K-ras gene mutations have been reported as early events in colorectal
tumorigenesis, but their role in tumor initiation and development is stillunclear. To analyze and compare K-ras mutational patterns betweencolorectal tissues at different stages of tumor progression in individualpatients, 65 colorectal tissue samples, including carcinoma, adenoma,
histologicallynormal mucosa,submucosalmuscularis propria, and histo.logically normal mucosa distant from tumor, were obtained from 13patients with colorectal cancer. In addition, normal mucosal tissues obtamed from four normal individuals were analyzed. Each ofthe 13 tumorswas shown previously to harbor a mutation in either codon 12 or 13 of theK-rat gene by direct sequencing. These tissues were reanalyzed, using therecently established mutant allele enrichment + denaturing gradient gel
electrophoresis method, which can detect one mutant allele in iO4—i0@normal alleles, thus allowing for the analysis of infrequent cells bearingmutations against the background of wild-type cells. No K-rat codon 12mutation was detected by this method in the histologicaHy normal mucosal
tissues sampled at the margin of resection distant from the tumor or inthose obtained from four normal individuals. On the other hand, thesemutations were detected in 9 of 10 adenoma and 6 of 10 mucosa samplesfrom 10 patients with known K-rat codon 12 mutations, and also in 2 of3 carcinoma, 2 of 3 adenoma, and 1 of 3 mucosa samples obtained from 3patients with known K-ms codon 13 mutations. Thus, K-ms codon 12mutations were found to occur with a high frequency (53.8%) in histologically normal mucosa adjacent to tumors of patients with K-rat mutation-positive colorectal cancer, suggesting that they may be useful biomarkers for early detection of colorectal cancer. Furthermore, multiple K-ratmutations were found in tissues ofnearly halfofthe 13 patients, indicatingthat distinct evolutionary subclones may be involved in the development oftumor in some patients with colorectal cancer.
INTRODUCTION
It has been widely accepted that progressive accumulation of mutations underlies the adenoma-to-carcinoma sequence in both FAP3and sporadic coborectal cancers. These neoplasms are thought to occurthrough cbonal evolution of a single cell initiated by a somatic mutation. Additional mutations accumulate in a subset of daughter cellsthat subsequently acquire a growth advantage. Further proliferation ofa clone ultimately results in the development of a macroscopic adenoma or polyp (1—3).Previous studies showed that an inherited defectin the APC (adenomatous polyposis coli) gene was the first necessarystep for coborectal polyp formation and the development of coloncancer in FAP patients (4). A similar defect in the APC gene was alsodetected as one of the earliest steps in many cases of sporadic
Received 6/24/96; accepted 4/15/97.The costs of publication of this article were defrayed in part by the payment of page
charges. This article must therefore be hereby marked advertisement in accordance with18 U.S.C. Section 1734 solely to indicate this fact.
I Supported by NIH Contract CN-l5393—02 (Early Detection Research Network) and
American Cancer Society Institutional Research Grant IRG-58-32.2Towhomrequestsforreprintsshouldbeaddressed,atDepartmentofEnvironmental
and Occupational Health, Graduate School of Public Health, University of Pittsburgh, 260Kappa Drive, Pittsburgh, PA 15238. Phone: (412) 967-6526; Fax: (412) 624-1020.
3 The abbreviations used are: FAP, familial adenomatous polyposis; ACF, aberrant
colorectal cancer (5). Mutations in the K-ras gene, found in more thanhalf of all colorectal tumors examined, were thought initially to occurafter inactivation of the APC gene and to represent a late event in theprogression of an adenomatous polyp toward malignancy (3, 6—8).Recently, K-rat mutations were identified in ACF and even in microscopically normal mucosa of patients with coborectal cancer, using
sensitive detection techniques (9—13). These observations suggest that
K-ras mutations might represent an early genetic event, which precedes the appearance of histologically detectable aberrations in coIonic epithelial cells.
Nevertheless, the biological significance of these early gene-mutational events in coborectal tumorigenesis remains unclear. Theoretically, if clonal expansion leading to carcinogenesis is derived from acobonic epithelial cell with pre-existing K-rat mutation(s), its frequency is expected to increase with the progression from the premalignant to the malignant state. Furthermore, exposure to environmental mutagens can cause multiple mutations of a single cancer-relatedgene (11), each of which may be involved in clonal expansion. Thus,it is possible that more than one mutation originating from differentevolutionary subclones may be detected with different frequencies ina single tissue sample from at least some, if not most, patients withcoborectal cancer. An informative study of the mutational patterns inthe K-ras gene should involve the types and frequencies of mutationsin colorectal tissues at different stages of tumorigenesis. This approach requires a sensitive and reliable technique capable of detectingmutation(s) in a single cell present among a large excess of nonmutated cells.
A variety of assays has been used to detect mutations in ras genes(8, 14, 15), including an enriched PCR, which is based on theintroduction of a restriction enzyme site at the codon of interest bymismatch PCR and RFLP analysis (14, 15). After elimination of thewild-type copies by restriction enzyme digestion, the assay allows forfurther mutant amplification and digestion, providing a sensitive approach to detecting ras gene mutations. To extend the application ofthis method to analyze multiple mutations, we have established aMAE+DGGE method (16, 17) by combining the enriched PCRtechnique with DGGE (18). This method allows the separation ofDNA fragments differing in melting properties caused by the presenceof a point mutation in their sequences, as a characteristic and, therefore, a recognizable pattern of homoduplex and heteroduplex bandsformed by the wild-type and mutant DNA (19).
In this study, we applied the MAE+DGGE method to analyzeK-rat codon 12 mutations in 65 colorectab tissue samples, at variousstages of tumorigenesis, obtained from patients known to harbormutations at either the 12th or the 13th codon of the K-ras gene in thecoborectal carcinoma and/or adjacent residual adenoma using theconventional PCR and DNA sequencing method (20). These samples,including carcinoma, adenoma, histologically normal mucosa adjacent to tumor and that taken from the margin of resection, andsubmucosal muscularis propria, were topographically microdissectedfrom paraffin-embedded tissues obtained from 13 patients with cobrectal cancer. The frequent occurrence of K-ras mutations in thesesamples may reflect early events in the evolution of specific colon
2485
K-ras Gene Mutations in Normal Colorectal Tissues from K-ras Mutation-positiveColorectal Cancer Patients'
Dan Zhu, Phouthone Keohavong,2 Sydney D. FinkeLstein, Patricia Swaisky, Anke Bakker, Joel Weissfeld,Sudhir Srivastava, and Theresa L. Whiteside
Departments of Environmental and Occupational Health (D. Z., P. K.J, Pathology (S. D. F., P. S., A. B.. T. L WI. and Epidemiolog@ If. W.J. and University of Pittsburgh Cancerinstitute (P. K., J. W, T. L W.], University of Pittsburgh, Pittsburgh, Pennsylvania 15238: and NIH, Bethesda, Maryland 20892 15. S.]
cancers or may be a sensitive barometer for the exposure and/or
sensitivityof thehumancolonicmucosato carcinogen.
MATERIALS AND METHODS
Tissue Specimens and PCR/Direct Sequencing of Microdissected Sam
plea. Surgical biopsy specimens from 13 patients with colorectal carcinomawere selected from the archives of the University of Pittsburgh Medical Center.Histological slides were reviewed, and the diagnosis of colorectal carcinomawas confirmed using established morphological criteria. Selected paraffinblocks representative of normal mucosa and submucosal muscularis propria,adenoma, or carcinoma were sectioned at a 4-,.tm thickness. Using lightmicroscopic features, tissues were removed from histological sections under
stereomicroscopic observation from multiple sites (less than 3 mm in diameter)corresponding to carcinoma, adenoma, or histologically normal mucosa adjacent to tumor. Histologically normal mucosa were also sampled at the margin
of resection at a distance of at least 5 cm from tumor. Post-topographyhistological sections were stained with H&E to confirm the accuracy of thesample selection. Initially, the tumor samples were analyzed for the presenceof K-ras exon I mutations, using PCR and a direct sequencing method (20). Agroup of these tumors yielded 10cases characterized by the presence of codon12, and 3 cases with codon 13 mutations. Using this subset of K-ms-mutated
specimens, tissue samples representative of carcinoma, adenoma, mucosa
adjacent to tumor, mucosa distant from tumor, or submucosal tissues were thenanalyzed similarly by PCR/direct sequencing for K-ras mutations. DNA obtamed from the same samples was used for subsequent MAE+DGGE analysis.The samples were coded and analyzed blindly. The code was broken after theMAE+DGGE analyses were completed. Negative controls (neonatal intestinaltissue; peripheral blood lymphocytes obtained from healthy volunteers; and
submucosal muscularis propria microdissected from slides along with the
mucosa, adenoma, and tumor tissues) and positive controls (known K-ras exon
1 mutations from colorectal cancer tissues) were included at all stages ofhistological and molecular analyses. In addition, histologically normal mucosa
obtained from four normal individuals were analyzed. These individuals wereselected from computerized pathology archives at University of PittsburghMedical Center. Each individual, ranging in age from 21 to 65, had sustaineda traumatic injury to normal colon, and none had history of cancer. Theabsence of cancerous or precancerous lesions was confirmed by histologicalexamination.
The MAE+DGGE Method. This methodwas establishedrecentlyfor thedetection of infrequent K-ras gene codon I2 mutations in human lung tissuesadjacent to the tumor (16) and in primary lung tumors (17). Briefly, for DNAamplification, each microdissected tissue was first deparaffinized with xylene
and ethanol, dried, incubated in the PCR reaction mixture in the presence ofproteinase K, and then boiled for 10 mm before being subjected to two sets ofPCR. The first PCR was performed in a 50-pJ reaction mixture containingDNA from 600-1000 cells and 0.2 @,tMof each primer (KII -1, 5'-TATTATAAGGCCTGCTGAAA-3' and the mismatched primer, PKB, 5'-AGGCACTCTTGCCTACGGCA-3'), 60 @Meach deoxynucleoside triphosphate,10 mMTris-HCI (pH 8.3), 50 msi KCI, 1.5 msi MgCI,, 0.01% (w/v) gelatin,and 2.5 units of GeneAmp Taq DNA polymerase (Perkin-Elmer Corp.,Branchburgh, NJ). The amplification was performed for 10 cycles (1 mm at94°C,1 mm at 54°C,and 2 mm at 72°C).Subsequently, the PCR product waspurified using a QlAquick PCR purification kit (Qiagen, Inc., Chatsworth, CA)and digested in a 50-@zlvolume with 10 units BanI restriction endonuclease(New England Biolabs, Beverly, MA) at 37°Cfor 2 h. One-fifth of thedigestion material was used as template in the second PCR for 30 cycles in asimilar reaction system, except that 4 pCi of [a-32P]dCTP were added, and the
primer KIl- 1 was replaced by a GC-Clamp primer (PKGC): 5'-GCCGCCTGCAGCCCGCGCCCCCCGTGCCCCCGCCCCGCCGCCGGCCCGGCCGCCTATAAGGCCTGCTGAAAATG-3'. After amplification, a 25-id aliquotfrom each PCR product was diluted to 250 @slwith BanI buffer and digestedwith 30 units BanI restriction enzyme at 37°C for 2 h. The digestion material
was recovered by ethanol precipitation, dissolved in 10 mMTris-HC1(pH 8.3)and I mM EDTA, and electrophoresed through a 10% polyacrylamide gel
(bisacrylamide:acrylamide, 1:19). The gel was autoradiographed. The position
of the DNA fragments in the gel was located by superimposing the autoradiogram on the gel. The portion of the gel containing the digestion-resistant
fragment was excised from the gel and migrated through a 12.5% polyacryl
amide(bisacrylamide:acrylamide,1:37.5)containinga 30—45%lineargradient of denaturant concentration [100% = 40% urea (wlv) + 40% formamide(v/v)]. The gel was dried and autoradiographed. Mutant alleles present in thegel were isolated and characterized further by a second DGGE and sequencinganalysis (19).
DNA Fragment Quantitation. Electrophoresisbands of DNA fragmentsvisualized by autoradiography were scanned and quantified using a densito
meter-computer system, model 300A (Molecular Dynamics, Sunnyvale, CA).
RESULTS
We identified K-rat codon 12 or 13 mutations in carcinoma and/oradenoma samples from a group of 13 patients with polyp-formingcoborectal cancer, using the direct DNA sequencing method (20).Sixty-five tissue samples, including carcinoma, adenoma, histobogically normal mucosa adjacent to tumor and that taken from the marginof resection, and submucosal muscularis propria, were microdissectedtopographically from paraffin-embedded tissues obtained from thesesame patients and analyzed using the more sensitive MAE+DGGEmethod (16). A representative example of multiple-target site sampling from a case of colorectal carcinoma included in this study isshown in Fig. 1. Tissue samples taken immediately outside the tumorborders were carefully reviewed histologically to confirm the absenceof histologically recognizabletumorscells. Given the 4-p@mthicknessof the microscopic sections, it is considered unlikely that tumor cellswere present but not observed. Control negative and positive tissues
Fig. 1. An example of tissue site sampling from paraffin-embedded tissue sections inone case of colorectal cancer (case 8; Table I). Topographic sampling of normal mucosa,adenoma, and carcinoma from histological sections is illustrated. Histopathological features of mucosa (N), adenoma (7), and carcinoma (Mi) were noted, and these tissue siteswere microdissected. Top, the accuracy of sampling for molecular analysis is confirmedby staining post-topographic tissue sections. Bottom. premicrodissection tissue section(H&E; original magnification, X2).
Fig. 2. An example of analysis of K-rat codon 12 mutations by MAE+DGGE. Tissue areas of mucosa (Mu), adenoma (Ad), and carcinoma (Ca) were obtained from patient case10 (Table 1). Wild-type DNA (Lanes Wu and Wc) obtained from normal peripheral blood lymphocytes was used as a negative control. In the MAE step, all DNA samples were subjectedto the two PCRs and two Ban! endonuclease digestions, as described in “Materialsand Methods.―The digestion-resistant fragment was analyzed by gel electrophoresis. Each lane showsa 126-bpfragment(digestion-resistantfragment)and/ora 107-bpfragment(cleavedfragmentcontainingwild-typeK-rascodon12).LaneWushowsthepositionofthe 126-bpwild-typefragment without treatment with BanI enzyme. After Ban! treatment (Lane Wc), the 126-bp wild-type fragment is cleaved into a 107-bp fragment and a l9-bp fragment (migrated outof the gel). A longer exposure of the gel revealed the 126-bp fragment as a very faint band in Lane Wc (not shown). In contrast, in Lanes Mu, Ad, and Ca. a significant fraction ofthe undigested 126-bp fragment is detectable, suggesting the presence of one or more K-rat codon 12 mutant alleles. In the DGGE step, each digestion-resistant l26-bp fragment fromthe MAE step was excited from the gel and separated by DGGE. Lane Wu shows a single wild-type band (G. glycine). The digestion-resistant fragment from the digested negativecontrol (Lane Wc) shows no major detectable band, whereas those from samples Mu. Ad. and Ca each show a specific pattem of bands, which were purified from the gel and sequenced.In case 10, a total of eight bands is detectable in the mucosa sample (each indicated by a line in Lane Mu). They corresponded to 3 K-rat codon I2 mutants, including an aspartate(D, GAT), a acme (S. AGT), and a valine (V. G'fl'). The pairs of bands, Dl-D2, 51-52, and Vl-V2, each corresponded to the two respective mutant/wild-type heteroduplexes formedbetweeneachof themutantfragmentscorrespondingto D, 5, or V.andthewild-typefragment.Theaspartatemutationin carcinoma(LaneCa)wasalsodetectedin adenoma(&.ineAd) and mucosa (Lane Mu), with an increasing intensity of the mutational signal from mucosa to carcinoma. The mutant homoduplexes corresponding to D and V were focused at thesame position of the denaturing gradient gel.
for K-ras mutations were processed histologically at the same time asthe patient samples, including topographic selection. After microdissection, the tissue sections were stained with H&E and compared tothe original histological sections to confirm the accuracy of tissue siteselection.
In the MAE+DGGE method (16), a downstream mismatchedprimer (PKB) is used to generate a BanI restriction endonucbeaserecognition site at the wild-type codon 12 (GGT) of the K-ras genein the PCR-amplified fragment (see “Materialsand Methods―).This codon can be mutated in either one or both of the twoguanines. Thereby, the large excess of the fragment with a wildtype K-rat codon 12 can be eliminated by digestion with BanIenzyme. The MAE step consists of two rounds of PCR digestion.The first PCR yielded a 75-bp product, which is subjected todigestion with BanI enzyme. Fragments with wild-type K-rascodon 12 are expected to be cleaved into 56- and 19-bp fragments.The remaining digestion-resistant fragments are amplified furtherby a second PCR using primers PKB and PKGC (GC-rich primer)to construct a high-temperature melting domain on the upstreampart of the fragment, yielding a 126-bp product. The batter isdigested further with Ban! enzyme, cleaving each fragment con
taming wild-type codon 12 into a 107- and a l9-bp fragment. After
elimination of the excess of wild-type product by gel electrophoresit separation, the digestion-resistant I 26-bp fragment is then
analyzed by DGGE (Fig. 2). Previously, we have evaluated thelevel of sensitivity of this method by carrying out reconstructionexperiments and found that this method allows for detection of onemutant allele among l0@—l0@normal alleles (16). This level ofsensitivity should allow us to easily detect one mutant allele amongthe 600-1000 cells (or mutant frequency of 1.6—1.0 X l0@)present in each tissue sample collected by topographic microdissection.
The results of K-ras mutation analysis in coborectal tissues, obtamed from 13 patients with colorectal cancer and from four normalindividuals without colorectal cancer, by either the direct sequencingor the MAE+DGGE method are listed in Table 1. Among the 65tissue samples obtained from the 13 patients, the direct sequencingmethod identified a K-rat codon 12 mutation in 15 samples, including10 carcinoma samples (all except those from cases 2, 5, and 9, whicheach contained a K-ras codon 13 mutation) and 5 adenoma samples(cases 1, 7, 8, 11, and 13). These mutations were all identifiedaccurately by the MAE+DGGE method (Table I). Thus, the
Table 1 Patient samples and K-ras gene mutations detected by MAE+DGGE and direct sequencing'@
Mutation Types―
CaseSexAgeTNMLesion1M59100
K.ras GENE MUTATIONS IN COLORECFAL CANCER
MAE + DGGE(codon 12)
DD
D, S@D, S
V
ASS
CD, S
V
AA
D, A, S
VV
D, V
SDS
DD, S
D, 5, V
A, DA, D
SS5,D
D
D
Dir/Seqc
12DI2D
-(l3Df-(13D)
12V
12A
—(13D)—(13D)
12C
12A12A
12V12V
-(13D)
l2D
12A12A
125
12Dl2D
Carcinoma
AdenomaMucosaDistant mucosa@'Muscularis propria
M 75 1 0 0 CarcinomaAdenomaMucosaDistant mucosa'@Muscularispropria
M 71 2 0 0 CarcinomaAdenoma
MucosaDistant mucosa'@Muscularis propria
F 63 2 0 0 CarcinomaAdenomaMucosaDistant mucosa'@Muscularit propria
F 64 3 0 0 CarcinomaAdenomaMucosaDistant mucosa'@Muscularis propria
M 57 3 1 0 CarcinomaAdenomaMucosaDistant mucovd'Muscularit propria
F 81 2 0 0 CarcinomaAdenomaMucosaDistant mucosa'@Muscularis propria
M 75 2 0 0 CarcinomaAdenomaMucosaDistant mucosa'@Muscularis propria
M 66 1 0 0 CarcinomaAdenomaMucosaDistant mucosa'@Muscularis propria
F 55 3 0 0 CarcinomaAdenomaMucosaDistant mucosa'@Muscularis propria
F 49 3 1 0 CarcinomaAdenomaMucosaDistant mucosa'@Muscularispropria
F 60 2 0 0 CarcinomaAdenomaMucosaDistant mucosa'@Muscularispropria
M 81 2 0 0 CarcinomaAdenomaMucosaDistant mucosa'@Muscularis propriaMucosaMucosaMucosaMucosa
a M, male; F, female; T, N, and M, clinical staging.
b A (OCr), C (TOT), D (GAT), S (AGT), and V (Gil'), amino acids of mutated codons; —, negative for mutated codons.CDir/Seq:analysisby directsequencing;12,13:mutatedcodonNo.d Distant mucosa corresponded to histologically normal mucosal tissue obtained from the margin of resection distant at least 5 cm from tumor.S The tissues analyzed corresponded to histologically normal mucosa from normal individuals without colorectal cancer.
e Multiple codon 12 mutations listed in an order from higher to lower relative frequencies.
I _ , negative at codon 12; (13D), positive at codon 13.
Fig. 3. Analysis of tissues obtained from patient case 11 by MAE+DGGE. For this patient, DGGE analysis showed (left) a pattern of seven bands (designated bands 1—7)in bothadenoma (Ad) and carcinoma (Ca). No mutation patterns were detected in the corresponding mucosa (Mu) and the negative control, tubmucosal muscularis propria (Su). For furthercharacterization, DNA in each band was purified from the gel, further amplified, and analyzed by a second DGGE (right). The number and kind of the mutation(s) were determinedby examining the DCIGE mutant patterns and by sequencing the mutant homoduplex DNA(s) appearing in each lane. Lane numbers (right) correspond to band numbers (left). Analysisof these bands showed either one mutant homoduplex, e.g., an alanine (A) in bands 4 and 7 and an aspartate (D) in bands 2, 3, and 6, or both A and D mutants in bands 1 and 5. Inaddition, a series of heteroduplexes appeared. They were formed between the wild-type and either mutant A (A1 and A2: band 4) or mutant D (DI and D2: bands 1—3),and betweenmutants A and D [(A+D)I and (A+D)2; bands I and 5]. DNA purified from each of bands 6 and 7 gave rise to a single band, indicating that they corresponded each to a homoduplexmutant (D and A, respectively). Lane Wu shows an undigested wild-type DNA (G) used as control.
MAE+DGGEmethodconfirmedwithouterrorsthe presenceof eachK-rat codon 12 mutation known to be present. In addition, in case 11,the MAE+DGGE method detected two K-rat mutations in both thecarcinoma and adenoma tissues (Fig. 3), only one of which had beenidentified previously by direct sequencing.
Importantly, among the 50 samples, which had been found previously to be negative for K-rat codon 12 mutations by direct sequencing, the MAE+DGGE method detected these mutations in 15 samples(30%; cases 2, 4, 6—10,and 12; Table 1). Interestingly, 8 of these 15samples (53%) showed multiple K-rat codon 12 mutations (cases 2,6—10,12; Table 1; Fig. 4), whereas the other 7 samples each containeda single K-rat codon 12 mutation. The samples from patient cases 2and 9 presented bow frequency of K-rat codon 12 mutations inaddition to the K-rat codon 13 mutations detected by the directsequencing method. Among the 10 samples with multiple K-rat
Fig. 4. Analysis of K-ms mutations in tissues from patientcases 6 and 9 by MAE+DGGE. Case 6 contained in K-rascodon 12 a valine (V) in the mucosa (Mu), a serine (5) andaspartate (D) in the adenoma (Ad), and a cysteine (C) in thecarcinoma (Ca). The cysteine was not detected in the adenoma or mucosa, which instead harbored other mutationtypes. Case 9 had been found previously to contain a K-rascodon 13 mutation in the carcinoma using the direct sequencing method. The MAE+DGGE method identified additionalK-ms codon 12 mutations, including a serine (S) in both themucosa and carcinoma and an aspartate (D) in the adenoma.Lane Wc corresponds to DNA containing wild-type K-ratcodon 12 used as control.
1v D,s C
Case 6
mutations (cases 2 and 6—12),only 3 (cases 2, 9, and I 1) werecarcinoma. These multiple mutations were, therefore, more frequentin adenoma or mucosa than in carcinoma.
A possible relationship was observed between the presence of K-ratcodon 12 mutations and the histology of the tissues. For instance,among the 50 tissue samples found to be negative for codon 12mutations by direct sequencing, the MAE+DGGE method detectedcodon 12 mutations in 2 of 3 carcinoma (cases 2 and 9; Table 1), 6 of8 adenoma (cases 2, 4, 6, 9, 10, and 12), and 7 of 13 mucosa cases(cases 4, 6—10,and 12). Overall, the MAE+DGGE method detectedK-ras codon 12 mutations in 11 of 13 adenoma (84.6%), in 7 of 13mucosa (53.8%), and in none of the normal mucosa taken from themargin of resection or the submucosal muscularis propria (Table 1).This result indicated that K-ras codon 12 mutations were more frequently detected in adenoma than in histologically normal mucosa
mutant frequency in each tissue sample was estimated to be 1.6—1.0 X l0@ (one mutant cell in 600-1000 normal cells), which was atleast 10 times higher than the limit of detection (one mutant cell inlO@—l0@normal cells) of the MAE+DGGE method, as demonstratedpreviously by our reconstruction experiments (16). For these reasons,we observed reproducibly that the background AGT and GAT mutations induced by Taq pobymerase in the negative controls were absentor, after prolonged autoradiography, appeared as much fainter bandscompared with those found in the samples considered to be positivefor these mutations (see Lane Wc in Figs. 2 and 4, and Lanes Su andMu in Fig. 3). Therefore, it appears unlikely that the multiple mutations found in our samples, especially AGT and/or GAT mutations(Table 1), were Taq-induced artifacts.
DISCUSSION
Abundant clinical and histological data suggest that colorectalcarcinomas arise from pre-existing benign tumors (adenoma) (20).The adenomatous elements are often found to be replaced by carcinoma at the time of surgery. However, in one-third of all surgicalspecimens, coborectal carcinoma may contain regions of both carcinoma and residual adenoma (21), providing an excellent system inwhich to search for and study the multistage genetic alterationsinvolved in the process of coborectal tumorigenesis (6, 7). The availability in these surgical specimens of histologically normal colonicmucosa adjacent to the pathologically identifiable lesion facilitatessuch studies. Nevertheless, the detection of a small proportion ofabnormal (mutated) cells in histologically normal mucosa requires ahighly sensitive and specific methodology. We have developed andapplied a novel technique, combining MAE and DGGE, to in situanalysis of K-rat mutations in cobonic mucosa obtained from patientswith coborectal cancer, whose tumors were shown previously to harbor mutations in codon 12 or 13 of the K-ras gene by direct sequencing.
Using the MAE+DGGE method, K-ras codon 12 mutations weredetected in most of the adenoma and more than half of the tumoradjacent mucosa obtained from 13 patients. Our findings, along withprevious reports (1, 3, 6, 7, 11—13),show that K-rat codon 12mutations are common events not only in coborectal tumors but also inpremalignant coborectab tissues obtained from patients with colorectalcancer. In addition, in five patients (cases 7, 8, 10, 12, and 13; Table1), the same mutation was found in the mucosa, adenoma, and/orcarcinoma, with a stepwise increase in the mutation frequency frommucosa to carcinoma evident in cases 7, 10, and 12 (Table 1; Fig. 2).Therefore, through comparison of the K-rat mutational pauerns incolorectal tissues at different stages of tumor development (normal —*adenoma —pcarcinoma) in each patient, we found that tumors from asignificant fraction of the patients analyzed were monoclonal, consistent with a previous report (2).
Although the frequency of K-rat mutation was found to be generally higher in carcinoma than in adenoma, in case 1, the samemutation (GGT to GAT) was detected at a lower frequency in carcinoma than in adenoma (Table 1; Fig. 5). One might speculate that theadenoma was heterogeneous with respect to the population of cellswith or without K-rat mutations, as reported previously (22). Additional mutation(s) might occur in an adenoma cell without a K-ratmutation, which thereafter acquires a growth advantage, expandscbonally, and progresses to carcinoma. In this case, residual adenomacells may be responsible for the much lower frequency of K-rasmutations observed in the carcinoma than adenoma cells. Alternatively, the low frequency of GGT-to-GAT mutation, which had notbeen involved in the initiation and early tumor progression, occurredin a subset of the carcinoma cells coincidentally just prior to surgery.
2490
Ad Ca Mu
@,MAE uu@—1O7bp
@—
I I
Cas.1Fig. 5. Gel electrophoresis analysis of codon 12 mutations in patient case 1 using the
MAE method. Each lane showed the digestion-resistant fragment (126 bp) and thedigested fragment (107 bp). The fraction of the mutation identified (an aspartate, OAT)was lower in the carcinoma (Lane Ca) than in the corresponding adenoma (Lane Ad). Themucosa (Lane Mu) showed no detectable digestion-resistant fragment.
adjacent to tumor among the 13 patients selected for this study basedon the presence of a K-rat mutation in tumor cells. The difference(84.6 versus 53.8%) reached a marginal statistical significance(P = 0.06 by exact test for paired data).
The presence of the same K-ras codon 12 mutations in both thecarcinoma and the adenoma and/or mucosa was observed in S of the13 patients (cases 7, 8, 10, 12 and 13; Table 1). However, in patients4, 6, and 9 (Table I ; Fig. 4), the carcinoma contained a K-rat codon
12 mutation different from those found in the corresponding adenomaand/or mucosa. Although the frequency of K-ras mutations was generally found to be higher in the carcinoma than in the adenoma, thereare some exceptions. For instance, in case 1, the aspartate mutationwas detected in both the carcinoma and adenoma but appeared at alower frequency in the carcinoma than in the adenoma (Table 1; Fig.5).
The results of MAE+DGGE analysis of histologically normalmucosa from the margin of resection in the 13 patients and that fromfour individuals without coborectal cancer were identical to thosefound in the submucosal muscularis propria, neonatal intestinal tissues, and peripheral blood lymphocytes from healthy volunteers.These data demonstrated the absence of K-rat codon 12 mutations inboth the submucosal muscularis propria from coborectal cancer patients and the mucosa from normal individuals without cancer (seeTable 1). The absence of mutations in these normal individualsconfirms the validity of our technique. The data also showed theabsence of K-ras codon 12 mutations in the mucosa obtained from themargin of resection in these patients, although these mutations wereidentified in the corresponding mucosa adjacent to tumor in sevenpatients (patients 4, 6—10,and 12; Table 1), suggesting that themutations may be generated from the exposure of these patients tocolon carcinogens.
In our previous studies (16, 17), the AGT and GAT mutations werefound to be also induced, but at very bow frequencies, by Taq DNApolymerase during DNA amplification of sequences with wild-typeK-ras gene codon 12 (GGT) present in barge excess in the originalamplification reaction. For this reason, it was important to ensure thatthe K-rat mutations detected in the present study by theMAE+DGGE method actually pre-existed in the tissue and were notderived from either the contamination or misincorporation by TaqDNA polymerase during PCR. Although all necessary precautions hadbeen taken to avoid such potential problems, we reexamined theoriginal tissue stock sample for each case (data not shown). Thereproducible mutation patterns and the negative results seen consistently in the negative controls, including the submucosal muscularispropria tissues microdissected from slides along with the mucosa,adenoma, and carcinoma tissues, analyzed side by side confirmed theaccuracy of the mutations listed in Table 1. Furthermore, the lowest
Both of these explanations allow for the coexistence of differentevolutionary subclones in an individual tumor.
Additional evidence for the hypothesis that multiple subclones maybe involvedin the coborectaltumorigenesisin tome patientscomesfrom the following observations. In two patients (cases 4 and 6; Table1), the codon 12 mutations present in the carcinoma were differentfrom those detected in the corresponding adenoma and mucosa. Theseobservations are in agreement with the recent identification of K-ratcodon 12 mutations in human coborectal ACF and normal-lookingcells in mucosa in patients with K-rat mutation-negative colorectalcancer(11, 12).Thesestudiesshowedthatmutationsfoundin theACF or mucosa may be different from those in the correspondingcarcinoma (1 1—13).An alternative hypothesis for these observations isthat multiple, independent K-rat mutations occur frequently in thecolonic mucosa of some individuals, and only some of these K-rat
mutated cells progress to cobonic neoplasms.Multiple mutations of a single gene in an individual tumor have
been reported in a variety of cancers, including acute myeloid leukemia (N-ras codon 12; Ref. 23), acute lymphoblastic leukemia (N-rat;Ref. 24), cholangiocarcinoma (K-ras codon 12; Ref. 14), hepatocellular carcinoma, and nonmalignant liver tissue from a patient withhepatocellubar carcinoma (p53 codon 249; Ref. 25). The individualmutations seemed to be present in distinct cell populations (23). In thepresent study, more than one mutation of K-ras in codon 12 was foundin two carcinoma, four adenoma, and four mucosa samples, andcoexistence of mutations at both codons 12 and 13 was detected intwo carcinoma, two adenoma, and one mucosa sample, obtained from13 patients (Table 1, Figs. 3 and 4). Notably, in case 11, two mutations, GGT to GCT and GGT to GAT, were detected in both carcinoma and adenoma (Fig. 3), reflecting the coexistence of two evolutionary subclones. Surprisingly, the relative frequencies of the twodistinct mutants were similar in carcinoma and adenoma, suggestingthat the proliferating rates in these two individual subclones mayremain similar during the tumor progression. This hypothesis is supported further by the similarity between the frequencies of the samemutation in adenoma and carcinoma in patients 7, 8, and 13 (Table 1).A well-known fact that progression to carcinoma is associated withincreased capability of invasion and poor differentiation (21) ratherthan changes in cell kinetics (26) is also consistent with the hypothesis. It has also been reported that in some colorectal cancer cases, amutation present in primary tumor was not detected in its metastasisor that a mutation in metastasis was absent in the primary tumor (27).Although mutations might occur after the tumor cells had metastasized, the possibility that multisubclonal composition pre-existedwithin the tumor prior to metastasis could not be excluded (27).
Our study of histologically normal mucosal tissues from the marginof resection did not show any K-ras codon 12 mutation among thepatients analyzed, although they all harbored these mutations in thecarcinoma, adenoma and/or tumor-adjacent histologically normal mucosa. These data suggest that mutations found in the tumor-adjacenthistologically normal mucosa in these patients may reflect the exposure to colon carcinogen(s) rather than resulting from pre-existingmutations.
In summary, we have found a high-frequency K-rat codon 12mutation in premalignant tissues from patients with colorectal cancer.Although K-ras gene mutations detectable in histologically normalmucosa may not be sufficient to provide a neoplastic growth advantage, it has been suggested that these mutations may place the host cellat high risk of cancer development (13). Therefore, these mutationsmay provide useful biomarkers for identifying individuals at a highrisk for developing coborectal cancer. Furthermore, we have found amultisubclonal composition in colorectal tissues from nearly half ofthe 13 patients harboring K-rat mutation in their coborectal cancer.
This finding is consistent with a recent report showing a polycbonaborigin of colonic adenomas in a FAP patient (28). Our results, alongwith previous findings, suggest that more than one evolutionarysubclone is involved in the coborectal tumorigenesis, at beast in somecases of coborectal cancer. These subcbones are present at a bowfrequency and, therefore, may not be detectable with conventionaltechniques. The application of more sensitive methods, such asMAE+DGGE, is necessary to confirm the presence of multiple orinfrequent K-rat mutations in patients with coborectal cancer. Finally,additional studies of normal mucosal tissue sampled at a distance fromboth K-rat-mutated and K-rat normal cancers involving a larger
number of individuals will be required to confirm our data, suggestingthat the population of K-rat-mutant cells determined in histologicallynormal mucosa may be a dosimeter of carcinogen exposure in theseindividuals.
ACKNOWLEDGMENTS
We thank Dr. Stephen Grant for critical review of the manuscript.
REFERENCES
1. Ponder, B. A., and Wilkinson, M. M. Direct examination of the clonality of carcinogen-induced colonic epithelial dysplasia in chimeric mice. J. Natl. Cancer Inst., 77:967—976,1986.
2. Fearon, E. R., Hamilton, S. R., and Vogelstein, B. Clonal analysis of human colorectaltumors. Science (Washington DC), 238: 193—197,1987.
3. Fearon, E. R., and Vogelstein, B. A genetic model for colorectal tumorigenesis. Cell.61: 759—767,1990.
4. Ichii, S., Hon. A., Nakatsuru, S., Furuyama, J., Utsunomiya, J., and Nakamura, Y.Inactivation of both APC alleles in an early ttage of colon adenomas in a patient withfamilial adenomatous polyposis (FAP). Hum. Mol. Genet., 1: 387—390,1992.
5. Powell, S. M., Zilz, N.. Bearer-Barclay, Y., Bryan, T. M., Hamilton, S. R.,Thibodeau, S. N., Vogelstein. B., and Kinzler, K. W. APC mutations occur earlyduring colorectal tumorigenesis. Nature (Land.), 359: 235—237,1992.
6. Bos, J. L., Fearon, E. R., Hamilton, S. R., Verlaan-de Vries, M., van Boom, J. H.,van der Eb, A. J., and Vogelstein, B. Prevalence of ras gene mutations in humancolorectal cancers. Nature (Land.), 327: 293—297,1987.
7. Vogelstein, B., Fearon, E. R., Hamilton, S. R., Kem, S. E., Preisinger, A. C., Leppert,M., Nakamura,Y., White, R., Smits, A. M. M., and Bos, J. L. Genetic alterationsduring colorectal-tumor development. N. EngI. J. Med.. 319: 525—532,1988.
8. Bos, J. L. ras oncogenes in human cancer: a review. Cancer Res., 49: 4682-4689,1989.
9. Pretlow, T. P., Barrow, B. J., Ashton, W. S., O'Riordan, M. A., Pretlow, T. 0.,Jurcisek, J. A., and Stellato, T. A. Aberrant crypts: putative preneoplastic foci inhuman colonic mucosa. Cancer Ret., 51: 1564—1567, 1991.
10. Pretlow, T. P., Brasitus, T. A., Fulton, N. C., Cheyer, C., and Kaplan, E. L. K-rasmutations in putative preneoplastic lesions in human colon. J. Nail. Cancer Inst., 85:2004—2007,1993.
11. Smith, A. J., Stem, H. S., Penner, M., Hay, K., Mitri, A., Bapat, B. V., and Gallinger,S. Somatic APC and K-ras codon 12 mutations in aberrant crypt foci from humancolons. Cancer Ret., 54: 5527—5530,1994.
12. Yamashita, N., Minamoto, T., Ochiai, A., Onda, M., and Esumi, H. Frequent andcharacteristic K-ras activation in aberrant crypt foci of colon. Is there preferenceamong K-ras mutants for malignant progression? Cancer (Phila.), 75: 1527—1533,1995.
13. Minamoto, T., Yamashita, N., Ochiai, A., Mai, M., Sugimura, T., Ronai, Z., andEsumi, H. Mutant K-ras in apparently normal mucosa of colorectal cancer patients:its potential as a biomarker of colorectal tumorigenesis. Cancer (Phila.). 75: 1520—1526, 1995.
14. Levi, S., Urbano-Ispizua, A., Gill, R., Thomas. D. M., Gilbertson, J., Foster, C., andMarshall, C. J. Multiple K-ras codon 12 mutations in cholangiocarcinomas demonstrated with a sensitive polymerase chain reaction technique. Cancer Res., 51:3497—3502,1991.
15. Jiang, W., Kahn, S. M., Guillem, J. 0., Lu, S., and Weinstein, I. B. Rapid detectionof ras oncogenes in human tumors: applications to colon, esophageal, and gastriccancer. Oncogene, 4: 923—928,1989.
16. Keohavong, P., Thu. D., Whiteside, T. L., Swalsky, P., Bakker, A., Elder, E. M.,Siegfried, J. M., Srivastava, S., and Finkelstein, S. D. Detection of infrequent andmultiple mutations in human tumors and tumor-adjacent tissues. Anal. Biochem.,247: 394—403, 1997.
17. Keohavong, P., Thu. D., Melacrinos, A. C., DeMichelle, M. A. A., Weyant, R. J.,Luketich, J. D., Testa, J. R., Fedder, M., and Siegfried, J. M. Detection of low fractionK-ras mutations in primary lung tumors using a sensitive method. Int. J. Cancer, 74:162—170,1997.
18. Fischer, S. 0., and Lerman, L. S. DNA fragments differing by single base-pairsubstitutions separated in denaturing gradient gels: correspondence with meltingtheory. Proc. Natl. Acad. Sci. USA, 80: 1579—1583, 1983.
19. Keohavong, P., and Thilly, W. 0. Mutational spectrometry: a general approach forhottpot point mutations in selectable genes. Proc. NatI. Acad. Sci. USA, 89: 4623—4627, 1992.
20. Finkelstein, S. D., Sayegh, R., Bakker, A., and Swalsky. P. Determination of tumoraggressiveness in colorectal cancer by K-ras-2 analysis. Arch. Surg., 128: 526—532,I993.
21. Liu, K. C., and Wright. N. A. Histogenesis of colorectal carcinoma. In: H. K. Seitz,U. A. Simanowski, and N. A. Wright (eds.), Colorectal Cancer: From Pathogenetit toPrevention, pp. 313—321.Berlin: Springer-Verlag, 1989.
22. Shibata,D., Schaeffer,J., Li,Z-H.,Capella,0., andPerucho,M.Geneticheterogeneity of the c-K-ras locus in colorectal adenomas but not in adenocarcinomat. J. NatI.Cancer Inst., 85: 1058—1063, 1993.
23. Neri, A., Knowles, D. M., Greco. A., McCormick, F., and Dalla-Favera, R. Analysisof ras oncogene mutations in human lymphoid malignancies. Proc. NatI. Acad. Sci.USA, 85: 9268—9272, 1988.
24. Fart. C. J., Saiki, R. K., Erlich, H. A., McCormick, F., and Marshall, C. J. Analysis
of ras genemutationsin acutemyeloidleukemiaby polymeratechainreactionandoligonucleotide probes. Proc. Natl. Acad. Sci. USA, 85: 1629—1633,1988.
25. Aguilar, F., Harris, C. C., Sun, T., Hollstein, M., and Cerutti, P. Geographic variationof p53 mutational profile in nonmalignant human liver. Science (Washington DC),264: 1317—1319,1994.
26. Deschner, E. E. Kinetics of normal, preneoplastic, and neoplastic colonic epithelium.In: M. P. Moyer and 0. H. Poste (eds.), Colon Cancer Cells, pp. 52—56.San Diego:Academic Press, Inc., 1990.
27. Oudejans, J. J., Slebot, R. J. C., Zoetmulder, F. A. N., Mooi, W. J., and Rodenhuis,S. Differential activation of ras genes by point mutation in human colon cancer withmetastases to either lung or liver. mt. J. Cancer, 49: 875—879, 1991.
28. Novelli, M. R., Williamson, J. A., Tomlinson, I. P. M., Elia, S., Hodgson, S. V.,Talbot, I. C., Bodmer, W. F., and Wright, N. A. Polyclonal origin of colonicadenomas in an XO/XY patient with FAP. Science (Washington DC), 272: 1187—1190, 1996.
