A
B
C
D
205 -
110 - 90 - 53 -
extracts. The use of synthetic peptides represents a substantial advance in the technology because there is, finally, a defined antigen to use in tumor studies. Similar studies with synthetic carbohy- drates have also been reported (19). It should now be possible to determine whether these second-generation anti- bodies do indeed represent a practical and useful advance. Serological tests should be easier to perform. Whether antipeptide antibodies have any selective in vivo ad- vantage over those reacting with nonpep- tide epitopes remains to be studied.
References
(/) GENDLER SJ. BURCHELL JM. DUHIG T. ET AL: Cloning of partial cDNA encoding differentia- tion and tumor-associated mucin glycoproteins expressed by human mammary epithelium. Proc Natl Acad Sci USA 84:6060-6064, 1987
(2) SIDDIQUI J, ABE M. HAYES D, ET AL: Isolation and sequencing of a cDNA coding for the human DF3 breast carcinoma-associated an- tigen. Proc Natl Acad Sci USA 85:2320-2323, 1988
(3) GUM JR. BYRD JC. HICKS JW. ET AL: Molecu- lar cloning of human intestinal mucin cDNAs: Sequence analysis and evidence for genetic polymorphism. J Biol Chem 264:6480-6487, 1989
(4) GUM JR. HICKS JW, SWALLOW DM, ET AL: Molecular cloning of cDNAs derived from a
novel human intestinal mucin gene. Biochem Biophys Res Commun 171:407-415, 1990
(5) PORCHET N, NGUYEN VC. DUFOSSE J. ET AL: Molecular cloning and chromosomal localiza- tion of a noval human tracheo-bronchial mucin cDNA containing tandemly repeated sequen- ces of 48 base pairs. Biochem Biophys Res Commun 175:414-422. 1991
(6) LIGTENBERG MJ, Vos HL, GENNISSEN AM, ET AL: Episialin, a carcinoma-associated mucin, is generated by a polymorphic gene encoding splice variants with alternative amino-termini. J Biol Chem 265:5573-5578. 1990
(7) PRICE MR. HUDECZ F. O’SULLIVAN C. ET AL: Immunological and structural features of the protein core of human polymorphic epithelial mucin. Mol Immunol 27:795-802, 1990
(8) BURCHELL J, TAYLOR-PAPADIMITRIOU J, BOSHELL M, ET AL: A short sequence, within the amino acid tandem repeat of a cancer-associated mucin, contains immunodominant epitopes. Int J Cancer 44:691-696, 1989
(9) XING P-X. PRENZOSKA J. QUELCH K, ET AL: Second generation anti-MUCI peptide mono- clonal antibodies. Cancer Res. In press
(10) XING P-X, REYNOLDS K, TJANDRA JJ, ET AL: Synthetic peptides reactive with anti-human milk fat globule membrane monoclonal an- tibodies. Cancer Res 50:89-96, 1990
(11) TEH JG, THOMPSON CH, MCKENZIE IF: Pro- duction and characterization of a new mono- clonal antibody to colorectal carcinoma. Immunol Cell Biol 68:253-262. 1990
(12) HOPP TP, WOODS KR: A computer program for predicting protein antigenic determinants. Mol Immunol 20:483-489. 1983
(13) WELLING GW. WEIJER WJ. VAN DER ZEE R. ET AL: Prediction of sequential antigenic regions in proteins. FEBS Lett 188:215-218. 1985
(14) STACKER SA. THOMPSON CH, RIGLAR C, ET AL: A new breast carcinoma antigen defined by a monoclonal antibody. JNCI 75:801-811, 1985
(15) AUGERON C. LABOISSE CL: Emergence of per- manently differentiated cell clones in a human colonic cancer cell line in culture after treat- ment with sodium butyrate. Cancer Res 44:3961-3969. 1984
(16) DEVINE PL. WARREN JA, WARD BG, ET AL: Glycosylation and the exposure of tumor as- sociated epitopes on mucins. J Tumor Marker Oncol 5:11-26. 1990
(17) MCKENZIE IF, XING P-X: Mucins in breast cancer. Recent immunological advances. Can- cer Cells 2:75-78, 1990
(18) JANY BH, GALLUP MW. YAN PS. ET AL: Human bronchus and intestine express the same mucin gene. J Clin Invest 87:77-82, 1991
(19) FUNG PY, MADEJ M, KOGANTY RR, ET AL: Ac- tive specific immunotherapy of a murine mam- mary adenocarcinoma using a synthetic tumor-associated glycoconjugate. Cancer Res 50:4308-4314, 1990
FREE CATALOG of Government Books Send for your copy today!
Free Catalog Box 37000 Washington DC 20013-7000
Detection of Novel Germ-line p53 Mutations in Diverse- Cancer-Prone Families Identified by Selecting Patients With Childhood Adrenocortical Carcinoma
Yuichi Sameshima, Yukiko Tsunematsu, Shaw Watanabe, Taiji Tsukamoto, Keisei Kawa-ha, Yoshiaki Hirata, Hideaki Mizoguchi, Takashi Sugimura, Masaaki Terada, Jun Yokota*
Background: Germ-line p53 mutations appear to be inherited among the mem- bers of families diagnosed with Li- Fraumeni syndrome (LFS). The mutations detected in those families to date have been clustered in exon 7 of the p53 gene and, typically, have been single-base substitutions resulting in amino acid changes. Purpose: Our aim was to define the spectrum of p53 mu- tations associated with LFS. Methods: From seven cancer-prone families iden- tified by selecting members with childhood adrenocortical carcinoma as probands, we chose two families, each of which had two members from whom specimens could be obtained for
Received July 10. 1991; revised January 31, 1992; accepted January 31, 1992.
Supported in part by a Grant-in-Aid for a Com- prehensive 10-Year Strategy for Cancer Control from the Ministry of Health and Welfare of Japan and Grants-in-Aid from the Ministry of Health and Welfare and from the Ministry of Education, Science, and Culture of Japan. Y. Sameshima is a recipient of a Research Resident Fellowship from the Foundation for Promotion of Cancer Research.
Y. Sameshima, S. Watanabe, T. Sugimura, M. Terada. J. Yokota, National Cancer Center Research Institute, Tokyo, Japan.
Y. Tsunematsu, National Children’s Hospital, Tokyo. T. Tsukamoto, Sapporo Medical College, Hok- kaido, Japan.
K. Kawa-ha, Osaka University Hospital, Osaka, Japan.
Y. Hirata, Hamamatsu Medical Center, Shizuoka, Japan.
H. Mizoguchi, Tokyo Women’s Medical College. *Correspondence to: Dr. J. Yokota, M.D., Na- tional Cancer Center Research Institute, 1-1, Tsukiji 5-chome, Chuo-ku, Tokyo 104, Japan.
Downloaded from http://jnci.oxfordjournals.org/ at UQ Library on June 14, 2015
genetic analysis. To detect germ-line p53 gene mutations in these indivi- duals, we performed polymerase chain reaction (PCR)-single-strand confor- mation polymorphism analysis with Taq polymerase and oligonucleotide primers specific for p53 gene sequen- ces. Genomic DNA extracted from fresh tissue samples and paraffin-em- bedded tumor samples was amplified, denatured, and electrophoresed on neutral polyacrylamide gels. PCR amplification was also carried out using total RNA from adrenocortical carcinoma samples of the proband in family 1. PCR products were purified, subcloned, and sequenced. Results: We detected novel germ-line p53 mutations in affected members of both cancer- prone families. In the proband of fami- ly 1, a single-base deletion was detected at the first nucleotide of codon 307 in exon 8 of the p53 gene, resulting in a premature stop codon in exon 10. In family 2, we detected an A to C trans- version at the second nucleotide of codon 286 in exon 8, both in DNA iso- lated from the adrenocortical tumor of the proband and in DNA isolated from the astrocytoma of the proband’s father. This single-base substitution resulted in an amino acid substitution of alanine for glutamic acid. Both of these mutations are located outside the highly conserved region of the p53 gene where mutations in patients with LFS have been reported previously. Con- clusion: Our results indicate that a wide range of germ-line p53 mutations is inherited in members of diverse-can- cer-prone families. [J Natl Cancer Inst 84: 703-707, 1992]
Li-Fraumeni syndrome (LFS) is a dom- inantly inherited rare familial cancer syndrome characterized by a high suscep- tibility to diverse malignant tumors (1,2). Recently, germ-line p53 mutations have been identified among members of six families with LFS (3,4), indicating that LFS is etiologically attributable to germ- line p53 mutations. However, since the criteria for LFS have not yet been pre- cisely defined, our purpose was to define the spectrum of p53 mutations associated with LFS to develop a gene-based clas- sification system. To accomplish this
goal, it was important to screen the mem- bers of diverse-cancer-prone families for the presence of germ-line p53 mutations and to identify the location and type of p53 mutations associated with LFS. We previously proposed that cancer-prone families could be effectively identified by selecting members with childhood adreno- cortical carcinoma, rather than sarcoma, as candidates for the probands because of the low incidence of adrenocortical car- cinoma in the general population and the frequent occurrence of this cancer in patients with LFS. From 47 families, we were able to identify seven cancer-prone ones that had members with childhood adrenocortical carcinoma (5). The cri- terion for the selection of these seven families was that three or more family members who were first-degree or sec- ond-degree relatives of the proband developed cancer before age 45.
We tested affected members of two families with LFS from whom tumor specimens could be obtained for the presence of germ-line p53 mutations. We used polymerase chain reaction-single- strand conformation polymorphism (PCR- SSCP) analysis, a rapid and sensitive method for detecting subtle mutations, and found novel p53 mutations in individuals from both of these families. This result indi- cates that a wide range of p53 mutations is inherited in diverse-cancer-prone families. The data obtained in this study should be valuable for the more precise classification of cancer family syndromes characterized by diverse tumors and associated with in- herited p53 mutations.
Subjects and Methods
Subjects
Seven cancer-prone families were identified from a total of 47 families. Those whose members had childhood adrenocortical carcinoma were selected as candidates for probands (5). In four in- dividuals, two each from two of these seven families, fresh surgical tissue speci- mens, paraffin-embedded tumor speci- mens, or fresh samples of peripheral lymphocytes were available for genetic analysis. Abridged pedigrees of these two families are shown in Fig. 1. The diag- noses of tumor types were either con- firmed by pathology records or obtained by family history, as indicated in Fig. 1
by the filled or hatched symbols, respec- tively. Although family 1 conformed to the original criteria for LFS proposed by Li et al. (2), family 2 may not, because we observed no family members with sar- coma or breast cancer.
Samples
From family 1, fresh samples of adreno- cortical carcinoma tissue and of periph- eral lymphocytes from the proband (III-I) and a fresh sample of peripheral lym- phocytes from his unaffected mother (II- 1) were obtained for genetic analysis (Fig. 1). From family 2, we obtained paraffin- embedded specimens from both the adrenocortical tumor of the proband (III- 4) and from the astrocytoma of the proband’s father (II-3), in addition to a fresh sample of peripheral lymphocytes from the proband’s father.
PCR-SSCP Analysis
Downloaded from http://jnci.oxfordjournals.org/ at UQ Library on June 14, 2015
High-molecular-weight DNA from fresh tissue samples was prepared by pro- teinase K digestion and phenol-chloro- form extraction (6). DNA from paraf- fin-embedded samples was extracted as described previously (7). PCR amplifica- tion was performed by using 250 ng of genomic DNA as a template, one unit of Taq polymerase (Perkin-Elmer-Cetus Corp., Norwalk, Conn.), and three pairs of p53-specific, unlabeled oligonucleo- tide primers. The primers for the follow- ing regions of DNA were synthesized with a model 391A DNA synthesizer (Applied Biosystems, Foster, Calif.) ac- cording to these published sequences (8): 5’-GGAATTCTTCCTCTTCCTGCAG- TACTC-3’ and 5’-GGAATTCAAAC-2 CAGACCTCAGGCGGCT-3’ for exons 5 and 6, 5’-GGAATTCCTAGCACTGCC- CAAC-3’ and 5’-GGAATTCCTGACC- TGGAGCT-3’ for exon 7, and 5’-GGAA- TTCCTATCCTGAGTAGTGG-T-3’ and 5’-GGAATCAAGACTTAGTACCTGA-3’ for exons 8 and 9.
Three separate amplification reactions were carried out, one with each pair of primers. In each reaction, the final con- centrations of the reagents in a total reac- tion volume of 20 uL were as follows: 10 mM Tris-HCI (pH 8.3), 50 mM KCI, 1.25 mM MgCl2, 0.01% gelatin, the four deoxynucleotide triphosphates at a concen- tration of 250 nM each, 1 L of [c .- 32P] deoxycytidine triphosphate (3000 Ci/mmol,
Family 1
Family 2
1
2
I
1
2
3
4
5
6
Br (57)
I
St (30)
St (33)
1
II
2
3
4
5
6
7
8
Br
Lu (45)
Os (5)
II
1
2
3
4
5
Pa (36)
BI (20)
As (41)
III
1
2
3
4
5
III
1
2
3
4
5
6
Ad (5)
Os
Ly
He
He
Ad
(2)
(4)
10 Ci/mL), and a pair of primers, each primer at a concentration of approximate- ly 100 nM (10 pmol/tube). Amplification was carried out for 35 cycles, each con- sisting of denaturation at 94 ℃ for 40 seconds, annealing at 55 ℃ for 40 seconds, and extension at 72 ℃ for 90 seconds. After the last cycle, samples were extended for 10 minutes at 72 ℃. PCR products were diluted 10-fold with diluent consisting of 0.1% sodium do- decyl sulfate, 20 mM EDTA, and 0.05% bromophenol blue, followed by a second 10-fold dilution with denaturing solution containing 96% formamide, 20 mM EDTA, and 0.05% bromophenol blue. A 1-uL aliquot of the diluted sample was then denatured by heating at 80 ℃ for 2 minutes and electrophoresed on a 6% nondenaturing polyacrylamide gel as de- scribed previously (9).
Reverse Transcriptase PCR
To determine whether the single-base deletion at the exon-intron junction of the p53 gene in the proband of family 1 resulted in a splicing abnormality or a frameshift mutation, we extracted total cellular RNA from a fresh sample of adrenocortical carcinoma tissue of the proband in family 1, as previously described (10). The first strand of com-
plementary DNA (cDNA) was synthe- sized from 5 µg of total RNA as the template, using 2.5 units of avian myelo- blastosis virus reverse transcriptase as the enzyme and oligo (dT)12-18, a mixture of oligomers of deoxythymidine triphos- phates ranging in size from 12 to 18 nucleotide bases, as the primer. The final concentrations of the reagents in a total reaction volume of 20 uL were as fol- lows: 50 mM Tris-HCI (pH 8.3), 75 mM KCI, 3 mM MgCl2, 10 mM dithiothreitol, 50 ng/uL of oligo (dT)12-18, and the four deoxynucleotide triphosphates at a con- centration of 1 mM each. The reaction mixture was incubated for 60 minutes at 37 ℃ and then diluted with 80 uL of PCR buffer to make the final concentration of the reagents as follows: 18 mM Tris-HCI, 55 mM KCI, 1.6 mM MgCl2, 0.01% gelatin, and each of the two primers at a concentration of approximately 100 nM (10 pmol/tube). Subsequent PCR amplifi- cation of a region of p53 cDNA was per- formed using a pair of oligonucleotides as primers: 5’-GGAATTCACCATCCAC- TACAACTACATGTGT-3’ and 5’- GGAATTCAGCCCTGCTCCCCCCGG- CTC-3’ (8), 2.5 units of Taq polymer- ase, and the same Thermocycler program used in the PCR-SSCP analysis above. (11).
Subcloning and Sequencing of PCR Products
Since the primers used for PCR or reverse transcriptase PCR had extraneous nucleotides comprising EcoRI sites at their 5’ ends, the PCR products were purified from agarose gels by phenol- chloroform extraction, digested with EcoRI, and inserted with T4 ligase into the EcoRI site of the plasmid pUC18. To confirm that sequence variations were not due to misincorporation by Taq poly- merase, we used DNA fragments pooled from 100 to 200 plasmid colonies as templates in the sequencing step (12). Se- quencing was performed by the dideoxy chain-termination method with a 7- DEAZA sequencing kit (version 2.0; U.S. Biochemical Corp., Cleveland, Ohio) (11). In each case, the presence of a mutation was confirmed by sequencing both strands of the DNA.
Results
Germ-line p53 Mutation in Family 1
In family 1, when amplified fragments of DNA from exons 8 to 9 in samples were electrophoresed, we observed a shift in mobility of the DNA fragments from the proband (III-1), compared with frag- ments from the. proband’s unaffected mother (II-1). Aberrantly migrating bands, in addition to the bands for the wild-type allele, were detected in the blood sample, while only abnormal bands were detected in the tumor sample from the proband (III-1). The unaffected mother (II-1) of the proband carried only the wild-type allele of the p53 gene (Fig. 2-family 1, panel A). When we se- quenced exons 8 and 9 of the p53 gene in this proband and his mother, we found a single-base deletion of G at either the first nucleotide of codon 307 in exon 8 or at the first nucleotide in intron 8 in the proband (III-1) (Fig. 2-family 1, panel B), but not in his mother (II-1). Since this mutation was located at the exon-intron junction, we further analyzed the p53 messenger RNA transcripts expressed in the tumor by reverse transcriptase PCR amplification and subsequent sequencing. We found that this deletion occurred at the first nucleotide of codon 307 in exon 8 and resulted in a frameshift mutation but not an abnormal splicing (//); conse-
Family 1
A
B
C
II-1 III-1
L
LT
AGCT
AGCT
G
G
Leu
C
A
T
(308)
G
Cys
A
intron 8
C
T
W
T
+
A
C
Ala (307)
C
M
G
His
G
G
A
+
G
C
A
G
exon 8
A
A
G
Arg
G
Arg
C
C
(306)
C
(306)
Family 2
frameshift
A
B
C
II-3
N
L
AGCT
AGCT
G
G
A
A
Glu
G
G
(287)
W
M
A
A
A
A
C
Glu (286)
A
C
Ala
G
G
G
G
G
A
A
Glu
G
G
(285)
quently, a premature stop codon appeared in exon 10. In conclusion, in one affected individual of family 1, we detected a germ-line p53 mutation that was a single- base deletion of G at the first nucleotide of codon 307 in exon 8 (Fig. 2-family 1, panel C). Furthermore, analysis of the tumor and peripheral lymphocyte samples from the proband revealed that loss of the wild-type allele had occurred during tumor development and that only a mutated allele was expressed in the tumor.
Germ-line p53 Mutation in Family 2
In family 2, we first searched for germ- line p53 mutations using a DNA isolated from a fresh sample of peripheral lym- phocytes from the proband’s father (II-3) (Fig. 1), whose diagnosis with an astro- cytoma had been histologically confirmed
and who was considered to be a carrier. By PCR-SSCP analysis, we were able to determine that a germ-line p53 mutation was present in the region between exon 8 and 9 in this patient (Fig. 2-family 2, panel A). Subsequent sequence analysis of the amplified sequences of DNA ex- tracted from the peripheral lymphocytes of this carrier (II-3) revealed that the type of mutation that was present was a single- base substitution of a C for an A at the second nucleotide of codon 286 in exon 8, resulting in the substitution of alanine for glutamic acid (Fig. 2-family 2, panel B). To determine whether this mutation was inherited, we extracted DNA from paraffin-embedded samples of both the adrenocortical tumor of the proband (III- 4) and the astrocytoma of her father (II- 3). After PCR amplification, the same single-base substitution was detected at
codon 286, both in DNA extracted from the adrenocortical carcinoma of the proband (III-4) and in DNA extracted from the astrocytoma from her father (II- 3) (Fig. 2-family 2, panel C). In con- clusion, the p53 mutation detected in family 2 by PCR amplification and sub- sequent sequencing of exon 8 did appear to be inherited. Sequence analysis sug- gested that wild-type p53 genes were retained in both of these tumors, although we could not exclude the possibility of contamination of the tumor specimens with normal cells.
Discussion
We found novel germ-line p53 muta- tions in members of two cancer-prone families. One was a G deletion at the first nucleotide of codon 307 in exon 8, outg side the highly conserved regions (codong 13-19, 117-142, 171-181, 236-258, and 270-286) of the p53 gene. The other was an A to C transversion at the second nucleotide of codon 286 in exon 8, the last codon of the highly conserved regiorg V (codons 270-286) (13) (see Table 1). le has been reported to date that p53 muta- tions in all six families with LFS were present only in the highly conserved region IV (codons 236-258) of exon (13) and were single-base substitutions. causing amino acid changes (3,4). This is the first report suggesting that either point mutation in region V or a frameshiff mutation of the p53 gene may predispose carriers in cancer-prone families to develop tumors.
Inheritance of germ-line p53 mutations was strongly indicated in family 2 be cause the proband and his father carried the same p53 mutation. However, we could not determine whether the mutation that we detected in family 1 originated with the proband or was inherited, be- cause no samples from any other affected members of family 1 were available for genetic analysis. Our study would have been greatly strengthened if we had been able to perform genetic analysis on sam- ples from other individuals in family 1 (e.g., III-2, III-3, III-4, or II-4).
We selected the two families in this study by identifying members with child- hood adrenocortical carcinoma rather than members with sarcoma as candidates for the probands. Family I was diagnosed
| p53 genotype, allele 1/allele 2 | ||||
|---|---|---|---|---|
| Individual | Type of tumor Lymphocytes | Tumors | Type of germ-line mutation | |
| Family 1 | ||||
| II-1 (noncarrier) | wt/wt | |||
| III-1 (affected) | Adrenocortical carcinoma | mu/wt | mu/- | One base deletion at codon 307 GCA (Ala) -> _CA (frameshift) |
| Family 2 | ||||
| II-3 (affected) | Astrocytoma | mu/wt | mu/ND | Point mutation at codon 286 |
| III-4 (affected) | Adrenocortical carcinoma | mu/ND | GAA (Glu) -> GCA (Ala) | |
*wt = wild type; mu = mutated type; - = allelic loss; ND = not determined.
by review of clinical records to be af- fected by typical LFS, but family 2 may not have been a typical LFS kindred be- cause we observed no member with either sarcoma or breast cancer (Fig. 1). How- ever, we observed only a limited part of the pedigree, and the tumors of certain family members were confirmed by his- tory only. In addition, the occurrence of other tumors (e.g., gastric carcinoma, common in Japan but atypical in mem- bers of families with LFS) is not enough reason to automatically exclude a diag- nosis of LFS for this family.
Molecular genetic analysis revealed the presence of germ-line p53 mutations in both of these families. Our results indi- cate that cancer-prone families with p53 mutations can be effectively identified by selecting members with childhood adreno- cortical carcinoma as candidates for the probands. In conclusion, the present data strongly suggest that we should genetical- ly screen diverse-cancer-prone families for the presence of germ-line p53 muta- tions to understand the pathogenic sig- nificance and etiologic role of p53 mutations in cancer.
(6) SAKAMOTO H, MORI M, TAIRA M, ET AL: Transforming gene from human stomach can- cers and a noncancerous portion of stomach mucosa. Proc Natl Acad Sci USA 83:3997- 4001, 1986
(7) TSUDA H, SHIMOSATO Y, UPTON MP, ET AL: Retrospective study on amplification of N-myc and c-myc genes in pediatric solid tumors and its association with prognosis and tumor dif- ferentiation. Lab Invest 59:321-327, 1988
(8) BUCHMAN VL, CHUMAKOV PM, NINKINA NN, ET AL: A variation in the structure of the protein-coding region of the human p53 gene. Gene 70:245-252, 1988
(9) ORITA M, SUZUKI Y, SEKIYA T, ET AL: Rapid and sensitive detection of point mutations and DNA polymorphisms using the polymerase chain reaction. Genomics 5:874-879, 1989
(10) YOKOTA J, AKIYAMA T, FUNG Y-K, ET AL: AI- tered expression of the retinoblastoma (RB) gene in small-cell carcinoma of the lung. On- cogene 3:471-475, 1988
(11) SAMESHIMA Y, AKIYAMA T, MORI N, ET AL: Point mutation of the p53 gene resulting in splicing inhibition in small-cell lung car- cinoma. Biochem Biophys Res Commun 173:697-703, 1990
(12) MORI N, YOKOTA J, AKIYAMA T, ET AL: Vari- able mutations of the RB gene in small-cell lung carcinoma. Oncogene 5:1713-1717, 1990
(13) SOUSSI T, CARON DE FROMENTEL C, MAY P: Structural aspects of the p53 protein in relation to gene evolution. Oncogene 5:945-952, 1990
References
(/) LI FP, FRAUMENI JF JR: Soft-tissue sarcomas, breast cancer, and other neoplasms. A familial syndrome? Ann Intern Med 71:747-752, 1969
(2) LI FP, FRAUMENI JF JR, MULVIHILL JJ, ET AL: A cancer family syndrome in twenty-four kindreds. Cancer Res 48:5358-5362, 1988
(3) MALKIN D, LI FP, STRONG LC, ET AL: Germ- line p53 mutations in a familial syndrome of breast cancer, sarcomas, and other neoplasms. Science 250:1233-1238, 1990
(4) SRIVASTAVA S, ZOU Z, PIROLLO K, ET AL: Germ-line transmission of a mutated p53 gene in a cancer-prone family with Li-Fraumeni syndrome. Nature 348:747-749, 1990
(5) TSUNEMATSU Y, WATANABE S, OKA T, ET AL: Familial aggregation of cancer from proband cases with childhood adrenal cortical car- cinoma. Jpn J Cancer Res 82:893-900, 1991
Detection of Hypoxic Cells in a Murine Tumor With the Use of the Comet Assay
Peggy L. Olive,* Ralph E. Durand
Background: Hypoxic cells within solid tumors are likely to limit tumor curability by radiation therapy and some chemotherapeutic agents. Pur- pose: To quantify a hypoxic fraction in
solid tumors, we developed a method which measures radiation-induced DNA single-strand breaks in individual tumor cells and makes use of the fact that three times more strand breaks are produced in aerobic than in hypoxic cells. Methods: Immediately after irradiation with doses of 4-20 Gy, SCCVII squamous cell carcinomas growing in C3H mice were removed and cooled, and a single-cell suspension was prepared. These cells were then embedded in agarose, lysed in an alkaline solution, subjected to electro- phoresis, and stained with a fluorescent DNA-binding dye. The amount and migration distance of damaged DNA from individual cells were scored by using a fluorescence image processing system, where differentially radio- sensitive aerobic and hypoxic cell populations resulted in bimodal damage distributions. Curve-fitting routines provided quantitative es- timates of the fraction of hypoxic cells. Results: After the mice were exposed to 10-20 Gy, the SCCVII tumors (450-600 mg) were shown to have a hypoxic fraction of 18.5% ± 10.6% (mean ± SD for 11 tumors), which compares well with the value of 11.6% observed using the paired survival curve method. Con- clusions: Our results indicate that this method, which requires only a few thousand cells, is a rapid and sensitive way to detect hypoxic cells in solid animal tumors. Implications: Estimat- ing hypoxia in accessible human tumors undergoing radiotherapy may be possible if the sensitivity of the method can be improved to allow de- tection of hypoxic cells after a dose of 2 Gy. [J Natl Cancer Inst 84:707-711, 1992]
Downloaded from http://jnci.oxfordjournals.org/ at UQ Library on June 14, 2015
Received October 21, 1991; revised January 14, 1992; accepted January 21, 1992.
Supported by Public Health Service grant CA- 37879 from the National Cancer Institute, National Institutes of Health, Department of Health and Human Services; and by the National Cancer In- stitute of Canada.
British Columbia Cancer Research Center, Van- couver, B.C., Canada.
We thank C. Vikse for technical assistance. *Correspondence to: Peggy L. Olive, Ph.D., Medical Biophysics Unit, B.C. Cancer Research Center, Vancouver, BC, V5Z 1L3, Canada.
Hypoxic cells are about three times more resistant to killing by ionizing radia- tion than are well-oxygenated cells. There is little question that hypoxic cells are present in many solid tumors in humans, and the presence of these cells can poten- tially jeopardize the success of radio- therapy and some types of chemotherapy (1-3). Whether any particular tumor con- tains hypoxic cells and whether those cells remain hypoxic throughout the course of treatment, however, are currently unknown. Moreover, the question of how best to quantify tumor hypoxia remains unresolved after decades of research, al- though there have been recent applications of new oxygen electrode technology, cryospectrophotometry, and radiolabeled chemical hypoxia markers in human tumors (4-6). Noninvasive imaging tech- niques (7-9) also show considerable promise, yet they remain subject to the uncertainties of resolution and the ability to identify biologically relevant (i.e., clonogenic) hypoxic cells.
We have recently described a method which measures DNA damage in indi- vidual mammalian cells. The “comet assay,” is based on microelectrophoresis as first described by Ostling and Johanson (10) and uses the principle that broken cellular DNA will migrate more readily in an electric field than undamaged DNA. Recent extensions of the methodology per- mit detection of DNA single-strand or double-strand breaks (11-13).
A linear relationship between damage and radiation dose has been observed (13) when using the more sensitive alkaline comet assay for single-strand break detec- tion; doses greater than about 1 Gy pro- duced significant DNA damage in individual cells. Since about three times less DNA damage is produced in anoxic than in aerobic cells when irradiated (14), it follows that damage should be detec- table in anoxic cells for doses exceeding about 3 Gy. Consequently, it should be possible to detect hypoxic cells in solid tumors, even with low doses of radiation, providing that tumor cells can be isolated and processed prior to the occurrence of significant repair, that little debris is pre- sent, and that the tumor contains a rea- sonable fraction of hypoxic cells. A practical assessment of this approach is presented here, using the murine SCCVII
squamous cell carcinoma growing in C3H mice.
Materials and Methods
SCCVII squamous cell carcinoma cells were transplanted subcutaneously over the sacral region of inbred male C3H/He mice that weighed approximately 30 g. Tumors were used for experimentation approximately 2 weeks later, when they had reached a weight of 450-600 mg. To provide a conventional estimate of hy- poxic fraction, the in vitro paired survival curve method (15) was employed as pre- viously described (16). A common ter- minal slope was used to obtain extrapolation numbers to the y axis for tumor cells from air-breathing and as- phyxiated mice, and confidence limits were obtained from the limits for these extrapolation numbers (15). For DNA damage assays, tumors were irradiated by placing unanesthetized mice in jigs and irradiating the tumor with 250-kilovolt- peak x rays, using parallel opposed fields. Immediately (within 30 seconds) follow- ing irradiation, mice were killed and tumors were rapidly excised and placed in ice-cold phosphate buffer. Preliminary experiments indicated that we could ob- tain good estimates of hypoxic fraction when tumors were excised 5 minutes after a dose of 10 Gy. A single-cell suspension was prepared by mincing the entire tumor on an ice-cold surface and filtering the suspension through a 70-um nylon mesh. Cells were diluted to a concentration of 4 x 104 cells/mL for the comet assay.
For the alkaline comet assay, a 0.5-mL cell suspension was mixed with 1.5 mL of a 1% solution of low-gelling-temperature agarose, and slides were carefully sub- mersed in an alkaline lysis solution con- taining 1 M NaCl and 0.03 M NaOH for 1 hour followed by a 1-hour wash in 0.03 M NaOH and 2 mM EDTA before electrophoresis in a fresh solution of 0.03 M NaOH and 2 mM EDTA at 0.5 V/cm (35 mA) for 25 minutes. Slides were rinsed and stained for 10 minutes in 2.5 ug/mL propidium iodide.
Individual “comets” were viewed using a Zeiss epifluorescence microscope (Carl Zeiss, Inc., Thornwood, N.Y.) attached to an intensified solid-state CCD camera from ITT Electro-Optical Prods. Div.,
Fort Wayne, Ind., and an image analysis system, an IBM AT computer and ITEX 100 board, from Imaging Technology Inc., Woburn, Mass., as previously described (11,13). The “tail moment,” proportional to the percentage of DNA in the tail mul- tiplied by the tail length, was used as the measure of DNA damage. Histograms of tail moments from 100 to 400 comets were constructed and analyzed mathe- matically by determining the best-fit parameters for two normal distributions with means separated by a factor of 1.8- 2.8 (the anticipated damage differential due to oxygen). Inadequate resolution is available to identify cells at intermediate oxygenation levels. To achieve the fitting, we used the Marquardt Levenburg algo- rithm, an interactive procedure.
Results
For comparative purposes, the fraction of radiation-resistant hypoxic cells in SCCVII tumors (0.45-0.6 g) was firse determined by the in vitro paired survival curve method, where the ratio of high dose survivals for cells from air-breathing and asphyxiated animals indicated hypoxic clonogenic fraction of 11.6% with 95% confidence limits of 5.6% 27.7% (Fig. 1). This value was deter
1.0
Surviving Fraction
asphyxiated
0.1
0.01
air-breathing
in vitro
0.001
0
10
20
30
Dose (Gy)
mined by fitting the 32 individual data points (analyzed for three doses) on the air-breathing curve and 12 data points on the asphyxiated curve, using a common slope, and determining the range in ex- trapolation numbers as described by Moulder and Rockwell (15).
Using the alkaline comet assay, we measured radiation-induced DNA dam- age in SCCVII tumor cells. For each ex- periment, at least 100 individual cells were analyzed for the distance of DNA migration and the fraction of DNA which migrated. The tail moment, a descriptor which reflects the fraction of DNA re- leased from the cell nucleus and the dis- tance migrated in the electric field, was used as a measure of DNA damage (11,13). Previous results using the comet assay have indicated a linear increase in tail moment as dose increases (13); the rate of increase is dependent on the choice of electrophoresis conditions. Average tail moment was significantly decreased in tumor cells of mice as- phyxiated before irradiation with doses ranging from 4 to 20 Gy (Fig. 2). In air- breathing mice, a subpopulation of less damaged tumor cells was clearly visible.
It should be possible to define hypoxic cells in a mixed population as comets having tail moments equivalent to those for anoxic cells receiving the same radia- tion dose. A technical problem arises, however, in that it is not possible to en- sure the same degree of damage or “re- pair” for each tumor excised. Differences in tumor temperature and in time before excision influence repair and small differ- ences in electrophoresis conditions (buff- er, voltage, and temperature) may alter the apparent initial damage. Therefore, the average amount of DNA damage (and consequently the tail moments) could vary for individual animals. For this reason, a mathematical analysis is preferable. Although this procedure is not directly dependent on the total amount of damage, it does include the implicit assumptions that a bimodal dis- tribution of damage is produced and that the histograms are normally distributed for both aerobic and hypoxic populations.
The proportion of these less damaged, presumably hypoxic cells was estimated by a curve-fitting program which con- strained the oxygen enhancement ratio to a value between 1.8 and 2.8-the damage
differential routinely observed for a three- fold difference in radiation dose for these electrophoresis conditions (damage for unirradiated cells was not subtracted). Results shown in Fig. 3 indicate good reproducibility for repetitive measure- ments from the same tumor cell suspen- sion. In 11 tumors exposed to 10-20 Gy,
the hypoxic fraction (mean ± SD) was calculated to be 18.5% ± 10.6%. Values obtained for six tumors using a 4-Gy dose indicate a higher (but not significantly different) hypoxic fraction of 27.4% ± 11.0%. Both of these values are some- what higher than the 11.6% hypoxic population measured using the paired sur-
15
4 Gy, air
10 Gy, air
20 Gy, air
10
Number of Comets
5
0
30
4 Gy, asphyx
10 Gy, asphyx
20
Gy, asphyx
20
10
0
0
10
20
0
10
20
0
10
20
Tail Moment
Number of Comets
18
Number of Comets
18
15
File 7
Anox= 14±2
15
Oxic= 86±5
File 10
12
OER= 2.8
12
9
9
Anox= 7±3
0xlc= 93±6
6
6
OER= 2.6
3
3
0
n
0
0
10
20
30
40
50
0
0
10
20
30
40
50
Number of Comets
18
Relative Tail Moment
15
File 8
Anox=16±4
12
Oxic=84+7
60
9
OER= 2.3
6
Number of Comets
50
Composite
3
40
0
0
10
20
30
40
50
Anox= 13±2
Number of Comets
Oxic= 87±3
18
30
OER= 2.4
15
File
Anox= 15±4
12
0xlc= 85±8
OER= 2.3
20
9
6
10
3
0
0.00.0.0 0 0
0
m
Л
n
10
20
30
20
30
50
0
10
40
50
0
40
Relative Tail Moment
Relative Tail Moment
vival curve method. The large range in hypoxic fraction and the use of different tumors for the two techniques prevent any firm conclusions about consistent dif- ferences between assays. It is also pos- sible that not all hypoxic cells detected using the comet assay can grow in the in vitro clonogenic assay. Alternatively, the rapid nonenzymatic dissociation proce- dure may release hypoxic cells more readily than well-oxygenated cells lo- cated in healthy stroma (15), again result- ing in a larger hypoxic fraction for the comet assay compared with the in vitro clonogencity method.
The ability of the comet assay to detect subsets of cells in known mixtures was examined by mixing aerobic cells exposed to 15 Gy with aerobic cells exposed to 5 Gy, to mimic an oxygen enhancement ratio of 3.0. The automated analysis pro- gram can identify hypoxic cells when these cells are present at proportions greater than 5% (Fig. 4). In addition to analysis of data using the image process- ing system, three investigators examined the same coded slides and scored comets as high or low in damage, based on comet tail length and fraction of DNA in the tail. Results shown in Fig. 4 are promising and
100
Observed % Hypoxic Cells
×
#3
#2
50
X
#1
FIPS
X
0
0
50
100
Expected % Hypoxic Cells
suggest that, at least for higher doses of radiation, it may be possible to quantify the hypoxic fraction without the use of an imaging system. In addition to increasing the versatility of the method, scoring slides by eye would increase the number of comets analyzed per hour and should, therefore, improve sensitivity for detect- ing low hypoxic fractions.
Discussion
The alkaline comet assay can detect hypoxic cells with good sensitivity, using only a small sample size. The accuracy of quantifying hypoxic cells naturally de- pends on the number of comets analyzed. Using a sample of only 100 cells, we might expect to reliably detect hypoxic fractions of 10% or greater. Scoring 100 comets requires about 20 minutes for identification, focus, capture, and image analysis; but an alternative method is simply to score comets by eye, using a fluorescence microscope. Results with viewers scoring slides “blind” indicated a good correlation between the expected and the measured hypoxic fraction using a 15-Gy exposure (Fig. 4). Doses of radiation as low as 4 Gy can be used to detect hypoxic cells, although resolution at these low doses is likely to be reduced, and the problem of repair time becomes more critical.
There are several advantages to the use of this method to detect hypoxic cells. It is considerably more rapid technically than conventional methods (e.g., paired survival curve and tumor growth delay) used to measure hypoxic fraction in murine tumors. The assay requires only a few thousand cells; therefore, it should be possible to use fine-needle aspirates and to perform multiple measurements over the course of radiotherapy. Combining multiple fine-needle aspirates would be feasible to ensure that a representative tumor sample is analyzed. The method does not rely on the preferential binding of a hypoxia marker and, thus, avoids problems of blood flow and drug delivery (17), binding to stromal and other normal cells (18), and nitroreductase activity of the tumor cells (18,19). Radiation damage itself is detected, and tumor cells, both acutely and chronically hypoxic at the time of radiation, are included in the analysis. By analysis of individual cells,
the extent of damage is not critical, since the response of the well-oxygenated cells can be used as an internal control for the hypoxic cells. Analysis of individual cells is also important if tumors contain cells at intermediate oxygenation.
Another important advantage of the comet method is that the number of DNA single-strand breaks produced by ionizing radiation is independent of cell cycle phase, cell size, or cell type (11). Inves- tigators (14) have shown that the relation- ship between the oxygen content of a cell and its radiosensitivity is similar, if not identical, for single-strand breakage and the more relevant clonogenic potential. Since DNA content is measured for each comet, it is also possible to separate the response of tetraploid SCCVII tumor cells from diploid nontumor cells (although we did not do so here). Even without thisg refinement, the presence of host cells ise not likely to be a limitation to the applica-& tion of this method. Both host and tumore. cells can be found in hypoxic regionsg (16), and both will be less responsive toz radiation and, therefore, indicative of tumor hypoxia.
Downloaded from http://jnci.oxfordjournals.org/ at UQ Library on June 14, 2015
There are, however, limitations to this method for detecting hypoxic cells ine human tumors. The most significant lim- itation is that a biopsy (fine-needle aspi- rate) is required immediately or shortlya after exposure to a dose of radiation in= excess of 4 Gy. We are exploring the ap-o plication of different types of agarose ande electrophoresis conditions which might allow detection of a hypoxic fraction afters a clinical dose of 2 Gy. A second limita- tion is that the curve-fitting program im- poses specific assumptions which cang have a significant impact on the con- clusions reached. A third problem is that information on the clonogenic potential of the hypoxic cells is not provided. With respect to this last limitation, it should be possible to at least assess viability, using fluorescent dyes which are excluded from intact cells or which must be metabolized by viable cells. Fluorescence-activated cell sorting, using fluorescent viability probes, can easily provide the small num- ber of cells required. While the feasibility of using the comet assay for human tumors remains to be explored, our results with measurement of a hypoxic fraction in an animal tumor model suggest consider- able potential.
References
(1) OVERGAARD J: Sensitization of hypoxic tumour cells-clinical experience. Int J Radiat Biol 56:801-811. 1989
(2) DISCHE S: Hypoxia and local tumour control. Part 2. Radiother Oncol 20(Suppl 1):9-11, 1991
(3) GONZÁLEZ DG: Hypoxia and local tumour control. Part 1. Radiother Oncol 20(Suppl 1):5-7, 1991
(4) MUELLER-KLIESER W, SCHLENGER KH, WALENTA S. ET AL: Pathophysiological approaches to identifying tumor hypoxia in patients. Radio- ther Oncol 20(Suppl 1):21-28, 1991
(5) URTASUN RC, KOCH CJ, FRANKO AJ, ET AL: A novel technique for measuring human tissue pO2 at the cellular level. Br J Cancer 54:453- 457,1986
(6) CHAPMAN JD: Measurement of tumor hypoxia by invasive and non-invasive procedures: A review of recent clinical studies. Radiother Oncol 20(Suppl 1):13-19, 1991
(7) MIRALDI F: Potential of NMR and PET for determining tumor metabolism. Int J Radiat Oncol Biol Phys 12:1033-1039, 1986
(8) GATENBY RA, KESSLER HB, ROSENBLUM JS, ET AL: Oxygen distribution in squamous cell carcinoma metastases and its relationship to outcome of radiation therapy. Int J Radiat Oncol Biol Phys 14:831-838, 1988
(9) ROSEN GM, HALPERN HJ, BRUNSTING LA, ET AL: Direct measurement of nitroxide pharma- cokinetics in isolated hearts situated in a low- frequency electron spin resonance spectrom- eter: Implications for spin trapping and in vivo oxymetry. Proc Natl Acad Sci USA 85:7772- 7776, 1988
(10) OSTLING O, JOHANSON KJ: Microelectro- phoretic study of radiation-induced DNA damages in individual mammalian cells. Biochem Biophys Res Commun 123:291- 298, 1984
(11) OLIVE PL, BANÁTH JP, DURAND RE: Heterogeneity in radiation-induced DNA da- mage and repair in tumor and normal cells measured using the “comet” assay. Radiat Res 122:86-94, 1990
(12) OLIVE PL. BANÁTH JP, DURAND RE: Detection of etoposide resistance by measuring DNA damage in individual Chinese hamster cells. J Natl Cancer Inst 82:779-783, 1990
(13) OLIVE PL, WLODEK D, DURAND RE, ET AL: Factors influencing DNA migration from in- dividual cells subjected to gel electrophoresis. Exp Cell Res 198:259-267, 1992
(14) CHAPMAN JD, DUGLE DL, REUVERS AP, ET AL: Studies on the radiosensitizing effect of oxygen in Chinese hamster cells. Int J Radiat Biol Relat Stud Phys Chem Med 26:383- 389, 1974
(15) MOULDER JE, ROCKWELL S: Hypoxic fractions of solid tumors: Experimental techniques, methods of analysis, and a survey of existing data. Int J Radiat Oncol Biol Phys 10:695- 712, 1984
(16) OLIVE PL: Distribution, oxygenation, and clonogenicity of macrophages in a murine tumor. Cancer Commun 2:93-100, 1989
(17) BLASBERG R, HOROWITZ M, STRONG J, ET AL: Regional measurements of [14C]misonidaozle distribution and blood flow in subcutaneous RT-9 experimental tumors. Cancer Res 45:1692-1701, 1985
(18) FRANKO AJ: Misonidazole and other hy- poxia markers: Metabolism and applica- tions. Int J Radiat Oncol Biol Phys 12:1195-1202, 1986
(19) OLIVE PL, CHAPLIN DJ: Oxygen and nitro- reductase dependent binding of AF-2 in spheroids and murine tumors. Int J Radiat Oncol Biol Phys 12:1247-1250, 1986
Various Methods of Analysis of mdr-1/P-Glycoprotein in Human Colon Cancer Cell Lines
Cynthia E. Herzog,* Jane B. Trepel, Lyn A. Mickley, Susan E. Bates, Antonio T. Fojo
Background: Multidrug resistance (MDR) mediated by high levels of mdr- I (also known as PGY1)/P-glycoprotein (Pgp) has been studied in tissue culture systems; however, most tumor samples which express mdr-l/Pgp have much lower levels. Purpose: We wanted to determine if levels seen clinically could be detected by commonly used methods and to determine if these levels con- ferred MDR reversible by Pgp antag- onists. Methods: We studied multi- drug-resistant cell lines and sublines with levels of mdr-l/Pgp expression comparable to those seen clinically. We evaluated the expression of mdr-I RNA by Northern blot analysis, slot blot analysis, polymerase chain reaction (PCR) analysis, and in situ hybridiza- tion. We evaluated protein expression by immunofluorescence, immunohis- tochemistry, fluorescence-activated cell sorting, and immunoblotting analyses. Drug resistance and reversibility were determined by cell growth during con- tinuous drug exposure. Results: In most cases, the low level of mdr-l/Pgp present in these cell lines could be detected by each method, but the as- says were at the limit of sensitivity for all methods except the PCR method. These low levels of mdr-l/Pgp are capable of conferring MDR, which can be antagonized by verapamil. Con- clusions: Levels of mdr-l/Pgp similar to those found in clinical samples can be detected by each of these methods, but the PCR method was the most sensitive and most reliably quantitative. Implica-
tions: In vitro sensitization by the addi- tion of verapamil in cell lines with these low levels of mdr-1/Pgp suggests that clinically detected levels may confer drug resistance in vivo. [J Natl Cancer Inst 84:711-716, 1992]
Effective chemotherapy is limited by drug resistance, which can occur de novo or at relapse. Resistance to several drugs is often acquired simultaneously-a phenom- enon termed multidrug resistance (MDR). P-glycoprotein (Pgp), a membrane phos- phoglycoprotein encoded by the mdr-1 gene (also known as PGY1), is an energy- dependent efflux pump, which confers resistance to numerous, naturally occur- ring hydrophobic agents (1-3). This resis- tance has been successfully reversed by several drugs, including verapamil (4,5). Although the clinical role of Pgp is un- known, its expression in tumors, either de novo or after chemotherapy (6-11), makes it a potential target for improved cancer therapy.
With the advent of clinical trials block- ing Pgp, determination of mdr-1/Pgp in samples is critical. To compare common- ly used methods, we studied eight cell lines expressing low levels of mdr-1/Pgp. The levels of mdr-1/Pgp in most clinical samples fall within the range of expres- sion observed in these cell lines. In this study, we characterize the drug resistance in these eight cell lines and evaluate four methods for detecting RNA and four methods for detecting protein. We used the KB 3-1 subclone of KB cells as a negative control.
Materials and Methods
Deoxycytidine 5’-[c .- 32P]triphosphate (3000 Ci/mmol) and uridine 5’-[a-32P]triphos- phate (3000 Ci/mmol) were from DuPont/ NEN Products (Boston, Mass.). [35S] Uridine 5’-(a-thio)triphosphate and 1251-
Received October 9, 1991; revised December 30, 1991; accepted January 21, 1992. -
C. E. Herzog (Pediatric Branch), J. B. Trepel, L. A. Mickley, S. E. Bates, A. T. Fojo (Medicine Branch), Division of Cancer Treatment, National Cancer Institute, Bethesda, Md.
*Correspondence to: Cynthia E. Herzog, M.D., National Institutes of Health, Bldg. 10, Rm. 13N240, Bethesda, MD 20892.