus the fitting used only the mean Y value of each time point. To judge and control the quality and precision of regression (goodness-of-fit), we used the R2, adjusted R2, and absolute sum-of- square values. To compare the fits, the extra sum-of-squares F test was applied, and the appropriate fit was chosen by selecting the simpler model unless p<0.05. The ‘linear growth followed by plateau followed by linear decrease’ model best fit the data (Figure 8—figure supplement 3A), but it is likely a simplification of a more complex model, so individual curves are shown for each embryo without any model fitted (average of >44 centrosomes per embryo) (Figure 8—figure supplement 1).

The data for Spd-2-CONS-GFP (Figure 8, red line) was fitted against the ‘linear growth followed by plateau followed by linear decrease’ model, as this was the preferred model for the WT data. As this data seemed to better be described as a straight line, this model was also compared to a standard straight line model (with no slope constraints)—function (5); or a user-defined constant line function (6) (Figure 8—figure supplement 3B):

Y = intercept1 + 0*X

The preferred model was the straight line—function (5) (Figure 8—figure supplement 3B)—although the average slope value was nearly zero (−0.0005 ± 0.0054; mean ± SD), indicating that the appropriate model in practice would be a constant line.

FCS (fluorescence correlation spectroscopy)

Request a detailed protocol

FCS measurements were obtained as previously described (Aydogan et al., 2019). For all measurements, the laser power was kept constant at 6.31 µW, and the heating unit of the microscope at 25 °C. All autocorrelation functions (ACFs) were fitted with the eight previously described models, and the best model was chosen based on the Bayesian information criterion. All fitting parameters and chosen diffusion models are stated in Table 3.

Table 3
Fitting parameters and chosen diffusion models for FCS experiments.

α , Anomalous subdiffusion parameter; AR, Structural parameter; ds, Diffusing species; bs, Blinking state of the fluorophore; ts, Triplet state of the fluorophore.

https://doi.org/10.7554/eLife.50130.051
ProteinFitting boundaries (ms)αARModel
mGFP4 × 10−4 - 1 × 1021.005one ds one ts
Spd-2-GFP4 × 10−4 - 1 × 1030.65one ds one ts
mNeonGreen (pPlk4)4 × 10−4 - 2 × 1020.70one ds one ts
mNeonGreen (pSas-6)4 × 10−4 - 2 × 1020.75one ds one ts
dNeonGreen4 × 10−4 - 4 × 1020.85one ds one bs one ts

Purified monomeric Spycatcher-GFP (kind gift from A. van der Merwe and colleagues) was measured in vitro in 1x PBS + 0.05% Tween20 at a similar concentration as Spd2-GFP in vivo. Embryos from mothers expressing GFP-tagged Spd2 were measured at the centrosomal plane and in nuclear cycles 11–14 at the beginning of S-phase.

The molecular brightness of the measured fluorophore (presented as photon count-rate per molecule (CPM)) was used to identify the oligomerization state of cytoplasmic, fluorescently-tagged Spd2 (in comparison to monomeric GFP). The CPM measurements were corrected for background fluorescence (with 10 control measurements of empty buffer for in vitro experiments, and ~20 recordings from WT embryos for in vivo experiments) using the following equation:

CPM= (Photon count rateSAMPLE- Photon count rateCONTROL)Number of particles in observation spot

In addition, control lines from mothers expressing monomeric or dimeric NeonGreen were measured to test the sensitivity of our CPM-based analysis. Identical to Spd2-GFP measurements, the recordings were taken within nuclear cycles 11–14 and at the beginning of S-phase. These flies also expressed Asl-mKate2 expressed from its endogenous promoter to identify the correct nuclear cycle stage and the centrosomal plane.

References

  1. 1
  2. 2
  3. 3
  4. 4
  5. 5
  6. 6
  7. 7
  8. 8
  9. 9
  10. 10
    PhosphoPep--a phosphoproteome resource for systems biology research in Drosophila Kc167 cells
    1. B Bodenmiller
    2. J Malmstrom
    3. B Gerrits
    4. D Campbell
    5. H Lam
    6. A Schmidt
    7. O Rinner
    8. LN Mueller
    9. PT Shannon
    10. PG Pedrioli
    11. C Panse
    12. HK Lee
    13. R Schlapbach
    14. R Aebersold
    (2007)
    Molecular Systems Biology, 3, 10.1038/msb4100182, 17940529.
  11. 11
  12. 12
  13. 13
  14. 14
  15. 15
  16. 16
  17. 17
  18. 18
  19. 19
  20. 20
  21. 21
  22. 22
  23. 23
  24. 24
  25. 25
  26. 26
  27. 27
    Studies of nuclear and cytoplasmic behaviour during the five mitotic cycles that precede gastrulation in Drosophila embryogenesis
    1. VE Foe
    2. BM Alberts
    (1983)
    Journal of Cell Science 61:31–70.
  28. 28
  29. 29
  30. 30
  31. 31
  32. 32
  33. 33
  34. 34
  35. 35
  36. 36
  37. 37
  38. 38
  39. 39
  40. 40
  41. 41
  42. 42
  43. 43
  44. 44
  45. 45
    Immunoisolation of centrosomes from Drosophila melanogaster
    1. V Lehmann
    2. H Müller
    3. BM Lange
    (2006)
    Current Protocols in Cell Biology, 3, 10.1002/0471143030.cb0317s29, 18228485.
  46. 46
  47. 47
  48. 48
  49. 49
  50. 50
  51. 51
  52. 52
  53. 53
  54. 54
  55. 55
  56. 56
  57. 57
  58. 58
  59. 59
  60. 60
  61. 61
  62. 62
    Drosophila: a Practical Approach
    1. EB Roberts
    (1998)
    Oxford University Press.
  63. 63
  64. 64
  65. 65
  66. 66
  67. 67
  68. 68
  69. 69
  70. 70
    Pericentriolar material structure and dynamics
    1. JB Woodruff
    2. O Wueseke
    3. AA Hyman
    (2014)
    Philosophical Transactions of the Royal Society B: Biological Sciences 369:20130459.
    https://doi.org/10.1098/rstb.2013.0459
  71. 71
  72. 72
  73. 73
  74. 74
  75. 75
  76. 76

Decision letter

  1. Yukiko M Yamashita
    Reviewing Editor; University of Michigan, United States
  2. Anna Akhmanova
    Senior Editor; Utrecht University, Netherlands

In the interests of transparency, eLife includes the editorial decision letter and accompanying author responses. A lightly edited version of the letter sent to the authors after peer review is shown, indicating the most substantive concerns; minor comments are not usually included.

[Editors’ note: a previous version of this study was rejected after peer review, but the authors submitted for reconsideration. The first decision letter after peer review is shown below.]

Thank you for submitting your work entitled "A positive feedback loop drives centrosome maturation in flies" for consideration by eLife. Your article has been reviewed by three peer reviewers, and the evaluation has been overseen by a Reviewing Editor and a Senior Editor. The reviewers have opted to remain anonymous.

Our decision has been reached after consultation between the reviewers. Based on these discussions and the individual reviews below, we regret to inform you that your work will not be considered further for publication in eLife.

As you can see in the individual reviewers' comments, all appreciated how well the study is done addressing a critical question in the field of centrosome biology. However, all remained concerned that the study did not provide a proof of Spd-2-Polo interaction being critical. They felt that proving this point is not necessarily straightforward in the scope of a simple revision (leaving the possibility open that those experiments disprove the model, or it might be not straightforward to design the experiments to prove this point).

With that said, all reviewers are enthusiastic about this study overall: although we cannot consider publication of your manuscript at this point per eLife's policy to invite revision only if the revision experiments are straightforward and the results are unlikely to change the major conclusions, we would be happy to reconsider your work as a new submission, if you're able to address the major points of reviewers.

Reviewer #1:

The manuscript "A positive feedback loop drives centrosome maturation in flies" by Alvarez-Rodrigo et al. investigates the regulation of centrosome maturation when cells prepare for mitosis, by focusing on the expansion of the pericentriolar material (PCM). In particular, the authors analyze the involvement of Spd-2 and Polo in this process and study potential phosphorylation sites in Spd-2 that may help recruit Polo and that seem to be important for pericentriolar scaffold assembly and expansion, including the recruitment of other essential PCM proteins such as γ-tubulin to the expanded scaffold. They propose that Spd-2 is an essential component of a feedback mechanism that involves recruitment of Polo to drive incorporation and phosphorylation of Cnn near centrioles and subsequent outward movement of Cnn-Spd-2, thereby contributing to assembly of an expanded PCM. This process is driven by continuous Spd-2-Polo-dependent incorporation of additional Cnn near centrioles and its outward flux.

Overall this is a very nice story that puts together various previous and new observations into an appealing model to explain the mechanism by which centrosomes mature, at least in flies.

The paper is very well written and the data are presented clearly.

However, I found one major issue with this work that needs to be addressed. In my opinion the authors do not provide convincing evidence for a core element of their model: the role of Spd-2 in Polo recruitment.

Specific points:

1) The authors state "Based on our observations below, we refer to these mutant proteins as being unable to recruit Polo." However, the recruitment of Polo to Spd-2 is never demonstrated in the first place. I understand that the interaction is supposed to happen specifically at the centrosome, so immunoprecipitation from soluble extract is most likely not an option, but can the author provide any evidence for a direct interaction of Polo with Spd-2?

2) The observations that follow the authors' statement do not demonstrate the inability of the mutants to recruit Polo: it is shown that the mutants partially rescue embryos lacking Spd-2, that mutants are less concentrated at centrosomes, that these centrosomes have less MTs, and that mutants and Polo localize around centrioles but do not assemble into an expanded pericentriolar scaffold. This latter finding is the only piece of data in support of the proposed role of Spd-2 in Polo binding. However, as the authors also state, in these experiments the scaffold may simply not assemble. In this case the reduced Polo signal could simply be the result of the absence of an expanded scaffold and not necessarily due to lack of binding to Spd-2.

3) The demonstration that Spd-2 is phosphorylated at centrosomes is very convincing (Figure 1). However, it was not tested whether this phosphorylation is at least partially Polo-dependent. This could indicate interaction with Polo at centrosomes.

4) Figure 2: Are the mutants still phosphorylated at centrosomes? This should be revealed by an analysis as in Figure 1.

Reviewer #2:

In the manuscript by Alvarez-Rodrigo et al. entitled "A positive feedback loop drives centrosome maturation in flies" the authors aim to show that the interactions between Spd2-Polo-Cnn drive an autocatalytic expansion of the PCM to ensure synchronized growth of sister centrosomes. Based on a minimal Polo box binding consensus sequence, the authors mutated Spd2 to prevent Polo binding and then characterized the effects on fly viability, Spd2 localization, and PCM expansion. They found that increasing the mutational burden on Spd2 reduces PCM expansion and localization of centrosome components during centrosome maturation. If the authors were able to show the mechanism in Figure 10, this would constitute a major contribution to the centrosome field. Unfortunately, the lack of justification or biological relevance behind the authors' approach prevent them from achieving this goal. Therefore, I do not believe the manuscript should be published without extensive changes to the approach or substantial work to justify and validate their approach.

The authors perform initial MS/MS to identify the phosphorylated residues on Spd2 which may be a prerequisite for Polo binding. They then mutated these residues and found mild Spd2 defects that did not recapitulate the phenotypes found in the CONS or ALL mutants. From these data, it appears that mutating the biologically relevant sites was not sufficient to produce a phenotype, so the authors moved on to mutating a substantial number of serine residues in Spd2 only because it appears in a minimal consensus sequence. This approach led to studying the effects of a massively mutated protein without regard to the biology of the system.

Furthermore, the authors do not provide evidence that the Spd2 mutants result in an inability of Polo to bind Spd2. Perhaps Polo can bind Spd2 through non-canonical sites? That may be the case because the authors present evidence that Polo is still recruited to the centriole and functions properly, as Cnn is phosphorylated and centrioles are still able to disengage properly. Are the Spd2 mutants still being phosphorylated, as in Figure 1? This can be answered with their established method in Figure 1 and by mass spectrometry. Is mutant Spd2 still able to homodimerize and interact with Cnn? The lab has previously established a yeast 2-hybrid assay and has used SEC-MALS, both of which could be used to test this. Spd2 also interacts with Asterless and PLP. Do the Spd CONS or ALL mutants disrupt these interactions, which may explain the mutant phenotypes? Further biochemical analysis of mutant Spd2 should be performed to show that the mutants do not cause protein misfolding.

Although their discussion and conclusions are intriguing and could provide insight into how centrosome maturation may be regulated, I do not feel they have provided sufficient evidence to support these conclusions. Additionally, while there is a striking phenotype in the manuscript, the authors have only shown that this is caused by mutating Spd2. They have not shown that this is due to disrupting the interaction with Polo, let alone disrupting a potential positive feedback loop, as implied by the title of their manuscript, which is misleading.

Additional concerns:

1) All experiments are performed in Drosophila embryos, whereas the authors claim that the mechanism is universal to the entire fly "in vivo". This should be toned down.

2) The manuscript would benefit from more details of the mitotic defects in Figure 2A.

3) Are Spd2ALL, Spd2-/- heterozygotes viable or can reduced dosage explain the phenotypes in Figure 5—figure supplement 1?

4) The images of centrosomes in Figures 4-6 appear to be S-phase centrosomes with flares, not mitotic centrosomes. The authors reference mitosis and maturation when, in fact, they show images of interphase (S-phase) centrosomes. The authors should show low magnification images of mitotic spindles and then high magnification insets of centrosomes to demonstrate that they are, in fact, examining mitotic centrosomes.

Reviewer #3:

In this manuscript, the authors show that Spd-2 is partially phosphorylated at centrosomes, allowing the recruitment of Polo to the centrosomes. Disabling the recruitment leads to outward pericentriolar scaffold expansion failure, and prevents centrosome maturation. The authors also demonstrate that Polo, Spd-2 and Cnn together can drives mitotic centrosome expansion by forming a positive feed-back loop. This is an interesting paper with data of high quality. I note the following issue the authors need to consider:

1) The authors should provide direct evidence about how much Spd-2-CONS and Spd-2-ALL are affects their binding with the Polo PBD compared to WT, which is critical for most of their conclusions. I'm concerned that mutating all 34 Serine sites may affect protein structure and this needs to be ruled out.

2) Do the authors now if the phosphorylation status is altered during the cell cycle?

3) Figure 4: authors conclude that the inability of mutant proteins to recruit Polo reduces the assembly and/or maintenance of this mixed scaffold, but the data shown can't exclude the possibility that mutant proteins are efficiently incorporated into the scaffold. The author should show a comparison of endogenous Spd-2 scaffold in all three conditions.

4) Figure 5: I'm not convinced by the authors' conclusion that the pericentriolar scaffold cannot expand because Spd-2-CON can't recruit Polo. First, they co-localize at the mother centriole, and second, Spd-2-CON itself is affecting scaffold formation in Spd-2 mutant background (Figure 4), so how can they be certain it's a Polo recruitment issue? I would potentially suggest a FRAP experiment to showcase the recruitment of Polo to the scaffold. In addition, apart from co-expressing with Polo-GFP, what's the difference between Figure 5B and Figure 4A (middle lane)? Why is there's a huge difference between the ratios (64/36 vs 9/91)? This was confusing to me.

5) Figure 7: what is the component of the scaffold? If the scaffold is made of Spd-2, when Spd-2 CONS is unable to form the scaffold, you won't be able to see other proteins there no matter how they are recruited. Is there any other marker for the scaffold?

6) Figure 9: The C-terminus of Spd-2 contains most of the conserved Serine sites, but the authors show that mutation N-terminal sites have the biggest effect? What do the authors think the molecular basis for this is?

[Editors’ note: what now follows is the decision letter after the authors submitted for further consideration.]

Thank you for submitting your article "Evidence that a positive feedback loop drives centrosome maturation in fly embryos" for consideration by eLife. Your article has been reviewed by three peer reviewers, and the evaluation has been overseen by a Reviewing Editor and Anna Akhmanova as the Senior Editor. The reviewers have opted to remain anonymous.

The reviewers have discussed the reviews with one another and the Reviewing Editor has drafted this decision to help you prepare a revised submission.

Summary:

In the manuscript by Alvarez-Rodrigo et al., the authors show that the interactions between Spd2-Polo-Cnn drive an autocatalytic expansion of the PCM to ensure synchronized growth of sister centrosomes.

This is a resubmission of previously rejected manuscript, where all the reviewers appreciated the rigor of the work and an elegant model to explain the process of centrosome maturation but found that the direct evidence to support the importance of Spd-2-Polo interaction was lacking.

This revised and newly submitted version was reviewed by original reviewers, and now they all agree that this work now has strong evidence that supports the conclusions. Some reviewers raised a few minor comments, which the authors may want to address prior to formal acceptance.

Overall this is a very nice story that puts together various previous and new observations into an appealing model to explain the mechanism by which centrosomes mature, which may be conserved in other species as well.

Reviewer #1:

In the manuscript by Alvarez-Rodrigo et al. entitled "A positive feedback loop drives centrosome maturation in flies" the authors show that the interactions between Spd2-Polo-Cnn drive an autocatalytic expansion of the PCM to ensure synchronized growth of sister centrosomes. I feel that the authors have addressed my original concerns with one remaining issue. Regarding the immunoprecipitation (IP) experiments between the MBP-phospho-Spd2 fragment and GST-PBD (new Figure 1C), the resolution of these immunoblots appear low and the brightness/contrast altered such that it is difficult to gauge the load between the samples. It appears that there is more GST-PBD in the MBP-Spd-2 WT immunoprecipitate, but how much more? Replacing these with higher resolution blots with accompanying quantitation would be appropriate. This is an important experiment/result because it demonstrates that the authors have in fact generated a Spd-2 Polo-binding mutant. Also, is the kinase being used human Plk1, whereas the Spd-2 and PBD proteins are fly? Whatever the species, the fact that the authors used proteins from different organisms should be stated in the Results section.

I also recommend a couple of text changes in this section. The authors state that binding of the 19T mutant was reduced to background levels. How do they know this is background? This could be real binding and, therefore, would remain at WT levels, not a reduced level. The authors end this section stating, "Thus, a fragment of Spd-2 can bind directly to the PBD when phosphorylated, and this binding is prevented when the S-S/T motifs are mutated to T-S/T". Based on the data, it looks to me that binding is not prevented. Instead, they should say that phosphorylation of Spd-2 increases PDB binding which, in my opinion, is what the data shows. The data also suggest that Polo activity primes it's binding to Spd-2 (as they state in the Discussion) but this important finding should also be emphasized here at the end of this Results section.

Reviewer #2:

In their revised manuscript the authors have addressed my concerns by providing new experiments and additional discussion/explanation.

My main concerns were the lack of evidence for interaction between Spd-2 and Polo and the conundrum that the reduced Polo signal at centrosomes containing Spd-2 mutants may be due to specific loss of interaction with Polo or simply due to the lack of any scaffold where Polo could be recruited.

The first concern was dealt with by providing in vitro biochemical interaction data (Figure 1C) and the second by performing a new type of quantification of the centrosomal Polo signal in cells expressing a mix of WT and mutant Spd-2, at a time point when Polo signal is just starting to become reduced (Figure 10). This approach allowed the authors to demonstrate reduced Polo signal even in cells where a Spd-2 PCM scaffold is still present. Although the data supports the authors' conclusion that Spd-2 mutants cause a specific loss of PCM associated Polo, the quantification would be more convincing if actual intensities would have been quantified for the Spd-2 scaffold and Polo signals rather than subjective classification into "weak" and "strong". This is because in contrast to a similar quantification in Figure 6, where an "all or nothing" effect is observed, here the relative intensities of PCM scaffold-associated Spd-2 vs Polo are crucial for the result. I would suggest to also adding quantification of intensities to the figure.

My other concerns were also addressed, not in all cases with experiments, but I recognize the technical challenges that have led to this decision.

I would now support publication.

Reviewer #3:

The revised version of the paper is much improved and the authors have addressed most of the issues raised in my initial review to the best of their abilities. The manuscript in my opinion is now suitable for publication in eLife.

https://doi.org/10.7554/eLife.50130.057

Author response

[Editors’ note: the author responses to the first round of peer review follow.]

Reviewer #1:

[…] Overall this is a very nice story that puts together various previous and new observations into an appealing model to explain the mechanism by which centrosomes mature, at least in flies.

The paper is very well written and the data are presented clearly.

However, I found one major issue with this work that needs to be addressed. In my opinion the authors do not provide convincing evidence for a core element of their model: the role of Spd-2 in Polo recruitment.

Specific points:

1) The authors state "Based on our observations below, we refer to these mutant proteins as being unable to recruit Polo." However, the recruitment of Polo to Spd-2 is never demonstrated in the first place. I understand that the interaction is supposed to happen specifically at the centrosome, so immunoprecipitation from soluble extract is most likely not an option, but can the author provide any evidence for a direct interaction of Polo with Spd-2?

To address this question, we have now purified an MBP-fusion containing a fragment of Spd-2 (aa352-758; the only recombinant fragment of Spd-2 that is relatively stable in our hands). This fragment contains 19 S-S/T residues and we show that it can bind to recombinant GST-PBD, but only when it has first been phosphorylated by recombinant Plk1 (New Figure 1C). Mutating the 19 S-S/T residues in this Spd-2 fragment to T-S/T abolishes PBD-binding. These findings provide strong support for our hypothesis that Drosophila Spd-2 can directly bind to the Polo-PBD when phosphorylated, and that this binding is prevented when the S-S/T sites are mutated to T-S/T.

2) The observations that follow the authors' statement do not demonstrate the inability of the mutants to recruit Polo: it is shown that the mutants partially rescue embryos lacking Spd-2, that mutants are less concentrated at centrosomes, that these centrosomes have less MTs, and that mutants and Polo localize around centrioles but do not assemble into an expanded pericentriolar scaffold. This latter finding is the only piece of data in support of the proposed role of Spd-2 in Polo binding. However, as the authors also state, in these experiments the scaffold may simply not assemble. In this case the reduced Polo signal could simply be the result of the absence of an expanded scaffold and not necessarily due to lack of binding to Spd-2.

To address this problem we have now injected mRNA encoding either WT or mutant forms of Spd-2-mKate2 into embryos expressing Polo-GFP. The mRNA is gradually translated, and so the Spd-2-mKate2 fusions gradually overwhelm the endogenous untagged Spd-2. We then looked for embryos injected with the mutant mRNA where the loss of Polo from the centrosome was first becoming apparent, and compared the centrosomal recruitment of the mutant Spd-2-mKate2 and Polo-GFP to similarly staged controls (injected with WT Spd-2-mKate2 mRNA). This analysis, scored blind, revealed that mutant Spd-2-mKate2 could still be detected in an expanded PCM even when the amount of Polo recruited to the scaffold was severely reduced (New Figure 10). This is presumably because there is enough WT Spd-2 present in these embryos to form a scaffold, but the mutant Spd-2 protein present in the scaffold cannot recruit Polo. This strongly supports our conclusion that the mutant Spd-2 proteins cannot recruit Polo to the expanded PCM. Nevertheless, we agree that we cannot formally exclude the possibility that the mutant Spd-2 proteins can still bind Polo in some way, and we have toned down our statements on this point throughout the manuscript.

3) The demonstration that Spd-2 is phosphorylated at centrosomes is very convincing (Figure 1). However, it was not tested whether this phosphorylation is at least partially Polo-dependent. This could indicate interaction with Polo at centrosomes.

This is a good question, but one that is difficult to address: a lack of Spd-2 phosphorylation after Polo inhibition might be a direct consequence of Polo failing to phosphorylate Spd-2, or an indirect consequence of there being no expanded PCM (as Spd-2 is mostly phosphorylated in the PCM). We do now provide evidence that Polo can phosphorylate a fragment of Spd-2 in vitro to create PBD binding sites (New Figure 1C), consistent with the possibility that at least some of the phosphorylation of Spd-2 at centrosomes is Polo-dependent.

4) Figure 2: Are the mutants still phosphorylated at centrosomes? This should be revealed by an analysis as in Figure 1.

This is a good question, but we have not attempted to answer it as it would be a lot of work to collect enough mutant material to do the biochemistry, and we think the answer may not be very informative: the S-S/T to T-S/T substitutions may or may not interfere with the phosphorylation of these motifs, but will perturb PBD-binding regardless.

Reviewer #2:

[…] If the authors were able to show the mechanism in Figure 10, this would constitute a major contribution to the centrosome field. Unfortunately, the lack of justification or biological relevance behind the authors' approach prevent them from achieving this goal. Therefore, I do not believe the manuscript should be published without extensive changes to the approach or substantial work to justify and validate their approach.

The authors perform initial MS/MS to identify the phosphorylated residues on Spd2 which may be a prerequisite for Polo binding. They then mutated these residues and found mild Spd2 defects that did not recapitulate the phenotypes found in the CONS or ALL mutants. From these data, it appears that mutating the biologically relevant sites was not sufficient to produce a phenotype, so the authors moved on to mutating a substantial number of serine residues in Spd2 only because it appears in a minimal consensus sequence. This approach led to studying the effects of a massively mutated protein without regard to the biology of the system.

This statement is based on the premise that we initially identified all the biologically relevant phosphorylation sites in our MS screen but that, when mutating these sites gave only a mild phenotype, we simply mutated many more sites (based only on a minimal S-S/T consensus) to generate a stronger phenotype. We disagree that these experiments were performed “without regard to the biology”. First, it is widely accepted that MS approaches may not identify all of the relevant phosphorylation sites in a protein, and this is demonstrably the case for Spd-2, where 4 independent MS screens (3 from embryos) have identified 41 phosphorylation sites in fly Spd-2, some of which were identified in multiple screens, but many of which were not (as summarised only for the sites that are potential PBD binding motifs in Table 2). Second, we considered a large body of evidence when devising this strategy: (1) The S-S/T(P) motif is widely accepted as a PBD-binding motif and the PBD is widely considered to be essential for targeting Polo/Plk1 to centrosomes1-7; (2) All of the previously identified sites in worm, human and frog SPD-2/Cep192 that recruit Plk1 to centrosomes conform to this consensus8-11; (3) The S-S/T to T-S/T substitution is widely accepted to prevent PBD binding5,6; (4) Ser-to-Thr substitutions are considered to be very conservative in nature, and so unlikely to lead to protein misfolding.

Furthermore, the authors do not provide evidence that the Spd2 mutants result in an inability of Polo to bind Spd2. Perhaps Polo can bind Spd2 through non-canonical sites? That may be the case because the authors present evidence that Polo is still recruited to the centriole and functions properly, as Cnn is phosphorylated and centrioles are still able to disengage properly. Are the Spd2 mutants still being phosphorylated, as in Figure 1? This can be answered with their established method in Figure 1 and by mass spectrometry. Is mutant Spd2 still able to homodimerize and interact with Cnn? The lab has previously established a yeast 2-hybrid assay and has used SEC-MALS, both of which could be used to test this. Spd2 also interacts with Asterless and PLP. Do the Spd CONS or ALL mutants disrupt these interactions, which may explain the mutant phenotypes? Further biochemical analysis of mutant Spd2 should be performed to show that the mutants do not cause protein misfolding.

In the third paragraph the reviewer raises several points. First, they ask for evidence that the Spd-2 mutants cannot bind Polo. As discussed in point 1 of our response to reviewer #1, we now provide compelling evidence that Spd-2 interacts directly with the PBD when phosphorylated, and this interaction is abolished when the putative PBD-binding motifs are mutated (New Figure 1C). Second, the reviewer highlights that the Spd-2 mutant proteins still colocalise with Polo at centrioles, suggesting that the mutant proteins are still recruiting Polo to centrioles. We now clarify that we believe Spd-2 is only required to recruit Polo to the PCM, not to the centrioles. There is strong evidence that Polo can be recruited to centrioles independently of Spd-2 by phosphorylated S-S/T(P) motifs in centriole proteins such as Sas-412. Third, the reviewer asks whether the Spd-2 mutant proteins are still phosphorylated. Please see point 4 in our response to reviewer #1.

We have taken two approaches to address this point. First, we tested whether Spd-2 is a homodimer in fly embryos. This is crucial as, if so, a largely misfolded mutant protein might still localise to centrioles and centrosomes if it can still homodimerize with the WT protein. Using Fluorescence Correlation Spectroscopy (FCS), we show that cytoplasmic Spd-2 is monomeric (New Figure 5—figure supplement 1)—as was also recently shown for SPD-2 in worm embryos13. Second, with this information in hand, we examined how WT and mutant Spd-2-GFP fusion-proteins are recruited to centrioles in the presence of WT Spd-2-mCherry (which serves as a marker for the WT scaffold and also allows a scaffold to assemble even in the presence of the mutant proteins). The recruitment of the WT and mutant Spd-2-GFP fusions to the WT Spd-2-mCherry scaffold (scored blind) was essentially indistinguishable (New Figure 5). Thus, the mutant Spd-2 proteins can interact with all the proteins that recruit and maintain Spd-2 at centrioles and centrosomes, and this is unlikely to be because they are simply homodimerizing with the WT protein. Finally, we more clearly explain the significance of the data showing that the mutant Spd-2 proteins can rescue the defect in pronuclear fusion in embryos lacking Spd-2 (Figure 2C), but that these embryos still die of mitotic defects during early embryo development. This suggests that the mutant proteins allow centrioles to assemble sufficient PCM to promote the relatively slow process of pronuclear fusion, but not enough PCM to support the very rapid rounds of mitosis that follow. Taken together, these data suggest that the mutant proteins are not misfolded.

Additional concerns:

1) All experiments are performed in Drosophila embryos, whereas the authors claim that the mechanism is universal to the entire fly "in vivo". This should be toned down.

The reviewer points out that all our results are from fly embryos and may not apply to other cell types. We now make this clear throughout the manuscript.

2) The manuscript would benefit from more details of the mitotic defects in Figure 2A.

As requested, we now comment in more detail about the nature of the mitotic defects shown in Figure 2A (now Figure 3A).

3) Are Spd2ALL, Spd2-/- heterozygotes viable or can reduced dosage explain the phenotypes in Figure 5—figure supplement 1?

The reviewer asks whether Spd2-ALL; Spd-2 -/- heterozygotes are viable, or whether reduced dosage can explain the phenotypes shown in Figure 5—figure supplement 1 (now Figure 6—figure supplement 1). These heterozygous flies are indeed viable, but this is to be expected as Spd-2-/- mutants are also viable (but female sterile). We suspect the reviewer is suggesting that because there is only one copy of Spd-2-ALL in these flies, the phenotypes we observe may be due to lowering the genetic dosage of Spd-2 rather than to the mutations we introduce into the genes. We think this is unlikely for two reasons: (1) In all the experiments where we assess Spd-2-ALL- or Spd-2-CONS-fusion protein function the control is always the WT Spd-2-fusion tested at the same gene dosage; (2) The Spd-2-11A mutant protein is expressed at much lower levels than the WT protein (Figure 1—figure supplement 1A), yet it localises to centrosomes nearly as strongly as the WT protein (Figure 1—figure supplement 1B-D). Thus, the reduction in the centrosomal levels of Spd-2-ALL and Spd-2-CONS is unlikely to simply be due a general reduction in their protein levels. We now discuss this point.

4) The images of centrosomes in Figures 4-6 appear to be S-phase centrosomes with flares, not mitotic centrosomes. The authors reference mitosis and maturation when, in fact, they show images of interphase (S-phase) centrosomes. The authors should show low magnification images of mitotic spindles and then high magnification insets of centrosomes to demonstrate that they are, in fact, examining mitotic centrosomes.

The reviewer questions whether we are examining centrosomes in interphase or mitosis. We apologise for not explaining this properly. In these rapidly developing embryos the nuclei progress through successive S- and M-phases without intervening Gap-phases. As a result, the centrosomes are essentially always either in mitosis, or are preparing to enter mitosis—they are never truly in an “interphase” state where they organise only the tiny amounts of PCM one might observe in a fly somatic cell in interphase. Thus, as soon as mitosis finishes the embryos enter S-phase and the centrosomes start to mature in preparation for the next round of mitosis. We now clarify this important point.

Reviewer #3:

[…] 1) The authors should provide direct evidence about how much Spd-2-CONS and Spd-2-ALL are affects their binding with the Polo PBD compared to WT, which is critical for most of their conclusions. I'm concerned that mutating all 34 Serine sites may affect protein structure and this needs to be ruled out.

Please see point 1 of our response to reviewer #1: we now show that a recombinant WT Spd-2 MBP-fusion protein can bind to the PBD when phosphorylated by Plk1, but mutation of the S-S/T motifs to T-S/T prevents binding (New Figure 1C). The reviewer was also concerned that the multiple Ser-The substitutions we introduce in the mutant proteins may affect protein structure. Please see point 3 in our response to reviewer #2: we now present several lines of evidence that indicate that the mutant proteins are not simply misfolded.

2) Do the authors now if the phosphorylation status is altered during the cell cycle?

This is a good question that we have tried to address by raising phospho-specific antibodies to Spd-2—but so far without success.

3) Figure 4: authors conclude that the inability of mutant proteins to recruit Polo reduces the assembly and/or maintenance of this mixed scaffold, but the data shown can't exclude the possibility that mutant proteins are efficiently incorporated into the scaffold. The author should show a comparison of endogenous Spd-2 scaffold in all three conditions.

We apologise that we were unclear on this, as we do indeed believe that the mutant Spd-2 proteins can incorporate into the scaffold formed by the WT protein. This is entirely consistent with our model: as long as there is some WT Spd-2 in the scaffold, it should recruit Polo, which can then phosphorylate Cnn to support the expanding Spd-2 scaffold; thus, there is no requirement that every Spd-2 molecule in the scaffold be able to recruit Polo. As requested by the reviewer, we now directly address this issue by expressing either the WT or mutant Spd-2-GFP fusions in the presence of WT Spd-2-mCherry (but in the absence of any endogenous unlabelled Spd-2). This allows us to directly compare the ability of the WT and mutant GFP-fusions to incorporate into the expanding scaffold formed by the WT Spd-2-mCherry (New Figure 5). The incorporation of the WT and mutant Spd-2-GFP fusion proteins into the WT Spd-2-mCherry scaffold is essentially indistinguishable—consistent with our model.

At a first glance, this new data may appear at odds with the data shown in Figure 4B of our original submission (now superseded by New Figure 5). In this earlier experiment we showed that in the presence of WT unlabelled Spd-2, the scaffold formed by the mutant-GFP fusion proteins appears somewhat weaker than that formed by the WT-GFP fusion protein—suggesting that the presence of the mutant protein may partially inhibit the assembly of the WT scaffold (which is also consistent with our model, as presumably less Polo will be recruited to the PCM in the presence of the mutant proteins). These two experiments, however, are not comparable: in the old experiment, the scaffold is formed from 2 copies of the endogenous unlabelled Spd-2 gene and two copies of the Spd-2-GFP fusion (Old Figure 4B) while in the new experiment it is formed from one copy of WT-Spd-2-mCherry and one copy of the Spd-2-GFP fusions in the absence of any endogenous (unlabelled) protein (new Figure 5). Thus, there are several possible explanations for this minor discrepancy (for example, unlabelled Spd-2 might more efficiently compete with mutant Spd-2-GFP for incorporation into the scaffold than WT-Spd-2-mCherry, so the ratio of WT/mutant proteins in the scaffolds may be very different).

4) Figure 5: I'm not convinced by the authors' conclusion that the pericentriolar scaffold cannot expand because Spd-2-CON can't recruit Polo. First, they co-localize at the mother centriole, and second, Spd-2-CON itself is affecting scaffold formation in Spd-2 mutant background (Figure 4), so how can they be certain it's a Polo recruitment issue? I would potentially suggest a FRAP experiment to showcase the recruitment of Polo to the scaffold. In addition, apart from co-expressing with Polo-GFP, what's the difference between Figure 5B and Figure 4A (middle lane)? Why is there's a huge difference between the ratios (64/36 vs 9/91)? This was confusing to me.

The reviewer raises two specific reasons for this scepticism. First, the mutant proteins still co-localise with Polo at the centriole (suggesting that they can still recruit Polo). Please see point 2 in our response to reviewer #2where we discuss why we think that Spd-2 is not required to recruit Polo to centrioles. Second, although the Spd-2 mutants effect scaffold assembly we cannot be certain that this is due to a Polo recruitment defect. Please see point 2 of our response to reviewer #1where we describe new experiments that address this problem.

The reviewer suggests that we use FRAP to measure the dynamics of Polo recruitment to the scaffold. This is a good experiment, but we believe it would be hard to interpret. This is because Polo may have different turnover rates at the centriole and in the PCM and the ratio of Polo in these two structures is very different in embryos expressing WT Spd-2-mCherry (where it is mostly in the PCM) and mutant Spd-2-mCherry (where it is mostly in the centriole).

The reviewer asks why the PCM recruitment of the mutant Spd-2-CONS protein is so much worse in embryos expressing Polo-GFP (New Figure 6B) than in embryos not expressing Polo-GFP (New Figure 4). We apologise for not explaining this more clearly. The Polo-GFP line we use here is the healthiest available (an exon-trap insertion into the endogenous Polo gene), but the GFP tag partially disrupts Polo function and this line is only viable in the presence of an untagged copy of Polo. As we now explain more fully, there appears to be a strong genetic interaction between mutant Spd-2 proteins and the partially functional Polo-GFP. Embryos lacking endogenous Spd-2 and expressing Polo-GFP are viable in the presence of WT Spd-2-mCherry, but die very early in the presence of Spd-2-ALL-mCherry or Spd-2-CONS-mCherry (and much earlier than embryos expressing just Spd-2-CONS-mCherry or Spd-2-ALL-mCherry without the Polo-GFP). We do not understand the basis for this genetic interaction, but it suggests an intimate link between Spd-2 and Polo that is somehow perturbed when Spd-2 is mutated in this way and Polo is GFP-tagged.

5) Figure 7: what is the component of the scaffold? If the scaffold is made of Spd-2, when Spd-2 CONS is unable to form the scaffold, you won't be able to see other proteins there no matter how they are recruited. Is there any other marker for the scaffold?

This is correct. However, although we have long proposed that Spd-2, Polo and Cnn cooperate to form the PCM scaffold in flies, we feel that this is not yet an established fact. Hence, we sought to test whether other PCM components might still be recruited to an expanded PCM that might form independently of Spd-2 and Polo. Our data suggest that this is not the case, supporting the view that Spd-2, Polo and Cnn are the major components of the PCM scaffold. The reviewer asks if there is any other marker of the scaffold we can test, but these are the only three markers that we are aware of.

6) Figure 9: The C-terminus of Spd-2 contains most of the conserved Serine sites, but the authors show that mutation N-terminal sites have the biggest effect? What do the authors think the molecular basis for this is?

We apologise for being unclear on this point. The N- and C-terminal halves of Spd-2 that we test here each have 17 S-S/T motifs (see Figure 2A) and we mutated all 17 of these (not just the conserved ones) in the experiments comparing the effect of the mutations in each half of the protein (Figure 9). We have now clarified this point. We do not know why the N-terminal mutations have a larger effect, but suspect that the N-terminal half of Spd-2 normally plays the greater part in recruiting Polo to the PCM in vivo.

[Editors' note: the author responses to the re-review follow.]

Reviewer #1:

[…] Regarding the immunoprecipitation (IP) experiments between the MBP-phospho-Spd2 fragment and GST-PBD (new Figure 1C), the resolution of these immunoblots appear low and the brightness/contrast altered such that it is difficult to gauge the load between the samples. It appears that there is more GST-PBD in the MBP-Spd-2 WT immunoprecipitate, but how much more? Replacing these with higher resolution blots with accompanying quantitation would be appropriate. This is an important experiment/result because it demonstrates that the authors have in fact generated a Spd-2 Polo-binding mutant. Also, is the kinase being used human Plk1, whereas the Spd-2 and PBD proteins are fly? Whatever the species, the fact that the authors used proteins from different organisms should be stated in the Results section.

I also recommend a couple of text changes in this section. The authors state that binding of the 19T mutant was reduced to background levels. How do they know this is background? This could be real binding and, therefore, would remain at WT levels, not a reduced level. The authors end this section stating, "Thus, a fragment of Spd-2 can bind directly to the PBD when phosphorylated, and this binding is prevented when the S-S/T motifs are mutated to T-S/T". Based on the data, it looks to me that binding is not prevented. Instead, they should say that phosphorylation of Spd-2 increases PDB binding which, in my opinion, is what the data shows. The data also suggest that Polo activity primes it's binding to Spd-2 (as they state in the Discussion) but this important finding should also be emphasized here at the end of this Results section.

The reviewer made a number of points about our presentation and description of the immunoprecipitation experiment showing that phosphorylated Spd-2 fragment can bind recombinant GST-PBD in vitro (Figure 1C). These points were all valid and we have made the following changes:

1) We now show the quantification from three repeats of the experiment and state in the Results section that we use recombinant human Plk1 and a fragment of Drosophila Spd-2 in these experiments.

2) We agree that our original description of this new experiment was unclear in several respects. We have now rewritten this section to focus on the conclusion that the MBP-Spd-2 fragment binds more GST-PBD when phosphorylated by Plk1, and that this increase in binding is not seen when the S-S/T sites have been mutated to T-S/T sites.

3) We now emphasise the important finding that Polo activity may prime its own binding to Spd-2 in both the Results and Discussion sections.

Reviewer #2:

[…] The first concern was dealt with by providing in vitro biochemical interaction data (Figure 1C) and the second by performing a new type of quantification of the centrosomal Polo signal in cells expressing a mix of WT and mutant Spd-2, at a time point when Polo signal is just starting to become reduced (Figure 10). This approach allowed the authors to demonstrate reduced Polo signal even in cells where a Spd-2 PCM scaffold is still present. Although the data supports the authors' conclusion that Spd-2 mutants cause a specific loss of PCM associated Polo, the quantification would be more convincing if actual intensities would have been quantified for the Spd-2 scaffold and Polo signals rather than subjective classification into "weak" and "strong". This is because in contrast to a similar quantification in Figure 6, where an "all or nothing" effect is observed, here the relative intensities of PCM scaffold-associated Spd-2 vs Polo are crucial for the result. I would suggest to also adding quantification of intensities to the figure.

My other concerns were also addressed, not in all cases with experiments, but I recognize the technical challenges that have led to this decision.

I would now support publication.

The reviewer requested that we provide some additional quantification of fluorescent intensities for the experiment shown in Figure 10, rather than the more qualitative classification of images into “weak” or “strong” PCM signal. The reason we do not do this (for this experiment and for several others where we use 3D-SIM to assess the localisation of proteins at both the centriole and in the PCM) is that these 3D-SIM images should not really be used for intensity quantification as they are computer-based reconstructions of what the computer “thinks” the original image looks like. Our 3D-SIM experts in the Micron Oxford Advanced Bioimaging Facility tell us that this makes it unreliable to compare absolute intensity levels from such images, particularly if they have been collected from different samples. This is why in all figures like this we show several representative images to cover the range of phenotypes we observe and quantify in our classification of the reconstructed images. As all scoring is performed blind, we are confident that this “semi-quantitative” scoring system provides a reproducible and accurate reflection of the underlying data.

In addition we have made a few minor textual clarifications, corrected a few typos, and inserted an extra reference to a very recently published paper by Cabral et al., 2019, that is relevant to this work.

References:

1. Lee, K. S., Grenfell, T. Z., Yarm, F. R. & Erikson, R. L. Mutation of the polo-box disrupts localization and mitotic functions of the mammalian polo kinase Plk. Proceedings of the National Academy of Sciences 95, 9301–9306 (1998).

2. Song, S., Grenfell, T. Z., Garfield, S., Erikson, R. L. & Lee, K. S. Essential function of the polo box of Cdc5 in subcellular localization and induction of cytokinetic structures. Mol Cell Biol 20, 286–298 (2000).

3. Seong, Y.-S. et al. A spindle checkpoint arrest and a cytokinesis failure by the dominant-negative polo-box domain of Plk1 in U-2 OS cells. J Biol Chem 277, 32282–32293 (2002).

4. Reynolds, N. & Ohkura, H. Polo boxes form a single functional domain that mediates interactions with multiple proteins in fission yeast polo kinase. J Cell Sci 116, 1377–1387 (2003).

5. Elia, A. E. H., Cantley, L. C. & Yaffe, M. B. Proteomic screen finds pSer/pThr-binding domain localizing Plk1 to mitotic substrates. Science 299, 1228–1231 (2003).

6. Elia, A. E. H. et al. The molecular basis for phosphodependent substrate targeting and regulation of Plks by the Polo-box domain. Cell 115, 83–95 (2003).

7. Liu, J., Lewellyn, A. L., Chen, L. G. & Maller, J. L. The polo box is required for multiple functions of Plx1 in mitosis. J Biol Chem 279, 21367–21373 (2004).

8. Decker, M. et al. Limiting amounts of centrosome material set centrosome size in C. elegans embryos. Curr. Biol 21, 1259–1267 (2011).

9. Joukov, V., De Nicolo, A., Rodriguez, A., Walter, J. C. & Livingston, D. M. Centrosomal protein of 192 kDa (Cep192) promotes centrosome-driven spindle assembly by engaging in organelle-specific Aurora A activation. Proc Natl Acad Sci USA 107, 21022–21027 (2010).

10. Joukov, V., Walter, J. C. & De Nicolo, A. The Cep192-organized aurora A-Plk1 cascade is essential for centrosome cycle and bipolar spindle assembly. Mol Cell 55, 578–591 (2014).

11. Meng, L. et al. Bimodal Interaction of Mammalian Polo-Like Kinase 1 and a Centrosomal Scaffold, Cep192, in the Regulation of Bipolar Spindle Formation. Mol Cell Biol 35, 2626–2640 (2015).

12. Novak, Z. A., Wainman, A., Gartenmann, L. & Raff, J. W. Cdk1 Phosphorylates Drosophila Sas-4 to Recruit Polo to Daughter Centrioles and Convert Them to Centrosomes. Dev Cell 37, 545–557 (2016).

13. Wueseke, O. et al. The Caenorhabditis elegans pericentriolar material components SPD-2 and SPD-5 are monomeric in the cytoplasm before incorporation into the PCM matrix. Mol Biol Cell 25, 2984–2992 (2014).

https://doi.org/10.7554/eLife.50130.058

Article and author information

Author details

  1. Ines Alvarez-Rodrigo

    The Sir William Dunn School of Pathology, University of Oxford, Oxford, United Kingdom
    Contribution
    Conceptualization, Resources, Formal analysis, Validation, Investigation, Visualization, Writing—original draft, Writing—review and editing
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0003-2181-5535
  2. Thomas L Steinacker

    The Sir William Dunn School of Pathology, University of Oxford, Oxford, United Kingdom
    Contribution
    Conceptualization, Resources, Formal analysis, Validation, Investigation, Visualization, Methodology
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-7244-5610
  3. Saroj Saurya

    The Sir William Dunn School of Pathology, University of Oxford, Oxford, United Kingdom
    Contribution
    Resources, Investigation
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0003-4057-0123
  4. Paul T Conduit

    The Sir William Dunn School of Pathology, University of Oxford, Oxford, United Kingdom
    Present address
    Department of Zoology, University of Cambridge, Cambridge, United Kingdom
    Contribution
    Formal analysis, Validation, Investigation, Visualization, Methodology
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-7822-1191
  5. Janina Baumbach

    The Sir William Dunn School of Pathology, University of Oxford, Oxford, United Kingdom
    Present address
    Structural and Computational Biology Unit, European Molecular Biology Laboratory, Heidelberg, Germany
    Contribution
    Resources, Formal analysis, Validation, Investigation, Visualization
    Competing interests
    No competing interests declared
  6. Zsofia A Novak

    The Sir William Dunn School of Pathology, University of Oxford, Oxford, United Kingdom
    Contribution
    Conceptualization, Resources, Formal analysis, Supervision, Methodology, Writing—review and editing
    Competing interests
    No competing interests declared
  7. Mustafa G Aydogan

    The Sir William Dunn School of Pathology, University of Oxford, Oxford, United Kingdom
    Contribution
    Formal analysis, Methodology, Writing—review and editing
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0003-1673-0596
  8. Alan Wainman

    The Sir William Dunn School of Pathology, University of Oxford, Oxford, United Kingdom
    Contribution
    Formal analysis, Supervision, Writing—review and editing
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-6292-4183
  9. Jordan W Raff

    The Sir William Dunn School of Pathology, University of Oxford, Oxford, United Kingdom
    Contribution
    Conceptualization, Supervision, Funding acquisition, Visualization, Methodology, Writing—original draft, Project administration, Writing—review and editing
    For correspondence
    jordan.raff@path.ox.ac.uk
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-4689-1297

Funding

Wellcome (Wellcome Trust Senior Investigator Award, 104575)

  • Thomas L Steinacker
  • Saroj Saurya
  • Paul T Conduit
  • Zsofia A Novak
  • Alan Wainman
  • Jordan W Raff

Wellcome (Wellcome Trust PhD studentship, 109096)

  • Ines Alvarez-Rodrigo

Wellcome (Wellcome Trust Strategic Award, 107457 (partially supported))

  • Alan Wainman

Edward Penley Abraham Scholarship (Graduate Student Scholarship)

  • Janina Baumbach
  • Mustafa G Aydogan

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Acknowledgements

We thank the members of the Raff Laboratory for critically reading the manuscript and stimulating discussions. We are grateful to Anna Caballe and Lisa Gartenmann for experimental advice; Omer Dushek for advice regarding statistical analysis; Francis Barr and Ricardo Nunes Bastos for help with the phosphopeptide enrichment and the in vitro kinase assays; Ben Thomas for mass spectrometry assistance and advice; Jonathan Bohlen for help with the radial profile analysis; Anton van der Merwe and his lab for the kind gift of purified monomeric Spycatcher-GFP for FCS analysis; Michael Barton and Andreas Haensele for taking part in the blind scoring and Andreas Haensele for help with protein purification; and the members of Micron Oxford for help and advice on live 3D structured illumination microscopy.

Superresolution microscopy was performed at the Micron Oxford Advanced Bioimaging Unit, funded by a Strategic Award from the Wellcome Trust (107457). The research was funded by a Wellcome Trust Senior Investigator Award (104575; T L Steinacker, S Saurya, A Wainman, ZA Novak, PT Conduit and JW Raff), a Wellcome Trust PhD studentship (I Alvarez-Rodrigo) and an Edward Penley Abraham Scholarship (to MG Aydogan and J Baumbach).

Senior Editor

  1. Anna Akhmanova, Utrecht University, Netherlands

Reviewing Editor

  1. Yukiko M Yamashita, University of Michigan, United States

Publication history

  1. Received: July 11, 2019
  2. Accepted: August 21, 2019
  3. Version of Record published: September 9, 2019 (version 1)

Copyright

© 2019, Alvarez-Rodrigo et al.

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.

Metrics

  • 481
    Page views
  • 71
    Downloads
  • 0
    Citations

Article citation count generated by polling the highest count across the following sources: Crossref, PubMed Central, Scopus.

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Download citations (links to download the citations from this article in formats compatible with various reference manager tools)

Open citations (links to open the citations from this article in various online reference manager services)

Further reading

    1. Cell Biology
    2. Evolutionary Biology
    Omaya Dudin et al.
    Research Article Updated
    1. Cell Biology
    2. Chromosomes and Gene Expression
    Leah F Rosin et al.
    Research Article Updated