Helicobacter pylori CagA Suppresses Apoptosis through Activation of AKT in a Nontransformed Epithelial Cell Model of Glandular Acini Formation.

H. pylori infection is the most important environmental risk to develop gastric cancer, mainly through its virulence factor CagA. In vitro models of CagA function have demonstrated a phosphoprotein activity targeting multiple cellular signaling pathways, while cagA transgenic mice develop carcinomas of the gastrointestinal tract, supporting oncogenic functions. However, it is still not completely clear how CagA alters cellular processes associated with carcinogenic events. In this study, we evaluated the capacity of H. pylori CagA positive and negative strains to alter nontransformed MCF-10A glandular acini formation. We found that CagA positive strains inhibited lumen formation arguing for an evasion of apoptosis activity of central acini cells. In agreement, CagA positive strains induced a cell survival activity that correlated with phosphorylation of AKT and of proapoptotic proteins BIM and BAD. Anoikis is a specific type of apoptosis characterized by AKT and BIM activation and it is the mechanism responsible for lumen formation of MCF-10A acini in vitro and mammary glands in vivo. Anoikis resistance is also a common mechanism of invading tumor cells. Our data support that CagA positive strains signaling function targets the AKT and BIM signaling pathway and this could contribute to its oncogenic activity through anoikis evasion.


Introduction
Helicobacter pylori (H. pylori) colonizes the human gastric epithelium and is considered the most important cause of chronic active gastritis, peptic ulcer, and gastric cancer [1]. The pathogenesis of H. pylori is importantly associated with the presence of the cag pathogenicity island (cagPAI) and the cagPAI effector protein, the cytotoxin-associated gene A 2 BioMed Research International (CagA) [2]. The cagPAI is a segment of DNA of about 40 kb that encodes a type IV secretion system (T4SS), which is necessary for CagA translocation into target epithelial cells. Once inside the cell, CagA is phosphorylated in tyrosine residues of the EPIYA motif by host cytoplasmic Src and c-Abl kinases, and phosphorylated and nonphosphorylated CagA interact with multiple signaling proteins [3][4][5][6][7][8].
H. pylori activation of the phosphoinositide 3-kinase (PI3K) and protein kinase B (PKB/AKT) signaling pathway has been previously documented in transformed gastric epithelial cells (AGS cells), although the mechanism by which this happens is not fully understood. On one hand, some studies support CagA phosphorylation dependent and independent roles [9][10][11]. On the other hand, a role for proinflammatory outer membrane (OipA) and vacuolating cytotoxin A (VacA) proteins has been proposed [12,13], ruling out a role for the cagPAI [14]. Also, multiple targets downstream of PI3K/AKT have been documented, including mammalian target for rapamycin (mTOR), forkhead box O (FoxO)-1 and -3a ERK mitogen activated kinase, and proapoptotic protein BAD [15][16][17][18][19]. Concordantly, the consequence of H. pylori activation of PI3K/AKT is also unclear, with different studies supporting deregulation of apoptosis, proliferation, or cell migration.
The use of transformed cells has been essential to understand H. pylori pathogenesis, but it may also contribute to the conflicting data as many signaling pathways and cellular processes associated with cell transformation are already deregulated. CagA-induced proliferation and altered cell polarity have also been shown in nontransformed Madindarby canine kidney epithelial cells (MDCK cells), but CagA's signaling has been partially described [20,21]. It was reported that CagA disrupts epithelial apical-basolateral polarity in MDCK cells by interacting with PAR1/MARK kinase, which prevents atypical protein kinase C-(aPKC-) mediated PAR1 phosphorylation [22]. More definitive evidence of the CagA oncogenic role comes from transgenic mice, in which CagA expression induced epithelial hyperplasia, polyp formation, and adenocarcinomas of the gastrointestinal tract [23,24]. Also, CagA transgenic expression in zebrafish induced epithelial cell proliferation and upregulation of cyclin D1, axin2, and the c-myc ortholog myca [25].
To better understand CagA interactions with cancerassociated signaling pathways and cellular processes, we studied CagA activity in a model of nontransformed epithelial cells. The epithelial cell line MCF-10A forms threedimensional (3D) acini-like spheroids with a hollow lumen and an apicobasal orientation when cultured in a simile of the extracellular matrix (ECM). These characteristics allow testing mechanisms of cell proliferation, cell survival and the cytoskeletal structure that yields the polarized spheroid architecture [26,27]. Hence, this 3D cellular system has been previously used to test cellular and viral oncogenes and has proved useful to decipher mechanisms of transformation and their targeted cellular signaling pathways [28,29]. We infected MCF-10A spheroids with CagA positive and negative H. pylori variants finding that CagA positive strains caused evasion of apoptosis that was associated with phosphorylation of AKT, BIM, and BAD, which suggests that CagA inhibits the anoikis form of apoptosis.

Material and Methods
2.1. Helicobacter pylori Strains and Culture. Two CagA positive H. pylori strains were used in this study: strain 11637 with a Western-type CagA (EPIYA ABCCC) that was obtained from the American Type Culture Collection (ATCC, Manassas, VA, USA No. 43504); and strain NY02-149 with an East-Asian-type CagA (EPIYA ABD) that was kindly donated by Dr. Guillermo Perez-Perez from New York University. Two additional H. pylori CagA negative variants were used as controls: strain 365A3, which has a partial cagPAI lacking the effector protein CagA and strain 254 that contains a nonfunctional cagPAI [30]. The latter will be referred as the cagPAI negative strain. The two CagA positive strains and the CagA negative variant are VacA s1/m1, while the cagPAI negative strain is positive to s1 and negative to the m region (Data no show). All H. pylori strains were grown on blood agar (BD, Bioscience, San Jose, CA, USA No. 211037) for 48 h at 5% CO 2 and 37 ∘ C.
2.2. PCR Assay. DNA was obtained from H. pylori strains using a Qiagen DNA extraction kit (Qiagen, Hilde, Germany No. 51306) according to the manufacturer protocols. DNA samples were subjected to PCR with primers cagA-F1 5 -ATGACTAACGAAACTATTGATCAA-3 and cagA-R7 5 -TTAAGATTTTTGGAAACCAC-3 for full-length cagA amplification and cagA-F5 5 -CCCTAGTCGGTA-ATGGGTTATC-3 and CagA-R7 for EPIYA region amplification (Figure 1(a)). PCRs for VacA were also performed with primers VA1F 5 -ATGGAAATACAACAAACACAC-3 and VA1R 5 -CTGCTTGAATGCGCCAAAC-3 for amplification of the S region and VAGF 5 -CAATCTGTCCAATCA-AGCGAG-3 and VAGR 5 -GCGTCAAAATAATTCCCA-AGG-3 for the M region. Infection was carried out with a MOI of 100 in MCF-10A and AGS cells at 80% subconfluency in nonsupplemented medium, unless otherwise specified. For proliferation assays MCF-10A cells were infected at 30% subconfluency in supplemented medium with 3% horse serum and the standard concentration of the other components. MCF-10A acini were of Matrigel. 400 L of culture medium was added per well and acini were allowed to form for 10-15 days, observing daily at the optical microscope and changing media every two days. Acini were grown in same medium as MCF-10A monolayers but with 2% of matrigel. For infection assays, MCF-10A cells were infected with a MOI of 100 in a monolayer for 2 h at 5% CO 2 and 37 ∘ C before recovering the cells and seeding them in the layer of matrigel. Cells were grown for 10 to 15 days changing medium every other day. Media of days 2, 4, and 6 also contained bacteria at the same MOI. To inhibit CagA phosphorylation-dependent activity, the PP2 Src kinase inhibitor was added on day 0, 2, 4, and 6 of acini formation and pictures were taken on day 10. 2.9. Statistical Analysis. The statistical differences between two variables were determined with the Student's t-test; differences among three or more continuous variables were compared by one-way ANOVA, followed by the Tukey test. Statistical significance was established at p ≤ 0.05.

H. pylori CagA Is Efficiently Translocated into Nontrans-
formed MCF-10A Cells. The two CagA positive H. pylori strains that were used in our study were able to induce a hummingbird phenotype and IL-8 secretion in AGS cells, as previously described [31,32], which are indirect markers of CagA translocation into these cells (Figures 1(b) and 1(c)). AGS and MCF-10A infections with analysis of hummingbird phenotype and IL-8 secretion were run in parallel through all experiments to monitor the viability of the H. pylori strains.
To determine if MCF-10A cells were permissive to H. pylori infection, infected cells were analyzed by immunofluorescence and Western blot. The immunofluorescence analysis showed that H. pylori, regardless of CagA status, adhered to the cell membrane of MCF-10A, with CagA signal observed within the cell (Figure 1(d)). Detection of phosphorylated-CagA confirmed its cellular translocation and activation (Figure 1(e)). Of note, the CagA-ABD strain showed stronger intensity of the phosphorylated CagA band and stronger activity than the CagA-ABCCC strain. These results indicated that MCF-10A cells are permissive for infection/CagA translocation with H. pylori strains.

MCF-10A Cells Evade Apoptosis after Infection with CagA
Positive H. pylori Strains. A consistent observation of MCF-10A acini infected with CagA positive H. pylori strains was the lack of a well-formed lumen. To explore the possibility that CagA induces survival of acini central cells, the number of nuclei was quantified at day 10 acini. Confocal microscopy transversal cuts 50% deep in the acini were used for nuclei count, finding significant differences between CagA positive H. pylori strains and mock infected acini (Figures 3(a)  and 3(b)). Acini morphogenesis in the presence of the Src kinases inhibitor PP2 indicated that the survival of the acini central cells was dependent on phosphorylation of CagA (Figures 3(c) and 3(d)). MCF-10A cells were also infected in suspension and apoptosis was determined by Anexin V and propidium iodide (PI) staining and flow cytometry analysis. Figures 3(e) and 3(f) show that, after 6 h of suspension, cells infected with CagA negative variants were mostly dead, while about 40% of the ones infected with CagA positive strains remained anexin V and PI negative. Overall, these data suggest that CagA promotes survival of cells that have lost substrate or ECM interactions.

AKT and BIM Are Important Targets during CagA Positive H. pylori Mediated Evasion of Apoptosis. Anoikis is a form of apoptosis responsible for death of the acini luminal cells after loss of ECM interactions. Anoikis resistance in MCF-
10A acini formation and mammary gland development has been associated with constitutive activity of the PI3K/AKT signaling pathway and phosphorylation-dependent inactivation of proapoptotic proteins mainly BIM but also BAD [33]. To evaluate whether CagA-induced activation of AKT was responsible for the anoikis resistance observed in MCF-10A acini luminal cells, we first tested AKT phosphorylation in response to H. pylori infection by Western blot analysis. MCF-10A cells were infected in single-cell suspensions with H. pylori strains for 4 h and cell lysates were blotted with the antiphosphorylated AKT. Figure 4(a) shows the specific activation of AKT by CagA positive strains. Also, a preferentially increased phosphorylation of BIM over BAD was observed (Figure 4(a)). We then inhibited AKT activation reducing the frequency of live MCF-10A cells in suspension (Figures 4(b) and 4(c)). These results argue in favor of an H. pylori mechanism of anoikis evasion through CagA-induced activation of AKT and inhibition of BIM.

Discussion
The 3D culture of MCF-10A epithelial cells interacting with a simile of ECM results in formation of acini-like polarized structures that recapitulate many of the characteristics of the in vivo mammary gland architecture. Acini formation involves programmed grow arrest, anoikis of luminal cells, and preservation of the cytoskeleton-dependent morphology, biological and biological processes that are usually deregulated in cancer cells. Therefore, this system has been extensively used to study cellular and viral oncogenes, including some unrelated to mammary tissue but whose transforming   mechanisms can be mirrored during acini formation [26,28,[34][35][36].
Oncogene expression often results in MCF-10A acini with solid lumens, similar to the ones observed in this study after infection with CagA positive H. pylori strains, and this is considered an indication of resistance to anoikis [37,38]. Anoikis is a form of apoptosis responsible for death of inner cell populations that have lost ECM interactions and grow factor stimulus, forming the hollow ducts in which milk is transported during mammary gland morphogenesis. MCF-10A in vitro acini formation has been critical to understand the in vivo mechanism of anoikis during mammary morphogenesis [26,27]. Anoikis resistance is very important during cancer initiation and progression since it is most likely BioMed Research International 9 responsible for the survival capacity of tumor cells filling the luminal glandular space in early carcinomas and of detached invasive and metastatic tumor cell [39], including gastric cancer cells [40]. In agreement, bonafide oncoproteins such as ERBB2 and CFS1R induce resistance to anoikis facilitating migration and metastasis of malignant cells [41][42][43].
In this paper we have shown that CagA positive H. pylori strains activate AKT resulting in evasion of apoptosis, a finding that correlates with previous reports in AGS transformed gastric tumor cells [9][10][11]. AKT is an important regulator of cell survival. AKT is activated by PI3K leading to inactivation of various proapoptotic proteins, including BIM and BAD [19]. AKT phosphorylates BAD in Ser136 and BIM in Ser87 in response to cell-integrin interactions. Phosphorylated BAD and BIM are then sequestered by the 14-3-3 complex counteracting its proapoptotic activity [44,45]. Loss of the AKT activity results in augmented apoptosis [38,46], and activating mutations in PI3K and AKT are often found in gastric tumors [47]. Furthermore, lumen formation by mammary glands and MCF-10A cells tightly depends on BIM activity [33,48].
Several studies also support an increased rate of cell apoptosis after H. pylori infection mediated by different bacterial factors, mainly VacA [49][50][51], gamma-glutamyl transpeptidase [52], the cagPAI [53], and CagA itself [54]. Cancer precursor gastric lesions, such as atrophic gastritis and metaplasia, are already characterized by loss of the glandular structure. These lesions may result from infection-triggered unbalanced cell apoptosis intimately linked to increased cellular regeneration to repair the gastric mucosa. CagA may help to counteract those apoptosis-inducing mechanisms triggered by bacterial factors or cell death resulting from chronic inflammation [55,56].
Our results showed that H. pylori infection does not increase proliferation of MCF-10A cells during acini or monolayer growth, which contrasts with previous studies showing that phosphorylated CagA interacts with GRB2 or SHP2 activating the Ras/MAPK pathway, leading to cell scattering and cell proliferation [4,57]. Other studies support that H. pylori promote proliferation through upregulating miR-222 [58]. In addition, zebrafish expressing transgenic CagA exhibited epithelial cell proliferation with significant upregulation of gene markers of proliferation CCND1 and the c-myc ortholog myca [25]. Similarly, CagA transgenic mice showed gastric epithelial hyperplasia [23]. We have no clear explanation for these differences, but we hypothesize that this may be due to the different genetic background of the cell lines used in those studies and our study. While in animal models multiple processes converge during carcinogenesis, such as inflammation and persistent tissue damage due to chronic CagA expression.
Studies in nontransformed MDCK cells found that H. pylori induced loss of cell polarity and mobilization of zonula occludens-1 (ZO-1) protein with loss of integrity of the tight junctions [20,21]. In contrast, we did not observe gross cytoskeletal changes in infected MCF-10A acini, which maintained the spheroid structure. Analysis of proteins participating in monolayer integrity and cell-to-cell contacts is needed together with nutrient deprivation studies to better understand the capacity of H. pylori and CagA to disturb these processes in MCF-10A acini.
The importance to test H. pylori pathogenicity in cellular models of organogenesis has been highlighted by the appearance of three recent papers in which mouse and human primary gastric organoids are tested for H. pylori infection and CagA signaling activity [59][60][61]. Two of those studies provide evidence of a CagA-induced epithelial to mesenchymal transition (EMT). Interestingly, EMT is importantly regulated by GSK3 inhibition of -catenin and GSK3 is negatively regulated by AKT. These studies and ours place AKT as an important target of CagA induced carcinogenesis.

Conclusion
We found that H. pylori CagA positive strains induce anoikis resistance in MCF-10A acini through the AKT signaling pathway, via phosphorylation and inactivation of proapoptotic proteins BIM and BAD. This CagA-dependent mechanism of anoikis resistance may contribute to the H. pylori/CagA carcinogenic potential. Our results also support the use of nontransformed cells and in vitro organogenesis in order to better understand the oncogenic mechanisms involved in H. pylori associated cancer development.