Gastrokine 1 (GKN1) is an 18 kDa stomach-specific protein that plays an important role for the physiological function of the gastric mucosa, for the renewal of the epithelial cell layer and in maintaining the integrity of the gastric mucosa. The down-regulation of the protein is associated with an increased risk of gastric cancer development. The protein is, in fact, down-regulated in Helicobacter pylori-infected gastric tissues and its expression is almost detected or completely lost in gastric cancer tissues and precancerous lesions   . So it is clear that the expression of GKN1 decreased throughout the progressive stages of neoplastic transformation, from gastritis to intestinal metaplasia, and that it plays an important role in the process of formation and development of gastric cancer. It has also been demonstrated that GKN1 expression induces apoptosis in gastric cancer cells  . This finding was also confirmed by the ability of recombinant human GKN1 (rhGKN1) to exert a higher antiproliferative effect on gastric cancer cells (AGS) compared to normal human embryonic kidney cells (HEK 293) and non-gastric cancer cells (H1355)  . The higher sensitivity of AGS cells to GKN1 exposure was likely linked to the role played by the protein in maintaining gastric mucosal integrity and to its function as a gastric tumor suppressor.
Regarding the mechanism by which GKN1 gene is down-regulated in gastric cancer, there still remains an open question even though epigenetic mechanisms seem to be involved in this process. In fact, we showed for the first time a possible mechanism of histone modifications that could lead to dysregulate GKN1 gene transcription in gastric cancer. We have found the massive presence of the repressive histone modification H3K9me3 (histone 3 trimethylation at lysine 9) and the recruitment of the specific histone methyltransferase SUV39H1 (histone-lysine N-methyltransferase) on the GKN1 gene promoter in human tumor tissues with respect to normal tissues  . Under this regard, both histone modification H3K4me2t (histone 3 dimethylation at lysine 4) and H3K4me3 (histone 3 trimethylation at lysine 4) are proposed to be epigenetic marks. In particular, H3K4me3 is associated with functional or open chromatin   whereas H3K9me2 is a mark linked with silenced/repressed euchromatin instead   . Therefore, H3K4 methylations levels could be indicative of chromatin in a poised state  .
Polycomb group (PcG) proteins act at different chromosomal sites by forming a complex, the polycomb repressive complexes (PRC) leading to the transcription repression of gene expression through an epigenetic mechanism  . Chromobox homolog 7 (CBX7) is a member of PcG family proteins that act in an independent manner during the progression of several types of cancer      . The association between the absence of CBX7 and the appearance of a highly malignant phenotype seems to be an oncologic characteristic. In fact, CBX7 was found overexpressed in gastric cancer cell lines and gastric tumors and correlated with clinical stage, patients’ age, and lymph node metastasis   . Loss of CBX7 expression in gastric cancer cells led to increased cellular senescence decreased cellular proliferation and migration ability  . However, the role of CBX7 in cancer development is still not clear since it has been demonstrated that it may behave as a tumor suppressor or act as an oncogene in relation to cancer type and cellular context      .
In this work, we focus on trimethylation of H3K4 (H3K4me3) as a possible epigenetic mark associated with GKN1 down-regulation and tried to highlight if CBX7 expression at protein level was also associated to GKN1 behavior in gastric cancer tissues.
2. Materials and Methods
Mouse GKN1 monoclonal antibody (M01), clone 2E5, was purchased from Abnova (Taipei, Taiwan). Rabbit polyclonal CBX7 (P-15) antibody [sc-70232] was from Santa Cruz Biotechnology (Dallas, TX, USA).
2.2. Cell Cultures, Human Tissues, and Western Blotting
Human gastric adenocarcinoma cell line (AGS) was grown in DMEM-F12 supplemented with heat-inactivated FBS, 1% penicillin/streptomycin and 1% L-glutamine at 37˚C in a 5% CO2 atmosphere.
Human GC tissues were from our collection of specimens as reported  . All patients were interviewed for their habit (alcohol intake, chronic use of drugs, smoking dependency). During surgery, Hospital Pathologist performed the dissection of non-tumor (N) and tumor tissues (T) and characterized the non-tumor gastric mucosa from the comparison between the macroscopic aspects of normal adjacent tissue versus the tumor one  . GC was staged and graded according to the American Joint Committee on Cancer criteria  . This study was approved by the University of Naples Federico II Ethics Committee (Comitato Etico Università Federico II) protocol number 34/15. All western blots were performed on 20 μg of cell extracts using mouse anti-GKN1 at 1:500, mouse anti-CBX7 and anti-α-Tubulin at 1:1000 dilutions. Gel bands were highlighted with the chemiluminescence detection kit (SuperSignal West Pico). Gel bands intensity was evaluated with ImageJ software Version 2.0.0-RC-43/1.51k.
2.3. Chromatin Immunoprecipitation (ChiP) Assay
ChiP assays on human specimens and cell line were carried out as already reported  . Samples were subjected to immunoprecipitation (IP) with the specific antibody against histone modification anti-tri methyl K9-Histone3, anti-tri methyl K4-Histone3, and anti-acetyl H3. qRT-PCR was performed as described  .
2.4. Statistical Analyses and Bioinformatics
Data were statistically analyzed with two-tailed paired Student’s t-test using KaleidaGraph 4.1.1 software and reported as means ± standard deviation (SD). The significance was accepted at the level of p < 0.05.
UCSC (University of California Santa Cruz) genome browser was explored for H3K4me3. CBX7 binding sites were searched “on-line” using TFbind software  (Suehiro, Tsurumi, Yokohama, Japan), LASAGNA-Search 2.0  , and Alggen Promo Version 3.0.2   . OpenEpi Version 3.01 was used for power calculation.
3. Results and Discussions
3.1. GKN1 Down-Regulation in GC Is Associated with Trimethylation of Histone 3 at Lysine 4 on the GKN1 Promoter
In previous work, we investigated a possible epigenetic mechanism underlying the GKN1 gene silencing in gastric carcinogenesis. We exemplified a functional model where a transcription factor might regulate the GKN1 promoter function by inducing histone methylation and deacetylation through SUV39H1 and HDACs (histone deacetylase), respectively and thus GKN1 gene inactivation. In fact, chromatin immunoprecipitation assays for H3K9me3 showed that GKN1 down-regulation in gastric cancer (GC) tissues were associated with a high level of H3K9me3. To highlight if also H3K4me3 was associated with the epigenetic mechanism leading to the inactivation of GKN1 gene, ChiP assays were performed. We chose to focus on H3K4me3 since together to H3K4me2, they are considered epigenetic marks linked with actively transcribed or transcriptionally poised genes    . As already reported, the experiment was carried out on three regions (A, B, and C) of the GKN1 promoter spacing for about 600 bp  using two paired non-tumor (N1-N2) and tumor (T1-T2) specimens of gastric tissues taken from patients after surgery (Table 1). We first analyzed in these tissues the expression level of GKN1. Compared to non-tumor tissues, the expression of GKN1 in the tumor tissues was completely absent (Figure 1A) whereas, a significant increase of H3K9me3 in tumor samples was observed  . ChiP assays performed on the same GKN1 promoter regions (A, B, and C) revealed, compared with non-tumor tissues, an increase of H3K4me3 in T1 tumor tissue (Figure 1B) that was much less pronounced in sample T2 (Figure 1C). In fact, a very low value of H3K4me3 enrichment and high values for the ChiP control samples were observed for sample T2. These results somehow reflect some controversial aspects regarding the role of H3K4me3 that is widely considered as an activating mark of transcription. In fact, evidence support that H3K4me3 is unlikely to be regulation of transcription initiation since its absence in the majority of cases seems to have no effect on nascent transcription  . The finding that the levels of H3K4 methylation could be a mark of chromatin that is kept in a “poised” state where is not actively transcribed, as also reported by Schneider et al., 2004  in the case of inactive genes within the β-globin locus. To search for existing histone modification tracks, UCSC genome browser platforms were explored. No H3K4me3 was present in various cell lines in the region surrounding the GKN1 transcription start site (TSS). Low level of GKN1 mRNA and Pol2 and absence of H3K4me3 was instead observed in HeLa S3 clone despite the presence of an enhancer of H3K4 mono-methylated of about 1 kb upstream the GKN1 TSS.
One criticism of the present study regards the number of GC specimens used. Although the results obtained on two GC patients were roughly similar, to be confident that a mechanism underlying GKN1 gene silencing is also based on the up-regulation of histone methylation H3K4me3 by SUV39H1 and deacetylation mediated by HDAC1, a larger number of samples should be analyzed. This aspect is not easy to afford because of the difficulty in the recruitment of GC patients, in the quality and quantity of tissues specimens that can be obtained after surgery. Moreover, also the intrinsic difficulty of the ChiP assay experimental procedure should be taken into consideration.
To further investigate the epigenetic modifications on GKN1 gene expression, we assayed the role of histone acetylation by using trichostatin A (TSA), an inhibitor of histone deacetylases (HDACs). As shown in Figure 2, ChiP assay performed on AGS gastric cancer cell line after treatment with TSA for 24 h showed a reduction of H3K9me3 repressive modification (Figure 2A) and of H3K4me3 (Figure 2B) that was associated by the increase of H3-acetylation-activating modification in the same three regions of the GKN1 promoter (Figure 2C). This result confirmed somehow the data obtained in human gastric tissues (Figure 1) and suggest that in this case H3K4me3 is not associated to transcription since the treatment of gastric cancer cell line MKN28 with TSA resulted in an increase of GKN1 mRNA level  .
Figure 1. H3K4me3 levels on human GKN1 gene promoter. ChiP assays performed on human non-tumor (N1-N2) and tumor (T1-T2) human gastric samples, respectively. H3K4me3 enrichment relative to input is reported as 2ΔCt × 100, where ΔCt is the difference between CtInput and CtIP. All quantitative ChiP data were derived from three independent experiments, and for each experiment, qPCR was performed in triplicate. * p < 0.05, compared to corresponding control.
Figure 2. TSA induces in AGS cells decreased levels of H3K9me3, H3K4me3, and increased levels of H3 acetylation. Levels of H3K9me3 (A), H3K4me3 (B), and H3 acetylation (C) determined by ChiP assays on AGS cells not treated (−TSA) and treated with TSA (+TSA). H3K9me3, H3K4me3, and H3 acetylation enrichment relative to input are reported as 2ΔCt × 100, where ΔCt is the difference between CtInput and CtIP. All quantitative ChiP data were derived from three independent experiments, and for each experiment, qRT-PCR was performed in triplicate. *p < 0.05, compared to corresponding control.
Table 1. Clinical and histopathological characteristics of gastric cancer patients*.
*Non-tumor (N3-N7) and tumor (T3-T7) specimens reported in Figure 3 were already used in previous work to analyze the GKN1 expression levels (Di Stadio et al., 2017). P = Patients.
3.2. CBX7 and GKN1 Expression Levels in GC Non-Tumor and Tumor Tissues
In previous work, we showed that RUNX3 was not a possible transcription factor functioning as a negative regulator able to induce GKN1 gene inactivation. Because CBX7 is reported to be overexpressed in gastric tumors  , we tried to highlight if CBX7 could be the specific transcription factor able to bind GKN1 promoter and thus to promote the formation of a multifactor complex including SUV39H1 and/or other histone modifier enzymes involved in GKN1 down-regulation during gastric carcinogenesis. Therefore, we analyzed by Western blot the expression level of CBX7 in a collection of paired non-tumor and tumor specimens from patients with gastric cancer during surgery (Table 1). The down-expression of GKN1 observed in gastric tumor tissues (Figure 3) was not always accompanied by the up-regulation of CBX7. This observation was established from the Western blot band intensity ratio (Figure 3) of non-tumor tissue and that of the corresponding tumor one minus 1 (N/T-1) (Figure 4). However, the intensity values evaluated with the Image J software represent only a trend since they were not statistically significant. This aspect was also confirmed by the power calculation of the average difference between the two populations (GKN1, and CBX7 band intensities). With a power of 80% at 95% confidence level, the sample size required was 301. Moreover, the calculated coefficient of variation (38.7% for GKN1, and 34.1% for CBX7) indicated a large biological variability within each group.
Figure 3. CBX7 and GKN1 expression levels in human gastric tissues. The expression levels of CBX7 in non-tumor (N3-N9) and tumor (T3-T9) paired gastric samples and corresponding expression levels of GKN1 were evaluated with ImageJ software. (A) AGS gastric cancer cell line. (B) Normal gastric mucosa. (a, b, c, d) average band intensity ± SD of three measurements. SD: Standard deviation.
Figure 4. Comparison between GKN1 and CBX7 expression levels of in the tumor tissues analyzed. The relative expression of GKN1 and CBX7 was determined from the band intensity ratio (N/T) of non-tumor tissue and that of the corresponding tumor one, minus 1 (N/T−1).
In present study, we addressed the role of H3K4me2 as a possible epigenetic marker involved in GKN1 gene silencing in gastric carcinogenesis. Our findings, and in particular those observed for sample N1/T1 and for AGS cells, as opposed to the proposed role of the K4 trimethylation of histone H3 as a transcription activator, reinforces the controversial hypothesis that its deposition is transcription independent. Moreover, our experimental evidence showed that compared to non-tumor tissues, the expression of CBX7 in tumor gastric tissues was not always overexpressed and thus associated with the down-regulation of GKN1. ChiP assay might clarify the possible binding of CBX7 on the GKN1 promoter since bioinformatic tools like TFbind, LASAGNA and PROMO did not highlight CBX7 consensus binding sequence on GKN1 promoter. In particular, PROMO home page predicted a large number of possible transcription factors but dissimilarity margin less or equal to 15%. The result strongly suggests the non-involvement of the transcription factor in the regulation of GKN1 silencing in gastric cancer.
The authors thank Dr. G. de Dominicis, Hospital Cardarelli, Naples, Italy, for tissue histopathological classification and immunohistochemistry. This work was supported by funds from Programmi di Ricerca Scientifica di Rilevante Interesse Nazionale (2012CK5RPF_004), PON Ricerca e Competitività 2007-2013 (PON01_02782) and POR Campania FSE 2007-2013, Project CRÈME.
CSDS and FA were involved in ChiP assays, AF performed qRT-PCR analysis and statistical evaluation, GM was involved in WB and cell extract preparation, MG takes care of patient recruitment, ER was involved in the preparation of a draft manuscript, PA in the reading and approval of the final manuscript.
ChiP: Chromatin Immunoprecipitation;
GC: Gastric Cancer;
GKN1: Gastrokine 1;
HDAC: Histone Deacetylase;
H3K4me3: Trimethylation of Histone 3 at Lysine 4;
H3K9me3: Trimethylation of Histone 3 at Lysine 9;
TSA: Trichostatin A;
TSS: Transcription Start Site;
SUV39H1: Histone-Lysine N-Methyltransferase.
*Equally contributed to the work.