Proc. Nat/. Acad. Sci. USA
`Vol. 92, pp. 5950-5954, June 1995
`Genetics
`
`Identification by representational difference analysis of a
`homozygous deletion in pancreatic carcinoma that lies
`within the BRCA2 region
`MIEKE SCHUTIE*, LUIS T. DA COSTAt, STEPHAN A. HAHN*, CHRIS MOSKALUK*, A. T. M. SHAMSUL HOQUE*,
`ESTER ROZENBLUM*, CRAIG L. WEINSTEIN*, MICHAEL BITTNER*, PAULS. MELTZER*, JEFFREY M. TRENT*,
`CHARLES J. YEo§, RALPH H. HRUBAN*, AND SCOTT E. KERN*~II
`Departments of *Pathology, §surgery, and 10ncology, and trhe Division o.f Toxic?logy and Program in Human Genetics, The Johns J:Iopkins Medical
`Institutions, Baltimore, MD 21205-2196; and *Laboratory of Cancer Genetics, National Center of Human Genome Research, The Natwnal
`Institutes of Health, Bethesda, MD 20892
`
`Communicated by Bert Vogelstein, The Johns Hopkins Oncology Center, Baltimore, MD, March 24, 1995
`
`Homozygous deletions have been central to
`ABSTRACT
`the discovery of several tumor-suppressor genes, but their
`finding bas often been either serendipitous or the result of a
`directed search. A recently described technique [Lisitsyn, N.,
`Lisitsyn, N. & Wigler, M. (1993) Science 259, 946-951] held out
`the potential to efficiently discover such events in an unbiased
`manner. Here we present the application of the representa(cid:173)
`tional difference analysis (RDA) to the study of cancer. We
`cloned two DNA fragments that identified a homozygous
`deletion in a human pancreatic adenocarcinoma, mapping to
`a 1-centimorgan region at chromosome 13q12.3 flanked by the
`markers D13S171 and D13S260. Interestingly, this lies within
`the 6-centimorgan region recently identified as the BRCA2
`locus of heritable breast cancer susceptibility. This suggests
`that the same gene may be involved in multiple tumor types
`and that its function is that of a tumor suppressor rather than
`that of a dominant oncogene.
`
`Tumor-suppressor genes play a crucial role in the control of
`cell growth and differentiation. Loss of the function of tumor(cid:173)
`suppressor genes is part of the cascade of genetic alterations
`which drive tumorigenesis (1). The biallelic inactivation of a
`tumor-suppressor gene typically involves an intragenic change
`(nucleotide substitution, small insertion, or microdeletion)
`within one allele, combined with inactivation ofthe other allele
`through the loss of a large chromosomal region. Although
`infrequent, sizable deletions involving both alleles have been
`observed. Such homozygous deletions have contributed to the
`discovery of several tumor-suppressor genes (RBl, DCC, and
`p16) (2-5).
`Despite the fact that pancreatic adenocarcinoma is one of
`the more common human cancers ( 6), little is known of the
`genetic alterations in these tumors. One of the reasons is that
`the tumors generally are diagnosed at a late stage of tumori(cid:173)
`genesis. This, together with the aggressive clinical course,
`severely limits the number of resected specimens available for
`research. Also, pancreatic adenocarcinomas characteristically
`exhibit an exuberant host desmoplastic response, resulting in a
`high admixture of nonneoplastic cells and hampering the
`molecular genetic analysis of primary tumor samples (7).
`Finally, familial patterns of pancreatic adenocarcinoma usually
`do not involve young ages of onset, high penetrance, or
`extensive pedigrees (8).
`We have circumvented some of these problems by the
`development of a xenograft model of pancreatic adenocarci(cid:173)
`noma that generates genetically stable cell expansions, free of
`infiltrating nonneoplastic human cells (9, 10). Molecular anal(cid:173)
`ysis of known oncogenes and tumor-suppressor genes has
`
`The publication costs of this article were defrayed in part by page charge
`payment. This article must therefore be hereby marked "advertisement" in
`accordance with 18 U.S.C. §1734 solely to indicate this fact.
`
`proven feasible; it is possible to identify both K-ras and p16
`alterations in over 80% of pancreatic adenocarcinomas (7, 10,
`11) and p53 mutations in at least 70% of the cases (12).
`However, a conventional search for novel loci of interest
`presented practical obstacles. Allelotyping had identified fre(cid:173)
`quent loss of heterozygosity (LOH; deletion of only one allele),
`mainly at sites of known genes, such as 9p (p16), 17p (p53), and
`18q (DCC) (7, 10). A limited number of xenografted speci(cid:173)
`mens, and the typically large areas involved by LOH, precluded
`a standard search for smaller consensus areas of deletion. An
`alternative approach for the identification of tumor-suppressor
`genes preferably would allow high-resolution genome scanning
`without the need for a statistical analysis of numerous tumor
`specimens. The newly described technique of representational
`difference analysis (RDA) (13) suggested a promising ap(cid:173)
`proach.
`RDA is a means for isolating DNA fragments that are
`present in only one of two nearly identical complex genomes.
`It utilizes a subtractive hybridization method but differs from
`conventional methods (14-16) by using "representations" of
`the genomes that have a reduction in complexity. Represen(cid:173)
`tations are generated by a PCR-based size selection applied to
`the restriction fragments of both genomes. Moreover, RDA
`takes advantage of both subtractive hybridization and DNA
`reassociation kinetics to favor the reiterated PCR amplifica(cid:173)
`tion of the difference among the two genomes. It has been
`demonstrated that RDA can enrich difference products over
`a millionfold after three rounds of selection (13).
`Here we apply RDA to the identification of DNA fragments
`that are deleted in neoplastic tissues. Normal tissue from the
`patient provides the "tester" sequences, and neoplastic cells
`provide the "driver" sequences in the hybridization reactions.
`RDA identifies a simple LOH, when a deletion involves a
`restriction fragment length polymorphism in such a way that
`the smaller fragment is deleted in the neoplasm and therefore
`is present only in the representation of the tester (normal)
`genome. Due to the PCR-based size exclusion, the larger allele
`is not present in either ofthe representations, and the 2:1 allele
`ratio seen upon comparison of the total genomic DNAs of
`normal and tumor is converted to a 1:0 ratio in the represen(cid:173)
`tations. Thus the existence of the larger allele in the driver will
`no longer prevent enrichment for the smaller allele in the tester
`(the "target," or deleted sequence in the tumor) (Fig. 1). In
`homozygously deleted regions, however, both alleles are ab(cid:173)
`sent from the driver genome and thus the target alleles do not
`
`Abbreviations: LOH, loss of heterozygosity; RDA, representational
`difference analysis; eM, centimorgan(s); STS, sequence-tagged site;
`Y AC, yeast artificial chromosome; FISH, fluorescence in situ hybrid(cid:173)
`ization.
`liTo whom reprint requests should be addressed at: The Johns Hopkins
`Medical Institutions, 628 Ross Building, Baltimore, MD 21205-2196.
`
`5950
`
`GeneDX 1024, pg. 1
`
`

`

`Genetics: Schutte et al
`
`Proc. Natl. Acad. Sci. USA 92 (1995)
`
`5951
`
`Homoz.
`Homoz.
`LOH
`LOH
`LOH
`Del.
`Del.
`A T N TN TN B TN TN
`
`t::=:2t:::::::t~c:;;:;:lc;:;:!~c=tc:::;:;3:
`
`c::::J~~~c::::l
`
`----------
`------ - 1.5 kb
`---------- - ------- - 0.1 kb
`
`Not detectable by RDA
`
`Detectable by RDA
`
`Identification by RDA of DNA sequences deleted in
`FIG. 1.
`tumors. The figure is a schematic representation of specific loci within
`electrophoretically separated restriction endonuclease-digested total
`genomic DNA from tumor (T) and corresponding normal tissue (N).
`The area between the broken lines depicts the PCR-based size
`selection, resulting in "representations" of the genomes. Hom oz. Del.,
`homozygous deletion. (A) RDA cannot identify losses of single alleles
`when the DNA fragments are nonpolymorphic in restriction fragment
`length (lanes 1 and 2), nor can it identify simple LOH wherein the
`remaining DNA fragment of the tumor lies within the boundaries of
`the size selection (lanes 4 and 5). Most DNA fragments in a region of
`homozygous deletion will lie outside the size selection area and
`therefore cannot be recovered by RDA (lanes 7 and 8). (B) RDA
`identifies DNA fragments at a site of simple LOH if the deletion
`involves the smaller fragment of a restriction fragment length poly(cid:173)
`morphism (lanes 1 and 2), and this technique detects a homozygously
`deleted DNA fragment provided that it lies within the representation
`(lanes 4 and 5).
`
`need to be polymorphic in restriction fragment length in order
`to be detectable by RDA.
`It can be reasoned that RDA would strongly favor the
`enrichment of homozygously deleted regions over areas of
`heterozygous loss in the tumor, allowing the identification of
`homozygous deletions even among the usually high back(cid:173)
`ground of LOH found in many malignancies. Assuming a
`polymorphism frequency in the human genome of 1 in 300 bp,
`and a necessity for the loss of the smaller of the two restriction
`fragments (half of the sites of LOH), the efficiency ratio for the
`identification by RDA of deleted fragments (comparing those
`within a homozygous deletion versus those within a site of
`simple LOH) will be 50:1 when a restriction endonuclease
`requiring a 6-bp recognition site at both ends of a fragment is
`used. That is, loss of a random DNA sequence should be
`detectable by RDA at least 50 times more often if the loss
`produces a homozygous deletion rather than simple LOH.
`Here we describe the identification of a homozygous dele(cid:173)
`tion in a pancreatic adenocarcinoma, using RDA. The ho(cid:173)
`mozygous deletion mapped to a 1-centimorgan (eM) region at
`chromosome 13q, flanked by the markers D13Sl71 and
`D13S260. The premise that a tumor-suppressor gene might be
`located within the region of the homozygous deletion is
`strengthened by the localization of the recently identified
`BRCA2 locus for heritable breast cancer susceptibility (17),
`which currently encompasses the entire region of the homozy(cid:173)
`gous deletion.
`
`MATERIALS AND METHODS
`Case Report. An 84-year-old woman presented with painless
`obstructive jaundice and was found to have a mass in the head
`of the pancreas without evidence of metastases. Her medical
`history included a right-sided colon carcinoma curatively
`resected at the age of 61. Her family history included multiple
`incidents of adenocarcinoma, including her mother, who had
`an adenocarcinoma of the colon resected and who died of
`breast carcinoma at age 80, her mother's sister, who died of
`
`breast carcinoma at age 94, her mother's brother, who died of
`"stomach" cancer in his 80s, and the patient's brother, who
`died of colorectal carcinoma at the age of 52. The only siblings
`in these two generations unaffected by cancer were the pa(cid:173)
`tient's sister (alive, age 76) and her mother's sister, who died
`at the age of 29 from tuberculosis. Both children of the patient
`are unaffected to date.
`Tissue Samples. Tissue specimens were obtained from the
`pancreas upon its resection at The Johns Hopkins Hospital.
`Histopathological examination revealed a moderately differ(cid:173)
`entiated primary pancreatic ductal adenocarcinoma. The pan(cid:173)
`creas cancer was histologically distinct from her previous
`colorectal carcinoma, slides of which were reviewed. At the
`time of surgery, normal duodenal mucosa was fresh-frozen at
`-80°C and xenografts were generated by implantation of
`2-mm3 pieces of the primary tumor into athymic nude mice.
`Xenografts were harvested at a size of 1 cm3, and DNA was
`prepared as described (10).
`RDA. RDA was performed essentially as described by
`Lisitsyn et al (13). The restriction endonuclease BamHI and
`corresponding anchor primers were used for digestion of the
`DNA samples and subsequent PCR amplifications. For the
`xenograft-driven RDA, hybridization times were increased to
`40 hr. A detailed protocol of the RDA procedure is available
`from the authors.
`The RDA round 2 difference products were cloned by using
`the pBluescript II plasmid vector (Stratagene ). Insert DNAs of
`individual clones were used as probes for Southern blots
`containing tester and driver amplicon DNA. These fragments
`were sequenced by the SequiTherm cycle sequencing method
`(Epicentre Technologies, Madison, WI) and 20-mer or 24-mer
`oligonucleotide pairs for sequence-tagged sites (STSs) were
`designed from these results.
`PCR. STSs were amplified by using 40 ng of genomic DNA
`in 67 mM Tris·HCl, pH 8.8/4 mM MgCh/16 mM
`(N~)zS04/10 mM 2-mercaptoethanol containing bovine se(cid:173)
`rum albumin at 100 #Lg/ml, dATP, dCTP, dGTP, and dTTP at
`200 ,...M each, each primer at 1 ,...M, and 2 units of Taq DNA
`polymerase (GIBCO/BRL) in a final reaction volume of 15~A-l.
`The enzyme was added after a preheating step of 2 min at 94°C.
`For 20-mers, 35 cycles of 94°C for 30 sec, 58°C for 1 min, and
`72°C for 1 min were followed by a final extension of 5 min at
`72°C. For 24-mers, the annealing step was omitted and the
`extension step was increased to 2 min. Primer sequences for
`DPCJ were 5' -CAGGTCTGAAACGTATAAAGG-3' and 5'(cid:173)
`GAGTCAAGGTAGGCTACTTC-3', and for DPC2, 5'-CTT(cid:173)
`CCCCAGTGCTTCTAATG-3' and 5'-CTCTCCTCATCTC(cid:173)
`TATTTCG-3'. Primer sequences for DPCJ' were 5'-TTCT(cid:173)
`CCATCTTCCCACCTAACAGG-3' and 5' -ATCAGCCATC(cid:173)
`TTGGCAGCAACTAG-3', and for DPC2', 5'-AAGCTTCC(cid:173)
`CCAGTGCTTCTAATGC-3' and 5'-TTTCCACGTAGGC(cid:173)
`TGTTGGTGTAG-3'. Primer sequences for LCOJ were 5'(cid:173)
`GCCTCCGGTAGGCTTTATTC-3' and 5'-GAGCGAGAC(cid:173)
`ACAGGGATTTG-3'. Dinucleotide markers and the Gene(cid:173)
`thon mega Y AC library were purchased from Research Ge(cid:173)
`netics (Huntsville, AL).
`
`RESULTS
`RDA. We performed RDA on a human pancreatic adeno(cid:173)
`carcinoma, essentially as described (13). The strategy is sche(cid:173)
`matically represented in Fig. 2. Tumor DNA was used to drive
`the subtractions, whereas corresponding normal DNA was
`used as the tester. Tissue from primary tumors, typically
`infiltrated with nonneoplastic cells, should not effectively drive
`the subtractions. We therefore used a carcinoma that had been
`propagated in an athymic nude mouse. Such xenografted
`tumors are genetically stable and do not contain detectable
`nonneoplastic human cells (9, 10). As these xenografts contain
`
`GeneDX 1024, pg. 2
`
`

`

`5952
`
`Genetics: Schutte et al
`
`Proc. Nat/. Acad. Sci. USA 92 (1995)
`
`c
`0
`~ ~
`Q. Q.
`E E ~ N
`c( c( "0 "0
`:I
`0
`
`0
`
`c ~ 0
`~ ~ c c * :I
`
`.~
`{!!. a: a:
`Q
`c( c(
`;,: ~ Q Q
`a: a:
`
`A
`
`Q
`
`Subtraction
`Clones
`
`Non(cid:173)
`Subtraction
`Clones
`
`===
`
`F
`
`Chromosomal Localization:
`PCR of Somatic Cell Hybrids
`FISH
`VAC address
`
`~-;-;H-~m-e~-.z-. --LO_H__,
`
`FIG. 2. Schematic strategy for analysis of RDA-generated clones.
`(A) RDA is performed. (B) Difference products are cloned by using
`a plasmid vector, and individual clones are picked. (C) Clones are
`evaluated by using them as probes in multiple Southern blots con(cid:173)
`taining driver (Dv) and tester (Ts) amplicons. Subtraction clones are
`thpse present in the tester but absent from the driver; the nonsub(cid:173)
`traction clones represent the background nontarget sequences which
`escape RDA enrichment. (D) Subtraction clones are sequenced and
`STS primer pairs are designed from separate positions within each
`sequence. (E) STS primer~ are used in PCR to evaluate the original
`total genomic DNA samples of tumor (T) and normal tissue (N) to
`excl~de those clones representing simple LOH. (F) STSs that identify
`sites of homozygous deletion are used in chromosomal localization
`techniques and yeast artificial chromosome (YAC) contig generation.
`FISH, fluorescent in situ hybridization.
`·
`
`up to 50% murine cells, we modified the RDA protocol of
`Lisitsyn et al by increasing the time of DNA annealing to 40 hr.
`Genomic representations of the xenograft and normal DNA
`were generated by using the restriction endonuclease BamHI.
`After two rounds of RDA, a distinct pattern of DNA fragments
`was visible upon electrophoretic separation of the difference
`product (Figs. 2A and 3A). The round 2 difference product was
`cloned by using a plasmid vector (Fig. 2B). True subtraction
`fragments were detected by using Southern blots of the. tester
`and driver representations (Fig. 2C). This analysis revealed
`that >80% of 60 randomly selected fragments were subtrac(cid:173)
`tion products-i.e., they were absent from driver and present
`in tester.
`The sequences of the cloned fragments were used to design
`primers to amplify STSs (Fig. 2D). Fourteen of 16 STSs derived
`from unique subtraction fragments were present in normal and
`xenograft total genomic DNA, consistent with sites of simple
`LOH in the carcinoma (Fig. 2£). Two STSs, designated DPCJ
`and DPC2, were present in normal but absent from xenograft
`DNA, indicating that they were homozygously deleted in the
`pancreatic carcinoma. As a control for DNA quality, duplex
`PCR was performed for both DPCJ (Fig. 3B) and DPC2 with
`concurrent use of STS primers for an irrelevant locus (LCOJ),
`
`A
`
`2 3 4 L
`
`B
`
`2 3
`
`L
`
`- <
`
`>
`
`(A) RDA of the pancreatic carcinoma xenograft. Lane 1,
`FIG. 3.
`PCR-generated amplicon of the xenograft (driver); lane 2, amplicon
`of normal DNA (tester); lanes 3 and 4, difference product after first
`and second round of hybridization-amplification, respectively; lane L,
`1-kb DNA ladder (GIBCO/BRL). The arrowhead marks 510 bp. (B)
`Duplex PCR analysis with the concurrent use of the STS primer pairs
`for DPCJ and for an irrelevant locus (LCOJ) which serves as a positive
`control for PCR. Lane 1, normal DNA as template; lane 2, xenograft
`DNA as template; lane 3, template-negative control; lane L, 1-kb DNA
`ladder. Arrowhead indicates the amplification product of the STS DPCJ.
`
`which localized to chromosome 14. To exclude simple inser(cid:173)
`tion/deletion polymorphisms, an adjacent sequence of each
`cloned fragment was amplified with additional STS primers,
`designated DPCJ' and DPC2'.
`As a control, we performed a parallel RDA in which the
`driver. DNA was provided by a cell line derived from the same
`pancreatic carcinoma. Seven of 8 unique subtraction fragments
`from this RDA had been identified in the xenograft-driven
`RDA. These fragments included DPCJ.
`Localization and YAC Contig. The STSs DPCJ and DPC2
`both localized to chromosome 13 upon PCR analysis of
`monochromosomal somatic cell hybrid DNAs of NIGMS
`mapping panel2 (Coriell Cell Repositories, Camden, NJ) (18).
`Both subtraction fragments, DPCJ and DPC2, were used to
`screen a chromosome 13 pnage library (American Type Cul(cid:173)
`ture Collection). Two-color FISH, using the whole phage
`DNAs as probes, localized DPCJ and DPC2 as distinct non(cid:173)
`overlapping nearby sites on a metaphase preparation, below
`the centromere of chromosome 13 (Fig. 2F).
`PCR screening of the Genethon megaYAC library (19)
`resulted in a Y AC contig, encompassing the BRCA2 region at
`13q12-13. YAC y886d8 contained both DPCJ and DPC2 and
`the marker D13S171. YAC y951a3 col)tained DPCJ and the
`markers D13S171 and D13S267, whereas y931f4 contained
`DPC2, D13S260, and D13S290. Five additional YACs con(cid:173)
`firmed the contig (Fig. 4). Analysis with the markers D13S289,
`S290, S260, Sl71, S267, S219, and S220 did not reveal inter(cid:173)
`stitial deletions within these YACs. Y ACs suspected to be
`chimeric, on the basis of Genethon data and our own data,
`were excluded from the contig.
`Dinucleotide markers D13S289, S290, S260, Sl71, S267,
`S219, and S220 in this region were all found to be present in
`the xenograft DNA, exhibiting a pattern of simple LOH upon
`comparison with normal DNA. Thus the entire homozygous
`deletion in the carcinoma mapped between the markers
`Dl3Sl71 and D13S260 at band 13q12.3 (Fig. 4). PCR analysis
`for the candidate tumor-suppressor genes Brush-1 (21) and
`RFC3 (22) revealed the expected PCR products in xenograft
`DNA. None of the eight Y ACs in the contig contained the
`Brush-] sequence. Microsatellite instability was not identified
`at any locus in the carcinoma.
`
`GeneDX 1024, pg. 3
`
`

`

`Genetics: Schutte et al
`
`Centromere
`
`ly755c1 t y753d5
`
`y951a3
`
`0138289
`0138290
`
`3cM
`
`0138260
`1 eM
`0138171
`
`2cM
`
`0138267
`
`2cM
`0138219,0138220
`
`- 20cM
`
`RB1
`
`Telomere
`
`FIG. 4. Schematic map of the region 13q12-13 and flanking
`markers. The gray area represents the DPC region homozygously
`deleted in the pancreatic carcinoma. The positions of D13S219, S220,
`S267, S171, S260, S289, and S290 markers and their genetic distances
`in eM (labeled on the heavy line) were adapted from the 1993-94
`Genethon human linkage map (20). The positions of STSs DPCJ and
`DPC2, which lie within the homozygous deletion of the pancreatic
`carcinoma, and the order of D13S289 and S290, are based on studies
`of the YAC contig (light lines) and on the demonstrated presence of
`one remaining allele ofthe D13S219, S220, S267, S171, S260, S289, and
`S290 markers in the xenograft DNA. The positions of the endpoints
`of the YACs are drawn arbitrarily.
`
`PCR analysis for DPCJ and DPC2 revealed that these STSs
`were present in all of 45 additional pancreatic adenocarcinoma
`xenografts and in 10 cell lines derived from pancreatic carci(cid:173)
`noma (ATCC; ref. 10). On the basis of the localization of the
`homozygous deletion, an analysis using the polymorphic dinu(cid:173)
`cleotide markers from chromosome 13q was performed (20).
`This revealed that 7 of 29 pancreatic adenocarcinoma xe(cid:173)
`nografts had LOH of chromosome 13q that spanned 13q12.3.
`One tumor, reported to have the cytogenetically identified
`translocation t(13;19)(q12;q13) (23), did not have LOH de(cid:173)
`tectable by using the available markers.
`
`DISCUSSION
`RDA has been described by Lisitsyn et al (13) as a means to
`isolate single-copy sequences that are present in only one of
`two otherwise nearly identical complex genomes. These inves(cid:173)
`tigators showed that RDA can identify binary polymorphisms
`and polymorphisms linked to a trait of interest (13, 24).
`Recently, it also has been shown that RDA can identify DNA
`losses and amplifications in tumors (25), as well as DNA
`sequences from unknown pathogens in infected tissues (26).
`Here we have applied RDA for the identification of DNA
`sequences that are deleted in tumors.
`
`Proc. Nat/. Acad. Sci. USA 92 (1995)
`
`5953
`
`Pancreatic carcinomas, as well as other carcinomas, can
`exhibit an average fractional allelic loss at least as high as 20%
`(7, 27). Overwhelmingly, the detected deletions are LOHs; that
`is, only one of the two alleles is deleted. Although complete
`data are not available, the occurrence of deletions involving
`both alleles is considered to be infrequent. Owing to the total
`loss of particular genetic information, the cellular effect of
`most homozygous deletions is assumed to be deleterious.
`Indeed, the homozygous deletions reported to date are rela(cid:173)
`tively small. The significance of the identification of a homozy(cid:173)
`gous deletion is best illustrated by their contribution to the
`discovery of several tumor-suppressor genes (RBJ, DCC, and
`p16) (2-5). The potential for identifying homozygous deletions
`among a high background of heterozygous losses suggests
`RDA as a powerful approach for the identification of novel
`tumor-suppressor genes.
`The homozygous deletion identified here by RDA maps to
`chromosome 13q12.3. Allelic loss at 13q is found in pancreatic
`carcinoma and in a wide variety of other tumor types. The
`tumor-suppressor gene RBJ, located at 13q14, is a candidate
`target gene within these areas of deletion. However, mutations
`or other evidence of inactivation of RBl have been found in
`only a subset of tumors (28-30). As for pancreatic adenocar(cid:173)
`cinoma, previous immunohistochemical analyses of Rb protein
`expression found no evidence of RBJ inactivation (7). The
`identification of a homozygous deletion at 13q12.3 in a pan(cid:173)
`creatic adenocarcinoma slrengthens the suspicion that, besides
`RBJ , at least one additional tumor-suppressor gene is located
`on chromosome 13q. Recently, a syndrome of familial breast
`cancer susceptibility (BRCA2) was linked to a 6-cM region at
`13q12-13, between the markers D13S267 and D13S289 (17).
`Although the BRCA2 candidate region encompasses the de(cid:173)
`letion we describe here, it as yet is not established whether the
`same genetic target is involved in pancreatic and breast
`carcinomas. If the target loci were postulated to be identical,
`the finding of a homozygous deletion would narrow the region
`for a gene search to the 1-cM region bounded by D13S171 and
`D13S260. It would also indicate that BRCA2 susceptibility is
`not due to a dominant oncogene (31) but could be attributed
`to a tumor-suppressor gene along the model proposed by
`Knudson, wherein both alleles must be inactivated to achieve
`the full tumorigenic phenotype (1).
`One of our carcinomas under study contains a translocation
`of 13q, with the breakpoint observed at or near the DPC locus
`(23). However, our analysis with dinucleotide markers did not
`reveal LOH at any flanking site of 13q in this particular
`carcinoma. LOH analysis might underestimate the fraction of
`cases with genomic alterations in a particular gene. It is also
`possible that additional cases of pancreatic carcinoma harbor(cid:173)
`ing a homozygous deletion would have gone undetected; since
`the markers flanking the DPC region are located 1 eM apart.
`We reported a possibly analogous situation with the p16
`tumor-suppressor gene, wherein we detected two pancreatic
`carcinomas as having a homozygous deletion upon the use of
`two flanking markers, and yet an additional eight homozygous
`deletions were identified only upon analysis of the p16 gene
`itself (10). Similarly, the majority of homozygous deletions
`involving RBJ are intragenic. Conversely, other tumor(cid:173)
`suppressor genes, like p53, rarely are inactivated by homozy(cid:173)
`gous deletion. Additional evidence for the involvement of a
`tumor-suppressor gene of general importance for pancreatic
`carcinoma includes our finding of LOH that spans 13q12.3 in
`nearly a quarter of the cases. This frequency of LOH at 13q is
`comparable with that found for breast carcinoma (32-34) and
`may be significant even though measurably less than frequen(cid:173)
`cies found at loci of some other tumor-suppressor genes. We
`postulate that a 1-cM region at 13q12.3, flanked by markers
`D13S171 and D13S260, contains a tumor-suppressor gene that
`is involved in pancreatic carcinoma.
`
`GeneDX 1024, pg. 4
`
`

`

`5954
`
`Genetics: Schutte et al
`
`Proc. Nat/. Acad. Sci. USA 92 (1995)
`
`The patient in the present study was a member of a familial
`clustering of adenocarcinomas of various organ sites (see Case
`Report). Two related points can be elaborated. First, the age of
`onset in this familial cluster is rather late. Indeed, an onset at
`older age is the pattern found for most familial pancreatic
`carcinoma pedigrees (8). Many, if not most, familial clusters of
`carcinoma in the general population do not reproducibly
`involve onset at young age. A comprehensive understanding of
`monogenic and polygenic influences on cancer susceptibility
`will have to include studies of these less distinctive phenotypic
`patterns of susceptibility (35). Second, it will be of interest to
`determine whether the individuals of the presently reported
`familial cluster are hemizygous in the region, which would
`suggest that the putative tumor-suppressor gene at 13q12.3
`might be involved in a variety of malignancies. This would be
`consistent with the frequent occurrence of allelic loss at 13q in
`multiple tumor types that is not readily attributable to inacti(cid:173)
`vation of the RBJ tumor-suppressor gene (7, 36-38).
`
`We thank the families in the Johns Hopkins National Familial
`Pancreatic Carcinoma Registry that cooperated in this study. We also
`thank Dr. Nikolai Lisitsyn for helpful suggestions and Dr. Elizabeth
`Jaffee and Karen Hauda for generation of pancreatic carcinoma cell
`lines. This work was supported by National Institutes of Health Grants
`CA56130 and CA62924, by the Dutch Cancer Society (K. W.F.) (M.S.),
`by Junta Nacional de lnvestiga~o Cientffica e Tecno16gica Scholar(cid:173)
`ship BD1508/91-ID (L.T.d.C.), and by Deutschen Krebshilfe (S.A.H.).
`S.E.K. is a McDonnell Foundation Scholar.
`
`1. Knudson, A. G. (1993) Proc. Natl. Acad. Sci. USA 90, 10914-
`10921.
`2. Dryja, T. P., Rapaport, J. M., Joyce, J. M. & Petersen, R. A.
`(1986) Proc. Natl. Acad. Sci. USA 83, 7391-7394.
`3. Fearon, E. R., Cho, K. R., Nigro, J. M., Kern, S. E., Simons,
`J. W., Ruppert, J. M., Hamilton, S. R., Preisinger, A. C., Thomas,
`G., Kinzler, K. W. & Vogelstein, B. (1990) Science 247, 49-56.
`4. Kamb, A., Gruis, N. A., Weaver-Feldhaus, J., Liu, Q., Harshman,
`K., Tavtigian, S. V., Stockert, E., Day, R. S., Johnson, B. E. &
`Skolnick, M. H. (1994) Science 264, 436-440.
`5. Diaz, M. 0., Ziemin, S., Le Beau, M. M., Pitha, P., Smith, S. D.,
`Chilcote, E. R. & Rowley, J. D. (1988) Proc. Natl. Acad. Sci. USA
`85, 5259-5263.
`6. Boring, C. C., Squires, T. S. & Tong, T. (1993) Ca Cancer J. Clin.
`43,7-26.
`7. Seymour, A. B., Hruban, R. H., Redston, M., Caldas, C., Powell,
`S.M., Kinzler, K. W., Yeo, C. J. & Kern, S. E. (1994) Cancer Res.
`54, 2761-2764.
`8. Lynch, H. T., Fitzsimmons, M. L., Smyrk, T. C., Lanspa, S. J.,
`Watson, P., McClellan, J. & Lynch, J. F. (1990) Am. J. Gastro(cid:173)
`enterol. 85, 54-60.
`9. McQueen, H. A., Wyllie, A. H., Piris, J., Foster, E. & Bird, C. C.
`(1991) Br. J. Cancer 63, 94-96.
`10. Caldas, C., Hahn, S. A., da Costa, L. T., Redston, M. S., Schutte,
`M., Seymour, A. B., Weinstein, C. L., Hruban, R. H., Yeo, C. J.
`& Kern, S. E. (1994) Nat. Genet. 8, 27-32.
`11. Almoguerra, C., Shibata, D., Forrester, K., Martin, J., Amheim,
`N. & Perucho, M. (1988) Cell 53, 549-554.
`12. Redston, M. S., Caldas, C., Seymour, A. B., Hruban, R. H., da
`Costa, L., Yeo, C. J. & Kern, S. E. (1994) Cancer Res. 54,
`3025-3033.
`13. Lisitsyn, N., Lisitsyn, N. & Wigler, M. (1993) Science 25!1,
`946-951.
`
`14. Kunkel, L. M., Monaco, A. P., Middlesworth, W., Ochs, H. D. &
`Latt, S. A. (1985) Proc. Natl. Acad. Sci. USA 82, 4778-4782.
`15. Wieland, 1., Bolger, G., Asouline, G. & Wigler, M. (1990) Proc.
`Natl. Acad. Sci. USA 81, 2720-2724.
`16. Wieland, I., Bohm, M. & Bogatz, S. (1992) Proc. Natl. Acad. Sci.
`USA 8!1, 9705-9709.
`17. Wooster, R., Neuhausen, S. L., Mangion, J., Quirk, Y., Ford, D.,
`et al. (1994) Science 265, 2088-2090.
`18. Drwinga, H. L., Toji, L. H., Kim, C. H., Greene, A. E. & Mulivor,
`R. A. (1993) Genomics 16, 311-314.
`19. Cohen, D., Chumakov, I. & Weissenbach, J. (1993) Nature
`(London) 366, 698-701.
`20. Gyapay, G., Morissette, J., Vigna!, A., Dib, C., Fizames, C.,
`Millasseau, P., Marc, S., Bernardi, G., Lathrop, M. & Weissen(cid:173)
`bach, J. (1994) Nat. Genet. 1, 246-339.
`21. Schott, D. R., Chang, J. N., Deng, G., Kurisu, W., Kuo, W. L.,
`Gray, J. & Smith, H. S. (1994) Cancer Res. 54, 1393-1396.
`22. Okumura, K., Nogami, M., Taguchi, H., Dean, F. B., Chen, M.,
`Pan, Z.-Q., Hurwitz, J., Shiratori, A., Murakami, Y., Ozawa, K.
`& Eki, T. (1995) Genomics 25, 274-278.
`23. Griffin, C. A., Hruban, R. H., Morsberger, L.A., Ellingham, T.,
`Long, P. P., Jaffee, E., Bohlander, S. & Yeo, C. J. (1995) Cancer
`Res., in press.
`24. Lisitsyn, N. A., Segre, J. A., Kusumi, K., Lisitsyn, N. M., Nadeau,
`J. H., Frankel, W. N., Wigler, W. H. & Lander, E. S. (1994) Nat.
`Genet. 6, 57-63.
`25. Lisitsyn, N. A., Lisitsina, N. M., Dalbagni, G., Barker, P.,
`Sanchez, C. A., Gnarra, J., Linehan, W. M., Reid, B. J. & Wigler,
`M. H. (1995) Proc. Natl. Acad. Sci. USA !12, 151-155.
`26. Chang, Y., Cesarman, E., Pessin, M.S., Lee, F., Culpepper, J.,
`Knowles, D. M. & Moore, P. S. (1994) Science 266, 1865-1869.
`27. Vogelstein, B., Fearon, E. R., Kern, S. E., Hamilton, S. R.,
`Preisinger, A. C., Nakamura, Y. & White, R. (1989) Science 244,
`207-211.
`28. Horowitz, J. M., Park, S.-H., Bogenmann, E., Cheng, J.-C.,
`Yandell, D. W., Kaye, F. J., Minna, J.D., Dryja, T. P. & Wein(cid:173)
`berg, R. A. (1990) Proc. Natl. Acad. Sci. USA 81, 2775-2779.
`29. Toguchida, J., Ishizaki, K., Sasaki, M. S., Nakamura, Y., Ikenaga,
`M., Kato, M., Sugimot, M., Kotoura, Y. & Yamamuro, T. (1989)
`Nature (London) 338, 156-158.
`30. Harbour, J. W., Lai, S.-L., Whang-Peng, J., Gazdar, A. D.,
`Minna, J.D. & Kaye, F. J. (1988) Science 241, 353-357.
`31. Mulligan, L. M., Eng, C., Healey, C. S., Clayton, D., Kwok, J. B.,
`Gardner, E., Ponder, M.A., Frilling, A.,