If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
Nashville Veterans Affairs Medical Center and the Epithelial Biology Center and the Section of Surgical Sciences, Vanderbilt University School of Medicine, Nashville, Tennessee
Department of Biochemistry, Jim Ayers Institute for Precancer Detection and Diagnosis, Vanderbilt University, Nashville, TennesseeDepartment of Pharmaceutical, Social, and Administrative Sciences, Belmont University College of Pharmacy, Nashville, Tennessee
Nashville Veterans Affairs Medical Center and the Epithelial Biology Center and the Section of Surgical Sciences, Vanderbilt University School of Medicine, Nashville, Tennessee
Department of Surgery, Seoul National University College of Medicine, Seoul, KoreaCancer Research Institute, Seoul National University College of Medicine, Seoul, South Korea
Department of Surgery, Seoul National University College of Medicine, Seoul, KoreaCancer Research Institute, Seoul National University College of Medicine, Seoul, South Korea
Cancer Research Institute, Seoul National University College of Medicine, Seoul, South KoreaDepartment of Pathology, Seoul National University College of Medicine, Seoul, Korea
Address reprint requests to James R. Goldenring, M.D., Ph.D., Section of Surgical Sciences, Epithelial Biology Center, Vanderbilt University School of Medicine, 10435G MRB-IV, 2213 Garland Ave, Nashville, TN 37232-2733
Nashville Veterans Affairs Medical Center and the Epithelial Biology Center and the Section of Surgical Sciences, Vanderbilt University School of Medicine, Nashville, Tennessee
Early diagnosis and curative resection are the predominant factors associated with increased survival in patients with gastric cancer. However, most gastric cancer cases are still diagnosed at later stages. Since most pathologic specimens are archived as FFPE samples, the ability to use them to generate expression profiles can greatly improve cancer biomarker discovery. We sought to uncover new biomarkers for stomach preneoplastic metaplasias and neoplastic lesions by generating proteome profiles using FFPE samples. We combined peptide isoelectric focusing and liquid chromatography–tandem mass spectrometry analysis to generate proteomic profiles from FFPE samples of intestinal-type gastric cancer, metaplasia, and normal mucosa. The expression patterns of selected proteins were analyzed by immunostaining first in single tissue sections from normal stomach, metaplasia, and gastric cancer and later in larger tissue array cohorts. We detected 60 proteins up-regulated and 87 proteins down-regulated during the progression from normal mucosa to metaplasia to gastric cancer. Two of the up-regulated proteins, LTF and DMBT1, were validated as specific markers for spasmolytic polypeptide–expressing metaplasia and intestinal metaplasia, respectively. In cancers, significantly lower levels of DMBT1 or LTF correlated with more advanced disease and worse prognosis. Thus, proteomic profiling using FFPE samples has led to the identification of two novel markers for stomach metaplasias and gastric cancer prognosis.
Despite the general advances in endoscopic screening and therapies, gastric cancer 5-year survival rates remain extremely poor,
representing the second leading cause of cancer-related death worldwide. The major proximate cause of gastric cancer is chronic Helicobacter pylori infection, which leads to a chronic inflammatory response and subsequent oxyntic atrophy (loss of acid-secreting parietal cells). In the fundus and corpus of the atrophic stomach, two types of metaplasia have been described: intestinal metaplasia (IM), characterized by the presence of cells with intestinal and goblet cell morphologic features, and spasmolytic polypeptide–expressing metaplasia (SPEM), which shows morphologic characteristics of the deep antral glands and expresses trefoil factor 2 (TFF2), originally designated spasmolytic polypeptide.
and are considered neoplastic precursors, although the mechanisms driving the progression from metaplasia to neoplasia remain unclear. Recent studies in mice have found that SPEM originates from the transdifferentiation of mature chief cells.
Other studies in Mongolian gerbils indicate that after H. pylori–induced parietal cell loss, SPEM is the first metaplastic lesion to evolve, whereas IM develops in the setting of preexisting SPEM.
Gene expression profile of gastric carcinoma: identification of genes and tags potentially involved in invasion, metastasis, and carcinogenesis by serial analysis of gene expression.
However, owing to the high heterogeneity of gastric tumors, no definitive markers have been established. High levels of REG4 were detected in patients with gastric cancer with metastasis, and its expression was correlated with worse prognosis.
Overexpression of RegIV in peritoneal dissemination of gastric cancer and its potential as a novel marker for the detection of peritoneal micrometastasis.
Other studies have noted that REG4 contributes to the resistance of gastric cancer to fluorouracil-based chemotherapy; in addition, patients with gastric cancer showed increased levels of serum REG4. However, despite a specificity of 99%, the diagnostic sensitivity was only 36%.
OLFM4 expression has been associated with intestinal-type gastric cancer. However, different proportions of OLFM4-positive tumors were observed in two independent studies (65%
Inclusion of the less heterogenous metaplastic lesions in the molecular profiling studies could allow a better understanding of the molecular alterations during gastric carcinogenesis and could lead to the development of early-stage gastric cancer biomarkers. Indeed, in a previous study, mRNA expression profiling of IM and SPEM identified several metaplasia and gastric cancer markers, including CDH17 and MUC13, as useful prognostic markers for stage I gastric cancer.
The combined loss of four metaplasia markers (CDH17, REG4, MUC13, and LGALS4) is an independent indicator of survival in undifferentiated or stage II/III gastric cancer.
Formalin-fixed, paraffin-embedded (FFPE) tissue samples are abundantly available in pathology archives, representing a valuable resource for biomarker discovery. Recent developments have facilitated the analysis of proteomic profiles from paraffin-embedded tissues with yields in the range of 90% of the peptides isolated from frozen tissue, showing good equivalence between the protein profiles from frozen and FFPE samples.
Equivalence of protein inventories obtained from formalin-fixed paraffin-embedded and frozen tissue in multidimensional liquid chromatography-tandem mass spectrometry shotgun proteomic analysis.
This innovation means that proteomic analysis can be performed on paraffin-embedded tissue samples with defined pathologic and histologic characteristics, allowing more specific analyses in better-characterized samples.
In the present studies, we performed proteomic profiling using macrodissected FFPE samples from intestinal-type gastric cancer, stomach metaplasia, and normal mucosa. These studies identified a variety of proteins that are up-regulated in metaplasia and cancer. We identified lactotransferrin (LTF) as a novel specific marker for SPEM and deleted in malignant brain tumor 1 (DMBT1) as a marker for IM. In addition, we found that expression of either LTF or DMBT1 influences the survival of patients with gastric cancer.
Materials and Methods
Sample Description and Preparation
We selected 10 paraffin-embedded blocks from gastric resections of intestinal-type tumors in the fundus where both cancer and metaplasia areas were present. Based on H&E staining of an adjacent section, the areas corresponding to metaplasia (predominantly IM) or cancer were macrodissected from 10-μm sections to minimize the stromal component. As a control group, normal fundic mucosa with no sign of atrophic gastritis, IM, SPEM, or gastric cancer was macrodissected from 10 blocks from patients without gastric cancer. Detailed patient information is shown in Supplemental Table S1 (available at http://ajp.amjpathol.org). All the samples were obtained from the Department of Surgery at Seoul National University Hospital (SNUH), Seoul, Korea, between July 1, 2007, and July 1, 2008. This study protocol was approved by the Institutional Review Board at SNUH, with written informed consent provided by all the patients. Use of the de-identified material was also approved by the Vanderbilt University (Nashville, TN) Institutional Review Board.
Sample Preparation and Shotgun Proteomic Analysis
Sample deparaffinization, digestion, and the subsequent steps of isoelectric focusing of peptides and reverse-phase liquid chromatography–tandem mass spectrometry (MS/MS) analysis were performed as previously described with minor modifications.
Equivalence of protein inventories obtained from formalin-fixed paraffin-embedded and frozen tissue in multidimensional liquid chromatography-tandem mass spectrometry shotgun proteomic analysis.
Macrodissected slices were placed in separate centrifuge tubes. Paraffin was removed with three washes in 1 mL of Sub-X clearing medium (Surgipath, Richmond, IL), and rehydration was achieved with three washes each in 1 mL of 100%, 85%, and 70% ethanol. The samples were then digested with trypsin (Promega Corp., Madison, WI) using a trifluoroethanol (Acros, Geel, Belgium) digestion procedure described by Wang et al
with minor modifications. The deparaffinized tissue was resuspended in 100 μL of ammonium bicarbonate (100 mmol/L, pH 8.0), heated to 80°C for 120 minutes, cooled on ice followed by the addition of 100 μL of trifluoroethanol, and sonicated twice for 20 seconds followed by 30 seconds of incubation on ice. The resulting homogenate was incubated at 60°C for 60 minutes, followed by the sonication steps as stated previously herein. The protein concentration was determined using a bicichoninic acid assay (Thermo Scientific Pierce, Rockford, IL). From the resulting homogenate, 50 μg of protein was reduced with 10 mmol/L carboxyethylphosphine (Thermo Scientific Pierce) and 25 mmol/L dithiothreitol (Acros) at 60°C for 30 minutes followed by alkylation with 50 mmol/L iodoacetamide (Sigma-Aldrich, St. Louis, MO) in the dark at room temperature for 20 minutes. The reduced and alkylated homogenate was diluted with 5 volumes of 50 mmol/L ammonium bicarbonate, pH 8.0 (Thermo Scientific Pierce), followed by the addition of trypsin at 1:50 (w/w). The digest was incubated overnight at 37°C, followed by freezing at 80°C and lyophilization. Samples were resuspended in 1 mL of high-pressure liquid chromatography water, desalted with Oasis HLB 96-well μElution plates (30 μm; Waters Corp., Milford, MA), and evaporated in vacuo using a SpeedVac sample concentrator (Thermo Fisher, Waltham, MA).
Isoelectric Focusing of Peptides
Isoelectric focusing of tryptic peptides was adapted from the method of Cargile et al.
Tryptic peptides (from 50 μg of protein) were resuspended in 155 μL of 6 mol/L urea and were loaded in custom-ordered 7-cm (pH 3.5 to 4.5) ZOOM pH strips (Invitrogen, Carlsbad, CA) in a ZOOM cassette and were allowed to rehydrate for 1 hour at room temperature. The loaded strips were focused at 21°C on a ZOOM IPGRunner system (Invitrogen) using the following program: step at 175 V for 15 minutes; gradient to 2000 V over 45 minutes and held at 2000 V for 105 minutes. The strips were then cut into 10 (0.7-cm) pieces and placed in separate wells of a 96-well enzyme-linked immunosorbent assay plate. Peptides were eluted from the strips as follows: 200 μL of 0.1% formic acid for 15 minutes; 200 μL of 50% acetonitrile/0.1% formic acid for 15 minutes; 200 μL of 100% acetonitrile/0.1% formic acid for 15 minutes. Solutions of extracted peptides were evaporated in vacuo, resuspended in 750 μL of 0.1% trifluoroacetic acid, desalted with Oasis HLB 96-well μElution plates (30 μm; Waters Corp.), and evaporated in vacuo using a SpeedVac sample concentrator (Thermo Fisher). Peptide solutions were resuspended in 25 μL of 0.1% formic acid and were placed in sample vials for liquid chromatography–MS/MS analysis.
Liquid chromatography–MS/MS analyses were performed on an LTQ Orbitrap hybrid mass spectrometer (Thermo Fisher Scientific, San Jose, CA) equipped with a nanoLC autosampler system (Eksigent, Dublin, CA). Peptides were resolved on a fused silica capillary column (100 μm × 11 cm; Polymicro Technologies, Phoenix, AZ) packed with Jupiter 5 μm 300 Å C18 (Phenomenex Inc., Torrance, CA) using an inline solid phase extraction column (100 mm × 4 cm) packed with the same C18 resin as that previously described.
Liquid chromatography was performed at room temperature at a flow rate of 0.6 μL/min using a gradient mixture of 0.1% (v/v) formic acid in water (solvent A) and 0.1% (v/v) formic acid in acetonitrile (solvent B). A 95-minute gradient was performed with a 15-minute washing period diverted to waste after the precolumn (100% solvent A for the first 10 minutes followed by a gradient to 98% solvent A at 15 minutes) to allow for solid phase extraction and removal of any residual salts. After the initial washing period, a 60-minute gradient was performed where the first 35 minutes was a slow, linear gradient from 98% solvent A to 75% solvent A, followed by a faster gradient to 10% solvent A at 65 minutes and an isocratic phase at 10% solvent A to 75 minutes.
Centroided MS/MS scans were acquired using the LTQ Orbitrap and an isolation width of 2 m/z, an activation time of 30 milliseconds, an activation q of 0.250, and normalized collision energy of 30% using one microscan with a maximum ion time of 100 milliseconds for each MS/MS scan and one microscan with a maximum ion time of 1000 milliseconds for each full MS scan. The mass spectrometer was tuned before analysis using the synthetic peptide TpepK (AVAGKAGAR) so that some parameters may have varied slightly from experiment to experiment, but typically the tune parameters were as follows: spray voltage of 2 kV, a capillary temperature of 150°C, a capillary voltage of 50 V, and a tube lens voltage of 120 V. The automatic gain control target value was set at 1,000,000 for the full MS and at 10,000 for the MS/MS spectra. A full scan obtained for eluting peptides in the range of 300 to 2000 atomic mass units was collected using the Orbitrap portion of the instrument at a resolution of 60,000, followed by eight data-dependent MS/MS scans (eighth most intense to most intense) using the LTQ portion of the instrument with a minimum threshold of 500 set to trigger the MS/MS spectra. MS/MS spectra were recorded using dynamic exclusion of previously analyzed precursors (±0.6) for 60 seconds with a repeat of 1 and a repeat duration of 1 and an exclusion list size of 50. Monoisotopic precursor selection, nonpeptide monoisotopic recognition, and charge state screening were enabled, with unassigned and single charge states being rejected.
Database Searching and Statistical Analysis of Spectral Counts
Proteins were identified using the MyriMatch algorithm,
Equivalence of protein inventories obtained from formalin-fixed paraffin-embedded and frozen tissue in multidimensional liquid chromatography-tandem mass spectrometry shotgun proteomic analysis.
to assess the possibility of protein changes demonstrating either an increasing or a decreasing trend from normal to metaplasia to cancer. The tool for gene ontology categorization in the WebGestalt toolkit (http://bioinfo.vanderbilt.edu/webgestalt) was used to generate graphics displaying the cell component categories of the proteins differentially expressed in the stomach during the progression from normal mucosa to metaplasia to cancer. In addition to the trend analysis, pairwise comparisons between cancer and normal, cancer and metaplasia, and metaplasia and normal were performed using QuasiTel software version 1.0 (Nashville, TN),
which uses the quasi-likelihood generalized linear modeling.
Immunostaining
Human stomach sections were deparaffinized and were submitted to antigen retrieval in a pressure cooker using the target retrieval solution (Dako North America Inc., Carpinteria, CA). Primary antibody incubation was performed overnight at 4°C. For immunohistochemical analysis, secondary antibody incubation and diaminobenzidine (DAB) development were performed using the Dako EnVision+ system horseradish peroxidase DAB according to the manufacturer's instructions. In the case of immunofluorescence, appropriate secondary antibodies conjugated with Alexa 488, Cy3, or Cy5 were used (1-hour incubation at room temperature). Since the antibodies against LTF, MIST1, and MPO were all made in rabbit, for the co-labeling of LTF/MIST1 or MPO/LTF, we used the TSA (tyramide signal amplification) plus fluorescein or cyanine 3 kits (PerkinElmer, Waltham, MA) to detect the first primary antibody of each pair (used in this case at a dilution 10- to 20-fold higher than in regular staining). After washes with 1X PBS and a second blocking step, we performed the incubation with the second primary antibody of each pair, which was then detected using Alexa 448– or Cy3-conjugated secondary antibodies. In both cases, we included a control slide for which the second primary antibody of each pair was omitted. Detailed information on primary antibodies and their dilutions are listed in Table 1.
For DMBT1 and LTF expression analysis throughout the gastrointestinal tract, we used cDNA samples from the human digestive system multiple tissue cDNA panel (Clontech, Palo Alto, CA). PCR amplifications were performed using the Advantage 2 PCR kit (Clontech), according to the manufacturer's instructions. The reaction was allowed to denature for 4 minutes at 94°C, followed by amplification (35 cycles): 45 seconds at 94°C, 1 minute at 55°C, and 1 minute at 68°C. Amplification of TBP (Tata-box binding protein) cDNA was performed as control for mRNA content. The following primers were used: DMBT1_foward: 5′-GCTTTTCTGCCAGTGACCTTG-3′, DMBT1_reverse: 5′-CTTGATGATGCCACCTGAGAC-3′; LTF_foward: 5′-CCACAGTATGTCGCAGGCATT-3′, LTF_reverse: 5′-CAGTGAATGGCTGAGGCTTTC-3′; and TBP_foward: 5′-GAGCTGTGATGTGAAGTTTCC-3′, TBP_reverse: 5′-TCTGGGTTTGATCATTCTGTAG-3′.
For real-time PCR, total RNA was extracted from frozen or RNAlater-preserved tissue samples using TRIzol (Invitrogen). Normal stomach samples were obtained from Vanderbilt University Hospital from organ donors. Metaplasia samples from gastric resections were obtained at Vanderbilt from the Cooperative Human Tissue Network or from the Department of Surgery at SNUH. cDNA from 1.0 μg of total RNA was prepared using a high-capacity cDNA reverse transcription kit (Applied Biosystems, Foster City, CA), according to the manufacturer's instructions. Equal amounts of each cDNA were analyzed by real-time PCR with specific primers for DMBT1, LTF, and TBP (same as described previously herein) and by Express SYBR GreenER quantitative PCR supermix (Invitrogen) in an ABI StepOne real-time PCR system (Applied Biosystems). Each sample was measured in triplicate.
Protein Expression Analysis on TMAs
The expression profiles of LTF and DMBT1 proteins were analyzed using two TMAs. One is a metaplasia microarray, previously described,
created originally from tissue samples from 33 Japanese patients. The cores that we were able to analyze included areas of SPEM (n = 12), IM (n = 9), foveolar hyperplasia (n = 10), normal fundic mucosa (n = 9), normal antral mucosa (n = 7), and normal duodenum (n = 2). The other TMA is a gastric cancer microarray (SNUH-2004-GC) representing a collection of 450 gastric adenocarcinomas resected at SNUH in 2004. Detailed patient data for this cohort are described elsewhere.
For the metaplasia microarray, DMBT1 staining was visualized by DAB development, whereas LTF was detected by immunofluorescence in a double staining with TFF2. For the gastric cancer microarray, both proteins were detected by immunofluorescence. In both cases, an antibody or an antibody mixture was included for keratin detection. The TMAs were imaged using an Ariol SL-50 platform (Genetix, San Jose, CA) with a 20X objective. The images from the metaplasia microarray were manually inspected and selected. For the gastric cancer array, the percentage of epithelial cells (keratin staining) also positive for DMBT1 or LTF in each sample was determined using a pipeline constructed in the CellProfiler cell image analysis software release 10997 (Cambridge, MA).
Mean protein level differences were analyzed by analysis of variance; prognostic indicators included stage, Lauren grade, metastasis, and histopathologic features. Pairwise differences were examined for all statistically significant prognostic indicators, and a Tukey adjustment for multiple comparisons was applied. We used Cox proportional hazards models to evaluate the difference in survival experience among the staining quartiles for DMBT1 and LTF; pairwise differences were tested using a Wald test statistic.
Results
Proteomic Profiling of Stomach Metaplasia and Gastric Cancer
We performed shotgun proteomic analysis on 10 macrodissected FFPE tissue samples collected from each of three groups: i) normal fundic mucosa, ii) metaplasia in the fundus, and iii) intestinal-type gastric adenocarcinoma. Since control samples had a more limited amount of tissue, we scaled down the isoelectric focusing preparation from a typical 200-μg input to analyze only 50 μg of protein.
Equivalence of protein inventories obtained from formalin-fixed paraffin-embedded and frozen tissue in multidimensional liquid chromatography-tandem mass spectrometry shotgun proteomic analysis.
This analysis yielded approximately 2350 protein groups per sample (see Supplemental Table S2 at http://ajp.amjpathol.org). The overlap of proteins identified in the different tissue groups was >99%, although the levels of proteins differed as assessed by spectral counts, which correlate quantitatively with protein levels.
According to the Jonckheere Terpstra trend test (P < 0.01), 60 proteins were found to be up-regulated and 87 proteins were found to be down-regulated across the progression from normal mucosa to gastric cancer (see Supplemental Tables S3 and S4 at http://ajp.amjpathol.org). The cohort of up-regulated proteins was predominantly represented by nuclear and membrane proteins, whereas, in accordance with the parietal cell loss observed during gastric metaplasia and cancer development, mitochondrial proteins were overrepresented in the group with decreased expression (see Supplemental Figure S2 at http://ajp.amjpathol.org). Box plots displaying the levels of the top 20 statistically significant up-regulated proteins from the trend analysis are shown in Figure 1. Reflecting the progressive alterations from normal mucosa to metaplasia and then to cancer, in pairwise comparisons (see Supplemental Tables S5 and S6 at http://ajp.amjpathol.org), we observed (P < 0.01 and fold change >1.5) 18 proteins up-regulated in metaplasia and 46 up-regulated in cancer, both compared with the normal mucosa. Nine of the proteins up-regulated in metaplasia were also among those up-regulated in cancer.
Figure 1Top 20 proteins up-regulated in the stomach across the progression from normal to metaplasia to cancer. For each of the top 20 proteins considered up-regulated using the Jonckheere-Terpstra trend test (P < 0.01), we show box plots representing the distribution of protein levels (normalized ratios of spectral counts) in each of the sample groups. The box plots show interquartile ranges and medians and whiskers indicate data value ranges excluding outliers. Each red circle represents a individual sample value. The number of samples in each group [normal, metaplasia (Meta), and cancer] was 10.
We examined the expression patterns of nine (NASP, DDX5, TNXB, NONO, RUVBL2, RCC1, LTF, DMBT1, and ESYT) of the top 20 up-regulated proteins according to the trend analysis in tissue sections from human normal mucosa, metaplasia, and gastric cancer samples. The first eight proteins were also in the cohort of those up-regulated in cancer, as indicated by the pairwise analysis. Two proteins, LTF and DMBT1, showed very specific staining for metaplastic or neoplastic lesions and presented low expression levels or no detection in the normal mucosa (Figure 2). All the other proteins analyzed by immunohistochemical analysis were detected in metaplasia and cancer but were also present in the normal mucosa (see Supplemental Figure S3 at http://ajp.amjpathol.org). LTF antibody labeled SPEM specifically but not IM (Figure 2A). The presence of LTF protein in SPEM lesions was confirmed using co-labeling with TFF2, whose ectopic expression toward the base of the oxyntic mucosa is a marker of SPEM (Figure 2B). In the normal mucosa, low to moderate expression of LTF was detected in scattered cells toward the bottom of the fundic glands. Those cells represented a subpopulation of mature chief cells because they immunostained for the transcription factor MIST1 (Figure 2C), a specific marker of mature chief cells,
and they did not express TFF2 (see Supplemental Figure S4 at http://ajp.amjpathol.org). LTF is a component of the secondary granules of neutrophils and, consequently, in addition to the epithelial staining, strong labeling was observed in scattered infiltrating neutrophils (confirmed by co-staining for myeloperoxidase) in the normal mucosa (see Supplemental Figure S5 at http://ajp.amjpathol.org).
Figure 2Expression of DMBT1 and LTF in metaplasia and cancer. A: Immunohistochemical analysis (DAB) using specific antibodies against DMBT1 (right panels) and LTF (left panels) in sections from normal stomach mucosa, stomach with metaplastic lesions, and gastric adenocarcinoma. Scale bars: 50 μm. B: To confirm the LTF staining in SPEM areas, we performed immunofluorescence co-labeling using the antibody against LTF (green; detected using an Alexa 488–conjugated secondary antibody) and an antibody against TFF2 (red; detected using a Cy3-conjugated secondary antibody). TFF2 staining at the base of fundic mucosa glands is a marker of SPEM. Scale bar = 50 μm. C: Immunofluorescence co-labeling of LTF (red; detected using the TSA plus cyanine 3 kit) and MIST1 (green; detected using an Alexa 488–conjugated secondary antibody) in normal fundic mucosa. D: DMBT1 and LTF mRNA expression in the gastrointestinal tract. Specific primers for DMBT1 or LTF messages were used for PCR amplifications in a panel of cDNAs from tissues throughout the gastrointestinal tract: esophagus (E), stomach (S), duodenum (D), jejunum (J), ileum (I), ileocecum (IC), cecum (C), ascending colon (Ca), transverse colon (Ct), descending colon (Cd), and rectum (R). Amplification of a TBP fragment was used as loading control. m, molecular standard 100 bp DNA ladder, showing bands from 100-500 bp; nc, negative or non-template control. E: Quantitative real-time PCR of DMBT1 and LTF in samples from normal stomach and stomach with metaplastic lesions. Expression levels are expressed as 2ΔCt, using TBP (endogenous control) levels for normalization. Statistical analysis using a Mann-Whitney test indicated that DMBT1 expression was significantly higher (*P < 0.05) in metaplasia than in normal mucosa, whereas the difference for LTF levels between normal mucosa and metaplasia was not significant (NS). Data are given as mean ± SD.
DMBT1 was not detected in the normal mucosa but showed strong expression in IM and cancer. The absence of DMBT1 expression in the normal stomach was confirmed by RT-PCR using a panel of normal human tissues from the gastrointestinal tract (Figure 2D). DMBT1 message was detected in all other regions of the gastrointestinal tract but not in the stomach. LTF message, however, was detected throughout the entire gastrointestinal tract, including the stomach (Figure 2D). In a previous mRNA profiling study, DMBT1 was found as one of the top 10 up-regulated genes in IM; however, no significant increase in LTF message was detected.
Using quantitative real-time PCR, we confirmed a dramatic up-regulation of DMBT1 message in stomach metaplasia (Figure 2E). For LTF message levels, however, the difference between the normal and metaplasia groups was not significant.
LTF and DMBT1 Are Early Markers for Different Metaplastic Lineages
We extended the analysis of LTF and DMBT1 expression to characterize their association with the metaplastic lesions. We used a small-scale tissue array (metaplasia array) containing fundic mucosa sections displaying SPEM and IM.
DMBT1 and LTF were detected as specific markers for IM and SPEM, respectively (Table 2 and Figure 3). In addition, we observed that both proteins were expressed early in the metaplasia progression. A small population of cells expressing DMBT1 was observed in all the sections with SPEM (12 of 12) and in most of the hyperplasia samples (9 of 10). These DMBT1-positive cells were present luminal to regions with basally located SPEM glands, in the upper third of the glands in areas with no evident IM (Figure 3A). Also, LTF expression was observed in all the hyperplasia samples, which already showed some evidence of SPEM development based on the extension of TFF2 expression toward the gland bottoms (Figure 3A). Strong LTF staining was observed even in glands without TFF2-expressing cells (Figure 3A), indicating that LTF expression may precede up-regulation of TFF2 in SPEM. We also noted that several of the glands with a few cells expressing DMBT1 (but no evident IM) also contained cells with low levels of LTF, usually in different cells (Figure 3, B and D). Co-staining with the two SPEM markers (LTF and TFF2) and DMBT1 showed that most of the cells expressing low levels of DMBT1 were TFF2 positive. We even observed rare cells expressing all three of the markers (see Supplemental Figure S6 at http://ajp.amjpathol.org). Also, in sections stained for DMBT1, TFF2, and MUC2, in addition to the cells expressing both TFF2 and DMBT1, we found cells expressing TFF2 and low levels of the two IM markers DMBT1 and MUC2 (Figure 4), suggesting that those hybrid glands represent a transitional stage between SPEM and IM. Using a recently defined SPEM marker, Clusterin,
in co-staining with DMBT1 and Ki-67, we observed that although cells in the glands expressing only the SPEM marker were mostly negative for Ki-67, cells in a transitional zone that express low levels of Clusterin and DMBT1 are proliferative (Figure 5).
Table 2DMBT1 and LTF Protein Expression in the Metaplasia Tissue Array
Figure 3DMBT1 and LTF are early markers of different metaplastic lineages. A TMA containing sections from normal stomach mucosa and cores with foveolar hyperplasia, SPEM, or IM were immunostained using specific antibodies against either DMBT1 (DAB) or LTF (green) with TFF2 (red). Immunofluorescence primary antibody detection was performed as described for Figure 2B. A: Area of foveolar hyperplasia with no morphologic evidence of IM but with evidence of initial SPEM (expansion of TFF2 toward the base). Note that although several glands at the base are not positive for TFF2, most of them already show LTF expression. B: Higher magnification of the area framed in A showing the presence of cells with low expression of LTF and DMBT1 in the same gland. C: Section displaying areas of IM with strong DMBT1 staining and areas of SPEM with strong LTF staining. D: Higher magnification of the area framed in C. Scale bars: 50 μm.
Figure 4Cells expressing the SPEM marker TFF2 and low levels of the intestinal markers DMBT1 and MUC2 support the concept that SPEM progresses to IM. A tissue section from human stomach with areas of SPEM and IM was immunostained using specific antibodies against DMBT1 (green; detected with an Alexa 488–conjugated secondary antibody), TFF2 (red; detected with a Cy3-conjugated secondary antibody), and MUC2 (pseudocolored to blue; detected with a Cy5-conjugated secondary antibody). DAPI staining was pseudocolored to gray. The green arrow illustrates the cells co-expressing TFF2 and DMBT1 (shown at higher magnification in the bottom right panel). The pink arrow illustrates a transitional gland with cells expressing the SPEM marker TFF2 and low levels of the intestinal markers DMBT1 and MUC2 (shown at higher magnification in the bottom left panel). Scale bars: 100 μm (top panel); 25 μm (bottom panel).
Figure 5Cells expressing the SPEM marker Clusterin and the IM marker DMBT1 are proliferative, in contrast to cells expressing the SPEM marker only. A tissue section from human fundic stomach displaying areas of SPEM and IM was immunostained using specific antibodies against Clusterin (green; detected with an Alexa 488–conjugated secondary antibody), Ki-67 (red; detected with a Cy3-conjugated secondary antibody), and DMBT1 (pseudocolored to blue; detected with a Cy5-conjugated secondary antibody). Higher magnifications of the areas depicted in the white squares in the top panel are shown in the bottom panels. Scale bars: 100 μm (top panels); 25 μm (bottom panels).
We next sought to investigate the expression pattern of LTF and DMBT1 proteins in a large number of gastric cancer samples. We immunostained a tissue array set (SNUH-2004-GC) containing 450 gastric cancer samples, most representing early-stage cases (45% at stage I, 20% at stage II, 18% at stage III, and 17% at stage IV). Using a cutoff value of 10% of positive cells (<10% was considered absent or very low expression), DMBT1 was detected in 71.7% of the cancer samples, whereas LTF was expressed in 44.1% of the cases. There was no significant correlation of LTF or DMBT1 expression with Lauren grade, World Health Organization grade, or the presence of metastasis. Although the expression levels (percentage of positive cells) of either DMBT1 or LTF were similar between intestinal- and diffuse-type cancers for both proteins, we observed two distinct staining patterns that were basically associated with the different histologic types. Representative staining for both proteins in intestinal- and diffuse-type cancers is shown in Figure 6A. Overall, DMBT1 showed cytoplasmic punctate staining in 55.3% of the positive samples and a predominant apical staining in 42.1% (2.6% of the positive samples did not show a clear pattern). The cytoplasmic punctate pattern was more frequently found in diffuse-type cancers (62.5% in diffuse type, 22.0% in intestinal type, and 15.5% in mixed type), whereas the apical staining was predominantly observed in intestinal-type cancers (72.0% in intestinal type, 3.8% in diffuse type, and 24.2% in mixed type). For LTF, the most common pattern (67.0%) was a moderate-intensity cytoplasmic punctate staining (Figure 6A). Diffuse-type cancers were predominant in the group displaying this pattern (54.3% of diffuse type, 26.1% of intestinal type, and 19.6% of mixed type). The other pattern (33.0% of the positive samples) was a very strong and diffuse cytoplasmic staining, which was more frequently observed for intestinal-type tumors (60.9% of intestinal type, 26.1% of diffuse type, and 13.0% of mixed type). According to the statistical analysis, the only clinicopathologic factor that correlated with DMBT1 or LTF expression was disease stage (Figure 6B). For both proteins, the percentage of positive cells was significantly higher in stage I cases. For survival analysis, the expression levels of the proteins (percentage of positive cells) were divided into quartiles. For either DMBT1 or LTF, the lowest expression level correlated with significantly poorer prognosis (Figure 6C). We did not detect any significant predictive benefit from the analysis of combined LTF and DMBT1 expression.
Figure 6DMBT1 and LTF expression in gastric cancer. DMBT1 and LTF expression (green; detected with an Alexa 488–conjugated secondary antibody) was evaluated in a TMA containing 450 gastric cancer samples. In both cases, the TMA was co-stained with an anti–pan-cytokeratin antibody (red; detected with a Cy3-conjugated secondary antibody) as a marker for epithelial cells. A: Representative pictures of DMBT1 and LTF staining in intestinal and diffuse types of gastric cancer. The LTF/keratin staining, (bottom left panel) shows the single channel image of keratin staining, included to show that the cells (indicated by the arrows) displaying strong LTF staining were keratin positive. Scale bar = 50 μm. B: DMBT1 and LTF expression inversely correlates with tumor stage. The percentage of DMBT1- and LTF-positive cells is significantly lower in stages II, III, and IV than in stage I. Data are given as mean ± SD. C: Survival impact of DMBT1 or LTF expression. Protein expression levels in patients, represented by percentage of positive cells, were divided into patient quartiles that were used to construct the survival curves. The log-rank test P value for DMBT1 was 0.0076 and for LTF was 0.0072. For pairwise analysis, the statistical models applied showed statistically significant differences between the first and second and first and third quartiles for DMBT1 (P = 0.015 and 0.002, respectively) and between the first and third and first and fourth quartiles for LTF (P = 0.012 and 0.003, respectively); all other pairwise differences were nonsignificant at the 0.05 level of significance. Overall, patients with the lowest expression of either LTF or DMBT1 (quartile 1) showed poorer survival.
As seen in the metaplasia array, we observed an inverse relationship for expression of DMBT1 and LTF in metaplastic glands, with an overlap in units that seemed to represent transitional glands in several metaplastic areas associated with the tumors in the SNUH-2004-GC tissue array (Figure 7A). Cells expressing DMBT1 and cells expressing LTF were observed in single glands, but they were usually in different domains, although in some cases we detected a very low expression level of LTF in cells expressing DMBT1.
Figure 7DMBT1 and LTF show an opposite expression gradient in the stomach during progression of metaplasia. A: Representative images of a metaplastic area adjacent to a gastric adenocarcinoma showing DMBT1 and LTF expression in different regions of the same gland displaying an opposite gradient of expression. LTF and DMBT1 were detected in green and keratin in red. B: Model of DMBT1 and LTF expression during stomach metaplasia and neoplastic progression.
Equivalence of protein inventories obtained from formalin-fixed paraffin-embedded and frozen tissue in multidimensional liquid chromatography-tandem mass spectrometry shotgun proteomic analysis.
have great potential to increase biomarker discovery. We used this approach herein for stomach tissues, generating consistent proteome profiles from normal mucosa, stomach metaplasia, and cancer using FFPE samples. The present data support the use of this method even when smaller amounts of total protein are available (50 μg instead of the 200 μg used in the previous report
Equivalence of protein inventories obtained from formalin-fixed paraffin-embedded and frozen tissue in multidimensional liquid chromatography-tandem mass spectrometry shotgun proteomic analysis.
). Loss of parietal cells induced by H. pylori infection is a prerequisite event in the development of metaplastic and neoplastic lesions in the stomach fundic mucosa. The fact that mitochondrial proteins (especially abundant in parietal cells) were the major class of down-regulated proteins in metaplasia and cancer indicates that the profiles we generated reflect the biological changes characteristic of the atrophic, metaplastic, and neoplastic stomach mucosa. Further supporting this conclusion, there were a higher number of up-regulated proteins in cancer than in metaplasia compared with normal.
IM is a well-established preneoplastic lesion for intestinal-type gastric cancer,
the repertory of biomarkers for SPEM diagnosis is still limited. We identified LTF as a new SPEM marker that may contribute to the early precise detection of SPEM in human stomach. LTF differs from other SPEM markers, including TFF2, HE4, and Clusterin, because it is expressed in a subpopulation of chief cells in the normal mucosa. This finding supports the concept that SPEM in humans derives, at least in part, from mature chief cells, as has been demonstrated in the mouse.
The presence of IM in the stomach is usually evident even in routine H&E-stained sections. Nevertheless, since DMBT1 expression seems to precede the appearance of the characteristic IM morphologic features, its expression analysis may help the diagnosis of transitions toward IM at an early stage. Sugano and colleagues
recently demonstrated that metaplasia markers (some intestinal markers and TFF2) could be detected very early during metaplasia development, preceding the morphologic changes characteristic of each lesion. In addition to the potential use of DMBT1 and LTF as early metaplasia markers, their expression pattern in gastric cancer makes them new candidate markers for predicting patient outcome.
LTF (lactoferrin or lactotransferrin) is an 80-kDa iron-binding glycoprotein expressed mainly in glandular epithelial cells and in the developing neutrophils, where it is stored in the secondary granules.
The biological functions of LTF go far beyond just iron homeostasis and include host defense, anti-inflammatory activity, cell growth and differentiation, and cancer development. LTF can activate gene expression via induction of intracellular signaling pathways, including NF-κB.
Although a recent study showed that LTF is responsible for the loss or reduction of ER-α, PR, and HER-2 in triple-negative breast cancer and contributes to increased invasive behavior of breast cancer cells,
The present data, showing that most cancers demonstrate low expression levels of LTF and that the lowest expression levels correlate with the poorest outcome, support the concept of a protective role of LTF against gastric cancer.
DMBT1 is a 340-kDa protein with 14 repeats of the scavenger receptor cysteine-rich domain, two C1r/C1s Uegf Bmp1 domains, and a carboxy-terminal zona pellucida domain. Other names for the DMBT1 gene products include salivary agglutinin, gp-340, CRP-ductin (mouse), muclin (mouse), and hensin (rabbit). It is a complex molecule that has been implicated mainly in innate immune defense and epithelial cell differentiation.
a process considered as an example of transdifferentiation. DMBT1 is predominantly expressed in epithelial cells, and in humans, the highest levels of DMBT1 expression were observed in trachea, lungs, and the gastrointestinal tract. Previous studies reported the presence of DMBT1 message or protein in normal human stomach.
Induction of DMBT1 expression by reduced ERK activity during a gastric mucosa differentiation-like process and its association with human gastric cancer.
In the present study, we confirmed abundant expression of DMBT1 message throughout the gastrointestinal tract but not in the normal stomach corpus. The use of whole stomach samples containing the antral region in those previous studies could explain this difference in results since we did observe DMBT1 protein expression in the antrum, in a subpopulation of cells in the isthmus/neck region. Down-regulation of DMBT1 expression has been described in breast,
Induction of DMBT1 expression by reduced ERK activity during a gastric mucosa differentiation-like process and its association with human gastric cancer.
Using a large-scale sample cohort, we confirmed DMBT1 up-regulation in gastric tumors compared with in normal stomach, and, in addition, we showed that the lowest expression level of DMBT1 was associated with poor patient survival.
Note that the proteins we detected as early markers for SPEM and IM (LTF and DMBT1, respectively) are both connected to the inflammatory response and cell differentiation. Inflammation is a key factor for metaplasia-neoplasia progression in the stomach. Studies in mouse models have revealed that although loss of parietal cells alone is sufficient to induce the appearance of SPEM, the progression to a more proliferative metaplasia and subsequent dysplasia requires the presence of inflammation.
Expression of LTF and DMBT1 as an initial response to infection and inflammation may lead to the induction of altered differentiation pathways in the cells of the gastric mucosa, contributing to the establishment of metaplastic lesions.
The expression pattern of LTF and of other SPEM markers and DMBT1 also supports the concept that human SPEM evolves into IM. Cells expressing both the SPEM marker Clusterin and DMBT1 are proliferative, in contrast to cells expressing Clusterin only. Those cells possibly represent the precursors of the IM lesions that are marked by high levels of DMBT1 expression and increased proliferation. Studies in animal models indicate that after parietal cell loss, SPEM is the first metaplastic lesion to evolve, whereas IM will develop from SPEM.
demonstrated that SPEM associated with long-term Helicobacter felis infection (>1 year) is associated with the emergence of intestinal markers. Thus, mouse SPEM cells in a more advanced metaplastic stage acquire a more intestinal-like expression profile. We recently found similar intestinalizing transcripts in SPEM lineages from 12-month H. felis–infected mice, including DMBT1.
In humans, the metaplasia-neoplasia cascade (Figure 7B) would be marked by early expression (or up-regulation) of LTF in SPEM (and possible pre-SPEM cells) that decreases when the cells progress to IM and start to express DMBT1. Consistent with previous studies showing that loss of metaplasia markers correlates with worse prognosis in patients with gastric cancer,
we showed herein that during cancer progression, the levels of both proteins decrease, and this has a negative effect on patient survival.
In summary, we demonstrated the utility of proteomic profiling of FFPE gastric tissues as an effective approach to identify relevant biomarkers. These investigations identified LTF and DMBT1 as strong markers of SPEM and IM, respectively. The patterns of LTF and DMBT1 expression support the concept that SPEM gives rise to IM. In addition, we showed that loss of LTF or DMBT1 expression in gastric tumors correlates with a poor prognosis.
Acknowledgments
We thank Dr. Joseph Roland for his assistance with the TMA imaging and for the use of his stomach gland illustrations in Figure 7 and Drs. Joseph Roland and Jared Weis for their assistance with the construction of a CellProfiler pipeline for image analysis.
Representative images of a TMA core and the output file after CellProfiler detection of epithelial cells positive for DMBT1 staining. The left panel is the TMA core original image with cytokeratin in red and DMBT1 in green. The right panel is the CellProfiler output image, indicating the cells considered positive for cytokeratin and DMBT1 staining.
Cell component categories of the proteins differentially expressed in the stomach during the progression from normal mucosa to metaplasia to cancer. The analysis was performed using the tool for gene ontology categorization in the WebGestalt toolkit (http://bioinfo.vanderbilt.edu/webgestalt).
Expression pattern of 7 of the top 20 up-regulated proteins in the progression from normal mucosa to metaplasia to cancer. Immunohistochemical analysis using specific in sections from normal stomach mucosa, stomach with metaplastic lesions, antibodies, as indicated, and gastric adenocarcinoma. Scale bars: 50 μm.
LTF-positive cells in the normal mucosa do not express TFF2. A tissue section from a normal human stomach was immunostained using specific antibodies against LTF (green; detected with an Alexa 488–conjugated secondary antibody) and TFF2 (red; detected with a Cy3-conjugated secondary antibody). Scale bars: 100 µm (left panel); 25 μm (right panel).
LTF also marks infiltrating neutrophils in the stomach mucosa. A tissue section from human normal stomach was immunostained using specific antibodies against myeloperoxidase (MPO; green; detected using the TSA plus fluorescein kit) and LTF (red; detected with a Cy3-conjugated secondary antibody). Scale bar = 100 μm.
Cells in a metaplasia transitional zone express both SPEM markers (TFF2 and LTF) and the early intestinal marker DMBT1. A tissue section from human stomach with areas of SPEM and IM was immunostained using specific antibodies against LTF (green; detected with an Alexa 488–conjugated secondary antibody), TFF2 (red; detected with a Cy3-conjugated secondary antibody), and DMBT1 (pseudocolored to blue; detected with a Cy5-conjugated secondary antibody) Arrows in the left panel indicate three transitional glands (expressing SPEM and IM markers) that are shown in more detail in the right panel. The arrow in the left panel indicates a single cell expressing the two SPEM markers (LTF and TFF2) and also the IM markers DMBT1.
Gene expression profile of gastric carcinoma: identification of genes and tags potentially involved in invasion, metastasis, and carcinogenesis by serial analysis of gene expression.
Overexpression of RegIV in peritoneal dissemination of gastric cancer and its potential as a novel marker for the detection of peritoneal micrometastasis.
Equivalence of protein inventories obtained from formalin-fixed paraffin-embedded and frozen tissue in multidimensional liquid chromatography-tandem mass spectrometry shotgun proteomic analysis.
Induction of DMBT1 expression by reduced ERK activity during a gastric mucosa differentiation-like process and its association with human gastric cancer.
These studies were supported by NIH grant RO1 DK071590 (J.R.G.), American Recovery & Reinvestment Act of 2009 Supplemental Funding grant RO1 DK071590-S1 (J.R.G.), and Core Resources supported by the Vanderbilt Digestive Disease Center grant P30 DK058404, and the Vanderbilt-Ingram Cancer Center through NCI Cancer Center support grant P30 CA068485 using the Pathology Shared Resource. These studies also used Ariol SL-50 imaging in the Epithelial Biology Center Shared Imaging Resource.
Early diagnosis and curative resection are associated with increased survival in gastric cancer patients. Yet, a majority of gastric cancers are still diagnosed at later stages. Sousa et al (Am J Pathol 2012, 181:1560−1572) uncovered novel biomarkers LTF and DNMT1 for spasmolytic polypeptide-expressing metaplasia and intestinal metaplasia, respectively, by generating proteome profiles using formalin-fixed, paraffin-embedded samples of interstitial-type gastric cancer, metaplasia, and normal mucosa by combining peptide isoelectric focusing and liquid chromatography-mass spectrometry/mass-spectrometry analysis.
00597-4&id=gr1.jpg){kind=link}
00597-4&id=gr2.jpg){kind=link}
00597-4&id=gr3.jpg){kind=link}
00597-4&id=gr4.jpg){kind=link}
00597-4&id=gr5.jpg){kind=link}
00597-4&id=gr6.jpg){kind=link}
00597-4&id=gr7.jpg){kind=link}