Phosphoproteomic analysis identifies supervillin as an ERK3 substrate regulating cytokinesis and cell ploidy

Extracellular signal‐regulated kinase 3 (ERK3) is a poorly characterized member of the mitogen‐activated protein (MAP) kinase family. Functional analysis of the ERK3 signaling pathway has been hampered by a lack of knowledge about the substrates and downstream effectors of the kinase. Here, we used large‐scale quantitative phosphoproteomics and targeted gene silencing to identify direct ERK3 substrates and gain insight into its cellular functions. Detailed validation of one candidate substrate identified the gelsolin/villin family member supervillin (SVIL) as a bona fide ERK3 substrate. We show that ERK3 phosphorylates SVIL on Ser245 to regulate myosin II activation and cytokinesis completion in dividing cells. Depletion of SVIL or ERK3 leads to increased cytokinesis failure and multinucleation, a phenotype rescued by wild type SVIL but not by the non‐phosphorylatable S245A mutant. Our results unveil a new function of the atypical MAP kinase ERK3 in cell division and the regulation of cell ploidy.


INTRODUCTION
Extracellular signal-regulated kinase 3 (ERK3) along with its paralog ERK4 define a distinct subfamily of atypical mitogen-activated protein (MAP) kinases (Coulombe and Meloche, 2007;Cargnello and Roux, 2011). By contrast to the classical MAP kinases ERK1/2, JNK1/2/3, p38a/b/g/d and ERK5, which are activated by dual phosphorylation of the Thr-Xxx-Tyr motif in the activation loop of the kinase domain, ERK3 and ERK4 contain a single phospho-acceptor site in their activation loop and are not substrates of MAP kinase kinases.
Much remains to be learned about the regulation, substrate repertoire and functions of these protein kinases. ERK3 is phosphorylated on activation loop Ser189 by group I p21-activated kinases and displays constitutive enzymatic activity in most cell types (Déleris et al., 2008;De La Mota-Peynado et al., 2011;Déléris et al., 2011). It has been postulated that ERK3 activity is regulated, at least in part, by its cellular abundance through the control of protein turnover . Notably, ERK3 is phosphorylated by cyclin B-CDK1 during mitosis entry leading to its stabilization and transient accumulation (Tanguay et al., 2010), although the functional consequence of this regulation is unknown.
Unlike classical MAP kinases, which are multifunctional Ser/Thr kinases that phosphorylate hundreds of protein substrates (Ünal et al., 2017;Trempolec et al., 2013;Zeke et al., 2016), ERK3 does not phosphorylate generic MAP kinase substrates, suggesting a more restricted substrate specificity (Cheng et al., 1996). To date, the only physiologically validated substrate of ERK3 is the MAP kinase-activated protein kinase (MK) family member MK5 (Seternes et al., 2004;Schumacher et al., 2004;Déleris et al., 2008). Of note, classical MAP kinases bind to downstream MKs and other substrates via a cluster of negatively charged amino acids, known as the common docking (CD) domain, found in the C-terminal half of the kinase domain (Tanoue et al., 2000). However, the CD domain of ERK3 is dispensable for its interaction with MK5. Instead, ERK3 binding to MK5 is mediated by a FRIEDE motif localized in the C34 domain (Coulombe and Meloche, 2007), which is essential for MK5 activating phosphorylation (Åberg et al., 2009). This mechanism may contribute to the difference in substrate selection between ERK3 and classical MAP kinases.
Studies with genetically-engineered mice have revealed that mice lacking ERK3 kinase activity or expression are born at normal mendelian frequency and survive to adulthood without any apparent health issue (Soulez et al., 2019). However, ERK3 kinase activity is required for optimal post-natal myogenesis and muscle regeneration by stimulating the proliferation of skeletal muscle myogenic progenitors (Soulez et al., 2022). Mice deficient in ERK4 are also viable and fertile and exhibit no gross physiological anomaly (Rousseau et al., 2010). Cellular studies point to important roles for ERK3 signaling in the regulation of cytoskeletal dynamics, cell migration and invasion, cell proliferation, and transcriptional control. Overexpression of ERK3 in epithelial cells induces a reorganization of the actin cytoskeleton, associated with a decrease in the number of stress fibers and reduced cellular adhesion (Al-Mahdi et al., 2015;Mathien et al., 2017). Consistently, ERK3 signaling enhances cellular migration in different epithelial and endothelial cell types (Long et al., 2012;Al-Mahdi et al., 2015;Mathien et al., 2017;Wang et al., 2014;Elkhadragy et al., 2017). ERK3 was also shown to stimulate leukocytes chemotaxis by inducing the transcription of CXCL8 and other chemokine-encoding genes (Bogucka et al., 2020). However, the downstream effectors and molecular mechanisms underlying these functions of ERK3 remain largely unknown. Interestingly, evidence is accumulating that ERK3 signaling may contribute to tumor progression. Indeed, ERK3 was reported to promote cancer cell invasion and metastasis, clonogenic cell proliferation, and tumor growth in xenograft models (Long et al., 2012;Bogucka et al., 2020;Cai et al., 2021).
These observations further highlight the importance of defining the substrate repertoire of ERK3 to elucidate its cellular functions.
To identify novel substrates of ERK3, we performed a large-scale quantitative phosphoproteomic analysis of breast cancer epithelial cells after genetic depletion of ERK3.
This led to the identification of supervillin (SVIL) as a candidate substrate of ERK3. SVIL is the largest member of the gelsolin/villin family of actin-binding proteins (Nag et al., 2013). It is a multi-domain protein that interacts with F-actin, myosin II and other cytoskeletal proteins to regulate cellular adhesion, cell motility, podosome turnover, and matrix invasion (Takizawa et al., 2006(Takizawa et al., , 2007Fang et al., 2010;Crowley et al., 2009;Bhuwania et al., 2012). SVIL is also required for cytokinesis (Smith et al., , 2013. We validated that ERK3 phosphorylates SVIL on Ser245 in vitro and in intact cells, a modification that is required for normal execution of cytokinesis. Our results uncover SVIL as a novel substrate of ERK3 and further implicate this atypical MAP kinase in the regulation of cytokinesis and cell ploidy.

Cell lines, cell culture and transfections
Hs578T (TP53 V157F), MDA-MB-231 (TP53 R280K) and HEK 293T cells were obtained from the American Type Culture Collection. Hs578T and MDA-MB-231 cell lines were recently authenticated by short tandem repeat profiling (ATCC). Hs578T cell pools depleted of ERK3 or SVIL by CRISPR/Cas9 gene editing were purchased from Synthego. The edited cell populations were generated by electroporation of Hs578T cells with SpCas9 and sgRNA delivered as ribonucleoproteins. The sgRNA sequences used were: MAPK6, "UUUGAGAUUCUUGGUCCCAG"; and SVIL, "GCUGGCUCUCAUGUAUCGAG". Gene edition was confirmed by DNA sequencing using Synthego's Inference of CRISPR Edits software. The editing efficiency after expansion of the cell pools was 78% for MAPK6 and 97% for SVIL. All cell lines were cultured in DMEM supplemented with 10% foetal bovine serum and antibiotics at 37°C. Primary mouse embryonic fibroblasts (MEFs) were isolated from ERK3-deficient mice and wild type littermates as previously described (Voisin et al., 2010).
The cells were grown in DMEM supplemented with 10% newborn calf serum and antibiotics.
Other cell lines were transfected using Lipofectamine 3000 (Invitrogen) according to the manufacturer's protocol.

Cell synchronization
Hs578T cells were plated at a density of 8 x10 5 cells per 10-cm dish for 24 h. The cells were synchronized in S phase by addition of 2 mM thymidine for 22 h. For synchronization in G2 phase, thymidine-treated cells were released for 6 h and treated with 10 µM RO-3306 for 3 h. For synchronization in M phase, thymidine-treated cells were released for 6 h, treated with 50 µg/ml nocodazole for 4 h, followed by treatment with 10 µM MG-132 for 2 h. The cells were washed three times before each treatment. For synchronization in G1 phase, the cells were serum-starved for 24 h.

RNA interference
For RNAi-mediated silencing of ERK3 or SVIL, the following siRNAs (Horizon
Full incorporation of isotopically labeled amino acids was confirmed by analyzing an aliquot of cell extracts by MS after 3 doublings. Light and heavy cell populations (4 biological replicates) were transfected with non-target siRNA or MAPK6 SmartPool siRNAs for 24 h.
The cells were scraped directly in the culture media, and light and heavy SILAC cultures were combined at this point. Cell pellets were washed several times in cold phosphate-buffered saline (PBS) and subsequently lysed in hypotonic buffer (10 mM NaCl, 10 mM Tris-HCl (pH 7.4), 3 mM MgCl2, 10% (v/v) glycerol and 0.3% (v/v) NP-40) containing protease and phosphatase inhibitors. Lysates were centrifuged at 500 x g for 5 min, and the supernatants (cytosol) were transferred to new tubes. Proteins from cytosolic fractions were precipitated by MeOH/CHCl3. The pellets (nuclei) were washed in isotonic buffer (8.5 g sucrose/100 mL, 10 mM Tris-HCl (pH 7.4) and 5 mM MgCl2). The proteins were solubilized in digestion buffer (1% sodium deoxycholate, 100 mM Tris-HCl (pH 8.5), 40 mM chloroacetamide and 10 mM TCEP) and incubated at 95 o C for 5 min, followed by overnight digestion with trypsin at 37 o C (protein:enzyme mass ratio 100:1). Digestion was halted by acidification with formic acid (FA) (final concentration of 0.5%) and the peptides were desalted and concentrated on tC18 Sep-Pak cartridges (Waters) according to the manufacturer's protocol.

MS analysis of phosphopeptides
Phosphopeptide enrichment was performed using TiO2beads as described (Kanshin et al., 2013). Sample loading, washing and elution steps were performed using custom spin columns made from 200-μl pipette tip containing a SDB-XC membrane (3M) frit and filled with TiO2 beads. The TiO2 resin was equilibrated in 250 mM lactic acid in 70% acetonitrile (ACN) and 3% TFA, and the same buffer was used for sample loading. Phosphopeptides were eluted from TiO2 beads with 500 mM phosphate buffer (pH 7.0). Peptides were desalted in 50 μl of 1% FA and subsequently eluted from spin columns using 50 μl of 50% ACN and 0.5% FA. Eluates were dried in a SpeedVac and the phosphopeptides were resuspended in 15 μl of 4% FA prior to MS analysis. MS analysis was performed on a Tribrid Orbitrap Fusion coupled to a Proxeon nanoflow HPLC system. Each sample was loaded on a reverse-phase precolumn (5 mm length, 360 μm i.d.) and separated on a reverse-phase analytical column (18 cm length, 150 μm i.d.) (Jupiter C18, 3 μm, 300 Å, Phenomenex). LC separations were performed at a flow rate of 0.6 μl/min using a linear gradient of 5-40% aqueous ACN (0.2% FA) in 120 min.
MS spectra were acquired with a resolution of 70,000 using a lock mass (m/z: 445.120025) followed by up to 10 MS/MS data-dependent scans on the most intense ions using high energy dissociation (HCD). AGC target values for MS and MS/MS scans were set to 1e6 (max fill time 500 ms) and 1e6 (max fill time 120 ms), respectively. The precursor isolation window was set to m/z 2 with a HCD-normalized collision energy of 25. The dynamic exclusion window was set to 30 s.

MS data processing and bioinformatics
MS data were analyzed using MaxQuant (Cox and Mann, 2008;Cox et al., 2011b) software version 1.3.0.3 and searched against the SwissProt subset of the H. Sapiens uniprot database (http://www.uniprot.org/). A list of 248 common laboratory contaminants included in MaxQuant was also added to the database as well as reversed versions of all sequences. The enzyme specificity was set to trypsin with a maximum number of missed cleavages set to 2.
The precursor mass tolerance was set to 20 ppm for the first search (used for nonlinear mass recalibration (Cox et al., 2011a)) and then to 6 ppm for the main search. Phosphorylation of serine, threonine and tyrosine residues was searched as variable modification; carbamidomethylation of cysteines was searched as a fixed modification. The false discovery rate (FDR) for peptide, protein and site identification was set to 1%, the minimum peptide length was set to 6, and the "peptide requantification" function was enabled. The option match between runs (1 min time tolerance after the alignment) was enabled to correlate identification and quantitation results across different runs.
Phosphorylation sites with a decrease in abundance of ≥ 1.2-fold upon ERK3 depletion and an associated t-test P value < 0.05 were qualified as ERK3 sensitive. To select for candidate ERK3 substrates, phosphopeptides were further filtered for the presence of the minimal MAP kinase consensus motif Ser/Thr-Pro. GO term analysis was performed using the STRING database. PhosphoLogo analysis (GE et al., 2004) was performed using the WebLogo application (weblogo.berkeley.edu) with standard settings.

In vitro phosphorylation assays
Human recombinant active ERK3 and ERK4 were expressed in Sf9 insect cells from a baculovirus vector and purified as described (Seternes et al., 2004;Perander et al., 2008).
Recombinant MK5 was produced as described previously (Déleris et al., 2008). Human SVIL protein was purified from HEK 293T cells transfected with pcDNA3-Flag-SVIL. Proteins were extracted in RIPA lysis buffer (50 mM Tris-HCl (pH 7.4), 150 nM NaCl, 1 mM EDTA, 1% Triton, 0.1% SDS), briefly sonicated, and subjected to immunoprecipitation with anti-Flag antibody overnight at 4 0 C. The purity and protein concentration of SVIL was determined by SDS-gel electrophoresis and Coomassie staining. The purified SVIL was then dephosphorylated by incubation with Lambda phosphatase for 30 min at 30°C according to manufacturer's protocol, followed by phosphatase inactivation for 1 h at 65°C. For in vitro phosphorylation assays, 5 µg of purified SVIL, MK5, MBP or histone H1 was incubated with either 0.2-1 µg of ERK3, 0.9 µg of ERK4, 25 ng of ERK1 or 100 ng of CDK1/cyclin B1 in kinase buffer (50 mM Tris-HCl (pH 7.4), 10 mM MgCl2, 2 mM dithiothreitol, 0.01% Brij 35, 100 µM ATP and 5 µCi of [γ 32 P]ATP) for 30 min at 30°C. Proteins were resolved by SDS-gel electrophoresis and SVIL phosphorylation was analyzed by autoradiography and quantified by Cerenkov counting (Tricarb 2800, Perkin-Elmer). Non-radioactive phosphorylation assays were performed in the absence of radioactive ATP and analyzed by immunoblotting with antiphospho-MAP kinase motif or anti-phospho-SVIL(S245) antibodies. For mapping of SVIL phosphorylation sites by MS, 20 µg of purified SVIL was incubated with 1 µg of active ERK3 in kinase buffer for 2 h at 30°C.

Immunoblot analysis and immunoprecipitation
Proteins were extracted in RIPA lysis buffer supplemented with protease and phosphatase inhibitors with brief sonication. Immunoblotting analysis was performed as described (Servant et al., 2000). For immunoprecipitation, proteins were extracted in Tris lysis buffer (50 mM Tris-HCl (pH 7.4), 250 mM NaCl, 5 mM EDTA, 0.5% NP-40, 10 mM NaF) or RIPA buffer and incubated overnight with indicated antibodies and Protein A or G magnetic beads (Bio-Rad Laboratories).

RNA extraction and quantitative real-time PCR
Total RNA was extracted with the RNeasy Mini Kit (Qiagen) and reverse transcribed with random primers and the High Capacity cDNA Archive Kit (Applied Biosystems) as described by the manufacturer. Quantitative real-time PCR was performed as previously described (Voisin et al., 2010). The threshold cycle (Ct) value for each gene was normalised to the Ct value of actin and GAPDH and the relative levels of expression were calculated.

Quantification of multinucleated cells
For quantification of multinucleated cells, the cells were plated on glass coverslips and transfected as described in figure legends. Then, the cells were fixed with 4% paraformaldehyde for 10 min and washed three times in PBS. The cells were blocked and permeabilized by incubation in 0.1% Triton X-100, 5% bovine serum albumin for 15 min at room temperature (RT). For staining of F-actin, the cells were incubated with Alexa Fluor 555 Phalloidin for 1 h. Nuclei were stained with DAPI for 5 min and the slides were mounted with Mowiol mounting medium. Images were analyzed on an Axio Imager (Zeiss) microscope. At least 1,000 cells/condition were counted in each experiment.

Indirect immunofluorescence and quantitative analysis
Hs578T cells cultured on glass coverslips were transfected using Lipofectamine 3000 and stained 48 h after plating. For staining of MEFs, the cells were directly plated on glass coverslips and stained the next day. The cells were fixed in 4% paraformaldehyde for 5 min at RT, washed three times in PBS, and permeabilized in 0.3 % Triton X-100 in PBS for 5 min at RT. After blocking non-specific sites for 20 min in PBS, 2% BSA, the cells were incubated with primary antibodies diluted in PBS, 0.5% BSA for 1 h at RT. After three washes in PBS, the cells were further incubated with Alexa fluor dyes-conjugated secondary antibodies (1/500) in PBS, 0.5% BSA for 1 h at RT. After three additional washes in PBS, coverslips were treated with a solution of Hoechst 33342 (1 µg/mL in PBS) for 5 min at RT, followed by two washes in PBS. Cells were mounted/sealed using Prolong Gold and allowed to dry overnight at RT before being imaged. For co-staining with goat anti-GFP, primary and secondary antibodies were added sequentially to diminish background due to nonspecific interactions. For the same reason, mouse anti-Myc Alexa 488 was always added last. Phalloidin Alexa 555 was applied to coverslip with secondary antibodies. Images were acquired with a Zeiss LSM700 laserscanning confocal microscope, illuminated in multi-track mode with 405, 488, 555, and 639 nm laser lines controlled by Zen software (Zeiss). The cells were visualized with a 40×/1.4 NA Plan-Apochromat oil objective and z-stacks (0.8 µm) sectioning the entire cell were taken for all conditions. Images were processed and analyzed from the original files using ImageJ software. For multinucleation experiment, at least 1,000 cells were quantified in at least three independent biological replicates.
To quantify the fluorescence intensity of GFP from Hs578T cells transiently transfected with different forms of EGFP-SVIL (wild type, S245A or S245D), ImageJ software was used to draw a region defining the whole cell to determine the average fluorescence intensity (a. u.) for three or five consecutive z-slices (sum of the three averages for fixed samples, sum of five averages for live-cell imaging). The average fluorescence intensity (a. u.) of a similar area of background signal was also quantified for the same z-slices (sum of the averages) and subtracted from the GFP fluorescence intensity signal. To quantify the level of RLC Ser19 phosphorylation at the equator of Hs578T cells, ImageJ software was used to draw a region defining the cell equator (5-7 micron wide) and determine the average fluorescence intensity (a. u.) for three consecutive z-slices (sum of the three averages). These values were ratioed over the average fluorescence intensity (a. u.) of the whole cell, for the same z-slices (sum of the three averages). 7-14 cells were quantified in three independent biological replicates. To quantify the accumulation of EGFP-SVIL at the cell equator, maximal intensity projections of the entire z-stack of images collected in live-cell confocal microscopy were generated using the ImageJ software. A 15 micron-thick line was drawn through the equator (i.e. where the furrowing of the contractile ring occurred) of both daughter cells to generate the integrated intensity of fluorescence (a. u.) along these lines. The values of the peaks of fluorescence at the equator (5 consecutive pixels corresponding to the cell cortex) was averaged and rationed over the average of three 5-pixels regions of the cytosolic fluorescence (define as the region in the middle of the line).

Time-lapse confocal microscopy
Hs578T cells were plated on 35-mm microscopy dishes (MatTek Corporation, cat# PG35G-1.5-14-C) and left to adhere overnight before co-transfection with H2B-mRFP and GFP alone or GFP-SVIL wild type, S245A or S245D using Lipofectamine 3000. After 48 h, asynchronous cells in prometaphase were selected and images were collected every 8 min for 18 h by live-cell microscopy at 37°C on a Zeiss LSM700 laser-scanning confocal microscope illuminated in single-track mode with 488 and 555 nm laser lines, controlled by Zen software.
Cells were visualized with a 40×/1.4 NA Plan-Apochromat oil objective and z-stacks (0.8 µm) sectioning the entire cell were taken for all conditions. Images were processed and analyzed from the original files using ImageJ software and presented as maximal intensity projections renderings.
To assess cytokinesis completion or failure, the cells were visualised throughout the entirety of the movie generated by Zen software and scored for "completion" (i.e. complete closure and stabilisation of the contractile ring without reopening) or "failure" (i.e. closure followed by regression of the contractile ring, no furrowing of the ring). Segregation defects were also analysed and cells undergoing multipolar mitosis were excluded. At least 25 cells per condition were imaged and scored.

Statistical analysis
Statistics were analysed using GraphPad Prism 5.0. Pairwise Student t-test was used for comparison of data in paired observations with two unmatched groups. For most of experiments, a one-way ANOVA with Bonferroni post hoc test was used for comparison of data in paired observation experiments containing three unmatched groups or more. For livecell imaging, a Fisher's exact test was applied to the different comparisons. For cell ploidy determination in ERK3-or SVIL deleted Hs578T, we applied a two-way ANOVA followed by a post hoc Dunnett's test. Assumptions of normality and equal variance were met for the data analyzed. In all cases, a two-tailed P value < 0.05 was considered significant. All results are expressed as average ± SEM. In all figures, P values are represented as follow: *P <0 .05, **P < 0.01, ***P < 0.001, and n.s. not significant.

Data availability
The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium via the PRIDE partner repository with the dataset identifier PXD028061. All other data generated during this study are included in the main text or supplementary material or are available from the authors on reasonable request.

Quantitative phosphoproteomic analysis identifies SVIL as a candidate ERK3 substrate
We performed quantitative phosphoproteomics analyses using stable isotope labeling by amino acids in cell culture (SILAC)-based phosphoproteomics to uncover novel substrates of ERK3. Exponentially proliferating Hs578T breast cancer cells were labeled with either light or heavy amino acids and depleted or not of ERK3 by short-term treatment with pooled MAPK6 siRNAs. ERK3 expression was efficiently silenced by RNAi (Fig. 1A). Lysates from differentially labeled cells were combined and, after cellular fractionation, proteins were digested with trypsin and phosphopeptides were enriched using TiO2 beads. For phosphoproteomic analysis, peptides were separated by reverse-phase chromatography and analyzed by liquid chromatography-tandem mass spectrometry (LC-MS/MS). Overall, we identified 6,760 unique phosphorylation sites on 2,442 proteins from 4 biological replicate experiments (Table S1).
To identify potential ERK3 substrate candidates, we first selected phosphosites whose abundance upon ERK3 depletion was decreased by ≥ 1.2-fold with an associated t-test P value < 0.05. This produced a list of 50 phosphosites corresponding to a total of 38 distinct proteins ( Fig. 1B; Table S2). From this list, we used the PhosphoLogo computational tool to predict a consensus phospho-acceptor motif for ERK3. This analysis revealed that ERK3-regulated phosphoproteins show a clear preference for a proline residue at the +1 position and, to a lesser extent, at the -2 position ( Fig. 1C; Table S2), in agreement with the minimal and optimal consensus phosphorylation motifs Ser/Thr-Pro and Pro-Xxx-Ser/Thr-Pro defined for MAP kinases (Gonzalez et al., 1991;Songyang et al., 1996). Gene ontology (GO) analysis of cellular components revealed that ERK3-regulated phospho-Ser/Thr-Pro proteins are enriched in the cytoskeleton and midbody categories (Table 1). One candidate substrate of special interest was the F-actin-and myosin II-binding protein SVIL. The phosphorylation of multiple SVIL residues was significantly decreased in ERK3-depleted Hs578T cells (Fig. 1B, D, E). We focused our subsequent analysis on the regulation of SVIL Ser245, as this residue lies in an optimal MAP kinase consensus motif.

ERK3 interacts with and phosphorylates SVIL on Ser245
As a first validation step, we tested the ability of ERK3 to phosphorylate SVIL in vitro. SVIL is a large protein of ~ 250 kDa that is not easily amenable to purification in bacteria. We therefore expressed full-length human SVIL as a Flag-tagged recombinant protein in HEK 293T cells and purified the protein to > 80% purity by affinity pull-down with anti-Flag M2 magnetic beads. Coomassie staining of the resulting gel revealed the presence of two major bands. The upper band was identified as SVIL isoform 4 by MS, while the lower band corresponded to a mix of SVIL isoform 1 and co-purifying myosin IX and myosin X proteins ( Fig. 2A). For in vitro phosphorylation assays, purified SVIL was first dephosphorylated with Lambda phosphatase, and then incubated with purified recombinant active ERK3 and ATP. We found that ERK3 phosphorylates SVIL in vitro in a dose-dependent manner ( Fig. 2A, B) and to an extent comparable to MK5, used as positive control (Fig. 2C). To evaluate the kinase selectivity of SVIL phosphorylation, we further examined the ability of the paralogous ERK4 and the multifunctional ERK1 MAP kinases to phosphorylate SVIL in vitro. The amount of each kinase used in the phosphorylation assay was normalized to their enzymatic activity on the common substrate MBP. Interestingly, purified recombinant ERK4 phosphorylated SVIL with an activity comparable to ERK3 (Fig. 2D) MS analysis identified Ser245 as a major in vitro phosphorylation site of SVIL by ERK3 (Fig. 2E). Ser245 is one of the most frequently phosphorylated sites of SVIL recorded in the PhosphoSitePlus database (www.phosphosite.org). This site is well conserved among the different SVIL isoforms and across other mammalian species (Fig. 2F). We generated a phospho-specific antibody against SVIL Ser245 to explore the regulation and functional impact of this modification. The phospho-SVIL(S245) antibody strongly reacted with ERK3phosphorylated immunopurified SVIL, and the immunoreactivity was completely abolished after substitution of Ser245 by Ala or by treatment with Lambda phosphatase, confirming the specificity of the antibody for phosphorylated Ser245 (Fig. 2G).
To substantiate these results, we evaluated if ERK3 phosphorylates SVIL on Ser245 in intact cells. We first examined if ERK3 can co-immunoprecipitate with SVIL. We showed that ERK3 and SVIL form stable complexes when the two proteins are ectopically expressed in HEK 293T cells (Fig. 3A). The two endogenous ERK3 and SVIL proteins were also found to co-immunoprecipitate from Hs578T cell extracts (Fig. 3B). Co-expressing the individual domains of ERK3 with SVIL in HEK 293T revealed that SVIL preferentially interacts with the ERK3(365-721) C-terminal domain (Fig. 3C). We used the phospho-SVIL(S245) antibody to assess if changes in ERK3 levels impact SVIL Ser245 phosphorylation in vivo. Co-expression of ERK3 with SVIL in HEK 293T cells significantly increased the phosphorylation of SVIL on Ser245, and this signal was completely abolished when expressing the non-phosphorylatable SVIL S245A mutant (Fig. 3D). Similar result were obtained after probing extracts with a phospho-MAP kinase motif antibody (Fig. 3D). RNAi-mediated depletion of ERK3 significantly decreased (but did not eliminate) the steady-state phosphorylation of SVIL on Ser245 (Fig. 3E). Together, these results identify SVIL as a novel bona fide ERK3 substrate and Ser245 as a major site of ERK3 activity.

ERK3 regulates cell ploidy through phosphorylation of SVIL on Ser245
SVIL is a member of the large family of gelsolin/villin proteins and was shown to regulate various properties of the cell cortex, in part through direct binding to F-actin and myosin II via its N-terminal region. One important activity of SVIL, demonstrated in HeLa cells, is to contribute to the proper initiation and completion of cytokinesis, thus ensuring the separation of chromosomes into distinct daughter cells at the end of mitosis (Smith et al., , 2013Hasegawa et al., 2013). We therefore asked whether phosphorylation of SVIL on Ser245 by ERK3 impacts cytokinesis. We first used immunofluorescence to assess whether RNAimediated depletion of ERK3 or SVIL in breast cancer cell lines resulted in an increase in the proportion of multinucleated cells, a typical consequence of cytokinetic failure. We found that transient depletion of ERK3 with three distinct siRNAs led to a small but significant increase in the proportion of multinucleated Hs578T cells ( Fig. S1A-C). Consistent with previous reports (Smith et al., , 2013Hasegawa et al., 2013), RNAi depletion of SVIL also increased the percentage of multinucleated Hs578T cells compared to control (Fig. S1A-C).
The percentage of multinucleated SVIL-depleted cells was smaller than observed previously in HeLa cells (Hasegawa et al., 2013), which could be due to cell type differences. Interestingly, re-expression of either wild type or phosphomimetic S245D forms of SVIL in SVIL-depleted cells significantly rescued the frequency of multinucleated cells to a degree comparable to controls, while expression of the non-phosphorylatable S245A mutant had no such effect ( We employed a pharmacological approach to further validate the impact of ERK3 on SVIL phosphorylation and cell ploidy. Treatment of Hs578T cells with a small molecule inihibitor of ERK3 (Grädler et al., 2020) reduced the phosphorylation of SVIL on Ser245 (Fig.   4A). Furthermore, pharmacological inhibition of ERK3 significantly increased the frequency of multinucleated Hs578T cells (Fig. 4B, C), demonstrating the requirement for the catalytic activity of ERK3 and providing orthogonal validation to the genetic depletion experiments).
Treatment with ERK3 inhibitor did not further increase the multinucleation of SVIL-depleted cells ( Fig. 4B; see also Fig. 5A), suggesting that ERK3 and SVIL lie on the same signaling axis.
These results demonstrate that both ERK3 and SVIL are required to maintain proper cell ploidy and suggest that phosphorylation of SVIL at Ser245 is functionally relevant for cell division.
Of note, no difference in cell ploidy was observed between wild type and ERK3-deficient primary MEFs (Fig. S2), suggesting that this regulation may be context-dependent (see Discussion).
To further demonstrate that ERK3 and SVIL function together to regulate cell ploidy, we used CRISPR/Cas9-mediated gene editing to generate Hs578T cells depleted of either gene product. The knockdown efficiency was estimated to be ~ 90% for ERK3 and > 80% for SVIL ( Fig. 5A). While ERK3-or SVIL-depleted Hs578T cells were viable, they showed an increase in the percentage of multinucleated cells compared to control Hs578T cells (Fig. 5B, C), consistent with the results obtained by RNAi and pharmacological inhibition. Expression of GFP-tagged wild type or phosphomimetic S245D forms of SVIL rescued the multinucleation phenotype associated with SVIL depletion, while expressing the non-phosphorylatable S245A form of SVIL had no such effect (Fig. 5B, C). Notably, expression of wild type or S245D forms of SVIL in ERK3-depleted cells also rescued multinucleation (Fig. 5B, C), consistent with SVIL regulating cell ploidy downstream of ERK3. The observation that overexpression of wild type SVIL can rescue ploidy in ERK3-depleted cells suggests that another kinase (other than ERK4 that is not expressed in Hs578T cells) can phosphorylate SVIL at Ser245 to compensate for the loss of ERK3. This is in agreement with our finding that RNAi depletion of ERK3 does not completely abolish SVIL Ser245 phosphorylation (see Fig. 3E). CDK1 also preferentially phosphorylates Ser/Thr residues followed by a proline, making it a candidate SVIL kinase.
However, we found that CDK1/cyclin B does not phosphorylate SVIL on Ser245 in vitro (Fig.   S3). Together, our results support the notion that Ser245 phosphorylation of SVIL by ERK3 is functionally relevant for SVIL activity in the regulation of cell division.

ERK3 phosphorylation of SVIL on Ser245 regulates cytokinesis
To determine whether the changes in Hs578T cell ploidy resulting from loss of ERK3 or SVIL were due to defects in cytokinesis, we used time-lapse confocal imaging of dividing control, ERK3-or SVIL-depleted Hs578T cells transiently expressing fluorescent proteintagged versions of alpha-tubulin and histone H2B to monitor mitotic progression (Movies S1-3). We also transiently co-expressed EGFP-tagged forms of SVIL (wild type, S245A or S245D) with histone H2B in control, ERK3-or SVIL-depleted Hs578T cells to further assay the role of Ser245 in SVIL function. All cell populations were viable and most individual cells monitored entered mitosis. In nearly all cell lines, we observed a small percentage of cells forming multipolar spindles after mitotic entry, consistent with our findings that ploidy is impacted in Hs578T cells and that this defect is exacerbated by loss of SVIL or ERK3; as this multipolar spindle defect likely arises of previous cell division failures, these cells were excluded from subsequent analyses. All other cells that entered mitosis formed a bipolar spindle that elongated normally at anaphase. While most mitoses completed with the formation of two distinct daughter cells, a fraction of the cells underwent cytokinetic failure, ending with a single, binucleated cell (Fig. 6A). In most of these cases, the cytokinetic furrow formed and initiated ingression normally but subsequently regressed. No furrow ingression was observed in a small proportion of mitotic cells. The frequency of cytokinetic failure significantly increased upon depletion of ERK3 or SVIL (Fig. 6A, B), indicating that both proteins impact cell ploidy via their regulatory effect on cytokinesis. As observed for cell ploidy, cytokinetic defects in ERK3and SVIL-depleted Hs578T cells were rescued by expressing either the wild type or S245D form of SVIL, but not the non-phosphorylatable S245A mutant (Fig. 6A, C, D). Importantly, measuring the total levels of GFP signal intensity revealed that there was no correlation between the expression levels of the different GFP-fused SVIL constructs and the extent of cytokinesis failure, arguing that the observed cytokinesis rescue arises of SVIL activity rather than level ( Fig. 6E). Altogether, these results indicate that phosphorylation of SVIL at Ser245 by ERK3 impacts Hs578T cell ploidy by ensuring proper mitotic progression and completion of cytokinesis.

Phosphorylation of SVIL on Ser245 is upregulated during mitosis
We sought to gain insight into the mechanism by which ERK3 phosphorylation of SVIL regulates cytokinesis completion. We first asked whether ERK3 signaling regulates the expression level of SVIL in the cell. No significant reduction of SVIL mRNA and protein levels was observed after RNAi depletion of ERK3 in Hs578T or MDA-MB-231 breast cancer cells ( Fig. S4A-D). In HeLa cells undergoing cytokinesis, SVIL was shown to localize at the leading edge of the ingressing furrow and become enriched at the midbody prior to abscission (Smith et al., , 2013. We used immunofluorescence analysis of Hs578T cells co-expressing epitope-tagged versions of ERK3 and SVIL to assess whether the two proteins colocalize at any time during mitotic progression. While no strong colocalization was observed, we found that the two ectopic proteins transiently accumulated at spindle poles in late anaphase, and SVIL localized at the midbody in telophase (Fig. 7A). Using antibodies specific to ERK3 and SVIL, we found that the two endogenous proteins wealy co-localized at the midbody in telophase (Fig.  7B). The small differences observed between these stainings may result from differences in epitope exposure, yet these results indicate that the two proteins can colocalize to discrete subcellular compartments during mitosis.
To determine whether SVIL is phosphorylated on Ser245 in a cell cycle-dependent manner, Hs578T cells were synchronized at different stages of the cell cycle and the phosphorylation of endogenous SVIL was monitored by immunoblotting. We observed that SVIL Ser245 phosphorylation is markedly upregulated in mitosis, concomitant to the accumulation of ERK3 and SVIL (Fig. 7C). These results are in agreement with our previous study showing the stabilization and accumulation of ERK3 in mitotic cells (Tanguay et al., 2010). Staining of SVIL-depleted Hs578T cells expressing wild type SVIL with the phospho-SVIL(S245) antibody revealed that this signal is enriched at spindle poles during anaphase and at the midbody in late telophase, while this signal is absent in cells reconstituted with SVIL S245A mutant (Fig. 7D). Together, these results support the notion that SVIL Ser245 phosphorylation by ERK3, possibly at or near the spindle pole in late anaphase, promotes the accumulation of phospho-SVIL at the ingressing furrow and midbody in telophase, where it contributes to coordinate the completion of cytokinesis.

ERK3 phosphorylation of SVIL Ser245 regulates activation of myosin II regulatory light chain during cytokinesis
Previous work demonstrated that SVIL regulates cytokinesis in part by controlling the activation of non-muscle myosin II at the cleavage furrow (Hasegawa et al., 2013). Myosin II activation is achieved by phosphorylation of its regulatory light chain (RLC) on Ser19, which is required for constriction of the contractile ring (Asano et al., 2009;Dean and Spudich, 2006;Komatsu et al., 2000). Thus, we hypothesized that the impact of SVIL phosphorylation by ERK3 on cytokinesis execution could be mediated through myosin II RLC phosphorylation and activation. To test this, we used immunofluorescence to monitor the levels of Ser19phosphorylated RLC in Hs578T cells undergoing mitosis. We found that Hs578T cells depleted of either ERK3 or SVIL have lower relative levels of myosin II phospho-RLC(S19) at the cell equator during anaphase as compared to control cells (Fig. 8A, B). Notably, the decreased accumulation of Ser19-phosphorylated RLC was rescued by re-expressing the wild type or S245D forms of SVIL, but not SVIL S245A (Fig. 8A, B). No correlation was observed between the expression levels of SVIL S245A rescuing construct and myosin II phospho-RLC(S19) levels in SVIL-depleted Hs578T cells (Fig. 8C). These results indicate that phosphorylation of SVIL at Ser245 contributes to maintaining high levels of Ser19-phosphorylated RLC at the cell equator, suggesting a mechanism by which phosphorylation of SVIL by ERK3 sustains myosin II activation at the cytokinetic ring.

DISCUSSION
The current study increases our understanding of the roles of the atypical MAP kinase ERK3 in cell physiology. Using a SILAC-based quantitative phosphoproteomics approach, we identified several candidate ERK3 substrates and downstream effectors in breast cancer epithelial cells, including the actomyosin scaffold protein SVIL. Using a combination of in vitro and in vivo assays, we confirmed that ERK3 interacts with and directly phosphorylates SVIL on Ser245. We further demonstrated that this phosphorylation event is functionally important for the sustained activation of non-muscle myosin II at the cleavage furrow and the proper completion of cytokinesis. These findings unveil a molecular mechanism by which the activity of ERK3 impacts cell division and ploidy.
Very little is known about the substrates and downstream effectors of the atypical MAP kinases ERK3 and ERK4, which has considerably hampered the understanding of their biological functions. No large-scale analysis of the ERK3 or ERK4 phosphoproteome has been reported to date. The first, and still only validated, substrate identified for ERK3 and ERK4 is the protein kinase MK5 (Seternes et al., 2004;Schumacher et al., 2004;Aberg et al., 2006;Kant et al., 2006). ERK3 was also reported to phosphorylate steroid receptor coactivator-3 (SRC-3) on Ser857, promoting its interaction with the ETS transcription factor PEA3 and gene expression (Long et al., 2012). However, Ser857 lies in a MK consensus phospho-acceptor motif and a subsequent study has shown that SRC-3 is instead phosphorylated at Ser857 by MK5 (Shrestha et al., 2020). Here, we have used a SILAC-based quantitative phosphoproteomics approach to discover novel substrates and effectors of ERK3. Along with SVIL, our analysis uncovered 37 proteins that show a significant decrease in phosphorylation upon depletion of ERK3 in Hs578T cells, considerably expanding our knowledge of this signaling network. One limitation of our study is that MS analysis was performed at steadystate in proliferating cells depleted of ERK3 by RNAi. Contrary to classical MAP kinases like ERK1/2, ERK3 displays constitutive kinase activity in cells and no upstream activator has been identified so far, which prevents the possibility of acutely stimulating the activity of the kinase.
Despite this inherent limitation, the following observations validate our approach and point to new areas of ERK3 biology research. First, PhosphoLogo analysis of ERK3-sensitive phosphoproteins revealed a strong preference for a proline residue at position +1 and -2, in agreement with the consensus phosphorylation motif of the MAP kinase family. Indeed, ERK3 and ERK4 phosphorylate MK5 at Thr182, which is followed by a Pro residue. Second, some ERK3-regulated phosphoproteins, such as septin family members, are already known to be regulated by ERK3-MK5 signaling (Brand et al., 2012;Shiryaev et al., 2012). Third, GO analysis of ERK3-regulated phospho-Ser/Thr-Pro proteins showed an enrichment in the cytoskeleton cellular component, consistent with the known functions of ERK3 in cellular adhesion, migration and invasion. The validation of SVIL as a bona fide ERK3 substrate confirms the predictive value of our analysis.
We identified Ser245, which lies in a full MAP kinase consensus motif, as a major SVIL phosphorylation site by ERK3. Notably, the phosphorylation of SVIL on Ser245 has been detected in 76 large-scale proteomic studies (www.phosphosite.org), but the kinase(s) involved in this regulation was not identified. We further showed that SVIL is phosphorylated by the paralogous kinase ERK4, but not by ERK1. ERK3 and ERK4 share a common C34 domain (Coulombe and Meloche, 2007), which is localized in the unique C-terminal extension and is absent from ERK1. The C-terminal extension of ERK3 is highly conserved in vertebrate evolution and was shown to mediate the interaction of ERK3 with its substrate MK5 (Åberg et al., 2009), as well as with other binding proteins such as cyclin D3, Cdc14A and septin 7 (Brand et al., 2012;Sun et al., 2006;Hansen et al., 2008). Similarly, we found that SVIL coimmunoprecipitates with the C-terminal extension of ERK3, which may contribute to the unique substrate selectivity of the atypical MAP kinases ERK3 and ERK4.
Our findings provide strong evidence that ERK3 and SVIL act in a linear pathway to regulate cytokinesis execution, unveiling a new cellular function of the poorly characterized MAP kinase ERK3 in cell division. We observed that Ser245-phosphorylated SVIL accumulates at the midbody, pointing to a role of ERK3 in controlling the localization of SVIL to coordinate cytokinesis execution. Mechanistically, we showed that phosphorylation of SVIL Ser245 by ERK3 is necessary to maintain high levels of Ser19-phosphorylated RLC at the cleavage furrow, suggesting that ERK3 signaling may regulate cytokinesis execution by promoting myosin II activation. Phosphorylation of RLC Ser19 by Rho-dependent kinases is required for the activation of myosin II at the cleavage furrow and the generation of inward contractile force driving furrow ingression and cell division (Matsumura, 2005). Further studies are needed to delineate the precise mechanism by which ERK3-mediated phosphorylation of SVIL is linked to myosin II activation at the cortex during cytokinesis. Moreover, we do not exclude the possibility that SVIL phosphorylation by ERK3 may impact other actin-driven cellular functions of SVIL.
Is is noteworthy that the loss of ERK3 did not increase the frequency of multinucleated cells in primary MEFs. One possible explanation is that ERK4, which is weakly expressed in MEFs (Rousseau et al., 2010), can compensate for the loss of ERK3. An alternative possibility is that the impact of ERK3 inactivation is dependent on the cellular context and becomes manifest in cancer cells, which exhibit a variety of genetic instability phenotypes and are prone to cell division errors. There is evidence that ERK3 signaling sustains the proliferation, anchorage-independent growth, and tumorigenesis of lung and breast cancer cells (Cai et al., 2021). Inactivation of ERK3 in some cancer cells may increase the rate of cell division errors leading to a decrease in cellular fitness and long-term cell proliferation by mechanisms that remain to be investigated.    Hs578T cells were treated with non-target (NT) or MAPK6 siRNAs for 24 h, and then transfected with Flag-SVIL. SVIL was immunoprecipitated with anti-Flag antibody and phosphorylation was analyzed by immunoblotting with anti-phospho-SVIL(S245) antibody.
Right, quantification of phosphorylation data (n=3). Data are expressed as mean ± SEM. **P < 0.01 by paired Student's t-test. ns, non significant.       Figure S1. ERK3 regulates cell ploidy through phosphorylation of SVIL on Ser245  Table S1. List of phosphorylation sites in Hs578T cells depleted or not of ERK3 Table S2. List of ERK3-regulated phosphosites in Hs578T cells

SUPPLEMENTARY MOVIES
Movie S1. Movie of control Hs578T cells expressing Citrine-a-Tubulin 1B (green) and H2B-mRFP (magenta). Images were captured at 8 min intervals and displayed at 4 frames per second.

!"#$%&'
(E) Effect of ERK3 depletion on the in vivo phosphorylation of SVIL. Hs578T cells were treated with non-target (NT) or MAPK6 siRNAs for 24 h, and then transfected with Flag-SVIL. SVIL was immunoprecipitated with anti-Flag antibody and phosphorylation was analyzed by immunoblotting with anti-phospho-SVIL(S245) antibody. Right, quantification of phosphorylation data (n=3). Data are expressed as mean ± SEM. **P < 0.01 by paired Student's t-test. ns, non significant.   Figure 6 A B  Figure 6. Phosphorylation of SVIL by ERK3 regulates cytokinesis (A) Cytokinesis failure in control, ERK3-and SVIL-depleted Hs578Tcells transfected or not with EGFP-SVIL wild type, S245A or S245D was quantified by time-lapse imaging of dividing cells. A total of 25-65 cells were scored for each condition (n>3 biological replicates). *P < 0.05, **P < 0.01 by Fisher's exact test. (B) Representative stills of movies S1-3 of control, ERK3-or SVIL-depleted Hs578T cells expressing Citrine-α-Tubulin 1B and H2B-mRFP. Time from anaphase onset (in min) is shown at the top left of each still. Scale bar = 10 µm. (C-D) Representative stills of movies of SVIL-depleted (C) or ERK3-depleted (D) Hs578T cells expressing EGFP-SVIL wild type, S245A or S245D forms of SVIL, and H2B-mRFP. Time from anaphase onset (in min) is shown at the top. Scale bar = 10 µm. (E) The expression level of EGFP-SVIL wild type, S245A or S245D forms in transfected control, ERK3-or SVIL-depleted Hs578T cells was quantified by measuring total GFP fluorescence intensity (a.u.) in individual cells and plotted as a function of the cell's ability to complete (success) or not (failure) cytokinesis. Pair-wise comparison (two-way ANOVA with Bonferonni post hoc test) between success or failure for the three different SVIL constructs is shown. ***P < 0.001. ns, non significant.