L. Statistics
III. RESULTS
Human gastric cancer cells were more tolerant of acidity in the culture media than normal cells.
To determine whether cancer cells tolerate acidic conditions better than noncancer cells, we cultured several gastric cancer cell lines (AGS, KATO III, MKN-28, MKN-45, SNU-1, and SNU-601) and noncancer cell lines (RGM-1, IEC-6, and COS-1) in media maintained at different pHs, including 7.4, 6.9, 6.4, 5.9, and 5.4. The cancer cells adapted well to lower pH, whereas the normal cells were much less tolerant of acidity in the culture media (Fig. 1A). At pH 5.4, most of cancer cell lines tested showed >80% viability, but normal cell lines RGM-1 significantly decreased viability by 2RGM-1.6%.
To investigate whether this difference in cell survival at lower pH was due to ability to dispose or disperse H+, we immunocytochemically stained two cancer cells (AGS and MKN-45) and two noncancer cells (RGM-1 and IEC-6) with antibodies against the gastric proton pump, α subunit of H+/K+-ATPase. We found that there was a much larger amount of proton pump in cancer than noncancer cells (Fig. 1B). Immunofluorescence images showed a predominant expression of the α subunit of H+/K+-ATPase in the membrane and cytoplasm of cancer cells. These results indicated the enhanced ability to dispose of H+ due to overexpression of H+/K+-ATPase, which might contribute to cancer cell survival in an acidic microenvironment.
Fig. 1. Effects of pH of culture medium on cell viability and cellular expression of H+/K+-ATPase.
Human gastric cancer cell lines (MKN-45, MKN-28, AGS, Kato III, SNU-601, and SNU-1) and noncancer cell lines (RGM-1, Cos-1, and IEC-6) were cultured in media at different pH for 24 hours, and cell viability was measured by the 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide assay (A). Expression of H+/K+-ATPase, α-subunit was detected with specific antibodies and visualized with Cy3-conjugated secondary antibodies.
Note that the α-subunit of H+/K+-ATPase was more highly expressed in the plasma membrane and cytoplasm of cancer than noncancer cells (B).
A.
B.
Treatment of PPZ significantly inhibits cancer cell viability in a pH-dependent manner.
Because PPZ is a protonatable weak base, which can convert to active form in an acidic environment with a low pH, we tested the effects of PPZ on cell growth in culture media maintained at various pHs (Fig. 2). Nevertheless, cancer cells showed tolerance of acidity in culture media as shown in Fig. 1A, and administration of the proton pump inhibitor PPZ led to significant reduction of cancer cell survival (Fig. 2A). At pH 5.0, PPZ treatment reduced cell viability by 6.9% on average in MKN-45 cancer cells (Fig. 2A). On the contrary, normal cell line RGM-1 did not showed a significant difference in cell viability between the PPZ-treated and nontreated groups (data not shown). Despite resistance of cancer cells to the acidic environment, cancer cells were much more susceptible to growth inhibition of PPZ at a lower pH.
We evaluated whether this attenuation of cancer cell viability is induced by a decrease of intracellular pH due to the blocking of disposal of intracellular H+ by specific protonation of this drug. As shown in Fig. 2B, PPZ treatment significantly increased acidity of intracellular pH in AGS, MKN-28, and MKN-45, but SNU-601 did not show a remarkable change (Fig. 2B). Of note, AGS decreased intracellular pH from 7.6 to 7.2. These findings showed that PPZ-induced cell death was caused by the specific protonation of this proton pump inhibitor.
We also investigated whether other proton pump inhibitors such as omeprazole and lansoprazole has a similar effect to pantoprazole on cancer cell viability. Human gastric MKN-45 cells were cultured with each proton pump inhibitor in pH 6.0 of media for 24 hours, and cell viability was determined by
3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide assay. The results showed that whereas the PPZ or omeprazole-treated cells remarkably reduced cell viability in a dose-dependent manner, lansoprazole did not show any changes in cancer cell viability (Fig. 3). This difference suggested that the decreased cell viability comes from the specific effect of proton pump inhibition.
Omeprazole and PPZ are available for parenteral injection, but lansoprazole is only available for oral administration, suggesting that lansoprazole lost its proton pump inhibiting ability in aqueous form.
Fig. 2. Inhibition of cancer cell viability by PPZ treatment.
In different pH media (7.4, 6.9, 6.4, 5.9, 5.0), gastric MKN-45 cancer cells were incubated in the presence or absence of 0.5 mmol/L PPZ, and cell viability was measured by 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide assay (A). Although MKN-45 showed resistance to acidity of culture media, PPZ significantly decreased this cancer cell viability in a pH-dependent manner. To evaluate effect of PPZ on intracellular pH (pHi) of cancer cell, gastric cancer cells were plated in a 6-well culture dish, treated with 0.5 mmol/L PPZ for 16 hours, and measured according to Materials and Methods as described previously (B).
A. B.
Fig. 3. Effect of proton pump inhibitors on gastric cancer cell viability.
To compare effect of other proton pump inhibitors (omeprazole and lansoprazole) with PPZ on cancer cell viability, MKN-45 cells were cultured in the presence of each proton pump inhibitor indicated at Ph 6.0 for 24 hours and were measured cell viability by 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide assay. Cell viability (%) was calculated by formula as follows ; Cell viability (%) = (sample absorbance - total absorbance)/(spontaneous absorbance - total absorbance) × 100. Total absorbance was obtained from absorbance of 0.1% Triton X-100-treated cells as negative control, and spontaneous absorbance indicated absorbance of PPZ-untreated control cells.
PPZ selectively induced apoptosis cell death in human gastric cancer cells.
To determine whether PPZ preferentially induces apoptosis in cancer cells, we administered this agent to two cell lines, a normal gastric mucosal cell line, RGM-1, and a gastric cancer cell line, MKN-45 (Fig. 4). Although normal gastric mucosal RGM-1 cells significantly reduced cell viability in a low pH media as shown in Fig. 1A, an apoptotic genomic DNA fragmentation was detected by neither pH change nor PPZ treatment (Fig. 4A).
However, human gastric cancer cell MKN-45 showed a significant apoptotic genomic DNA fragmentation, caspase-3 and its substrate, PARP, were cleaved by PPZ treatment in a dose-dependent manner in gastric cancer cells but not detected in normal gastric mucosal cell (Fig.
4B). PPZ also induced alteration of phosphatidylserine distribution and permeability of plasma membrane only in cancer that which were detected by annexin V/propidium iodide staining (Fig. 4C). These results indicated much higher vulnerability and selective sensitivity to apoptosis in gastric cancer MKN-45 cells than normal mucosal RGM-1 cells.
Fig. 4. Selective induction of apoptosis with PPZ.
Cells were cultured in the presence or absence of PPZ for 24 hours, and induction of apoptosis was detected by genomic DNA fragmentation (A), caspase-3 activation and PARP cleavage (B), and localization of annexin V and propidium iodide (C). Annexin V-positive cells were stained green, and red represented propidium iodide-positive cells.
A.
B.
C.
PPZ suppressed ERK phosphorylation in human gastric cancer cells.
To assay the involvement of MAPKs in PPZ-induced apoptosis, phosphorylation of ERKs, JNKs/stress-activated kinases, and p38 after PPZ treatment was detected (Fig. 5A and B). In MKN-45 cells, PPZ significantly and selectively inhibited the phosphorylation of ERK and increased the phosphorylation of p38 but had no apparent effects on the phosphorylation of JNK. After 8-hours treatment of 0.5 mmolL PPZ, ERK phosphorylation was completely inhibited in MKN-45 cells despite an equal amount of total proteins controlled by α-tubulin (Fig. 5A). In adition, inhibition of p38 by SB203580 blocked PPZ-induced apoptosis in MKN-45 cells (data not shown). As shown is Fig. 5B, RGM-1 cells did not change the degree of phosphorylation of ERK and p38, but phosphorylation of JNK was significantly decreased in a time-dependent manner. Taken together, these findings suggest that inhibition of ERK phosphorlyation may be responsible for the attenuation of cancer cell survival, whereas activation of p38 contributes to proton pump inhibitor-induced apoptosis.
Fig. 5. Distinct MAPK signaling in cancer cells (A) and noncancer cells (B) by PPZ.
Cells were treated in the presence or absence of 0.5 mmol/L PPZ for various times, and total proteins were extracted from the cells, electrophoresed on 12% SDS-PAGE gels, and transferred to polyvinylidene difluoride membranes. The membranes were probed with specific antibodies for p-ERK, p-P38, p-JNK, and α-tubulin, respectively.
A. B.
Overexpression of HSP27 and HSP70 played a cytoprotective role in PPZ-induced apoptosis of normal gastric mucosal cells.
Interestingly, we found that HSP70 and HSP27 were overexpression in RGM-1 after PPZ treatment but not in MKN-45 cells (Fig. 6A and B). Notably, treatment of 0.6 mmol/L PPZ in RGM-1 cells showed a 3.3- and 36.3-fold increased of HSP27 and HSP70 compared with that of nontreated cells, respectively (Fig. 6C). However, HSP27 and HSP70, as well as HSP60, were not grossly changed in PPZ-treated gastric cancer MKN-45 cells (Fig. 6A). To compare the HSP induction, we also treated geranylgeranylacetone (GGA), which is well known as a HSP inducer in gastric mucosal cells. GGA slightly increased HSP60 and HSP27 in MKN-45 cell, but deceased HSP60 and HSP27 in RGM-1. In comparisom with GGA, the ablilty of PPZ to induce HSP70 is much higher at 5.6-fold than GGA. These data suggest that overexpression of HSP27 and HSP70 might play a cytoprotective role in PPZ-induced apoptosis of normal gastric mucosal cells.
Fig. 6. Effect of PPZ on expression of HSP70 and HSP27 in gastric cancer and noncancer cells.
RGM-1 (A) and MKN-45 cells (B) were treated with PPZ or generylgenerylacetone (GGA) for 24 hours, and expression of HSP70, HSP60, and HSP27 was assayed by immunoblotting with specific antibodies. Note that PPZ induced expression of HSP70 and HSP27 in RGM-1 normal gastric mucosal cells but not in MKN-45 gastric cancer cells. Relative intensities of HSP70 and HSP27 compared with nontreated control cells were represented in C.
A. B.
C.
In a xenograft model of nude mice, administration of PPZ significantly inhibited tumorigenesis and induced large-scale apoptosis of tumor cells.
We also evaluated the antitumorigenic effects of PPZ in a gastric cancer xenograft model (Fig. 7). The animals were randomly divided into two groups: an intratumoral injection of PBS and an intratumoral injection of 0.4 ㎎/㎏ PPZ. Intratumor administration of PPZ significantly suppressed tumor growth in athymic nude mice, with decreased in tumor volume at day 22 of 44.69% compared with mice injected with PBS (Fig. 7A and B).
The isolated tumor from mice with intratumor administration of PPZ was remarkably smaller than that of mice intramorally injected with PBS (Fig. 7C). Histopathological examination revealed that PPZ treatment provoked considerable apoptosis. Only a few tumor cells survived, and most tumor cells were replaced by apoptosis cells (Fig. 7D). TUNEL staining also showed considerably higher apoptotic cell death in the PPZ-treated group compared with the control group (Fig. 7E).
Fig. 7. Antitumorigenic potency of PPZ in xenograft model.
MKN-45 human gastric cancer cells were inoculated s.c. into the backs of athymic nude mice. PPZ (40mg/kg) was injected intrautmorally (IT) 14 days after xenograft. Mice were sacrificed 22 days after tumor inoculation. Tumor volume was measured and represented (A and B). The isolated tumors (C) were sectioned and processed for H&E staining (D, ×100 and ×400 magnification) and TUNEL assay (E, ×100 and ×200 magnification); bars, ±SD.
A
B
C
D
E
Distinct expression of CD34+ blood vessels between H.pylori positive and negative gastric patients
Stomach tissue samples were obtained from gastritis patients during endoscopy examination and were evaluated histologically by haematoxylin-eosin staining. Finally, we choose 20 H.pylori positive gastritis and 18 H.pylori negative gastritis cases with a similar degree of gastritis, scored according to the modified Sydney classification, as the degree of gastric inflammations itself can affect angiogenesis. Immunohistochemical staining using antibodies against the CD34 endothelial cell marker was performed to evaluate the difference in angiogenesis according to H.pylori infection. The results showed that patients with H.pylori positive gastritis (Fig. 8A (b, d)) showed significantly higher expression of CD34 positive blood vessels in the gastric mucosa layer than that of patients with H.pylori negative gastritis (Fig. 8A (a, c)). The number of blood vessels was counted in three sites, for each specimen, with equal dimensions, and mean levels are shown in Fig. 8B. While H.pylori negative gastritis samples had a mean of 7.2 (SEM 0.8) blood vessels per x100 magnified field, H.pylori positive gastritis samples had a significant increased number (mean 40.9 (SEM 4.4)) of blood vessels (p<0.01). Moreover, blood vessels observed in cases with H.pylori positive gastritis (Fig. 8A (b, d), arrow) were thicker and larger than those of H.pylori negative cases (Fig. 8A (a, c), arrow). Interestingly, blood vessels were found more abundantly in the mesenchymal stromal layer below the mocosa layer but the number of blood vessels in the stromal layer was not different between H.pylori positive gastritis and H.pylori negative gastritis cases, suggesting that H.pylori infection may be associated with induction of angiogenesis in H.pylori infected gastric mucosa.
We then evaluated if there were any differences in expression of HIF-1α, the potent angiogenic transcriptional factor (Fig. 9). As a control, we used gastric biopsies obtained from five cases diagnosed with functional dyspepsia with no significant abnormal gastroscopic findings and no H.pylori infection. Compared with HIF-1α mRNA from five normal stomachs, expression was not altered in five H.pylori negative gastritis but was significantly increased in H.pylori positive gastritis (p<0.01), suggesting that HIF-1α is responsible for H.pylori induced angiogenesis (Fig. 9B).
0
specimens of Helicobacter Helicobacter Helicobacter Helicobacter pyloripyloripylori negative (a, pylori negative (a, negative (a, ×100; c, negative (a, 100; c, 100; c, 100; c, ×200) and 200) and 200) and 200) and H.pyloriH.pyloriH.pyloriH.pylori positive (b, positive (b, positive (b, positive (b,
×100; d, 100; d, 100; d, 100; d, ×200) gastritis. Specimens obtained from 200) gastritis. Specimens obtained from 200) gastritis. Specimens obtained from 200) gastritis. Specimens obtained from H.pyloriH.pyloriH.pyloriH.pylori negative gastritis or negative gastritis or negative gastritis or negative gastritis or H.pylori
H.pyloriH.pylori
H.pylori positive gastritis during endoscopic examination were positive gastritis during endoscopic examination were used for positive gastritis during endoscopic examination were used for positive gastritis during endoscopic examination were used for used for immunohistological analysis with anti
immunohistological analysis with antiimmunohistological analysis with anti
immunohistological analysis with anti----CD34CD34CD34 antibodies. CD34CD34 antibodies. CD34 antibodies. CD34 antibodies. CD34++++ blood vessels were blood vessels were blood vessels were blood vessels were stained by a dark red color (arrow), counterstained with haematoxylin showing a blue stained by a dark red color (arrow), counterstained with haematoxylin showing a blue stained by a dark red color (arrow), counterstained with haematoxylin showing a blue stained by a dark red color (arrow), counterstained with haematoxylin showing a blue
color. ( color. (color. (
color. (BBB) Mean number of blood vessels stained with CD34. Number of blood vessels B) Mean number of blood vessels stained with CD34. Number of blood vessels ) Mean number of blood vessels stained with CD34. Number of blood vessels ) Mean number of blood vessels stained with CD34. Number of blood vessels with equal dimensions were counted in triplicate f
with equal dimensions were counted in triplicate fwith equal dimensions were counted in triplicate f
with equal dimensions were counted in triplicate for each sample and are represented or each sample and are represented or each sample and are represented or each sample and are represented
as means (SD). There was a statistically significant difference between the two groups as means (SD). There was a statistically significant difference between the two groups as means (SD). There was a statistically significant difference between the two groups as means (SD). There was a statistically significant difference between the two groups
(p<0.01).
(p<0.01). (p<0.01).
(p<0.01). Hp, Helicobacter pyloriHp, Helicobacter pyloriHp, Helicobacter pylori.Hp, Helicobacter pylori...
Normal HP--gastritis HP+- gastritis
HIF-1αααα
RNA isolated from Helicobacter pyloriHelicobacter pyloriHelicobacter pyloriHelicobacter pylori ( ( ( (HpHpHpHp) negative gastritis (n=5), ) negative gastritis (n=5), ) negative gastritis (n=5), H.pylori) negative gastritis (n=5), H.pyloriH.pyloriH.pylori positive positive positive positive gastritis (n=5), and normal stomachs
gastritis (n=5), and normal stomachs gastritis (n=5), and normal stomachs
gastritis (n=5), and normal stomachs (n=5). ((n=5). ((n=5). ((n=5). (BBB) Relative expression of HIFB) Relative expression of HIF) Relative expression of HIF) Relative expression of HIF----1111α mRNA mRNA mRNA mRNA
is represented as mean intensity/mm is represented as mean intensity/mmis represented as mean intensity/mm
is represented as mean intensity/mm2222. HIF. HIF. HIF. HIF---1-11α expression was significantly higher in 1 expression was significantly higher in expression was significantly higher in expression was significantly higher in gastric mucosa of
gastric mucosa of gastric mucosa of
gastric mucosa of H. pyloriH. pyloriH. pyloriH. pylori positive gastritis than in positive gastritis than in positive gastritis than in H.pylori positive gastritis than in H.pyloriH.pylori negative gastritis in spite H.pylori negative gastritis in spite negative gastritis in spite negative gastritis in spite
of similar expression of GAPDH (p<
of similar expression of GAPDH (p<of similar expression of GAPDH (p<
of similar expression of GAPDH (p<0.01). 0.01). 0.01). 0.01).
Suppression of H.pylori induced in vitro angiogenesis by gastric PPZ
To prove a direct effect of H.pylori infection on angiogenesis, we performed an in vitro angiogenesis assay. Conditioned media obtained from H.pylori infected AGS cells were added to HUVEC culture flasks and morphological changes in the endothelial cells were observed. After nine days, HUVEC became long in shape and formed a tubular structure (Fig.
10B) compared with conditioned media of non-H.pylori infected AGS (Fig. 10A). CD31 immunofluorescence staining showed a dense intensity of CD31 molecules in HUVEC cells incubated with the culture supernatants of H.pylori infected AGS (Fig. 10C) while control media obtained from non-H.pylori infected AGS stimulated neither tubular formation of HUVEC nor expression of the CD31 endothelial cell marker (Fig. 10A, and F).
Results of in vitro angiogenesis assay strongly suggested that H.pylori infection stimulated infected gastric epithelial cells to secrete proangiogenic factors which induce growth and differentiation of endothelial HUVEC. Interestingly, pretreatment with PPZ (200µM or 400µM, for eight hours) on AGS cells prior to H.pylori inoculation significantly inhibited tubular formation (Fig. 10D, and E) and CD31 expression of endothelial HUVEC (Fig. 10I, and J). However, no significant changes were noted in HUVEC incubated with 400µM PPZ alone (Fig. 10C, and H), suggesting that PPZ itself did not influence tubular formation of HUVEC. The data clearly indicate that PPZ suppressed H.pylori induced in vitro angiogenesis, suggesting that antiangiogenic treatment with PPZ could be a promising therapeutic approach for H.pylori associated carcinogenesis.
Control H. pylori (24 h)
H. pylori (24 h) +PPI (200 µµµµM, 8 h)
H. pylori (24 h) +PPI (400 µµµµM, 8 h)
Phase-contrast αααα-CD31Ab
A B C D E
F G H I J
PPI (400 µµµµM, 8 h)
Fig. 10. In vitro angiogenesis assay.
AGS cells (1×107 cells/100mm2 culture dish) were incubated with 0, 200, or 400 µM proton pump inhibitor (PPZ) for eight hours, washed with phosphate buffered saline three time, and inoculated with Helicobacter pylori (5×107 CFU/ml) for 24 hours. Conditioned media were prepared from 1:1 dilution of the cell culture supernatant and the human umbilical vein endothelial cell (HUVEC) medium. Conditioned media were filtered through a 0.4 µM pore filter to remove H.pylori and then added to the HUVEC culture which was change every three days. After nine days, HUVEC were observed in a tubular formation under microscopy (A-E) and expression of the endothelial cell marker, CD31, was confirmed by immunocytofluorescence staining (F-J). (A, F) Control HUVEC cells; (B, G) HUVEC cells
incubated with conditioned media of H.pylori infected AGS; (C, H) HUVEC cells incubated with conditioned media of 400 µM PPZ treated AGS; (D, I) HUVEC cells incubated with conditioned media of 200 µM PPZ/H.pylori infected AGS; (E, J) HUVEC cells incubated with conditioned media of 400 µM PPZ/H.pylori infected AGS.
Production of proangiogenic factors from H.pylori infected gastric epithelium and its inhibition by PPZ
Following H.pylori infection, AGS cells significantly secreted VEGF and IL-8, well characterized as proangiogenic factors, in a time dependent manner (Fig. 11A). Maximal induction of IL-8 (mean 1019 (SEM 278) pg/ml) and VEGF (1597 (94) pg/ml) was observed after 24 hours of inoculation (Fig. 11A). We also examined mRNA expression of these angiogenic factors using RT-PCR analysis (Fig. 11B). Expression of VEGF mRNA, one of the HIF-1α target genes, was induced after 16 hours of H.pylori infection, showing the correlation with HIF-1α expression (Fig. 11B). IL-8 mRNA was also significantly induced after H.pylori infection (Fig. 11B). All of these results suggest that synthesis of angiogenic epithelial cells, which could induce proliferation and differentiation of endothelial cells.
Because PPZ showed a strong antiangiogenic action in the in vitro angiogenesis assay (Fig. 10), we measured the effect of PPZ on expression of these angiogenic factors (Fig. 12).
Secretion of IL-8 in the supernatants of H.pylori infected AGS cells was found to be remarkably suppressed after PPZ treatment in a dose dependent manner (Fig. 12A).
Following eight hours of infection with H.pylori, IL-8 production increased up to 870 pg/ml but this increment in IL-8 production was significantly attenuated by PPZ pretreatment.
Pretreatment with PPZ showed a considerable regulatory effect on H.pylori mediated VEGF synthesis (Fig. 12A). Suppression of thses angiogenic factors by PPZ was evidenced by transcriptional inhibition of the genes (Fig. 12B). At 400 µM of PPZ, expression of VEGF and HIF-1α seemed to decline relevant to that of control AGS cells.
A.
Relative intensity (-folds)Relative intensity (-folds)Relative intensity (-folds)Times after H. pylori 0 0.25 0.5 1 2 4 8 24h
C.
Fig. 11. Release and expression of vascular endothelial growth factor (VEGF) and interleukin 8 (IL-8) from Helicobacter pylori infected AGS cells.
(A) Production of VEGF (top) and IL-8 (bottom) was measured by ELISA with the culture supernatant of AGS cells infected with H.pylori for the indicated times. (B) Induction of
(A) Production of VEGF (top) and IL-8 (bottom) was measured by ELISA with the culture supernatant of AGS cells infected with H.pylori for the indicated times. (B) Induction of