We assessed the feasibility of clonality analysiswith human androgen receptor gene polymerasechain reaction in terms of the sensitivity andspecificity for normal and cancerous colonic tis-sues taken from fourteen informative casesselected from 22 women with colonic adenocar-cinoma. Ten crypts microdissected from each 10-Òm-thick cryostat sections and whole tissueswere used as samples. DNA was extracted from thesamples and amplified with and without priorenzyme digestion. These products were analyzedby capillary electrophoresis for clonality. Of thewhole-tissue DNA, none of the normal tissues andseven (50.0%) of the cancerous tissues showedmonoclonality. Of the microdissected samples,monoclonality was found in 88.4% (107/121) ofnormal crypts and 95.9% (117/122) of cancerouscrypts. Samples composed of crypts with shortand long alleles were found in eight of the 14 nor-mal colonic mucosae, but in none of the canceroustissues. We concluded that the sensitivity of thismethod is limited for both whole-tissue DNA andmicrodissected-tissue DNA, because monoclonali-ty from small samples does not always indicatemonoclonality of the entire lesion. The high speci-ficity of this method, however, allows polyclonalresults in whole tissues to be confirmed by addi-tional analysis of microdissected tissues.

Key words: clonality analysis, HUMARA,microdissection, colonic mucosa

Introduction

The highly polymorphic trinucleotide repeat of thehuman androgen receptor gene (HUMARA) makes itthe most attractive probe for clonality analysis amongthe currently available polymorphic DNA probescloned from X-chromosomal loci.1-4 HUMARA on the X-chromosome has a trinucleotide (CAG) repeat poly-morphism in its sequence. The number of this repetitionis extremely varied and by analyzing the sequence ofthe repetition in the 2 strands of X-chromosomes in awoman, in many cases, each paternal and maternal X-chromosome can be identified as having either long orshort alleles. Because, one of the X-chromosomes in awoman is inactivated by DNA methylation at an earlystage in its development and the restriction enzymeHpaII cannot breakdown methylated DNA, if themakeup of the target HUMARA DNA is disproportion-ately paternal or maternal, in other words monoclonal,the use of HpaII results in predominantly long alleles orshort alleles, revealing the clonality of the target cellpopulation.

The results obtained by using HUMARA, however,are conflicting. Clonality analysis using HUMARA hasbeen effective in establishing the clonality of homoge-neous samples such as blood, but its reliability forassessing the clonality of samples from the epitheliaand epithelial cancers remains questionable. Somestudies using HUMARA have found leukemia,5 cancersof the stomach,6 pancreas,7 and breast8 to be mono-clonal, whereas others have shown frequent nonran-dom X-inactivation in normal blood, skin, muscle,9 and

Original Article

Clonality analysis for normal and cancerous colon tissues with human androgenreceptor gene polymerase chain reaction

Kisu Kim, Jiro Kumagai, Yoshinobu Eishi, Ikuo Ishige, Yuki Ishige and Morio Koike

Department of Human Pathology, Graduate School, Tokyo Medical and Dental University

J Med Dent Sci 2005; 52: 163–170

Corresponding Author: Jiro KumagaiDepartment of Human Pathology, Graduate School, Tokyo Medicaland Dental University1-5-45, Yushima, Bunkyo-ku, Tokyo 113-8519, JapanTel: +81-3-5803-5176 Fax: +81-3-5803-0123 E-mail: j.kuma.pth1@tmd.ac.jpReceived April 1; Accepted June 10, 2005

nonneoplastic lesions such as endometriosis10 andatherosclerosis.11 These discrepancies suggest thereare limitations to this testing method.

The inconsistent findings are partly due to theanatomical structure of tissues. Samples taken from theepithelia or epithelial tumors are prone to contaminationby stromal components during the samplingprocess.12 Using large whole tissue samples thusincreases false negative results and decreases sensi-tivity. To address this issue, studies have been carriedout using small samples acquired by laser capturemicrodissection, a recently introduced sampling tech-nique that can reduce the risk of stromal contamination.This, however, leads to a bias toward monoclonalfindings and a decrease in specificity.

The most notable inconsistencies have been foundamong the studies of colonic lesions, highlighting thedifficulty of clonality analysis for colonic mucosa andcancers. Because X inactivation occurs at a relativelyearly stage of embryonic development, most progeny ofsingle X-inactivated embryonic cells in the adulthuman are believed to exist in groups of monoclonalcells.13,14 These monoclonal cell groups, or patches,have been suspected to influence the evaluation of cer-tain organs such as the colon.15,16 A recent study hasfound the patches in the human colon to be much larg-er than previously expected, leading to the conclusionthat studies using small samples obtained from thecolonic mucosa or cancers are biased toward showingmonoclonality.17

The purpose of our study was to evaluate the relia-bility of the HUMARA polymerase chain reaction(PCR) analysis using both whole tissues andmicrodissected single crypts. To achieve this, wedetermined the sensitivity and specificity for normal andcancerous colon tissues. This investigation shouldprovide useful additional information about the feasibil-ity of this method for analyzing the clonality of thecolonic mucosa and cancer.

Materials and Methods

Patients and samplesSamples were taken from the normal colonic

mucosa and cancer of 22 women who underwentcolectomy at the Kawaguchi Kogyo General Hospital.Samples were embedded in an optimum cutting tem-perature (OCT) compound. Three frozen sectionswere cut from each sample as follows: one 5-Òm-thicksection for hematoxyline-eosin staining to confirm the

orientation of the tissue and two 10-Òm-thick sections.Of the 10-Òm-thick sections, one was preserved in amicrotube for whole-tissue analysis and the other wasplaced on PALM membrane slide glasses (P.A.L.M.Mikrolaser Technologie AG, Bernried, Germany) forlaser capture microdissection. The ethics committee ofthe hospital approved the study.

Extraction of DNA and selection of heterozygouscases

DNA was extracted from 10-Òm-thick whole-tissuesections by using QIAamp DNA Mini Kit (Qiagen,Hilden, Germany) following the manufacturer’sinstructions. The heterozygosity (difference of CAGrepeats between the X chromosomes) of the DNA sam-ples from the whole tissues was determined by directsequencing.

PCR was carried out in a reaction mixture (50 ÒL)containing 5 ÒL of template DNA (The DNA extractedwas minimal in amount and could not be analyzed), 5ÒL of 10×PCR buffer (without MgCl2), 3 ÒL of 25 mMMgCl2, 5 ÒL of dNTP mixture, 5 ÒL of each HUMARAprimer (10 pmol/ÒL; forward: 5’-GTGCGCGAAGT-GATCCAGAACC, reverse: 5’-TACGATGGGCTTGGGGAGAACC), 0.5 ÒL of Taq Gold polymerase, and26.75 ÒL of DDW. The thermal cycles consisted of aninitial cycle of 95°C for 10 minutes. This was followed by30 cycles of the following steps: 95°C for 45 seconds,65°C for 45 seconds, and 72°C for 45 seconds. Afterthe last cycle, a final extension step was performed at72°C for 10 minutes. To detect PCR contamination, ablank control without DNA was always included in everyPCR reaction.

These PCR products were purified for DNAsequencing using the MagExtractor (Toyobo, Co.,Ltd., Osaka, Japan). The cycle sequencing was carriedout using BigDye Terminator v3.1 Cycle Sequencing Kit(Applied Biosystems, Foster City, California, USA) asfollows: a reaction mixture for cycle sequencing (20 ÒL)containing 10 ÒL of purified PCR product, 4 ÒL ofReady Reaction Premix (Applied Biosystems), 2 ÒL of5×BigDye Sequencing Buffer (Applied Biosystems),0.4 ÒL of primer (10 pmol/ÒL; forward: 5’-GTGCGC-GAAGTGATCCAGAACC) and 3.6 ÒL of DDW. Thecycle sequencing consisted of 25 cycles of 96°C for 10seconds, 50°C for 5 seconds and 60°C for 4 minutes.

To remove excess DyeDeoxy terminator, thesecycle sequencing products were purified before DNAsequencing using the CENTRI-SEP COLUMNS(Princeton Separations, Inc., Adelphia, New Jersey,USA). These were dissolved into 20 ÒL of formamide

K. KIM et al. J Med Dent Sci164

(Applied Biosystems), denatured at 95°C for 2 minutes,then immediately cooled on ice. Direct sequencing wasperformed using an ABI PRISM 3100 GeneticAnalyzer (Applied Biosystems).

Distinctly identical peak patterns were obtained forthe two X chromosomes in homozygous cases, but forheterozygous cases, a divergence of the patterns wasnoted. Because ABI PRISM 310 Genetic Analyzer(Applied Biosystems) detects the difference of tworepeats or more, a finding of two CAG repeats or morewas determined to indicate that a case was heterozy-gous, or informative.

Tissue processing and staining DNA for the HUMARA analysis was extracted from

fresh frozen tissues as follows. The 10-Òm-thick sec-tions of heterozygous cases were briefly stained with0.05% toluidine blue for microdissection. Using theP.A.L.M. Robot Microbeam (P.A.L.M. MikrolaserTechnologie AG), 10 crypts were microdissected fromeach tissue section. DNA was extracted from thesecrypts using the Pinpoint Slide DNA IsolationSystemTM (Zymo Research, Orange, California, USA)following the manufacturer’s instructions.

HUMARA PCR analysisFive microliters of DNA from whole tissues and

microdissected tissues were subjected to enzymaticdigestion with 5 units of HpaII in the buffer supplied bythe manufacturer in a final volume of 6.5 ÒL. The sam-ples were incubated for 16 hours at 37°C, subjected toheat inactivation at 100°C for 10 minutes, and thencooled on ice.

PCR was performed in a final volume of 50 ÒL con-taining 5 ÒL of template DNA (The DNA extracted wasminimal in amount and could not be analyzed), 5 ÒL of10×PCR Buffer (without MgCl2), 3 ÒL of 25 mMMgCl2, 5 ÒL of dNTP mixture, 5 ÒL of each HUMARAprimer (10 pmol/ÒL; forward IA: 6-FAM labeled 5’-GTGCGCGAAGTGATCCAGAACC, reverse: 5’-TAC-GATGGGCTTGGGGAGAACC), 0.5 ÒL of Taq Goldpolymerase, and 26.75 ÒL of DDW. Reactions werestarted with an initial denaturation step at 95°C for 10minutes. This was followed by 35 cycles of the followingsteps: 95°C for 45 seconds, 65°C for 45 seconds, and72°C for 45 seconds. After the last cycle, a finalextension step was carried out at 72°C for 10 minutes.All samples were amplified in triplicate.

Clonality analysis using an ABI PRISM 310 GeneticAnalyzer (Applied Biosystems) was performed as fol-lows: 1 ÒL of PCR product, 12 ÒL of deionized for-

mamide (Applied Biosystems), and 0.5 ÒL of 400HD[ROX] size standard (Applied Biosystems) were mixedand denatured at 95°C for 2 minutes, then immediate-ly cooled on ice. The running conditions were as fol-lows: 15 kV through a capillary with a length of 47 cmand an inner diameter of 50 Òm (Applied Biosystems)filled with 310 Genetic Analyzer 10×Buffer with EDTA(Applied Biosystems) and Genetic AnalyzerPerformance Optimized Polymer 4; POP-4 (AppliedBiosystems) at 60°C for 24 minutes. After elec-trophoresis, peak patterns were automatically ana-lyzed by GeneScan Analysis software (version 3.1.2,Applied Biosystems).

Quantification of clonalityThe integration analysis of the peaks allowed the

quantification of the data. The ratio of the two peaks inthe undigested sample (A/B) was divided by the ratio ofthe two peaks in the HpaII-digested sample (a/b). Asample was defined to be monoclonal when the cor-rected ratio indicated that either of the two alleles waspresent in excess of 50% (corrected ratio＞3) (Figure.1).

165CLONALITY ANALYSIS WITH HUMARA IN HUMAN COLON

Figure 1. Quantification of clonality. The ratio of the two peaks in the undigested sample (A/B) divided bythe ratio of the two peaks in the HpaII-digested sample (a/b). Polyclonal composition with random X-chromosomes would beexpected to show ratios equal to or close to 1.0. A sample wasdefined to be monoclonal when the corrected ratio indicated thateither of the two alleles was present in excess of 50% (corrected ratio＞ 3).

Results

Selection of heterozygous cases for colonicmucosa and cancer

DNA from whole tissue samples of normal colonicmucosae and cancerous tissues from 22 patientswere successfully extracted. The number of CAGrepeats in DNA extracted from the colonic mucosasamples determined by direct sequencing rangedfrom 5 to 38 (mean 21.5). Fourteen (63.6%) of the 22samples showed a difference of two or more CAGrepeats between the two X chromosomes (Table,Figure. 2) and were determined to be heterozygous.The normal mucosae and cancerous tissues of the 14patients were used for clonality analysis.

Clonality of whole tissue samplesNone of the whole tissue samples taken from the 14

normal mucosae were shown to be of monoclonal ori-gin. Seven (50.0%) of the 14 cancerous tissue sampleswere determined to be of monoclonal origin. Of these,five were of the short-allele type and two were of thelong-allele type.

Clonality of microdissected samplesEach of the crypts that were obtained using laser

capture microdissection had an average of 50 cells

(Figure. 3). Of the crypts obtained from the 14 infor-mative cases, PCR amplification was unsuccessful in13.6% (19/140) of the normal crypts and 12.9%(18/140) of cancerous crypts.

Analysis of the cell population of the crypts indicatedmonoclonality in 107 (88.4%) normal crypts and 117(95.9%) cancerous crypts. Fourteen (10.0%) normalcrypts and five (3.6%) cancerous crypts showed aclonality ratio below 3 and were determined to be poly-clonal.

Six of the 14 normal mucosae were determined to bemonoclonal, because all the crypts in these mucosaehad identical alleles (Table). Of these six, four had shortalleles and two had long alleles. The remaining eightmucosa samples had a mixture of short and long alle-les and were determined to be polyclonal. All 14 can-cerous tissue samples were determined to be mono-clonal with either a short or long allele, the ratio being 9to 5 (Figure. 4).

Discussion

Our study confirms that in the colonic mucosa, thesensitivity of clonality analysis with HUMARA is limited.Irrespective of the size of the samples, conclusive find-ings on the monoclonality of an entire lesion cannot be

K. KIM et al. J Med Dent Sci166

Table. Results of clonality analysis for whole-tissue DNA and microdissected-tissue DNA from normal and can-cerous colonic tissues

aThe number of CAG repeats of the human androgen receptor genes from X chromosomes were determined bydirect sequencing. The difference in the repeat numbers between two chromosomes are listed in this column; mono-S:Monoclonal with the short allele remaining after digestion; mono-L: Monoclonal with the long allele remainingafter digestion; poly: Polyclonal; N.D.: Clonality could not be determined because of PCR failure

obtained by this method of analysis. Large tissuesamples are prone to contamination by stromal com-ponents or other cells and small samples are affectedby the existence of patches. With the specificity beinghigh, results indicating a polyclonal population inwhole tissue samples can be interpreted to show theactual polyclonal nature of the epithelial tissues if theyare confirmed by the analysis of microdissected sam-ples.

Our analysis of whole tissue samples showed a lim-ited sensitivity of 50% for determining monoclonality.For the 50% of whole tissues which were shown to bemonoclonal, we accepted the findings without additionalanalysis for validation, because the effect of patch sizecan be ruled out in large samples such as whole tissuesamples. Results indicating the presence of a poly-clonal population in the remaining 50% of the samples,however, were considered to be inconclusive,

because it was not possible to determine whether theseresult were due to stromal contamination or the actualpolyclonal nature of the tissues.

The sensitivity was determined to be 88.4%(107/121) for microsdissected normal crypts and95.9% (117/122) for microdissected cancerous crypts.Because this was based on the findings that normalcrypts and cancerous crypts obtained by microdissec-tion are monoclonal, this sensitivity represents sensi-tivity, not for the entire tissue sample or lesion, but forsmall groups of cells for which clonality has beendetermined.

The existence of patches is the primary reason theresults cannot be conclusive for the entire lesion.Animal studies have shown that monoclonal groups ofup to 450 colonic crypts were distributed over thecolonic mucosa, suggesting the analysis of thecolonic mucosa to be especially susceptible to the exis-

167CLONALITY ANALYSIS WITH HUMARA IN HUMAN COLON

Figure 2. Selection of heterozygous female.A, B. Homozygous female. A No difference in the location of the final sequences of CAG repeat in the HUMARA gene(CAAGAGA) on the two X chromosomes. B The alleles are observed as a single peak. C, D. Heterozygous female.C The location of the terminal sequence of the two X chromosomes are different due to the difference in the numberof CAG repeats. D Long and short alleles are observed as two separate peaks.

tence of patches.15,16,18 The presence of patches inhuman colonic mucosa has been verified usingenzyme histochemistry in a study of heterozygouswomen with one X chromosome coded for inactivemutant G6PD.17 Reporting these patches to be largerthan previously thought, the study concluded that it wasvirtually impossible to establish the clonality of theentire lesion from the analysis of just a few crypts. Ouranalysis showed that in six out of 14 normal mucosasamples, all the crypts had the same type of allele, indi-cating monoclonality and supporting previous studieswhich have reported the presence of patches. To ourknowledge, the present study is the first to show theexistence of patches using samples taken from het-erozygous women without relying on rare mutations.

The specificity of analysis with HUMARA can beevaluated for normal whole tissues by determining theclonality of a random group of cells considered to bepolyclonal. In our study, because all normal whole tis-sue samples were determined to be polyclonal, we con-cluded that this confirms a specificity of 100% for wholetissues.

Using microdissected samples can reduce the risk ofstromal contamination, but because the patch size does

not allow a strictly random group of cells to beobtained, specificity cannot be determined as forwhole tissues. The finding that a whole tissue is poly-clonal, however, is confirmed if analysis of microdis-sected samples from that tissue also finds it to be poly-clonal. In our study, seven cancerous whole tissue sam-ples were determined to be of monoclonal origin andthe remaining seven to be of polyclonal origin, but sub-sequent analysis of microdissected crypts showed allthese samples to be monoclonal, rendering no conclu-sive results. All normal whole tissue samples werefound to be polyclonal and additional analysis ofmicrodissected samples confirmed these findings ineight of these samples.

Although the reason remains unclear, varyingdegrees of non-random X-chromosome inactivationhas been shown in normal tissues.9,14 We also found20% of 100 peripheral blood samples from healthywomen (data not shown) to be monoclonal. This con-founds the notion that monoclonality indicates malig-nancy. In our analysis, however, no normal mucosasamples were found to be monoclonal. This conse-quently facilitated the interpretation of our results.

The analysis of microdissected samples found

K. KIM et al. J Med Dent Sci168

Figure 3. A-F. Photomicrographs of toluidine-blue stained, OCT-embedded sections of colon tissue (A-C normal colon musosa; D-F colon can-cer lesion) Specimens are shown: A, D Before microdissection. B, E After laser pressure cell transfers. C, F Microdissected crypts. A, B, D,E×10. C, F×20

10.0% of normal crypts and 3.6% of cancerous cryptsto be polyclonal, leading us to suspect that the detec-tion of monoclonal populations was hampered bystromal contamination that occurred during microdis-section or by inflammatory cell infiltration in theepithelium. Another explanation for these results is con-tamination of the microdissected crypts by cells from anadjacent crypt composed of a different type of cell.12

The possibility that endocrine cells in colonic cryptsaffected the results should also be considered,because these cells have not been confirmed to be of

the same clonal population as epithelial cells.Sakurazawa et al assessed the clonality of

microdissected crypts taken from aberrant crypt foci inthe human colon.19 Their assessment using HUMARAshowed 38% of the crypts to be polyclonal, while theinvestigation of K-ras mutations showed the crypts tobe monoclonal. They concluded that because DNAmethylation in aberrant crypt foci is an unstable eventwhich can occur de novo, monoclonal foci can attimes appear to be polyclonal. In our study, only a smallpercentage of crypts were determined to be formed

169CLONALITY ANALYSIS WITH HUMARA IN HUMAN COLON

Figure 4. Clonality analysis of normal mucosa and cancer. Typical methylation patterns before digestion and after HpaII digestion. A-C wholetissue DNA. D-F microdissected tissue DNA. A Normal mucosa. Decrease in alleles insufficient to indicate monoclonality. B Monoclonal can-cer tissue sample showing a decrease in long alleles. C Polyclonal cancer tissue sample not showing a decrease in either allele. D Monoclonalmicrodissected normal crypt showing: a decrease in short alleles (second from the top); a decrease in long alleles (second from the bottom);and no decrease in either allele (bottom). E microdissected cancerous crypts showing a decrease in long alleles (center) and showing nodecrease in alleles (bottom).

from polyclonal cell populations: 10% of normal cryptsand 3.5% of cancerous crypts. Although we suspect theeffect of de novo methylation to be minimal, this pointwarrants further investigation.

Our method of analysis may shed some light on con-troversial issues on clonality such as the clonal natureof certain colonic tumors. Tatematsu et al.20 found intheir study on chemical carcinogenesis of chimeramice that cancers with a mixture of two strains of cryptscan develop. They surmised this may be a result of acollision of two or more lesions, because such obser-vations were made when clusters of numerouslesions developed in the colon. If, however, sporadiclesions like the ones we studied are shown to be poly-clonal, it would confirm that some cancers are indeedpolyclonal. Bjerknes et al.21 studied familial adenoma-tous polyposis patients using somatic mutation andconcluded that adenomas were not clonally pure. Itshould be noted, however, that their definition of clon-ality is not clonality as it is generally understood,because their marker contains an acquired mutation. Ifthis adenoma is indeed polyclonal, our analysisshould confirm its polyclonal nature.

In conclusion, when using HUMARA for clonalityanalysis of normal and cancerous colonic tissues,monoclonality can be determined in whole tissueswith a 50% sensitivity, but beyond that, even withadditional analysis of microdissected samples, conclu-sive findings on monoclonality cannot be obtained. Thehigh specificity of the analysis, however, guaranteesthat when a whole-tissue sample is found to be poly-clonal and additional analysis of microdissected sam-ples also shows the same, this rules out the possibilityof stromal contamination and confirms the polyclonalnature of the tissue. This method of analysis may pro-vide conclusive findings on the possible polyclonalnature of certain cancers and adenomas.

Acknowledgments

The authors thank N. Ando of the Department ofHuman Pathology, Tokyo Medical and DentalUniversity Graduate School, and Y. Takeichi and N.Katsuta of the Laboratory of Pathology, KawaguchiKogyo General Hospital, for technical assistance.

References1. Allen RC, Zoghibi HY, Moseley AB, et al. Methylation of Hpa II

and Hha I sites near the polymorphic CAG repeat in the andro-

gen-receptor gene correalates with X chromosome inactiva-tion. Am J Hum Gen 1992;51:1229-39

2. Mashal RD, Lester SC, Sklar J. Clonal analysis by study of Xchromosome inactivation in formaline-fixed paraffin-embeddedtissue. Cancer Res 1993;53:4676-9

3. Busque L, Zhu J, Dehart D, et al. An expression based clon-ality assay at the human receptor locus (HUMARA) on chro-mosome X. Nucleic Acids Res 1994;22:697-8

4. Delabesse E, Aral S, Kamoun P, et al. Quantitative non-radioactive clonality analysis of human leukemic cells andprogenitors using the human androgen receptor (AR) gene.Leukemia 1995;9:1578-82

5. Gale RE, Mein CA, Kinch DC. Quantification of X-chromosomeinactivation patterns in haematological samples using the DNAPCR-based HUMARA assay. Leukemia 1996;10:362-7

6. Bamba M, Sugihara H, Okada K, et al. Clonal analysis ofsuperficial depressed-type gastric carcinoma in human.Cancer 1998;83:867-75

7. Izawa T, Obara T, Tanno S, et al. Clonality and field cancer-ization in intraductal papillary-mucinous tumors of the pan-creas. Cancer 2001;92:1807-17

8. Diallo R, Schaefer KL, Poremba C, et al. Monoclonality in nor-mal epithelium and in hyperplastic and neoplastic lesions ofthe breast. J Pathol 2001;193:27-32

9. Gale RE, Wheadon H, Boulos P, et al. Tissue specificity of X-chromosome inactivation patterns. Blood 1994;83:2899-905

10. Jimbo H, Hitomi Y, Yoshikawa H, et al. Evidence for mono-clonal expansion of epithelial cells in ovarian endometrial cyc-sts. Am J Pathol 1997;150:1173-78

11. Schwartz SM, Murry CE. Proliferation and the monoclonal ori-gins of atherotic lesions. Annu Rev Med 1998;49:437-60

12. Paradis V, Dargere D, Bonvoust F, et al. Clonal analysis ofmicronodules in virus C-induced liver chirrhosis using lasercapture microdissection (LCM) assay and HUMARA assay.Lab Invest 2000;80:1553-9

13. Lyon MF. Gene action in the X-chromosome of the mouse.Nature 1961;190:371-3

14. Fey MF, Leichti-Gallati S, von Rohr A, et al. Clonality an X-inactivation patterns in haematopoietic cell populationsdetected by the highly informative M27ÇDNA probe. Blood1994;83:931-8

15. Ponder BA, Schmidt GH, Wilkinson MM, et al. Derivation ofmouse intestinal crypts from single progenitor cells. Nature1985;313:689-91

16. Kusakabe M, Yokoyama M, Sakakura T, et al. A novelmetodology for analysis of cell distribution in chimeric mouseorgans using a strain specific antibody. J Cell Biol1988;107:257-65

17. Novelli M, Cossu A, Oukrif D, et al. X-inactivation patch size inhuman female tissue confounds the assessment of tumor clon-ality. Proc Natl Acad Sci USA 2003;100:3311-4

18. Brittan M, Wright NA. Stem cell in gastrointestinal structureand neoplastic development. Gut 2004;53:899-910

19. Sakurazawa N, Tanaka N, Onda M, Esumi H. Instability of Xchromosome methlation in aberrant crypt foci of the humancolon. Cancer Res 2000;60:3165-9

20. Tatematsu M, Masui T, Fukami H, et al. Primary monoclonaland secondary polyclonal growth of colon neoplastic lesions inC3H/HeN—BALB/c chimeric mice treated with DMH. Int JCancer 1996;66:234-8

21. Bjerknes M, Cheng H, Kim H, et al. Clonality of dysplasticepithelium in colorectal adenomas from familial adenomatouspolyposis patients. Cancer Res 1997;57:355-361

K. KIM et al. J Med Dent Sci170