Identification of proteins regulated by 14-3-3 from Drosophila oocytes

We recently proposed a new mechanism of spatial regulation in oocytes, in which Kinesin-14/Ncd is regulated by the combined action of the phospho-docking protein 14-3-3 in the ooplasm and Aurora B kinase on chromosomes [18].

We hypothesise that this mechanism involving 14-3-3 and Aurora B provides a general way to spatially activate many spindle proteins only when near the chromosomes in oocytes. Here we tested this hypothesis by developing a new biochemical method to systematically identify proteins whose ability to bind microtubules is regulated by 14-3-3 in oocytes.

First, we have developed a novel method of purifying microtubule-associated proteins (MAPs) from Drosophila ovaries (Fig 1A) by adapting a previous method for embryos [23]. We took advantage of Drosophila ovaries being mainly made up of mature oocytes (arrested in metaphase I) by volume with a minimal contribution from mitotic cells. A hundred pairs of ovaries were dissected from mature flies and homogenised on ice to depolymerise microtubules. After the resulting lysate was cleared by ultracentrifugation, endogenous tubulin was polymerised by incubating with paclitaxel and GTP at room temperature. Microtubules, together with their associated proteins, were pelleted through a sucrose cushion by ultracentrifugation. The pellet predominantly consists of tubulins and minimal amounts of other proteins (S1 Fig)

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Identification of proteins whose microtubule binding is regulated by 14-3-3.

(A) Dissected ovaries were homogenised and cleared by centrifugation to make soluble extract. It was separated into two and a 14-3-3 inhibitor (R18) was added to one. After microtubules are polymerised by addition of Taxol, they are sedimented by centrifugation through sucrose cushion. The pellets containing microtubules and their associated proteins were analysed by label-free quantitative mass-spectrometry. (B) Volcano plot showing the fold changes of the amounts of each protein detected in microtubule fraction in the presence of the 14-3-3 inhibitor in comparison to its absence on the X axis and the significance (p-value) on the Y axis. The red box contains 47 proteins that significantly increased their microtubule binding under 14-3-3 inhibition (the fold change >. 1.5 and the p < 0.05).

https://doi.org/10.1371/journal.pgen.1009995.g001

Next we used a 14-3-3 inhibitor, R18. R18 is a synthetic peptide with 20 residues that has been shown to competitively bind to the 14-3-3 phospho-docking site with a higher affinity than native substrates [24]. This inhibitor has been shown to bind to various 14-3-3 isoforms, including ε and ζ in vertebrates [24–26]. We expect it can inhibit both 14-3-3 isoforms (ε and ζ) in Drosophila oocytes, as the same isoforms in Drosophila and vertebrates are more similar to each other than between two different isoforms in the same species. To identify MAPs regulated by 14-3-3 in oocytes, soluble ovary extract from wild type was divided into two aliquots. The 14-3-3 inhibitor R18 was added to one, and as a control, water was added to the other. After incubation with paclitaxel and GTP, microtubules and associated proteins were pelleted from both samples.

These two microtubule fractions were then analysed by LC-MS/MS for identification and label-free quantification.

Quantitative mass-spectrometry has identified proteins potentially regulated by 14-3-3

To determine the relative amounts of each protein in the microtubule fraction with or without 14-3-3 inhibition, six pairs of biological replicates were analysed by label-free quantification using LC-MS/MS (S2–S7 Tables). As a result, we have detected 3,564 proteins in ovaries, including 24 out of 32 known spindle MAPs in our dataset. Among them, we were able to quantify 1,504 proteins (42%) in at least 2 out of the 6 experiments (S1 Table). These proteins were plotted as a volcano plot with fold differences on the X axis and the confidence level (p-value) on the Y axis (Fig 1B and S1 Table). In this plot, proteins on the top right increased their binding to microtubules under 14-3-3 inhibition. In total, 47 proteins were detected in microtubule fractions at a higher abundance in the presence of the 14-3-3 inhibitor than in its absence with a good confidence (defined by p<0.05 and ratio>1.5; Fig 1B). 14-3-3 is likely to suppress microtubule binding of these proteins in oocytes. Using equivalent criteria, 20 proteins were detected in microtubule fractions at a lower abundance in the presence of the 14-3-3 inhibitor than in its absence (S2 Fig).

It is possible that some identified proteins are not regulated directly by 14-3-3, and instead form a complex with a protein regulated by 14-3-3. To test this possibility, we examined whether any of these 14-3-3 regulated proteins or their orthologues are known to physically interact with each other using STRING database ([27]; Figs 2A and S2). Among the 47 proteins that increased microtubule binding in the presence of the 14-3-3 inhibitor, we found 48 known protein-protein interactions, which is significantly higher than expected from a random set of 47 proteins (17 interactions; p = 1.2e^-09). This is consistent with the possibility that microtubule binding of some proteins is indirectly regulated by 14-3-3 through interactions with other proteins.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Bioinformatics analysis.

(A) Physical protein-protein interactions known among 47 proteins that significantly increased their microtubule binding under 14-3-3 inhibition (the fold change >1.5 and p<0.05; the boxed area in Fig 1C). Red and blue indicate proteins with at least one predicted 14-3-3 binding sites and without them, respectively. The lines indicate physical interactions and their thicknesses indicate confidence levels of evidence. (B) The numbers and proportions of proteins with or without predicted 14-3-3 sites that increased or decreased in the microtubule fractions by 14-3-3 inhibition. (C) Sequences around predicted 14-3-3 binding sties among proteins with top 20%, middle 60% and bottom 20% of all detected proteins in the order of the fold change in microtubule fraction in the presence of the 14-3-3 inhibitor. Coloured boxes indicate the residues whose frequencies are significantly different among the top 20% than the frequencies among the rest of proteins. ** indicates p<0.01.

https://doi.org/10.1371/journal.pgen.1009995.g002

One of the proteins that increased binding to microtubules under 14-3-3 inhibition with a high degree and confidence was a known 14-3-3 regulated protein in oocytes, Kinesin-14/Ncd (Fig 1B). In addition, both subunits of centralspindlin (MKlp1/Pav, RacGAP/Tum) increased their binding to microtubules under 14-3-3 inhibition (Figs 1B and 2A). This is consistent with a previous study that shows MKlp1/Pav is directly regulated by 14-3-3 in human mitotic cells [28], although this has not been shown in oocytes. These results confirmed that our new method can successfully identify 14-3-3 regulated proteins. Similarly, three out of the four CPC subunits (Aurora B, Incenp and Borealin) significantly increased microtubule binding under 14-3-3 inhibition, except the smallest subunit Survivin/Deterin (Figs 1B and 2A). The CPC is not yet known to be regulated by 14-3-3 in any organism, and therefore was investigated further.

A broader question is whether 14-3-3 regulates only a few microtubule-associated proteins or many proteins in oocytes. To gain an insight into this question, we tested whether the presence of predicted 14-3-3 binding sites correlated with either an increase or decrease in microtubule binding under 14-3-3 inhibition (Fig 2B). If any changes under 14-3-3 inhibition are purely due to random statistical variations rather than 14-3-3 regulation, the number of proteins which increase or decrease in microtubule binding would be 50:50, regardless of whether proteins can bind 14-3-3. Among 1,504 proteins quantifiable in our experiments, 600 proteins had at least one site predicted at a high confidence to bind 14-3-3. Interestingly, among these 600 proteins, 375 increased (62%) and 225 decreased (38%) microtubule binding under 14-3-3 inhibition, showing proteins with predicted 14-3-3 binding sites are much more likely to increase microtubule binding (a difference of 150 proteins; p<0.001; Fig 2B). In contrast, among the remaining 904 proteins without predicted 14-3-3 binding sites, 448 increased (49%) and 456 decreased (51%) microtubule binding, which is not significantly different from 50:50 (p = 0.82; Fig 2B). This suggests that 14-3-3 potentially affects microtubule binding of many proteins in oocytes, and more often suppresses their microtubule binding than promotes it.

In the case of Kinesin-14/Ncd, 14-3-3 binding is regulated by phosphorylation at two sites, and regulating 14-3-3 binding is important for spatially controlling the microtubule binding activity of Kinesin-14/Ncd in oocytes. If 14-3-3 binding to a subset of proteins is regulated by a common mechanism such as phosphorylation by the same kinases, we may see a bias in sequences surrounding 14-3-3 binding sites in addition to the 14-3-3 binding motif. To gain an insight into regulation of 14-3-3 binding, sequences surrounding predicted 14-3-3 binding sites were compared between proteins whose microtubule binding is, or is not, regulated by 14-3-3. All identified proteins were divided into 5 bins according to the fold change in microtubule association under 14-3-3 inhibition. Sequences around predicted 14-3-3 sites on proteins in each bin were pooled together. As the three middle bins are similar, the top 20% (increase under inhibition), the middle 60% and the bottom 20% (decrease under inhibition) were compared (Fig 2C and S8 Table). Serine at -2, serine at 0 and lysine at -1 are significantly overrepresented among the top 20%, while proline at +2 is underrepresented. As previously shown for Kinesin-14/Ncd, Serine at -2 can be phosphorylated by a kinase and this would negatively regulate 14-3-3 binding. Overrepresentation of serine over threonine at 0 (14-3-3 binding site) might also be interesting, as the phosphatase PP2A-B55 is known to prefer phospho-threonine over phospho-serine for dephosphorylation [29].

Two phosphorylations regulate 14-3-3 interaction with the disordered region of Borealin

Our results showed that three out of the four CPC subunits (Aurora B, Incenp and Borealin) significantly increased their quantities in microtubule fractions under 14-3-3 inhibition (Fig 1C). Previous studies from us and others showed that the CPC has crucial roles in oocytes, including spindle microtubule assembly, spindle organisation, and bi-orientation of homologous centromeres [30–35]. The CPC dynamically localises to chromosomes, centromeres and the spindle equator in oocytes [33,36]. However, the CPC has not been reported to be regulated by 14-3-3 in any organism.

Among the CPC subunits, we identified 3 potential sites in Borealin and Incenp that match a consensus 14-3-3 binding sequence (Fig 3A) by bioinformatic analysis [37]. Among them, one site on Borealin (S163) not only matches the 14-3-3 binding consensus, but also shares a high similarity with the 14-3-3 binding sites of Kinesin-14/Ncd and Kinesin-6/MKlp1/Pav (Fig 3B). A distinct feature of these sites is the serine residue at the -2 position (S161) that matches the Aurora B phosphorylation consensus (R/KxS/T). Indeed, both S161 and S163 are phosphorylated in vivo, according to previous phosphoproteomics studies in Drosophila [38]. This potential 14-3-3 binding site (S163) on Borealin is located in a disordered region between the Incenp/Survivin interaction domain and the Borealin dimerisation domain [39,40] (Fig 3A).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. 14-3-3 binding to the disordered region of Borealin is regulated by two phosphorylations.

(A) Diagram of domain organisation of Borealin and Incenp proteins. Known protein-protein interactions are indicated as double arrows. Three predicted 14-3-3 binding sites (red circles; Borealin S163 and Incenp S163 and S740) are identified among the CPC subunits. Borealin S163 is located within an intrinsically disordered region between domains interacting with the CPC subunits. MT; microtubule. (B) Sequences surrounding predicted 14-3-3 binding sites. The sequence surrounding Borealin has high similarity to Ncd S96 and Pav S744, containing serine at -2 position that is potentially phosphorylated by Aurora B kinase. (C) Testing interaction between the phospho-docking protein 14-3-3 and the disordered region of Borealin (Borr). Purified MBP-tagged Borealin fragment was incubated with or without PKD2, together with BSA. GST-14-3-3ε was added and pulled down by centrifugation using glutathione beads. Proteins bound to the beads and unbound fraction (supernatant) were analysed by western blotting using an MBP antibody. (D) MBP-Borealin(113–221) interacts with GST-14-3-3ε in a manner dependent on phosphorylation at S163. MBP-Borealin(113–221) and MBP-Borealin(113–221) with S163A mutation were incubated with or without human PKD2 kinase before pulldown by glutathione beads coupled with GST-14-3-3ε. A western blot of proteins bound to the glutathione beads and unbound fractions (supernatants) was carried out using an MBP antibody and total protein staining. Quantification and statistical analysis from triplicated experiments are shown in S4 Fig. (E) An additional phosphorylation by Aurora B prevents PKD2-phosphorylated MBP-Borealin(113–221) from interacting with GST-14-3-3ε. Borealin(113–221) was incubated with human PKD2 kinase alone, human Aurora B kinase alone, both kinases or without kinases, and tested for pull down using GST or GST-14-3-3ε. Bound or unbound proteins were analysed by a western blot using an MBP antibody or total protein staining. PKD2-phosphorylated MBP-Borealin(113–221) specifically interacted with GST-14-3-3ε, while MBP-Borealin(113–221) doubly phosphorylated by PKD2 and Aurora B did not. Quantification and statistical analysis from triplicated experiments are shown in S4 Fig. (F) Aurora B cannot prevent interaction between GST-14-3-3ε and PKD2-phosphorylated MBP-Borealin(113–221) with S161A mutation. MBP-Borealin(113–221) and MBP-Borealin(113–221) with S161A mutation were incubated with or without human Aurora B in the presence of human PKD2 kinase before pulldown by glutathione beads coupled with GST-14-3-3ε. A western blot of proteins bound to the glutathione beads and unbound fractions (supernatants) was carried out using an MBP antibody and total protein staining. Quantification and statistical analysis from triplicated experiments are shown in S4 Fig.

https://doi.org/10.1371/journal.pgen.1009995.g003

To experimentally test whether 14-3-3 can bind Borealin at phosphorylated S163, this disordered region (residues 113–221) was produced with MBP-tag in bacteria (S3 Fig). As a control, the same region was produced with S163 mutated to a non-phosphorylatable alanine residue (S163A). After purification (S3 Fig), MBP-Borealin(113–221) and the S163A mutant were incubated with human PKD2 kinase. PKD2 was used as it is known to efficiently phosphorylate the similar 14-3-3 binding site in Kinesin-14/Ncd in vitro [18]. After the reaction, MBP-Borealin(113–221) and the S163A mutant were incubated with bacterially produced GST-14-3-3ε (S3 Fig). GST-14-3-3ε was then pulled down by glutathione-beads and analysed by western blot using an MBP antibody (Fig 3C). We found that MBP-Borealin(113–221) was efficiently pulled down with GST-14-3-3ε only after treatment with PKD2 (Fig 3D). In contrast, much less of the non-phosphorylatable S163A mutant was pulled down than wild-type Borealin even after PKD2 treatment (Figs 3D and S4 and S9 Table). This demonstrated that 14-3-3ε binds to Borealin at S163 in a phospho-dependent manner.

Next, we tested whether an additional phosphorylation by Aurora B can prevent Borealin from binding to 14-3-3, as a potential Aurora B phosphorylation site (S161) is located near the 14-3-3 binding site (S163). MBP-Borealin(113–221) was first incubated with PKD2, Aurora B, both kinases or no kinases, and then tested for 14-3-3 binding by pulling down with GST-14-3-3ε. We found that MBP-Borealin(113–221) phosphorylated by both PKD2 and Aurora B bound poorly to 14-3-3ε, much less than when phosphorylated by PKD2 alone (Figs 3E and S4 and S9 Table). This indicates that additional phosphorylation by Aurora B prevents 14-3-3ε interaction with Borealin already phosphorylated at S163. To confirm that this Aurora B phosphorylation is at S161, a non-phosphorylatable S161A mutation was introduced to MBP-Borealin(113–221). In a 14-3-3ε pulldown assay, Aurora B failed to prevent binding of 14-3-3ε to PKD2-phosphorylated MBP-Borealin(113–221) carrying this S161A mutation (Figs 3F and S4 and S9 Table). Therefore, the additional Aurora B phosphorylation at S161 prevents binding of 14-3-3ε to Borealin phosphorylated at S163.

Phospho-regulation of 14-3-3 binding is important for efficient Borealin localisation

Our biochemical analysis showed that 14-3-3 binding to Borealin is differentially regulated by two phosphorylations (Fig 3B and 3E). To define the in vivo role for this phospho-regulation in oocytes, we generated GFP-tagged versions of two non-phosphorylatable mutants at the 14-3-3 binding site (S163A) and the Aurora B phosphorylation site (S161A).

Their localisation was examined in the absence of the endogenous protein by expressing the GFP-tagged RNAi-resistant wild-type and non-phosphorylatable mutants together with a short hairpin RNA (shRNA) against the endogenous gene, and the chromosomal marker Rcc1-mCherry in oocytes. Live imaging showed that wild-type Borealin localised to the spindle equator and centromeres (Fig 4A and S10 Table), as reported previously for the CPC subunits [33,41].

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. Non-phosphorylatable mutations compromises the localisation of Borealin and bi-orientation of centromeres in oocytes.

(A) Non-phosphorylatable mutations (S163A, S161A) reduce the Borealin localisation to the spindle and centromeres. Fluorescence was observed in live oocytes expressing the GFP-tagged RNAi-resistant wild-type or non-phosphorylatable mutants together with a short hairpin RNA (shRNA) against the endogenous gene and a red chromosomal marker, Rcc1-mCherry. The images were presented using the same condition of capture and contrast adjustment for comparison. (B) The total signals of GFP-Borealin and Rcc1-mCherry on the spindle and centromeres above the background was quantified. Red lines indicate the median signal intensities. *** and ** indicate p<0.001 and p<0.01, respectively (Wilcoxon rank sum test). (C) Western blot of ovaries expressing GFP, wild-type ovaries without expression of GFP, and ovaries expressing wild-type or mutant GFP-Borr (S163A or S161A) in the background of borr RNAi, probed with an anti-GFP antibody and protein staining. (D) Non-phosphorylatable mutations (S163A, S161A) compromises the Borealin function in the bi-orientation of centromeres. α-tubulin and peri-centromere 3 (dodecasatellite; arrowheads) were visualised by Immunostaining combined with in situ hybridisation in borr RNAi oocytes with or without expression of the wild-type or non-phosphorylatable Borealin mutants. (E) Frequencies of bi-orientation of homologous centromeres of chromosome 3. Two separate dodecasatelite signals near both ends of the chromosome mass are considered as bi-oriented centromeres, and one or closely located two signals on the one side of the chromosome mass are considered as mono-oriented centromeres. *** and * indicate p<0.001 and p<0.05, respectively (Fisher exact test).

https://doi.org/10.1371/journal.pgen.1009995.g004

A non-phosphorylatable Borealin mutant at the 14-3-3 binding site (S163A) nearly abolished the localisation on both the spindle equator and centromeres (Fig 4A). The non-phosphorylatable Borealin mutant at the Aurora B phosphorylation site (S161A) still localised, but with a reduced intensity, on both the spindle equator and centromeres in comparison to wild-type Borealin (Fig 4A). For quantification, a total intensity of GFP signals including both the spindle equator and centromeres above the background were measured. The S163A mutation resulted in a dramatic drop in the GFP signal intensity by 82% in comparison to wild-type Borealin (p<0.001), while the S161A mutation resulted in a less dramatic but significant reduction (by 59%; p<0.001) (Fig 4B). In contrast, intensities of the chromosome signal (Rcc1-mCherry) in the phospho-mutants were not significantly different from those in wild-type Borealin, suggesting that the difference in Borealin signal intensity is due to the reduction in localisation, rather than differences in imaging (Fig 4B). Immunoblotting suggested that these differences in signal intensity are not explained by the expression levels of GFP-Borealin mutant proteins (Fig 4C).

These results revealed that phospho-regulation of 14-3-3 binding to Borealin is important for Borealin localisation in oocytes. The phospho-dependent 14-3-3 binding is crucial for Borealin localisation, while the phosphorylation by Aurora B is less crucial but still important.

14-3-3 binding to Borealin is important for bi-orienting homologous centromeres

To establish the functional significance of the phospho-regulation of 14-3-3 binding in oocytes, we further examined the oocytes expressing the wild-type or mutant Borealin transgene in a background in which the endogenous gene is silenced. We then examined spindle morphology and centromere positions in mostly metaphase I arrested oocytes by immunostaining tubulin combined with in situ hybridisation using a peri-centromere probe specific to chromosome 3 (dodecasatellite; [42]).

We examined oocytes expressing an shRNA against Borealin and without Borealin transgenes alongside a control without shRNA expression (Fig 4D and 4E). In the control without shRNA expression, one centromere 3 signal was found on each side of the chromosome mass in nearly all (95%) oocytes, showing a pair of homologous centromeres that are bi-oriented and pulled apart to both poles. Only 5% of oocytes showed a mono-orientation in the control. In contrast, 30% of RNAi oocytes showed mono-orientation (p<0.001). The spindle morphologies look similar between Borealin RNAi oocytes and the control. This is likely to reflect partial depletion of Borealin, although we cannot exclude the possibility that Borealin is not essential for normal spindle morphology.

Expression of GFP-tagged wild-type Borealin fully rescued the centromere bi-orientation defect of Borealin RNAi, and the frequency of mono-orientation went back to a similar level to oocytes without RNAi. In contrast, expression of GFP-tagged Borealin(S163A) failed to rescue the defect, and the frequency of mono-orientation stayed at a similar level to the RNAi oocytes without a transgene. This is consistent with the large reduction in localisation to the spindle/centromeres we observed above (Fig 4D and 4E and S11 Table). Expression of GFP-tagged Borealin(S161A) showed an intermediate effect, which is also consistent with partial loss of localisation. These results showed that phosphorylation for 14-3-3 binding at S163 is essential for Borealin function, while phosphorylation of the flanking site (S161) is not essential but appears to play a role. These results showed that phospho-regulation of 14-3-3 binding is important for the centromere bi-orientation function of Borealin.

Fly maturation and generation of GFP-Borr flies

Standard fly techniques were followed [45]. Drosophila melanogaster stocks were cultured on standard cornmeal medium at 25°C. w¹¹¹⁸ was used as wild-type. Less than 1 day old female adults were matured for 3 days at 25°C in the presence of males before dissection of ovaries.

To generate Borealin transgenes, the full-length Borr open reading frame was amplified from LD36125 by PCR using PrimeStar (Takara) and cloned between AscI and NotI sites of pENTR (ThermoFisher) using Gibson assembly (HiFi; NEB). To make the gene resistant to RNAi, silent mutations (from GCG GTG TTC to GCC GTC TTT) were introduced by PCR using primers containing the mutations and followed by Gibson assembly. Further mutations (TCC to GCC for S163A; AGT to GCC for S161A) were introduced using the same method. The absence of unwanted mutations were confirmed by Sanger sequencing. These entry clones were recombined with the destination vector φMGW [36] by LR Gateway Clonase II (ThermoFisher).

Transgenic flies expressing RNAi-resistant GFP-Borr (wild-type/S163A/S161A) under the maternal α -tubulin promoter were generated using φC31 integrase-mediated transgenesis at the VK18 site [46] on the second chromosome, performed by BestGene Inc. Each transgene of GFP-Borr or variant was then recombined with another transgene at the attP40 site [47] expressing short hairpin (sh) RNA against borr (HMC04381; [48]; 55942 from Bloomington Drosophila Stock Center).

Microtubule cosedimentation with the 14-3-3 inhibitor R18

Microtubules and their associated MAPs were purified from ovaries by modifying protocols previously used in embryos [49] and in S2 cells [23]. Ovaries from matured female w¹¹¹⁸ flies were dissected in 1x BRB80 buffer (80 mM PIPES-KOH pH 6.8, 1 mM MgCl₂, 1 mM Na₃EGTA) supplemented with phosphatase inhibitors (1mM DTT (Promega), 1mM PMSF, Complete EDTA-free Protease Inhibitor Mixture Tablets MINI, diluted according to manufacturer instructions (Roche), 15 mM Na₃VO₄, 10 mM p-nitrophenyl phosphate, 1 μM okadaic acid). Ovaries were snap frozen in liquid nitrogen and stored at -80°C. About 100 pairs of ovaries (~150 mg in wet weight) were defrosted and pooled. An equal volume of 2x BRB80+phosphatase inhibitors was added. Ovaries were homogenised in a pre-cooled Dounce homogeniser and incubated on ice for 30 minutes to depolymerise microtubules, then cleared by centrifugation at 13,000 rpm/13,628 x g for 15 minutes at 4°C. The supernatant was cleared again by 2 rounds of ultracentrifugation at 100,000 rpm (358,400 x g; Beckman TLA-120.2) for 10 minutes at 0°C, and then incubated at 30°C for 20 minutes, followed by ultracentrifugation at 100,000 rpm for 10 minutes at 22°C. Actin was depolymerised by the addition of 1 μg/mL Latrunculin A (Cambridge Bioscience), 2 μg/mL Latrunculin B (Cambridge Bioscience), 0.1 mM DTT (Promega) and 1 mM GTP (Sigma), and an aliquot of the lysate was kept for analysis (‘input’).

The remaining supernatant was split into two portions, one of which (‘sample’) was treated with R18 peptide (R18 trifluoroacetate, Sigma) at a final concentration of 100 μM to inhibit 14-3-3 activity. To the other (‘control’) was added an equal volume of water. Both were then incubated for 5 minutes at room temperature. Paclitaxel (Sigma-Aldrich) was subsequently added to both sample and control at a final concentration of 20 μM, and incubated for 30 minutes to allow microtubules to polymerise. Microtubules were pelleted twice through 50% sucrose cushions containing 20 μM paclitaxel via ultracentrifugation at 50,000 x g for 35 minutes at 22°C. Pellets were resuspended in 50 μl of 1x BRB80 + phosphatase inhibitors supplemented with 20 μM paclitaxel. An equal volume of 3x SDS sample buffer + 5% β-mercaptoethanol was added to samples before boiling at 95°C for 2 minutes to denature proteins. Samples were stored at -20°C. Aliquots of samples were analysed by western blot. Total proteins on the membrane were stained with MemCode reversible staining kit (Thermo-Fisher) first, and then with a rat polyclonal antibody against α-tubulin (1:500) followed by IRDye 800CW conjugated goat anti-mouse IgG antibody (LI-COR). The signals were detected with an Odyssey CLx imaging scanner (LI-COR). The brightness and contrast were adjusted uniformly across the entire area in a linear manner without removing or altering features.

Mass spectrometry

Pellets from microtubule cosedimentation were separated using a NuPAGE 12% Bis-Tris gel with MOPS running buffer and stained with Colloidal Blue Staining Kit (Invitrogen). 2 lanes were run for each pellet. The gel region containing the tubulin band was excised and discarded from both sample and control lanes. The remaining gel regions were trypsin-digested and reduced/alkylated using standard procedures [50]. Following digestion, samples were diluted with an equal volume of 0.1% TFA and spun onto StageTips as described by Rappsilber et al. [51]. Peptides were eluted in 80% acetonitrile in 0.1% TFA and concentrated 40x by vacuum centrifugation.

Samples were prepared for LC-MS/MS analysis by diluting them to 5 μL with 0.1% TFA. MS-analysis was performed on an Orbitrap Fusion Lumos tribrid mass spectrometer (Thermo Fisher Scientific, UK) coupled on-line to Ultimate 3000 RSLCnano Systems (Dionex, Thermo Fisher Scientific, UK). Peptides were separated by a PepMap RSLC C18 EasySpray column (2 μm, 100Å, 75 μm x 50 cm) (Thermo Fisher Scientific, UK), operating at 50°C. Parameters are described in S12 Table. The peptide gradient was: 2 to 40% buffer B in 140 min, then to 95% in 11 min. The percentage of buffer B remained at 95 for 5 minutes and returned back at 2 one minute after. Peptides were selected and fragmented by higher-collisional energy dissociation (HCD) [52] with normalised collision energy of 30. Raw files were processed using MaxQuant software platform [53] version 1.5.2.8 against the complete Drosophila melanogaster proteome (Uniprot, released in September 2016), using Andromeda [54]. Raw mass spectrometry proteomics data were deposited to the ProteomeXchange Consortium (http://proteomecentral.proteomexchange.org) via the PRIDE partner repository [55] with the dataset identifier PXD030445.

Bioinformatic analysis

Peptides were assigned to Uniprot accession numbers, and arranged as one accession per row. When more than one accession was assigned per gene, the accession with the lower score was removed. Accessions assigned only one peptide cannot be confidently identified and so were removed. Accessions denoting different RNA transcripts from the same gene were collapsed into one accession, again keeping the highest scored accession in the event of duplicates. The ratio between sample and control intensities for each gene was calculated as log₂(sample)–log₂(control). When either the sample or control intensity was 0, these accessions were removed from the graph and the statistical analysis and analysed separately.

To determine the consistency between replicates, log₂(sample) and log₂(control) value distributions were compared between replicates, as well as the line of best fit for the graph of log₂(sample) vs log₂(control). 6 out of 8 replicates were determined to be similar enough to continue with analysis. To compare the sample and control intensities for each accession, the given accession had to be detected in at least two replicates for both sample and control. The ratios were calculated as above, and the p-value for each ratio calculated using a two-tailed unpaired t-test. Then the ratio and p-value for each protein were plotted in log₂ as a volcano plot (Fig 2C).

Gene ontology terms were downloaded from Flybase [56]. Data manipulation and statistical analyses were performed using Excel (Microsoft) and RStudio (RStudio Team, 2020) using the R scripting language [57]. Graphs were plotted using ggplot2 [58]. Candidate 14-3-3 binding sites were predicted using the 14-3-3 Pred tool [37] on the basis of a consensus score above 0.9 and individual scores above the default thresholds (ANN 0.55; PSSM 0.80; SVM 0.25) for all 3 prediction methods. Likely kinase target sequences were predicted in Borealin sequence using the GPS 3.0 programme [59] with medium threshold. Protein fasta sequences were obtained from UniProt. Protein interaction data for network analysis was downloaded from STRING-db [27] on 29 Nov 2021. Included interactions were limited to physical subnetwork above medium confidence (0.400) sourced from experiments and databases. 14-3-3 site sequence comparison analysed and presented using Weblogo (https://weblogo.berkeley.edu/, [60]).

GST-14-3-3 binding to MBP-Borr including phosphorylation

To generate a construct expressing MBP-Borr(113–221) in E. coli, Borr open reading frame corresponding to amino acids 113–221 was first amplified with a stop codon from LD36125 by PCR using PrimeStar (Takara) and cloned between AscI and NotI sites of pENTR (ThermoFisher) using Gibson assembly (HiFi; NEB). The S161A mutation (AGT to GCC) or the S163A mutation (TCC to GCC) was introduced to this entry clone by PCR-amplifying the entire plasmid using primers carrying the mutation followed by Gibson assembly (HiFi; NEB). These entry clones were recombined using LR clonase II (ThermoFisher) with a destination vector modified from pMAL-c2 (NEB) by insertion of the Gateway cassette (ThermoFisher) at the XmnI site. GST-14-3-3ε was produced using a construct previously reported [18].

To test if Borealin is bound by 14-3-3, MBP-Borr(113–221) and GST-14-3-3ε were purified from E. coli (BL21/pLysS) carrying the expression constructs using amylose beads (NEB) and S-glutathione beads (GE Healthcare), respectively. Fractions were eluted using 10 mM reduced glutathione through a polypropylene column (Qiagen), and those with the highest concentrations combined. Combined elutes were then dialysed overnight using Slide-a-Lyser dialysis cassettes (Thermo) in 1 l of dialysis buffer (50 mM Na-phosphate buffer pH7.6, 250 mM KCl, 1 mM MgCl₂, 5 mM β-mercaptoethanol). Subsequently, the dialysed purified GST-14-3-3ε was concentrated to 500 μl total using Ultra-4 Spin Columns (Amicon), before being aliquoted, supplemented with 10% final concentration of glycerol, snap frozen in liquid nitrogen and stored at -70°C.

After thawing on ice, 3 μg (3.75 μl) of purified MBP-Borr(113–221) protein (wild-type, S161A and S163A) were cleared of aggregates by centrifugation for 1 minute at 14,000 rpm (13,628 x g), and incubated with 20 μl washed S-glutathione beads for 30 minutes at 4°C. After the beads were removed by centrifugation, 7.5 μg of MBP-Borr(113–221) was incubated with either 450 ng Aurora B (Cambridge Biosciences), 337.5 ng of PKD2 (Cambridge Biosciences), both Aurora B and PKD2 or no kinase. Phosphorylation was carried out in a two-step incubation at 30°C in phosphorylation buffer (20 mM HEPES pH 7.4, 2 mM MgCl₂, 1 mM ATP, 27.5 mM KCl, 1 mM DTT, 0.2 mg/mL BSA, 1 mM EGTA), with Aurora B added for the first 30 minutes, followed by addition of PKD2 and a further 60 minutes at 30°C. Phosphorylated fragments (5 μl) were then mixed with 20 μg GST or GST-14-3-3 in 500 μl Pulldown buffer (25 mM Tris-Cl pH = 7.6, 150 mM NaCl, 0.5% Triton X-100, 0.3 mM Na₃VO₄) and allowed to bind for 30 minutes on ice. They were then incubated with 20 μl washed S-glutathione beads on a rotator at 4°C for 1 hour. The supernatant/unbound fraction was removed and kept for analysis, and beads were then washed 3 times in 800 μl of Pulldown buffer. 2x SDS loading buffer was added to beads and supernatant samples at 1:1 ratio, supplemented with 5% final volume of β-mercaptoethanol, and were boiled for 2 minutes at 95°C, before analysis by western blot. Proteins were separated by SDS-PAGE and transferred onto a nitrocellulose membrane using the Mini-PROTEAN and Mini Trans-Blot system (Biorad). Total proteins on the membrane were stained with MemCode reversible staining kit (Thermo-Fisher). After being destained, the membrane was incubated with a rat polyclonal antibody against MBP (1:500) followed by IRDye 800CW conjugated goat anti-rat IgG antibody (LI-COR). The signals were detected with an Odyssey CLx imaging scanner (LI-COR). The brightness and contrast were adjusted uniformly across the entire area in a linear manner without removing or altering features. The band intensities over the background were quantified by the following method. Two rectangular boxes (S and L) were drawn. Box S includes mainly the band of interest, and Box L includes the band and background areas above and below the band. The total signal intensity over the background was calculated by the formula I_S − (N_S x [I_L − I_S] / [N_L − N_S]), where I and N are the total pixel intensity and pixel number in the specified box (S or L), respectively.

Live-imaging of oocytes

To visualise the localisation of GFP-Borr in live oocytes, flies carrying two transgenes that express GFP-tagged Borr (wild-type, S163A or S161A) under the maternal α-tubulin promoter and shRNA against borr under the UASp promoter were crossed with flies carrying GAL4-VP16 (V37; Bloomington Drosophila Stock Centre 7063) and Rcc1-mCherry [36] both under the maternal α-tubulin promoter. Ovaries from 3-day matured flies were dissected one at a time in Halocarbon oil (700; Halocarbon) on a cover slip. Stage 14 oocytes were identified by their morphology, namely a well-formed chorion and long dorsal appendages. Oocytes which had entered the oviduct were discarded along with other stages. These oocytes were observed under a microscope (Axiovert 200M; Zeiss) attached to a spinning disk confocal head (CSU-X1; Yokogawa) controlled by Volocity (PerkinElmer). A Plan-Apochromat objective lens (63x/1.4 numerical aperture) was used with Immersol 518F oil (Zeiss). Images were taken with a Z-slice interval of 0.8 μm and using 40% laser intensity in the red channel, 100% intensity in the green channel. Maximum intensity projections of the Z-stacks are presented as figures and were used for analyses. Images were exported in the TIFF format and signal intensity measurements carried out in ImageJ as followed. The total signal intensities of GFP-Borealin and Rcc1-mChery on the spindle were estimated using the following method. Two areas (S and L) were drawn on the maximum-intensity projection made from Z series of images. Area S includes mainly the central spindle and centromeres or all chromosomes, and Area L includes this area and surrounding region. The total signal intensity over the background was calculated by the formula I_S − (N_S x [I_L − I_S] / [N_L − N_S]), where I and N are the total pixel intensity and pixel number in the specified area (S or L), respectively.

Immunostaining and fluorescence in situ hybridisation

For immunostaining, ovaries were dissected out of mature flies in methanol as previously described [61]. Sonication was used to disrupt remove the chorion and, which can interfere with antibody penetration. Ovaries were sonicated at 38% amplitude (Vibra Cell VCX500; Sonics) for three 1-second pulses and selected for removal of chorion and vitelline membranes. This was repeated by a further one or two times. The resultant oocytes were washed in 1x PBS, and blocked for 1 hour in blocking buffer (10% foetal calf serum in PBS-T) on a rotator. Oocytes were incubated overnight with primary antibodies diluted in blocking buffer. After washing with PBS-T, fluorophore-conjugated secondary antibodies (Alexa 488, 1:250, or Cy3, 1:1000, in PBS-T) along with DAPI (0.2 μg/ml, Sigma) were added, and incubated for 2 hours in the dark. Finally, oocytes were washed in PBS-T and mounted on microscopy slides in glycerol.

For FISH combined with immunostaining in oocytes, stage 14 oocytes were prepared as for immunostaining, and then post-fixed in 8% formaldehyde as described in [62,63]. An oligonucleotide (CCCGTACTGGT)₄ for dodecasatellite [42] near centromere 3 was used. 100 pmol of probe was end-labelled with 2 nmol Alexa546-conjugated dUTP (Invitrogen), 16 nmol of unlabelled dTTP (Promega) and 30 units of terminal deoxynucleotidyl transferase (Promega) in 20 μl of transferase buffer at 37°C for one hour. The reaction was halted by incubation at 70°C for 10 minutes, and then remaining free dTTP was removed using a G25 Mini Quick spin column (Roche). 4 μl of this labelled oligonucleotide was added to ovaries in 40 μl of hybridisation buffer (0.1 g/ml dextran sulphate, 50% formamide, 3xSSC) at 30°C overnight. After washing twice in washing buffer (50% formamide, 2x SSC, 0.1% Triton X100) at 30°C for total 30 minutes, oocytes were further washed three times in 2xSSC+0.01% Triton X100 and once in PBS. After blocking in PBS+0.1% Triton containing 10% fetal calf serum, the immunostaining procedure was followed for antibody staining.

These slides were then observed via a confocal microscopy (LSM800 on AxioObserver Z1; Zeiss) using a Plan-Apochromat objective lens (63x/1.4 numerical aperture) with Immersol 518F oil (Zeiss). Images spanning an entire spindle were taken with a Z-slice interval of 0.5 μm and Zoom 2.0 (pixel size 0.1 μm). Maximum intensity projections of the Z-stacks are presented as figures.

Expression of GFP-Borr in ovaries were assessed by western blot using a rabbit anti-GFP antibody (1:100, Invitrogen, A11122) and IRDye 800CW conjugated goat anti-mouse IgG antibody (LI-COR). The signals were detected with an Odyssey CLx imaging scanner (LI-COR). MemCode reversible staining kit (Thermo-Fisher) was used to visualise any proteins. The brightness and contrast were adjusted uniformly across the entire area in a linear manner without removing or altering features.