|Home | About | Journals | Submit | Contact Us | Français|
Snail1 is a major factor for epithelial-mesenchymal transition (EMT), an important event in tumor metastasis and in other pathologies. Snail1 is tightly regulated at transcriptional and posttranscriptional levels. Control of Snail1 protein stability and nuclear export by GSK3β phosphorylation is important for Snail1 functionality. Stabilization mechanisms independent of GSK3β have also been reported, including interaction with LOXL2 or regulation of the COP9 signalosome by inflammatory signals. To get further insights into the role of Snail1 phosphorylation, we have performed an in-depth analysis of in vivo human Snail1 phosphorylation combined with mutational studies. We identify new phosphorylation sites at serines 11, 82, and 92 and confirmed previously suggested phosphorylations at serine 104 and 107. Serines 11 and 92 participate in the control of Snail1 stability and positively regulate Snail1 repressive function and its interaction with mSin3A corepressor. Furthermore, serines 11 and 92 are required for Snail1-mediated EMT and cell viability, respectively. PKA and CK2 have been characterized as the main kinases responsible for in vitro Snail1 phosphorylation at serine 11 and 92, respectively. These results highlight serines 11 and 92 as new players in Snail1 regulation and suggest the participation of CK2 and PKA in the modulation of Snail1 functionality.
Epithelial-mesenchymal transition (EMT) is a process whereby cells lose cell–cell interactions and other epithelial properties while acquiring a more migratory and mesenchymal phenotype. EMT occurs at several stages of early development and an EMT-like process is thought to occur during tumor progression (Thiery, 2002 ; Gupta and Massagué, 2006 ; Moreno-Bueno et al., 2008 ; Yang and Weinberg, 2008 ). A hallmark of EMTs is the functional loss of E-cadherin (CDH1) which mediates cell-cell interactions. Snail1 is one of a family of transcription factors shown to repress transcription of the CDH1 gene (Cano et al., 2000 ; Batlle et al., 2000 ) and has been shown to be crucial to development in mice, because Snail1 knockout mutants die at gastrulation (Carver et al., 2001 ). Snail1 also regulates the expression of different target genes involved in EMT and other functions, such as cell survival (reviewed in Barrallo-Gimeno and Nieto, 2005 ). Importantly, Snail1 has been shown to be up-regulated in various tumor types (reviewed in Peinado et al., 2007 ) and has more recently been observed expressed at the tumor stromal interface (Francí et al., 2006 , 2009 ).
To date, several studies have examined the posttranslational modifications that control Snail1 cell function. Snail1 consists of four C-terminal DNA-binding zinc fingers and a regulatory region spanning amino acids 1–150 (Sefton et al., 1998 ), comprising an N-terminal SNAG (SNAIL and Gfi conserved) domain (1–9 amino acids) important for corepressor interaction (Hemavathy et al., 2000 ; Peinado et al., 2004 ), and the serine-rich domain (SRD), which appears to control the function of a short leucine-rich nuclear export sequence (NES) recognized by the CRM1 transporter (Domínguez et al., 2003 ; Zhou et al., 2004 ). Initial studies demonstrated that the SRD was important for subcellular localization and that mutation of all serines-to-alanine residues in that region resulted in cytoplasmic localization of Snail1 and consequent repressive dysfunction (Domínguez et al., 2003 ). Additionally, two separate investigations (Zhou et al., 2004 ; Yook et al., 2005 ) identified a short sequence in the SRD important for Snail1 posttranslational regulation. This short sequence is conserved in several proteins, including β-catenin and IKκB, and is termed destruction box (DB). The DB in the case of β-catenin contains two serines, the phosphorylation of which is critical for recognition and ubiquitination by the β-TrCP ubiquitin ligase (Nelson and Nusse, 2004 ). This leads to proteasomal degradation and thereby constitutes an important shared control mechanism for the government of several transcriptional regulators simultaneously. In Snail1 these serines lie at positions 96 and 100 (D96SGKG100SQPP), are evolutionary conserved, and are the target for the protein kinase glycogen synthase kinase 3-beta (GSK3β). Mutation of these serines to alanine leads to increased Snail1 stability (Zhou et al., 2004 ). Yook et al. also identified two additional conserved serines (QPP104SPP107SPAP) as targets for the serine/threonine protein kinase GSK3β. Based on those results two models were proposed to explain the regulation of Snail1 by GSK3β. Yook et al. (2005) proposed that phosphorylation of serines 104 and 107 would prime GSK3β phosphorylation of serines 96 and 100 in the DB, thus leading to ubiquitination and degradation of Snail1. In support of this model they found that GSK3β was one of the components of a regulatory complex, including Axin2 that facilitates the shuttling of GSK3β between nucleus and cytoplasm (Yook et al., 2006 ). Zhou et al. (2004) proposed a slightly different model whereby nuclear GSK3β phosphorylation of serines 107, 111, 115, and 119 prompts Snail1 nuclear export. Cytoplasmic Snail1 is then phosphorylated at serines 96 and 100 by GSK3β promoting ubiquitination and degradation. An additional proposed modulator of Snail1 was p21-activated kinase (PAK1), which may phosphorylate Snail1 at serine 246 (TF246SRM) to favor its nuclear localization (Yang et al., 2005 ). This serine is conserved in all the members of the Snail family and is located within the C-terminal zinc finger domain (Manzanares et al., 2001 ; Nieto, 2002 ). Posttranslational regulation of Snail1 stability by GSK3β-independent mechanisms has also been reported, including stabilization by lysyl oxidase like-2 (LOXL2) interaction (Peinado et al., 2005 ), mediated by TNF-α/NF-κB and COP9 signalosome (Wu et al., 2009a ), and cap-independent translational regulation by YB-1 factor (Evdokimova et al., 2009 ). On the other hand, interaction between Snail and the F-box protein PpA has also been implicated in Snail2 stabilization (Vernon and Labonne, 2006 ).
The phosphorylation of Snail1 in previous studies has been analyzed by mutation of the serine residues or by alterations of the mobility shift in Western blots, but not by in vivo analysis of Snail1 phosphorylation sites. In the present study, we have performed a complete analysis of Snail1 phosphorylation in vivo to clarify and complement previous investigations. We have identified several new sites of phosphorylation (S11, S82, and S92) with regulatory implications in Snail1 functionality as well as confirming some of those previously described (S104 and S107). We also propose that, in addition to GSK3β, the ubiquitous serine/threonine protein kinase CK2 (casein kinase-2) and the cAMP-activated kinase PKA (protein kinase A) may play a role in Snail1 function and regulation.
The pcDNA3-HA plasmid containing human SNAIL1 cDNA described previously (Peinado et al., 2005 ) has been used as a template for site-directed mutagenesis. Site-directed mutagenesis was performed as described (Zheng et al., 2004 ). After mutagenesis, the entire SNAIL1-HA was sequenced to verify that only the nucleotide changes introduced by the mutagenic oligonucleotide were obtained. Vectors for expression of Snail1 fused to the GST (glutathione S-transferase) protein were generated through PCR amplification of human Snail1 coding region from pcDNA3-Snail1-HA plasmid using oligonucleotides 5′-gcggatcccatgccgcgctctttc-3′ and 5′-gcgaattcagcggggacatcctgagc-3′ containing restriction sites for BamHI and EcoRI, respectively. The resulting product was digested and cloned into pGEX4T3 and sequenced to confirm identity. The mSin3A-myc-pcDNA3 expression vector was previously described (Peinado et al., 2005 ).
HEK293T and MDCK cells were maintained in DMEM medium supplemented with 10% fetal calf serum and antibiotics (100 μg/ml ampicillin, 32 μg/ml gentamicin, Sigma-Aldrich, St. Louis, MO). Stable and transient transfections were performed using Lipofectamine reagent (Invitrogen, Carlsbad, CA) according to manufacturer's instructions. For generation of stable clones MDCK cells (1 × 106 cells) grown in P100 plates were transfected with the indicated Snail1-HA wild-type, mutant constructs, or control pcDNA3 construct (cytomegalovirus [CMV]) and grown in the presence of G418 (500 μg/ml) for 2–3 wk; individual colonies were then selected and grown individually. At least 10 independent clones were selected and characterized from each transfection with the pcDNA3-Snail1-HA mutants. Two independent clones were additionally selected from pcDNA3-Snail1-HA wild-type or control transfections and compared with similarly isolated clones from our previous studies (Peinado et al., 2005 ).
The two-hybrid screen was performed as previously described (Peinado et al., 2005 ), using the Matchmaker system 3 (Clontech, Palo Alto, CA). The bait protein was the N-terminus of mouse Snail1 (residues 1-150). A fibroblast NIH3T3 library in pACT2 vector (Clontech) was used. Positive colonies were isolated based on their capacity to express the markers ADE2, HIS3, and MEL1.
Human embryonic kidney HEK293T cells were grown in 30-mm-diameter plates and transfected with pcDNA3-Snail1-HA wild-type or the indicated pcDNA3-Snail1-HA mutants. After 24–36 h, cells were washed three times with DMEM phosphate-free medium, supplemented with 10% fetal calf serum, and subsequently incubated in the same medium containing 1 mCi/ml [32P]orthophosphate (Amersham, Piscataway, NJ) for 4 h. Cells were lysed with ice-cold RIPA buffer (0.1% SDS, 0.5% sodium deoxicholate, 1% NP-40, 150 mM NaCl, 50 mM Tris-HCl, pH 8.0) containing protease (2 μg/ml aprotinin, 1 μg/ml leupeptin, 1 mM PMSF) and phosphatase inhibitors (1 mM sodium vanadate, 10 mM sodium fluoride), the resulting lysate centrifuged at 14,000 × g for 15 min, and the supernatant was incubated overnight at 4°C with 1 μg anti-HA and 25 μl of protein G-Sepharose (Amersham). Immunoprecipitates were collected and washed four times with ice-cold RIPA buffer. Precipitated proteins were resolved by SDS-PAGE and transferred to nitrocellulose membrane before exposure to x-ray film overnight to view radioactively labeled proteins.
Phosphopeptide mapping was performed as described (Boley et al., 1991 ; Van Der Geer and Hunter, 1994 ). Briefly, phosphorylated Snail1 bands were excised from the membrane and blocked for 30 min in 0.5% polyvinylpirrolidon K30 (Sigma-Aldrich), 0.6% acetic acid at 37°C. After several washes with distilled water, bands were each incubated overnight in 50 mM NH4HCO3, pH 8.0, with 1 μg modified sequencing grade trypsin (Promega, Madison, WI) at 37°C. The solution containing eluted peptides was transferred to a fresh tube and the radioactivity measured to ensure efficient elution. The peptides were then frozen and dried by vacuum evaporation. They were then oxidized in performic acid for 1 h on ice in the dark. The oxidization was stopped through dilution in water, the solution was frozen, and vacuum evaporated. A further overnight digestion with trypsin was required to ensure efficient trypsin digestion. The peptides were then washed twice in water, with vacuum-evaporating in between, and then were resuspended in 200 μl electrophoresis running buffer (2.5% formic acid, 7.5% acetic acid, pH 1.9), centrifuged 15 min at 15,000 × g, to remove remaining insoluble material. The supernatant was then dried and resuspended in 10 μl electrophoresis running buffer and carefully spotted onto a 20 × 20-cm cellulose TLC plate (TLC plates cellulose were precoated, without fluorescent indicator and with layer thickness 0.1 mm; cat. no. 5716, Merck, Rahway, NJ). Electrophoresis was performed using HTLE-700 high-voltage electrophoresis apparatus (CBS Scientific, Del Mar, CA) for 30 min at 2000 V. The plates were then allowed to dry for a minimum of 6 h before chromatography. Separation of peptides by chromatography was performed for 14–16 h in isobutyric acid buffer (62.5% isobutyric acid, 3% acetic acid, 2% n-butanol, 5% pyridine). After drying for at least 6 h, the plates were exposed to phosphoimaging screen (Fujifilm, Tokyo, Japan). The results were observed using a Typhoon variable mode imager (GE Healthcare, Waukesha, WI) or Fuji FLA-3000 PhosphorImager (Fuji) and the images adjusted using Adobe Photoshop version 7.0 (San Jose, CA).
Phosphopeptides were eluted from the plates in pH 1.9 electrophoresis buffer and lyophilyzed. The fractions were then subjected to automated Edmann degradation. For Edmann degradation, phosphopeptides (40–500 cpm) were coupled to Sequelon-AA membranes (Millipore, Bedford, MA) according to the manufacturer's instructions and sequenced on an Applied Biosystems Gas Phase Sequencer model 470A (Foster City, CA). Released phenylthiohydantoin amino acid derivatives from each cycle were spotted onto TLC plates. The radioactivity in each spot was quantitated by exposure to a screen and scanning in a FLA-300 PhosphorImager (Fuji).
In vitro kinases assays were performed by incubation of 1 μg of GST-Snail1 fusion protein or GST alone with 10 U of the stated kinase in the supplied buffer, supplemented with 100 μM ATP and 1 μCi [γ-32P]ATP. Reactions were left for 30 min at 30°C and stopped by addition of 2× Laemmli buffer. Proteins were resolved by 10% PAGE-SDS electrophoresis and transferred to nitrocellulose membranes, which were then exposed to x-ray film. Human recombinant CK2 holoenzyme and human PKA catalytic subunit were purchased from Calbiochem (Darmstadt, Germany; cat. no. 218701) and Sigma-Aldrich (cat. no. C8482), respectively.
Coimmunoprecipitation analyses were performed as described previously (Peinado et al., 2004 ). Basically, HEK293T cells were transiently transfected with the indicated vectors for 48 h. Lysates were then obtained in immunoprecipitation buffer (50 mM Tris-HCl, pH 8.0, 150 mM NaCl, 5 mM EDTA, 0.5% NP-40) containing protease and phosphatase inhibitors (2 μg/ml aprotinin, 1 μg/ml leupeptin, 1 mM PMSF, 1 mM sodium vanadate, 10 mM sodium fluoride) and precleared with Sepharose G-beads. Supernatants were subjected to overnight incubation with anti-HA affinity matrix (Roche, Indianapolis, IN) or Sepharose G-beads coated with anti-rat IgG as immunoprecipitation control. Immunoprecipitates were resolved in 7.5–12% PAGE-SDS gels, transferred to membranes, and incubated with the indicated antibodies. Membranes were finally developed using ECL reagent following manufacturer's instructions (Amersham). Blots were incubated with rat anti-HA (Roche; 1:400), or anti-myc (Santa Cruz Biotechnology, Santa Cruz, CA; 1:200). For characterization studies, whole cell extracts were fractionated in SDS-PAGE gels and subjected to Western blot analyses with anti-E-cadherin, anti-HA, and anti-α-tubulin as described (Cano et al., 2000 ; Peinado et al., 2005 ). When indicated, MAb anti-Snail1 EC3 (1:100; Francí et al., 2006 ) was used. The secondary antibodies used were HRP-coupled sheep anti-mouse (1:1000), goat anti-rabbit (1:4000; Amersham), or anti-rat (1:10,000; Pierce, Rockford, IL).
Snail1-HA wild-type or the indicated Snail1-HA mutant constructs were transiently transfected into HEK293T cells, and 24 h later cells were treated with 20 mM cycloheximide (Sigma-Aldrich) for the indicated time intervals. Cells were lysed in RIPA buffer, containing protease and phosphatase inhibitors (2 μg/ml aprotinin, 1 μg/ml leupeptin, 1 mM PMSF, 1 mM sodium vanadate, 10 mM sodium fluoride), and the expression of Snail1 was analyzed by Western blotting using anti-HA antibody (Roche) as above, using α-tubulin as loading control.
Immunofluorescence analysis for E-cadherin, vimentin, and Snail1-HA were performed on cells grown on glass coverslips and fixed in cold methanol as previously described (Cano et al., 2000 ; Peinado et al., 2005 ). The secondary antibodies used were anti-rat Alexa594 and anti-mouse Alexa488 (1:800; Molecular Probes, Eugene, OR). Preparations were visualized in a Nikon N90i microscope (Melville, NY) equipped with epifluorescence.
HEK293T cells transiently transfected with pcDNA-CK2α-HA were lysed in RIPA buffer after a 48-h expression period. Lysates were then incubated with 10 μg of recombinant GST-Snail1 or GST control protein coupled to 50 μl glutathione-Sepharose for 1 h at 4°C. The beads were then washed four times for 15 min each in 1 ml RIPA buffer containing 0.5 M NaCl and then resuspended in 2× Laemmli buffer. The samples were then boiled for 5 min before be analyzed in 10% PAGE-SDS gels to detect associated CK2α with anti-HA antibodies (Roche).
The human and mouse E-cadherin promoters (−178 to +92 in both cases) fused to luciferase were used to determine the activity of the E-cadherin promoter as described previously (Peinado et al., 2004 , 2005 ). The human claudin-1 promoter (−748 to +252) fused to luciferase was also used as previously described (Martinez-Estrada et al., 2006 ). Cotransfections were carried out in the presence of the indicated amounts of Snail1-HA or Snail1-HA mutant cDNAs, all cloned in the pcDNA3 vector. The amount of total transfected DNA was normalized with empty pcDNA3 vector. Luciferase and renilla activities were measured using the dual luciferase reporter assay kit (Promega) and normalized to the wild-type promoter activity in mock (pcDNA3) transfected cells, as previously described (Peinado et al., 2004 , 2005 ).
Although analysis of in vivo Snail1 phosphorylation had previously been carried out (Domínguez et al., 2003 ), detailed analyses of specific sites had not. It had been demonstrated that the majority of Snail1 phosphorylation took place on serine residues and that although many sites resided within the serine rich domain, others lay outside this region. Therefore, to obtain additional information we undertook phosphopeptide mapping of in vivo phosphorylated human Snail1 after being transiently expressed in HEK293T cells.
Phosphorylated Snail1 appeared as one main band with further fainter bands also apparent in the parallel Western blot control, possibly due to partial degradation (Figure 1A). The main phosphorylated band was excised and subjected to trypsin digestion, and the resulting peptides separated in two-dimensions by electrophoresis and chromatography. The resultant phosphopeptide map showed a minimum of eight distinct phosphopeptides, suggesting several phosphorylation sites in Snail1 (Figure 1B).
Each phosphopeptide was extracted from the cellulose plate and subjected to Edmann's degradation with the resultant residues spotted on TLC cellulose plates (Figure 1C). This analysis indicated that phosphopeptide 1 was phosphorylated on the 3rd residue, phosphopeptides 2 and 4 on the 8th residue, phosphopeptide 3 on the 4th residue, and phosphopeptides 5 and 6 on the 6th and 9th residues (Figure 1C). Phosphopeptides 7 and 8 could not be sufficiently eluted to allow reliable sequencing (data not shown). This analysis suggested multiple possible phosphorylation sites (Figure 1D) when compared with the all potential predicted peptides produced by trypsin digest of Snail1 protein as shown in Figure 1E.
To confirm the specific phosphorylation sites in Snail1 for each possible phosphorylation site, serine-to-alanine mutants were generated by site directed mutagenesis of the appropriate residues. The resulting S-to-A mutants were compared with the wild-type Snail1 protein for differences in the in vivo phosphorylation pattern. The disappearance of a particular spot in the S-to-A mutant relative to wild type being taken as confirmation of serine phosphorylation at this site. For each of phosphopeptides 1–6, a corresponding S-to-A mutant was found lacking the corresponding phosphopeptide.
Peptide 1 represented phosphorylation at serine 11 of Snail1 (Figure 2B, cf. Figure 2A) adjacent to the SNAG domain (Figure 1E) and proximal to lysine residues thought to be involved in nuclear entry (Ko et al., 2007 ).
Phosphopeptide 2 (Figure 1C) represented phosphorylation at serine 92 at the N-terminal of the SRD (Figure 2D). There appeared to be no additional phosphorylations on this peptide possibly expected at serine 96 (Figure 2D).
Phosphopeptide 3 represents phosphorylation at serine 82 at the N-terminal of the SRD (Figure 2C). Interestingly, phosphopeptide 3 is not detected in mutant S92A (Figure 2D), suggesting that the phosphorylation at serine 92 is required for phosphorylation at serine 82. Phosphopeptide 4 was weakly detected in most Snail1-HA mutants, making its identification difficult. However, its Edmann′s degradation pattern was similar to that of peptide 2 (Figure 1C), and its complete disappearance in S11A and S92A mutants suggests it may correspond to a miscut or other posttranslational modifications.
Phosphopeptides 5 and 6 both represent phosphorylations at serines 104 and 107 (Figure 2E). Indeed, there appear to be several peptide species with high hydrophobicity and low electrophoretic mobility. This group of phosphopeptides was poorly resolved in most of the analyzed mutants (see Figure 2, B–F), and these are thought to represent various modes of posttranslational modification of peptide 99-137, in particular affecting serine 104 and serine 107, because they disappear in the S104A/S107A mutants (Figure 2E). Peptide 99-137 exhibits no charge and a high hydrophobicity (aprox. 0.61), and phosphorylation of one to three serines in the peptide would have a very weak impact on migration in chromatography, due to the considerable length of the peptide.
Of note was the absence of peptides phosphorylated at serines 96 or 100, the two serines within the DB motif, proposed previously to be phosphorylated by GSK3β (Zhou et al., 2004 ; Yook et al., 2005 ). One possible explanation may be that subsequent ubiquitination and degradation of phospho-serine 96/100 protein occurs so quickly as to prevent attainment of these phosphopeptides. Although not observed, it is important not to discount other possible phosphorylations within the SRD as the large size of this peptide and progressive weakness of signal obtained from Edmann's sequencing may have obscured further downstream phosphorylations occurring at serines 111, 115, and 119.
Interestingly, the triple mutation of serines 92, 104, and 107 almost completely ablated phosphopeptides expected to represent phosphorylation in the SRD (Figure 2F). Therefore, it appears that either further phosphorylation in this region does not occur or that such phosphorylations are principally dependent on phosphorylations of serines 92, 104, and 107.
It was previously reported that Snail1 is phosphorylated at serine 246 by PAK1 (Yang et al., 2005 ). However, we were unable to detect any potential phosphopeptide corresponding to TF246SR, even under shorter electrophoretic run (to prevent the loss of high mobility peptides; Supplemental Figure S1). Furthermore, analysis of in vivo phosphorylation of S246A mutation did not reveal any change in its phosphorylation pattern compared with wild-type Snail1 (Supplemental Figure S2), suggesting that no phosphorylation occurs at serine 246 in our experimental system. It is possible, however, that there was insufficient PAK1 activity in HEK293T cells, yet because this phosphorylation was deemed critical for Snail1 function, we would expect to observe it in wild-type Snail1. The only phosphopeptide corresponding to serine 246 phosphorylation would coincide with phosphopeptide 1 after trypsin digestion (Figure 1D). As mentioned above, phosphopeptide 1 disappears in S11A mutant (Figure 2B), but was unaffected in S246A mutant (Supplemental Figure S2).
Also of note is that, with the exception of Snail1S92A mutation effecting phosphorylation at serine 82 (Figure 2D), the ablation of the various serines did not have significant effect on other phosphorylations, suggesting they occur independently of each other, although effects on global phosphorylation level cannot be discounted.
Overall, the present results indicate that Snail1 is phosphorylated in vivo at positions 11, 82, 92, 104, and 107 and that although other sites may be present, they are less frequently phosphorylated.
To identify possible serine/threonine kinase candidates for the phosphorylated serines in Snail1 we carried out in silico analysis of Snail1 protein using NetphosK 1.0 server (http://www.cbs.dtu.dk/services/NetPhosK).
A probability score of 0.7 was given for PKA at serine 11 and 0.54 for CK2 at serine 92. For serines 104 and 107 the predicted kinase was GSK3β, and extensive investigations into the phosphorylation of these sites by this kinase have already been carried out (Zhou et al., 2004 ; Yook et al., 2005 , 2006 ). There was no clear candidate for serine 82 phosphorylation and this will not be discussed further.
Therefore, we carried out in vitro phosphorylation reactions with PKA and CK2 on wild-type GST-Snail1 and the respective mutants, GST-Snail1S11A and GST-Snail1S92A. In the case of CK2, GST-Snail1 is readily phosphorylated by CK2 in vitro with relatively little phosphorylation of the GST-Snail1S92A mutant (Figure 3A). Phosphopeptide analysis identified a major spot from wild-type GST-Snail1 in the in vitro reaction of approximately the same mobility as that representing serine 92 phosphorylation in vivo (phosphopeptide 2; Figure 3B, left). In fact, Edmann's degradation of the main CK2 in vitro phosphopeptide rendered the same pattern as for in vivo phosphopeptide #2 (Figure 3C, cf. Figure 1C). Furthermore, this major spot was absent in the phosphopeptide mapping of Snail1S92A mutant (Figure 3B, right), confirming serine 92 as the main in vitro substrate of CK2.
In the case of PKA, a significant in vitro phosphorylation of wild-type GST-Snail1 was detected that was almost lacking in the GST-Snail1S11A mutant (Figure 4A). Furthermore, phosphopeptide analysis of wild-type GST-Snail1 in vitro phosphorylated by PKA identified a phosphorylated peptide of approximately the same mobility as that representing in vivo phosphorylation of serine 11 (phosphopeptide 1; Figure 4B, left) that was missing in the GST-Snail1S11A mutant (Figure 4B, right). Therefore, these results support serine 11 as the main in vitro phosphorylation site of PKA.
Interestingly, a previous two-hybrid screening performed in our group to detect Snail1 interacting proteins identified several potential partners of Snail1, including LOXL2/3 proteins (Peinado et al., 2005 ) and CK2α. Using the N-terminal regulatory domain of Snail1 (1–150) as bait, CK2α was confirmed as a positive interactor in the yeast system (Figure 5A). CK2 holoenzyme exists as a heterodimer consisting of two alpha and two beta subunits. CK2α is one of two possible alpha subunits, the other being CK2α′. To confirm the Snail1/CK2α interaction, pulldown assays were carried out using GST-Snail1 protein as bait and lysates from HEK293T cells transiently expressing CK2α-HA. We found that GST-Snail1 indeed interacts with CK2α-HA, whereas GST alone did not (Figure 5B). In vivo interaction was examined by coimmunoprecipitation analysis, after transient cotransfection of Flag-tagged Snail1 and CK2α-HA, but the results were negative, suggesting that the interaction in vivo is weak or transient in nature.
Overall, the evidence of in vivo and in vitro phosphorylation of Snail1 on the same residue and the yeast two-hybrid results strongly suggest that CK2 interacts with and may phosphorylate Snail1 at serine 92 in vivo.
To gain insight into the contribution of serine 11 and serine 92 in Snail1 functionality, the activity of the Snail1S11A and Snail1S92A mutants on the E-cadherin promoter was analyzed. Both mutants exhibited strongly decreased repressor activity (~20%) compared with the wild-type Snail1 protein on both the human (Figure 6A) or mouse E-cadherin promoter (Supplemental Figure S3A), indicating that serine 11 and serine 92 residues are required for an efficient Snail1 transcriptional repression on E-cadherin. Additional studies of other Snail1 target genes, such as claudin-1 (Martinez-Estrada et al., 2006 ), showed that the Snail1S11A and SnailS92A mutants also exhibited decreased repressor activity on claudin-1 promoter, although a milder effect was observed for the Snail1S92A mutant activity compared with Snail1 wild type (Supplemental Figure S3B), suggesting an effective derepression action of the Snail1 mutants on additional target genes, apart from E-cadherin.
Snail1 mediated E-cadherin repression depends on the SNAG domain and requires the recruitment of the mSin3A/HDAC1/2 corepressor complex (Peinado et al., 2004 ). Because serine 11 is located adjacent to the SNAG domain (Figure 1D), the interaction of Snail1S11A with the mSin3A corepressor was then studied by coimmunoprecipitation assays on HEK293T cells transiently transfected with Snail1-HA and mSin3A-myc. Results indicate that the Snail1S11A mutant has almost completely lost the interaction with mSin3A (Figure 6B, top middle panels). Interaction analysis of mSin3A corepressor with the Snail1S92A mutant also indicated a markedly reduced interaction with mSin3A compared with wild-type Snail1 (Figure 6B, top right panels). On the other hand, no significant changes in the nuclear localization of Snail1S11A or Snail1S92A mutants were observed (data not shown). These results indicate that phosphorylation of serine 11 or serine 92 does not affect Snail1 nucleo-cytoplasmic traffic machinery, yet positively influences the interaction of Snail1 with mSin3A corepressor, thus contributing to its transcriptional repression.
As the importance of specific serines in the SRD to Snail1 stability had been previously shown, we additionally examined the stability of Snail1S11A and Snail1S92A mutants, in comparison with wild-type Snail1, when transiently transfected in HEK293T cells. We found that the Snail1S92A mutation led to greatly increased stability compared with wild-type Snail1 (Figure 7, A and B), suggesting that phosphorylation at serine 92 forms part of the Snail1 degradation mechanism. Similar mutations at serine 96, 104, and 107 had previously been shown to increase Snail1 stability (Zhou et al., 2004 ; Yook et al., 2005 ) and, therefore, it appears that several phosphorylation sites inside the DB motif, or at its proximity, are involved in Snail1 degradation. In agreement with this, the double Snail1S104A/S107A mutant shows increased protein stability (Figure 7, C and D) and exhibits a repressive activity on the human E-cadherin promoter similar to that of Snail1 wild type in HEK293T cells (Supplemental Figure S4). In contrast to this behavior, the increased stability of the Snail1S92A mutant is not sufficient to maintain an efficient E-cadherin promoter repression (Figure 6A and Supplemental Figures S3–S5). Significantly, the phosphorylation mimicking Snail1S92E mutant rescues the repressor activity on E-cadherin promoter to levels similar to the Snail1 wild type (Supplemental Figure S5), indicating that electrostatic modification of serine 92 produced by phosphorylation is required for efficient Snail1-mediated repression. Interestingly, the Snail1S11A mutant also showed a marked increased stability compared with wild-type Snail1 (Figure 7, C and D), despite its strongly decreased repressor activity (Figure 6A).
To further explore whether the Snail1S11A and Snail1S92A mutations have any in vivo consequence we evaluated the competence of the Snail1 mutants to achieve EMT. To this end, MDCK cells were stably transfected with human Snail1-HA wild-type, Snail1S11A-HA, or Snail1S92A-HA versions. MDCK cells expressing Snail1-HA suffered EMT (80% of the clones) with complete loss of E-cadherin and increased expression and organization of vimentin (Figure 8, A, b, f, and j, and B) in agreement with previous observations (Peinado et al., 2005 ). In contrast, MDCK cells expressing the Snail1S11A-HA mutant exhibited an unaltered epithelial phenotype (75% of the clones) similar to that of the mock-control cells (Figure 8A, cf. c and d with a) and maintain the expression of E-cadherin (Figure 8B) organized in cell–cell junctions (Figure 8A, cf. g and h, with e) and basal vimentin levels (Figure 8A, cf. k and l with i). Importantly, the majority of MCDK-Snail1S11A-HA clones express the mutant Snail1S11A protein to levels similar to those of Snail1-HA wild type present in MDCK-Snail1 suffering EMT (Figure 8C). Strikingly, repeated attempts to stably express the Snail1S92A-HA mutant were unsuccessful or resulted in a very low number of stable colonies, suggesting that mutation of serine 92 has a deleterious effect on survival of MDCK cells. Indeed, analysis of the isolated clones indicated that although most of them maintain the epithelial phenotype with absence of EMT features (Supplemental Figure S6, A and B), they express almost undetectable levels of Snail1-S92A-HA protein even lower of basal endogenous Snail1 (Supplemental Figure S6C). The basal nonfunctional level of endogenous Snail1 detected in control MDCK-CMV cells is in agreement with our previous observations (Peinado et al., 2005 ). These results reinforce the functional requirement of intact S11 and S92 residues for Snail1 function in EMT and cell survival, respectively.
Taken together, these results strongly suggest that phosphorylation of serine 92 and serine 11 positively regulates Snail1functionality and point to the participation of CK2 and PKA, respectively, in those phosphorylation events.
The role of Snail1 in development and tumor invasion is well established and understanding the regulatory mechanisms controlling Snail1 functionality will facilitate the development of new strategies and targets for cancer treatment. To date, several studies had examined the control of Snail1 through phosphorylation, supporting the action of negative (GSK3β) and positive (PAK1) kinase players (Zhou et al., 2004 ; Yang et al., 2005 ; Yook et al., 2005 ). Recently, identification of the small C-terminal domain phosphatase (SCP) as responsible for dephosphorylation of Snail1-GSK3β phosphorylated has added a new level of regulation (Wu et al., 2009b ). An alternative mechanism for Snail1 stabilization by TNF-α/NF-κB pathway independent of GSK3β phosphorylation has been recently described (Wu et al., 2009a ), indicating the existence of several pathways to control Snail1 stability in response to different cues. However, in none of these studies was there a detailed analysis of in vivo Snail1 phosphorylation and its influence in Snail1 repressor and functional activity. Here, we presented a detailed analysis of human Snail1 in vivo phosphorylation. Phosphopeptide and mutational analysis demonstrate that Snail1 serines 11, 82, 92, 104, and 107 were phosphorylated in vivo. Importantly, phosphorylations of S11 and S92 are independently required for Snail1 repressor activity of E-cadherin promoter and for efficient recruitment of the mSin3A corepressor. Furthermore, intact serine 11 and serine 92 are required for in vivo Snail1 function in induction of EMT and/or cell survival, respectively.
It had been initially shown, through in vivo phosphoamino acid analysis that the majority of Snail1 phosphorylations occurred on serine residues (Domínguez et al., 2003 ). Furthermore, those previous studies demonstrated a marked reduction in phosphorylation of Snail1 mutants generated either by mutating all serines to alanine between residues 82-123 or by deleting this region entirely (Domínguez et al., 2003 ). Our present results indicate that this is due to the lack of phosphorylation at serines 82, 92, 104, and 107, and the remaining phosphorylation observed in those mutants would be attributed to serine 11. Two other investigations have previously examined Snail1 SRD phosphorylation (Zhou et al., 2004 ; Yook et al., 2005 ), demonstrating a role for GSK3β in SRD phosphorylation and regulation of Snail1 stability, through slightly different models.
Notably, Yook et al. (2005) found serines 104 and 107 to be the only double mutant with significant effect on Snail1 stability, emphasizing the importance of these two residues. In agreement with those previous observations, we found both serines 104 and 107 to be phosphorylated together in vivo. Zhou and colleagues proposed a slightly different model where GSK3β initially phosphorylates Snail1 in the nucleus at serines 119, 115, 111, and 107, promoting its nuclear export to the cytoplasm where GSK3β further phosphorylates the DB serines (S96 and S100), eliciting ubiquitination and degradation. However, we observed serines 104 and 107 phosphorylated together without detectable phosphorylation of serines 111, 115, and 119. Furthermore, mutations of Snail1 at serines 111, 115, and 119 did not result in obvious changes in the in vivo phosphorylation pattern or abrogation of the phosphopeptides corresponding to S104 and S107, suggesting that these are not “priming” phosphorylations (data not shown). Overall, our in vivo data tend to support the model originally proposed by Yook et al. (2005) , favoring that GSK3β probably phosphorylates serines 104 and 107 and then 96 and 100 to promote degradation. However, we should not completely discount the possibility of phosphorylations at serines 111, 115, and 119 because the resolution of the larger peptide species containing the SRD was poor and sequencing may have missed phosphorylations so distant from the N-termini of the peptide. Surprisingly, we could not detect in vivo phosphorylation at serines 96 and 100, although we suspect that the ubiquitination and degradation of this Snail1 species prevented observance of these phosphorylations.
The serine at position 82 appears more cryptic and is only weakly phosphorylated relative to other serines in Snail1. In silico analysis did not reveal any likely kinase candidates although GSK3β and p38MAPK had the highest probability scores of 0.48. Despite this, phosphorylation at serine 82 appears to be influenced by that on serine 92, because it completely disappears in mutant Snail1S92A, perhaps by promoting a conformational change. The functional significance of serine 82 phosphorylation remains to be investigated.
Also of interest, was the lack of detection in our analysis of in vivo phosphorylation at serine 246 previously speculated to be critical for Snail1 function (Yang et al., 2005 ). In our hands, mutation at this site resulted in no effect on Snail1's phosphorylation pattern, cellular localization, or function (Supplemental Figures S1 and S2; data not shown). It may be that the effects observed previously in this mutant are due to secondary effects present in particular cellular contexts, due to changes in zinc finger binding or on nuclear import sequences lying nearby (Mingot et al., 2009 ).
Our present study provides evidence for new regulatory phosphorylation events in Snail1 at serine 92 and at serine 11 with consequences in Snail1 stability and, more importantly, in Snail1-mediated repression and functional activity. Interestingly, phosphorylation of serine 92 does not appear to prime or stimulate phosphorylation in the SRD region, because no change in the phosphorylation pattern in this region was observed in the Snail1S92A mutant. However, we observed that, despite its increased stability, the Snail1S92A mutant has strongly decreased repressor activity on the E-cadherin promoter relative to Snail1 wild type. Importantly, this was associated to a marked decreased interaction with mSin3A corepressor. These observations indicate that phosphorylation at serine 92 positively regulates Snail1 transcriptional repression by a mechanism dependent on efficient recruitment of mSin3A corepressor.
We also observed phosphorylation at serine 11, adjacent to the SNAG domain known to be critical for Snail1 repressive action and interaction with corepressors (Hemavathy et al., 2002; Peinado et al., 2004 ). Ablation of S11 phosphorylation, despite inducing increased stability, strongly attenuated Snail1 repression on E-cadherin and claudin-1 promoters, suggesting that serine 11 phosphorylation is necessary for correct interaction with corepressors. Indeed almost complete abrogation of interaction with mSin3A was detected in the Snail1S11A mutant. Mutation of this site did not result in any change in subcellular localization despite its proximity to lysines (K9, K16) identified as part of a nuclear import signal (Ko et al., 2007 ), although recent studies also indicate the participation of residues in the zinc finger region for efficient Snail1 nuclear import (Mingot et al., 2009 ). Interestingly, the in vitro phosphorylation studies identified serine 92 and serine 11 as the main phosphorylated substrates of CK2 and PKA, respectively, suggesting that both kinases may be involved in Snail1 regulation in vivo by directly modulating Snail1-mSin3A interaction.
The direct implication of serine 11 in Snail1 activity has been demonstrated in functional transfection studies, because mutation of serine 11 abrogates the EMT inducing activity of Snail1, indicating that intact serine 11 is required for Snail1 to drive the EMT program, affecting not only E-cadherin repression but the regulation of other Snail1 target genes, like claudin-1. Mutation of serine 92 in Snail1 also results in abrogation of the EMT program and, strikingly, in cell lethality, reinforcing the functional requirement of serine 92 for EMT and additional Snail1 functions as cell survival (Vega et al., 2004 ), although further studies are required to characterize the involvement of serine 92 in Snail1-mediated regulation of cell survival.
The results presented here strongly support that phosphorylation of Snail1 at serine 11 and serine 92, likely mediated by PKA and CK2, respectively, emerge as new positive regulatory pathways of Snail1 repressor activity both influencing mSin3A corepressor recruitment and functional activity. In this context, it is interesting to note that β-catenin is also regulated by CK2 and PKA phosphorylation, both directly and through its interaction with APC (Song et al., 2003 ; Hildesheim et al., 2005 ; Taurin et al., 2006 ). Regulation of the tumor suppressor PML by CK2 has also been shown (Scaglioni et al., 2006 ; Stehmeier and Muller, 2009 ). The positive regulation of Snail1 function by CK2 proposed here, provides new mechanistic aspects to support the participation of this kinase in tumorigenesis.
Taken together, the present results reinforce the existence of a complex regulatory network in Snail1 posttranslational regulation involving activating and inhibitory phosphorylation mechanisms and putting into the stage new players as serines 11 and 92 and, potentially, CK2 and PKA kinases. These observations, together with recently reported alternative mechanisms for Snail1 stabilization (Evdokimova et al., 2009 ; Wu et al., 2009a ), provide potential novel targets to influence Snail1 function in tumor progression.
The authors thank Ismo Virtanen (University of Helsinki) for generous gift of Snail1 antibodies, Ihor Yakymovych for valuable help with 2D phosphopeptide mapping, Vanesa Santos for generation of stable clones, and members of A. Cano lab for helpful discussions. This work was supported by the EU and the Spanish Ministry of Science and Innovation (MRTN-CT-2004-005428, SAF2007-06351, Consolider Ingenio CDS2007-00017 grants to AC) and Fundación Mutua Madrileña. M.McP. was supported by the EU Marie Curie program, P.O. is supported by a FPI fellowship from the Spanish Ministry of Science and Innovation.
This article was published online ahead of print in MBC in Press (http://www.molbiolcell.org/cgi/doi/10.1091/mbc.E09-06-0504) on November 18, 2009.