Phosphorylation in the intrinsically disordered region of F-BAR protein Imp2 regulates its contractile ring recruitment

Dynamic cellular processes such as endocytosis, cytokinesis and motility require reorganization of the plasma membrane in concert with cytoskeletal changes. The Bin/Amphiphysin/Rvs (BAR) family of proteins are central players in these events (reviewed in Carman and Dominguez, 2018; Salzer et al., 2017). BAR domains are membrane-binding modules that fold as crescent-shaped dimers. A subset of the BAR domain superfamily, Fer/Cip4 homology (FCH) BAR domains (F-BARs) form more elongated, less curved structures (reviewed in Carman and Dominguez, 2018; Salzer et al., 2017; Almeida-Souza et al., 2018; Hanawa-Suetsugu et al., 2019; Snider et al., 2020). By oligomerizing via their F-BAR domains, F-BAR proteins can concentrate on membranes to scaffold protein assemblies that may result in membrane deformation (reviewed in McDonald and Gould, 2016b; Snider et al., 2021). Interestingly, several biological processes, such as endocytosis and cytokinesis, involve multiple BAR proteins (Taylor et al., 2011), but why cells deploy multiple family members and whether each plays unique or redundant roles is unclear.

Two paralogous F-BAR proteins contribute to Schizosaccharomyces pombe cytokinesis. Cdc15 and Imp2 share domain organization – an N-terminal F-BAR domain, C-terminal SH3 domain and central predicted intrinsically disordered region (IDR) – but they have distinct functions. Cdc15 localizes to the division site prior to contractile ring (CR) formation and serves as a scaffold essential for stability and organization of the CR (Arasada and Pollard, 2014; Carnahan and Gould, 2003; Fankhauser et al., 1995; McDonald et al., 2015; Roberts-Galbraith et al., 2009; Snider et al., 2020; Wachtler et al., 2006; Wu et al., 2003). The F-BAR domain of Cdc15 directly binds to the formin Cdc12 (Carnahan and Gould, 2003; Willet et al., 2015) and paxillin-like Pxl1 (Snider et al., 2020), while the SH3 domain of Cdc15 binds multiple partners (Ren et al., 2015; Roberts-Galbraith et al., 2009).

Imp2 arrives at the CR after Cdc15 but prior to CR constriction (Ren et al., 2011). The CR forms normally in the absence of Imp2, but CR constriction and septation are severely compromised (Demeter and Sazer, 1998). The SH3 domains of Cdc15 and Imp2 are functionally interchangeable (Ren et al., 2015; Roberts-Galbraith et al., 2009). The F-BAR domain of Cdc15 can also substitute for that of Imp2 (McDonald et al., 2016), and the Imp2 F-BAR domain can substitute for the essential function of the Cdc15 F-BAR domain (Mangione et al., 2019). Thus, we reasoned that the remaining domains, which are predicted IDRs, must provide unique function or regulation. Extensive phosphorylation of the Cdc15 IDR inhibits its membrane binding, oligomerization and binding to Pxl1 (Bhattacharjee et al., 2020; Lee et al., 2018; Magliozzi et al., 2020; Roberts-Galbraith et al., 2010). Thus, we asked whether Imp2 is similarly phosphoregulated.

We identified Imp2 in a pilot phosphoproteomic screen designed to identify substrates of the S. pombe casein kinase 1 (CK1) protein kinases Hhp1 and Hhp2 (referred to collectively as Hhp1/2; Dhillon and Hoekstra, 1994). In all, we identified 28 phosphorylation sites matching the CK1 consensus sequence and the Cdk1 (also known as Cdc2) consensus sequence, all of which cluster within the Imp2 IDR. Our genetic, cell biological and biochemical analysis of phosphoeliminating and phosphomimetic mutants indicate a strikingly different role for phosphorylation in Imp2 compared with that in Cdc15.

We took a quantitative phosphoproteomics approach to identify mitotic substrates of Hhp1 and Hhp2 that are involved in a mitotic checkpoint (Fig. 1A) (Johnson et al., 2013). Control and hhp1-as hhp2Δ cells were arrested in prometaphase and treated for 1 h with DMSO or with 1-NM-PP1 to inhibit Hhp1. Cells were then lysed, proteins were digested with trypsin and peptides were labeled with iTRAQ reagents prior to phosphopeptide enrichment using TiO₂ beads. Peptides were analyzed using liquid chromatography–tandem mass spectrometry (LC-MS/MS), and iTRAQ reporter ion intensities were quantitated (Fig. 1B). In two replicate experiments, we identified ∼3500 unique phosphorylated peptides. Phosphopeptides with more than a 2-fold reduction in hhp1-as hhp2Δ cells with 1-NM-PP1 treatment were considered candidate CK1 substrates (red ratios in Fig. 1B).

We were intrigued to find Imp2 as a top hit in this pilot experiment (Fig. 1C). Consistent with our MS-based identification, many Imp2 phosphopeptides have been identified in global phosphoproteomics screens (Carpy et al., 2014; Kettenbach et al., 2015; Koch et al., 2011; Swaffer et al., 2016; Tay et al., 2019; Wilson-Grady et al., 2008; Wu et al., 2020). HA-tagged Imp2 (Imp2–HA₃) also migrated as multiple species during SDS–PAGE, and these collapsed to a single band upon phosphatase treatment (Fig. 1D).

To determine whether Imp2 phosphorylation varied during the cell cycle, we examined Imp2 SDS–PAGE mobility using lysates prepared from synchronized cells. Although there was a relative increase in phosphorylation during mitosis (Fig. 1E, 45 min time point), we detected Imp2–HA₃ phosphorylation throughout the cell cycle. We also used conditional cell cycle mutants to examine cell cycle effects on Imp2 phosphorylation. Again, Imp2 was most highly phosphorylated in prometaphase, but was present in multiple phosphorylated forms at all cell cycle stages (Fig. 1F).

We isolated Imp2–TAP from cells blocked in prometaphase and then used LC-MS/MS to comprehensively identify phosphorylation sites. Most of the mapped sites fit either CK1 ([pS/pT]xx[S/T] or [E/D]xx[S/T]) or Cdk1 ([S/T]P) consensus motifs (Marin et al., 1994; Songyang et al., 1994). They included one site (S383) identified in the Hhp1/2 substrate screen (Fig. S1) and many sites previously identified as targeted by Cdk1 (Swaffer et al., 2016). In total, we identified eight possible Cdk1 sites and 15 potential CK1 sites, all of which reside in the region of the protein predicted to be intrinsically disordered (Fig. 2A).

To test whether Hhp1/2 (CK1) or Cdk1 can phosphorylate Imp2, wild-type Imp2 or various mutant Imp2 proteins in which the identified phosphorylation sites had been mutated to alanine were produced recombinantly and used as substrates in kinase reactions. Imp2 was phosphorylated by Hhp1 in vitro, and mutation of the 15 identified sites and two more CK1 consensus sites eliminated this phosphorylation (Fig. 2B). Imp2 was also phosphorylated by recombinant Cdk1–cyclin complex (Cdc2–Cdc13) in vitro. Mutation of the eight identified sites matching the Cdk1 consensus did not completely prevent phosphorylation in vitro (Fig. 2C, Imp2-8A). However, mutation of three additional Cdk1 consensus sites abolished Imp2 phosphorylation by Cdk1 (Fig. 2C, Imp2-11A). Thus, Hhp1/2 and Cdk1 can phosphorylate Imp2 on the sites identified in our phosphoproteomics experiments.

Next, we examined the contributions of Cdk1 and Hhp1/2 to the phosphorylation state of Imp2 in vivo using the cdc2-as mutant (Dischinger et al., 2008), an hhp1-as hhp2Δ mutant or the triple cdc2-as hhp1-as hhp2Δ mutant, all arrested in prometaphase. Wild-type and mutant strains were treated with DMSO or the ATP analog 1-NM-PP1 for 15 or 30 min. Even without inhibitor treatment, Imp2 SDS–PAGE mobility was increased in the cdc2-as strain (Fig. 2D), reflecting the hypomorphic nature of this allele (Dischinger et al., 2008). Inhibitor treatment led to a further increase in Imp2 SDS–PAGE mobility, also consistent with Cdk1 contributing to Imp2 phosphorylation (Fig. 2D). In contrast, inhibitor treatment of hhp1-as hhp2Δ did not result in a detectable change in Imp2 mobility (Fig. 2E), whereas Imp2 mobilities in the triple mutant were the same as those in the cdc2-as single mutant (Fig. 2F). Although the experiment designed to identify Hhp1 substrates had design limitations (for example, the inhibitor treatment was too long and the phosphoproteomic workflow had not been optimized), Imp2 was reproducibly identified as a substrate. Hhp1 is found throughout the cytoplasm and at the division site (Elmore et al., 2018), so it is a plausible kinase for Imp2. Although MS is more sensitive than gel shift, the lack of a significant Imp2 mobility change in the hhp1-as hhp2Δ mutant raises the possibility that other kinases contribute to the phosphorylation of these sites in vivo or that they are relatively immune to phosphatase activity.

To determine whether phosphorylation affects Imp2 function, we replaced imp2 at the endogenous locus with imp2-11A or imp2-11E (substitutions at the 11 Cdk1 sites), imp2-17A or imp2-17E (substitutions at the 17 CK1 sites), or imp2-28A or imp2-28E (substitutions at both Cdk1 and CK1 sites). With increasing numbers of alanine mutations, the gel mobility of Imp2 increased, and reciprocally, with increasing numbers of glutamate substitutions, the gel mobility of Imp2 decreased (Fig. 2G). Furthermore, while phosphatase treatment of immunoprecipitated Imp2 resulted in increased mobility, the mobility of Imp2-28A treated with phosphatase did not change (Fig. 2H). These results indicate that the 28 identified sites account for the majority of Imp2 phosphorylation.

Phosphorylation of the Cdc15 IDR inhibits the ability of Cdc15 to bind membranes in vitro and the plasma membrane in vivo (Bhattacharjee et al., 2020; Roberts-Galbraith et al., 2010). Thus, we tested whether Imp2 phosphostatus affected membrane-binding ability (McDonald et al., 2016). Phosphorylation by Cdk1 or CK1 did not disrupt the ability of Imp2 to co-pellet with Folch fraction liposomes in vitro (Fig. 2I,J).

Deletion of imp2 causes severe CR disassembly and cell separation defects (Demeter and Sazer, 1998). Even when grown at 36°C, a condition that exacerbates the defects in imp2Δ cells, there was no detectable defect in morphology or cell division in imp2-11A, imp2-11E, imp2-17E, imp2-28A or imp2-28E cells, although some imp2-17A cells failed to separate, forming chains (Fig. S2A,B). We also measured the septation index of imp2 phosphomutant cells and found no difference compared to that of wild-type cells (Fig. S2C), indicating that none of the imp2 phosphomutants exhibit a significant loss-of-function phenotype.

Cdk1 activity inhibits cytokinesis while cells are in mitosis, and multiple cytokinetic proteins are Cdk1 substrates (Wheatley et al., 1997; reviewed in Bohnert and Gould, 2011; Wolf et al., 2007). In S. pombe, Cdk1-mediated phosphorylation of formin Cdc12 (Willet et al., 2018) and IQGAP Rng2 (Morita et al., 2021) antagonizes their CR localization. To test whether this was also the case for Imp2, we endogenously tagged either the Cdk1 site phosphomutants or the CK1 site phosphomutants with sequences encoding mNeonGreen (mNG) (Shaner et al., 2013), combined them with sad1–mCherry to visualize the spindle pole body (SPB) and imaged them at 25°C. Imp2-11A–mNG and Imp2-17A–mNG accumulated in fully formed CRs to similar levels as wild-type Imp2–mNG (Fig. S2D,E). In contrast, the levels of Imp2-11E–mNG and Imp2-17E–mNG at the CR were reduced. Imp2-17E–mNG also had reduced overall protein levels, comparable to the reduction in CR localization (∼50%), suggesting that either the glutamate substitutions or constitutive phosphorylation at all 17 sites destabilizes Imp2. In contrast, the overall protein level of Imp2-11E–mNG was similar to that of wild-type Imp2–mNG (Fig. S2D,F) suggesting that Cdk1-mediated Imp2 phosphorylation may inhibit Imp2 CR localization.

We next analyzed the kinetics and extent of Imp2 phosphomutant recruitment to the division site in two ways. First, SPB separation was set as time 0, and the first frame that Imp2 appeared at the CR for each strain was determined (Fig. 3A,B). Second, maximum spindle length was set as time 0, and Imp2 fluorescence intensity was measured at the division site over time (Fig. 3C,D). Imp2-11A–mNG was recruited to the CR earlier (∼4 min) than Imp2–mNG (Fig. 3A–C). Imp2-11E–mNG was not recruited to the CR later than Imp2–mNG, but its peak accumulation was delayed compared to that of the wild-type protein (Fig. 3A–C). Taken together, these findings implicate Cdk1 in modulating the timing of Imp2 localization to the CR and are consistent with the general theme of Cdk1 inhibiting cytokinesis until chromosome segregation is complete (reviewed in Bohnert and Gould, 2011; Wolf et al., 2007). Whereas Imp2-17A–mNG did not differ from Imp2–mNG in its kinetics, Imp2-17E–mNG was recruited later and had delayed accumulation to peak CR intensity compared to the wild-type protein (Fig. 3A,B,D), results concordant with its reduced levels.

To determine whether Imp2 mutations affected cytokinesis, we measured CR dynamics using myosin regulatory light chain Rlc1–mNG and Sid4–mNG as markers of the CR and SPB, respectively. The timing of CR formation (from SPB separation to the appearance of a coherent CR), maturation (from formation to the beginning of constriction) and constriction in wild-type, imp2-11A, imp2-11E, and imp2-17A cells was similar (Fig. 4A,B). However, imp2-17E cells exhibited slower CR constriction. In fact, CR remnants remained in 62% of imp2-17E cells for the duration of our observations. Interestingly, we noticed during imaging that imp2-17A rlc1–mNG sid4–mNG cells displayed CR sliding events where the CR formed in the middle of the cell but then slid towards one cell tip (6/18 cells; Fig. S3A). Concordantly, imp2-17A cells grown at 36°C had significantly more off-center septa than wild-type cells (Fig. S3B,C). This phenotype suggests that phosphorylation of the 17 CK1 consensus sites in Imp2 promotes the medial anchoring of the CR on the membrane, possibly by stabilizing an unknown interaction involving the Imp2 IDR. As expected, due to the lower levels of protein, imp2-17E displayed a negative genetic interaction with cdc15-140 (Fig. S4A) similar to the genetic interaction of imp2Δ with cdc15-140 (Demeter and Sazer, 1998; McDonald et al., 2016). To try to distinguish whether Imp2-17E affects myosin II contractility or cell wall assembly, we combined imp2-17E with myo2-E1, cps1-191 or myp2Δ (Fig. S4B). However, imp2-17E was synthetically lethal or sick with all three mutations, preventing a functional assignment.

In summary, we have built upon previous results implicating the central IDR of Imp2 as being key to the functional uniqueness of Imp2, distinguishing Imp2 from its paralog Cdc15 (Mangione et al., 2019; McDonald et al., 2016; Roberts-Galbraith et al., 2009). We find that while the Imp2 IDR is highly phosphorylated by multiple kinases, like that of Cdc15, the role of Imp2 IDR phosphorylation differs substantially from that of Cdc15. Whereas Cdc15 is subject to extensive cell cycle-regulated phosphorylation that controls its oligomerization, localization and interactions with protein partners (Bhattacharjee et al., 2020; Lee et al., 2018; Magliozzi et al., 2020; Roberts-Galbraith et al., 2010), the majority of Imp2 phosphorylation, which occurs on sites matching the CK1 consensus, is constitutive and appears to have modest regulatory consequences for localization kinetics and CR anchoring. It will be interesting to determine what features of the Imp2 and Cdc15 IDRs underlie their uniqueness in future studies.

All S. pombe strains used in this study (Table S1) were cultured using standard methods in YE medium, at 32°C unless otherwise noted (Moreno et al., 1991). Strain construction was accomplished through tetrad analysis using standard methods (Moreno et al., 1991). Cells were fixed with 70% ethanol for DAPI and Methyl Blue staining as described previously (Roberts-Galbraith et al., 2009). For cdc25-22 block-and-release experiments, cells were arrested in G2 at 36°C for 3.5 h and released to 25°C at the 0 time point before sample collection. For other cell cycle arrests, except for nda3-KM311, cells were arrested for 3 h at 36°C. For the nda3-KM311 arrest, cells were shifted to 18°C for 6 h. For phosphoproteomic experiments, overnight cultures were diluted to 0.2 OD₅₉₅ and shifted to 18°C for 6 h to block cells in prometaphase with the cold sensitive β-tubulin allele nda3-KM311 (Hiraoka et al., 1984) prior to chemical inhibition of the Hhp1 kinase. To inhibit the hhp1-as allele (Gregan et al., 2007) in the phosphoproteomic experiments, cells were treated with 10 μM of the ATP analog 1-NM-PP1 (Sigma-Aldrich, St Louis, MO, USA) for 1 h. In other experiments to inhibit the hhp1-as, hhp2-as and cdc2-as alleles (Dischinger et al., 2008), cells were treated with 25 μM 1-NM-PP1 (Sigma-Aldrich, St Louis, MO, USA) for 30 min.

hhp1⁺ and hhp2⁺ genes [including 500 bp upstream and downstream of the open reading frames (ORFs)] were amplified from genomic DNA by PCR, ligated into a PCR-Blunt vector (Life Technologies) and subcloned into pIRT2 (Hindley et al., 1987). The hhp1 and hhp2 analog-sensitive mutants used in Fig. 2 were created by mutagenizing the predicted gatekeeper residues to glycine (http://shokatlab.ucsf.edu/ksd/) (hhp1-M84G; hhp2-M85G) using a site-directed mutagenesis kit (QuikChange kit, Agilent Technologies). The imp2 phosphomutants were ordered as gene fragments from Life Technologies and subcloned into PCRblunt or pIRT2. To integrate the hhp1-M84G, hhp2-M85G and imp2 alleles into the genome, haploid hhp1::ura4⁺, hhp2::ura4⁺ and imp2::ura4⁺ strains, respectively, were transformed with the linear gene fragments (digested with BamHI and PstI from pIRT2-hhp1 and pIRT2-hhp2, or with PstI and Sac1 from pIRT2-imp2-28A and PCRblunt-imp2-11A, 11E, 17A or 17E) using standard lithium acetate transformations (Keeney and Boeke, 1994). Integrants were selected based on resistance to 1.5 mg/ml 5-fluoroorotic acid (Thermo Fisher Scientific) and validated by colony PCR using primers homologous to endogenous sequences that flank the genomic clones within the vectors in combination with those within the ORF. All constructs and integrants were sequenced to ensure their accuracy. The Imp2-11A and Imp2-11E mutants contained substitutions at residues 346, 367, 384, 397, 402, 426, 451, 467, 490, 511 and 531. The Imp2-17A and Imp2-17E mutants contained substitutions at residues 383, 396, 438, 454, 456, 469, 480, 503, 520, 524, 554, 560, 564, 565, 584, 595 and 596. The 28-site mutations combined the 11 and 17 sites.

Tagged strains were generated by endogenously tagging the 3′ end of ORFs with sequences encoding the appropriate epitope tag or fluorescent protein and/or resistance or auxotrophic selection cassette (kan^R, nat^R, ura4⁺ or hyg^R) using pFA6 cassettes, as previously described (Bähler et al., 1998), using lithium acetate transformations (Keeney and Boeke, 1994). G418, hygromycin or nourseothiricin (clonNAT) (100 mg/ml; Sigma-Aldrich, St Louis, MO, USA) were used for selection of kan^R, hyg^R, or nat^R cells, respectively. All fusion proteins were expressed from their native promoters at their chromosomal loci.

Cells were grown at 25°C prior to live-cell imaging unless otherwise stated. Images of S. pombe cells were acquired with a Personal DeltaVision microscope system (Applied Precision) that includes an Olympus IX71 microscope, 60× NA 1.42 Plan Apo oil immersion objective, a Photometrics CoolSnap HQ2 camera and softWoRx imaging software. Live-cell imaging was performed in YE medium in a CellAsic ONIX microchannel plate (EMD Millipore). Images in figures are deconvolved maximum intensity projections of z sections spaced at 0.5 µm, unless otherwise indicated. Images used for fluorescence quantification were not deconvolved. Quantitative analysis of microscopy data was performed using Fiji (a version of ImageJ software available at https://fiji.sc; Schindelin et al., 2012). For all intensity measurements, a region of interest (ROI) was drawn around the object of interest to obtain the raw intensity measurement. The image background was measured by obtaining an intensity measurement in a region of the image where there were no cells. The background intensity was then subtracted from the ROI (Waters, 2009).

Recombinant Hhp1 kinase, Imp2 and variants thereof were produced in E. coli Rosetta2 DE3 pLys cells (Novagen) grown in TB (12 g/l tryptone, 24 g/l yeast extract, 4 ml/l glycerol, 2.4 g/l KH₂PO₄ and 16.5 g/l K₂HPO₄). MBP-tagged Hhp1ΔC (amino acids 1–296) was purified over amylose beads (New England Biolabs) as per the manufacturer's protocol. His₆-tagged Imp2 protein and variants were purified over His-c0mplete (Roche) resin in the presence of 1% NP-40 and 5 mM β-mercaptoethanol, as per the manufacturer's protocol.

Cdk1 kinase assays were performed with 500 ng Imp2, Imp2-8A or Imp2-11A in 20 µl of 50 mM Tris-HCl pH 7.5, 100 µM unlabeled ATP, 0.8 µCi γ-³²P-ATP, 10 mM MgCl₂ and 5 mM dithiothreitol (DTT) at 30°C for 30 min. Active or kinase-dead Cdc2–Cdc13 complexes purified from baculovirus-infected Sf9 cells were used in Cdk1 kinase assays (Wolfe et al., 2006). CK1 kinase assays were performed with 300 ng MBP–Hhp1ΔC that was added to 4 μg Imp2 or Imp2-17A in 30 μl 1× PMP buffer (50 mM HEPES, 100 mM NaCl, 2 mM DTT, 0.01% Brij 35, pH 7.5; New England Biolabs) supplemented with 10 mM MgCl₂, 100 µM unlabeled ATP and 1 µCi γ-³²P-ATP for 30 min at 30°C. Reactions were stopped with the addition of SDS sample buffer, boiled and resolved using SDS–PAGE. Gels were stained with Coomassie Blue, and dried gels were subjected to autoradiography to detect phosphorylation.

To prepare liposomes, lipids were obtained from Sigma-Aldrich (B1502, brain extract from bovine brain, type I, Folch fraction). In vitro liposome preparation was performed as previously described (McDonald and Gould, 2016a). Radioactive in vitro kinase assays using recombinant His–Imp2 protein as substrate were performed either in Cdk1 buffer (20 mM Tris-HCl pH 7.4, 150 mM NaCl, 1 mM EDTA, 1 mM DTT, 10 mM MgCl₂ and 100 µM ATP) or in Hhp1 buffer [1× PMP buffer (New England Biolabs), 10 mM MgCl₂ and 100 µM ATP] with 1 µCi γ-³²P-ATP (PerkinElmer BLU002250UC). For each 25 µl kinase reaction, 1 µg of kinase (active Cdc2–Cdc13 complex purified from baculovirus-infected Sf9 cells or MBP–Hhp1ΔC) was used with 10 µg of substrate (His–Imp2). After 30 min at 30°C, reactions were added to 100 µl of 1 mg/ml liposome samples. Then the volume was increased to 200 µl gently with 50 mM Tris-HCl pH 7.4 and 150 mM NaCl, and reactions were incubated at room temperature for 15 min. Supernatants (unbound fraction) were collected from the reaction after a spin at 150,000 g for 15 min at 25°C. The pellets (bound fraction) were resuspended in 200 µl 50 mM Tris-HCl pH 7.4 and 150 mM NaCl. SDS sample buffer was added to both fractions, and 20 µl samples from all experimental conditions were run on SDS–PAGE for Coomassie Blue staining (Sigma-Aldrich; B0770). ³²P-labeled proteins were detected by autoradiography.

Cell pellets (20–30 OD₅₉₅) were snap frozen and lysed by bead disruption in NP-40 lysis buffer (Gould et al., 1991) or urea buffer (8 M urea, 300 mM NaCl, 50 mM NaPO₄, 0.5% NP40 and 4 mM imidazole, pH 8) containing protease inhibitors (cOmplete, EDTA-free Protease Inhibitor Cocktail; Roche). A FastPrep cell homogenizer (MP Biomedicals) was used for bead beating for 20–30 s. For immunoprecipitation, anti-Imp2 antibody (VU484; McDonald et al., 2016) conjugated to protein A beads was added to cleared cell lysates, then beads were washed three times with NP-40 buffer and proteins were eluted with SDS sample buffer. Proteins were separated on 10% Bis-Tris or 8% Tris-glycine gels, transferred to Immobilon-P PVDF (Millipore) membranes, and immunoblotted with anti-Imp2 (VU484; 1:1000) or anti-α-tubulin (Sigma-Aldrich; T5168; 1:5000) followed by fluorescently labeled anti-rabbit IgG or anti-mouse IgG secondary antibodies (LI-COR Biosciences), used according to the manufacturer's instructions. HA₃- and FLAG₃-tagged proteins were immunoprecipitated from NP-40 lysates with 4 μg of 12CA5 antibody (Vanderbilt Antibody and Protein Resource Core) or 4 μg anti-FLAG M2 antibody (Sigma-Aldrich) and detected by immunoblotting with the 12CA5 (1:1000) and anti-FLAG M2 antibodies (1:1000), respectively, followed by fluorescently labeled anti-mouse IgG secondary antibodies (LI-COR Biosciences). For phosphatase treatments, immunoprecipitated Imp2 was incubated with λ phosphatase (New England Biolabs) in 25 mM HEPES-NaOH (pH 7.4), 150 mM NaCl and 1 mM MnCl₂ for 30 min at 30°C prior to addition of SDS sample buffer, boiling, and separation by SDS–PAGE.

Imp2–TAP was purified as previously described (Tasto et al., 2001). Proteins were precipitated using trichloroacetic acid (TCA), digested with trypsin and analyzed by 2D-LC-MS/MS using a Thermo LTQ (Thermo Fisher Scientific, West Palm Beach, FL) as previously described (Chen et al., 2015; McDonald et al., 2002; Roberts-Galbraith et al., 2009). Peptide identifications were filtered and assembled using Scaffold 3 (v3.6.5; Proteome Software, Portland, OR, USA) using the following filters: minimum of 99.9% protein identification probability; minimum of two unique peptides; minimum of 95% peptide identification probability, resulting in protein and peptide level FDRs of 0% and 0.24%, respectively. Phosphorylation sites were analyzed using Scaffold PTM (v3.1.0).

Cells arrested in prometaphase using the nda3-KM311 β-tubulin mutant (Hiraoka et al., 1984) were treated with 10 µM 1-NM-PP1 for 1 h to inhibit the analog-sensitive allele hhp1-as. Cells were then harvested by centrifugation, snap frozen and stored at −80°C.

Denaturing lysates of control and mutant cells were made as described previously (Gould et al., 1991) using only phenylmethanesulfonylfluoride (1 mM, Sigma-Aldrich) and benzamidine (1 mM, Sigma-Aldrich) protease inhibitors. Protein concentrations were determined using a BCA kit (Thermo Fisher Scientific), and 500 µg of protein from control and mutant kinase strains was precipitated by adding six volumes of ice-cold acetone, vortexed and incubated at −20°C overnight. Proteins were pelleted by centrifugation, washed once with ice-cold acetone, dried and stored at −20°C.

Proteins were resolubilized in 8 M urea and 250 mM triethylammonium bicarbonate (TEAB), and 0.5 µg of α-casein (internal control) was added. Cysteines were reduced using 5 mM Tris(2-carboxyethyl)phosphine hydrochloride (TCEP, Sigma-Aldrich; 30 min at room temperature) and alkylated using 10 mM methyl methanethiosulfonate (MMTS, Thermo Fisher Scientific) for 15 min in the dark at room temperature. Samples were diluted to reduce urea concentration to 1 M and digested with 4 µg trypsin overnight (Promega; 37°C). Peptides were labeled with appropriate iTRAQ reagent (AB Sciex) for 1 h, as per the manufacturer's instructions. Excess label was removed using a Sep-Pak light C18 cartridge (WAT023501, Waters), and eluted peptides were dried using vacuum centrifugation.

Peptides were resuspended in 1 ml of 0.05% heptafluorobutyric acid (HFBA), 2% acetonitrile (ACN) and 300 mg/ml lactic acid then incubated with 30 mg of Titanosphere TiO₂ 5 µm beads (GL Sciences, Torrance, CA, USA; equilibrated with 0.05% HFBA and 80% ACN) for 30 min. The beads were washed once with 0.05% HFBA, 80% ACN and 300 mg/ml lactic acid and then were washed twice with 0.05% HFBA and 80% ACN. Phosphopeptides were eluted from the beads twice, once with 0.5 M ammonia and once with 5 M ammonia, then eluates were combined and dried using vacuum centrifugation.

Peptides were resuspended in 0.1% formic acid and loaded onto a self-packed biphasic C18/SCX MudPIT column using a Helium-pressurized cell (pressure bomb). The MudPIT column consisted of 360×150 µm internal diameter (i.d.) fused silica, which was fitted with a filter-end fitting (IDEX Health & Science) and packed with 5 cm of Luna SCX material (5 µm, 100 Å) followed by 4 cm of Jupiter C18 material (5 µm, 300 Å; Phenomenex). Once the sample was loaded, the MudPIT column was connected using an M-520 microfilter union (IDEX Health & Science) to an analytical column (360 µm×100 µm i.d.), equipped with a laser-pulled emitter tip and packed with 20 cm of C18 reverse phase material (Jupiter, 3 µm beads, 300 Å; Phenomenex). Using an Eksigent NanoLC Ultra HPLC and Autosampler, MudPIT analysis was performed with a 12-step salt-pulse gradient (0, 25, 50, 75, 100, 150, 200, 250, 300, 500, 750 mM and 1 M ammonium acetate). Following each salt pulse, peptides were gradient-eluted from the reverse analytical column at a flow rate of 500 nl/min. Mobile phase solvents consisted of 0.1% formic acid, 99.9% water (solvent A) and 0.1% formic acid, 99.9% ACN (solvent B). For peptides from the first 11 SCX fractions, the reverse phase gradient consisted of 2–45% solvent B in 90 min, followed by a 15 min equilibration at 2% solvent B. For the last SCX-eluted peptide fraction, the peptides were eluted using a gradient of 2–55% solvent B in 70 min, 55–95% solvent B in 15 min, 95% solvent B hold for 5 min and a 15 min equilibration at 2% solvent B. Peptides were introduced via nanospray into an LTQ-Orbitrap Velos mass spectrometer (Thermo Fisher Scientific), and the data were collected using a data-dependent method. Full scans (m/z 300–2000) were acquired with the Orbitrap as the mass analyzer (resolution 60,000), and the eight most abundant ions in each MS scan were selected for collision-induced dissociation (CID) in the Velos LTQ. Each CID MS/MS scan event was followed by acquisition of a higher-energy collisional dissociation (HCD) scan event on the same precursor as the preceding CID MS/MS scan. An isolation width of 2 m/z, activation time of 10 ms, 35% normalized collision energy and MSⁿ AGC target of 1×10⁴ were used to generate CID MS/MS spectra. For HCD spectra, 45% collision energy and an MSⁿ AGC target of 1×10⁵ were used. Dynamic exclusion was enabled, using a repeat count of one within 10 s and exclusion duration of 15 s.

Mass spectra were processed using the Spectrum Mill software package (version B.04.00; Agilent Technologies). Consecutive CID and HCD MS/MS spectra acquired on the same precursor were merged; the composite spectra contain the extracted iTRAQ reporter region from a given HCD spectrum and the preceding CID spectrum. MS/MS spectra of poor quality that failed the quality filter by not having a sequence tag length >3 were excluded from searching. A minimum matched peak intensity requirement was set at 50%. For peptide identification, MS/MS spectra were searched against an S. pombe database (pombase.org). Additional search parameters included: trypsin enzyme specificity with a maximum of three missed cleavages, ±20 p.p.m. precursor mass tolerance, and fixed modifications including MMTS alkylation of cysteines (+45.99) and iTRAQ labeling (+144.20) of lysines and peptide N-termini. Phosphorylation (+79.99) of serine, threonine and tyrosine, and oxidation (+15.99) of methionine were allowed as variable modifications. Autovalidation was performed with a maximum target–decoy-based false-discovery rate (FDR) of 2.0%.

All statistical analysis was performed in Prism 8 (GraphPad software).

This paper is dedicated to the memory of Anna Feoktistova, who will never be forgotten. We are grateful to Dr Juraj Gregan for generously providing one of the hhp1-as alleles used in this study.