Long-Term Helicobacter pylori Infection Switches Gastric Epithelium Reprogramming towards Cancer Stem Cell-Related Differentiation Program in Hp-Activated Gastric Fibroblast-TGFβ Dependent Manner

Helicobacter pylori (Hp)-induced inflammatory reaction leads to a persistent disturbance of gastric mucosa and chronic gastritis evidenced by deregulation of tissue self-renewal and local fibrosis with the crucial role of epithelial–mesenchymal transition (EMT) in this process. As we reported before, Hp activated gastric fibroblasts into cells possessing cancer-associated fibroblast properties (CAFs), which secreted factors responsible for EMT process initiation in normal gastric epithelial RGM1 cells. Here, we showed that the long-term incubation of RGM1 cells in the presence of Hp-activated gastric fibroblast (Hp-AGF) secretome induced their shift towards plastic LGR5+/Oct4high/Sox-2high/c-Mychigh/Klf4low phenotype (l.t.EMT+RGM1 cells), while Hp-non-infected gastric fibroblast (GF) secretome prompted a permanent epithelial–myofibroblast transition (EMyoT) of RGM1 cells favoring LGR−/Oct4high/Sox2low/c-Myclow/Klf4high phenotype (l.t.EMT−RGM1 cells). TGFβ1 rich secretome from Hp-reprogrammed fibroblasts prompted phenotypic plasticity and EMT of gastric epithelium, inducing pro-neoplastic expansion of post-EMT cells in the presence of low TGFβR1 and TGFβR2 activity. In turn, TGFβR1 activity along with GF-induced TGFβR2 activation in l.t.EMT−RGM1 cells prompted their stromal phenotype. Collectively, our data show that infected and non-infected gastric fibroblast secretome induces alternative differentiation programs in gastric epithelium at least partially dependent on TGFβ signaling. Hp infection-activated fibroblasts can switch gastric epithelium microevolution towards cancer stem cell-related differentiation program that can potentially initiate gastric neoplasm.


Introduction
Gastric cancer (GC) remains the fourth most common cause of cancer-related deaths worldwide. It is estimated that 90% of all stomach tumors are malignant, with gastric adenocarcinoma comprising 95% of the total number of malignancies [1]. Despite the improvement in surgery, chemo-radiotherapy, reprogramming of gastric epithelial cells. In particular, we addressed the role of Hp signaling in the determination of Hp-AGF-mediated pluripotency of EMT-committed gastric epithelial cells, with a special emphasis to the role of TGFβ signaling in this process.

Bacterial Hp Strain
The Hp strain expressing CagA and VacA cytotoxins (43504 Hp cagA+vacA+ (s1/m1)) was purchased from American Type Culture Collection [36,37]. Stock cultures were maintained at −70 • C in Brucella Broth (Becton Dickinson, New Jersey, NJ, USA), supplemented with 10% fetal bovine serum and 10% glycerol. The cultures of bacteria were grown on Columbia Agar with 5% fresh horse blood (BioMerieux, Marcy l'Etoile, France). The plates were incubated under microaerophilic conditions at 37 • C for 3-5 days. Before the co-incubation with fibroblasts, Hp strain was suspended in PBS (Sigma-Aldrich, Saint Louis, MO, USA) and immediately transferred to the dishes containing fibroblasts.

Technique of Rat Gastric Fibroblasts Isolation and Their Activation towards Fibroblasts Possessing CAFs Characteristic
Gastric samples were harvested from 8-week-old Spraque-Dowley rats and extensively washed with sterile PBS to remove contaminating debris. Primary and secondary fibroblast culture was established as described previously [36]. The cells were cultured in DMEM containing 10% FBS and antibiotics (penicillin, streptomycin, and amphotericin B; Sigma A5955, Sigma-Aldrich, Saint Louis, MO, USA). The flasks were maintained in a humidified atmosphere of 5% CO 2 at 37 • C, and the medium was changed every 2 days [36]. Before the co-incubation with fibroblasts, Hp strain was suspended in PBS, counted with Densimat Densitometer (bioMerieux, Marcy l'Etoile, France), and immediately transferred to the dishes containing fibroblasts. Then, 0.25 × 10 6 fibroblasts were infected with 1 × 10 9 of live Hp and incubated in humidified atmosphere for 72 h (Hp-AGFs). As for the control, gastric fibroblasts were cultured for 72 h in DMEM + 10% FBS (GF) [22,[35][36][37].

Isolation of Hp-AGF and Normal Gastric Fibroblast (GF)-Conditioned Media
Gastric fibroblasts were co-cultured with Hp (cagA+vacA+) strain for 72 h [35]. Then, the Hp was washed out from fibroblasts and the medium was changed into DMEM + 10% FBS and antibiotics (penicillin, streptomycin, and amphotericin B; Sigma A5955, Marcy l'Etoile, France). The culture dishes were maintained in a humidified atmosphere of 5% CO 2 at 37 • C for 4 h, and then the incubation fluid was again replaced with fresh portion of the medium. Fibroblasts were then left in fresh medium for 96 h. After 96 h, the supernatant was collected (abbreviated in this paper as Hp-AGF supernatant). The same procedure was applied to the control, normal gastric fibroblast culture (herein abbreviated as GF). To confirm the absence of viable bacteria, the Hp-AGF supernatants were applied on the plates with Columbia Agar with 5% fresh horse blood (BioMerieux, Marcy l'Etoile, France). The plates were incubated under microaerophilic conditions at 37 • C for up to 7 days. No visible signs of bacterial growth were detected. Additionally, the Hp-AGF secretome-target epithelial cells were checked for Hp 16S rRNA expression. All supernatants were stored in 4 • C up to two weeks.

Long-Term Influence of Supernatants Collected from Hp-AGFs and GFs on Epithelial RGM1 Cells
RGM1 rat gastric epithelial cells [38,39] were cultured in 5 mL of neat Hp-AGF supernatant for 96 h on 6-well plates [22,37], then trypsinized (standard trypsinization technique) and resuspended in DMEM with antibiotics and without FBS (Sigma-Aldrich, Saint Louis, MO, USA). Then, the cells were seeded on the upper side of Transwell inserts containing native microporous membranes-pore diameter: 8 µm (Corning, Corning, NY, USA), at a density of 1 × 10 4 and allowed to transmigrate into the Hp-AGF supernatant during 24 h. The most activated post-EMT RGM1 cells which successfully transmigrated through the cell culture inserts were then checked for E-and N-cadherin expression, the indicators of the EMT. E-cadherin negative, N-cadherin positive post-EMT RGM1 cells (abbreviated in our present work as s.t.EMT + RGM1 cells) were then cultured in the supernatant from Hp-AGFs for at least 30 days (described in our present work as long-term RGM1 cells originating from EMT-positive short-term RGM1 cells l.t.EMT + RGM1 cells) in 25 cm 2 flasks. The neat Hp-AGF supernatant was added in an amount of 4 mL and changed every 3 days. Simultaneously, RGM1 cells cultured for 96 h in GF supernatant (abbreviated as s.t.EMT − RGM1 cells) were then cultured in GF supernatant for at least 30 days (described in our present study as long-term RGM1 cells originating from short-term EMT negative RGM1 cells (l.t.EMT − RGM1 cells). RGM1 cells were cultured in DMEM + 10% FBS (RGM1) for identical period of time (Scheme 1). During all experiments, the amounts of the media as well as the procedures applied to l.t.EMT + RGM1 cells, l.t.EMT − RGM1, and RGM1 cells were standardized. The cells were then harvested and used for the mRNA expression analysis by RT-PCR and for protein expression (immunofluorescence and Western blot analysis).

RT-PCR Technique
Total cellular RNA was isolated according to the Chomczynski and Sacchi method [40] using Trizol Reagent (Invitrogen, Carlsbad, CA, USA). First strand cDNA was synthesized from total cellular RNA (2 µg) using Reverse Transcription System (Promega, Madison, WI, USA). Bacterial DNA was isolated from Hp strain 43504 using DNAzol Reagent (Life Technologies, Carlsbad, CA, USA) according to the manufacturer's protocol. The PCR was carried out from l.t.EMT + RGM1, l.t.EMT − RGM1, and RGM1 cells, using 1 µg cDNA and Promega PCR reagent. Specific primers for Ki67, Cyclin D1, Bax, Bcl2, c-Myc, Oct4, Sox2, Klf4, p53, TGFβR1, and TGFβR2 and 18S RNA (the verification of the RNA integrity) and 16S bacterial RNA were used (Sigma-Aldrich, Saint Louis, MO, USA). Sequences and annealing temperatures are listed in Table 1. PCR products were separated by electrophoresis in 2% agarose gel containing 0.5 µg/mL ethidium bromide and then visualized under UV light. Location of predicted PCR product was confirmed by using O'Gene Ruler 50 bp DNA ladder (ThermoFisher, Waltham, MA, USA) as standard marker. Table 1. Rat oligonucleotide primers for detection of mRNA and Hp oligonucleotide primer for detection of 16S rRNA by RT-PCR and PCR, annealing temperature and size of PCR products employed in the experimental protocol.

Cell Proliferation
For proliferation assays, l.t.EMT + RGM1, l.t.EMT − RGM1, and RGM1 cells were seeded onto 24-well plates at the density of 0.5 × 10 3 cells/cm 2 and incubated in 2 mL of Hp-AGF supernatant, GF supernatant, and DMEM + 10% FBS, respectively, for 24, 48, 72, and 96 h. Cells were then harvested with trypsin and counted with Bürker's hemocytometer. The same procedure was applied to l.t.EMT + RGM1, l.t.EMT − RGM1, and RGM1 cells with the addition of SB-431542 (10 µM/L) (Sigma-Aldrich, Saint Louis, MO, USA). Additionally, the l.t.EMT + RGM1 cells were placed in Hp-AGF supernatant and in DMEM + 10% FBS and recorded for 12 h using the Leica DMI6000B time-lapse system equipped with a temperature/CO 2 chamber, the interference modulation contrast optics, and a cooled, digital DFC360FX CCD camera. Then, the number of cell divisions was referred to the total cell count.

Determination of the Cells Ability to Release TGFβ1
L.t.EMT + RGM1, l.t.EMT − RGM1, and RGM1 cells were placed on the 6 well plates in DMEM + 10% (0.3 × 10 6 cells per well) and allowed to release TGFβ for 48 h. Then, the cells supernatants were collected and the concentration of TGFβ1 was measured by ELISA kit no. orb50104 (Biorbyt, St. Louis, MO, USA) according to manufacturer protocol.

Image Acquisition and Immunofluorescence
Image acquisition was performed with a Leica DMI6000B microscope equipped with the total internal reflection fluorescence (TIRF), Nomarski interference contrast (DIC) modules. LAS-AF deconvolution software was used for image processing. For phenotype plasticity L.t.EMT + RGM1, l.t.EMT − RGM1 and RGM1 cells were trypsinized and seeded at 8 × 10 6 cells per well and each of them were cultured for 24 h in all 3 types of media: Hp-AGF supernatant, GF supernatant, and DMEM + 10% FBS. Then, the cells were photographed.

Statistical Analyses
Statistical analysis of the data was performed with the use of Excel Software. Each variable was expressed as the mean (±S.E.M.). Statistical significance of difference was determined using analysis of variance (one-way ANOVA) test (Statistica Software). Further statistical analysis for post hoc comparisons was carried out with the Newman-Keuls test. Differences were considered statistically at p < 0.05.

Long-Term Influence of Hp-AGF Secretome Induces Phenotypical Changes in Pro-Invasive s.t.EMT + RGM1 Cells
We have previously shown that Hp activates paracrine loops between gastric fibroblasts and epithelial cells that result in the phenotypic shifts of RGM1 cells towards rear-front polarized, motile phenotype [14]. To estimate the long-term effects of these loops on the phenotype of RGM1 cells, we allowed these cells to transmigrate through microporous membranes [22]. The cells that most readily (24 h) penetrated the pores were characterized for E-and N-cadherin expression (s.t.EMT + RGM1; Figure 1A,B). These cells showed cadherin switch from E-to N-cadherin, which is characteristic for the EMT process. The progeny of these cells was then allowed to expand in the presence of the secretome from Hp-AGFs for 30 days (l.t.EMT + RGM1; Figure 1C). At the same time, the RGM1 cells were cultured for 96 h in normal gastric fibroblast (GF) supernatant ( Figure 1D). In these cells, N-cadherin expression was almost absent, while E-cadherin was expressed (s.t.EMT − RGM1; Figure 1A), which remains in agreement with our previous results [22]. Their progeny was allowed to expand in GF supernatant for 30 days. Such long-term incubation in the presence of GF secretome enabled us to establish a fibroblastoid RGM1 lineage (l.t.EMT − RGM1; Figure 1E). Western blot analysis revealed the induction of N-cadherin expression in these cells ( Figure 1F). Analysis of α-SMA expression revealed that these cells incorporated α-SMA into stress fibers, indicating their myofibroblastic phenotype and GF secretome-induced epithelial-myofibroblast transition (EMyoT; Figure 1G). Concomitantly, an induction of N-cadherin observed after the short-term incubation (s.t.EMT + RGM1 cells) in Hp-AGF-conditioned medium ( Figure 1A) was followed by its down-regulation in the cells that underwent the long-term incubation (l.t.EMT + RGM1; Figure 1F). Moreover, α-SMA did not incorporate into stress fibers in l.t.EMT + RGM1 cells ( Figure 1G). These phenotypic shifts were accompanied by the increased TGFβ1 levels in Hp-AGF secretome [22]. Notably, l.t.EMT + RGM1 cells were able to release detectable amounts of TGFβ1, as documented for cells allowed to secrete factors in DMEM + 10% FBS for 48 h by ELISA tests ( Figure 1H). 3.2. The Phenotypic Plasticity of l.t.EMT + RGM1 Cells L.t.EMT + RGM1 cells were characterized by plastic phenotype, which was modified by GF secretome. In the presence of Hp-AGF secretome, l.t.EMT + RGM1 acquired elongated mesenchymal-like shape with long, thin protrusions ( Figure 2A). In the presence of GF secretome, l.t.EMT + RGM1 cells retained the elongated, mesenchymal shape with long protrusions or acquired epithelial-like phenotype similar to that of untreated cells (Figure 2A,D). This effect was more pronounced in DMEM + 10% FBS (  The cells were counted according to cell morphology (non-polarized epithelioid cells vs rear-front polarized fibroblastoid cells). Blue arrows: sample fibroblastoid cells, red arrows: sample epithelioid cells. Results are mean ± SEM of three to six different independent experimental repeats. Asterisk indicates a significant difference between epithelial and fibroblastoid populations (p < 0.05).

TGFβR1 Activation Participates in Both Pro-Pluripotent and Pro-Fibrotic RGM1 Reprogramming
The presence of TGFβ1 in Hp-AGF-conditioned medium prompted us to identify the role of this cytokine in the regulation of Hp-AGF-reprogrammed cell plasticity. High levels of LGR5 were retained in l.t.EMT + RGM1 cell lineage ( Figure 3A). However, they were accompanied by the altered (in comparison to naive RGM1 cells) expression of Yamanaka factors at the mRNA level ( Figure 3B).
LGR5 + l.t.EMT + RGM1 cells were characterized by the induction of Oct4 mRNA expression, while it was significantly lower in l.t.EMT − RGM1 cells and almost absent in RGM1 cells ( Figure 3B) pointing to CSC-like phenotype of l.t.EMT + RGM1 cells. Concomitantly, l.t.EMT + RGM1 cells retained considerable Sox2 mRNA expression (the factor responsible for maintaining progenitor cell self-renewal) and slightly lowered, but still pronounced expression of c-Myc. Additionally, l.t.EMT + RGM1 cells displayed strongly down-regulated Klf4 encoding mRNA levels ( Figure 3B). Thus, l.t.EMT + RGM1 cells were characterized by overall LGR5 + /Oct4 high /Sox2 high /c-Myc high /Klf4 low pattern ( Figure 3B). The involvement of TGFβ and TGFβR1 activation in the stem-like commitment of l.t.EMT + RGM1 cells was further confirmed by the sensitivity of Oct4, Sox2, c-Myc, and Klf4 to the chemical inhibition of TGFβR1 signaling by SB431542. Consequently, the cells switched to overall Oct4 low /Sox2 low /c-Myc high /Klf4 high pattern ( Figure 3C,D). On the other hand, LGR5 − l.t.EMT − RGM1 displayed high Oct4 mRNA expression levels, accompanied by low Sox2 and c-Myc mRNA expression as well as Klf4 gene up-regulation (overall Oct4 + /Sox2 − /c-Myc − /Klf4 + phenotype; Figure 3B). Again, Oct4, Sox2, c-Myc, and Klf4 mRNA levels were sensitive to TGFβR1 inhibition by SB431542, which resulted in overall Oct4 − /Sox2 + /c-Myc + /Klf4 + pattern ( Figure 3C,D). Collectively, these data confirm that Hp-AGF and naive GF secretome induce alternative reprogramming scenarios, which at least partly depend on TGFβR1 signaling and result in differential LGR5/Oct4/Sox2/c-Myc/Klf4 expression pattern. Results are mean ± SEM of three to six independent experimental repeats. Asterisk indicates a significant change (p < 0.05) as compared to the control RGM1 value. Hash indicates a significant change (p < 0.05) as compared to the corresponding values without SB-431542.

Hp-AGF Secretome Induces Potentially Pro-Proliferative Phenotype of l.t.EMT + RGM1 Cells
We have checked the pro-proliferative potential of l.t.EMT + RGM1 cells cultured in Hp-AGF supernatant, l.t.EMT − RGM1 cells cultured in GF supernatant, and RGM1 cells cultured in DMEM + 10% FBS by evaluation of MEK1 protein expression. Western blot analysis showed highly accelerated MEK1 expression in l.t.EMT + RGM1 cells (Figure 4).  Figure 5B). Thus, anti-apoptotic effect of TGFβR1, which is present in Hp-AGF secretome, may account for the propagation of l.t.EMT + RGM1 cells under the selective pressure of Hp-AGF secretome. Furthermore, slight inhibition of l.t.EMT + RGM1 proliferation ( Figure 5C) was accompanied by the up-regulation of Bax upon SB431542 administration ( Figure 5D,E). Together with the effects on the expression of Ki67, Cyclin D1 mRNA, and less pronounced effect on Bcl2 mRNA ( Figure 5D,E). These findings confirm the important role of TGFβR1 in this process. The percentage of dividing l.t.EMT + RGM1 cells allowed proliferating in TGFβ1 rich Hp-AGF secretome and in DMEM + 10% FBS for 48 h was significantly higher in DMEM ( Figure 5F). In GF-conditioned medium, l.t.EMT − RGM1 cells expressed considerably lower mRNA levels encoding Ki67 and Cyclin D1 than their l.t.EMT + RGM1 counterparts ( Figure 5B), which correlated with their lower proliferation rate ( Figure 5A). In contrast to l.t.EMT + RGM1 cells, l.t.EMT − RGM1 cells reacted to SB-431542 with the induction of proliferation ( Figure 5C), correlated with the Ki67 and Cyclin D1 gene up-regulation ( Figure 5D,E). The expression of Bax mRNA underwent slight increase, while the increase of Bcl2 mRNA was pronounced, which points to the inhibition of apoptosis ( Figure 5D,E). Thus, TGFβR1 participates in the induction of anti-apoptotic effect in l.t.EMT + RGM1 cells, while it inhibits proliferation and induces apoptosis of l.t.EMT − RGM1 cells. Collectively, our data confirm that Hp-AGF secretome prompts the enhanced proliferation of l.t.EMT + RGM1 under TGFβ1 stress, thus potentially facilitating Hp-induced gastric neoplasia.

Interrelations between TGFβR1 and TGFβR2 Activity Underlie Differential Microevolution Pattern of l.t.EMT + RGM1 and l.t.EMT − RGM1 Cells
Apparently, secretome of gastric fibroblasts can switch between alternative developmental programs in RGM1 cells, depending on gastric fibroblast activation by Hp. Further experiments were performed to address the TGFβ-related mechanisms underlying differential reactivity of l.t.EMT + RGM1 and l.t.EMT − RGM1 cells to Hp-AGF and GF secretomes. Even though TGFβR1 mRNA levels were slightly down-regulated in l.t.EMT + RGM1 cells, TGFβR1 and TGFβR2 genes were activated in all tested cell types, both in the presence/absence of Hp-AGF secretome ( Figure 6A). A more pronounced down-regulation of TGFβR2 mRNA expression in l.t.EMT + RGM1 cells ( Figure 6A) was accompanied by the presence of two bands in Western blot analysis ( Figure 6B). The lower band (ca. 65 kDa) had the molecular mass of unmodified TGFβR2 protein, whereas the second (75-80 kDa) corresponded to the glycosylated (active) TGFβR2 [41] ( Figure 6B). These data point to partial inactivation of TGFβR2 in l.t.EMT + RGM1 cells as indicated by decreased levels of glycosylated TGFβR2 compared to l.t.EMT − RGM1 and RGM1 cells ( Figure 6B). l.t.EMT + RGM1 cells were also characterized by decreased p53 mRNA level ( Figure 6B). The data indicate that the collective TGFβR1/2 activation determines the expansion of pro-invasive RGM1 cells. Apparently, remaining TGFβR1 activity along with TGFβR2 stimulation may negatively affect the proliferation of l.t.EMT − RGM1 cells in the presence of GFs secretome. This mechanism is potentially involved in the regulation of l.t.EMT − RGM1 cell stromal phenotype. In turn, TGFβ1 can induce pro-neoplastic expansion of l.t.EMT + RGM1 cells under conditions of the reduced TGFβR2 signaling. These findings underline the importance of TGFβ signaling balance for the regulation of phenotypic fate of gastric epithelial cells.

Discussion
Tumors comprise heterogeneous cell populations with distinct phenotypes, functions, and gene expression profiles [42]. Currently, a growing body of evidence supports the view that gastric cancer is initiated by single self-renewing CSC-like cells [30,43,44]. CSCs have the potential to self-renew and differentiate, thus initiating and sustaining the primary tumor growth and establishing metastases [33,45,46]. However, CSC origins still remain a matter of debate. During tumor development, CSCs cooperate with stromal fibroblasts that constitute the milieu for neoplastic loci and the barriers for immune system, concomitantly providing paracrine signals for tumor promotion [18,[26][27][28][29]. We have previously reported that normal gastric epithelial RGM1 cells undergo EMT-like phenotypic shifts, accompanied by inhibition of proliferation upon short-term (96 h) exposure to the secretome from Hp-infected gastric fibroblasts (Hp-AGFs) [22]. Here, we demonstrate that the prolonged cultivation of these post-EMT RGM1 cells in the presence of Hp-AGF secretome induces their reprogramming towards overall LGR5/Oct4 high phenotype, characteristic for phenotypically plastic stem-like neoplastic cells. In turn, the induction of RGM1 reprogramming towards the stromal α-SMA+ phenotype was seen in the presence of non-infected gastric fibroblast (GF) secretome, confirming the role of Hp in biasing the GF secretome towards pro-EMT activity. Apparently, TGFβR1/TGFβR2-dependent signaling switches participate in these alternative scenarios. Thus, paracrine loops within pre-cancerous loci govern the developmental fate of gastric epithelium via directing epithelial cells towards neoplastic or stromal differentiation program in aTGFβR1/R2-dependent/Hp regulated manner.
Apparently, "trans differentiation" processes and the induction of pluripotency factors promote the phenotypic plasticity of the cells within pre-neoplastic loci that facilitates the adaptation of alternative differentiation programs by gastric epithelial cells. We show that RGM1 cells can undergo alternative reprogramming events, depending on the Hp-modulated quality of humoral factors secreted by gastric fibroblasts. In the long-term presence of Hp-AGF secretome, RGM1 cells evolve towards the plastic Oct4 + /LGR5 + phenotype (l.t.EMT + RGM1). Oct4 stands at the top of a hierarchy governing the regulation of both pluripotency and de-differentiation [30,[47][48][49]. In addition, LGR5 + GC cells have previously been shown to display stem cell-like features [50,51] and to regulate Oct4 levels. The overall LGR5/Oct4 expression in phenotypically plastic l.t.EMT + RGM1 cells suggests their stem-cell like pluripotency commitment, which is potentially relevant for tumor development [30,52,53]. Thus, Hp induces fibroblast activation towards CAFs [35], which secrete factors increasing LGR5/Oct4-dependent stem cell-like potential of epithelial RGM1 cells, while non-infected fibroblasts are more predisposed to direct gastric epithelium towards the stromal (pro-fibrotic) phenotype. Such microevolution of fibroblastoid cells with overall Oct4 + /LGR5 − phenotype was observed in RGM1 populations exposed for long periods to the GF secretome. l.t.EMT − RGM1 EMyoT-related phenotypic commitment is illustrated by α-SMA incorporation into stress fibers, l.t.EMT − RGM1 relative dormancy and stable (niche-secretome independent) myofibroblastoid phenotype.
Induction of complementary developmental scenarios in gastric epithelium may have a profound significance for Hp-induced initiation of gastric tumorogenesis. Nevertheless, further studies are necessary to address the multipotency and invasiveness of l.t.EMT + RGM1 cells, their Oct4 + /LGR5 + expression accompanied by transient N-cadherin up-regulation, the post-EMT morphology/motility and apparent phenotypic plasticity suggest their CSC-like properties. Thus, we show the interrelations between mesenchymal phenotype and plasticity of cancer cells in the in situ GC model. Incomplete, partial EMT has previously been shown to account for the phenotypic plasticity (the ability to produce more than one phenotype) of cancer cells [24,30,33,[54][55][56][57][58][59]. These interrelations also participate in the EMT/MET-dependent GC progression, where epithelioid cancer cells undergo EMT to enter the metastatic cascade, whereas post-EMT CSCs undergo MET-directed differentiation after metastatic niche nesting [24,30,60,61]. Gastrointestinal CSCs were shown to have elevated expression levels of genes associated with stemness and pluripotency. Apart from Lgr5, Oct4, these include c-Myc, Sox2, and Klf4 [30,[62][63][64][65][66] discrete overall expression patterns of these factors in l.t.EMT + RGM1 and l.t.EMT − RGM1 cells confirm their involvement in the alternative EMT/EMyoT-related differentiation scenarios induced in gastric epithelial cells by their paracrine looping with Hp-AGFs/GFs, respectively. Apparently, prominent Klf4 and less pronounced c-Myc down-regulation facilitates pre-neoplastic phenotype of otherwise Oct4 + /LGR5 + l.t.EMT + RGM1 cells, whereas LGR5/Sox2/c-Myc overall down-regulation is linked to pro-fibrotic/stromal phenotype of l.t.EMT − RGM1 cells. Consequently, these key modulators of pluripotency [33,[67][68][69] are closely linked not only to Hp-AGF-induced EMT of gastric epithelial cells but also participate in their non-infected fibroblast-induced EMyoT.
It has also been stated, that MEK1 and MEK2, which are related protein kinases involved in the Ras/Raf/MEK/ERK signal transduction cascade, participate in the regulation of cell cycle progression, cell differentiation, and proliferation [70], which suggests that the effect of the supernatant on proliferation rate could not account for the nutrient depletion by fibroblasts. The Ras-dependent ERK1/2 MAP kinase signaling pathway plays a central role in cell proliferation control and is frequently activated in different human cancers. Early studies have shown that expression of activated alleles of MEK1 is sufficient to deregulate the proliferation and trigger the morphological transformation of cell lines [70][71][72][73][74]. Moreover, ERKs have been shown to be vital for CSC tumorogenicity [75].
Previously, short-time (96 h) incubation of RGM1 cells in the presence of Hp-AGF secretome has been shown to induce TGFβR1/R2 mRNA up-regulation in these cells (especially TGFβR2) [37]. It is well established that TGFβ1 induces EMT during pathologic and physiologic fibrosis of normal tissues [22][23][24][25]37,[76][77][78][79][80][81][82]. Pleiotropic effects of TGFβ1 are also observed in cancer systems, where TGFβ signaling can suppress or promote tumor initiation [80][81][82] and progression [83][84][85]. Relatively high Lgr5/Oct4/Sox2 mRNA levels, accompanied by p53 down-regulation [86,87], high proliferation rate, and TGFβ1 secretion by l.t.EMT + RGM1 cells indicate that TGFβ1 is involved in the induction of their CSC-like/proliferative state. The p53 down-regulation in l.t.EMT + RGM1 cells may sustain the activity of TGFβ1 signaling in l.t.EMT + RGM1 cells at the levels promoting their viability. Furthermore, Klf4 down-regulation in l.t.EMT + RGM1 cells remains in concordance with this concept, as Klf4 promotes differentiation through TGFβ1 target gene activation and myofibroblast formation [88]. In a parallel manner, Klf4 suppresses oncogenic TGFβ1 signaling [89] via interactions with TGFβ1 control elements (TCEs) [90]. TGFβ1 is considered an immune biomarker which has also been linked to myocardial function and remodeling due to the regulation of valve fibrosis and calcification processes [91]. Notably, Klf4 has been shown to be down-regulated during tumor initiation and consecutively lost during GC progression [92][93][94]. Furthermore, the involvement of TGFβ1 in the regulation of pro-neoplastic l.t.EMT + RGM1 phenotype is evidenced by the shifts in overall Oct4/c-Myc/Sox2/Klf4 mRNA levels following TGFβR1 inhibition. Additionally, anti-apoptotic activity of TGFβR1 is evidenced by some slight inhibition of l.t.EMT + RGM1 cell proliferation and Bax up-regulation induced by SB-431542. Thus, Hp-AGFs help to sustain the EMT-related CSC commitment of RGM1 cells and their viability in a TGFβ1-mediated, TGFβR1-dependent, and Hp-modulated manner. This hypothesis was further confirmed by relatively strong impact of TGFβR1 inhibition on Oct4/c-Myc and Sox2 expression pattern in l.t.EMT − RGM1 cells. This finding indicates that TGFβR1 signaling remains active in these cells even in the environment characterized by lower level of TGFβ1 [22] and is necessary to sustain their myofibroblastic phenotype. This observation was accompanied by SB431542-induced l.t.EMT − RGM1 proliferation, which was correlated with Ki67, Cyclin D1, and Bax mRNA up-regulation back to the levels characteristic for naive RGM1 cells. The suppression of nuclear translocation of β-catenin into gastric cells along with the expression of the β-catenin target survival genes c-myc and cyclin D1 along with induction of apoptosis has recently been implicated as a potential mechanism of anti-cancer therapy [95]. Collectively, these observations confirm that non-infected GF secretome contains the range of TGFβR1 ligands that determine the array of developmental fates undertaken by RGM1 cells [96]. In conjunction with apparently opposite (inducing) effects of TGFβ1/TGFβR1 on the phenotype/proliferation of l.t.EMT + RGM1 cells, this observation also indicates that these ligands cooperate with TGFβ1 in biasing RGM reprogramming towards pro-neoplastic phenotype, while prompting EMyoT in the lower concentrations of TGFβ1. Relatively low TGFβR2 expression/activity and p53 impairment [87,97,98] in l.t.EMT + RGM1 cells may suggest the role of TGFβR1/R2 interactions in the differential effects of Hp-AGF/GF secretome on RGM1 phenotype. Accordingly, the combination of TGFβ1 with the other TGFβR influencing ligands impairs TGFβR2 activity, prompting the neoplastic program in RGM1 cells, whereas their myofibroblastic differentiation is induced in the low level of TGFβ1/presence of active TGFβR2. These findings need further research comprising direct and indirect influence of Hp on reciprocal interactions between different cellular components of gastric tissue taking place in vivo.

Conclusions
Collectively, our data confirm the primary role of the signals from Helicobacter pylori in the remodeling of gastric niche towards pro-neoplastic activity. In the absence of Hp, occasional lesions can temporarily help to establish potential paracrine loops between gastric stromal fibroblasts and epithelial cells, induce EMyoT in gastric epithelium, and prompt local tissue regeneration. On the perimeter of chronic Hp-induced lesions, where gastric fibroblasts are activated by local inflammation but not directly exposed to Hp signals, EMyoT of gastric epithelium can turn into the chronic state. On the other hand, direct Hp-generated stress results in further reprogramming of gastric fibroblasts and stabilization of paracrine loops between Hp-infected stromal fibroblasts possessing CAF-like characteristic [35] and gastric epithelium. Hp-infected stromal fibroblasts prompt reprogramming of normal gastric epithelial cells towards precancerous phenotype. These events are related to TGFβ1 release by Hp-AGF/epithelium, which modulates TGFβR1/R2-dependent signaling, induces EMT in non-cancerous epithelial cells, consequently prompting their pro-CSC-like phenotype. Consequently, the post-EMT cells are reprogrammed towards TGFβ1-dependent expansion and CSC-like commitment, providing the basis for their further EMT/MET plasticity. The Hp-activated fibroblast reprogramming of epithelial cells may thus potentiate the canonical pathway of Hp infection influence on epithelial cells.