Gastrokine 1 induces senescence through p16/Rb pathway activation in gastric cancer cells
- Rui Xing1,
- Wenmei Li2,
- Jiantao Cui2,
- Jun Zhang1,
- Bin Kang1,
- Yuan Wang1,
- Zhaohui Wang1,
- Siqi Liu1,
- Youyong Lu1,2
- 1Beijing Institute of Genomics, Chinese Academy of Sciences, Beijing, China
- 2Laboratory of Molecular Oncology, Key Laboratory of Carcinogenesis and Translational Research (Ministry of Education), Peking University School of Oncology, Beijing Cancer Hospital/Institute, Beijing, China
- Correspondence to Dr Youyong Lu, Peking University School of Oncology, Beijing Cancer Hospital/Institute, 52 Fucheng Road, Haidian District, Beijing 100142, China; Dr Siqi Liu, Beijing Institute of Genomics, Chinese Academy of Sciences, Beijing Airport Industrial Zone B-6, Shunyi District, Beijing 101318, China;
- Revised 8 April 2011
- Accepted 17 April 2011
- Published Online First 14 June 2011
Background and aims Gastrokine 1 (GKN1) is a stomach-specific protein that is normally expressed in gastric mucosa but not in primary tumours and cell lines. Based on this evidence, it was presumed that GKN1 might play a role in gastric cancer development; however, its function and molecular mechanism are not clear. A systematic study was initiated that combined multiple approaches to define the molecular mechanism of GKN1 in gastric cancer cells.
Method Proteomics, western blotting and immunohistochemistry were used to measure the expression level of GKN1. Western blotting combined with immunofluorescence was used to monitor the secretory process of this protein. Subsequently, the function and molecular mechanism of GKN1 was explored in vitro and in vivo.
Results It was shown that GKN1 is an autocrine/paracrine protein and inhibits cell growth due to senescence, which resulted from activation of p16/Rb and p21waf pathways. Furthermore, sustained activation of Ras/Raf/MEK/ERK signalling was characterised in gastric cancer cells and a xenograft nude mouse model following GKN1 treatment.
Conclusion These results provide comprehensive molecular evidence of GKN1 in inducing senescence of gastric cancer cells, and indicate that GKN1 might be a potential novel target for gastric cancer therapeutics.
Significance of this study
What is already known about this subject?
Gastrokine 1 (GKN1) is a stomach-specific protein and its expression is downregulated in gastric cancer.
GKN1 has a signal peptide at the N-terminus, which means that it may be a secreted protein, but there is no direct evidence of this to date.
Conflicting data about the function of GKN1 have been reported from different laboratories. Martin et al have proposed that GKN1 exerts a mitogenic effect on IEC-6 cells, when compared with epidermal growth factor; whereas Shiozaki et al have found that GKN1 can inhibit growth of MKN28 cells.
What are the new findings?
GKN1 is either an autocrine or paracrine protein, which is secreted via exocytosis after being packaged in Golgi vesicles and re-attached to the cellular membrane as an exogenous ligand.
GKN1 has dual functions: it promotes the growth of normal cells but inhibits the growth of gastric cancer cells.
In gastric cancer cells, GKN1 is able to induce cell senescence and its core pathways, the p16/Rb and p21waf pathways, via the Ras/Raf/MEK/ERK pathway.
How might it impact on clinical practice in the foreseeable future?
The stomach-specific protein GKN1 is a secreted protein; furthermore, it is able to promote the growth of normal cells but inhibit the growth of gastric cancer cells. We suggest that GKN1 could be a diagnostic and therapeutic target for gastric disease, including precancerous lesions and tumours.
Gastric cancer (GC) has one of the highest incidences and the second highest mortality among all cancers in China.1 Approximately 70% of GC occurs in the gastric antrum/pylorus region2; hence, the proteins expressed specifically in the antrum region might play a critical role in maintaining normal gastric mucosal structure and function. The protein gastrokine 1 (GKN1) (also named AMP-18) is a member of the BRICHOS superfamily,3 and is mainly located around the antrum.4 5 Moreover, GKN1 is specifically expressed in gastric mucosa.6
Previously, Jang et al have observed that GKN1 is downregulated in GC tissues and cell lines.7–9 A recent report has demonstrated a progressive decrease of GKN1 from chronic gastritis to atrophic intestinal metaplasia and from Helicobacter pylori-infected tissues.10 Walsh-Reitz et al have shown that GKN1 can stimulate growth of IEC-6 cells in a mouse model.11 However, the molecular mechanism of GKN1 and its role in tumorigenesis of gastric mucosa remain unclear.
In the present study, we investigated two aspects of GKN1. First, the secretory pathway of GKN1 was determined in a cell model. Although data from bioinformatics analysis and a few preliminary studies have supported GKN1 as a secreted protein, definitive evidence is still inconclusive.4–6 9 Secondly, the function and molecular mechanism of GKN1 remain to be addressed. Conflicting results have been reported. Martin et al have proposed that GKN1 exerts a mitogenic effect on IEC-6 cells,4 5 whereas Shiozaki et al have found that GKN1 can inhibit growth of MKN28 cells.9 To clarify these issues, we initiated a systematic study to elucidate how GKN1 participates in the processes of secretion, growth and signal transduction in gastric cells.
Proteomics analysis and immunohistochemistry (IHC) staining
The techniques of two-dimensional electrophoresis (2DE) and MALDI TOF/TOF MS (matrix-assisted laser desorption ionisation time of flight/time of flight mass spectrometry) were performed as previously reported.12 The 2DE image analysis was carried out by a combination of manual inspection and software analysis with ImageMaster Platinum 5.0 (Amersham Biosciences, Uppsala, Sweden). The relative spot volumes were normalised to the total spot volumes with a multiplication factor of 100.
For IHC assessment, tissue microarrays were constructed as described previously.13 Eighty-six pairs of tissues were collected from the Beijing Cancer Hospital & Institute. Most samples, including tumours and matched normal tissues, were derived from the antrum. The study was conducted in accordance with the Helsinki Declaration, and was approved by the Institutional Ethical Standards Committee with the patients' informed consent. IHC staining was performed using an EnVision+ Kit (Dako, Glostrup, Denmark). The slices were incubated with primary antibody against GKN1. In the negative control group, 1% bovine serum albumin (BSA) was used.
Total RNA was isolated from seven gastric cells and seven pairs of human gastric tissues. First-strand cDNA was synthesised by reverse transcriptase (Invitrogen, Carlsbad, California, USA) using total RNA as a template. Quantification of gene expression of GKN1 and glyceraldehyde phosphate dehydrogenase (GAPDH) was conducted with an ABI PRISM 7300 system (Foster City, California, USA). The primer sequences were: forward, 5′-GCCCTCCATTCAATCC-3′; reverse, 5′-GACTTTGTTTGGGTTGACTGA-3′. PCR was carried out with programmed parameters, heating at 95°C for 3 min, followed by 40 cycles of a four-stage temperature profile of 95°C for 15 s, 60°C for 20 s, 72°C for 15 s and 78°C for 30 s, to collect the fluorescent signals. The melting curves for each PCR were carefully analysed to avoid non-specific amplification in the PCR products. GKN1 expression of each sample was transformed using the ΔΔCt formula and normalised with GAPDH expression.
Molecular cloning and protein expression
Coding sequence regions that lacked the N-terminal 20 amino acid coding sequences of human GKN1 were amplified from the cDNA library of human stomach tissue, and cloned into the pQE30 vector, which was confirmed by sequencing. The His-tagged recombinant proteins of wild-type GKN1, without its first 20 amino acids, were expressed in Escherichia coli JM109 strain, and purified by Ni-affinity chromatography. The supernatant fluid of JM109 transfected with the pQE30 vector was used as the control buffer. The specific RNA interference (RNAi) vectors for p16INK4a, p21waf and GKN1 were constructed based on pSliencer3.1-H1-neo, and the targeting DNA sequences were 5′-AGAACCAGAGAGGCTCTGA-3′, 5′-CATACTGGCCTGGACTGTT-3′ and 5′-GCCCAAACAAAGTCGATGAC-3′, respectively. Scrambled short hairpin (shRNA) was transfected into cells as a control.
Cell culture and GKN1 treatment
BGC-823, MGC-803, SGC-7901 and GES-1 cells were provided by Beijing Cancer Hospital. GES-1 cells were generated and immortalised from normal stomach mucosal cells and transformed with SV4014; BGC-823 is a human gastric adenocarcinoma-derived cell line, which is used widely for studying GC.14–17 N87 and AGS cell lines were purchased from the American Type Culture Collection (ATCC). The cells were cultured in Dulbecco's modified Eagle's medium that contained 10% fetal bovine serum. To monitor GKN1 effects on cell growth, 1 μg/ml (60 nM) GKN1 was added to cells after they had reached ∼80% confluence and were starved overnight. Cell culture was continued for 24–48 h, followed by examination of cellular activity.
Cellular fractionation and western blotting (WB)
Cellular fractions were prepared by differential centrifugation: nuclei were pelleted at 1000 g for 30 min; mitochondria at 8000 g for 15 min; Golgi/endoplasmic reticulum (ER) at 32 000 g for 30 min; membranes at 1 500 000 g for 2 h; and the final supernatant collected was the cytoplasm. The antibodies used were ERK2 (sc-81457; Santa Cruz Biotechnology, Santa Cruz, California, USA), p16INK4a (MS-383; NeoMarkers, California, USA), p21waf (sc-817; Santa Cruz Biotechnology), retinoblastoma protein (Rb) (sc-50; Santa Cruz Biotechnology), pRb (sc-7986; Santa Cruz Biotechnology), pERK (Tyr204) (sc-7383; Santa Cruz Biotechnology), pRaf (sc-81513; Santa Cruz Biotechnology) and actin (sc-1616; Santa Cruz Biotechnology). The polyclonal antibody against GKN1 was a generous gift from Professor TE Martin of the University of Chicago.
Immunofluorescence image analysis by confocal microscopy
Cells were fixed, perforated and blocked by 1% BSA. GKN1 antibody was incubated with the treated cells for 2 h. After a thorough wash in Tris-buffered saline–Tween (TTBS), cells were incubated with the fluorescein isothiocyanate (FITC)-conjugated secondary antibody for 1 h. The fluorescent image was acquired using a confocal laser scanning microscope (LSM510; Zeiss, Toronto, Canada) and analysed with the LSM5 Image Browser program (Carl Zeiss Micro-Imaging, Toronto, Canada). To trace the secretion pathway of GKN1, the vector enhanced green fluorescent protein (EGFP)-N1–GKN1 was transfected into BGC-823 cells and stained with fluorescent ceramides (Invitrogen), an indicator for the Golgi. The co-localisation of EGFP–GKN1 and the Golgi was examined by confocal microscopy.
Cell growth determined by MTT assay
Approximately 5000 BGC-823 cells were plated into 96-well plates and starved overnight. The cells were treated with GKN1 or control buffer. MTT (5 mg/ml) solution was added to each well for 4 h, followed by dimethylsulfoxide (DMSO). The reaction products were detected at 570/630 nm.
Cell cycle estimated by flow cytometry
Cells were fixed overnight, resuspended in phosphate-buffered saline (PBS), and stained with propidium iodide in the dark for 30 min. The DNA content was measured by fluorescence-activated cell sorting (FACS) on a Becton-Dickinson FACScan flow cytometry system.
Senescence-associated β-galactosidase (SA-β-Gal) activity assay
Cells that were seeded in the dishes were washed twice with PBS and fixed in PBS that contained 2% formaldehyde/0.2% glutaraldehyde for 7 min at room temperature. SA-β-Gal staining was performed in fresh senescence-associated X-Gal staining solution at 37°C (no CO2). Incubation typically lasted for 12 h. Cells were rinsed in PBS and stored in PBS with 70% glycerol. Cells were then examined under a microscope for blue-green staining of the cytoplasm that was indicative of senescence.
Tumorigenicity in nude mice
Approximately 500 000 BGC-823 cells were suspended in Hank's buffer and were bilaterally injected into the subcutaneous fat of each female BalB/c nude mouse, aged 5 weeks. After 1 week, 25 μmol/kg U012618 or DMSO was injected intraperitoneally, and 0.1 μg GKN1 was injected locally in one side seeded with BGC-823 cells, while the supernatant fluid of JM109 transfected with pQE30 vector was injected in the other side seeded with BGC-823 cells as a control.
All calculations were performed with SPSS version 11.5. All the statistical analysis was conducted using the t test, while the significance of IHC images was evaluated with the χ2 test. All comparisons were two-tailed, and p<0.05 was considered to be significant.
Expression status of GKN1 in GC tissues and cells
We employed multiple approaches to assess the abundance of GKN1 in gastric tissues, at either the mRNA or protein levels. By combining the data from 2-DE (data not shown) and WB (figure 1A, upper panels), we found that GKN1 protein abundance in the adjacent normal tissues was dramatically higher than that in the tumours. IHC data were obtained from the tissue array that consisted of 86 pairs of samples, primary GC and corresponding adjacent regions. Seventy-six adjacent tissues exhibited a high level of GKN1 expression, whereas the 73 tumour tissue samples showed nearly undetectable amounts of GKN1 (figure 1B). We examined the GKN1 expression level in an immortalised gastric epithelial cell line (GES-1) and five cancer cell lines. As depicted in figure 1A (lower panel), the signal for GKN1 was detected in GES-1 cells but was almost undetectable in all the GC cell lines. Moreover, we assessed GKN1 abundance at the transcriptional level in tissues and cell lines using real-time PCR. Similar to the protein level, the GKN1 mRNA level in all adjacent normal tissues and GES-1 cells was higher than that in GC tissues and cell lines (figure 1C). Collectively, the above results demonstrated that GKN1 expression, at either the protein or mRNA level, was significantly attenuated in the primary GC tissues and GC cell lines, compared with adjacent normal tissue and normal cells.
Secretory pathway of GKN1 in GC cells
Although GKN1 is predicted to have a secretory signal peptide, the secretory property of this protein has not yet been proven experimentally.4–6 9 We transfected pEGFP-N1-GKN1 into BGC-823 cells, collected the culture medium, and measured the level of GKN1. WB showed the presence of GKN1 in the medium (figure 2A, upper panel). The cellular localisation of GKN1 was monitored using confocal microscopy. As seen in figure 2B, confocal images reveal that the immuno-signals against GKN1 were merged with the Golgi. We next performed differential centrifugation to check the abundance of GKN1 in cellular organelle preparations. Figure 2A (middle panel) indicates that the immunologically identified GKN1 was found in the cytoplasmic membrane and Golgi/ER fractions, but not in other organelles. With three approaches, we have experimentally demonstrated that GKN1 is a secretory protein that is probably secreted through the ER/Golgi secretory pathway.
The phenomenon of GKN1 rebound to the cell membrane suggests that it plays an autocrine role. To confirm this, truncated GKN1 without its secretory signal peptide was added to the culture medium of BGC-823 and HeLa cells. After 30 min, immunologically tagged GKN1 and cadherin, a typical indicator of the cytoplasmic membrane, were observed in the cells by confocal microscopy. Figure 2C shows that the GKN1 signals were superposed with cadherin fluorescence in BGC-823 cells, whereas the signal was not found on the membrane or in the cytoplasm of HeLa cells. Hence, the truncated GKN1 probably binds to BGC-823 cell membranes, possibly through a membrane receptor. In contrast, HeLa cells failed to bind GKN1 because of the lack of a receptor on the membrane. Furthermore, we prepared membrane and cytoplasmic fractions by differential centrifugation of BGC-823 cells incubated with GKN1, and determined its presence in the fractions. The immunologically identified GKN1 was only seen in the membrane, but not in the cytoplasm (figure 2A, lower panel). Therefore, we provided convincing evidence to support the notion of GKN1 being an autocrine/paracrine protein in gastric cells.
Impact of GKN1 on cell growth and cell cycle control
Figure 3A summarises the cell growth rates of gastric cells in response to GKN1 treatment. BGC-823 cell growth was significantly inhibited (p<0.001), but GES-1 cell growth was enhanced (p<0.01). The inhibitory effect of GKN1 on BGC-823 cells was dose dependent, with a half-maximal inhibitory concentration (IC50) at 1 μg/ml (equivalent to 60 nM). Consistent with the prior HeLa observation, GKN1 incubation did not exert any effect on HeLa cell growth (p>0.05). We also conducted a parallel experiment to monitor cell growth by bromodeoxyuridine (BrdU) incorporation (data not shown). The result coincided with the observation from the MTT assay, which demonstrated that GKN1 inhibited the growth of BGC-823 cells. Two groups used GKN1 to treat different cells and found that it caused opposite impacts on cell growth.4 5 9 We observed that GKN1 had different effects on cell growth even when the cells were from a similar background, such as GES-1 and BGC-823. The function of GKN1 seems to depend upon the cell type or the receptor of the cytoplasmic membrane.
The inhibition of cell growth by GKN1 raises the question as to whether these effects are associated with alterations in the cell cycle. The cell cycle was examined by flow cytometry. Compared with cells treated with control buffer, BGC-823 cells were arrested in G1 phase after GKN1 treatment (figure 3B).
GKN1 enables induction of senescence through activation of two senescent pathways
One of the biological consequences of G1 cell cycle arrest is senescence. We used SA-β-Gal staining to test if GKN1 could cause BGC-823 cells to senesce. Figure 3C shows that after incubating BGC-823 cells with GKN1 for 48 h, the cells displayed obvious senescent characteristics, whereas cells treated with control buffer did not exhibit any such signs. Furthermore, we sought to determine whether GKN1 natively expressed from gastric cells could induce senescence. The vector RNAi-GKN1 and the control, scrambled shRNA, were transfected into GES-1 cells, termed GKN1(–) and Mock, respectively (figure 3D, upper panel). We collected the medium after culturing Mock and GKN1(–) cells, and used them to treat BGC-823 cells for 48 h. Figure 3D (lower panel) shows that medium from Mock cells induced BGC-823 cell senescence but medium from GKN1(–) cells lacked this ability. This demonstrated that GKN1 generated from E coli or mammalian cells can exert the same effect on BGC-823 cells, which leads to changes in senescent phenotype.
GKN1 can stimulate senescence of BGC-823 cells; therefore, the next question is which senescent pathways are involved in this process. It is accepted that activation of the p16/Rb and/or p21waf pathway results in senescence. We selected three protein biomarkers, p16INK4a and Rb/pRb, both of which indicate the p16/Rb pathway, and p21waf, which represents the p21waf pathway. In the p16/Rb pathway, pRb, which is an inactive form of Rb, inhibits the function of transcriptional factor E2F. As shown in figure 4A, after incubation with GKN1, the levels of p16INK4a, p21waf and Rb were increased, whereas pRb was decreased, which indicated that both senescent pathways were activated. In GES-1 and HeLa cells, none of the biomarkers showed any changes in response to GKN1 (figure 4B,C). To further explore the senescent pathways related to GKN1, p16INK4a, p21waf alone or both were knocked-down in BGC-823 cells, termed p16RNAi, p21RNAi or p16/p21RNAi, respectively, and scrambled shRNA-transfected cells as a control termed control-RNAi. p16RNAi cells treated with GKN1 displayed an increased level of pRaf, pERK and p21waf when compared with control buffer treatment (figure 6B). In response to GKN1, p16RNAi cells grew slightly faster and showed decreased senescence when compared with control-RNAi cells treated with GKN1 (figure 4D, lower left); whereas the p21RNAi cells showed similar behaviour but with increased p16INK4a abundance (figure 4D, upper right, and figure 6). In contrast to the partial knock-down senescent pathways in BGC-823 cells, GKN1-treated p16/p21RNAi cells grew faster than control buffer-treated p16/p21RNAi cells or control-RNAi cells (figure 4D lower right). All the data reinforced the argument that activation of the p16/Rb and p21waf pathway plays a crucial role in the process of BGC-823 cell growth inhibition induced by GKN1.
Involvement of GKN1 in activation of the Ras/Raf/MEK/ERK pathway
The Ras/Raf/MEK/ERK pathway possesses dual functions: stimulation and inhibition of cell growth. Is this pathway involved in the regulation of gastric cell growth after incubation with GKN1? We examined whether pERK, which is an indicator for the Ras/Raf/MEK/ERK pathway, was enhanced in BGC-823 cells treated with GKN1. Figure 5A shows that pERK levels steadily increased during the incubation period of 48 h, which indicated that GKN1 probably activated this signalling pathway. Of the pERK activators, pRaf is believed to activate ERK and induce senescence. WB demonstrated that consistent with the kinetics of pERK, pRaf was also elevated after 48 h incubation with GKN1 (figure 5A). In contrast to sustainable activation of Ras/Raf/MEK/ERK in BGC-823 cells by GKN1, ERK and Raf were transiently phosphorylated within 30 min of being induced by GKN1 in GES-1 cells (figure 5B). We further checked if pERK and pRaf in HeLa cells were induced by GKN1 incubation, but GKN1 did not affect pERK and pRaf levels (figure 5C, right). To avoid any false-negative results caused by mutation in HeLa cells, we assessed the activation of the Ras/Raf/MEK/ERK pathway in response to epidermal growth factor (EGF). pRaf and pERK were clearly increased after 30 min incubation with EGF (figure 5C, left), which was consistent with a previous study.19 Taking these results together, we postulate that HeLa cells do not possess the GKN1 receptor on their membranes, whereas BGC-823 and GES-1 cells might share a similar membrane receptor for GKN1, but have different downstream signalling that leads to distinct responses in cell behaviour.
Involvement of GKN1 activation in the Ras/Raf/MEK/ERK pathway was further validated by the following inhibition experiment. We incubated BGC-823 cells with U0126, an inhibitor of pERK, followed by GKN1 treatment, and monitored the levels of ERK, pERK, p16INK4a and p21waf. As illustrated in figure 6A, pERK in BGC-823 cells treated with control buffer was attenuated by 6 μM U0126 and almost disappeared with 12 μM U0126. However, pERK in BGC-823 cells treated with GKN1 was also decreased by U0126 in a dose-dependent manner, but was still detectable with 12 μM U0126. Furthermore, we investigated how the senescence pathways responded to changes in ERK signalling. The changes in levels of p16INK4a and p21waf in BGC-823 cells treated with U0126 and GKN1 coincided with the pERK mode, while those in cells treated with U0126 and control buffer showed almost no change (figure 6A). This implies that the senescence pathways in BGC-823 cells, p16/Rb and p21waf, are located downstream of the Ras/Raf/MEK/ERK signalling pathway and are regulated by the activation status of their upstream effectors.
We established whether BGC-823 cell growth regulated by GKN1 was correlated with activation of pERK signalling. As shown in figure 4D (upper left), the growth curve for cells treated with U0126 and GKN1 was comparable with that of cells treated with control buffer; however, cell growth was significantly inhibited by incubation with GKN1 or U0126 alone. This means that GKN1 inhibits cell growth through activation of Ras/Raf/MEK/ERK signalling.
Suppression of BGC-823 tumorigenicity by GKN1 in xenograft nude mice
The results of the tumorigenicity study in nude mice following bilateral injection of BGC-823 cells treated with GKN1 or the supernatant fluid of E coli JM109 transfected with the pQE30 vector (as mock) are summarised in figure 7. As compared with mock treatment, the tumours were significantly shrunken when they were treated with GKN1 (figure 7A,B), the mitotic cells almost disappeared in the GKN1-treated mice (H&E staining; figure 7C), GKN1 was detectable on the cell membrane of these tumours (IHC images; figure 7D, arrow) and the levels of the protein markers related to the senescence pathways were significantly increased in the GKN1-treated tumour sections (figure 7D).
To confirm that GKN1 inhibited cell growth through the Ras/Raf/MEK/ERK pathway, we treated nude mice with U0126 by intraperitoneal injection. A MEK-inhibited animal model was constructed, because the expression level of pERK almost disappeared after U0126 injection (figure 7D). More importantly, GKN1 failed to inhibit tumour formation (figure 7A) or to affect the expression levels of the p16/Rb, p21waf and Ras/Raf/MEK/ERK pathways under these conditions (figure 7D). Clearly, these results are in agreement and consistent with those in our in vitro experiments described above.
From the outset of this study, we systematically confirmed the downregulation of GKN1 expression in GC tissues and cell lines.6 20 We further defined the mechanism of GKN1-induced senescence. Our aim was to explore in depth how GKN1 affects the growth of gastric cells, and show that it is an autocrine/paracrine protein that can induce senescence through the Ras/Raf/MEK/ERK pathway and its downstream p16/Rb and p21waf effectors in BGC-823 cells.
The presence of GKN1 in the antral mucosa has been reported previously.4–6 Oien et al have found that GKN1 tracking yields granular cytoplasmic staining with perinuclear accentuation that is suggestive of the Golgi apparatus.6 There has been a lack of direct evidence, however, as to how GKN1 is secreted from gastric cells. By confocal microscopy, using an endogenous source, we showed GKN1 co-mergence with the Golgi; using an exogenous source, GKN1 was only bound to BGC-823 cell membranes. The results of these investigations offered substantial evidence that GKN1 is an autocrine protein that is transported via the ER/Golgi secretory pathway. It is commonly accepted that secretory proteins play important roles in maintaining the physiological function of the gastric mucosa. Most secretory proteins from gastric cells are capable of autocrine/paracrine functions and are upregulated during GC development.21 22 For instance, heparin-binding EGF (HB-EGF) is present at a much higher level in GC compared with normal gastric tissues. Incubation of RGM-1 cells with HB-EGF results in activation of the Ras/Raf/MEK/ERK pathway and entry into S phase.21 Gastrin is generally acknowledged as an autocrine/paracrine protein and is a stimulatory factor for malignant development. In GC cells, gastrin is able to interact with its receptor and activate Ras/Raf/MEK/ERK pathways, which enhance cell growth.22 In contrast to these gastric autocrine/paracrine proteins, GKN1 is a novel autocrine/paracrine protein that exhibits markedly different specific biological patterns and functions, such as having a dramatically attenuated endogenous expression level, and its exogenous expression results in G1 phase cell cycle arrest in GC cells.
We found that GKN1 caused senescence, which is an important mechanism of tumour suppression, through p21waf and/or p16/Rb activation. This phenomenon is observed in GC cells.23 24 Based on the in vitro and in vivo experiments (figures 4, 6, 7), we confirmed that the senescence process induced by GKN1 in BGC-823 cells probably occurs through activation of the Ras/Raf/MEK/ERK pathway. It is generally accepted that transient activation of the ERK pathway favours cell growth.25 26 However, in certain circumstances, activation of the Ras/Raf/MEK/ERK pathway can lead to cell cycle arrest rather than cell growth, particularly for consecutive stimuli. For example, constitutive activation of a component of the Ras/Raf/MEK/ERK pathway induces p53 and p16INK4a, and leads to senescence of normal human fibroblasts.27 Matrix metalloproteinase-9 expression in Daoy cells is knocked down by transfection with small interfering RNA (siRNA). After 36 h transfection, cells show significant senescence and cell cycle arrest in the G1 phase, along with increased expression of p16INK4a and p21waf, and a consistently activated Ras/Raf/MEK/ERK pathway.28 As yet there is no explanation for why cell behaviour that is affected by transient activation of the ERK pathway remains so different from that affected by constitutive activation of the ERK pathway. However, our data offer another viewpoint from which to explore the relevant mechanisms. As seen in figure 5, incubation of BGC-823 cells with GKN1 results in steady activation of ERK within 48 h, which in turn brings on senescence. Similar incubation of GES-1 cells with GKN1 only results in transient activation of ERK, which leads to an increase in cell growth. Although the two cell lines are both derived from gastric tissues, GES-1 cells show relatively normal properties, whereas BGC-823 cells have the features of GC. Thus, it is clear that GKN1 is unable to exert the same effects on the two different types of gastric cells. This prompted us to investigate further and look for GKN1 receptors on the different membranes of BGC-823 and GES-1 cells, which will be described in a future study.
We are grateful to the Tissue Bank of Beijing Cancer Hospital for specimens. We also thank Dr Richard Ascione (Department of Biochemistry and Molecular Biology, Medical School of Georgetown University USA) and Dr Jun Yu (Department of Medicine and Therapeutic, Chinese University of Hong Kong) for critical reviewing and editing of the manuscript.
Siqi Liu and Youyong Lu are co-corresponding authors.
Funding This work was supported by the National Bio-Tech 86-3 (2006AA02A402), 97-3 program (2009CB522204) and Science and Technology Project of Beijing, China (D0905001040332).
Competing interests None.
Ethics approval This study was conducted with the approval of the Institutional Ethical Standards Committee.
Provenance and peer review Not commissioned; externally peer reviewed.