Enterochromaffin-like (ECL) cell hyperplasia and then irreversible neoplasia can be generated in the African rodent Mastomys natalensis using the H2 receptor blocker, loxtidine, for 8–16 wk. We used a GeneChip approach complemented by standard technologies to identify gene expression alterations in the gastric mucosa during gastrin-mediated ECL cell transformation. Gastric mucosa (mucosal scrapping) and ECL cell-enriched fractions were obtained from untreated Mastomys (controls) and from animals treated with loxtidine for 8 wk (hyperplasia). Tumor ECL cells were obtained by hand-dissection of gastric ECL cell nodules from animals treated with loxtidine for >16 wk and from a spontaneously developed ECL cell tumor. RNA was isolated, examined on rat U34A GeneChips, and comparison analysis was performed to identify altered gene expression. Alterations in gene expressions were examined further by immunohistochemistry, quantitative RT-PCR (Q-RT-PCR), sequencing and Western blot. GeneSpring analysis demonstrated alterations in few genes (<20) in hyperplastic and tumor mucosa. The histamine H1 receptor was consistently increased in proliferating mucosa. This gene change was confirmed by Q-RT-PCR. Other genes showing alterations included neural-(chromogranin A and somatostatin), cell-cycle-, and AP-1-associated genes. Immunostaining confirmed alterations in neural markers. Cluster analysis of ECL cell-enriched samples demonstrated that c-fos and junD were differently regulated. Q-RT-PCR and Western blot in prospectively collected gastric mucosal samples confirmed the differential expression of Fos and Jun. The negative regulators of AP-1, JunD, and Menin were decreased in tumor mucosa. A missense of unknown function was noted in the menin gene. Hypergastrinemia in an animal model of gastric carcinoids differentially altered the histamine type 1 receptor and gene expression and protein composition of AP-1. These results suggest that expression of this receptor and an altered composition of AP-1 with a loss of inhibition play a role in ECL cell transformation.
- enterochromaffin-like cell
the enterochromaffin-like (ECL) cell is the main regulatory neuroendocrine cell of the fundic gastric mucosa and plays a pivotal role in acid secretion (13, 26). It is also associated with pathology of the fundic mucosa, where it exhibits hyperplasia and even neoplasia (as gastric carcinoid), with initiation by low-acid states (26). As this gastric milieu is associated with common clinical events such as long-term pharmacological management of gastro-esophageal reflux disease (GERD), the control of ECL cell growth is an important issue in gastric mucosal biology.
Lifelong sustained acid suppression is known to generate growth perturbations in ECL cell populations in the gastric mucosa of the rat. A unique animal ECL cell tumor model, Mastomys natalensis, with an advantageously short experimental time frame (4 mo) has been developed in our laboratory to investigate ECL tumor biology (4, 29–31). This sub-Saharan African muroid rodent, with a closer phylogenetic affinity to the mouse than the rat (15), has a genetic propensity to spontaneously develop gastric carcinoid tumors (in ∼20% of animals) by the age of 18–24 mo (29). This lesion is histamine-secreting, is not driven by gastrin, and invades the submucosa, characteristics which render it closely comparable to human type III gastric carcinoid tumors. Administration of acid-inhibitory drugs at standard human therapeutic dosages rapidly accelerates this process, and after 8 wk of administration, all M. natalensis stomachs exhibit ECL cell hyperplasia; by 16 wk, more than 80% of animals have macroscopically evident carcinoid tumors. Normalization of gastrin levels by withdrawal of acid-inhibitory drugs after 8 wk of administration results in a reversal of gastrin-mediated mucosal alteration (apoptosis and proliferation) (19). We have previously characterized these gastrin-driven tumors in detail as neuroendocrine lesions of ECL cell origin and as histamine-secreting carcinoid tumors (30, 31); their growth is initiated by gastrin and subsequently becomes autonomous of gastrin (34). Removal of the gastric antrum ablates ECL cell tumor formation in vivo (27). The evolution of the tumor is thus similar to the human type I lesion.
Loss of function of the MEN-1 tumor-suppressing gene usually due to germ line mutations or loss of heterozygosity is associated with neuroendocrine cell proliferation characterized by the multiple endocrine neoplasia-1 syndrome (6). Most germ line or somatic mutations in the MEN1 gene predict truncation or absence of encoded menin, while a 11q13 loss of heterozygosity in a tumor predicts inactivation of the other MEN1 copy (3). MEN1 somatic mutations are prevalent in nonhereditary, MEN1-like tumor types. The MEN-1 protein product, menin, contains nuclear localization signals, translocates from the nucleus to the cytoplasm during the cell cycle (14, 16), and binds JunD, a repressor transcription factor that negatively alters the function of the activator protein-1 (AP-1) complex (2). The AP-1 transcription factor family is a fundamental class of transcriptional regulators comprising either Jun homodimers or Fos/Jun heterodimeric complexes (36) and is the paradigm for transcription factors that regulate aspects of cell physiology in response to environmental changes or growth factor signals (6, 7). There is evidence that AP-1 proteins, mostly those belonging to the Jun group, control cell life and death through their ability to regulate the expression and function of various cell cycle regulators (7). AP-1 function and regulation may be important in ECL cell proliferation and transformation.
One gastric receptor whose transcription is regulated by AP-1 is the histamine H1 receptor (8). The effects of histamine itself are mediated via AP-1 and a shift in AP-1 composition to Fos/Jun heterodimers (18). We have previously demonstrated that histamine, via activation of the H1 receptor, may play a role in Mastomys ECL cell tumorigenesis (28). In these studies, the selective H1-histamine antagonist, terfenadine, specifically inhibited the effect of loxtidine-induced ECL cell proliferation in vivo. This suggests that histamine, through an H1 receptor, positively modulates gastric ECL cell proliferation. These effects are potentially modulated by the AP-1 complex.
In this study, a GeneChip approach was employed to examine gene message in Mastomys gastric mucosa and in selectively enriched ECL cells to identify genes altered during gastrin-mediated ECL cell proliferation. We next examined the hypothesis that alterations in AP-1 may be associated with ECL cell tumorigenesis. Gene alterations were further explored using quantitative RT-PCR (Q-RT-PCR), DNA sequencing, immunohistochemistry, and Western blot approaches.
MATERIALS AND METHODS
Pronase E, collagenase I, and Taq DNA polymerase were from Boehringer (Mannheim, Germany). Nycodenz was from Accurate Chemical and Scientific (Westbury, NY). The total RNA isolation kit was the RNeasy kit from Qiagen (Valencia, CA), and the mRNA isolation kit was from Invitrogen (San Diego, CA). The Roche PCR Core Kit was used for PCR. Oligonucleotide primers were synthesized at the W. M. Keck Biotechnology Resource Laboratory (Yale University, New Haven, CT), and the QIAquick gel purification kit was obtained from Qiagen. The High-Capacity cDNA Archive Kit for Q-RT-PCR, TaqMan Universal Master Mix, and PCR probes were obtained from Applied Biosystems, Foster City, CA. The rat genome U34A and mouse genome U74A GeneChips were obtained from Affymetrix (Santa Clara, CA). Cyclin D1 sense and antisense primers were obtained from the Oligonucleotide Synthesis Resource (W. M. Keck Biotechnology Resource Laboratory, Yale University). Antibodies for immunohistochemistry were obtained from Accurate Chemical and Scientific [rabbit-anti somatostatin (SST) polyclonal] and from Dr. B. Erikson and K. Oberg, Uppsala, Sweden (rabbit anti-chromogranin A polyclonal), and those for Western blot were from Santa Cruz Biotechnology (Santa Cruz, CA) [rabbit monoclonal antibodies: c-Fos (6-2H-2F), c-Jun/AP1 (sc-45), menin (C-19), JunD (D-10), and β-actin antibody (C-11)]. Biotinylated secondary antibodies and horseradish peroxidase (HRP)-conjugated streptavidin were obtained from Dako, Carpinteria, CA. The ECL Western blot kit was obtained from Amersham (UK), and X-OMAT-AR film was from Kodak. All other chemicals including the histamine type 1 receptor antagonist, terfenadine, were purchased from Sigma (St. Louis, MO) or were of the highest quality available. Loxtidine was obtained from Glaxo Wellcome.
Loxtidine, an oral irreversible histamine type 2 receptor antagonist, was utilized to induce acid inhibition and hypergastrinemia (31). A total of 49 animals (3–6 mo of age) with equal sex distribution were randomly assigned to two groups: control and loxtidine (1 mg·kg−1·day−1). The animals were maintained on oral loxtidine treatment for 8–16 wk to generate hypergastrinemia, hyperplasia, and ECL tumors as described previously (5, 19, 22, 24, 28, 34). A gastric tumor sample was obtained from one additional aged Mastomys with a spontaneous ECL cell tumor.
Fundus (n = 15).
Preparations of total fundus (containing ECL cells) were obtained by mucosal scraping from five untreated animals, five animals treated with loxtidine for 8 wk, and five animals treated with loxtidine for 16 wk.
ECL cells (n = 5).
ECL cells were isolated from nontreated animals (n = 10 animals/preparation; n = 4 preparations) and animals treated for 8 wk (n = 10 animals/preparation; n = 1 preparation) using pronase digestion of everted gastric sacs, in alternating calcium-free and respiration media followed by elutriation and Nycodenz gradient centrifugation (30). This resulted in a 65–80% ECL cell population (∼1 × 106 cells, an enrichment of 32-fold over nonenriched gastric cells).
Tumor ECL cells (n = 6).
Preparations of tumor ECL cells were obtained from animals (n = 1 animal/preparation) treated with loxtidine (n = 5) or which had spontaneously developed a carcinoid (n = 1). The macroscopic gastric nodules were visually dissected, and the resulting cell mass was subjected to pronase (1.6 mg/ml)/collagenase (1.0 mg/ml) digestion and counterflow elutriation (34). This resulted in a 75–90% ECL cell population (∼3 × 106 cells; an enrichment of 25-fold over a nonenriched gastric mucosal cell preparation). Four of the five ECL cell tumor preparations were used in the histamine H1 receptor and cyclin D1 functional studies. The remaining sample was used in the DNA microarray study.
DNA Microarray Studies
Total RNA was isolated from each preparation (n = 16) using the RNeasy kit. Target cRNAs were generated and hybridized to either rat U34A (8,000 potential genes) or mouse U74A (11,000 potential genes) Affymetrix GeneChips. The data was analyzed using Affymetrix MAS 5.0 GeneChip software. The data generated using MAS 5.0 was further analyzed using various statistical software packages as described below.
Affymetrix MAS 5.0.
The Affymetrix MAS 5.0 software package (1) is the standard approach for analyzing gene alterations identified on Affymetrix arrays. A comparison analysis was undertaken between transcripts from different M. natalensis RNA samples. In a comparison analysis, two samples, hybridized to two GeneChip probe arrays of the same type, are compared against each other to detect and quantify changes in gene expression (1). One array was designated as the baseline (untreated) and the other as experimental (loxtidine treated). The Wilcoxon signed rank test used the differences between perfect-match and mismatch intensities, as well as the differences between perfect-match intensities and background levels to compute each change P value.
Data generated using MAS software was further analyzed using the GeneSpring program (35). The unsupervised two-dimensional cluster analysis was carried out as described in Virtanen et al. (35) using GeneSpring. In this method, clustering and gene clustering was performed independently using an agglomerative hierarchical clustering algorithm. Initially, genes were filtered according to their expression in the control channel. Genes in which >50% of the sample had an absorption measurement of <200 units were removed from the analysis. A Welch ANOVA was performed at the 0.01% significance level to find genes that vary significantly across samples.
The National Center for Biotechnology Information (NCBI) Blast program was used to examine the message of expressed sequence tags (ESTs) to identify homology and potential gene function against the working draft sequences of human, mouse or rat genomes. This was augmented by data available online at the Affymetrix website (http://www.affymetrix.com). The genes identified were further evaluated using an algorithm that specifically identified functional gene families using an in-house database application (Access 2000; Microsoft, Redmond, WA).
Hierarchical clustering was determined using the weighted-pair group method with centroid average as implemented by the Cluster 2.02 software (9). The distance matrixes used were of Pearson correlations for clustering the arrays, and the inner products were normalized to magnitude 1 for the genes involved. The subsequent results were analyzed using TreeView 1.45 software (9).
PCR was prospectively performed for selected genes using the ABI 7900 sequence detection system, as described (23), in samples from normal (n = 3), hyperplastic (n = 3), and tumor (n = 3) mucosa. Total RNA from each sample was subjected to reverse transcription using the cDNA Archive Kit following the manufacturer’s suggestions. Briefly 2 μg of total RNA in 50 μl of water was mixed with 50 μl of 2× RT mix containing reverse transcription buffer, dNTPs, random primers, and MultiScribe reverse transcriptase. RT reaction was carried out in a thermal cycler for 10 min at 25°C followed by 120 min at 37°C. Real-time PCR analysis was then performed in duplicate. Briefly, cDNA in 7.2 μl of water was mixed with 0.8 μl of 20× Assays on Demand primer and probe mix and 8 μl of 2× TaqMan Universal Master Mix in a 384-well optical reaction plate. The following PCR conditions were used: 50°C for 2 min, then 95°C for 10 min, followed by 40 cycles at 95°C for 0.15 min and 60°C for 1 min. A standard curve was generated for each gene using cDNA obtained by pooling equal amounts from each sample (n = 9). The expression level of the target genes was normalized to internal GAPDH. Data was analyzed using Microsoft Excel and calculated using the relative standard curve method (ABI, User Bulletin no. 2). The Assays on Demand primers used were as follows: for rat, GAPDH = Rn99999916, chromogranin A = Rn00572200, junD = Rn00824678, v-jun = Rn00572991, histamine H1 receptor = Rn00566691, and histamine H2 receptor = Rn00564216; for mouse, GAPDH = Mm99999915, chromogranin A = Mm00514341, SST = Mm00436671, fos = Mm00487425, and menin = Mm00484963.
Western Blot Analysis
Harvested cells from gastric mucosae from each of the treatment groups (nontreated, n = 1; 16 wk treated, n = 2) were boiled in SDS buffer for 5 min as described (34). Protein levels were measured using the Bio-Rad protein assay. Equal aliquots (∼40 μg) of total protein from the whole-cell extracts were fractionated on a 12% denaturing SDS-polyacrylamide gel and transferred to nitrocellulose membranes. Nonspecific interactions were blocked with 5% nonfat milk/0.1% Tween-20. Proteins were identified using rabbit monoclonal antibodies against c-Fos, c-Jun/AP1, Menin, and JunD (all 1:500 concentration), with a 1-h incubation at room temperature followed by incubation with peroxidase-conjugated secondary antibodies (goat anti-rabbit 40 mU) for 30 min. Membrane-bound antibodies were detected using a luminol-based chemiluminescent system (ECL Western blot kit). Immunoblots were exposed on X-OMAT-AR film under standard conditions. Equivalent loading of proteins was checked using β-actin antibody (1:500 concentration).
Stomachs were obtained after death and washed in PBS; the fundus was excised from the midportion of each stomach and immersed in 4% formaldehyde overnight (28). Fixed tissue was mounted edge-on in paraffin blocks. Deparaffinized sections were studied using hematoxylin and eosin staining for morphology and assessment of ECL cell distribution as described (28). Sections were incubated in 1% hydrogen peroxide for 10 min at 25°C to quench endogenous peroxidase activity, blocked in 10% normal porcine serum for 30 min at 25°C, and then incubated with rabbit anti-chromogranin A and anti-SST polyclonal antibodies. After rinsing, the biotinylated secondary antibodies and HRP-conjugated streptavidin were each subsequently applied to the sections at room temperature for 60 min. The peroxidase activity was visualized by applying diaminobenzidine chromogen containing 0.05% hydrogen peroxide for 5 min at 25°C. The sections were counterstained with hematoxylin, dehydrated, and mounted. The sections were examined and photographed using a Zeiss light microscope. The number of chromogranin A- and SST-positive cells in the gastric mucosa were assessed by examining sections (magnification of 450×) using a square grid of 10 × 10 mm (American Optical, Buffalo, NY) and noted as the number of cells per 1-mm2 zone. Only the areas with full-thickness mucosa were measured. An average of 45 (fundus) and 6 (antrum) grids from each section were counted. The cell density of chromogranin A- and SST-positive cells was calculated as follows (expressed as cell number/mm2): density = (total no. positive cells in all grids counted/number of grids) × (450/100 mm2).
Mastomys Menin Gene Sequencing
The Mastomys menin gene and transcript from the gastric mucosa was analyzed using direct sequencing of PCR products. Briefly, DNA (normal mucosa, n = 3; tumor mucosa, n = 7) or cDNA (normal mucosa, n = 3; tumor mucosa, n = 3) was PCR amplified at five segments of the menin gene spanning nucleotides −49 to +1670, which includes 557 of the 610/611 amino acid coding region (11). Primers used to amplify the menin gene are listed in Table 1. Conditions consisted of 35 cycles at 94°C for 1 min, 62°C for 1 min, and 72°C for 1 min. PCR reactions were performed in 25-μl reaction volumes containing 0.2 mM of each dNTP, 1.5 mM MgCl2, 1× reaction buffer (Roche PCR Core Kit), 0.2 μM of forward and reverse primer, 0.5 U Taq DNA polymerase (Roche), and 100 ng template DNA. PCR amplification products were excised and gel-purified using the QIAquick gel purification kit (Qiagen) per the manufacturer’s instructions, and product was eluted in 30 μl of water. Purified PCR products were sequenced and analyzed by the W. M. Keck Biotechnology Resource Laboratory at Yale University using an automated Applied Biosystems 373A Stretch DNA sequencer (PerkinElmer, Norwalk, CT). PCR products were initially sequenced using the forward primer. If ambiguous peaks were present, then the sequence was confirmed with the reverse primer.
Cell Culture and DNA Synthesis/Antisense Studies
Freshly isolated tumor ECL cells were rinsed in culture medium at 37°C (DMEM-F12 with 12% BSA at pH 7.4), and 5 × 104 cells were added to each well of a collagen I-coated 96-well plate in growth medium (culture medium plus 2% FCS, 0.5 mg/100 ml ITS, 10 nM hydrocortisone, and 0.1 mg/100 ml gentamicin) and cultured for 24 h (37°C, 5% CO2) as described (22). For the measurement of the effects of the H1 receptor, tumor cells were incubated with the H1 receptor antagonist terfenadine (10−12–10−6 M) for 24 h in the presence of bromodeoxyuridine (BrdU) 1:1,000. Thereafter, cells were fixed and wells incubated sequentially with anti-BrdU antibody/nuclease (60 min), rabbit anti-mouse secondary antibody (30 min), and peroxidase substrate. The color development (as a measure of BrdU uptake) was quantified at 405 nm (microplate reader 450, Bio-Rad).
For the effects of cyclin D1, 24-h cultured cells were incubated for 24 h in serum-deprived growth medium that contained antisense or non-sense oligonucleotides (0.1 μM). The cells were then challenged with the known ECL cell tumor cell growth stimulants TGFα (10−7 M) (20) or pituitary adenylate cyclase activating polypeptide (PACAP; 10−12 M) (22) for a further 24 h in the presence of BrdU, and uptake of this nucleotide was detected as described above.
Results were expressed as means ± SE; n indicates the number of ECL cell preparations (10 naive or hyperplastic animals or 1 tumor animal/ECL cell preparation). The error in gene expression for enriched ECL cell populations was the coefficient of variation (CV). This was calculated as: CV = e − 1, where e = standard deviation of the difference in gene expression/. Data are represented as percentages. Statistical significance was calculated by the two-tailed Student’s test for paired and unpaired values as appropriate, with a probability of <0.05 representing significance.
GeneChip and GeneSpring Examination of Mastomys Gastric Mucosa During ECL Cell Tumorigenesis
Gene alterations in gastric mucosa from animals exposed to hypergastrinemia of variable duration were initially examined. Hypergastrinemia in M. natalensis has been demonstrated to affect gastric mucosal cell proliferation (typically manifested as increases of up to 400%) and ECL cell proliferation (with more than 70% of ECL cells undergoing DNA synthesis) and concomitant increases in apoptosis (2-fold) (19).
We identified 1,359 genes to be expressed (present call) in mucosal samples from normal, hyperplastic, and tumor gastric mucosa (Fig. 1). The distribution of these genes in the different states is indicated in the Venn diagram. GeneSpring analysis was performed to identify which genes were altered in each state. Twenty genes were significantly altered in the hyperplastic mucosa. These are tabulated in Fig. 1. Genes identified to be upregulated in both hyperplasia and neoplasia included step II splicing factor SLU7, the histamine H1 receptor, transgelin, and RIKEN cDNA11100038D17. Genes downregulated included tachykinin receptor 1, tryptophanyl-tRNA synthetase, and an EST similar to serine-threonine protein kinase, pim3. Seven genes up- and downregulated were specific to the hyperplastic mucosa, whereas three up- and two downregulated genes were specific to the tumor mucosa. An examination of the overlap of genes identified to be differently expressed using Affymetrix (Venn diagram) and GeneSpring software programs indicated that 2/21 genes were concordant for the hyperplastic mucosa and 1/13 for the tumor mucosa. Genes identified to be altered by Affymetrix software in the hyperplastic mucosa included chromogranin A and gastrin, as well as a number of genes associated with mucosal remodeling (collagen, tubulin, and actin genes), and carbonic anhydrase II, and fibroblast growth factor-14. An examination of the genes identified to be altered in tumor mucosa included chromogranin A (Table 2), indicating an increase in ECL cell number. A plethora of other neuroendocrine-specific genes were also demonstrated to be present in the enriched ECL cell fraction and genes associated with the control of the cell cycle (cyclin G and p57KIP2). In addition, a decrease in marker genes for other gastric cell types, including SST, was noted.
The genes identified to be differently expressed by GeneSpring analysis were confirmed by Q-RT-PCR in Mastomys mucosa. Message levels (Fig. 2) for the histamine type 1 receptor were significantly elevated in hyperplastic (P < 0.05) and tumor (P < 0.01) samples, whereas no changes were noted in the histamine type 2 receptor.
Genes identified to be differently expressed including the markers chromogranin A and SST were confirmed by immunohistochemistry and Q-RT-CPR in Mastomys mucosa. Chromogranin A-positive cells significantly increased (P < 0.01) during hypergastrinemia (Fig. 3A), whereas the number of SST-positive cells was significantly (P < 0.01) decreased. Message levels (Fig. 3B) for chromogranin A were significantly elevated in hyperplastic (P < 0.002) and tumor (P < 0.0003) samples, whereas SST levels were significantly decreased (P < 0.05) in the tumor mucosa.
GeneChip Analysis of Enriched ECL Cells from the Mastomys
The method for isolation and enrichment of ECL cells from the Mastomys is well established. This process results in a 65–80% ECL cell population (an enrichment of 32-fold over nonenriched gastric cells) (30). Total RNA from seven different Mastomys ECL cell samples was subjected to GeneChip analysis using rat U34A arrays. A range of 1,041–1,723 (means ± SE: 1,327 ± 95) (∼17% of potential genes on the rat GeneChip) transcripts was identified as present in this experiment. The coefficient of variation for these samples was 9.7% across treatment types and 4.1% within treatment types. These samples were found to express the genes encoding preprochromogranin A (+1.0 → +5.8-fold), SST-14, and also for H+-K+-ATPase and pepsinogen. Fifteen percent of these genes were ESTs. Because of the presence of gene markers for other cell types, a cluster analysis was performed with the seven Mastomys samples and genes present in an enriched rat parietal cell and ECL cell databases as described (21). An examination of the 531 shared genes demonstrated that three of the seven Mastomys cell samples clustered in the same group as the rat parietal cells (Fig. 4). This demonstrated that these preparations contained insufficient ECL cell RNA for analysis and was confirmed by the observation that these samples had low chromogranin A levels [decreased 4- to 13-fold vs. the control (untreated) Mastomys ECL cell sample] and high H+-K+-ATPase (increased 7-fold), suggesting parietal cell contamination. In these cases, the low abundance of ECL cell mRNA species was overshadowed by the higher abundance of contaminating parietal cell mRNA species; these samples were thus excluded from further analysis. The Mastomys samples that contained sufficient ECL cell gene expression clustered separately from those tissues already mentioned. Two tumor samples [one from a hypergastrinemic animal (ML1) and one spontaneous tumor (MS1)] were more closely related than the ECL sample from the animals treated with loxtidine for 8 wk (MH1).
Comparing the ECL cell gene expression in these three proliferating ECL cell samples with the control (untreated) Mastomys ECL cell sample demonstrated the following alterations in gene expression.
Eight weeks of loxtidine treatment (MH1) resulted in upregulation of 46 genes, including insulin-like growth factor binding protein 5 (+1.7, P < 0.0005), the insulin I gene (+1.1, P < 0.0001), and furin (+1.7, P < 0.0005), and downregulation of 120 genes, including fos-transformation effector protein (−4.3, P = 0.0006) and the early response gene NGFI-A (−3.5, P = 0.0004). A large number of ribosomal genes were also downregulated. Hypergastrinemia for 8 wk, therefore, increased insulin-mediated cell function and altered fos in ECL cells.
Neoplasia: gastrin driven.
Sixteen weeks of loxtidine treatment (ML1) resulted in upregulation of 139 genes, including parathyroid hormone receptor (+15, P < 0.0007), furin (+4.6, P < 0.0001), and c-erbA thyroid hormone receptor (+2.1, P < 0.001), and downregulation of 103 genes including NGFI-A (−32.0, P < 0.000001), SST (−11.3, P < 0.000001), mucin-5 (−1.7, P < 0.002), and genes involved in the AP-1 pathway (Table 3). The insulin I and II genes were upregulated, whereas a large number of ribosomal genes were downregulated. Gastrin-mediated ECL transformation resulted in alterations in insulin-mediated cell function, a dysregulation of ribosomal gene expression, and alterations in the AP-1 pathway.
A comparison of MH1 and ML1 to identify gastrin-responsive (predicted to be in MH1) and gastrin-autonomous (predicted to be in ML1) genes demonstrated that the early response gene krox-24 ( −5.3, P < 0.0001) was downregulated in ML1 compared with MH1. This result demonstrated that Krox was expressed as a consequence of hypergastrinemia and was not a marker for ECL cell tumor autonomy. In contrast, the ornithine decarboxylase (ODC) antizyme (+4.0, P < 0.0001) and the fos-effector protein (+2.1, P < 0.002) were upregulated in the autonomous (ML1) lesion, indicating that genes regulating fos (AP-1) activity may be potential markers of ECL cell tumor autonomy.
Normogastrinemic ECL cell tumor (MS1), which is analogous to the human type III tumor, demonstrated upregulation of 148 genes including furin (+3.2, P < 0.00001), insulin I and II genes (+1.3–1.6, P < 0.0001), and trefoil peptide pS2 (+1.1, P < 0.0006) and downregulation of 99 genes including NGFI-A (−11.3, P < 0.000001) and genes involved in the AP-1 pathway (Table 3). A large number of ribosomal genes were also downregulated.
These data demonstrate that a number of common pathways (insulin, ribosomal, furin) are altered during ECL cell proliferation irrespective of whether cells are hyperplastic or autonomous. In contrast, both types of Mastomys ECL cell tumors exhibited alterations in the AP-1 gene pathway (fos, jun) (Table 3).
Examination of the AP-1 Pathway
The AP-1 pathway and its potential regulators were next examined in more detail. Alterations in gene and protein expression were examined in prospectively collected gastric mucosal samples. First, the alterations in gene expression of the components of AP-1, jun and fos, and its regulators junD and menin were measured using Q-RT-PCR (Fig. 5). Consistent alterations in message were noted in tumor mucosa. Specifically, both jun and fos were significantly (P < 0.05) increased in tumor mucosa compared with normal mucosa. In contrast, the regulators junD and menin were significantly decreased in tumor mucosa (P < 0.05). The possibility that protein levels of these genes were altered was next evaluated (Fig. 6). The AP-1 complex in the naive gastric mucosa was signified by the presence of Fos but not Jun/AP-1 gene products (N) (Fig. 6A). In contrast, this was altered in tumor mucosal samples (T1, T2); both Fos and Jun/AP-1 proteins were specifically upregulated. Message levels of Jun were low in both normal and hyperplastic samples, a finding probably reflected in the lack of signal noted in the Western blot. Protein levels of the regulators of AP-1 were significantly repressed in both tumor samples (Fig. 6, B and C).
Mastomys Menin Gene Sequence
The nucleotide sequence of the protein-coding region of Mastomys menin in gastric mucosa isolated from untreated and loxtidine-treated animals was then determined. We examined the first 1,669 nucleotides of cDNA and identified 59 base pair changes compared with the mouse menin cDNA. Seven nucleotide changes involved the first codon position changes, three of which were silent. The other four first codon nucleotide alterations results in the following amino acid changes: Thr → Ala385, Thr → Pro390, Gly → Pro529, and Gly → Ser532. None of these changes resulted in novel amino acid substitutions, as prolines at position 390 and 529, alanine at 385, and serine at 532 were also identified in the human and/or rat menin protein. One nucleotide alteration in a second codon position, GGG → GCG, resulted in a Gly → Ala511 amino acid change. This was a novel missense mutation and was not noted in any of the other three species examined (mouse, rat, and human). Fifty-one nucleotide alterations were noted in third codon positions. The deduced amino acid sequence of M. natalensis menin is highly homologous to rat, mouse, and human menin (Fig. 7). One missense but no frame-shift mutations were noted in the M. natalensis menin protein. No alterations were noted in the JunD binding region or the nuclear localization signals within Mastomys menin. The functionality of the missense at codon 511 is not known.
Functional Examination of the AP-1 Target Gene: Histamine H1 Receptor
First, the alterations in gene expression of histamine H1 receptor were measured using Q-RT-PCR (Fig. 8A). This was significantly (P < 0.05) increased in enriched tumor ECL cells compared with normal ECL cells. The role of histamine and specifically histamine H1 receptor in mediating cell proliferation in tumor ECL cells was then evaluated by examining the effects of the addition of the specific H1 receptor antagonist, terfenadine, against TGFα-mediated DNA synthesis. As expected, TGFα stimulated BrdU uptake into tumor cells (Fig. 8B). Although terfenadine alone (10−12–10−6 M) had no effect on tumor DNA synthesis, it selectively inhibited TGFα-mediated DNA synthesis with an IC50 of ∼10−10 M.
Functional Examination of the AP-1 Target Gene: Cyclin D1
First, the alterations in gene expression of cyclin D1 were measured using Q-RT-PCR (Fig. 9A). This was significantly (P < 0.05) increased in both tumor mucosa and in enriched tumor ECL cells compared with normal mucosa or normal ECL cells, respectively. The role of cyclin D1 in mediating cell proliferation in tumor ECL cells was evaluated by examining the effects of the addition of antisense (D1/AS) against this gene on TGFα- and PACAP-mediated DNA synthesis. As expected, both TGFα and PACAP stimulated BrdU uptake into tumor cells (Fig. 9, B and C). Addition of 0.1 μM cyclin D1 antisense significantly decreased the proliferative effect; this was not seen when non-sense oligonucleotide was used.
In this study, a GeneChip expression analysis approach was used to identify gene pathways that were altered during gastric ECL cell tumorigenesis in the Mastomys model.
Statistical analyses of the genes identified in the gastric mucosa identified a subset of genes that were consistently altered in both hyperplasia and neoplasia in this model. One gene that may be of physiological relevance and was consistently altered in both states was the histamine type 1 receptor. We have previously demonstrated that this receptor may play an important role in gastrin-driven ECL cell tumorigenesis in vivo (28). The results in the current manuscript identify that this gene is overexpressed, which was confirmed by Q-RT-PCR and functionally by the determination that the H1 receptor antagonist, terfenadine, selectively reversed DNA synthesis in tumor cells in vitro.
Histamine is a well-known mitogen, and receptor antagonism (particularly of the parietal cell expressed type 2 receptor) has been suggested as important in gastric growth (37). Nothing is known about the proliferative effects of the H1 receptor in the gastric mucosa. One study in astrocytoma cells (U373 MG cell line) demonstrated that activation of the histamine type 1 receptor resulted in DNA synthesis that was mediated via the MAPK pathway (12). The GeneSpring data suggest that the neuroendocrine ECL cell-enriched mucosal fractions express the H1 receptor. The in vitro cell proliferation studies in enriched tumor ECL cells support this function.
Affymetrix microarray analysis identified a range of other transcripts that were variably expressed in the gastric mucosa. These included gastrin in hyperplastic mucosa. Serum gastrin levels are significantly increased in these animals (19), and an increase in gastrin mRNA levels would be expected. Other genes altered by hypergastrinemia included markers for neuroendocrine cells. Specifically, ECL cell numbers were increased, but both SST mRNA levels as well as the number of SST-expressing cells were identified to be decreased. SST-producing D cells exist in relatively large numbers in both the fundus and the antrum of the stomach; the primary effects of SST are to inhibit gastrin release and negatively modulate gastric acid secretion (13). In addition, SST is also a known inhibitor of ECL cell proliferation both in vivo and in vitro (5). The former observation is thought to occur via inhibition of gastrin and/or local growth factor release (5). The mechanism for the decrease in D cells in the gastric mucosa during hypergastrinemia is not known, but a sustained increase in plasma gastrin levels (for at least 8 wk) results in the removal or reduction of this important negative regulatory hormone (SST) in Mastomys gastric mucosa. The findings of alterations in cell-cycle regulators coupled with the decrease in SST in the present study suggest an impaired feedback control of proliferation in ECL cell tumorigenesis in M. natalensis.
Genes identified to be altered in enriched ECL cells included upregulation of the early response gene krox-24, genes involved in insulin biosynthesis and signaling, and furin, and downregulation of ribosomal genes and the immediate early response gene NGFI-A. The importance of these gene expression alterations is largely unknown, but suggests the ECL cells are responding to gastrin in an acute response fashion (perhaps to fluctuating circulating levels of the hormone). Neoplastic ECL cells were signified by alterations in insulin-mediated cell function, identification of the H1 receptor and alterations in the AP-1 pathway. The former is of interest, as an earlier study from our group has demonstrated that insulin-like growth factor and its receptor may be an important mitogenic factor for the ECL cell (33). Specifically, IGF-1 stimulated tumor ECL cell DNA synthesis with an EC50 of 2 × 10−12 M. IGF-1 immunopositivity, while absent in tumor cells per se, was evident in cell populations in the transformed gastric mucosa, and the IGF-1 receptor was present on the Mastomys ECL cell. A recent study has demonstrated the presence of the IGF-1 receptor on a human gastric ECL cell carcinoid (38). The insulin pathway is known to be obligatory for cell transformation (39), and the AP-1 pathway is implicated in expression of this gene (17). The identification of the H1 receptor and alterations in AP-1 are of interest because this suggests a mechanism by which ECL cell proliferation is modulated.
An examination of the AP-1 pathway in the transformed Mastomys mucosa demonstrated that this was altered during gastrin-driven ECL cell proliferation. An examination of the proteins encoding the Fos and Jun components of this molecular machinery demonstrated that both of these proteins were upregulated in tumor mucosa compared with normal mucosa. This was associated with a corresponding alteration in message for these factors. The alteration in AP-1 composition is similar to that identified in keratinocytes stimulated with histamine (18). In an earlier, preliminary study, our group demonstrated that fos/jun message was also upregulated by gastrin in rat ECL cells maintained in short-term culture (10).
AP-1 activity is negatively regulated by the transcriptional repressor, JunD. Both the protein and gene levels of this factor were decreased in tumor tissue. In addition, its catalytic partner menin was also decreased. Examination of the first 1,669 base pairs of the coding region identified 54 synonymous alterations and five missense mutations in the Mastomys menin gene. Alignment of the predicted protein sequences with mouse (accession no. NM_008583) indicated that the percentage DNA homology was 96.5%. Four of the five amino acid differences were largely limited to interspecies differences and were not consistent with known human germ line mutations (6). One missense was unique to M. natalensis. A G → A transition at coding site 1464 resulted in a Gly → Ala substitution. The effect of this on menin function is not known. The possibility that this results in a significant decrease in protein expression in tumor specimens is unlikely, as this was identified also in normal specimens. The decrease in mRNA and protein expression probably reflects a transcriptional modification. The absence (downregulation) of JunD and menin in tumor ECL cells suggests a mechanism for impaired/altered control of AP-1 function. The modification in AP-1 composition itself supports a conclusion that the function of this transcription machinery may be altered during tumorigenesis. Earlier studies have determined that the Jun group of AP-1 proteins (c-Jun in particular), largely control cell life and death through their ability to regulate the expression and function of the cell cycle regulators such as cyclin D1 and the negative regulator, p16 (7, 36). We have demonstrated that both message and protein levels of cyclin D1 are increased during ECL cell tumorigenesis (40). In the current study we demonstrated that cyclin D1 was critical for tumor cell proliferation in vitro. Antisense strategies against this gene completely reversed the proliferative effect of a number of known ECL cell mitogens.
The spontaneous ECL cell tumor (proliferation initiated as a function of age and not due to gastrin stimulation), shared a number of characteristics with the gastrin-driven tumor but was signified by upregulation of the trefoil factor pS2. This factor protects the gastric epithelium from injury, promotes mucosal repair through restitution after injury has occurred (32), and is upregulated during ulcerogenesis (25). Aged animals that develop carcinoids also develop duodenal ulcers (29) and could be expected to develop mucosal injuries requiring altered expression of trefoil factors.
In conclusion, this study using both GeneChip microarray analysis and more conventional technologies has identified that gastrin-driven ECL cell tumorigenesis in the Mastomys model is associated with an increase in chromogranin A, a decrease in SST levels, an upregulation of the histamine type 1 receptor, and alterations in AP-1. AP-1 directs mucosal ECL cell cycle progression by altering cyclin D1 expression. Findings of alterations (decrease in expression) of the negative regulators of AP-1 (JunD and Menin) provide a potential mechanism for ECL cell growth (Fig. 10). It appears likely that the trophic stimulus of gastrin in the Mastomys initiates ECL cell proliferation via modifications in AP-1 and that this effect may be augmented by histamine and the type 1 receptor. Long-term hypergastrinemia is associated with the loss of inhibition of and a compositional change in AP-1, with a concomitant alteration in transcriptional regulation of the ECL cell that appears to drive the neoplastic process in the stomach.
This work was funded by National Institutes of Health Grant R01-CA-097050-01 (to I. M. Modlin).
Article published online before print. See web site for date of publication (http://physiolgenomics.physiology.org).
Address for reprint requests and other correspondence: M. Kidd, Yale Univ. School of Medicine, Dept. of Surgery, TMP203, 333 Cedar St., New Haven, CT 06520 (E-mail:).
- Copyright © 2004 the American Physiological Society