The gastric cancer cell lines AGS (ATCC CRL-1793), MKN74 (JCRB0255), and KATO III (ATCC HTB-103) were cultured in RPMI 1640 medium (Gibco) supplemented with 10% Fetal Bovine Serum (Biological Industries) and antibiotics (10,000 U/ml penicillin, 10 μg/ml streptomycin). Cells were incubated at 37°C in a humidified atmosphere with 5% CO₂.

The completely sequenced H. pylori 26695 (ATCC 700392) strain was cultured in trypticase soy agar plates supplemented with 5% horse serum (Biological Industries), nutritive supplement Vitox (Oxoid) and selective supplement Dent (Oxoid) for 24 h at 37°C in a humidified atmosphere with 5% CO₂.

AGS cells were transfected with sh-RNA-Scramble plasmid (Santa Cruz Biotechnology #108065) or sh-RNA-HIF-1α plasmid (Santa Cruz Biotechnology #35561) using the reagent Fugene 6®, following instructions provided by the manufacturer (Promega, Madison, WI). Post-transfection, cells were grown in a selection medium containing 2 μg/ml puromycin for 2 weeks. Then, clones were isolated from batch-transfected cell populations in order to obtain cells with improved silencing of HIF-1α. A clone with effective silencing of HIF-1α (AGS sh-HIF-1α Clone7) and control clone (AGS sh-Scr Clone2) were selected (see Supplementary Figure 5) and used to perform the indicated experiments.

8 × 10⁵ cells were seeded in 60 mm culture plates and cultivated for 24 h. For infection assays, culture media was replaced by serum supplemented RPMI without antibiotics at least 4 h prior to bacterial exposure. The bacteria were harvested, washed, and resuspended in PBS. OD₅₆₀ = 0.4 was considered equivalent to 3 × 10⁸ bacteria. Cells were infected at the multiplicity of infection (MOI) indicated in each experiment.

To obtain the total protein extracts, cells were lysed with a buffer containing 20 mM HEPES pH 7.4, 12.5 μg/ml leupeptin, 10 μg/ml antipain, 100 μg/ml benzamidine, 1 mM phenylmethylsufonyl fluoride, 1 mM sodium orthovanadate and 10 mM sodium fluoride, 0.1% SDS, 0.05% NP-40 (IGEPAL). Then, lysates were sonicated and centrifuged. The resulting supernatants, referred to as total protein extracts, were stored at −20°C.

Nuclear and the cytoplasmatic fractions were essentially obtained as previously described (Valenzuela et al., 2014). Briefly, cells were lysed in a hypotonic buffer containing 10 mM Hepes-KOH pH 7.9, 2 mM MgCl₂, 0.1 mM EDTA, 10 mM KCl, 0.5% NP-40 (IGEPAL) and supplemented with 1 mM DTT, as well as protease and phosphatase inhibitors (12.5 μg/ml leupeptin, 10 μg/ml antipain, 100 μg/ml benzamidine, 1 mM phenylmethylsufonyl fluoride, 1 mM sodium orthovanadate and 10 mM sodium fluoride). The lysate was centrifuged at 16,000 × g to obtain the cytoplasmic protein fraction in the supernatant and the nuclear protein fraction in the pellet. The nuclear pellet fraction was solubilized in a hypertonic buffer containing 50 mM Hepes-KOH pH 7.9, 2 mM MgCl₂, 0.1 mM EDTA, 50 mM KCl, 400 mM NaCl, 10% glycerol, supplemented with 1 mM DTT, as well as the protease and phosphatase inhibitors indicated above. The resulting lysate was centrifuged at 16,000 × g to obtain the nuclear proteins in the supernatant.

Proteins (50–75 μg total per lane) were separated in 10% SDS polyacrylamide gels and transferred to a nitrocellulose membrane, as previously described (Valenzuela et al., 2010). Blots were blocked with 5% milk in 0.1%-Tween20 TBS or 5% BSA in 0.1%-Tween20 TBS for phosphorylated proteins. Then, blots were probed with either mouse anti-HIF-1α monoclonal antibody (BD Bioscience #610958, dilution 1:500), mouse anti-β-actin monoclonal antibody (Sigma #A5316, dilution 1:20,000), rabbit anti-phospho-S473-Akt polyclonal antibody (Cell Signaling Biotechnology #4060, dilution 1:1,000), rabbit anti-Akt polyclonal antibody (Cell Signaling Biotechnology #9272, dilution 1:1,000), mouse anti-Cyclin D1 monoclonal antibody (Santa Cruz Biotechnology #20044, dilution 1:1,000), rabbit anti-phospho-S235/236-S6 polyclonal antibody (Cell Signaling #48585, dilution 1:1,000), mouse anti-S6 monoclonal antibody (Cell Signaling Biotechnology #2317, dilution 1:1,000), mouse anti-LAP2A monoclonal antibody (BD Bioscience #611000, dilution 1:3,000) or rabbit anti-HSP90 polyclonal antibody (Santa Cruz Biotechnology #7947, dilution 1:3,000). Primary antibodies were detected with horseradish peroxidase-conjugated anti-mouse (Kirkergaard and Perry Laboratories #214-1806, dilution 1:5,000) or anti-rabbit (Kirkergaard and Perry Laboratories #214-1516, dilution 1:5,000) secondary antibodies and the ECL system.

Protein levels were essentially quantified as previously described (Valenzuela et al., 2013) by scanning densitometric analysis of Western blots using the software UnScan-it v. 6.1. Values obtained for the respective proteins of interest (HIF1a, Cyclin D1) were standardized to β-actin levels in the same lane. Results shown in graphs are averages from at least three different experiments with their respective standard errors of mean (SEM).

To analyze the HIF-1α transcriptional activity, 1.5 × 10⁵ cells were transfected with 0.25 μg of the respective plasmids: pGL3-HRE (HIF-1α reporter kindly provided by Dr. Kaye Williams, University of Manchester (Brown et al., 2006) and pON (β-galactosidase). Post-transfection, after 16 h, culture media was replaced by supplemented RPMI 1640 without antibiotics for inoculation with bacteria. After infection, cells were lysed in a buffer containing 0.1 M KH₂PO₄ (pH 7.9), 0.5% Triton X-100, and 1 mM DTT. The resulting lysate was centrifuged at 1640 × g and supernatants were used to measure either luciferase or β-galactosidase activities. Luciferase activity was detected using a luminometer (Synergy HT Multi-Mode Microplate reader, BioTek Instruments, Winooski, VT) after addition of KTME buffer containing 100 mM Tris HCl, 10 mM MgSO₄, 2 mM EDTA, 5 mM Na₂ATP, and 0.1 mM luciferin. β-galactosidase activity was determined by spectrophotometry (410 nm) after addition of substrate 2-nitrophenyl-β-D-galactopyranoside (ONPG). The values reported for luciferase activity were standarized to β-galactosidase activity.

Total RNA was isolated with the reagent TriZOL® following instructions provided by the manufacturer. RNA samples were spectrophotometrically quantified, digested with DNAse I and characterized by electrophoresis in 2% agarose gels (quality control) and then used as templates to generate cDNA.

The real-time quantitative PCR (qPCR) analysis was performed using the Brilliant II SYBR® Green QPCR Master Mix (Agilent Technologies, La Jolla, CA), cDNA template and the following primers: VEGFA: sense primer 5′-ATCCGGGTTTTATCCCTCTTC-3′ and anti-sense primer 5′-TCTCTCTGGAGCTCTTGCTA-3′; GLUT-1: sense primer 5′- AAGGAAGAGAGTCGGCAGAT-3′ and anti-sense primer 5- TCGAAGATGCTCGTGGAGTA-3'; Bcl-xL: sense primer 5- CACTAACCAGAGACGAGACT-3′ and anti-sense primer 5′- GTCCTGTTCTCTTCCACATC-3′; LDHA: sense primer 5′- ACGTCAGCAAGAGGGAGAA-3′ and anti-sense primer 5′- TCTTCCAAGCCACGTAGGT-3′; GAPDH: sense primer 5′- GCGAGATCCCTCCAAAATCA-3′ and anti-sense primer 5′- ATGGTTCACACCCATGACGA-3′; Cyclin D1: sense primer 5′-GGTGAACAAGCTCAAGTGGA-3′ and anti-sense primer 5′-GAGGGCGGATTGGAAATGAA-3′; RPS13 housekeeping gene: sense primer 5′-CTCTCCTTTCGTTGCCTGAT-3′ and anti-sense primer 5′- TGAAGGAGTAAGGCCCTTCT-3′. The PCR reactions were carried out using the Agilent Mx3000P QPCR System (Agilent Technologies, La Jolla, CA) following suggestions of the manufacturer. The relative gene expression levels were calculated by the 2^−ΔΔCT method (Livak and Schmittgen, 2001). VEGFA, GLUT-1, Bcl-xL, LDHA, GAPDH, and Cyclin D1 levels mRNA levels were normalized to RPS13 mRNA (housekeeping gene). All data were expressed relative to values obtained for non-infected cells.

After infection, cells were incubated with 200 μg/ml gentamycin at 37°C for 1 h to eliminate adherent bacteria. Then, cells were washed, harvested and fixed in chilled methanol at −20°C for 10 min. Fixed cells were treated with 100 μg/ml RNAase A for 30 min and then stained with 10 μg/ml propidium iodide. Cells were analyzed by flow cytometry (FACS Canto, BD) and the DNA content was quantified as propidium iodide fluorescence intensity histograms using the FACSDiva program version 6.1.3.

The PI3K inhibitor LY294002 (Enzo Life Science #BML-ST420-005) was used at concentrations of 1, 5 and 10 μM; the Akt inhibitor VIII (Calbiochem #124018) was used at 1 μM and the mTOR inhibitor Rapamycin (Sigma #R8781) at 50 nM. The inhibitors were added to culture media 30 min previous to infection. All inhibitors were solubilized in DMSO.

To determine the Cyclin D1 half-life, the protein synthesis inhibitor cycloheximide (CHX) 100 μg/ml (Sigma #C0934) was added to infected cells after 8 h of exposure to H. pylori and protein extracts were then prepared at the indicated time points after CHX addition.

To determine the role of the proteasome in Cyclin D1 decrease, the proteasome inhibitor MG132 (20 μM) (Enzo Life Science #BML-PI102-005) was added to infected cells after 8 h of infection, in the absence and presence of protein synthesis inhibitor CHX, for 15 min. Then, protein extracts were prepared for western blot analysis.

Results from at least 3 independent experiments were analyzed by applying 1-way ANOVA analysis and Dunnett post-test to compare multiple conditions or Student's t-test to compare two conditions, using the GraphPad Prism software (GraphPad Software, San Diego, CA). P < 0.05 was considered significant.

Previously, H. pylori infection of gastric cells has been linked to HIF-1α induction via ROS-dependent mechanisms, VEGF production and angiogenesis (Park et al., 2003; Bhattacharyya et al., 2010; Kang et al., 2014). However, alternative pathways reportedly also contribute to HIF-1α induction in normoxia, such as activation of PI3K/Akt. Here we evaluated the possibility that PI3K/Akt activation by H. pylori might favor HIF-1α induction and sought to identify the cell responses associated with HIF-1 activation. To this end, AGS cells were infected with H. pylori 26695 at MOI 100 and the induction of HIF-1α protein levels, as well as its intracellular localization were evaluated by western blotting following infection. HIF-1α protein levels increased until 8 h, but then decreased to essentially basal levels after 16 h of H. pylori exposure (Figure 1A). Consistent with this, the nuclear localization of HIF-1α, analyzed by nucleus/cytoplasm fractionation, increased after 8 h, but then decreased at 24 h post-infection (Figure 1B). Similar results were obtained by indirect immunofluorescence analysis of HIF-1α subcellular localization (Supplementary Figure 1). Likewise, H. pylori-induced HIF-1α expression also was observed in MKN74 and KATO III gastric cells infected for 8 h, albeit to varying extents (Supplementary Figure 2). We then determined whether HIF-1α induction and translocation to the nucleus were linked to enhanced transcriptional activity, using a luciferase reporter assay. In AGS cells infected with H. pylori a MOI-dependent increase in HIF-1 reporter activity was observed after 8 h (Figure 1C).

H. pylori infection has been linked to PI3K/Akt activation, which regulates many cellular responses, such as cell cycle progression and proliferation (Nakayama et al., 2009; Suzuki et al., 2009; Tabassam et al., 2009). In normoxia, activation of this pathway downstream of growth factor receptors can induce HIF-1α. Thus, we wondered whether a similar connection might exist following H. pylori infection. To that end, the levels of p-Ser473-Akt were evaluated as an indicator of PI3K activity at different time points following infection with H. pylori. For AGS cells infected with H. pylori 26695 at MOI 100, a substantial increase in p-Ser473-Akt levels was detected after 4 and 8 h of infection, which then decreased 16 and 24 h post-infection (Figure 2A). These changes paralleled those observed for HIF-1α protein levels following H. pylori infection (Figure 1A). Thus, PI3K/Akt activation and HIF-1α induction occurred with similar kinetics following H. pylori infection.

To evaluate whether HIF-1α induction was due to PI3K activation, cells were incubated with LY294002, a selective pharmacological inhibitor of PI3K. In AGS cells infected by H. pylori, LY294002 (at 5 and 10 μM) reduced significantly HIF-1α induction observed 4 and 8 h post-infection (Figures 2B,C, respectively). Under these conditions, LY294002 prevented PI3K activation by H. pylori, since phosphorylation of Akt S473 was almost completely abolished (Figures 2B,C). Alternatively, the adhesion of H. pylori to gastric cells was not altered in the presence of the PI3K inhibitor (Supplementary Figures 3A,B), suggesting that the observed results with HIF-1α were not attributable to non-specific effects of the inhibitor on the bacterial adherence to AGS cells. In addition, infection in the presence of the PI3K inhibitor did not affect the translocation of virulence factors to the host cells, since intracellular protein levels of CagA, a virulence factor secreted by the bacteria into the host cell, was not modified by PI3K inhibition (Supplementary Figure 4A). Also, immunoprecipitation of intracellular CagA showed that phosphorylation on tyrosine of CagA, which occurs intracellularly, was not affected by presence of the PI3K inhibitor (Supplementary Figure 4B).

In fibroblasts, as well as breast and colon cancer cells, stimuli leading to PI3K activation increase HIF-1α protein levels through the downstream activation of Akt and mTOR (Laughner et al., 2001; Fukuda et al., 2002). To determine which elements downstream of PI3K were required for H. pylori-mediated HIF-1α induction, cells were incubated with the mTOR inhibitor Rapamycin (50 nM) during infection by 4 h or 8 h and mTOR inhibition was assessed by determining the levels of S6 phosphorylation on S235/236 (p-S235/236-S6) due to S6K, a direct target of mTOR (Semenza, 2012). Indeed, Rapamycin for either 4 h or 8 h effectively inhibited mTOR activity and precluded the H. pylori-induced HIF-1α protein levels in AGS cells (Figures 3A,B, respectively). Again, as a control, Rapamycin (50 nM) was shown neither to affect H. pylori adhesion to AGS cells (Supplementary Figure 3C,D), or the phosphorylation of immunoprecipitated CagA (Supplementary Figure 4C). Interestingly, the Akt inhibitor (1 μM) was not significantly effective in reducing HIF-1α induction following 4 or 8 h H. pylori infection (Figures 3C,D, respectively), suggesting that mTOR activation downstream of PI3K is the major pathway responsible for HIF-1α induction following H. pylori infection.

Since our results showed an increase in the nuclear localization of HIF-1α and an increase in the reporter activity after 8 h of infection by H. pylori (Figures 1B,C), we next evaluated whether some of the known target genes of HIF-1α were induced by H. pylori infection. Unexpectedly, H. pylori promoted only few and rather modest changes in the mRNA levels of known downstream target genes of HIF-1α. Specifically, GLUT-1, LDHA, and GAPDH did not change, while for Bcl-xL a trend but not a significant increase was observed (Figure 4). This suggested to us that HIF-1α induction following H. pylori infection may be more closely linked to non-canonical functions.

Thus, to determine whether HIF-1α was required to increase transcription of the indicated genes, we also isolated clonal populations of AGS cells in which HIF-1α had been silenced. AGS cells were transfected with sh-RNA that specifically silenced HIF-1α expression or with a sh-RNA Scramble as a control and then infected with H. pylori. To evaluate the efficiency of the HIF-1α silencing, stably transfected cells were treated with the chemical inducer of HIF-1α, Deferroxamine (DFO) (200 μg/ml), and the levels of HIF-1α were determined by western blotting (Supplementary Figure 5). The stably transfected clones AGS sh-HIF-1α C7 and AGS sh-Scr C2 were chosen for infection by H. pylori 26695 at MOI 100 for 24 h and induction of mRNA of several HIF-1α target genes were evaluated. In these AGS sh-Scr C2 (clonal control) cells, some differences with respect to the wild-type cells were noted. For instance, we observed modest changes in the mRNA levels of GLUT-1 after 24 h of H. pylori infection. Importantly, however, similar changes were also detected in sh-HIF-1α C7 cells lacking HIF-1α, indicating that they were HIF-1α-independent. For LDHA, a minor increase was observed after 8 h of bacterial infection in cells expressing HIF-1α. However, overall a trend toward decreased LDHA was detectable after 24 h both in the cells expressing or lacking HIF-1α (Figure 5). Moreover, for VEGFA substantially elevated expression was noticeable after 24 h, but again these changes were seen in cells expressing or not HIF-1α (Figure 5). Taken together, these results suggested that the increase in nuclear HIF-1α protein levels and reporter activity (Figures 1B,C, respectively) was likely to be associated with additional functions beyond the transcriptional activation of common target genes.

Although H. pylori promotes activation of signaling pathways linked to cell cycle progression and proliferation (Suzuki et al., 2009), the bacteria is also known to induce G1 cell cycle arrest (Shirin et al., 1999; Ahmed et al., 2000). To evaluate in our hands how H. pylori modulates the cell cycle, AGS cells were infected with the bacteria at MOI 100 for 24 h. This time point was chosen post-infection considering that it represented roughly the amount of time necessary to complete one cycle of duplication. Then, cell cycle distribution was analyzed by flow cytometry. As is shown (Figure 6), infection increased the number of cells in G0/G1 and decreased the number of cells in G2/M compared to non-treated cells. Alternatively, the number of cells in S phase was not significantly altered by infection.

Our observations shown in Figure 5 indicated that HIF-1α induction by H. pylori may be triggering an alternative response, not necessarily linked to an increase in the expression of specific genes. Indeed, in more recent years, a non-canonical role for HIF-1α, independent of its transcriptional activity, has emerged, involving the inhibition of DNA replication and promotion of G1 cell cycle arrest (Huang, 2013). To test whether H. pylori- induced HIF-1α is able to induce cell cycle arrest, the stably transfected clones AGS sh-HIF-1α C7 and AGS sh-Scr C2 were infected by H. pylori 26695 at MOI 100 for 24 h and cell cycle distribution was determined by flow cytometry. We observed that infection with H. pylori increased the percentage of cells in the G0/G1 phase in sh-Scr C2 AGS cells (Figures 7A–C), while the percentage of cells in S phase increased in sh-HIF-1α C7 AGS cells (Figures 7–C). Moreover, infection decreased the percentage of both control cells and HIF-1α knockdown cells in G2/M phase to the same extent (Figures 7–C). Taken together, these data suggest that H. pylori-induced HIF-1α prevents G1/S cell cycle progression, but does not affect transition to the G2/M phase.

Cyclin D1 is required for progression of cells through the G1 phase of the cell cycle, and decreased expression has been linked to an arrest in this phase (Malumbres, 2011). To evaluate whether HIF-1α modulated Cyclin D1 levels and promoted cell cycle arrest in G0/G1 phase, sh-HIF-1α C7 and sh-Scr C2 AGS cells were infected with H. pylori for 8 and 24 h and Cyclin D1 levels were measured. We observed that Cyclin D1 protein levels decreased in both cell lines after 24 h of infection; however, for HIF-1α knockdown cells, Cyclin D1 levels were higher compared to control cells at this time point of infection (Figure 8A), suggesting that HIF-1α contributes to the decrease in Cyclin D1 triggered by H. pylori infection and cell cycle arrest. To evaluate whether the decrease in Cyclin D1 protein levels mediated by H. pylori-induced HIF-1α was attributable to a transcriptional mechanism, the Cyclin D1 mRNA in sh-HIF-1α C7 and sh-Scr C2 AGS cells infected with H. pylori were determined. Our results revealed that infection did not reduce the transcription of Cyclin D1 in either of the stably transfected cell lines. If anything, a tendency toward increased expression was observed 24 h post-infection by H. pylori, and this trend was similar for cells expressing or not HIF-1α (Figure 8B). To analyze whether H. pylori-induced HIF-1α affected Cyclin D1 at the post-translational level, sh-HIF-1α C7 and sh-Scr C2 AGS cells were infected by H. pylori for 8 h, treated with cycloheximide (CHX) to block the protein synthesis and then the kinetics of Cyclin D1 protein loss were determined. As shown in Figure 8C, Cyclin D1 protein levels declined more rapidly for the sh-Scr C2 than for sh-HIF-1α C7 AGS cells. Protein half-life was found to be approximately 9 min (t_1/2 = 9 min) in sh-Scr C2 AGS cells, compared to the 20 min (t_1/2 = 20 min) for sh-HIF-1α C7 cells. These results suggest that H. pylori-induced HIF-1α promotes the decline in Cyclin D1 by a post-translational mechanism.

Finally, to determine whether the proteasome pathway is involved in this mechanism, AGS cells infected with H. pylori for 8 h were treated with CHX together with MG132, a specific proteasome inhibitor. Our results showed that infection in the presence of MG132 and CHX partially restored Cyclin D1 levels compared to the treatment with CHX alone (Figure 8D), suggesting that Cyclin D1 decrease mediated by H. pylori infection is attributable in part to degradation via the proteasome. This recovery is apparently more effective in AGS sh-HIF-1α C7 cells; however, the degree of recovery (expressed in percent) in Cyclin D1 in the presence of MG132 is similar for both AGS sh-Scr C2 and sh-HIF-1α C7 cells (Figure 8D). This can be taken to suggest that the presence of HIF1α does not alter proteasome-mediated turnover, but rather some other post-translational process.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.