The E-cadherin/β-catenin signaling pathway plays a critical role in the maintenance of epithelial architecture and regulation of tumor progression. Normally, E-cadherin locates on the cell surface with its cytosolic domain linking to the actin cytoskeleton through interaction with catenins. Although the nuclear localization of E-cadherin has been frequently observed in various types of cancers, little is known regarding the functional consequences of its nuclear translocation. Here, we showed that in colorectal cancer samples and cell lines, E-cadherin localized in the nucleus; and the nuclear localization was mediated through protein interaction with CTNND1. In the nucleus, E-cadherin was acetylated by CREB-binding protein at Lysine870 and Lysine871 in its β-catenin–binding domain, and the acetylation can be reversed by SIRT2. Acetylation of nuclear E-cadherin attenuated its interaction with β-catenin, which therefore released β-catenin from the complex, resulting in increased expression of its downstream genes and accelerated tumor growth and migration. Further study showed that acetylation level of nuclear E-cadherin had high prognostic significance in clinical colorectal samples. Taken together, our findings reveal a novel mechanism of tumor progression through posttranslational modification of E-cadherin, which may serve as a potential drug target of tumor therapy.
This finding that acetylation of nuclear E-cadherin regulates β-catenin activity expands our understanding of the acetylation of E-cadherin promotes colorectal cancer cell growth and suggests novel therapeutic approaches of targeting acetylation in tumors.
E-cadherin is a well-known homotypic adhesion molecule that mainly expresses in epithelial cells and serves as a suppressor of tumor growth and metastasis (1, 2). In adherent junctions, E-cadherin stabilizes the surfaces of neighboring cells by forming adhesive complexes in which the first cadherin domain (extracellular cadherin domain 1; EC1) of E-cadherin plays a key role (3). Loss of E-cadherin expression on the membrane results in metastasis by disrupting cell–cell adhesion and intracellular signaling, which is one of the main characteristics of epithelial–mesenchymal transition (EMT; refs. 4–6). The intracellular domain of E-cadherin interacts with β-catenin–α-catenin complex and participates in the regulation of actin-containing cytoskeletal filaments (2). In addition, E-cadherin recruits β-catenin to the cell membrane and inhibits its nuclear transcription activity (7, 8). Another E-cadherin–binding protein, CTNND1, also named as p120-catenin, interacts with the membrane-proximal region of E-cadherin (9–11).
Aberrant subcellular localization of E-cadherin, a potential diagnosis biomarker, has previously been examined in solid pseudo-papillary tumors (12), Merkel cell carcinomas (13), pancreatic endocrine tumors (14), and lung cancer (15). Su and colleagues detected and identified a possible link between cancer stem-like properties and loss of E-cadherin expression. They found that high-grade primary and metastatic lung cancers showed low levels of total (nuclear/membranous) E-cadherin; whereas E-cadherin nuclear expression was significantly inversely correlated with CD133 expression, which is a stem-like marker of multiple cancers (15).
E-cadherin was previously found to be regulated by posttranslational modifications (PTM), including phosphorylation (16–18), ubiquitination (19, 20), and binding of O-linked β-N-acetylglucosamine (O-GlcNAc; ref. 21). In the cytoplasmic domain of E-cadherin, there are 10 serine residues (2, 4), which are abundantly phosphorylated by the nonreceptor tyrosine kinase Src (17) and a cytosolic serine/threonine kinase casein kinase II (CKII; ref. 16), respectively. Activated Src phosphorylates E-cadherin, which in turn induces the Hakai-regulated ubiquitination of E-cadherin, resulting in endocytosis of E-cadherin and disruption of cell–cell contacts (19). CKII triggers phosphorylation of a series of serine residues within the serine-rich domains in the cytoplasmic region, which promotes the interaction of E-cadherin with β-catenin. Consistently, E-cadherin mutant exhibits a reduced β-catenin association and a concomitant effect on cell–cell adhesion (16).
We report here that E-cadherin possessed aberrant nuclear localization in colorectal cancer cells, which requires CTNND1′s activity. The nuclear E-cadherin was then found to be acetylated by CREB-binding protein (CBP), and the acetylation attenuated the association between E-cadherin and β-catenin, leading to enhanced transcription through β-catenin and accelerated colorectal tumorigenesis.
Materials and Methods
Antibodies and reagents
Antibodies against LAMIN B (sc-3652), β-ACTIN (sc-130656), HA-probe (sc-805), His-probe (sc-803), p120-catenin (sc-13957), SIRT2 (sc-20966), and β-catenin (sc-7199) were purchased from Santa Cruz Biotechnology. Anti–E-cadherin (#3195), Hsp90 (#4874), pan-acetylated-Lysine (#9441), and acetyl-α-tubulin (Lys40; #3971) were purchased from Cell Signaling Technology. mAbs against α-tubulin (T5168) and Flag (F1804) were purchased from Sigma-Aldrich.
Trichostatin A (TSA, S1045) and AGK2 (S7577) were purchased from Selleckchem. Nicotinamide (NAM; N0636) and β-Nicotinamide adenine dinucleotide hydrate (N7004) were purchased from Sigma-Aldrich. E-cadherin acetyl-K870.871 peptide [CDYLNEWGNRFK(Ac)K(Ac)LAD, purity > 98%] and unmodified peptide (CDYLNEWGNRFKKLAD) were synthesized by GL Biochem. The specific acetylated E-cadherin antibodies were prepared by Abclonal.
All the clinical samples were collected from patients who were surgically treated at Jinan Central Hospital from January 2009 to December 2012. All the patients were informed and consented before collecting specimens. All the samples were verified by two experienced pathologic staffs in a random and blinded manner. The study was approved by Jinan Central Hospital Affiliated to Shandong University (Jinan, China).
HEK293T, SW480, HCT116, A549, MD-MBA-231, and RKO cells were purchased from ATCC, and authenticated by the vendor. All cells were Mycoplasma free with regular checks performed by a LookOut Mycoplasma PCR Detection Kit (MP0035, Sigma-Aldrich).
HEK293T, RKO, MD-MBA-231, SW480, and HCT116 cells were maintained in DMEM (Hyclone) supplemented with 10% [volume for volume (v/v)] fetal calf serum (FCS, Gibco) and 100 U/mL penicillin and streptomycin. A549 was cultured in RPMI1640 medium (Hyclone) supplemented with 10% (v/v) FBS and 100 U/mL penicillin and streptomycin. All cells were purchased from the cell bank of Chinese Academy of Sciences and maintained at 37°C with 5% CO2.
Cell subcellular fraction
For the detection of E-cadherin subcellular localization, membranous, cytoplasmic, and nuclear proteins were fractioned using the Subcellular Protein Fractionation Kit (Thermo Fisher Scientific Inc) according to the manufacturer's instructions. Also noncommercial experimental buffers were used to extract cytoplasmic and nuclear proteins. Briefly, 2 × 106 cells were collected and spun down at 3,000 rpm in PBS. Pellets were suspended in 200 μL Buffer A (50 mmol/L HEPES, pH 7.9, 10 mmol/L KCl, 2 mmol/L MgCl2, 0.1 mmol/L EDTA, 0.5 mmol/L DTT, and 0.2% NP-40–containing protease inhibitors cocktail) and centrifuged at 13,200 rpm for 5 minutes after incubating on ice for 0.5 hour. The supernatants were cytoplasmic sections. After washing with 1 mL Buffer A, the pellets were lysed in 40 μL Buffer B (100 mmol/L HEPES, 600 mmol/L NaCl, 1.5 mmol/L MgCl2, 0.1 mmol/L EDTA, 0.5 mmol/L DTT, and 2.5% glycerol-containing protease inhibitors cocktail) thoroughly and spun at 13,200 rpm at 4°C for 15 minutes. Supernatant was collected as nuclear sections. For nuclear proteins immunoprecipitation, nuclear fractions were diluted with 4-fold PBS to reduce the salt concentration.
Standard Western blotting assay was used. Samples were heated at 95°C for 8 minutes, separated on a SDS-PAGE gel, and transferred onto nitrocellulose membranes. The related proteins were detected by specific antibodies. Analysis was conducted using Fluorescence Detection Methods (Odyssey Infrared Imaging System). Western blot assay pictures were normalized by Image Studio 3.1 software.
P120-catenin cDNA was a gift from Han's Laboratory (http://hanlab.xmu.edu.cn/), the code number is 8138. Open reading frame was cloned into pcDNA3.0 empty vector.
RNAi and transient transfection
CTNNB1 and CTNND1 knockdown were carried out using siRNA oligonucleotides synthesized from GenePharma. The following targeted sequences were chosen. siCTNNB1 #1: 5′-GCAGUUGUAAACUUGAUUATT, siCTNNB1 #2: 5′-CCCAAGCUUUAGUAAAUAUTT, siCTNNB1 #3: 5′-GGACACAGCAGCAAUUUGUTT, siCTNND1 #1: 5′-GCUUUCAUCCAGAGCCUUATT, siCTNND1 #2: 5′-GCAUGAGCUAGGAAGUUUATT, and siCTNND1 #3: 5′-GCUCGUGAUAUGGACCUUATT.
For siRNA transfection, HCT116 cells were plated at 50%–60% confluence and transfected with indicated siRNA oligoes using Lipofectamine 2000 (Invitrogen) according to the manufacturer's protocol.
Generation of specific antiacetylated E-cadherin antibody
The specific acetylated E-cadherin antibody was prepared by Abclonal according to standard protocol. Briefly, to generate antiacetylated E-cadherin antibody, peptides around K870.871 were synthetized by GL Biochem including the unmodified (CDYLNEWGNRFKKLAD) and acetyl-K870.871 [CDYLNEWGNRFK(Ac)K(Ac)LAD]. The acetyl-peptide was coupled to KLH as antigen to immunize rabbit. Antiserum was collected and purified after four doses of immunization and characterized by slot blotting and Western blotting.
Recombinant SIRT2 protein purification and in vitro deacetylation assay
SIRT2 was cloned into pET28a vector containing an N-terminal hexa-histidine tag. The plasmid was transformed into E. coli BL21 (DE3) grown in Luria-Bertani medium. The cells were cultured at 37°C to an OD600 of approximately 0.6. IPTG (0.2 mmol/L) was used to induce SIRT2 protein expression and the culture was grown for 18 hours at 20°C. Cells were collected by centrifugation at 8,500 × g for 10 minutes and then lysed by sonication on ice in Lysis Buffer (20 mmol/L Tris-HCl, pH 7.5, 200 mmol/L NaCl). After centrifugation at 12,000 × g for 1 hour at 4°C, the resulting supernatant was loaded onto a Ni-NTA (Qiagen) column and the column was washed with the buffer containing 20 mmol/L Tirs-HCl, 200 mmol/L NaCl, and 20 mmol/L imidazole, pH 7.5. The bound SIRT2 protein was eluted with Elution Buffer (20 mmol/L Tirs-HCl, 200 mmol/L NaCl, and 250 mmol/L imidazole, pH 7.5). The eluted SIRT2 was desalinated with Amicon Ultra-15 Centrifugal Filter Units (EMD Millipore). The purified protein was stored at −80°C.
The in vitro deacetylation of acetylated E-cadherin peptides was carried out essentially. Purified SIRT2 (0.1 μg) was incubated with 300 ng peptides in the deacetylation buffer containing 50 mmol/L Tris-HCl, pH 8.0, 137 mmol/L NaCl, 2.7 mmol/L KCl, 1 mmol/L MgCl2, 1 mg/mL BSA, and 3 mmol/L NAD+ or NAM, and warmed at 37°C for 2 hours in water bath. After reaction, 2 μL samples were spotted onto nitrocellulose membrane and blocked with 5% BSA, then detected as Western blotting procedure.
Total RNA was extracted by TRIzol RNA Isolation Reagent (Invitrogen) and reverse transcribed. Quantitative PCR was carried out using SYBR Select Master Mix (Applied Biosystems), following manufacturer's protocol. Before total RNA was harvested, RKO-stable cell lines were transfected with SIRT2-Flag plasmids for 36 hours. Primers used in qRT-PCR for each gene were as following: 5′- AAAGCGGCTGTTAGTCACTGG-3′ and 5′- CGAGTCATTGCATACTGTCCAT-3′ for human CTNNB1; 5′- GTGACAACACGGACAGTACAG-3′ and 5′- TTCTTGCGGAAATCACGACCC-3′ for human CTNND1; 5′- TGCGGAACTTATTCTCCCAGA-3′ and 5′- GAGAGCGAAAGTCGGGGAT-3′ for human SIRT2; 5′- CAACACCAGGCGGAACGAA-3′ and 5′- GCCCAATAAGGAGTGTAAGGACT-3′ for human AXIN2; 5′- GGCTCCTGGCAAAAGGTCA-3′ and 5′- CTGCGTAGTTGTGCTGATGT-3′ for human MYC; 5′- GCTGCGAAGTGGAAACCATC-3′ and 5′- CCTCCTTCTGCACACATTTGAA-3′ for human CCND1; 5′- AGAACACCCCGATGACGGA-3′ and 5′- GGCATCATTATGTACCCGGAAT-3′ for human LEF1; 5′- CTGGCTTCTACTCCCTGACCT-3′ and 5′- ACCAGAACCTAGCATCAAGGA-3′ for human TCF7; 5′- GGAGCGAGATCCCTCCAAAAT-3′ and 5′- GGCTGTTGTCATACTTCTCATGG-3′ for human GAPDH; and 5′- CATGTACGTTGCTATCCAGGC-3′ and 5′- CTCCTTAATGTCACGCACGAT-3′ for human ACTB. Data of triplicate average Ct were normalized with that of GAPDH or actin.
Preparation of K870.871-acetylated E-cadherin intracellular domain proteins
K870.871-acetylated E-cadherin intracellular domain was generated by the protocol described previously (22). Briefly, Lys 870 and 871 were replaced with amber codon, then wild-type and mutated intracellular domains of E-cadherin were cloned into pTEV8 plasmid. The constructions of pTEV8–E-cadherin-intra were transformed with pACKRS-3 and pCDF PylT-1 to E. coli BL21 (DE3). Bacterial cells were grown in Luria-Bertani medium supplemented with ampicillin (150 mg/mL), kanamycin (50 mg/mL), and spectinomycin (50 mg/mL), when the concentration of E. coli cells reached to OD600 of 0.6, bacterial cells were induced with 0.5 mmol/L IPTG, 20 mmol/L NAM, and 2 mmol/L Nϵ-acetyl-lysine for another 12 hours. E. coli cells were harvested, both E-cadherin intracellular domain proteins were purified by nickel beads.
Cell colony formation and wound-healing assay
For single-cell colony formation assay, 1,000 RKO cells were seeded onto 6-well plate in triplicate, and cells’ numbers were counted under optical microscope with crystal violet staining after 1-week culture.
Monolayer cells were scratched by a sterile tip and cell migration distances were observed and normalized by microscopy 24 or 48 hours later.
Statistical analysis of data was performed by Student t test using GraphPad Prism 5.0 software. All data are expressed as mean ± SEM and values of P < 0.05 were considered to represent a significant difference statistically.
Aberrant nuclear localization of E-cadherin in colorectal cancer
To investigate the nuclear localization of E-cadherin, we fractionated two colorectal cancer cell lines, HCT116 and SW480, into membranous, cytoplasmic, and nuclear fractions and expression profile of E-cadherin was detected by Western blotting. The results showed that E-cadherin mainly localized in membranous and nuclear fractions, rather than in the cytoplasmic fraction (Fig. 1A). A recent report suggested that E-cadherin nuclear localization had close correlation with lung tumorigenesis (15). To test whether E-cadherin was enriched in the nuclei of lung cancer cells, A549 lung cancer cells were used and fractioned for E-cadherin detection. Similarly, expression of E-cadherin was mainly detected in the nuclear fractions of A549 cells (Fig. 1A). To further corroborate, we next measured E-cadherin expression in HCT116 cells using immunofluorescence (IF) staining and found that E-cadherin primarily localized in the nuclei (Fig. 1B). Furthermore, the expression and localization of E-cadherin in human colorectal cancer samples were also evaluated using IHC staining and E-cadherin showed nuclear localization in 16 of 29 cases (Fig. 1C and D). Together, these results demonstrated that E-cadherin was enriched in nuclei in colorectal cancer.
CTNND1 was essential for E-cadherin nuclear localization
Because E-cadherin is a typical type I transmembrane protein and does not have any nuclear localization signal sequence, we next assessed the possible mechanism of E-cadherin nuclear localization. Two protein-binding domains were previously found in the intracellular part of E-cadherin, which are required for the interaction with CTNND1 and β-catenin (encoded by CTNNB1 gene), respectively. To study whether CTNND1 or β-catenin is involved in the nuclear localization of E-cadherin, we knocked down either CTNNB1 or CTNND1 in HCT116 cells and fractionated the cells to obtain the cytoplasmic and nuclear proteins. Western blotting results showed that knockdown of CTNND1, but not CTNNB1, reduced the amount of E-cadherin in the nuclear fraction (Fig. 2A–D). This result was consistent to the previous findings that E-cadherin C-terminal fragment 2 was cleaved by γ-secretase and released into the cytosol and nuclei, which was enhanced by CTNND1 (23). Furthermore, both of Western blot and IF results indicated that ectopic expression of CTNND1 in HCT116 cells increased the nuclear localization of E-cadherin (Fig. 2E–H). Collectively, these data indicate that CTNND1 is required for the nuclear translocation of E-cadherin.
E-cadherin was acetylated by CBP at Lysine870 and Lysine871
Histone acetylation is well-known to play important roles in transcriptional activation (24–26). Besides histone modification, histone acetyltransferases are also found to be able to acetylate numerous non-histone nuclear proteins (27–30). To test whether E-cadherin undergoes regulation through acetylation modification, we treated HCT116 and SW480 cells with the deacetylase inhibitors trichostatin A (TSA) and NAM, and E-cadherin was immunoprecipitated and detected by Western blotting using an anti–pan-acetyl-Lysine antibody. The acetylation of E-cadherin was detected at the basal level, which markedly increased upon TSA and NAM treatment (Fig. 3A). To identify which enzyme(s) was/were responsible for E-cadherin acetylation, the ectopic CBP and E1A-binding protein 300 kDa (p300) were overexpressed and the acetylation of E-cadherin was examined. The results showed that overexpression of ectopic CBP, but not p300, dramatically enhanced E-cadherin acetylation, indicating that CBP was the acetyltransferase for E-cadherin acetylation (Fig. 3B). Moreover, to confirm which lysine site(s) was/were acetylated, we mutated every lysine (K) into arginine (R), which cannot be acetylated and keeps the positive charge of the amino acid side chain. We found that the K-to-R mutation at either Lysine870 (Lys870) or Lysine871 (Lys871) decreased E-cadherin acetylation while the double mutation of these two lysine sites abolished the acetylation signal, suggesting that both the Lys870 and Lys871 residues were simultaneously acetylated by CBP (Fig. 3C). The schematic diagram of primary structure of E-cadherin showed that these two residues were located in the β-catenin–binding domain (Fig. 3D).
We then generated a specific antibody against acetylated Lys870/Lys871 (anti–acetyl-E-cad), and the quality and specificity of the antibody was verified. Immunoblotting results showed that anti–acetyl-E-cad antibody specially recognized the synthetic acetylated peptides as well as an E. coli expressed recombinant E-cadherin intracellular domain with its Lys870 and Lys871 acetylated (Fig. 3E and F). Because E-cadherin was translocated into the nuclei in HCT116, the localization of acetylated E-cadherin was evaluated by immunostaining using anti–acetyl-E-cad and the results showed that acetylated E-cadherin mainly localized in cell nuclei (Fig. 3G). Together, these results suggested that E-cadherin was acetylated by CBP at both Lys870 and Lys871.
SIRT2 was the deacetylase for E-cadherin
TSA is an inhibitor of histone deacetylases, while NAM is the pan-inhibitor of sirtuin family, which catalyzes the deacetylation of multiple proteins in the presence of NAD (31–36). As NAM, but not TSA, was able to enhance endogenous E-cadherin acetylation (Fig. 3A), we screened for the deacetylase of E-cadherin among sirtuin family members. There are seven SIRT proteins (SIRT1–7) in mammal cells, and different SIRT proteins have their specific subcellular localizations (37–39). As SIRT3, SIRT4, and SIRT5 mainly locate in mitochondria, thus may not be the deacetylase of E-cadherin, we chose SIRT1, SIRT2, SIRT6, and SIRT7 for our further study. The ectopic SIRTs were overexpressed and the E-cadherin acetylation was measured. The results showed that among all SIRTs tested, only overexpression of SIRT2 decreased the acetylation level of E-cadherin (Fig. 4A). Then we utilized five methods to confirm that SIRT2 is the specific deacetylase for E-cadherin. First, expression of the catalytically inactive SIRT2 H187Y mutant (40) showed no effect to the acetylation level of E-cadherin (Fig. 4B). Second, HCT116 and SW480 cells were treated with the SIRT2-specific inhibitor AGK2, and endogenous E-cadherin acetylation was measured (41). Inhibition of SIRT2 by AGK2 markedly enhanced E-cadherin acetylation. As tubulin is a known substrate of SIRT2, acetylation of tubulin was also increased when SIRT2 was inhibited by AGK2 (Fig. 4C; ref. 42). Third, in vitro deacetylation assay showed that an E. coli expressed SIRT2, specifically deacetylated the synthetic acetylated E-cadherin peptides, and this process was blocked by the addition of SIRTs inhibitor NAM (Fig. 4D). Fourth, overexpression of SIRT2 decreased E-cadherin acetylation in HCT116 cells (Fig. 4E).Finally, inhibition of SIRT2 by siRNAs dramatically upregulated the acetylation signal of endogenous E-cadherin (Fig. 4F and G). Collectively, these results suggested that SIRT2 is the deacetylase for E-cadherin.
Moreover, to examine the relationship between E-cadherin acetylation and its nuclear translocation, firstly we transfected Flag-SIRT2 in HCT116 cells, as overexpression of ectopic SIRT2 decreased E-cadherin acetylation, it cause no significant change of E-cadherin nuclear translocation (Fig. 4H and I), suggesting that acetylation of E-cadherin did not affect its nuclear translocation. On the other hand, we also inhibited the expression of CTNNB1 and CTNND1 by siRNAs in HCT116 cells, and examined the acetylation level of E-cadherin in different fractions. Western blotting results showed a significant decrease in acetylation of nuclear E-cadherin in CTNND1 knockdown cells but not CTNNB1 knockdown cells (Fig. 4J and K), suggesting that the nuclear translocation of E-cadherin promoted its acetylation.
Colorectal tumorigenesis was associated with E-cadherin acetylation
Then we evaluated E-cadherin acetylation levels in five pairs of colorectal tumor and their normal adjacent tissues. Immunoblot analysis results showed increased E-cadherin acetylation levels in three tumor samples (#138, #156, and #204), in two of which the enhanced E-cadherin acetylation was accompanied by reduced SIRT2 expression (#156 and #204; Fig. 5A). IHC staining results in 92 pairs of colorectal tumor and their normal adjacent tissues showed that in tumor tissues the E-cadherin acetylation levels were higher than in normal adjacent tissues (Fig. 5B and C). Furthermore, the higher acetylation of E-cadherin in the tumors was with poor prognosis (Fig. 5D).
Furthermore, we utilized RKO cell line, a low-differentiated colorectal cancer cell line, which does not express endogenous E-cadherin (43, 44), to generate stable cell lines overexpressing either wild-type E-cadherin or E-cadherin with K870R/K871R (KKRR) mutations. RKO cells became flatter and cell adhesion was tighter with wild-type E-cadherin overexpression; and similar phenotypic changes were observed with the KKRR mutation overexpression (Fig. 5E). Consistently, the KKRR mutation did not influence the formation of E-cadherin homodimers (data not shown). Because E-cadherin acetylation was previously suggested to be associated with colorectal tumorigenesis (Fig. 5A and B), we therefore performed colony formation assay to check whether E-cadherin acetylation affected the tumorigenesis of RKO cells. As E-cadherin is a known tumor repressor, RKO cells stably overexpressing wild-type E-cadherin exhibited a significantly lower colony-forming capacity; moreover, E-cadherin KKRR mutant displayed an even stronger inhibitory effect on colony formation than wild-type E-cadherin (Fig. 5F). Furthermore, overexpression of ectopic SIRT2 suppressed colony formation capacity of the control RKO cells or those containing wild-type E-cadherin, but could not suppress KKRR mutant any longer (Fig. 5F), suggesting that the suppressing capacity of SIRT2 is through its deacetylation on E-cadherin Lys870/Lys871. We also evaluated KKRR mutations’ effect on the cell migration activity. RKO cells were serum-starved overnight to exclude the interference by cell proliferation; monolayer cells were scratched off and the migration distance was examined. When compared with the wild-type E-cadherin, E-cadherin KKRR mutant also displayed a significantly stronger inhibitory effect on cell migration (Fig. 5G). Also, overexpression of ectopic SIRT2 could no longer suppress cell migration of RKO cells containing KKRR mutant (Fig. 5G), suggesting the suppressing of cell migration by SIRT2 is dependent on E-cadherin acetylation. We conclude that E-cadherin acetylation is associated with and may promote colorectal tumorigenesis and cell migration.
E-cadherin acetylation increased the activity of β-catenin
Of the E-cadherin cytoplasmic domain, β-catenin is one of the most important binding factors. Through β-catenin transition, E-cadherin affects a variety of cell functions, such as proliferation, migration, and actin accumulation (1, 45, 46). To test whether acetylation of E-cadherin–mediated tumorigenesis through affecting β-catenin, expression of several β-catenin downstream targeting genes, Axin2, c-myc, CCND1, LEF1, and TCF7, were detected by qRT-PCR. The results showed that the KKRR mutant of E-cadherin exhibited an enhanced inhibiting capacity on the expression of these β-catenin–responsive genes, when compared with the wild-type E-cadherin (Fig. 6A).
As shown before, Lys870 and Lys871 are located in the β-catenin–binding domain of E-cadherin (Fig. 3E), we tested whether E-cadherin acetylation regulated the interaction between β-catenin and E-cadherin. As ectopically expressed E-cadherin could form dimers with endogenous proteins, to exclude such effects, mutants of E-cadherin with cadherin domain 1 truncation (ΔEC1), which is essential for E-cadherin homodimer formation (3), were constructed, including wild-type, KKRR, and KKQQ (Lys870/Lys871 mutated into Glutamine to mimic Lysine acetylation). HEK293T cells were transfected with indicated plasmids, and the interaction between β-catenin and E-cadherin was analyzed by coimmunoprecipitation. Results showed reduced interaction between KKQQ-mutant form E-cadherin and β-catenin (Fig. 6B). As phosphorylation at Serine838/840 was reported to enhance E-cadherin's association with β-catenin (16), we therefore mutated these two serines (S) into alanines (A) to block the phosphorylation. Similarly, the SA mutation also suppressed the interaction between E-cadherin and β-catenin (Fig. 6C). MDA-MB-231 cells do not express endogenous E-cadherin, we further reperformed the coimmunoprecipitation assay in this cell lines to exclude any effects of endogenous E-cadherin on the protein association. In addition, consistently, the KKQQ mutation reduced the interaction between these proteins (Fig. 6E).
Then we purified the acetylated E-cadherin recombinant protein to examine whether the acetylation affected the direct binding between E-cadherin and β-catenin. GST pull-down assay showed that much less of acetylated E-cadherin than unacetylated controls was precipitated by the GST–β-catenin (Fig. 6D). Taken together, these results suggest that acetylation of E-cadherin interferes its association with β-catenin.
Finally, RKO-stable cell lines treated with or without TSA and NAM were fragmented into cytoplasmic and nuclear fractions. E-cadherin was precipitated and endogenous β-catenin was detected. In cytoplasmic panels, the interaction between these two proteins was not affected by NAM and TSA treatment. Whereas in the nuclear fractions, wild-type E-cadherin, but not the KKRR mutant, bound less β-catenin upon NAM and TSA treatment (Fig. 6F). Taken together, these results suggest that acetylation of E-cadherin in nuclear interferes its association with β-catenin, resulting in increased activity of β-catenin.
Our current study demonstrated the aberrant nuclear localization of E-cadherin in colorectal cancers. Further study showed that nuclear E-cadherin was regulated by acetylation modification, and CBP and SIRT2 were identified to be the major acetyltransferase and the deacetylase in this process. Acetylation of E-cadherin then led to increased β-catenin's transcription regulatory activity and consequently, enhanced β-catenin–dependent tumor growth and migration. Although aberrant nuclear localization of E-cadherin was previously observed in multiple tumors, and its function was proposed to be correlated with cancer stem cell formation and tumor invasion, our study revealed the first piece of evidence that aberrant nuclear localization of E-cadherin also existed in colorectal cancer cell lines and tumor samples, and functionally promoted colorectal tumor progression, suggesting a conventional role of nuclear E-cadherin in tumorigenesis of multiple tumors.
Our study indicated that nuclear E-cadherin facilitates the transcription regulatory capacity of β-catenin and promotes β-catenin–dependent cell growth and migration. This seems contradictory to the previous discovery that nuclear localization of E-cadherin was a negative regulator of Wnt/β-catenin, and promoted the CSC phenotype in lung cancers. Despite the difference between these two tumors, one possible explanation is that acetylation of E-cadherin by CBP is pivotal for its function in the enhancement of β-catenin activity. Numerous studies showed that in multiple tumors including lymphoma, breast cancer, and lung cancer, CBP is a pro-oncogene by acetylating Histone H3, as well as various transcriptional factors and facilitating the expression of multiple protumor genes such as c-myc and E1A (47, 48). Our finding that CBP-mediated acetylation of E-cadherin promoted tumorigenesis through enhancing β-catenin transcriptional capacity expanded the experimental evidences to the theoretical role of CBP-mediated acetylation in promotion of tumor growth and metastasis.
Nuclear β-catenin exerts gene transcriptional initiating activity, which is involved in Wnt signaling; however, E-cadherin–anchored membrane β-catenin does not (49, 50). E-cadherin shares the binding domain of β-catenin with BCL9, which promotes β-catenin transcriptional activity (51–54). When full-length E-cadherin translocate into nuclei, it keeps the ability to interact with β-catenin and to inhibit its function, which is contradictory with previous conclusion that nuclear localization of E-cadherin was associated with high Wnt signaling activation. One possible explanation to this phenomenon postulates that E-cadherin activity is regulated by PTM. Our study showed that E-cadherin is acetylated at Lys870 and Lys871 in the β-catenin–binding domain by acetyltransferase CBP and that acetylated E-cadherin reduces the interaction with β-catenin; whereas released β-catenin is able to initiate the transcription of downstream genes.
According to the previously published structure data (55), the association between E-cadherin and β-catenin is dependent on the electrostatic force between positively charged Lys871 of E-cadherin and negatively charged Glu155 of β-catenin. Our finding showed that acetylation of E-cadherin at Lys870 and Lys871 dramatically disrupted the E-cadherin–β-catenin association, most likely due to the neutralization effect of acetylation on Lysine residue.
There are numerous studies showing that the loss of expression of E-cadherin on the cell surface is a key step in the transdifferentiation of epithelial cells to a mesenchymal phenotype, which occurs in cancer progression. Our finding showed that the increased nuclear localization of E-cadherin was associated with the loss of membranous E-cadherin, which was consistent with previous publications. In addition to this, our study also suggest that the nuclear localization of E-cadherin may as well promote cell motility via enhancing β-catenin transcriptional activity, indicating a potent cumulative effect of the aberrant nuclear localization of E-cadherin on tumor metastasis.
Disclosure of Potential Conflicts of Interest
No potential conflicts of interest were disclosed.
Conception and design: Y. Zhao, Y. Qin, F. Liu
Development of methodology: Y. Zhao, T. Yu, Y. Qin
Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): Y. Zhao, T. Yu, N. Zhang, J. Chen, P. Zhang, S. Li, L. Luo, Z. Cui
Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): Y. Zhao, T. Yu, J. Chen, S. Li, Z. Cui, Y. Qin, F. Liu
Writing, review, and/or revision of the manuscript: Y. Zhao, Y. Qin, F. Liu
Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): Y. Zhao, Y. Qin, F. Liu
Study supervision: Y. Qin, F. Liu
The authors would like to thank LetPub (www.letpub.com) for providing linguistic assistance during the preparation of this manuscript. We also thank Dr. Qiurong Ding for her kind advice on the project and the manuscript preparation. This research was supported by grants 81501373 (to F. Liu), 8150002 (to X. Xu) from the National Science Foundation of China, and grant 17ZR1423400 (to Z. Cui) from Natural Science Foundation of Shanghai.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.