The PP2A‐B56 Binding Site LxxIxE Contributes to Asp‐Mediated Spindle Pole Stability

ABSTRACT

The organization of microtubules into a mitotic spindle is critical for animal cell proliferation and involves the cooperation of hundreds of proteins whose molecular roles and regulation are not fully understood. The protein product of the Drosophila gene abnormal spindle, Asp, is a microtubule‐associated protein required for correct mitotic spindle formation. To better understand the contribution of Asp to microtubule organization during spindle formation, we reverse‐engineered flies to express a version of Asp (Asp^LIE), predicted to have lost its ability to bind the phosphatase trimer PP2A‐B56. We demonstrated that the Asp^LIE mutation reduced an interaction with the Drosophila PP2A‐B56 regulatory subunit Widerborst (Wdb), as well as other proteins with known roles in spindle formation. Asp^LIE flies exhibited less robust microtubule minus‐end cohesion at neural stem cell spindle poles, which was accompanied by a substantial developmental delay but no microcephaly. Predictive structural modeling suggests that the presence of Wdb alters the conformation of an Asp interaction with a tubulin dimer in a manner similar to that of the Asp^LIE mutation. Protein localization in the Drosophila embryo, in addition to in vitro microtubule organization experiments, suggests that a role of PP2A may be to prevent Asp from contributing to microtubule cross‐linking at spindle microtubule plus ends. Together, these findings add new insights to mechanisms underlying microtubule organization within the mitotic spindle.

1. Introduction

A critical step of animal cell proliferation is the organization of microtubules into a mitotic spindle, allowing accurate segregation of duplicated DNA. This organization requires the regulation of hundreds of proteins to cooperate within the dynamic spindle structure, whose cell cycle phase‐dependent interaction partners are dictated by a tightly controlled balance between kinase and phosphatase activity. The Drosophila gene abnormal spindle (asp) encodes a microtubule minus‐end‐associated protein enriched at metaphase spindle poles and the telophase midbody, contributing to metaphase pole‐focusing, spindle flux, and cytokinesis. Deregulated Asp expression is associated with cell proliferation problems leading to microcephaly or cancer (Razuvaeva et al. 2023). How Asp is regulated through the various stages of microtubule organization during mitosis is not understood.

The serine/threonine Protein Phosphatase 2A (PP2A) complex is a major contributor to cell cycle progression and spindle dynamics across eukaryotes (Moura and Conde 2019). PP2A holoenzymes are highly conserved heterotrimers comprised of a catalytic C subunit (Mts in Drosophila), a scaffolding A subunit (PP2A‐29B in Drosophila), and one of various regulatory B subunits with distinct substrate specificities. A conserved short linear motif, LxxIxE, facilitates substrate recognition by the PP2A‐B56 subfamily (Hertz et al. 2016; Wang et al. 2016). Here, we show that a version of the Asp protein lacking this motif has a reduced interaction with the Drosophila PP2A‐B56 regulatory subunits Widerborst (Wdb) and Well‐rounded (Wrd) plus other spindle pole‐focusing factors. When this version was expressed in a CRISPR asp knockout background, development was substantially delayed, accompanied by metaphase delay and less robust spindle pole formation. Microtubule aster formation in vitro was more robust in the presence of wildtype versus mutant recombinant Asp truncations, and structural modeling predicts that either the presence of Wdb or the loss of the LxxIxE PP2A‐B′ binding site similarly alters the conformation of the Asp‐tubulin dimer interaction. Comparison of GFP‐tagged Wdb vs. Asp localization in the Drosophila syncytial embryo suggests that the PP2A heterotrimer may function to prevent Asp from interacting with microtubules at the plus end. These data open up new research avenues into the regulation controlling microtubule organization during spindle pole formation.

2. Materials and Methods

2.1. Fly Lines and Cloning

Flies were housed according to standard procedures at room temperature except during assays indicated in figure legends. The initial AP‐MS of mitotically arrested embryos expressing Asp‐GFP was undertaken in a fly line expressing Asp‐GFP in an asp ^t25 mutant background (a gift from Renata Basto). All other experiments were carried out using GFP‐Asp,^WT and asp ^t25 lines were gifts from Nasser Rusan (Schoborg et al. 2015). The Asp^LIE line was generated by Quikchange PCR of the pENTR vector containing asp ^WT cDNA (gift from Nasser Rusan; Schoborg et al. 2015), sequence validated, and recombined a pUBq‐GFP Destination vector (gift from Jordan Raff), then injected into w¹¹¹⁸ embryos by BestGene Inc. (Chino Hills, CA, USA), resulting in random insertion of a poly‐ubiquitin driven GFP‐Asp^LIE, matching the previously generated GFP‐Asp^WT. A similar approach was taken to generate the pUBq‐GFP‐Wdb fly line, using the full‐length coding sequence of Wdb. Flies expressing α‐Tubulin‐GFP were obtained from the Bloomington Stock Center.

2.2. Protein Expression, Western Blots, Antibodies

Coding sequences for Asp truncations were amplified by PCR from asp pENTR vectors described above. The cDNA of full‐length wdb was purchased from the Drosophila Genome Resource Center (DGRC). The asp cDNA was inserted into a pMAL‐C2X vector and the wdb cDNA was inserted into a pNIC‐BSA vector using standard restriction enzyme cloning techniques and validated by Sanger sequencing. Proteins were expressed in Rosetta‐2(DE3) competent cells (Novagen) in LB media with appropriate antibiotic. MBP‐tagged proteins were extracted by sonication in column buffer (50 mM Hepes pH 7.4, 50 mM KCl, 1 mM MgCl₂, 1 mM EGTA, 0.1% NP40 plus protease/phosphatase inhibitors and lysozyme), column purified with amylose resin, and eluted in column buffer with 20 mM maltose. His‐tagged Wdb was purified using the MagneHis Protein Purification System (Promega), washed on magnetic nickel beads, and then incubated with equal protein amounts of respective MBP protein elutions in column buffer for 2 h. After incubation, beads were washed 3 times in column buffer, boiled in Laemmli, and resolved by SDS‐PAGE. Proteins were transferred to PVDF membranes, blocked in 5% BSA, and washed in PBS with 0.5% Triton X‐100. Proteins were labeled using anti‐MBP mouse mAb (NEB E8032S) at 1:2000 and Alexa Fluor 647 secondary (Invitrogen A21235) at 1:1000, or anti‐His rabbit pAb (Invitrogen PA5‐141016) and Alexa Fluor 790 secondary (Invitrogen A11369) at 1:1000. Proteins were visualized using a Sapphire Biomolecular Imager (Azure Biosystems) and analyzed by AzureSpot Pro software.

2.3. Mass Spectrometry

Embryos (0–3 h post egg‐lay) were collected on apple juice/agar plates in collection chambers, bleach dechorionated, washed in PBS, flash frozen in liquid nitrogen, and stored at −80°C. Clarified extract (high speed supernatant) was recovered from 400 mg of collected embryos and incubated with 30 μL of GFP‐TRAP‐A bead slurry (Chromotek, Germany) in lysis buffer (50 mM Hepes pH 7.3, 50 mM KCl, 1 mM MgCl₂, 1 mM EGTA, 0.1% NP40 with protease inhibitors) for 2 h at 4°C with rotation. Beads were washed three times and frozen in liquid nitrogen. Samples from five separate preparations per treatment group were TMT‐labeled and processed by the Bristol Proteomics Facility using Orbitrap nano‐LC MSMS. Relative abundance was determined for proteins identified in all samples, omitting “one‐hit wonders” and normalized against wildtype levels for each paired repetition. For the initial screen to identify mitotic interactors of GFP‐Asp, single peptide hits and any protein ID with a score of < 30 were removed. The resultant IDs were then cross‐referenced against an in‐house database of nonspecific (false‐positive) proteins identified from a bank of control experiments (Palumbo et al. 2015). Table 1 constitutes proteins that were: (i) either not identified in negative controls or were identified in negative controls with scores of at least 2‐fold less than in the GFP‐Asp experiment. For the comparative analysis of GFP‐Asp^WT and GFP‐Asp^LIE interactomes, a volcano plot was generated using VolcanoR freeware (Naumov et al. 2017) and significantly altered Asp‐interactions were identified if proteins met a minimum 2‐fold abundance change and 10% False Discovery Rate (Benjamini‐Hochberg corrected).

TABLE 1.

Mitotic interactors of GFP‐Asp (AP/MS).

Protein name	Accession	% coverage	# peptides	MW [kDa]	Score	Area	Function
Asp	FBpp0084071	20	34	230	1452	4.09E9	Spindle pole focusing
Nap1	FBpp0072128	26	9	43	442	2.30E8	Histone chaperone
Nlp	FBpp0084918	24	3	17	103	1.15E8	Histone‐binding protein
Pp2A‐29B	FBpp0099974	23	11	65	195	8.06E7	Catalytic subunit of PP2A
Ubi‐p5E	FBpp0070894	62	5	60	152	7.02E7	Ubiquitin
Set	FBpp0082521	22	4	31	117	6.45E7	Inhibitor of Histone acetyltransferase
Nph	FBpp0084971	45	5	17	130	4.45E7	Histone binding protein
CG2061	FBpp0073285	17	5	48	63	2.08E7	LAN‐C homolog
CG12702	FBpp0074578	24	15	97	131	1.94E7	Inhibitor of PP2A homolog
Wdb	FBpp0084573	21	9	60	57	1.86E7	Regulatory subunit of PP2A
Wrd	FBpp0082976	32	13	64	152	1.78E7	Regulatory subunit of PP2A
Cam	FBpp0087109	22	3	17	65	1.40E7	Calmodulin
Mlc‐c	FBpp0070749	42	7	17	38	1.34E7	Myosin light chain

2.4. Larval Brain Dissection and Immunofluorescence

Brains were dissected from third instar larvae at 96 h post egg lay and fixed for 20 min in 4% Formaldehyde in PBS with 1 mM EGTA and 10 mM MgCl₂. Brains were washed in PBS plus 1% Triton X‐100, blocked in 3% BSA, incubated with primary antibody overnight followed by washing and incubation with species‐appropriate secondary antibody at 1:200 (Alexa Fluor 555, Invitrogen) for at least 2 h, washing, and incubation in 50% glycerol for 2 h, then 70% glycerol overnight. The Asp‐GFP signal was retained through this process. Primary antibodies were as follows: anti‐deadpan (guinea pig), used 1:3000, a gift from Christian Q Doe; and anti‐Bazooka (rabbit), used 1:1000, a gift from Andreas Wodarz. Brains were mounted in 35% glycerol, 50% Vectashield Antifade Mounting Medium with DAPI (Vector Laboratories, UK) in PBS, mounted inside reinforcement rings, and coverslipped for imaging.

2.5. Development Assays

Larval development was assessed at 25°C for the Asp^LIE vs. Asp^WT comparisons. Adult flies were left to egg lay in tubes for 1 h at 25°C. The number of non‐wandering 3rd instar larvae was quantified at regular time intervals until 8 days post egg lay (192 h post egg lay) to measure developmental time course until pupation.

2.6. Quantitative Neural Stem Cell Analyses

Imaging was performed using a Leica SP8 confocal laser‐scanning microscope with Leica Application Suite X. For each of 3 biological reps, z‐stacks of both lobes from at least 3 larval brains were imaged per genotype and timepoint. All image processing was performed using FIJI (ImageJ Inc.; Schindelin et al. 2012). For all metaphase spindles detected within each z‐stack, the planes which comprised the spindle were selected and projected into a single maximum‐intensity image. Metaphase spindle pole splaying was scored as positive or negative by a blinded experimenter according to the Asp‐GFP signal. Ambiguous spindles were excluded from analysis (13 removed out of 609 total spindles). Results were analyzed for the “bi‐polar focused spindles only” condition by one‐way ANOVA with Tukey's post hoc. Asp distribution along spindles was measured using the Line Selection tool in FIJI (ImageJ Inc.) to draw a line intersecting both spindle poles, which was then measured for fluorescence intensity using the Plot Profile function. The lowest intensity recorded in each line scan was considered as background signal and removed. Intensities were then normalized on a scale from 0 to 1. For all intensity curves, distance normalization was performed in order to align the 2 highest fluorescence peaks (corresponding to the spindle poles). A Matlab program was created to interpolate all intensity curves, which allowed the mean fluorescence intensity and standard deviation to be calculated along the length of the spindle.

2.7. Live Imaging of Spindle Formation

Images were acquired using an Olympus spinning disc confocal microscope (model IX81, with Yokogawa CSUX1 spinning disc head and CoolSNAP HQ2 CCD camera) and a 60x objective (Oil UPlanS APO 1.3 NA). Embryos were manually dechorionated and mounted using heptane glue on 22 mm × 22 mm coverslips, then covered with a 1:1 mixture of Halocarbon oil 700 and Halocarbon oil 27 (Sigma). Larval brains were dissected as described above and mounted in Schneider's medium inside a spacing ring between two coverslips. Z‐stacks were acquired at a room temperature of 22°C every 5 s (embryo) or 20 s (brain). Image processing and analysis were performed on FIJI. Fluorescence loss caused by bleaching was corrected using the Bleach Corrector macro (developed by Kota Miura, European Molecular Biology Laboratory).

2.8. In Vitro Tubulin Assay

Rhodamine‐ or Alexa 647‐labeled tubulin (Cytoskeleton Inc. Denver, CO) was mixed at a 1:10 ratio with unlabeled porcine tubulin for a final concentration of 2 mg/ml in GPEM buffer (80 mM PIPES pH 6.9, 2 mM MgCl₂, 0.5 mM EGTA, 1 mM GTP) and incubated at 37°C for 10 min. For each 10 μL reaction, 2.5 μL of 1 mM Taxol (Cytoskeleton Inc.) was added and incubated for a further 5 min at 37°C. After microtubule stabilization, 1 μL of GPEM buffer or eluted protein (described above) was added and incubated for 10 min at room temperature. The mixture was gently mixed, and 2 μL was spotted onto a coverslip using a wide‐cut pipette tip, then inverted onto a slide for immediate imaging using a Leica TCS SP8 or Zeiss LSM900 confocal microscope. For quantification, asters were outlined in Image J (NIH) using the ellipse tool and analyzed by area and mean gray value. Background mean gray value was measured in an adjacent area not including the aster. For each aster, the mean gray value minus background was multiplied by area to produce a score representing both size and microtubule density.

2.9. Structural Predictions

Amino acid sequences were obtained from NCBI for Drosophila Asp, Wdb isoform D, β‐tubulin isoform at 60D, isoform A, and α‐tubulin at 84B. Figure 2 structures were predicted by ColabFold v1.5.2 (Mirdita et al. 2022) and all other structures were predicted using Alphafold 3 (Abramson et al. 2024) and false‐colored using UCSF ChimeraX (Meng et al. 2023).

FIGURE 2.

Abolishing the LxxIxE binding site on Asp results in developmental delay and disrupts the integrity of Asp spindle pole localization. (A) 3rd instar larval brains (dashed outline) showed no major differences in gross morphology judged by Asp‐GFP expression. Scale = 150 μm. (B) Cross‐section of Deadpan immuno‐labeled 3rd instar optic lobes showing extreme examples of optic lobe neuroblast numbers at the same developmental timepoint (96 hpf). Inset shows GFP co‐label in metaphase neuroblasts (arrowheads). Samples were fixed immediately after dissection. Scale = 30 μm. Quantification of neuroblasts is shown to the right. A single Z‐plane containing the maximum Deadpan‐labeled cells was scored per optic lobe (t(10) = 3.18, ** = p < 0.01). (C) Development of GFP‐Asp^LIE larvae was over 10 h slower than for GFP‐Asp^WT larvae. Non‐wandering larvae were scored from spiracle formation at 25°C. Solid line = mean; shaded area = standard deviation. (D, E) Examples of GFP distribution in GFP‐Asp^WT vs. GFP‐Asp^LIE optic lobe stem cell spindles 20 min (D) and 40 min (E) post dissection. Dashed lines in (E) represent line scan measurements of GFP signal intensity taken of > 100 spindles per genotype imaged from > 9 brains from 3 biological replicates. Line scan quantification shows that at 40 min post dissection GFP‐Asp^LIE spindles has a greater distribution of GFP across the entire spindle compared to GFP‐Asp^WT spindles. Lower panels: Peaks correspond to spindle poles. Solid line = mean; shaded area = standard deviation. Upper panels: Red = bazooka; green = Asp‐GFP; blue = DAPI. Scale = 5 μm; maximum projection 3 × 1 μm. (F) Quantification of > 100 spindles scored per genotype from > 9 brains over 3 biological replicates. Results are shown for 20 and 40 min dissection time windows as indicated. Significantly fewer neural stem cells in GFP‐Asp^LIE compared to GFP‐Asp^WT larval brains had 2 focused poles after 40 min of dissection time (F(3,8) = 11.73, p = 0.006; * = p < 0.05, ** = p < 0.01).

2.10. Statistical Analyses

Mass spectrometry analysis is described in the methods subsection above. Other relevant experiments were analyzed by paired t‐tests or one‐way ANOVA with Tukey's post hoc tests as indicated in figure legends and in the methods subsections above.

3. Results

3.1. The PP2A‐B56 Subunit Wdb Interacts With Asp at the LxxIxE Motif

We have previously demonstrated the power of GFP‐TRAP‐based affinity purification and mass spectrometry (AP‐MS) using transgenic Drosophila embryos to identify interaction partners of mitotic proteins (Chen et al. 2017; Cicconi et al. 2017; Hayward et al. 2014; Maccallini et al. 2020; Palumbo et al. 2015, 2020; Pellacani et al. 2018). We undertook a similar analysis of the Drosophila spindle pole‐focusing protein, Abnormal Spindle (Asp), using 0–3 h embryo extracts treated with the 26S proteasomal inhibitor MG132 to arrest embryos in a metaphase‐like state, as described in (Tariq et al. 2020). A highly stringent analysis identified 15 interacting proteins, including both catalytic and regulatory subunits of protein phosphatase 2A (PP2A) as Asp‐associated proteins (Table 1). Sequence analysis of the Drosophila Asp protein identified a PP2A‐B56 binding motif LxxIxE (Hertz et al. 2016; Wang et al. 2016) at the N‐terminus (amino acids 350–356), located within a microtubule binding region (Saunders et al. 1997). We hypothesized that the microtubule organizing roles of Asp during spindle formation may be influenced by interaction with the PP2A heterotrimer. We therefore pursued genetic, biochemical, and predictive modeling strategies to test the functional relevance of the PP2A‐B′ binding motif to modify the role of Asp within the mitotic spindle, in each case comparing a version of Asp containing the wildtype sequence to a version where key residues of the LxxIxE motif were mutated to alanine (resultant sequence AACAHA, Figure 1A).

FIGURE 1.

Asp interaction with Wdb is reduced when the LxxIxE binding site is abolished. (A) Schematic representation of known Asp domains and Asp^LIE sequence modification (MT = microtubule binding; CH=Calponin homology; IQ = IQ‐rich region; HR = heat repeats). The LxxIxE binding site is located within the MT region predicted by Saunders et al. 1997. (B) A recombinant MBP.Asp^WT 1–500 truncation co‐immunoprecipitated in vitro with His‐tagged Wdb, while MBP alone did not. (C) A recombinant MBP.Asp^LIE 1–500 truncation co‐immunoprecipitated in vitro with His.Wdb to a lesser extent than the MBP.Asp^WT 1–500 truncation. (D, E) Left panel: Predictive structural modeling showing the Asp 1–500 truncation (light blue) and Wdb (green). Center panel: Colors represent structural prediction confidence (dark blue = very high confidence (plDDT> 90), light blue = confident (plDDT 80), red = very low confidence plDDT< 50). Both Asp^WT 1–500 (D) and Asp^LIE 1–500 (E) were predicted as a β‐sheet (ASH/Hydin domain) with very high confidence surrounded by peripheral areas of low confidence, and Wdb was predicted as a very high confidence series of α‐helices arranged to form a gradual bend. The LxxIxE PP2AB56‐ binding motif was predicted with confidence to be located directly between the Asp^WT truncation and Wdb α‐helices (D). The mutated PP2A binding motif was no longer localized next to Wdb; instead, it was predicted as a region of low confidence external to the Asp‐Wdb interaction, and the Asp β‐sheet was displaced from the center of the gradual bend of Wdb α‐helices (E). (F) Wdb raw abundance scores determined by TMT‐labeled quantitative mass spectrometry were reduced across 5 experiments when co‐precipitated by GFP‐Asp^LIE compared to GFP‐Asp^WT expressed in Drosophila embryos (t(4) = 3.811; p < 0.05). (G) Wrd raw abundance scores determined by TMT‐labeled quantitative mass spectrometry were reduced across 5 experiments when co‐precipitated by GFP‐Asp^LIE compared to GFP‐Asp^WT expressed in Drosophila embryos (t(4) = 3.501; p < 0.05). (H) Volcano plot illustrating protein abundance change for all high confidence hits co‐precipitated by GFP‐Asp^LIE vs. GFP‐Asp^WT, determined by TMT‐labeled quantitative mass spectrometry (n = 5). Negative fold change reflects a loss of co‐precipitation with GFP‐Asp^LIE compared to GFP‐Asp^WT. Significantly altered hits are labeled (2‐fold abundance change at 10% false discovery rate (FDR)). BH=Benjamini‐Hochberg.

To validate the Asp‐PP2A‐B56 complex interaction by biochemistry, we co‐precipitated recombinant His‐tagged PP2A‐B56 subunit Wdb with a soluble MBP‐tagged Asp truncation containing the LxxIxE binding site. There was a robust interaction between recombinant purified wildtype MBP.Asp 1–500 and His.Wdb isoform D (Figures 1B and S1A,B). We then sought to determine whether this interaction was lost in the LxxIxE‐mutant version of MBP.Asp 1–500. Overall, there was a reduction in the His.Wdb‐MBP.Asp^LIE protein–protein interaction (Figure 1C); however, there were inconsistencies across 5 experiments (see quantification in Figure S1A). Predictive structural modeling for both the 1–500 truncations (Figure 1D,E) and full‐length Asp (Figure S1) suggests that the LxxIxE region mediates the interaction between Wdb and wildtype Asp, and that the Asp^LIE mutation interrupts this conformation but does not fully abolish the Asp interaction with Wdb. Therefore, our biochemical co‐precipitation data reflect that the MBP.Asp^LIE‐His.Wdb interaction is less efficient than that which occurs for wildtype MBP.Asp‐His.Wdb, and may be influenced by other events within the cell.

As reported elsewhere (Ito and Goshima 2015; Schoborg et al. 2015), attempts to visualize a reproducible single band by SDS‐PAGE corresponding to full‐length Asp were not successful, and we could not abolish multiple MBP‐tagged nonspecific bands, which likely represent cleaved protein products due to over‐expression in a heterologous bacterial system (Figure S1B,C). Therefore, to test whether the Asp^LIE mutation reduced this interaction for full‐length versions of the proteins, we proceeded with experiments in vivo and generated flies that rescued a previously validated asp CRISPR knockout (asp ^t25 ; Schoborg et al. 2015) with the LxxIxE binding site‐mutant version of GFP‐tagged Asp (Figure 1A; GFP‐Asp^LIE) expressed constitutively. We compared these flies to asp ^t25 knockout flies expressing GFP‐tagged wildtype Asp (GFP‐Asp^WT). We used quantitative mass spectrometry to measure the in vivo interactions between full‐length GFP‐Asp^WT/GFP‐Asp^LIE and the two Drosophila PP2A‐B56 regulatory subunits, Wdb and Well‐rounded (Wrd). We found that GFP‐Asp^WT co‐precipitated both Wdb and Wrd, and that the raw abundance scores for the co‐precipitated PP2A‐B56 subunits were consistently reduced when GFP‐Asp^LIE was precipitated (Figure 1F,G). To determine whether abolishing the LxxIxE binding motif altered the Asp interaction profile, we analyzed the full quantitative mass spectrometry data set to identify proteins with a greater than 2‐fold change in co‐precipitation with GFP‐Asp^WT vs. GFP‐Asp^LIE (Figure 1H). To account for high variance of abundance scores in the raw data, significance was assigned according to Benjamini‐Hochberg corrected pairwise t‐tests on the data normalized to percentage decrease for each of five GFP‐Asp^WT/GFP‐Asp^LIE co‐precipitation comparisons. This strategy allowed identification of proteins consistently achieving greater than 2‐fold change in co‐precipitation status regardless of overall abundance. Within these parameters, only four proteins consistently interacted at least two‐fold less with GFP‐Asp^LIE compared to GFP‐Asp^WT, including both of the PP2A‐B56 subunits Wrd and Wdb, as well as No poles (Nopo), loss of which causes mitotic arrest with barrel‐shaped spindles (Merkle et al. 2009), and Calmodulin (CaM), previously identified as an Asp‐interacting protein that also contributes to spindle pole formation and focusing (Goshima et al. 2007; Schoborg et al. 2015). It should be noted that the raw abundance scores for interacting proteins CaM and Nopo were 10‐ to 50‐fold more in this experimental setup than for Wrd and Wdb, likely reflecting spindle localization (see discussion below).

3.2. Mutation of the LxxIxE Motif on Asp Disrupts Spindle Pole Integrity in Larval Neuroblasts

Because the lack of Drosophila asp expression is associated with mitotic spindle defects and microcephaly, we investigated whether GFP‐Asp^LIE flies showed any defects in spindle formation during brain development. It has been demonstrated that an asp hypomorph phenotype dramatically affects optic lobe size due to interrupted neuroepithelial progenitor cell expansion and subsequent loss of optic lobe neuroblasts and lamina neurons (Rujano et al. 2013). We therefore investigated whether replacing wildtype Asp with GFP‐Asp^LIE would result in altered spindle morphology in larval optic lobe neuroepithelial progenitor cells and optic lobe neuroblasts. GFP‐Asp^LIE flies were viable and fertile and did not show any obvious abnormalities post eclosion. Gross morphology was similar in GFP‐Asp^WT and GFP‐Asp^LIE developing larval brains, and GFP expression was abundant in the optic lobe (Figure 2A). No microcephaly was observed in adults; however, GFP‐Asp^LIE larval brains had fewer neuroblasts than their GFP‐Asp^WT counterparts at the time of brain dissection (Figure 2B). To determine whether this reflected defective or delayed cell proliferation, we assessed the timing of larval development. Larval development was substantially delayed in the Asp^LIE genotype, indicative of less efficient, but not defective, tissue development (Figure 2C). Optic lobe neuroblast spindles appeared normal at the time of brain dissection (Figure 2D) but prolonging the time between dissection and fixation led to altered morphology in the GFP‐Asp^LIE but not the GFP‐Asp^WT genotype (Figure 2E). Systematic quantification of > 100 spindles per genotype revealed that, while brains fixed within 20 min of dissection showed no differences, GFP‐Asp^LIE brains kept in dissection medium for 40 min prior to fixation showed splayed spindle poles and a concomitant redistribution of GFP from a highly localized concentration at the metaphase pole to the entire spindle (Figure 2D–F). In contrast, GFP‐Asp^WT brains retained the Asp‐GFP signal concentrated at a focused pole at both time points, demonstrating this ex vivo phenotype to be specific to the GFP‐Asp^LIE genotype. The ability of the GFP‐Asp^LIE protein to localize to the spindle structure suggests that Asp‐microtubule binding still occurs, but the increased area of the GFP‐Asp^LIE signal at the spindle poles suggests the microtubule minus‐end focusing to be less robust. Qualitative live imaging of mitosis in optic lobe neural stem cells expressing GFP‐Asp^LIE showed spindles with an initial polar accumulation of GFP‐Asp^LIE protein subsequently becoming unfocused and persisting in prolonged metaphase (Figure S2). Within the same imaging time frame, GFP‐Asp^WT spindles always maintained a polar accumulation of the GFP signal, corroborating that an ex vivo reduction in spindle pole cohesion was specific to the GFP‐Asp^LIE genotype.

3.3. Wdb Masks the LxxIxE Motif to Prevent Asp‐Mediated Microtubule Bundling

The above data suggests that the integrity of the LxxIxE motif within the Asp protein structure contributes to the stability of the focused spindle pole; however, in the syncytial embryo, Wdb localizes to the region of the spindle proximal to the chromatin, while Asp localizes diffusely along the spindle and concentrates at the spindle poles (Figure 3A). We therefore questioned whether the loss of the Asp LxxIxE motif was sufficient to alter the influence of Asp on microtubule organization. We assessed in vitro microtubule organization in the presence of the MBP.Asp^WT and MBP.Asp^LIE purified protein fragments described above. After allowing tubulin to polymerize in vitro, the addition of an MBP.Asp^WT 1–500 fragment led to robust formation of microtubule asters, including “super‐asters” of large size and high microtubule density that were not frequently observed in other experimental conditions; this augmentation was reduced in the presence of His.Wdb. The addition of an MBP.Asp^LIE 1–500 fragment also enhanced aster formation, but to a substantially lesser extent, and was not altered by the presence of His.Wdb (Figure 3B). Together this suggests that the ex vivo phenotype could be due to a reduced ability of Asp to bundle microtubules when the LxxIxE motif is not available due to mutation; in vivo, the LxxIxE motif could be made unavailable by Wdb binding in spindle regions where Asp and Wdb both localize diffusely. In silico modeling supports this by predicting that the full‐length Asp^LIE protein interacts with an alpha tubulin/beta‐tubulin dimer in a manner similar to Asp^WT in the presence of Wdb but not Asp^WT alone (Figure 3C, discussed below).

FIGURE 3.

The LxxIxE binding site influences the ability of Asp to arrange microtubules. (A) Still images of Drosophila syncytial embryos expressing GFP transgenes as labeled. Scale bar = 2 μm. No difference was observed between Asp^WT vs. Asp^LIE localization. GFP.Wdb localizes as a diffuse cloud at the spindle midzone, overlapping with a diffuse GFP.Asp signal approaching the spindle midzone. (B) Asters formed in vitro by taxol‐stabilized microtubules incubated with recombinant His‐ or MBP‐tagged proteins as indicated. Max projections, Scale bar = 10 μm. (C) Quantification of microtubule asters represented in B. A minimum of 30 asters across 3 biological replicates were scored for each condition as “area × mean pixel” to compare both size and microtubule density in one metric. Asters were significantly increased in the presence of the MBP.AspWT 1–500 truncation compared to MBP alone (F(5,265) = 3.025; p = 0.0113; Tukey's post hoc: ** = p < 0.01). Overlaid histogram shows mean + SEM. (D) Predicted structures of Asp^WT and an α‐tubulin/β‐tubulin dimer (left panels), Asp^LIE and an α‐tubulin/β‐tubulin dimer (right panels), and Asp^WT with an α‐tubulin/β‐tubulin dimer and Wdb (center panels). Top images show confidence predictions (dark blue = very high confidence (plDDT> 90), light blue = confident (90 > plDDT > 70), yellow = low confidence 70 > plDDT > 50, orange = very low confidence plDDT < 50). The Asp N‐terminus is predicted with low confidence and all other structures are predicted with high to very high confidence. The lower images identify protein structures as follows: Gold = Asp, red = LxxIxE motif, dark & light magenta alpha helices = CH domains on Asp, pink beta pleated sheet = ASH/Hydin domain on Asp; dark green = beta‐tubulin; light green = alpha tubulin; blue = Wdb). Without Wdb, the tubulin dimer is flanked by Asp^WT CH domains (left panel). When Wdb is present, the LxxIxE motif and both Asp CH domains localize close to Wdb, and the ASH/Hydin domain localizes close to the tubulin dimer (center panel). Without Wdb, both CH domains of Asp^LIE are close together and the ASH/Hydin domain is close to the tubulin dimer (right panel).

4. Discussion

The PP2A‐B56 holoenzymes have been shown to recognize a conserved LxxIxE short linear motif (Hertz et al. 2016; Wang et al. 2016). Our biochemical data demonstrated that a direct protein–protein interaction between the PP2A‐B56 Drosophila subunit Wdb and an Asp truncation can occur and is reduced though not abolished when the LxxIxE motif is mutated (Figure 1A–C and S1A–C). Predictive structural modeling suggests that the LxxIxE site is important for the Asp Hydin domain to localize in close proximity to Wdb (Figures 1D,E and S1D,E). This data was supported by quantitative mass spectrometry, which showed in vivo that loss of the LxxIxE binding site could interrupt interaction between full‐length Asp and both PP2A‐B56 subunits Wrd and Wdb, as well as the spindle‐associated proteins No poles and Calmodulin (Figure 1F–H). Although the PP2A‐B55 subunit Twins was also identified by mass spectrometry as an Asp‐interacting protein, this interaction was not significantly influenced by the loss of the LxxIxE PP2A‐B56 binding site (log2 fold change = 0.1). This is in line with B56 subunit specificity for the LxxIxE binding site and could provide a route to compensatory PP2A holoenzyme binding. Many validated LxxIxE motifs have additional acidic residues immediately C‐terminal to the core motif (Hertz et al. 2016) which serve to reinforce interactions (Wu et al. 2017). As noted in Figure 1A, Drosophila Asp has 2 additional glutamic acid residues at positions 7 and 8. This LxxIxE motif is also found in both isoforms of the human ortholog assembly factor for spindle microtubules (ASPM), located at the ASPM protein C‐terminus (isoform 1 amino acids 3266–3271; isoform 2 amino acids 1681–1686) which is frequently truncated in disease (Létard et al. 2018). Neither human isoform has additional acidic residues immediately C‐terminal to the core motif, which may contribute to differences in ASPM vs. Asp regulation. Since humans cannot extend gestational development in an analogous manner to the extended Drosophila larval development, it will be important to also assess the relevance of this regulatory motif in a mammalian or human cell culture environment, as well as alterations in the Asp/ASPM phospho‐peptide profile.

Drosophila Asp is a microtubule minus end‐binding protein that contributes to spindle pole cohesion by cross‐linking microtubules (Ito and Goshima 2015; Schoborg et al. 2015; Wakefield et al. 2001). Here, we have demonstrated that the integrity of the LxxIxE binding motif within the Drosophila Asp protein sequence is relevant to this role. When rescuing the asp ^t25 CRISPR knockout (Schoborg et al. 2015), GFP‐Asp^WT expressing optic lobe neuroblast spindles displayed a robust polar accumulation of the GFP signal post brain dissection. This was not the case for GFP‐Asp^LIE expressing spindles, where the GFP signal was progressively re‐distributed from the pole to throughout the spindle after dissection and spindle poles became less focused (Figures 2D–F and S2). This phenotype was only detected ex vivo suggesting that compensation exists in vivo, supported by the observed developmental delay but not failure of GFP‐Asp^LIE larval development (Figure 2A–C). Possibly, localization of GFP‐AspLIE at the spindle pole requires a continuous input of energy which would be compromised after brain dissection, whereas GFP‐AspWT does not require additional ATP input to be maintained at the spindle pole. Analysis of molecular pathways with robust compensation may require pushing a system “to tipping point” prior to observing a phenotype.

In human and Drosophila mitotic metaphase, PP2A‐B56 is enriched at the centromere region of unattached chromosomes and is involved in generating kinetochore‐microtubule attachments and protecting centromeric cohesion from premature phosphorylation (F. Chen et al. 2007; Foley et al. 2011; Kruse et al. 2013; Suijkerbuijk et al. 2012). In Drosophila meiosis, PP2A‐B56 subunits Wdb and Wrd also maintain sister chromatid cohesion and have been shown to promote end‐on microtubule‐microtubule attachments rather than lateral attachments, as well as having a non‐kinetochore spindle role (Jang et al. 2021). Our immunofluorescence data (Figure 3A) similarly suggest that in Drosophila embryo mitotic spindles, GFP‐Wdb localizes to kinetochores as well as the metaphase spindle region proximal to the chromosomes. Conversely, both GFP‐Asp^WT and GFP‐Asp^LIE localize along the length of the metaphase spindle, concentrating at the spindle poles. Drosophila Asp has been shown to cross‐link microtubule minus ends at spindle poles and at minus ends throughout the spindle (Ito and Goshima 2015) including proximal to chromatin. In these regions, Asp would have the opportunity to interact with spindle‐localized Wdb, which would obscure the LxxIxE motif and prevent Asp from contributing to lateral microtubule‐microtubule connections within the spindle. Wdb does not localize to the spindle pole, which would allow Asp to better facilitate microtubule bundling since the LxxIxE motif would be available.

Our in vitro analysis shows that a recombinant MBP‐Asp^WT 1–500 truncation augments microtubule aster size and robustness compared to including recombinant MBP alone, and that this augmentation is blocked by the inclusion of recombinant His.Wdb. Recombinant MBP‐Asp^LIE 1–500 truncation does not achieve the same effect and is not influenced by the presence of His.Wdb. Given that Wdb and Asp concentrate at different regions within the metaphase spindle, we propose that the integrity of the LxxIxE binding motif itself is important for the ability of Asp to cross‐link microtubules and maintain a robust spindle pole, and that a role for spindle‐localized PP2A‐B56 could be to mask this site through Asp interaction with Wdb, preventing Asp from cross‐linking microtubules at their plus ends. This hypothesis is supported by predictive structural modeling showing that the conformation of CH domains, which can bind both actin and microtubules (Yin et al. 2020), that are located on Asp interact with a tubulin dimer in a manner that is similar when the LxxIxE binding motif is abolished or when Wdb is also interacting with Asp, whereas both of these differ from the CH‐domain conformation when the wildtype version of Asp is interacting with a tubulin dimer (Figure 3C, magenta alpha helices). Furthermore, a conserved microtubule‐associated hydin/ASH domain (ASPM, SPD‐2, Hydin; Ponting 2006; Rujano et al. 2013; Schou et al. 2014) localizes in close proximity to Wdb when no tubulin dimer is present (Figure 1D,S,1D), or close to an unstructured region of the N‐terminus when tubulin but no Wdb is present (Figure 3C, pink beta pleated sheets). In the presence of a tubulin dimer and Wdb, the ASH domain re‐locates toward this tubulin dimer, which could alter microtubule bundling properties. This ASH domain is likewise re‐located toward the tubulin dimer when the LxxIxE motif is abolished, suggesting that an impact of mutating the LxxIxE motif on microtubule organization could be the same as masking this motif by Wdb interaction. As is typical for short linear motifs, the LxxIxE motif is located within a broad unstructured region of low confidence prediction; thus, further experimental data will be needed to further validate this hypothesis.

5. Conclusion

In summary, this work validates an interaction between Drosophila PP2A‐B56 subunit Wdb and the microtubule minus‐end spindle‐associated factor Asp that is mediated in part by a conserved LxxIxE motif. Based on in vivo and in vitro data, we propose that the integrity of the LxxIxE motif supports the ability of Asp to bundle microtubule minus ends at the metaphase spindle pole, and that an interaction with Wdb could prevent Asp from bundling microtubules close to the chromosome region. The human homolog ASPM is overexpressed in many tumors, and mutations are the leading genetic cause of microcephaly; therefore, understanding the regulation of the Asp/ASPM proteins is important to further understanding spindle formation, as is required for cell proliferation.

Author Contributions

Conceptualization: L.B., J.G.W., M.Q. Data curation: L.B., M.Q. Formal analysis: L.B., M.Q. Funding acquisition L.B., J.G.W., M.C.B., A.M. Investigation L.B., M.Q., M.C.B., A.M., K.J., S.J.S. Methodology L.B., M.Q., J.G.W., M.C.B., A.M., K.J. Writing (original draft: L.B.) and writing review and editing: L.B., J.G.W., M.Q., M.C.B., A.M., K.J. Project administration L.B., J.G.W. Resources L.B., J.G.W. Software L.B., J.G.W. Supervision L.B., J.G.W.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Figure S1.

Figure S2A.

Figure S2B.

Quiniou, M. , Burns M. C., McDermott A., et al. 2025. “The PP2A‐B56 Binding Site LxxIxE Contributes to Asp‐Mediated Spindle Pole Stability.” Cytoskeleton 82, no. 12: 804–814. 10.1002/cm.22013.

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.