- Email: [email protected]
The Clonal Origins of Dysplasia From Intestinal Metaplasia in the Human Stomach
GASTROENTEROLOGY 2011;140:1251–1260
The Clonal Origins of Dysplasia From Intestinal Metaplasia in the Human Stomach LYDIA GUTIERREZ–GONZALEZ,*,‡ TREVOR A. GRAHAM,* MANUEL RODRIGUEZ–JUSTO,§ SIMON J. LEEDHAM,* MARCO R. NOVELLI,§ LAURA J. GAY,储 TANIA VENTAYOL–GARCIA,储 ALICIA GREEN,储 IAN MITCHELL,¶ DAVID L. STOKER,¶ SEAN L. PRESTON,# SHIGEKI BAMBA,** EIJI YAMADA,‡‡ YUUKI KISHI,§§ REBECCA HARRISON,储 储 JANUSZ A. JANKOWSKI,储 NICHOLAS A. WRIGHT,*,储 and STUART A. C. McDONALD*,储 *Histopathology Laboratory, London Research Institute, Cancer Research UK, London, United Kingdom; ‡Health Sciences Institute of Aragon, CIBERehd, Zaragoza, Spain; §Department of Histopathology, University College London, London; 储Centre for Digestive Diseases, Blizard Institute of Cell and Molecular Sciences, Barts and the London Medical School, Queen Mary, University of London, London; ¶Department of Surgery, University College London, London; #Digestive Diseases Clinical Academic Unit, Barts and the London NHS Trust, London, United Kingdom; **Department of Gastroenterology, Shiga University of Medical Science Seta-Tsukinowa, Ohtsu, Shiga, Japan; ‡‡Department of Pathology and §§Department of Internal Medicine, Hikone Muncipal Hospital, Hikone, Japan; 储 储Department of Pathology, University Hospitals Leicester, Leicester, United Kingdom
Keywords: Stomach Cancer; Tumor Development; Neoplasia; Oncogenesis. View this article’s video abstract at www.gastrojournal.org
G
astric adenocarcinoma (GA) is a major cause of cancer-related mortality worldwide (http://info .cancerresearchuk.org/cancerstats; Cancer Research UK, 2010, London, UK, viewed 6th March 2011). There are a multitude of risk factors: within these, infection with Helicobacter pylori and the development of intestinal metaplasia (IM) are prominent.1,2 IM of the human stomach is the replacement of the normal gastric glands with an intestinal crypt phenotype. Current dogma proposes that there is a stepwise progression from chronic atrophic gastritis to intestinal metaplasia, with subsequent progression to low-/high-grade dysplasia, eventually culminating in the intestinal type of GA.3,4 Using whole tissue specimens, previous studies have suggested that metaplasia can contain similar genetic abnormalities to the associated GA,3,5–7 suggesting that IM is a premalignant condition from which a carcinoma may develop. Whereas some studies demonstrate a genetic link,7 some reported abnormalities were also present in adjacent normal tissue and were therefore most probably polymorphisms.3 To date, a clear genetic link, on a gland-by-gland basis, describing the evolution of metaplasia into dysplasia has not been produced. The clonal nature of metaplastic glands in the human stomach is a source of some controversy. We have previously shown that human IM glands from patients undergoing resection for GA can be entirely deficient in cytochrome c oxidase (CCO).8 This deficiency has been shown to be highly suggestive of an underlying mutation in the mitochondrial genome resulting in CCO deficiency; therefore, IM glands from the human stomach appear clonal. Other groups have suggested otherwise; Abbreviations used in this paper: CCO, cytochrome c oxidase; GA, gastric adenocarcinoma; IM, intestinal metaplasia; LOH, loss of heterozygosity; mtDNA, mitochondrial DNA; SPEM, spasmolytic polypeptide-expressing metaplasia; TFF, trefoil factor. © 2011 by the AGA Institute 0016-5085/$36.00 doi:10.1053/j.gastro.2010.12.051
BASIC– ALIMENTARY TRACT
BACKGROUND & AIMS: Studies of the clonal architecture of gastric glands with intestinal metaplasia are important in our understanding of the progression from metaplasia to dysplasia. It is not clear if dysplasias are derived from intestinal metaplasia or how dysplasias expand. We investigated whether cells within a metaplastic gland share a common origin, whether glands clonally expand by fission, and determine if such metaplastic glands are genetically related to the associated dysplasia. We also examined the clonal architecture of entire dysplastic lesions and the genetic changes associated with progression within dysplasia. METHODS: Cytochrome c oxidase-deficient (CCO⫺) metaplastic glands were identified using a dual enzyme histochemical assay. Clonality was assessed by laser capture of multiple cells throughout CCO⫺ glands and polymerase chain reaction sequencing of the entire mitochondrial DNA (mtDNA) genome. Nuclear DNA abnormalities in individual glands were identified by laser capture microdissection polymerase chain reaction sequencing for mutation hot spots and microsatellite loss of heterozygosity analysis. RESULTS: Metaplastic glands were derived from the same clone—all lineages shared a common mtDNA mutation. Mutated glands were found in patches that had developed through gland fission. Metaplastic and dysplastic glands can be genetically related, indicating the clonal origin of dysplasia from metaplasia. Entire dysplastic fields contained a founder mutation from which multiple, distinct subclones developed. CONCLUSIONS: There is evidence for a distinct clonal evolution from metaplasia to dysplasia in the human stomach. By field cancerization, a single clone can expand to form an entire dysplastic lesion. Over time, this field appears to become genetically diverse, indicating that gastric cancer can arise from a subclone of the founder mutation.
1252
GUTIERREZ–GONZALEZ ET AL
BASIC– ALIMENTARY TRACT
Mihara et al9 demonstrated that, within single IM glands, patterns of CpG methylation in the promoter regions of the ZIK1 gene, which is normally silenced in the human stomach, are highly polymorphic, suggesting that metaplastic glands develop multifocally, arising from multiple stem cells. On the other hand, methylation patterns are known to change over time, and the variability observed by Mihara et al9 may reflect rapid diversification rather than stable clones. To understand the progression and expansion of intestinal metaplasia and its relationship to dysplasia, the stem cell and clonal architecture of IM become important. Early gastric intestinal type adenocarcinomas are usually associated with fields of dysplastic glands.10 Again, it is unknown how these fields develop and spread through the stomach, namely whether by multiple dysplasia-initiating events occurring independently or through extensive clonal expansion of a single dysplastic gland through a large area of gastric mucosa. There is precedent for both scenarios in the human gastrointestinal tract: in patients with ulcerative colitis-associated neoplasias, a single mutated crypt is able to spread through large areas of colonic mucosa encompassing the entire lesion,11 a socalled founder mutation. On the other hand, Barrett’s esophagus appears to be a genetically heterogeneous disease where there are multiple clones within the Barrett’s mucosa.12 Even leaving aside the intrinsic interest of the true nature of the development of intestinal-type gastric carcinoma, the lack of such knowledge severely impairs our ability to identify reliable biomarkers to assess the risk of gastric carcinogenesis. Here, we show, by using either cell-by-cell or gland-bygland laser capture microdissection of human gastric epithelium that intestinal metaplastic glands are clonal, the intestinal metaplasia clonally expands by crypt fission to form large patches, and that dysplasia can arise from a single clone of mutated intestinal metaplastic glands and expand to form the entire dysplastic lesion. Furthermore, we show that, within this clonal dysplastic field, genetic diversity can arise, raising the possibility that cancer develops through genetic progression of a subclone arising from the founder mutation.
Patients and Methods Patients Intestinal metaplastic and dysplastic formalinfixed, paraffin-embedded and frozen specimens were obtained from patients undergoing resection for either GA or high-grade dysplasia and from pathology archives. A total of 23 patient specimens (age range, 57–72 years) were used in this study. Ethical approval was sought and obtained as per United Kingdom Human Tissue Act (2006, reference 07/Q1604/17) and Hikone Municipal Hospital (Japan) regulations. Histology of each specimen was confirmed by 2 pathologists (N.A.W. and M.R.J.). All
GASTROENTEROLOGY Vol. 140, No. 4
specimens showed IM surrounding or in between large areas of dysplasia (or early gastric cancer) and was consistent through multiple blocks from each patient.
Laser Capture Microdissection Ten serial sections (5–20 m) of each specimen were prepared. Sections 1–5 were cut onto normal glass slides, and sections 6 –10 were cut onto P.A.L.M. MembraneSlide 1.0 PEN slides (Zeiss Microimaging, Munich, Germany). Paraffin sections were left overnight at 37°C, dewaxed in xylene, rehydrated through decreasing ethanol concentrations to water, and then air-dried. Frozen sections were allowed to air-dry immediately after cutting. Each membrane section underwent light staining with methylene green (2%; Sigma, Poole, UK). Paraffin sections 1–3 were used to screen for genomic mutations. Sections 4 and 5 were used for H&E or alcian blue/ periodic acid—Schiff/diastase staining. On sections 6 –10, areas of interest were delineated using P.A.L.M. Robosoftware (Zeiss Microimaging) and cut into 0.5-mL AdhesiveCap tubes using a P.A.L.M. Laser capture microdissection system (Zeiss Microimaging). The majority of the frozen sections was used for enzyme histochemistry with an occasional section (approximately 1 per 20 sections cut) used for H&E. Captured specimens were incubated overnight (paraffin) or 3 hours (frozen) in PicoPure DNA extraction buffer (Arcturus Bioscience, Mountain View, CA) and were then denatured at 95°C for 10 minutes.
Enzyme Histochemistry Sections of frozen gastric mucosa were subjected to a dual enzyme histochemical staining to assay activity for CCO and succinate dehydrogenase as per published protocols.13,14 Briefly, sections were incubated in 0.2 mol/L phosphate buffer (pH 7.0) containing 100 mmol/L cytochrome c, 4 mmol/L diaminobenzidine tetrahydrochloride, and 20 g/mL catalase (All Sigma, Poole, UK) for approximately 50 minutes at 37°C in a humid chamber. Sections were washed 3 times for 5 minutes each in phosphate-buffered saline (PBS; Sigma) then incubated in 0.2 mol/L phosphate buffer (pH 7.0) containing 130 mmol/L sodium succinate, 200 mmol/L phenazine methosulfate, and 1.5 mmol/L nitroblue tetrazolium (all Sigma) for 45 to 90 minutes at 37°C until a blue color develops. Sections were then washed in PBS 3 times for 5 minutes and dehydrated through increasing ethanol concentrations and left in 100% ethanol for 10 minutes and allowed to dry. Sections that were stained on normal glass slides were mounted in Permount (Fisher Scientific, Fairlawn, NJ) and then coverslipped.
Immunohistochemistry Immunohistochemistry was performed by formalin-fixed, paraffin-embedded sections as per previously published protocols.8 Briefly, the primary antibody (mouse anti-human p53 [1:100] and -catenin [1:100];
DAKO, Ely, UK) were made up in 5% fetal calf serum (FCS) and PBS and applied to the sections for 35 minutes at room temperature then washed 3 times in PBS. The secondary rabbit anti-mouse antibody (DAKO) was also diluted in 5% FCS in PBS and applied for 30 minutes and washed as per the primary antibody. Avidin peroxidase (1:500 in PBS) was then applied for 30 minutes and washed in PBS 3 times and developed in liquid diaminobenzidine for approximately 2to 5 minutes. The sections were then dehydrated in increasing ethanol concentrations, cleared twice in xylene, and mounted in DePeX (Sigma) and coverslipped.
Mitochondrial DNA Polymerase Chain Reaction Sequencing The entire mitochondrial genome was sequenced in cells laser captured from frozen sections. To get sufficient product to sequence, a 2-round, nested polymerase chain reaction (PCR) protocol that amplified the entire mitochondrial DNA (mtDNA) in a series of overlapping fragments was used as per previous publications detailing primer sequences.14,15 Briefly, the mitochondrial genome was amplified in overlapping fragments by using a series of M13-tailed oligonucleotides. PCR products were sequenced by using BigDye3.1 terminator cycle sequencing chemistries on an ABI Prism 3730 DNA Analyzer (Applied Biosystems, Foster City CA) and compared directly with the revised Cambridge mtDNA reference sequence. Variations in the sequences of CCO-deficient cells, not present in neighboring CCO-normal cells, were considered to be clonal markers. All sequencing was repeated 3 times.
Genomic PCR Sequencing To screen for genomic DNA mutations, areas of dysplasia were identified on an H&E section, which was then traced onto the 3 preceding serial, dewaxed but unstained serial sections. Needle macrodissection of these areas was digested in 30 L PicoPure as described above. The lysate was used to screen for mutations in APC (adenamtous polyposis coli mutation cluster region), TP53 (exons 5– 8), KRAS (codons 12 and 13), CTNNB1 (-catenin, exon 11), CDKN2A (p16, exon 2), and PTEN (phosphatase and tensin homolog exons 5, 6, and 8). These genes account for approximately 85% of all somatic mutations previously reported in gastric adenocarcinomas according to the catalogue of somatic mutations in cancer (COSMIC) database (www.sanger .ac.uk/genetics/cgp/cosmic). Most mutations within the COSMIC database have been identified in cancer specimens and may be late events in the dysplasia:carcinoma sequence. After a mutation had been detected in tissue lysate, every laser-captured gland would then be screened individually for that mutation using a nested PCR protocol. Primers and PCR reaction conditions can be found in Supplementary Table1. DNA extracted
CLONAL EXPANSION IN THE HUMAN STOMACH
1253
from each gland was divided into 2 separate tubes to repeat each sequencing reaction. Furthermore, single cells were laser captured and subjected to PCR and sequencing for each gene. These were entirely unsuccessful, and, therefore, the contribution of any contaminating cells that were extraneous to each lasercaptured gland was negligible.
Loss of Heterozygosity Analysis Loss of heterozygosity (LOH) analysis was performed on individual gland lysates using a multiplexed microsatellite assay. Sixteen highly polymorphic microsatellites, located on chromosomes 3p (FHIT), 5q (APC), 9p (CDKN2A), 17p (TP53), 17q, and 18q (SMAD4), were amplified in 4 separate reactions using a multiplex PCR kit (Qiagen, Crawley, UK). Amplification of constitutional and individual gland lysate was done in parallel to control for PCR variability. Constitutional DNA was extracted from normal smooth muscle. Marker accession numbers and primer and reaction details are shown in Supplementary Table 2. Primers were tagged at the 5= end with either a FAM (tetrachloro-6-carboxyfluorescin) or HEX (hexachloro-6-carboxyfluorescin) fluorescent marker. Successfully amplified PCR products were analyzed on an ABI 3100 Genetic analyser (Applied Biosystems) using Genotyper 2.5 software (Applied Biosystems). LOH was then considered present if the area under 1 allelic peak was more than twice that of the other, after normalizing the peak areas relative to the constitutional DNA.
Mapping of Genomic Abnormalities All laser-captured glands were genotyped for the mutation discovered in the screening phase and analyzed for LOH at informative markers. Using an H&E image of the serial section to a postlaser-captured section as a guide, topographical maps of mutation spread were created with each phenotype and genotype assigned a unique color code.
Results Intestinal Metaplastic Glands Are Clonal To determine the clonal architecture of intestinal metaplastic glands, sections of snap-frozen gastric resection tissue were assayed by dual-enzyme histochemistry for CCO activity. Cells from glands that were entirely CCO deficient (blue, Figure 1A) were laser captured from the base to the surface of the gland (Figure 1B and C). Sequencing revealed the same mutation (7588G⬎A within the cytochrome c oxidase subunit II [MT-CO2] gene) in every CCO-deficient region captured, whereas sequencing cells captured along the length of a neighboring CCO-normal gland revealed an entirely wild-type
BASIC– ALIMENTARY TRACT
April 2011
1254
GUTIERREZ–GONZALEZ ET AL
GASTROENTEROLOGY Vol. 140, No. 4
Figure 1. Intestinal metaplastic glands from the human stomach are clonal. (A) CCO (brown) and SDH (blue) enzyme activity within an area of intestinal metaplasia. The blue gland is entirely deficient in CCO. (B and C) Pre- and postlaser capture microdissection sections of the same gland in (A). (D) Sequence trace from a laser-captured cell from the top of the CCO-deficient gland showing the same mtDNA mutation (7588 G⬎A transition in the MT-CO2 gene) as that from the bottom of the gland (E). (F) A neighboring CCO-normal gland shows a wild-type sequence. BASIC– ALIMENTARY TRACT
genotype (Figure 1D and E and Supplementary Figure 1). This provides conclusive evidence that intestinal metaplastic glands from the human stomach are indeed clonal. We observed a large patch of approximately 15 CCOdeficient metaplastic glands (Figure 2A), surprising because of the rarity of such CCO-deficient glands within the gastric mucosa. Sequencing the mitochondrial genome from cells taken from this large patch of CCOdeficient metaplastic glands revealed that each individual gland within the patch contained the same mutation (8503C⬎T within the mitochondrially-encoded 16S RNA (MT-RNR2) gene, Figure 2A–C). The surrounding CCOnormal glands did not contain this mutation (Figure 2D). We therefore conclude that every metaplastic gland within this patch is clonally derived from the same founder gland.
Dysplasia Can Originate From Metaplasia in the Human Stomach To investigate the origins of dysplasia within the human stomach, areas of dysplasia in each specimen were needle macrodissected and screened for mutations in genes known to be mutated in GA. In 23 patients screened, 3 (all of Japanese origin) showed a mutation common to both metaplasia and dysplasia. In patient 1, a c.4680insA mutation in APC was found in 7 of 42 metaplastic glands. Figure 3A–F shows the results of laser-capture microdissection of metaplastic, dysplastic,
and hyperplastic gastric glands from this specimen. Both the metaplastic (Figure 1A, [M]) glands adjacent to the dysplastic glands (Figure 1A, [D]) contained the c.4680insA APC mutation. Hyperplastic gastric glands (Figure 1A, [G]) were APC-wild type. This demonstrates that dysplasia was genetically related to the surrounding metaplasia. Patient 2 (c.4688insA mutation in APC) showed that 5 of 78 metaplastic glands shared the same mutation as the surrounding dysplasia and that, in patient 3 (c.643G⬎A mutation in TP53), there were 3 of 72 mutated metaplastic glands. This is summarized in Supplementary Table 3.
Whole Tissue Clonal Analysis Reveals a Founder Mutation in Fields of Dysplasia To determine how dysplasia develops and spreads in the human stomach, numerous glands of varying phenotypes were individually laser-capture microdissected and mapped according to their genotype. In patient 2, the spatial location of the c.4688insA in APC mutation in an area of mucosa displaying metaplasia and dysplasia is shown in Figure 4. Figure 4A is an H&E (in higher power, 4B), and Figure 4C shows the post-microdissected specimen (serial section to 4A) with the genotype of each gland color mapped. Although a vast majority of metaplastic glands were genetically wild type, the occasional metaplastic gland contained the APC mutation. Importantly, every dysplastic gland contained this mutation.
CLONAL EXPANSION IN THE HUMAN STOMACH
1255
Figure 2. Patches of mutated intestinal metaplastic glands arise from a founder gland. (A) Prelaser capture section of CCO-normal (brown) and CCO-deficient (blue) intestinal metaplastic gland. (B) Postlaser capture. (C) Representative sequence trace showing that each CCO-deficient gland contained the same mtDNA mutation (8503 C⬎T transition in the MT-RNR2 region). (D) Representative sequence trace showing a wild-type genotype for neighboring CCO-normal glands.
Supplementary Figure 2 demonstrates the mapping of an entire block from patient 1, showing that every dysplastic gland contained a c.4680insA in APC. Patient 3 (c.643A⬎G transition in TP53) was also mapped (data not shown). In each of these 3 patients, there was a single founder mutation throughout the dysplastic field (patient 1, 22/22 dysplastic glands; patient 2, 67/67; and patient 3, 158/158). This is compelling evidence, suggesting that the entire dysplastic lesion was derived from a single metaplastic gland. Furthermore, it is clear that this mutation was not the dysplasia-initiating event because mutated metaplastic glands with the same mutation are present. Supplementary Table 3 summarizes all data from these patients. It should be noted that, although 3 informative patients were found, the remaining 20 patients screened did not have a mutation in any of the genes that were screened. Spasmolytic polypeptide-expressing metaplasia (SPEM) has been implicated as a condition that can either give rise to IM or GA itself. To this end, we have performed immunohistochemistry for trefoil factor 2 (TFF2) in an attempt to detect SPEM. In each specimen TFF2 was detected at the base of normal pyloric glands but was generally absent from metaplastic or dysplastic glands. However, 1 specimen (patient 1, c.4680insA in APC) did show the rare intestinal metaplastic gland that expressed TFF2 (Supplementary Figure 3).
Genetic Diversity Is Subsequent to Clonal Expansion of the Founder Mutation To attempt to understand how a field of dysplasia may develop into cancer, genetic diversity was sought within fields of dysplasia by searching for further mutations and LOH within regions of genomic DNA that are associated with genes known to be mutated within gastric cancer. Figure 5A shows a strip of mucosa displaying both metaplasia and dysplasia from patient 1 (c.4680insA in APC). Figure 5B maps all glands with significant LOH of 2 microsatellite markers associated with TP53. Subclones consisting of 8 of 17 dysplastic glands in this strip of mucosa were identified by their TP53 LoH. Figure 5C shows representative LOH in both microsatellites associated with TP53. Patient 3 showed similar findings, and all 3 patients are summarized in Supplementary Table 4. Additional mutations can contribute to genetic diversity within fields of dysplasia. Figure 6 shows in patient 1 (c.4680insA APC founder mutation) isolated groups of p53-positive glands demonstrated by immunohistochemistry. When those p53-positive crypts were laser-capture microdissected, a TP53 mutation was found (c.818 G⬎A transition resulting in a p.272R⬎H amino acid change). Those neighboring dysplastic glands, which were p53negative or contained a few p53-positive cells, were TP53
BASIC– ALIMENTARY TRACT
April 2011
1256
GUTIERREZ–GONZALEZ ET AL
BASIC– ALIMENTARY TRACT
Figure 3. Intestinal metaplastic glands can be genetically related to surrounding dysplastic glands. (A) H&E of an area of gastric mucosa showing metaplastic (M), dysplastic (D), and hyperplastic gastric (G) glands. (B) Prelaser capture serial section. (C) Postlaser capture section. (D and E) Two representative sequence traces showing the presence of a c.4680insA in APC in both (D) metaplastic and (E) dysplastic glands from patient 1. (F) Hyperplastic gastric glands were APC wild type.
wild type. All other patients with dysplasia displaying founder mutations showed widespread p53 staining, making it impossible to phenotypically identify glands that contain further TP53 mutations. Further attempts to identify further genetic diversity by -catenin immunohistochemistry were unsuccessful because all specimens showed consistent membranous or frequent nuclear staining (Supplementary Figure 4). Supplementary Table 4 summarizes the genetic diversity found in the 3 informative specimens investigated.
Discussion Here, we have conclusively demonstrated that metaplastic glands within the human stomach are clonal and can clonally expand through the stomach. Previous work from this laboratory8 has questioned initial observations by Mihara et al9 that such glands were polyclonal as defined by methylation signatures of promoters of nonexpressed genes. It is known that CpG methylation does change over time16; moreover, we have shown, using partially CCO-deficient colonic crypts as a marker of clonal expansion, that CpG methylation does record the immediate ancestry of small clones within individual human colonic crypts. As these clones increase in size through the process of niche succession and monoclonal
GASTROENTEROLOGY Vol. 140, No. 4
conversion,13,17 their methylation signatures quickly diversify, and, by the time a crypt divides into 2 daughter crypts, their methylation signatures are no more similar than crypts distant to each other18 (unpublished observations). This suggests that the use of CpG methylation signatures as markers of clonal expansion is likely to be uninformative when considering the clonality of whole glands. We consider that a specific point mutation (here within the mitochondrial genome) is a reliable marker of clonal expansion and that the presence of the same mutation throughout an intestinal metaplastic gland confirms clonality. Field cancerization is the process by which a field of epithelial cells is preconditioned to allow the development of tumors.19 It is thought to expand by repeated gland fission events within the gastric mucosa.8 Gland fission, first proposed by Hattori and Fjuita,20 is the process by which a gland bifurcates at the neck/isthmus and divides into 2 distinct daughter glands. Here, we describe a large patch of CCO-deficient metaplastic glands, containing approximately 15 glands, each containing the same mtDNA mutation (and therefore derived from a founder gland). Intestinal metaplastic glands of course resemble intestinal crypts rather than gastric glands, but we have shown that clonally mutated patches arise in the human colon also through the mechanism of crypt fission.13 Individual CCO-deficient metaplastic glands in the stomach are rare when compared with those in the colon. In the normal human colon, the average patch size at the age of the patient we describe here is 2.5 crypts,13 and it is therefore interesting to observe such a large patch of mutated intestinal metaplastic glands. The reason for this is of course unknown, but it is tempting to relate it to the inflamed microenvironment,21 a concept supported by the prominent crypt fission seen in patients with active ulcerative colitis.22 We are limited in this analysis because of the rarity of CCOdeficient glands in the gastric mucosa. Evidence that metaplasia is an immediate precursor of dysplasia in the human stomach is limited. Previous studies looking at the genetic relationship between the 2 tend to focus on whole tissue DNA analysis investigating individual tumor suppressor genes.3,5,6 This study has, for the first time, attempted to identify genetic abnormalities on a gland-by-gland basis in genes accounting for approximately 85% of nuclear DNA mutations previously reported (COSMIC) in human GA. In our cohort, mutations in these genes were rarely detected strongly, suggesting that, if 2 glands shared the same mutation, they are very likely to have originated from the same gland, regardless of phenotype. Interestingly, most of the patients in this study displayed no obvious genetic relationship between metaplasia and dysplasia. These patients did not show any mutations in our panel of tumor suppressor genes in their entire and, in many cases, very extensive areas of dysplas-
CLONAL EXPANSION IN THE HUMAN STOMACH
1257
Figure 4. All dysplastic glands share a founder mutation. (A) A composite H&E of an entire strip of gastric mucosa showing intestinal metaplasia and dysplasia from patient 2 (c.4688insA in APC). (B) Higher power of highlighted areas. (C) Serial section, showing postlaser capture microdissection. The space left by the laser-capture microdissected gland has been colored to reflect the genotype. Blue filled areas are dysplastic glands positive for the c.4688insA mutation, red are positive metaplastic glands, and green are negative metaplastic glands.
tic mucosa. An explanation to why most specimens do not show this could be that, despite screening the most common genes known to be mutated in GA, there may be as yet undiscovered genes that are mutated are high frequency in GA. Methylation of tumor suppressor gene
promoters also plays an important role in carcinogenesis, and this could possibly explain the lack of mutations seen in our early gastric cancer patients. For example, in the esophagus, Eads et al have demonstrated that there is increased promoter methylation in genes such as APC
Figure 5. Genetic diversity develops subsequent to clonal expansion of the founder mutation. (A) A composite H&E of an area of gastric mucosa showing intestinal metaplasia and dysplasia in patient 1 (c.4680insA in APC). (B) Serial postlaser capture microdissection section. Mutated dysplastic gland without TP53 LOH (identified using microsatellite marker D17S1832) is distinguished by blue with green border, mutated dysplastic with TP53 LOH by blue with white border, mutated metaplastic without LOH by red with white border, and wild-type metaplastic with no LOH by green with white border. (C) Representative genotype analysis demonstrating LOH. DNA from underlying gastric muscle (constitutional DNA, top) and a mutated dysplastic gland (bottom) with clear LOH of 1 allele in each of the 2 markers associated with TP53.
BASIC– ALIMENTARY TRACT
April 2011
1258
GUTIERREZ–GONZALEZ ET AL
GASTROENTEROLOGY Vol. 140, No. 4
BASIC– ALIMENTARY TRACT
Figure 6. (A and B) P53 immunohistochemistry of an area of dysplasia showing a p53-positive dysplastic gland (*) in an otherwise p53-negative (§) specimen from patient 1 (c.4680insA APC founder mutation). (C) Pre- and postlaser capture serial section. (D) Sequencing revealing a c.818 G⬎A transition mutation in TP53 in the p53-positive gland (top) and the neighboring p53-negative gland is wild type (bottom). (E and F) Representative p53 expression is widespread in mutated dysplastic glands from the other patients.
and CDKN2A in the progression of Barrett’s esophagus to esophageal adenocarcinoma.23,24 Somatic changes in methylation may drive gastric carcinogenesis, and work is underway to address this. It is also pertinent to consider whether IM is the only pathway to dysplasia. In cases in which a mutation is shared between IM and dysplasia, there is a definite sequence of genetic events that are inherited as the disease progresses, and the origins of dysplasia are clear. We cannot discount the possibility that other, undetected, (epi-)mutations link metaplasia and dysplasia in the cases that were wild type for our panel of genes. However, IM may not be the only precursor of dysplasia, and, in this respect, there is much current interest in SPEM.25 Our study provides a precedent to perform analogous clonal analysis of concomitant SPEM, IM, and dysplastic lesions
to determine the biologic and clinical significance of SPEM plaques. Large-scale clonal expansion of dysplastic glands has previously been observed in the human gastrointestinal tract. Leedham et al demonstrated that a single mutation was able to spread through an entire lesion of ulcerative colitis-associated neoplasia.11 However, the existence of founder mutations in a lesion appears to be dependent on the specific lesion. For example, Thirlwell et al26 have recently shown that colonic adenomas from patients with familial adenomatous polyposis are polyclonal containing multiple clones with distinct patterns of APC inactivation and that, whereas sporadic adenomas frequently contain multiple clones, only one progresses to the resulting invasive carcinoma. Furthermore, genetic heterogeneity is a feature of Barrett’s esophagus and is
thought to promote tumor progression,27,28 although this has been disputed.29,30 Dysplasia within the human stomach appears to differ slightly from these examples. There is a clonal proliferation of a single mutated metaplastic gland that eventually becomes dysplastic, and this expands forming an entirely clonal dysplastic lesion. Further genetic abnormalities accrue and expand on top of the founder mutation. We know that these early mutations are not dysplasia-initiating events because, if that were the case, we would not observe any mutated metaplastic glands. We have no evidence to suggest that specific mutations would influence the rate of expansion by fission, although we would predict that, as more genetic abnormalities are acquired, that the fission index would increase. Analysis of clonal expansion is important when we consider how a premalignant lesion progresses into cancer. The greater the expansion, the greater the probability of acquiring a tumor-initiating defect. Competition between clones has been suggested to be the driving factor of the development of esophageal adenocarcinoma arising from Barrett’s esophagus.27 Such models predict that the more clones present in premalignant lesion, the more likely it will progress to cancer. Does this accord with what is seen in the human gastric mucosa? The answer is that it probably does; however, such clonal competition only arises when a substantial field of genetic heterogeneity exists. Here, we show an entire field of dysplastic glands exhibit a founder mutation. This would indicate that the lesion is clonal and competition between dysplastic glands may not be prominent but rather have a selective advantage over surrounding wild-type metaplastic glands. Over time, genetic diversity does occur with glands exhibiting LOH or additional mutations arising on the back of the founder mutation. Clonal competition could therefore become more of a driving force, and we would predict that lesions with more diversity would be more likely to develop into cancer. Clonal ordering of genetic abnormalities allows us to describe the genetic ancestry of dysplastic fields and permits the creation of phylogenetic trees (Figure 7). In this Figure, we show the time line by which each abnormality occurs in each informative patient as they progress from chronic gastritis to metaplasia to dysplasia. It is clear that a vast majority of metaplastic glands are genetically wild type with only a small percentage displaying a mutation. When a mutation does arise in a metaplastic gland, the initiation of dysplasia must occur rapidly because the proportion of mutated metaplastic to dysplastic glands is very small. Although the presence of mutated metaplastic glands indicates that the acquisition of such mutations is not a dysplasia-initiating event, an overwhelming number of mutated glands are nevertheless dysplastic. It is important to note that, although mutated cases are limited in number, we do not think that there is a specific order of mutations within metaplasia:dysplasia sequence
CLONAL EXPANSION IN THE HUMAN STOMACH
1259
Figure 7. A phylogenetic tree demonstrating linear progression of genetic defects in the development of dysplasia in the human stomach. Patients 1–3 demonstrate our working hypothesis that the dysplasia develops from a mutated metaplastic gland and that every dysplastic gland can trace its ancestry to this metaplastic gland of origin. Over time, genetic diversity can develop, and we hypothesize that the greater the extent of diversity, the more likely cancer will develop. Patient 4 is entirely wild type and accounts for the majority of patients in this study suggesting that there may be mutations in genes hitherto unreported, tumor suppressor genes silenced by methylation, or that dysplasia arises independently of the surrounding metaplasia.
and that an APC founder mutation may be as frequently detected as a TP53 founder mutation. To conclude, these data demonstrate that intestinal metaplastic glands from the human stomach are clonal and can expand by fission. When a mutation within a tumor suppressor gene occurs in an intestinal metaplastic gland, it quickly becomes dysplastic and then expands, rapidly forming a clonal field of dysplasia. Such fields promote the development of additional genetic abnormalities resulting in multiple independent subclones. This process is not apparently universal and may be one of several mechanisms behind the development of gastric carcinogenesis.
Supplementary Material Note: To access the supplementary material accompanying this article, visit the online version of Gastroenterology at www.gastrojournal.org, and at doi: 10.1053/j.gastro.2010.12.051. References 1. Compare D, Rocco A, Nardone G. Risk factors in gastric cancer. Eur Rev Med Pharmacol Sci 2010;14:302–308. 2. Uemura N, Okamoto S, Yamamoto S, et al. Helicobacter pylori infection and the development of gastric cancer. N Engl J Med 2001;345:784 –789. 3. Correa P, Shiao YH. Phenotypic and genotypic events in gastric carcinogenesis. Cancer Res 1994;54:S1941–S1943.
BASIC– ALIMENTARY TRACT
April 2011
1260
GUTIERREZ–GONZALEZ ET AL
BASIC– ALIMENTARY TRACT
4. Correa P. A human model of gastric carcinogenesis. Cancer Res 1988;48:3554 –3560. 5. Fenoglio-Preiser CM, Wang J, Stemmermann GN, et al. TP53 and gastric carcinoma: a review. Hum Mutat 2003;21:258 –270. 6. Segal F, Kaspary APB, Prolla JC, et al. p53 Protein overexpression and p53 mutation analysis in patients with intestinal metaplasia of the cardia and Barrett’s esophagus. Cancer Lett 2004;210: 213–218. 7. Gong C, Mera R, Bravo JC, et al. KRAS mutations predict progression of preneoplastic gastric lesions. Cancer Epidemiol Biomarkers Prev 1999;8:167–171. 8. McDonald SAC, Greaves LC, Gutierrez-Gonzalez L, et al. Mechanisms of field cancerization in the human stomach: the expansion and spread of mutated gastric stem cells. Gastroenterology 2008;134:500 –510. 9. Mihara M, Yoshida Y, Tsukamoto T, et al. Methylation of multiple genes in gastric glands with intestinal metaplasia: a disorder with polyclonal origins. Am J Pathol 2006;169:1643–1651. 10. de Dombal FT, Price AB, Thompson H, et al. The British Society of Gastroenterology early gastric cancer/dysplasia survey: an interim report. Gut 1990;31:115–120. 11. Leedham SJ, Graham TA, Oukrif D, et al. Clonality, founder mutations, and field cancerization in human ulcerative colitis-associated neoplasia. Gastroenterology 2009;136:542–550. 12. Leedham SJ, Preston SL, McDonald SAC, et al. Individual crypt genetic heterogeneity and the origin of metaplastic glandular epithelium in human Barrett’s oesophagus. Gut 2008;57:1041– 1048. 13. Greaves LC, Preston SL, Tadrous PJ, et al. Mitochondrial DNA mutations are established in human colonic stem cells, and mutated clones expand by crypt fission. Proc Natl Acad Sci U S A 2006;103:714 –719. 14. Taylor RW, Barron MJ, Borthwick GM, et al. Mitochondrial DNA mutations in human colonic crypt stem cells. J Clin Invest 2003; 112:1351–1360. 15. Taylor RW, Taylor GA, Durham SE, et al. The determination of complete human mitochondrial DNA sequences in single cells: implications for the study of somatic mitochondrial DNA point mutations. Nucleic Acids Res 2001;29:E74-E74. 16. Graff JR, Gabrielson E, Fujii H, et al. Methylation patterns of the E-cadherin 5= CpG island are unstable and reflect the dynamic, heterogeneous loss of E-cadherin expression during metastatic progression. J Biol Chem 2000;275:2727–2732. 17. McDonald SAC, Preston SL, Greaves LC, et al. Clonal expansion in the human gut: mitochondrial DNA mutations show us the way. Cell Cycle 2006;5:808 – 811. 18. Yatabe Y, Tavaré S, Shibata D. Investigating stem cells in human colon by using methylation patterns. Proc Natl Acad Sci U S A 2001;98:10839 –10844. 19. Slaughter DP, Southwick HW, Smejkal W. Field cancerization in oral stratified squamous epithelium; clinical implications of multicentric origin. Cancer 1953;6:963–968.
GASTROENTEROLOGY Vol. 140, No. 4
20. Hattori T, Fjuita S. Fractographic study on the growth and multiplication of the gastric gland of the hamster. The gland division cycle. Cell Tissue Res 1974;153:145–149. 21. Wong W-M, Mandir N, Goodlad RA, et al. Histogenesis of human colorectal adenomas and hyperplastic polyps: the role of cell proliferation and crypt fission. Gut 2002;50:212–217. 22. Chen R. The initiation of colon cancer in a chronic inflammatory setting. Carcinogenesis 2005;26:1513–1519. 23. Eads CA, Lord RV, Kurumboor SK, et al. Fields of aberrant CpG island hypermethylation in Barrett’s esophagus and associated adenocarcinoma. Cancer Res 2000;60:5021–5026. 24. Eads CA, Lord RV, Wickramasinghe K, et al. Epigenetic patterns in the progression of esophageal adenocarcinoma. Cancer Res 2001;61:3410 –3418. 25. Goldenring JR, Nam KT, Wang TC, et al. Spasmolytic polypeptideexpressing metaplasia and intestinal metaplasia: time for reevaluation of metaplasias and the origins of gastric cancer. Gastroenterology 2010;138:2207–2210. 26. Thirlwell C, Will OC, Domingo E, et al. Clonality assessment and clonal ordering of individual neoplastic crypts shows polyclonality of colorectal adenomas. Gastroenterology 2010;138:1441– 1454. 27. Maley CC, Galipeau PC, Finley JC, et al. Genetic clonal diversity predicts progression to esophageal adenocarcinoma. Nat Genet 2006;38:468 – 473. 28. Merlo LMF, Pepper JW, Reid BJ, et al. Cancer as an evolutionary and ecological process. Nat Rev Cancer 2006;6:924 –935. 29. Barrett MT, Sanchez CA, Prevo LJ, et al. Evolution of neoplastic cell lineages in Barrett oesophagus. Nat Genet 1999;22:106 – 109. 30. Wong DJ, Paulson TG, Prevo LJ, et al. p16(INK4a) lesions are common, early abnormalities that undergo clonal expansion in Barrett’s metaplastic epithelium. Cancer Res 2001;61:8284 – 8289.
Received September 7, 2010. Accepted December 23, 2010. Reprint requests Address requests for reprints to: Stuart A. C. McDonald, PhD, Centre for Digestive Diseases, Blizard Institute of Cell and Molecular Sciences, Barts and the London School of Medicine and Dentistry, 4 Newark Street, Whitechapel, London, United Kingdom E1 2AT. e-mail: [email protected]. Conflicts of interest The authors disclose no conflicts. Funding Supported by the Medical Research Council (to S.A.C.M.) and Cancer Research UK (to T.A.G., J.A.Z., N.A.W.).
April 2011
CLONAL EXPANSION IN THE HUMAN STOMACH
1260.e1
Supplementary Figure 1. H&E stain of a region of intestinal metaplasia associated with that shown in printed article Figure 1. Original magnification, ⫻100.
Supplementary Figure 2. A complete map of 1 block of the founder APC mutation found in patient 1 (4680insA). Glands colored dark blue are mutated and dysplastic. Glands colored red are metaplastic and mutated. Green and light blue represent wild-type metaplastic and gastric glands, respectively. Every dissected dysplastic gland contained the same mutation. Only a minority of the metaplastic glands were mutated.
1260.e2
GUTIERREZ–GONZALEZ ET AL
GASTROENTEROLOGY Vol. 140, No. 4
Supplementary Figure 3. Immunohistochemistry for trefoil factor 2. Two examples (from patient 1, APC mutated) of contiguous trefoil factor (TFF) 2-positive cells (arrows) and goblet cell-containing epithelium, suggesting the presence of spasmolytic polypeptide-expressing metaplasia or UACL (ulcer-associated cell lineage). (A) Example 1, low-power magnification, ⫻100. (B) Example 1, high-power magnification, ⫻200. (C) Example 2, low-power magnification, ⫻100. (D) Example 2, high-power magnification, ⫻200. Hybridoma supernatant mouse immunoglobulin M anti-human TFF2 generated in-house at Cancer Research UK, Histopathology Department, London, UK) was used as a primary antibody.
April 2011
CLONAL EXPANSION IN THE HUMAN STOMACH
1260.e3
Supplementary Figure 4. -Catenin immunohistochemistry in sections containing intestinal metaplasia (A) and dysplasia (B–E). In patient 2 (4668insA APC mutated) revealed frequent nuclear -catenin staining in each dysplastic gland (B and C). Patient 1 (4680insA APC mutated) showed less nuclear and more membranous -catenin expression (D and E). In each case, the expression of -catenin was consistent throughout each dysplastic gland.
1260.e4
GUTIERREZ–GONZALEZ ET AL
GASTROENTEROLOGY Vol. 140, No. 4
Supplementary Table 1. Primer Details and Reaction Conditions for Sequenced Loci Primer
Sequence 5= to 3=
p16-2A 1st F p16-2A 1st R p16-2B 1st F p16-2B 1st R p16-2A 2nd F p16-2A 2nd R p16-2B 2nd F p16-2B 2nd R K-ras 1st F K-ras 1st R K-ras 2nd F K-ras 2nd R p53-5 1st F p53-5 1st R p53-6 1st F p53-6 1st R p53-7 1st F p53-7 1st R p53-8 1st F p53-8 1st R p53-5 2nd F p53-5 2nd R p53-6 2nd F p53-6 2nd R p53-7 2nd F p53-7 2nd R p53-8 2nd F p53-8 2nd R APC-1 1st F APC-1 1st R APC-2 1st F APC-2 1st R APC-3 1st F APC-3 1st R APC-4 1st F APC-4 1st R APC-5 1st F APC-5 1st R APC-6 1st F APC-6 1st R APC-7 1st F APC-7 1st R APC-8 1st F APC-8 1st R APC-9 1st F APC-9 1st R APC-10 1st F APC-10 1st R APC-11 1st F APC-11 1st R APC-12 1st F APC-12 1st R APC-1 2nd F APC-1 2nd R APC-2 2nd F APC-2 2nd R APC-3 2nd F APC-3 2nd R APC-4 2nd F APC-4 2nd R APC-5 2nd F
GCTTCCTTTCCGTCATGC CAGGTACCGTGCGACATC CTGTTCTCTCTGGCAGGTCA TGTGCTGGAAAATGAATGCT CCTGGCTCTGACCATTCTGT CAGCTCCTCAGCCAGGTC CTTCCTGGACACGCTGGT TGGAAGCTCTCAGGGTACAAA GAGTTTGTATTAAAAGGTACTGGTGGA ATCAAAGAATGGTCCTGCAC TTTGATAGTGTATTAACCTTAT TATTAAAACAAGATTTACCTC CACTTGTGCCCTGACTTTCA GAGCAATCAGTGAGGAATCAGA AGAGACGACAGGGCTGGTT TGGAGGGCCACTGACAAC TGCTTGCCACAGGTCTCC GGTCAGAGGCAAGCAGAGG TTTTTAAATGGGACAGGTAGGA CACCCTTGGTCTCCTCCAC TCTGTCTCCTTCCTCTTCCTACA AACCAGCCCTGTCGTCTCT CAGGCCTCTGATTCCTCACT CTTAACCCCTCCTCCCAGAG CTTGGGCCTGTGTTATCTCC GTGTGCAGGGTGGCAAGT GCCTCTTGCTTCTCTTTTCC GCTTCTTGTCCTGCTTGCTT GGACAAAGCAGTAAAACCGAAC AACTACATCTTGAAAAACATATTGGA AAGTGGTCAGCCTCAAAAGG GCTATTTGCAGGGTATTAGCA GATACTCCAATATGTTTTTCAAGATG GCCTGGCTGATTCTGAAGAT CCCTGCAAATAGCAGAAATAAAA AACATGAGTGGGGTCTCCTG CAGACTGCAGGGTTCTAGTTTATC CATTCCACTGCATGGTTCAC CCAAAAGTGGTGCTCAGACA CATGGTTTGTCCAGGGCTAT TTTGAGAGTCGTTCGATTGC TCTCTTTTCAGCAGTAGGTGCTT CATGCAGTGGAATGGTAAGT GCAGCATTTACTGCAGCTT CAAGCGAGAAGTACCTAAAAA TTCTGTATAAATGGCTCATCG GGTTCTTCCAGATGCTGATA CTTGGTTTTCATTTGATTCTTT AATTAAGAATAATGCCTCCAGT TTTACGTGATGACTTTGTTGG CCAAGAGAAAGAGGCAGAAAAA TGATGGTAGAAGTTTGTACACAGG CAAGCAGTGAGAATACGTCCA TTTCTTGGTTAATAGAAGAAACTTTGC GCCACTTGCAAAGTTTCTTCT TGCTTCCTGTGTCGTCTGA CAGACGACACAGGAAGCAGA TGGAACTTCGCTCACAGGAT GATCCTGTGAGCGAAGTTCC CTGAGCACCACTTTTGGAGG AGAATCAGCCAGGCACAAAG
Reaction conditions 60/2/5 60/2/5 60/2/5 60/2/5 60/2/5 55/2/5 55/1/5 60/2/5 60/1/5 60/2/5 60/1/5 60/1/0 60/1/5 60/2/0 55/1/0 60/3/5 55/1/0 55/2/5 60/2/0 60/2/0 60/1/0 55/2/0 55/2/0 55/1/5 55/2/0 55/1/5 55/1/0 55/2/5 60/1/0 60/1/0 55/1/0
Supplementary Table 1. Continued Primer
Sequence 5= to 3=
APC-5 2nd R APC-6 2nd F APC-6 2nd R APC-7 2nd F APC-7 2nd R APC-8 2nd F APC-8 2nd R APC-9 2nd F APC-9 2nd R APC-10 2nd F APC-10 2nd R APC-11 2nd F APC-11 2nd R APC-12 2nd F APC-12 2nd R PTEN-5 1st F PTEN-5 1st R PTEN-5 2nd F PTEN-5 2nd R PTEN-6 1st F PTEN-6 1st R PTEN-6 2nd F PTEN-6 2nd R PTEN-8b 1st F PTEN-8b 1st R PTEN-8b 2nd F PTEN-8b 2nd R
GCAATCGAACGACTCTCAAA CACTATGTTCAGGAGACCCCA TGGAAGATCACTGGGGCTTA GTGAACCATGCAGTGGAATG ACTTCTCGCTTGGTTTGAGC GCTTAGGTCCACTCTCTCTCTT GGCATTATAAGCCCCAGT CCTAAAAATAAAGCACCTACTGCTG CACTCAGGCTGGATGAACAA ACATTTTGCCACGGAAAGTA GGCTGCTCTGATTCTGTTTC CAGGAAAATGACAATGGGAAT TGGCATGGCAGAAATAATACA GGACCTATTAGATGATTCAGATGATG ACTTGGTTTCCTTGCCACAG TGCAACATTTCTAAAGTTACCTACTTG GAAACCCAAAATCTGTTTTCCA TTCTGAGGTTATCTTTTTACCACA GGAAGAGGAAAGGAAAAACATC GGCTACGACCCAGTTACCAT CCTGCATAAATTTCAAATGTGG TTTTTCAATTTGGCTTCTCTTTTT TGTTCCAATACATGGAAGGATG CAAAATGTTTCACTTTTGGGTAAA CTCCTAGAATTAAACACACATCACA ACCAGGACCAGAGGAAACCT AGTCAACAACCCCCACAAAA
Reaction conditions 60/1/0 60/1/0 60/1/0 60/3/5 55/2/0 55/2/0 55/2/0 55/1.5/5 55/1.5/5 60/1.5/0 60/2.5/5 55/2.5/0 60/1.5/5
NOTE. Naming nomenclature is as follows: gene-exon, polymerase chain reaction round, primer direction. Large exons, consisting of more than 250 nucleotides, were subdivided into separate regions for amplification; a letter following the exon number denotes these regions (eg, p16-2A). For APC primers, the numbers do not refer to exons but identify regions of the mutation cluster region. Reaction condition nomenclature is annealing temperature/magnesium concentration (mmol/L)/addition of Q-solution. F, forward; R, reverse.
April 2011
CLONAL EXPANSION IN THE HUMAN STOMACH
1260.e5
Supplementary Table 2. Microsatellite Loss of Heterozygosity Analysis Primers and Reaction Details Primer Multiplex 1 D18S58 F D18S58 R D5S346 F D5S346 R D9S932 F D9S932 R D3S1300 F D3S1300 R Multiplex 2 D17S250 F D17S250 R D18S474 F D18S474 R D17S1832 F D17S1832 R D3S1313 F D3S1313 R Multiplex 3 D17S1176 F D17S1176 R D17S1678 F D17S1678 R D17S1881 F D17S1881 R Multiplex 4 D9S942 F D9S942 R D5S2001 F D5S2001 R D17S1506E F D17S1506E R D5S489 F D5S489 R
Sequence 5= to 3=
Conditions 57/0
GCTCCCGGCTGGTTTT GCAGGAAATCGCAGGAACTT ACTCACTCTAGTGATAAATCGGG AGCAGATAAGACAGTATTACTAGTT CTCCCTTTGTATTTCTGTTCTATT AAGCTATGATGGTGCCACC ACAAAGGAACGTCATGTGGTAGG GCTGTTTATTCTTCGTGGAATGCC 57/0 GGAAGAATCAAATAGACAAT GCTGGCCATATATATATTTAAACC CTCCACCCACTAGATGTCAG ACTTGCTTAAGCCTTGGACT ACGCCTTGACATAGTTGC TGTGTGACTGTTCAGCCTC TACTTTCCTTCAGATCCTTGG AACTAGGGGCCATGAATAAG 57/0 ACTTCATATACATATCACGTGC TCAATGGAGAATTACGATAGTG TTTGGGTCTTTGAACCCTTG CCACAACAAAACACCAGTGC CCCAGTTTAAGGAGTTTGGC TAGGGCAGTCAGCCTTGTG 57/5 GCAAGATTCCAAACAGTA CTCATCCTGCGGAAACCATT GCCAAGATGGTCTCGATCTC TCTGAACAGGTGATGGCAAC TGTGGGATGGGGTGAGATTTC CTGTTGGTCGGTGGGTTG GGGCTTTTGTGTTGTTTCTA GAAAACCCATAACCAGACTTG
NOTE. Reaction condition nomenclature is annealing temperate/Qsolution used in reaction. Other conditions were performed according to manufacturer’s instructions.
Supplementary Table 3. Summary of Mutation Status of All Hyperplastic Gastric, Metaplastic, and Dysplastic Glands Laser Captured From at Least 2 Blocks From Each Patient Phenotype (No. glands) Hyperplastic
Metaplastic (%)
Dysplastic
Patient
Wild type (%)
Mutated
Wild type
Mutated
Wild type
Mutated (%)
1 2 3
10 (100) 15 (100) 14 (100)
0 0 0
35 (83.3) 73 (93.6) 69 (95.8)
7 (16.6) 5 (6.4) 3 (4.2)
0 0 0
22 (100) 67 (100) 158 (100)
NOTE. Mutation: Patient 1 ⫽ c.4680 Ains APC, Patient 2 ⫽ c.4668 Ains APC, Patient 3 ⫽ c.643A⬎G TP53. Data expressed as number of glands and (percentage). All samples had a small number of mutated metaplastic glands; this became a founder mutation for the entire dysplastic field.
1260.e6
GUTIERREZ–GONZALEZ ET AL
GASTROENTEROLOGY Vol. 140, No. 4
Supplementary Table 4. Summary of Genetic Diversity in a Clonal Field of Dysplasia From All Informative Patients No. dysplastic glands Genetic abnormality TP53 LoH TP53 mutation APC LoH APC mutation
Patient 1 (%)
Patient 2 (%)
Patient 3 (%)
8/25 (32) 0 2/25 (8.0) 0
20/60 (30) 5/60 (8.3) 0 0
29/63 (46) 0 0 0
NOTE. Founder mutations: Patient 1 ⫽ c.4680 Ains APC, Patient 2 ⫽ c.4668 Ains APC, Patient 3 ⫽ c.643A⬎G TP53. Data expressed as number of glands and (percentage). Although all dysplastic crypts contain a founder mutation, genetic heterogeneity does develop over time. LoH, loss of heterozygosity.
The Clonal Origins of Dysplasia From Intestinal Metaplasia in the Human Stomach LYDIA GUTIERREZ–GONZALEZ,*,‡ TREVOR A. GRAHAM,* MANUEL RODRIGUEZ–JUSTO,§ SIMON J. LEEDHAM,* MARCO R. NOVELLI,§ LAURA J. GAY,储 TANIA VENTAYOL–GARCIA,储 ALICIA GREEN,储 IAN MITCHELL,¶ DAVID L. STOKER,¶ SEAN L. PRESTON,# SHIGEKI BAMBA,** EIJI YAMADA,‡‡ YUUKI KISHI,§§ REBECCA HARRISON,储 储 JANUSZ A. JANKOWSKI,储 NICHOLAS A. WRIGHT,*,储 and STUART A. C. McDONALD*,储 *Histopathology Laboratory, London Research Institute, Cancer Research UK, London, United Kingdom; ‡Health Sciences Institute of Aragon, CIBERehd, Zaragoza, Spain; §Department of Histopathology, University College London, London; 储Centre for Digestive Diseases, Blizard Institute of Cell and Molecular Sciences, Barts and the London Medical School, Queen Mary, University of London, London; ¶Department of Surgery, University College London, London; #Digestive Diseases Clinical Academic Unit, Barts and the London NHS Trust, London, United Kingdom; **Department of Gastroenterology, Shiga University of Medical Science Seta-Tsukinowa, Ohtsu, Shiga, Japan; ‡‡Department of Pathology and §§Department of Internal Medicine, Hikone Muncipal Hospital, Hikone, Japan; 储 储Department of Pathology, University Hospitals Leicester, Leicester, United Kingdom
Keywords: Stomach Cancer; Tumor Development; Neoplasia; Oncogenesis. View this article’s video abstract at www.gastrojournal.org
G
astric adenocarcinoma (GA) is a major cause of cancer-related mortality worldwide (http://info .cancerresearchuk.org/cancerstats; Cancer Research UK, 2010, London, UK, viewed 6th March 2011). There are a multitude of risk factors: within these, infection with Helicobacter pylori and the development of intestinal metaplasia (IM) are prominent.1,2 IM of the human stomach is the replacement of the normal gastric glands with an intestinal crypt phenotype. Current dogma proposes that there is a stepwise progression from chronic atrophic gastritis to intestinal metaplasia, with subsequent progression to low-/high-grade dysplasia, eventually culminating in the intestinal type of GA.3,4 Using whole tissue specimens, previous studies have suggested that metaplasia can contain similar genetic abnormalities to the associated GA,3,5–7 suggesting that IM is a premalignant condition from which a carcinoma may develop. Whereas some studies demonstrate a genetic link,7 some reported abnormalities were also present in adjacent normal tissue and were therefore most probably polymorphisms.3 To date, a clear genetic link, on a gland-by-gland basis, describing the evolution of metaplasia into dysplasia has not been produced. The clonal nature of metaplastic glands in the human stomach is a source of some controversy. We have previously shown that human IM glands from patients undergoing resection for GA can be entirely deficient in cytochrome c oxidase (CCO).8 This deficiency has been shown to be highly suggestive of an underlying mutation in the mitochondrial genome resulting in CCO deficiency; therefore, IM glands from the human stomach appear clonal. Other groups have suggested otherwise; Abbreviations used in this paper: CCO, cytochrome c oxidase; GA, gastric adenocarcinoma; IM, intestinal metaplasia; LOH, loss of heterozygosity; mtDNA, mitochondrial DNA; SPEM, spasmolytic polypeptide-expressing metaplasia; TFF, trefoil factor. © 2011 by the AGA Institute 0016-5085/$36.00 doi:10.1053/j.gastro.2010.12.051
BASIC– ALIMENTARY TRACT
BACKGROUND & AIMS: Studies of the clonal architecture of gastric glands with intestinal metaplasia are important in our understanding of the progression from metaplasia to dysplasia. It is not clear if dysplasias are derived from intestinal metaplasia or how dysplasias expand. We investigated whether cells within a metaplastic gland share a common origin, whether glands clonally expand by fission, and determine if such metaplastic glands are genetically related to the associated dysplasia. We also examined the clonal architecture of entire dysplastic lesions and the genetic changes associated with progression within dysplasia. METHODS: Cytochrome c oxidase-deficient (CCO⫺) metaplastic glands were identified using a dual enzyme histochemical assay. Clonality was assessed by laser capture of multiple cells throughout CCO⫺ glands and polymerase chain reaction sequencing of the entire mitochondrial DNA (mtDNA) genome. Nuclear DNA abnormalities in individual glands were identified by laser capture microdissection polymerase chain reaction sequencing for mutation hot spots and microsatellite loss of heterozygosity analysis. RESULTS: Metaplastic glands were derived from the same clone—all lineages shared a common mtDNA mutation. Mutated glands were found in patches that had developed through gland fission. Metaplastic and dysplastic glands can be genetically related, indicating the clonal origin of dysplasia from metaplasia. Entire dysplastic fields contained a founder mutation from which multiple, distinct subclones developed. CONCLUSIONS: There is evidence for a distinct clonal evolution from metaplasia to dysplasia in the human stomach. By field cancerization, a single clone can expand to form an entire dysplastic lesion. Over time, this field appears to become genetically diverse, indicating that gastric cancer can arise from a subclone of the founder mutation.
1252
GUTIERREZ–GONZALEZ ET AL
BASIC– ALIMENTARY TRACT
Mihara et al9 demonstrated that, within single IM glands, patterns of CpG methylation in the promoter regions of the ZIK1 gene, which is normally silenced in the human stomach, are highly polymorphic, suggesting that metaplastic glands develop multifocally, arising from multiple stem cells. On the other hand, methylation patterns are known to change over time, and the variability observed by Mihara et al9 may reflect rapid diversification rather than stable clones. To understand the progression and expansion of intestinal metaplasia and its relationship to dysplasia, the stem cell and clonal architecture of IM become important. Early gastric intestinal type adenocarcinomas are usually associated with fields of dysplastic glands.10 Again, it is unknown how these fields develop and spread through the stomach, namely whether by multiple dysplasia-initiating events occurring independently or through extensive clonal expansion of a single dysplastic gland through a large area of gastric mucosa. There is precedent for both scenarios in the human gastrointestinal tract: in patients with ulcerative colitis-associated neoplasias, a single mutated crypt is able to spread through large areas of colonic mucosa encompassing the entire lesion,11 a socalled founder mutation. On the other hand, Barrett’s esophagus appears to be a genetically heterogeneous disease where there are multiple clones within the Barrett’s mucosa.12 Even leaving aside the intrinsic interest of the true nature of the development of intestinal-type gastric carcinoma, the lack of such knowledge severely impairs our ability to identify reliable biomarkers to assess the risk of gastric carcinogenesis. Here, we show, by using either cell-by-cell or gland-bygland laser capture microdissection of human gastric epithelium that intestinal metaplastic glands are clonal, the intestinal metaplasia clonally expands by crypt fission to form large patches, and that dysplasia can arise from a single clone of mutated intestinal metaplastic glands and expand to form the entire dysplastic lesion. Furthermore, we show that, within this clonal dysplastic field, genetic diversity can arise, raising the possibility that cancer develops through genetic progression of a subclone arising from the founder mutation.
Patients and Methods Patients Intestinal metaplastic and dysplastic formalinfixed, paraffin-embedded and frozen specimens were obtained from patients undergoing resection for either GA or high-grade dysplasia and from pathology archives. A total of 23 patient specimens (age range, 57–72 years) were used in this study. Ethical approval was sought and obtained as per United Kingdom Human Tissue Act (2006, reference 07/Q1604/17) and Hikone Municipal Hospital (Japan) regulations. Histology of each specimen was confirmed by 2 pathologists (N.A.W. and M.R.J.). All
GASTROENTEROLOGY Vol. 140, No. 4
specimens showed IM surrounding or in between large areas of dysplasia (or early gastric cancer) and was consistent through multiple blocks from each patient.
Laser Capture Microdissection Ten serial sections (5–20 m) of each specimen were prepared. Sections 1–5 were cut onto normal glass slides, and sections 6 –10 were cut onto P.A.L.M. MembraneSlide 1.0 PEN slides (Zeiss Microimaging, Munich, Germany). Paraffin sections were left overnight at 37°C, dewaxed in xylene, rehydrated through decreasing ethanol concentrations to water, and then air-dried. Frozen sections were allowed to air-dry immediately after cutting. Each membrane section underwent light staining with methylene green (2%; Sigma, Poole, UK). Paraffin sections 1–3 were used to screen for genomic mutations. Sections 4 and 5 were used for H&E or alcian blue/ periodic acid—Schiff/diastase staining. On sections 6 –10, areas of interest were delineated using P.A.L.M. Robosoftware (Zeiss Microimaging) and cut into 0.5-mL AdhesiveCap tubes using a P.A.L.M. Laser capture microdissection system (Zeiss Microimaging). The majority of the frozen sections was used for enzyme histochemistry with an occasional section (approximately 1 per 20 sections cut) used for H&E. Captured specimens were incubated overnight (paraffin) or 3 hours (frozen) in PicoPure DNA extraction buffer (Arcturus Bioscience, Mountain View, CA) and were then denatured at 95°C for 10 minutes.
Enzyme Histochemistry Sections of frozen gastric mucosa were subjected to a dual enzyme histochemical staining to assay activity for CCO and succinate dehydrogenase as per published protocols.13,14 Briefly, sections were incubated in 0.2 mol/L phosphate buffer (pH 7.0) containing 100 mmol/L cytochrome c, 4 mmol/L diaminobenzidine tetrahydrochloride, and 20 g/mL catalase (All Sigma, Poole, UK) for approximately 50 minutes at 37°C in a humid chamber. Sections were washed 3 times for 5 minutes each in phosphate-buffered saline (PBS; Sigma) then incubated in 0.2 mol/L phosphate buffer (pH 7.0) containing 130 mmol/L sodium succinate, 200 mmol/L phenazine methosulfate, and 1.5 mmol/L nitroblue tetrazolium (all Sigma) for 45 to 90 minutes at 37°C until a blue color develops. Sections were then washed in PBS 3 times for 5 minutes and dehydrated through increasing ethanol concentrations and left in 100% ethanol for 10 minutes and allowed to dry. Sections that were stained on normal glass slides were mounted in Permount (Fisher Scientific, Fairlawn, NJ) and then coverslipped.
Immunohistochemistry Immunohistochemistry was performed by formalin-fixed, paraffin-embedded sections as per previously published protocols.8 Briefly, the primary antibody (mouse anti-human p53 [1:100] and -catenin [1:100];
DAKO, Ely, UK) were made up in 5% fetal calf serum (FCS) and PBS and applied to the sections for 35 minutes at room temperature then washed 3 times in PBS. The secondary rabbit anti-mouse antibody (DAKO) was also diluted in 5% FCS in PBS and applied for 30 minutes and washed as per the primary antibody. Avidin peroxidase (1:500 in PBS) was then applied for 30 minutes and washed in PBS 3 times and developed in liquid diaminobenzidine for approximately 2to 5 minutes. The sections were then dehydrated in increasing ethanol concentrations, cleared twice in xylene, and mounted in DePeX (Sigma) and coverslipped.
Mitochondrial DNA Polymerase Chain Reaction Sequencing The entire mitochondrial genome was sequenced in cells laser captured from frozen sections. To get sufficient product to sequence, a 2-round, nested polymerase chain reaction (PCR) protocol that amplified the entire mitochondrial DNA (mtDNA) in a series of overlapping fragments was used as per previous publications detailing primer sequences.14,15 Briefly, the mitochondrial genome was amplified in overlapping fragments by using a series of M13-tailed oligonucleotides. PCR products were sequenced by using BigDye3.1 terminator cycle sequencing chemistries on an ABI Prism 3730 DNA Analyzer (Applied Biosystems, Foster City CA) and compared directly with the revised Cambridge mtDNA reference sequence. Variations in the sequences of CCO-deficient cells, not present in neighboring CCO-normal cells, were considered to be clonal markers. All sequencing was repeated 3 times.
Genomic PCR Sequencing To screen for genomic DNA mutations, areas of dysplasia were identified on an H&E section, which was then traced onto the 3 preceding serial, dewaxed but unstained serial sections. Needle macrodissection of these areas was digested in 30 L PicoPure as described above. The lysate was used to screen for mutations in APC (adenamtous polyposis coli mutation cluster region), TP53 (exons 5– 8), KRAS (codons 12 and 13), CTNNB1 (-catenin, exon 11), CDKN2A (p16, exon 2), and PTEN (phosphatase and tensin homolog exons 5, 6, and 8). These genes account for approximately 85% of all somatic mutations previously reported in gastric adenocarcinomas according to the catalogue of somatic mutations in cancer (COSMIC) database (www.sanger .ac.uk/genetics/cgp/cosmic). Most mutations within the COSMIC database have been identified in cancer specimens and may be late events in the dysplasia:carcinoma sequence. After a mutation had been detected in tissue lysate, every laser-captured gland would then be screened individually for that mutation using a nested PCR protocol. Primers and PCR reaction conditions can be found in Supplementary Table1. DNA extracted
CLONAL EXPANSION IN THE HUMAN STOMACH
1253
from each gland was divided into 2 separate tubes to repeat each sequencing reaction. Furthermore, single cells were laser captured and subjected to PCR and sequencing for each gene. These were entirely unsuccessful, and, therefore, the contribution of any contaminating cells that were extraneous to each lasercaptured gland was negligible.
Loss of Heterozygosity Analysis Loss of heterozygosity (LOH) analysis was performed on individual gland lysates using a multiplexed microsatellite assay. Sixteen highly polymorphic microsatellites, located on chromosomes 3p (FHIT), 5q (APC), 9p (CDKN2A), 17p (TP53), 17q, and 18q (SMAD4), were amplified in 4 separate reactions using a multiplex PCR kit (Qiagen, Crawley, UK). Amplification of constitutional and individual gland lysate was done in parallel to control for PCR variability. Constitutional DNA was extracted from normal smooth muscle. Marker accession numbers and primer and reaction details are shown in Supplementary Table 2. Primers were tagged at the 5= end with either a FAM (tetrachloro-6-carboxyfluorescin) or HEX (hexachloro-6-carboxyfluorescin) fluorescent marker. Successfully amplified PCR products were analyzed on an ABI 3100 Genetic analyser (Applied Biosystems) using Genotyper 2.5 software (Applied Biosystems). LOH was then considered present if the area under 1 allelic peak was more than twice that of the other, after normalizing the peak areas relative to the constitutional DNA.
Mapping of Genomic Abnormalities All laser-captured glands were genotyped for the mutation discovered in the screening phase and analyzed for LOH at informative markers. Using an H&E image of the serial section to a postlaser-captured section as a guide, topographical maps of mutation spread were created with each phenotype and genotype assigned a unique color code.
Results Intestinal Metaplastic Glands Are Clonal To determine the clonal architecture of intestinal metaplastic glands, sections of snap-frozen gastric resection tissue were assayed by dual-enzyme histochemistry for CCO activity. Cells from glands that were entirely CCO deficient (blue, Figure 1A) were laser captured from the base to the surface of the gland (Figure 1B and C). Sequencing revealed the same mutation (7588G⬎A within the cytochrome c oxidase subunit II [MT-CO2] gene) in every CCO-deficient region captured, whereas sequencing cells captured along the length of a neighboring CCO-normal gland revealed an entirely wild-type
BASIC– ALIMENTARY TRACT
April 2011
1254
GUTIERREZ–GONZALEZ ET AL
GASTROENTEROLOGY Vol. 140, No. 4
Figure 1. Intestinal metaplastic glands from the human stomach are clonal. (A) CCO (brown) and SDH (blue) enzyme activity within an area of intestinal metaplasia. The blue gland is entirely deficient in CCO. (B and C) Pre- and postlaser capture microdissection sections of the same gland in (A). (D) Sequence trace from a laser-captured cell from the top of the CCO-deficient gland showing the same mtDNA mutation (7588 G⬎A transition in the MT-CO2 gene) as that from the bottom of the gland (E). (F) A neighboring CCO-normal gland shows a wild-type sequence. BASIC– ALIMENTARY TRACT
genotype (Figure 1D and E and Supplementary Figure 1). This provides conclusive evidence that intestinal metaplastic glands from the human stomach are indeed clonal. We observed a large patch of approximately 15 CCOdeficient metaplastic glands (Figure 2A), surprising because of the rarity of such CCO-deficient glands within the gastric mucosa. Sequencing the mitochondrial genome from cells taken from this large patch of CCOdeficient metaplastic glands revealed that each individual gland within the patch contained the same mutation (8503C⬎T within the mitochondrially-encoded 16S RNA (MT-RNR2) gene, Figure 2A–C). The surrounding CCOnormal glands did not contain this mutation (Figure 2D). We therefore conclude that every metaplastic gland within this patch is clonally derived from the same founder gland.
Dysplasia Can Originate From Metaplasia in the Human Stomach To investigate the origins of dysplasia within the human stomach, areas of dysplasia in each specimen were needle macrodissected and screened for mutations in genes known to be mutated in GA. In 23 patients screened, 3 (all of Japanese origin) showed a mutation common to both metaplasia and dysplasia. In patient 1, a c.4680insA mutation in APC was found in 7 of 42 metaplastic glands. Figure 3A–F shows the results of laser-capture microdissection of metaplastic, dysplastic,
and hyperplastic gastric glands from this specimen. Both the metaplastic (Figure 1A, [M]) glands adjacent to the dysplastic glands (Figure 1A, [D]) contained the c.4680insA APC mutation. Hyperplastic gastric glands (Figure 1A, [G]) were APC-wild type. This demonstrates that dysplasia was genetically related to the surrounding metaplasia. Patient 2 (c.4688insA mutation in APC) showed that 5 of 78 metaplastic glands shared the same mutation as the surrounding dysplasia and that, in patient 3 (c.643G⬎A mutation in TP53), there were 3 of 72 mutated metaplastic glands. This is summarized in Supplementary Table 3.
Whole Tissue Clonal Analysis Reveals a Founder Mutation in Fields of Dysplasia To determine how dysplasia develops and spreads in the human stomach, numerous glands of varying phenotypes were individually laser-capture microdissected and mapped according to their genotype. In patient 2, the spatial location of the c.4688insA in APC mutation in an area of mucosa displaying metaplasia and dysplasia is shown in Figure 4. Figure 4A is an H&E (in higher power, 4B), and Figure 4C shows the post-microdissected specimen (serial section to 4A) with the genotype of each gland color mapped. Although a vast majority of metaplastic glands were genetically wild type, the occasional metaplastic gland contained the APC mutation. Importantly, every dysplastic gland contained this mutation.
CLONAL EXPANSION IN THE HUMAN STOMACH
1255
Figure 2. Patches of mutated intestinal metaplastic glands arise from a founder gland. (A) Prelaser capture section of CCO-normal (brown) and CCO-deficient (blue) intestinal metaplastic gland. (B) Postlaser capture. (C) Representative sequence trace showing that each CCO-deficient gland contained the same mtDNA mutation (8503 C⬎T transition in the MT-RNR2 region). (D) Representative sequence trace showing a wild-type genotype for neighboring CCO-normal glands.
Supplementary Figure 2 demonstrates the mapping of an entire block from patient 1, showing that every dysplastic gland contained a c.4680insA in APC. Patient 3 (c.643A⬎G transition in TP53) was also mapped (data not shown). In each of these 3 patients, there was a single founder mutation throughout the dysplastic field (patient 1, 22/22 dysplastic glands; patient 2, 67/67; and patient 3, 158/158). This is compelling evidence, suggesting that the entire dysplastic lesion was derived from a single metaplastic gland. Furthermore, it is clear that this mutation was not the dysplasia-initiating event because mutated metaplastic glands with the same mutation are present. Supplementary Table 3 summarizes all data from these patients. It should be noted that, although 3 informative patients were found, the remaining 20 patients screened did not have a mutation in any of the genes that were screened. Spasmolytic polypeptide-expressing metaplasia (SPEM) has been implicated as a condition that can either give rise to IM or GA itself. To this end, we have performed immunohistochemistry for trefoil factor 2 (TFF2) in an attempt to detect SPEM. In each specimen TFF2 was detected at the base of normal pyloric glands but was generally absent from metaplastic or dysplastic glands. However, 1 specimen (patient 1, c.4680insA in APC) did show the rare intestinal metaplastic gland that expressed TFF2 (Supplementary Figure 3).
Genetic Diversity Is Subsequent to Clonal Expansion of the Founder Mutation To attempt to understand how a field of dysplasia may develop into cancer, genetic diversity was sought within fields of dysplasia by searching for further mutations and LOH within regions of genomic DNA that are associated with genes known to be mutated within gastric cancer. Figure 5A shows a strip of mucosa displaying both metaplasia and dysplasia from patient 1 (c.4680insA in APC). Figure 5B maps all glands with significant LOH of 2 microsatellite markers associated with TP53. Subclones consisting of 8 of 17 dysplastic glands in this strip of mucosa were identified by their TP53 LoH. Figure 5C shows representative LOH in both microsatellites associated with TP53. Patient 3 showed similar findings, and all 3 patients are summarized in Supplementary Table 4. Additional mutations can contribute to genetic diversity within fields of dysplasia. Figure 6 shows in patient 1 (c.4680insA APC founder mutation) isolated groups of p53-positive glands demonstrated by immunohistochemistry. When those p53-positive crypts were laser-capture microdissected, a TP53 mutation was found (c.818 G⬎A transition resulting in a p.272R⬎H amino acid change). Those neighboring dysplastic glands, which were p53negative or contained a few p53-positive cells, were TP53
BASIC– ALIMENTARY TRACT
April 2011
1256
GUTIERREZ–GONZALEZ ET AL
BASIC– ALIMENTARY TRACT
Figure 3. Intestinal metaplastic glands can be genetically related to surrounding dysplastic glands. (A) H&E of an area of gastric mucosa showing metaplastic (M), dysplastic (D), and hyperplastic gastric (G) glands. (B) Prelaser capture serial section. (C) Postlaser capture section. (D and E) Two representative sequence traces showing the presence of a c.4680insA in APC in both (D) metaplastic and (E) dysplastic glands from patient 1. (F) Hyperplastic gastric glands were APC wild type.
wild type. All other patients with dysplasia displaying founder mutations showed widespread p53 staining, making it impossible to phenotypically identify glands that contain further TP53 mutations. Further attempts to identify further genetic diversity by -catenin immunohistochemistry were unsuccessful because all specimens showed consistent membranous or frequent nuclear staining (Supplementary Figure 4). Supplementary Table 4 summarizes the genetic diversity found in the 3 informative specimens investigated.
Discussion Here, we have conclusively demonstrated that metaplastic glands within the human stomach are clonal and can clonally expand through the stomach. Previous work from this laboratory8 has questioned initial observations by Mihara et al9 that such glands were polyclonal as defined by methylation signatures of promoters of nonexpressed genes. It is known that CpG methylation does change over time16; moreover, we have shown, using partially CCO-deficient colonic crypts as a marker of clonal expansion, that CpG methylation does record the immediate ancestry of small clones within individual human colonic crypts. As these clones increase in size through the process of niche succession and monoclonal
GASTROENTEROLOGY Vol. 140, No. 4
conversion,13,17 their methylation signatures quickly diversify, and, by the time a crypt divides into 2 daughter crypts, their methylation signatures are no more similar than crypts distant to each other18 (unpublished observations). This suggests that the use of CpG methylation signatures as markers of clonal expansion is likely to be uninformative when considering the clonality of whole glands. We consider that a specific point mutation (here within the mitochondrial genome) is a reliable marker of clonal expansion and that the presence of the same mutation throughout an intestinal metaplastic gland confirms clonality. Field cancerization is the process by which a field of epithelial cells is preconditioned to allow the development of tumors.19 It is thought to expand by repeated gland fission events within the gastric mucosa.8 Gland fission, first proposed by Hattori and Fjuita,20 is the process by which a gland bifurcates at the neck/isthmus and divides into 2 distinct daughter glands. Here, we describe a large patch of CCO-deficient metaplastic glands, containing approximately 15 glands, each containing the same mtDNA mutation (and therefore derived from a founder gland). Intestinal metaplastic glands of course resemble intestinal crypts rather than gastric glands, but we have shown that clonally mutated patches arise in the human colon also through the mechanism of crypt fission.13 Individual CCO-deficient metaplastic glands in the stomach are rare when compared with those in the colon. In the normal human colon, the average patch size at the age of the patient we describe here is 2.5 crypts,13 and it is therefore interesting to observe such a large patch of mutated intestinal metaplastic glands. The reason for this is of course unknown, but it is tempting to relate it to the inflamed microenvironment,21 a concept supported by the prominent crypt fission seen in patients with active ulcerative colitis.22 We are limited in this analysis because of the rarity of CCOdeficient glands in the gastric mucosa. Evidence that metaplasia is an immediate precursor of dysplasia in the human stomach is limited. Previous studies looking at the genetic relationship between the 2 tend to focus on whole tissue DNA analysis investigating individual tumor suppressor genes.3,5,6 This study has, for the first time, attempted to identify genetic abnormalities on a gland-by-gland basis in genes accounting for approximately 85% of nuclear DNA mutations previously reported (COSMIC) in human GA. In our cohort, mutations in these genes were rarely detected strongly, suggesting that, if 2 glands shared the same mutation, they are very likely to have originated from the same gland, regardless of phenotype. Interestingly, most of the patients in this study displayed no obvious genetic relationship between metaplasia and dysplasia. These patients did not show any mutations in our panel of tumor suppressor genes in their entire and, in many cases, very extensive areas of dysplas-
CLONAL EXPANSION IN THE HUMAN STOMACH
1257
Figure 4. All dysplastic glands share a founder mutation. (A) A composite H&E of an entire strip of gastric mucosa showing intestinal metaplasia and dysplasia from patient 2 (c.4688insA in APC). (B) Higher power of highlighted areas. (C) Serial section, showing postlaser capture microdissection. The space left by the laser-capture microdissected gland has been colored to reflect the genotype. Blue filled areas are dysplastic glands positive for the c.4688insA mutation, red are positive metaplastic glands, and green are negative metaplastic glands.
tic mucosa. An explanation to why most specimens do not show this could be that, despite screening the most common genes known to be mutated in GA, there may be as yet undiscovered genes that are mutated are high frequency in GA. Methylation of tumor suppressor gene
promoters also plays an important role in carcinogenesis, and this could possibly explain the lack of mutations seen in our early gastric cancer patients. For example, in the esophagus, Eads et al have demonstrated that there is increased promoter methylation in genes such as APC
Figure 5. Genetic diversity develops subsequent to clonal expansion of the founder mutation. (A) A composite H&E of an area of gastric mucosa showing intestinal metaplasia and dysplasia in patient 1 (c.4680insA in APC). (B) Serial postlaser capture microdissection section. Mutated dysplastic gland without TP53 LOH (identified using microsatellite marker D17S1832) is distinguished by blue with green border, mutated dysplastic with TP53 LOH by blue with white border, mutated metaplastic without LOH by red with white border, and wild-type metaplastic with no LOH by green with white border. (C) Representative genotype analysis demonstrating LOH. DNA from underlying gastric muscle (constitutional DNA, top) and a mutated dysplastic gland (bottom) with clear LOH of 1 allele in each of the 2 markers associated with TP53.
BASIC– ALIMENTARY TRACT
April 2011
1258
GUTIERREZ–GONZALEZ ET AL
GASTROENTEROLOGY Vol. 140, No. 4
BASIC– ALIMENTARY TRACT
Figure 6. (A and B) P53 immunohistochemistry of an area of dysplasia showing a p53-positive dysplastic gland (*) in an otherwise p53-negative (§) specimen from patient 1 (c.4680insA APC founder mutation). (C) Pre- and postlaser capture serial section. (D) Sequencing revealing a c.818 G⬎A transition mutation in TP53 in the p53-positive gland (top) and the neighboring p53-negative gland is wild type (bottom). (E and F) Representative p53 expression is widespread in mutated dysplastic glands from the other patients.
and CDKN2A in the progression of Barrett’s esophagus to esophageal adenocarcinoma.23,24 Somatic changes in methylation may drive gastric carcinogenesis, and work is underway to address this. It is also pertinent to consider whether IM is the only pathway to dysplasia. In cases in which a mutation is shared between IM and dysplasia, there is a definite sequence of genetic events that are inherited as the disease progresses, and the origins of dysplasia are clear. We cannot discount the possibility that other, undetected, (epi-)mutations link metaplasia and dysplasia in the cases that were wild type for our panel of genes. However, IM may not be the only precursor of dysplasia, and, in this respect, there is much current interest in SPEM.25 Our study provides a precedent to perform analogous clonal analysis of concomitant SPEM, IM, and dysplastic lesions
to determine the biologic and clinical significance of SPEM plaques. Large-scale clonal expansion of dysplastic glands has previously been observed in the human gastrointestinal tract. Leedham et al demonstrated that a single mutation was able to spread through an entire lesion of ulcerative colitis-associated neoplasia.11 However, the existence of founder mutations in a lesion appears to be dependent on the specific lesion. For example, Thirlwell et al26 have recently shown that colonic adenomas from patients with familial adenomatous polyposis are polyclonal containing multiple clones with distinct patterns of APC inactivation and that, whereas sporadic adenomas frequently contain multiple clones, only one progresses to the resulting invasive carcinoma. Furthermore, genetic heterogeneity is a feature of Barrett’s esophagus and is
thought to promote tumor progression,27,28 although this has been disputed.29,30 Dysplasia within the human stomach appears to differ slightly from these examples. There is a clonal proliferation of a single mutated metaplastic gland that eventually becomes dysplastic, and this expands forming an entirely clonal dysplastic lesion. Further genetic abnormalities accrue and expand on top of the founder mutation. We know that these early mutations are not dysplasia-initiating events because, if that were the case, we would not observe any mutated metaplastic glands. We have no evidence to suggest that specific mutations would influence the rate of expansion by fission, although we would predict that, as more genetic abnormalities are acquired, that the fission index would increase. Analysis of clonal expansion is important when we consider how a premalignant lesion progresses into cancer. The greater the expansion, the greater the probability of acquiring a tumor-initiating defect. Competition between clones has been suggested to be the driving factor of the development of esophageal adenocarcinoma arising from Barrett’s esophagus.27 Such models predict that the more clones present in premalignant lesion, the more likely it will progress to cancer. Does this accord with what is seen in the human gastric mucosa? The answer is that it probably does; however, such clonal competition only arises when a substantial field of genetic heterogeneity exists. Here, we show an entire field of dysplastic glands exhibit a founder mutation. This would indicate that the lesion is clonal and competition between dysplastic glands may not be prominent but rather have a selective advantage over surrounding wild-type metaplastic glands. Over time, genetic diversity does occur with glands exhibiting LOH or additional mutations arising on the back of the founder mutation. Clonal competition could therefore become more of a driving force, and we would predict that lesions with more diversity would be more likely to develop into cancer. Clonal ordering of genetic abnormalities allows us to describe the genetic ancestry of dysplastic fields and permits the creation of phylogenetic trees (Figure 7). In this Figure, we show the time line by which each abnormality occurs in each informative patient as they progress from chronic gastritis to metaplasia to dysplasia. It is clear that a vast majority of metaplastic glands are genetically wild type with only a small percentage displaying a mutation. When a mutation does arise in a metaplastic gland, the initiation of dysplasia must occur rapidly because the proportion of mutated metaplastic to dysplastic glands is very small. Although the presence of mutated metaplastic glands indicates that the acquisition of such mutations is not a dysplasia-initiating event, an overwhelming number of mutated glands are nevertheless dysplastic. It is important to note that, although mutated cases are limited in number, we do not think that there is a specific order of mutations within metaplasia:dysplasia sequence
CLONAL EXPANSION IN THE HUMAN STOMACH
1259
Figure 7. A phylogenetic tree demonstrating linear progression of genetic defects in the development of dysplasia in the human stomach. Patients 1–3 demonstrate our working hypothesis that the dysplasia develops from a mutated metaplastic gland and that every dysplastic gland can trace its ancestry to this metaplastic gland of origin. Over time, genetic diversity can develop, and we hypothesize that the greater the extent of diversity, the more likely cancer will develop. Patient 4 is entirely wild type and accounts for the majority of patients in this study suggesting that there may be mutations in genes hitherto unreported, tumor suppressor genes silenced by methylation, or that dysplasia arises independently of the surrounding metaplasia.
and that an APC founder mutation may be as frequently detected as a TP53 founder mutation. To conclude, these data demonstrate that intestinal metaplastic glands from the human stomach are clonal and can expand by fission. When a mutation within a tumor suppressor gene occurs in an intestinal metaplastic gland, it quickly becomes dysplastic and then expands, rapidly forming a clonal field of dysplasia. Such fields promote the development of additional genetic abnormalities resulting in multiple independent subclones. This process is not apparently universal and may be one of several mechanisms behind the development of gastric carcinogenesis.
Supplementary Material Note: To access the supplementary material accompanying this article, visit the online version of Gastroenterology at www.gastrojournal.org, and at doi: 10.1053/j.gastro.2010.12.051. References 1. Compare D, Rocco A, Nardone G. Risk factors in gastric cancer. Eur Rev Med Pharmacol Sci 2010;14:302–308. 2. Uemura N, Okamoto S, Yamamoto S, et al. Helicobacter pylori infection and the development of gastric cancer. N Engl J Med 2001;345:784 –789. 3. Correa P, Shiao YH. Phenotypic and genotypic events in gastric carcinogenesis. Cancer Res 1994;54:S1941–S1943.
BASIC– ALIMENTARY TRACT
April 2011
1260
GUTIERREZ–GONZALEZ ET AL
BASIC– ALIMENTARY TRACT
4. Correa P. A human model of gastric carcinogenesis. Cancer Res 1988;48:3554 –3560. 5. Fenoglio-Preiser CM, Wang J, Stemmermann GN, et al. TP53 and gastric carcinoma: a review. Hum Mutat 2003;21:258 –270. 6. Segal F, Kaspary APB, Prolla JC, et al. p53 Protein overexpression and p53 mutation analysis in patients with intestinal metaplasia of the cardia and Barrett’s esophagus. Cancer Lett 2004;210: 213–218. 7. Gong C, Mera R, Bravo JC, et al. KRAS mutations predict progression of preneoplastic gastric lesions. Cancer Epidemiol Biomarkers Prev 1999;8:167–171. 8. McDonald SAC, Greaves LC, Gutierrez-Gonzalez L, et al. Mechanisms of field cancerization in the human stomach: the expansion and spread of mutated gastric stem cells. Gastroenterology 2008;134:500 –510. 9. Mihara M, Yoshida Y, Tsukamoto T, et al. Methylation of multiple genes in gastric glands with intestinal metaplasia: a disorder with polyclonal origins. Am J Pathol 2006;169:1643–1651. 10. de Dombal FT, Price AB, Thompson H, et al. The British Society of Gastroenterology early gastric cancer/dysplasia survey: an interim report. Gut 1990;31:115–120. 11. Leedham SJ, Graham TA, Oukrif D, et al. Clonality, founder mutations, and field cancerization in human ulcerative colitis-associated neoplasia. Gastroenterology 2009;136:542–550. 12. Leedham SJ, Preston SL, McDonald SAC, et al. Individual crypt genetic heterogeneity and the origin of metaplastic glandular epithelium in human Barrett’s oesophagus. Gut 2008;57:1041– 1048. 13. Greaves LC, Preston SL, Tadrous PJ, et al. Mitochondrial DNA mutations are established in human colonic stem cells, and mutated clones expand by crypt fission. Proc Natl Acad Sci U S A 2006;103:714 –719. 14. Taylor RW, Barron MJ, Borthwick GM, et al. Mitochondrial DNA mutations in human colonic crypt stem cells. J Clin Invest 2003; 112:1351–1360. 15. Taylor RW, Taylor GA, Durham SE, et al. The determination of complete human mitochondrial DNA sequences in single cells: implications for the study of somatic mitochondrial DNA point mutations. Nucleic Acids Res 2001;29:E74-E74. 16. Graff JR, Gabrielson E, Fujii H, et al. Methylation patterns of the E-cadherin 5= CpG island are unstable and reflect the dynamic, heterogeneous loss of E-cadherin expression during metastatic progression. J Biol Chem 2000;275:2727–2732. 17. McDonald SAC, Preston SL, Greaves LC, et al. Clonal expansion in the human gut: mitochondrial DNA mutations show us the way. Cell Cycle 2006;5:808 – 811. 18. Yatabe Y, Tavaré S, Shibata D. Investigating stem cells in human colon by using methylation patterns. Proc Natl Acad Sci U S A 2001;98:10839 –10844. 19. Slaughter DP, Southwick HW, Smejkal W. Field cancerization in oral stratified squamous epithelium; clinical implications of multicentric origin. Cancer 1953;6:963–968.
GASTROENTEROLOGY Vol. 140, No. 4
20. Hattori T, Fjuita S. Fractographic study on the growth and multiplication of the gastric gland of the hamster. The gland division cycle. Cell Tissue Res 1974;153:145–149. 21. Wong W-M, Mandir N, Goodlad RA, et al. Histogenesis of human colorectal adenomas and hyperplastic polyps: the role of cell proliferation and crypt fission. Gut 2002;50:212–217. 22. Chen R. The initiation of colon cancer in a chronic inflammatory setting. Carcinogenesis 2005;26:1513–1519. 23. Eads CA, Lord RV, Kurumboor SK, et al. Fields of aberrant CpG island hypermethylation in Barrett’s esophagus and associated adenocarcinoma. Cancer Res 2000;60:5021–5026. 24. Eads CA, Lord RV, Wickramasinghe K, et al. Epigenetic patterns in the progression of esophageal adenocarcinoma. Cancer Res 2001;61:3410 –3418. 25. Goldenring JR, Nam KT, Wang TC, et al. Spasmolytic polypeptideexpressing metaplasia and intestinal metaplasia: time for reevaluation of metaplasias and the origins of gastric cancer. Gastroenterology 2010;138:2207–2210. 26. Thirlwell C, Will OC, Domingo E, et al. Clonality assessment and clonal ordering of individual neoplastic crypts shows polyclonality of colorectal adenomas. Gastroenterology 2010;138:1441– 1454. 27. Maley CC, Galipeau PC, Finley JC, et al. Genetic clonal diversity predicts progression to esophageal adenocarcinoma. Nat Genet 2006;38:468 – 473. 28. Merlo LMF, Pepper JW, Reid BJ, et al. Cancer as an evolutionary and ecological process. Nat Rev Cancer 2006;6:924 –935. 29. Barrett MT, Sanchez CA, Prevo LJ, et al. Evolution of neoplastic cell lineages in Barrett oesophagus. Nat Genet 1999;22:106 – 109. 30. Wong DJ, Paulson TG, Prevo LJ, et al. p16(INK4a) lesions are common, early abnormalities that undergo clonal expansion in Barrett’s metaplastic epithelium. Cancer Res 2001;61:8284 – 8289.
Received September 7, 2010. Accepted December 23, 2010. Reprint requests Address requests for reprints to: Stuart A. C. McDonald, PhD, Centre for Digestive Diseases, Blizard Institute of Cell and Molecular Sciences, Barts and the London School of Medicine and Dentistry, 4 Newark Street, Whitechapel, London, United Kingdom E1 2AT. e-mail: [email protected]. Conflicts of interest The authors disclose no conflicts. Funding Supported by the Medical Research Council (to S.A.C.M.) and Cancer Research UK (to T.A.G., J.A.Z., N.A.W.).
April 2011
CLONAL EXPANSION IN THE HUMAN STOMACH
1260.e1
Supplementary Figure 1. H&E stain of a region of intestinal metaplasia associated with that shown in printed article Figure 1. Original magnification, ⫻100.
Supplementary Figure 2. A complete map of 1 block of the founder APC mutation found in patient 1 (4680insA). Glands colored dark blue are mutated and dysplastic. Glands colored red are metaplastic and mutated. Green and light blue represent wild-type metaplastic and gastric glands, respectively. Every dissected dysplastic gland contained the same mutation. Only a minority of the metaplastic glands were mutated.
1260.e2
GUTIERREZ–GONZALEZ ET AL
GASTROENTEROLOGY Vol. 140, No. 4
Supplementary Figure 3. Immunohistochemistry for trefoil factor 2. Two examples (from patient 1, APC mutated) of contiguous trefoil factor (TFF) 2-positive cells (arrows) and goblet cell-containing epithelium, suggesting the presence of spasmolytic polypeptide-expressing metaplasia or UACL (ulcer-associated cell lineage). (A) Example 1, low-power magnification, ⫻100. (B) Example 1, high-power magnification, ⫻200. (C) Example 2, low-power magnification, ⫻100. (D) Example 2, high-power magnification, ⫻200. Hybridoma supernatant mouse immunoglobulin M anti-human TFF2 generated in-house at Cancer Research UK, Histopathology Department, London, UK) was used as a primary antibody.
April 2011
CLONAL EXPANSION IN THE HUMAN STOMACH
1260.e3
Supplementary Figure 4. -Catenin immunohistochemistry in sections containing intestinal metaplasia (A) and dysplasia (B–E). In patient 2 (4668insA APC mutated) revealed frequent nuclear -catenin staining in each dysplastic gland (B and C). Patient 1 (4680insA APC mutated) showed less nuclear and more membranous -catenin expression (D and E). In each case, the expression of -catenin was consistent throughout each dysplastic gland.
1260.e4
GUTIERREZ–GONZALEZ ET AL
GASTROENTEROLOGY Vol. 140, No. 4
Supplementary Table 1. Primer Details and Reaction Conditions for Sequenced Loci Primer
Sequence 5= to 3=
p16-2A 1st F p16-2A 1st R p16-2B 1st F p16-2B 1st R p16-2A 2nd F p16-2A 2nd R p16-2B 2nd F p16-2B 2nd R K-ras 1st F K-ras 1st R K-ras 2nd F K-ras 2nd R p53-5 1st F p53-5 1st R p53-6 1st F p53-6 1st R p53-7 1st F p53-7 1st R p53-8 1st F p53-8 1st R p53-5 2nd F p53-5 2nd R p53-6 2nd F p53-6 2nd R p53-7 2nd F p53-7 2nd R p53-8 2nd F p53-8 2nd R APC-1 1st F APC-1 1st R APC-2 1st F APC-2 1st R APC-3 1st F APC-3 1st R APC-4 1st F APC-4 1st R APC-5 1st F APC-5 1st R APC-6 1st F APC-6 1st R APC-7 1st F APC-7 1st R APC-8 1st F APC-8 1st R APC-9 1st F APC-9 1st R APC-10 1st F APC-10 1st R APC-11 1st F APC-11 1st R APC-12 1st F APC-12 1st R APC-1 2nd F APC-1 2nd R APC-2 2nd F APC-2 2nd R APC-3 2nd F APC-3 2nd R APC-4 2nd F APC-4 2nd R APC-5 2nd F
GCTTCCTTTCCGTCATGC CAGGTACCGTGCGACATC CTGTTCTCTCTGGCAGGTCA TGTGCTGGAAAATGAATGCT CCTGGCTCTGACCATTCTGT CAGCTCCTCAGCCAGGTC CTTCCTGGACACGCTGGT TGGAAGCTCTCAGGGTACAAA GAGTTTGTATTAAAAGGTACTGGTGGA ATCAAAGAATGGTCCTGCAC TTTGATAGTGTATTAACCTTAT TATTAAAACAAGATTTACCTC CACTTGTGCCCTGACTTTCA GAGCAATCAGTGAGGAATCAGA AGAGACGACAGGGCTGGTT TGGAGGGCCACTGACAAC TGCTTGCCACAGGTCTCC GGTCAGAGGCAAGCAGAGG TTTTTAAATGGGACAGGTAGGA CACCCTTGGTCTCCTCCAC TCTGTCTCCTTCCTCTTCCTACA AACCAGCCCTGTCGTCTCT CAGGCCTCTGATTCCTCACT CTTAACCCCTCCTCCCAGAG CTTGGGCCTGTGTTATCTCC GTGTGCAGGGTGGCAAGT GCCTCTTGCTTCTCTTTTCC GCTTCTTGTCCTGCTTGCTT GGACAAAGCAGTAAAACCGAAC AACTACATCTTGAAAAACATATTGGA AAGTGGTCAGCCTCAAAAGG GCTATTTGCAGGGTATTAGCA GATACTCCAATATGTTTTTCAAGATG GCCTGGCTGATTCTGAAGAT CCCTGCAAATAGCAGAAATAAAA AACATGAGTGGGGTCTCCTG CAGACTGCAGGGTTCTAGTTTATC CATTCCACTGCATGGTTCAC CCAAAAGTGGTGCTCAGACA CATGGTTTGTCCAGGGCTAT TTTGAGAGTCGTTCGATTGC TCTCTTTTCAGCAGTAGGTGCTT CATGCAGTGGAATGGTAAGT GCAGCATTTACTGCAGCTT CAAGCGAGAAGTACCTAAAAA TTCTGTATAAATGGCTCATCG GGTTCTTCCAGATGCTGATA CTTGGTTTTCATTTGATTCTTT AATTAAGAATAATGCCTCCAGT TTTACGTGATGACTTTGTTGG CCAAGAGAAAGAGGCAGAAAAA TGATGGTAGAAGTTTGTACACAGG CAAGCAGTGAGAATACGTCCA TTTCTTGGTTAATAGAAGAAACTTTGC GCCACTTGCAAAGTTTCTTCT TGCTTCCTGTGTCGTCTGA CAGACGACACAGGAAGCAGA TGGAACTTCGCTCACAGGAT GATCCTGTGAGCGAAGTTCC CTGAGCACCACTTTTGGAGG AGAATCAGCCAGGCACAAAG
Reaction conditions 60/2/5 60/2/5 60/2/5 60/2/5 60/2/5 55/2/5 55/1/5 60/2/5 60/1/5 60/2/5 60/1/5 60/1/0 60/1/5 60/2/0 55/1/0 60/3/5 55/1/0 55/2/5 60/2/0 60/2/0 60/1/0 55/2/0 55/2/0 55/1/5 55/2/0 55/1/5 55/1/0 55/2/5 60/1/0 60/1/0 55/1/0
Supplementary Table 1. Continued Primer
Sequence 5= to 3=
APC-5 2nd R APC-6 2nd F APC-6 2nd R APC-7 2nd F APC-7 2nd R APC-8 2nd F APC-8 2nd R APC-9 2nd F APC-9 2nd R APC-10 2nd F APC-10 2nd R APC-11 2nd F APC-11 2nd R APC-12 2nd F APC-12 2nd R PTEN-5 1st F PTEN-5 1st R PTEN-5 2nd F PTEN-5 2nd R PTEN-6 1st F PTEN-6 1st R PTEN-6 2nd F PTEN-6 2nd R PTEN-8b 1st F PTEN-8b 1st R PTEN-8b 2nd F PTEN-8b 2nd R
GCAATCGAACGACTCTCAAA CACTATGTTCAGGAGACCCCA TGGAAGATCACTGGGGCTTA GTGAACCATGCAGTGGAATG ACTTCTCGCTTGGTTTGAGC GCTTAGGTCCACTCTCTCTCTT GGCATTATAAGCCCCAGT CCTAAAAATAAAGCACCTACTGCTG CACTCAGGCTGGATGAACAA ACATTTTGCCACGGAAAGTA GGCTGCTCTGATTCTGTTTC CAGGAAAATGACAATGGGAAT TGGCATGGCAGAAATAATACA GGACCTATTAGATGATTCAGATGATG ACTTGGTTTCCTTGCCACAG TGCAACATTTCTAAAGTTACCTACTTG GAAACCCAAAATCTGTTTTCCA TTCTGAGGTTATCTTTTTACCACA GGAAGAGGAAAGGAAAAACATC GGCTACGACCCAGTTACCAT CCTGCATAAATTTCAAATGTGG TTTTTCAATTTGGCTTCTCTTTTT TGTTCCAATACATGGAAGGATG CAAAATGTTTCACTTTTGGGTAAA CTCCTAGAATTAAACACACATCACA ACCAGGACCAGAGGAAACCT AGTCAACAACCCCCACAAAA
Reaction conditions 60/1/0 60/1/0 60/1/0 60/3/5 55/2/0 55/2/0 55/2/0 55/1.5/5 55/1.5/5 60/1.5/0 60/2.5/5 55/2.5/0 60/1.5/5
NOTE. Naming nomenclature is as follows: gene-exon, polymerase chain reaction round, primer direction. Large exons, consisting of more than 250 nucleotides, were subdivided into separate regions for amplification; a letter following the exon number denotes these regions (eg, p16-2A). For APC primers, the numbers do not refer to exons but identify regions of the mutation cluster region. Reaction condition nomenclature is annealing temperature/magnesium concentration (mmol/L)/addition of Q-solution. F, forward; R, reverse.
April 2011
CLONAL EXPANSION IN THE HUMAN STOMACH
1260.e5
Supplementary Table 2. Microsatellite Loss of Heterozygosity Analysis Primers and Reaction Details Primer Multiplex 1 D18S58 F D18S58 R D5S346 F D5S346 R D9S932 F D9S932 R D3S1300 F D3S1300 R Multiplex 2 D17S250 F D17S250 R D18S474 F D18S474 R D17S1832 F D17S1832 R D3S1313 F D3S1313 R Multiplex 3 D17S1176 F D17S1176 R D17S1678 F D17S1678 R D17S1881 F D17S1881 R Multiplex 4 D9S942 F D9S942 R D5S2001 F D5S2001 R D17S1506E F D17S1506E R D5S489 F D5S489 R
Sequence 5= to 3=
Conditions 57/0
GCTCCCGGCTGGTTTT GCAGGAAATCGCAGGAACTT ACTCACTCTAGTGATAAATCGGG AGCAGATAAGACAGTATTACTAGTT CTCCCTTTGTATTTCTGTTCTATT AAGCTATGATGGTGCCACC ACAAAGGAACGTCATGTGGTAGG GCTGTTTATTCTTCGTGGAATGCC 57/0 GGAAGAATCAAATAGACAAT GCTGGCCATATATATATTTAAACC CTCCACCCACTAGATGTCAG ACTTGCTTAAGCCTTGGACT ACGCCTTGACATAGTTGC TGTGTGACTGTTCAGCCTC TACTTTCCTTCAGATCCTTGG AACTAGGGGCCATGAATAAG 57/0 ACTTCATATACATATCACGTGC TCAATGGAGAATTACGATAGTG TTTGGGTCTTTGAACCCTTG CCACAACAAAACACCAGTGC CCCAGTTTAAGGAGTTTGGC TAGGGCAGTCAGCCTTGTG 57/5 GCAAGATTCCAAACAGTA CTCATCCTGCGGAAACCATT GCCAAGATGGTCTCGATCTC TCTGAACAGGTGATGGCAAC TGTGGGATGGGGTGAGATTTC CTGTTGGTCGGTGGGTTG GGGCTTTTGTGTTGTTTCTA GAAAACCCATAACCAGACTTG
NOTE. Reaction condition nomenclature is annealing temperate/Qsolution used in reaction. Other conditions were performed according to manufacturer’s instructions.
Supplementary Table 3. Summary of Mutation Status of All Hyperplastic Gastric, Metaplastic, and Dysplastic Glands Laser Captured From at Least 2 Blocks From Each Patient Phenotype (No. glands) Hyperplastic
Metaplastic (%)
Dysplastic
Patient
Wild type (%)
Mutated
Wild type
Mutated
Wild type
Mutated (%)
1 2 3
10 (100) 15 (100) 14 (100)
0 0 0
35 (83.3) 73 (93.6) 69 (95.8)
7 (16.6) 5 (6.4) 3 (4.2)
0 0 0
22 (100) 67 (100) 158 (100)
NOTE. Mutation: Patient 1 ⫽ c.4680 Ains APC, Patient 2 ⫽ c.4668 Ains APC, Patient 3 ⫽ c.643A⬎G TP53. Data expressed as number of glands and (percentage). All samples had a small number of mutated metaplastic glands; this became a founder mutation for the entire dysplastic field.
1260.e6
GUTIERREZ–GONZALEZ ET AL
GASTROENTEROLOGY Vol. 140, No. 4
Supplementary Table 4. Summary of Genetic Diversity in a Clonal Field of Dysplasia From All Informative Patients No. dysplastic glands Genetic abnormality TP53 LoH TP53 mutation APC LoH APC mutation
Patient 1 (%)
Patient 2 (%)
Patient 3 (%)
8/25 (32) 0 2/25 (8.0) 0
20/60 (30) 5/60 (8.3) 0 0
29/63 (46) 0 0 0
NOTE. Founder mutations: Patient 1 ⫽ c.4680 Ains APC, Patient 2 ⫽ c.4668 Ains APC, Patient 3 ⫽ c.643A⬎G TP53. Data expressed as number of glands and (percentage). Although all dysplastic crypts contain a founder mutation, genetic heterogeneity does develop over time. LoH, loss of heterozygosity.