Am. J. Hum. Genet. 47:892-895, 1990
1989 Allen Award Address: The American Society of Human Genetics Annual Meeting, Baltimore Ray White Howard Hughes Medical Institute, University of Utah, Salt Lake City
To participate in the conversion of a new idea into a working reality, from enthusiastic discussion to the commitment of years of effort at the laboratory bench, is one of the real pleasures of scientific life. My recruitment to the concept and the subsequent effort to map the human genome with linked DNA markers followed a classic route, the telephone call from a friend and colleague. Maury Fox, my former mentor, called me upon his return from Washington, where he had attended a meeting of a task force on breast cancer; Mark Skolnick had also been present. Skolnick had described to Fox a meeting held a few days previously at Snowbird, UT, which had resulted in the proposition that human linkage markers could be derived from normal and presumably ubiquitous DNA sequence polymorphisms. Coming from the genetic tradition of molecular biology, I found the notion of developing a general system for mapping human genes by linkage to be natural enough. However, my initial response was one of intellectual irritation at the grandiose and presumptuous scale of this idea which I had not conceived. If the approach could be realized, an entire field would be transformed and the consequences for our understanding of the genetics of the human would be substantial. Heady stuff. But as I chewed through the logic and estimated the likelihood of success, the idea assumed a more and more promising character until by the next morning I was fully captured by the romance and promise of the new venture. The proposal seemed highly likely to contribute enormously to our understanding of the genetics of the most important organism from our point of view, man. A phone call to David Botstein confirmed that there was a role for me in the project: I was already Received September 17, 1990. Address for correspondence and reprints: Ray White, Ph.D., Howard Hughes Medical Institute, Wintrobe Building, University of Utah, Salt Lake City, UT 84132. o 1990 by The American Society of Human Genetics. All rights reserved. 0002-9297/90/4706-0002$02.00
892
on his list. Additional reassurance came as I witnessed the reaction of colleagues: the frequency with which they expressed annoyance and/or dismissed the proposal established its importance. The history of marker development in my laboratory follows the record of contributions of the fellows and students who have accepted the challenge. The first of these was Arlene Wyman. It was her work that identified the first arbitrary DNA sequence polymorphism, on the basis of a screen of arbitrary loci (Wyman and White 1980), although as her work got underway we read with great interest the reports-first from Y. W. Kan (Kan and Dozy 1978) and shortly thereafter from Alec Jeffreys (1979)-of polymorphisms within the P-globin locus. By happy chance, Arlene's polymorphism had the relatively rare and highly hopeful characteristic of reflecting not just the two alleles expected of a restriction-site variant but several alleles; higher-resolution gels revealed that the several was actually many. On rereading that first paper, I note that our primary interpretation of this phenomenon was in terms of transposon insertion, which stemmed from our conviction that the key to finding abundant polymorphism in the human would be the identification of transposons. This idea was well founded in the genetics of essentially all other systems studied to that time, including the mouse genome, which contains abundant polymorphism based on retroviral insertion. The only thing wrong with the idea was that it was not true; transpositional activity is almost nonexistent in the human. In the next screen for arbitrary polymorphisms, however, Mirielle Schafer and David Barker were able to find a number of the expected restriction-site polymorphisms (Barker et al. 1984). These had the curious feature of being strongly biased toward two of the many known restriction enzymes, TaqI and MspI. The reason became apparent, however, when one considered the CG dimer sequence that is present in the recognition sites for both enzymes. We knew that the cytosine
1989 Allen Award Address
of CG dimers was often methylated in mammalian DNA. Barker quickly pointed out that in bacteria methylated cytosines are hot spots for mutation; this property could explain the apparent high frequency of polymorphism found at sites recognized by these enzymes. Although careful bookkeeping has not been done, this interpretation has been supported by observations that a significant proportion of new mutations in human DNA are the result of C-to-T transitions from CG dimers. With the assurance of an abundance of markers, the next problem was to find a means of ordering them into linkage maps: family resources were now urgently required. For efficiency in gathering genotypic data, we needed to test DNA from families with many children; moreover, grandparental genotypes were desirable to help define allelic phase. We began to look for families with these characteristics; Mark Leppert was able to anecdotally ascertain and to sample more than 30 families with eight or more children and living grandparents. Our Utah population had not disappointed us. We chose a small set of informative markers from the short arm of chromosome 11 for our pilot run and were able to construct a good map with strong support for the order of the markers (White et al. 1985). Shortly thereafter, Dennis Drayna constructed the first map spanning an entire chromosome, with his markers for the X chromosome (Drayna and White 1985). This experience acquainted us with the complexities inherent in the analysis of linkage data, especially when more than a few markers are involved. It was therefore with delight that I learned upon first meeting Jean-Marc Lalouel that he and Mark Lathrop were about to complete a comprehensive new set of computer programs for linkage analysis, a package called LINKAGE (Lathrop et al. 1985), which would have the capacity to analyze many markers jointly. This made possible, for the first time, a true multilocus approach to linkage analysis. In addition to his message of multilocus analysis, Lalouel brought with him a surprising request; we in Utah should join Jean Dausset in his effort to develop the CEPH (Centre d'Etude de Polymorphism Humain), which he had conceived as an international collaboration to construct a linkage map of the human genome. The substance of the collaboration would be in sharing a common set of family DNAs for linkage mapping, each group returning its primary genotypic data to a common data base. We agreed and sent to Paris the cell lines we had established for 30 large families, to form the core of the CEPH family set.
893
About this time it became apparent that the basic proposition had been tested and confirmed: there were sufficient polymorphisms in human DNA, they could be detected, maps could be made, and genes bearing disease alleles could be located by linkage to DNA markers. It was appropriate, therefore, to consider how best to generate the large number of markers that would be needed to cover the genome. We were developing a coalition of funding sources that reflected interest in a variety of genetic disorders, when the Howard Hughes Medical Institute unexpectedly asserted a desire to support a marker-development effort of increased scope; it was an offer we could not refuse. To some extent this project presaged the much larger vision of the Genome Project; it also required building a group of greater organizational complexity than is the custom in academic fields. Fortunately, however, several of the associates in the group - notably Mark Leppert, Peter O'Connell, and Yusuke Nakamura-were prepared to help in developing a laboratory that would be able to meet the challenge of developing DNA markers on a large scale. With the sense of the doable nature of the project also came a recognition of the need for more highly informative marker systems, like the pAW101 found by Arlene Wyman. Sequencing studies (Bell et al. 1982) had, since her original paper, revealed that the molecular basis for the multiallelic systems was variation in the number of copies of tandemly repeated oligonucleotide sequences. However, arbitrary searches had turned up only few such loci in human DNA. It was, therefore, of great interest to us when Jeffreys reported his discovery that repeating oligonucleotide sequences, although unique to each locus, shared a common core of homologous sequence which could serve as the basis for a more rational ascertainment of new VNTR loci. Yusuke Nakamura implemented this logic, added his own developments, and was able to ascertain a large collection of new, single-copy VNTR loci (Nakamura et al. 1987). These markers have proven as robust and informative as we had hoped and have formed the basis for a number of mapping studies. At present, more than 1,000 polymorphic marker systems have been developed by our laboratory and have been ordered into marker sets that span each of the human chromosomes (examples include O'Connell et al. 1987, 1989a, 1989b; Nakamura et al. 1988a, 1988b; Lathrop et al. 1989; Julier et al. 1990). Our interest has not been solely an altruistic drive to develop tools for others to use; our main purpose was to further our own research. Our first opportunity to use these tools came not in a linkage study but in
894 the application of the new markers to a problem of somatic cell genetics: the molecular genetic description of carcinogenesis. We had been aware of the role of constitutional deletions in mapping the retinoblastoma gene, but those observations could have been interpreted as reflecting either loss of a gene or activation of a gene adjacent to the deletions. A description by Louise Strong and Vince Riccardi of a family in which retinoblastoma occurred with an unusual pattern of segregation (cousins were affected, but generations were often skipped) tipped the balance. Cytogenetic examination provided the explanation: a balanced interstitial translocation was segregating in the family. Those who received the balancer chromosome and the deleted chromosome 13 were free from disease; those who received only the deleted chromosome 13 had retinoblastoma. This "coinplementation" by the balancer chromosome suggested that the deletion was in fact a recessive lesion; furthermore, if that were the case, large deletions or chromosome losses might be triggers for development of tumors. DNA markers in the vicinity of the retinoblastoma gene could provide definitive evidence to support this model. Web Cavenee set out to test this hypothesis for his postdoctoral work in our lab and developed a set of DNA markers for chromosome 13. The experiment was to compare the DNA of a retinoblastoma tumor with that of its host: if there were a large deletion or chromosome loss, one of the two alleles found in the host DNA would not be found in the tumor DNA. To the surprise and delight of both of us, his first comparison of a small panel of tumor DNAs revealed that several had lost marker alleles. Even more remarkably, one of the tumors, in appearance karyotypically normal, had lost alleles distal to the retinoblastoma locus. This is what one would expect following a mitotic recombination event, but although we had discussed this mechanism as a formal possibility, mitotic exchanges had not previously been seen in mammalian systems and the finding was a complete surprise. The result provided strong support for the cell-recessive model in retinoblastoma. It also suggested an experimental approach to the mapping of cell-recessive cancer genes that has now been explored in a wide variety of tumors and that has revealed the chromosomal locations of a number of important cancer genes. We have been engaged in a number of linkage studies with disease genes, but perhaps the most interesting in terms of genetics has come from the study of the familial polyposis gene. This autosomal dominant genetic disorder strongly prediposes to colon cancer
White through the appearance of large numbers of adenomatous polyps of the colon. One of the interesting questions raised when genes with rare alleles have striking, highly penetrant phenotypes that result in common diseases -early coronary disease in consequence of a defective LDL-receptor gene is an example - is whether attenuated alleles, less penetrant but more common, might exist in the population. With respect to colon cancer, might there exist attenuated alleles of the polyposis gene that cause not hundreds or thousands of polyps in the colon but only a few? If so, might such alleles be common in the population and account for some significant proportion of colon cancers? Exploration of the first part of this hypothesis led Mark Leppert in our group to examine a Utah pedigree having a heterogeneous polyp syndrome (several members have more than 100 polyps, but many others have only a few upon examination). Linkage studies have now confirmed that the syndrome in this pedigree indeed maps to the polyposis locus. The first part of the hypothesis is, therefore, established: attenuated alleles do exist in the gene responsible for polyposis. Determination of their frequency in the population remains to be established. The primary goal of gene mapping is to identify the gene causing a disorder, i.e., to clone it, sequence it, and learn how its molecular lesions result in disease. Dramatic successes in these steps with the genes responsible for retinoblastoma, Duchenne muscular dystrophy, and chronic granulomatous disease have now been followed by the identification of the cystic fibrosis gene. Peter O'Connell in our group has been pursuing the same paradigm to identify the neurofibromatosis gene. Linkage mapping (Barker et al. 1987; O'Connell et al. 1989c) had indicated a map location on the proximal part of the long arm of chromosome 17. Recently, comparative gene mapping opened a surprising possibility: a mouse oncogene identified by retroviral insertion was mapped to mouse chromosome 11, known to possess conserved syntenic relationships with a number of loci on human chromosome 17. Mapping studies have now shown that the human homologue of this gene lies between the sites of different mutations (translocation breakpoints), about 60 kb apart, that originally appeared on chromosome 17 in the karyotypes of two individuals with neurofibromatosis. Sequencing studies must subsequently determine whether this gene, or other candidates from the region, are the sites of mutations leading to neurofibromatosis. That several such candidate genes have now emerged, however, is a dramatic reflection of the rate of progress and the poten-
1989 Allen Award Address
tial power of gene mapping as a crucial step toward identifying unknown genes. Linkage analysis on the basis of DNA markers has proven its worth. Many disease genes have been mapped in laboratories around the world, and the first few that have been cloned and sequenced are providing insights into human pathophysiology. I am very pleased to have had the opportunity to be a part of the work that is making this progress possible.
References Barker D, Schafer M, White R (1984) Restriction sites containing CpG show a higher frequency of polymorphism in human DNA. Cell 36:131-138 Barker D, Wright E, Nguyen K, Cannon L, Fain P, Goldgar D, Bishop DT, et al (1987) Gene for von Recklinghausen neurofibromatosis is in the pericentromeric region of chromosome 17. Science 236:1100-1102 Bell GI, Selby JJ, Rutter WJ (1982) The highly polymorphic region near the insulin gene is composed of singletandemly repeated sequences. Nature 295:31-35 Drayna D, White R (1985) The genetic linkage map of the human X chromosome. Science 230:753-758 Jeffreys AJ (1979) DNA sequence variants in the Ggamma, Agamma, delta, and 0-globin genes of man. Cell 18:1-10 Julier C, Nakamura Y, Lathrop M, O'Connell P, Leppert M, Mohandas T, Lalouel J-M, et al (1990) A primary map of 24 loci on human chromosome 16. Genomics 6:419-427 Kan YW, Dozy AM (1978) Polymorphism of DNA sequence adjacent to the human 3-globin structural gene: relationship to sickle mutation. Proc Natl Acad Sci USA 75: 5631-5635 Lathrop GM, Lalouel JM, Julier C, Ott J (1985) Multilocus linkage analysis in humans: detection of linkage and estimation of recombination. Am J Hum Genet 37:482-498 Lathrop GM, O'Connell P, Leppert M, Nakamura Y, Farrall
895 M, Tsui L-C, Lalouel J-M, et al (1989) Twenty-five loci form a continuous linkage map of markers for human chromosome 7. Genomics 5:866-873 Nakamura Y, Lathrop M, Bragg T, Leppert M, O'Connell P, Jones C, Lalouel J-M, et al (1988a) An extended genetic linkage map of markers for human chromosome 10. Genomics 3:389-392 Nakamura Y, Lathrop M, O'Connell P, Leppert M, Barker D, Wright E, Skolnick M, et al (1988b) A mapped set of DNA markers for human chromosome 17. Genomics 2: 302-309 Nakamura Y, Leppert M, O'Connell P, Wolff R, Holm T, Culver M, Martin C, et al (1987) Variable number of tandem repeat (VNTR) markers for human gene mapping. Science 235:1616-1622 O'Connell P, Lathrop GM, Law M, Leppert M, Nakamura Y, Hoff M, Kumlin E, et al (1987) A primary genetic linkage map for human chromosome 12. Genomics 1:93-102 O'Connell P, Lathrop GM, Nakamura Y, Leppert M, Ardinger R. Murray J, Lalouel J-M, et al (1989a) Twenty-eight loci form a continuous linkage map of markers for human chromosome 1. Genomics 4:12-20 O'Connell P, Lathrop GM, Nakamura Y, Leppert M, Lalouel J-M, White R (1989b) Twenty loci form a continuous linkage map of markers for human chromosome 2. Genomics 5:738-745 O'Connell P, Leach RJ; Ledbetter DH, Cawthon RM, Culver M, EldridgeJR, Frej A-K, et al (1989c) Fine structure DNA mapping studies of the chromosomal region harboring the genetic defect in neurofibromatosis type 1. Am J Hum Genet 44:51-57 White R, Leppert M, Bishop DT, Barker D, BerkowitzJ, Brown C, Callahan P, et al (1985) Construction of linkage maps with DNA markers for human chromosomes. Nature 313:101-105 Wyman A, White R (1980) A highly polymorphic locus in human DNA. Proc Natl Acad Sci USA 77:6754-6758
1989 Allen Award Address: The American Society of Human Genetics Annual Meeting, Baltimore Ray White Howard Hughes Medical Institute, University of Utah, Salt Lake City
To participate in the conversion of a new idea into a working reality, from enthusiastic discussion to the commitment of years of effort at the laboratory bench, is one of the real pleasures of scientific life. My recruitment to the concept and the subsequent effort to map the human genome with linked DNA markers followed a classic route, the telephone call from a friend and colleague. Maury Fox, my former mentor, called me upon his return from Washington, where he had attended a meeting of a task force on breast cancer; Mark Skolnick had also been present. Skolnick had described to Fox a meeting held a few days previously at Snowbird, UT, which had resulted in the proposition that human linkage markers could be derived from normal and presumably ubiquitous DNA sequence polymorphisms. Coming from the genetic tradition of molecular biology, I found the notion of developing a general system for mapping human genes by linkage to be natural enough. However, my initial response was one of intellectual irritation at the grandiose and presumptuous scale of this idea which I had not conceived. If the approach could be realized, an entire field would be transformed and the consequences for our understanding of the genetics of the human would be substantial. Heady stuff. But as I chewed through the logic and estimated the likelihood of success, the idea assumed a more and more promising character until by the next morning I was fully captured by the romance and promise of the new venture. The proposal seemed highly likely to contribute enormously to our understanding of the genetics of the most important organism from our point of view, man. A phone call to David Botstein confirmed that there was a role for me in the project: I was already Received September 17, 1990. Address for correspondence and reprints: Ray White, Ph.D., Howard Hughes Medical Institute, Wintrobe Building, University of Utah, Salt Lake City, UT 84132. o 1990 by The American Society of Human Genetics. All rights reserved. 0002-9297/90/4706-0002$02.00
892
on his list. Additional reassurance came as I witnessed the reaction of colleagues: the frequency with which they expressed annoyance and/or dismissed the proposal established its importance. The history of marker development in my laboratory follows the record of contributions of the fellows and students who have accepted the challenge. The first of these was Arlene Wyman. It was her work that identified the first arbitrary DNA sequence polymorphism, on the basis of a screen of arbitrary loci (Wyman and White 1980), although as her work got underway we read with great interest the reports-first from Y. W. Kan (Kan and Dozy 1978) and shortly thereafter from Alec Jeffreys (1979)-of polymorphisms within the P-globin locus. By happy chance, Arlene's polymorphism had the relatively rare and highly hopeful characteristic of reflecting not just the two alleles expected of a restriction-site variant but several alleles; higher-resolution gels revealed that the several was actually many. On rereading that first paper, I note that our primary interpretation of this phenomenon was in terms of transposon insertion, which stemmed from our conviction that the key to finding abundant polymorphism in the human would be the identification of transposons. This idea was well founded in the genetics of essentially all other systems studied to that time, including the mouse genome, which contains abundant polymorphism based on retroviral insertion. The only thing wrong with the idea was that it was not true; transpositional activity is almost nonexistent in the human. In the next screen for arbitrary polymorphisms, however, Mirielle Schafer and David Barker were able to find a number of the expected restriction-site polymorphisms (Barker et al. 1984). These had the curious feature of being strongly biased toward two of the many known restriction enzymes, TaqI and MspI. The reason became apparent, however, when one considered the CG dimer sequence that is present in the recognition sites for both enzymes. We knew that the cytosine
1989 Allen Award Address
of CG dimers was often methylated in mammalian DNA. Barker quickly pointed out that in bacteria methylated cytosines are hot spots for mutation; this property could explain the apparent high frequency of polymorphism found at sites recognized by these enzymes. Although careful bookkeeping has not been done, this interpretation has been supported by observations that a significant proportion of new mutations in human DNA are the result of C-to-T transitions from CG dimers. With the assurance of an abundance of markers, the next problem was to find a means of ordering them into linkage maps: family resources were now urgently required. For efficiency in gathering genotypic data, we needed to test DNA from families with many children; moreover, grandparental genotypes were desirable to help define allelic phase. We began to look for families with these characteristics; Mark Leppert was able to anecdotally ascertain and to sample more than 30 families with eight or more children and living grandparents. Our Utah population had not disappointed us. We chose a small set of informative markers from the short arm of chromosome 11 for our pilot run and were able to construct a good map with strong support for the order of the markers (White et al. 1985). Shortly thereafter, Dennis Drayna constructed the first map spanning an entire chromosome, with his markers for the X chromosome (Drayna and White 1985). This experience acquainted us with the complexities inherent in the analysis of linkage data, especially when more than a few markers are involved. It was therefore with delight that I learned upon first meeting Jean-Marc Lalouel that he and Mark Lathrop were about to complete a comprehensive new set of computer programs for linkage analysis, a package called LINKAGE (Lathrop et al. 1985), which would have the capacity to analyze many markers jointly. This made possible, for the first time, a true multilocus approach to linkage analysis. In addition to his message of multilocus analysis, Lalouel brought with him a surprising request; we in Utah should join Jean Dausset in his effort to develop the CEPH (Centre d'Etude de Polymorphism Humain), which he had conceived as an international collaboration to construct a linkage map of the human genome. The substance of the collaboration would be in sharing a common set of family DNAs for linkage mapping, each group returning its primary genotypic data to a common data base. We agreed and sent to Paris the cell lines we had established for 30 large families, to form the core of the CEPH family set.
893
About this time it became apparent that the basic proposition had been tested and confirmed: there were sufficient polymorphisms in human DNA, they could be detected, maps could be made, and genes bearing disease alleles could be located by linkage to DNA markers. It was appropriate, therefore, to consider how best to generate the large number of markers that would be needed to cover the genome. We were developing a coalition of funding sources that reflected interest in a variety of genetic disorders, when the Howard Hughes Medical Institute unexpectedly asserted a desire to support a marker-development effort of increased scope; it was an offer we could not refuse. To some extent this project presaged the much larger vision of the Genome Project; it also required building a group of greater organizational complexity than is the custom in academic fields. Fortunately, however, several of the associates in the group - notably Mark Leppert, Peter O'Connell, and Yusuke Nakamura-were prepared to help in developing a laboratory that would be able to meet the challenge of developing DNA markers on a large scale. With the sense of the doable nature of the project also came a recognition of the need for more highly informative marker systems, like the pAW101 found by Arlene Wyman. Sequencing studies (Bell et al. 1982) had, since her original paper, revealed that the molecular basis for the multiallelic systems was variation in the number of copies of tandemly repeated oligonucleotide sequences. However, arbitrary searches had turned up only few such loci in human DNA. It was, therefore, of great interest to us when Jeffreys reported his discovery that repeating oligonucleotide sequences, although unique to each locus, shared a common core of homologous sequence which could serve as the basis for a more rational ascertainment of new VNTR loci. Yusuke Nakamura implemented this logic, added his own developments, and was able to ascertain a large collection of new, single-copy VNTR loci (Nakamura et al. 1987). These markers have proven as robust and informative as we had hoped and have formed the basis for a number of mapping studies. At present, more than 1,000 polymorphic marker systems have been developed by our laboratory and have been ordered into marker sets that span each of the human chromosomes (examples include O'Connell et al. 1987, 1989a, 1989b; Nakamura et al. 1988a, 1988b; Lathrop et al. 1989; Julier et al. 1990). Our interest has not been solely an altruistic drive to develop tools for others to use; our main purpose was to further our own research. Our first opportunity to use these tools came not in a linkage study but in
894 the application of the new markers to a problem of somatic cell genetics: the molecular genetic description of carcinogenesis. We had been aware of the role of constitutional deletions in mapping the retinoblastoma gene, but those observations could have been interpreted as reflecting either loss of a gene or activation of a gene adjacent to the deletions. A description by Louise Strong and Vince Riccardi of a family in which retinoblastoma occurred with an unusual pattern of segregation (cousins were affected, but generations were often skipped) tipped the balance. Cytogenetic examination provided the explanation: a balanced interstitial translocation was segregating in the family. Those who received the balancer chromosome and the deleted chromosome 13 were free from disease; those who received only the deleted chromosome 13 had retinoblastoma. This "coinplementation" by the balancer chromosome suggested that the deletion was in fact a recessive lesion; furthermore, if that were the case, large deletions or chromosome losses might be triggers for development of tumors. DNA markers in the vicinity of the retinoblastoma gene could provide definitive evidence to support this model. Web Cavenee set out to test this hypothesis for his postdoctoral work in our lab and developed a set of DNA markers for chromosome 13. The experiment was to compare the DNA of a retinoblastoma tumor with that of its host: if there were a large deletion or chromosome loss, one of the two alleles found in the host DNA would not be found in the tumor DNA. To the surprise and delight of both of us, his first comparison of a small panel of tumor DNAs revealed that several had lost marker alleles. Even more remarkably, one of the tumors, in appearance karyotypically normal, had lost alleles distal to the retinoblastoma locus. This is what one would expect following a mitotic recombination event, but although we had discussed this mechanism as a formal possibility, mitotic exchanges had not previously been seen in mammalian systems and the finding was a complete surprise. The result provided strong support for the cell-recessive model in retinoblastoma. It also suggested an experimental approach to the mapping of cell-recessive cancer genes that has now been explored in a wide variety of tumors and that has revealed the chromosomal locations of a number of important cancer genes. We have been engaged in a number of linkage studies with disease genes, but perhaps the most interesting in terms of genetics has come from the study of the familial polyposis gene. This autosomal dominant genetic disorder strongly prediposes to colon cancer
White through the appearance of large numbers of adenomatous polyps of the colon. One of the interesting questions raised when genes with rare alleles have striking, highly penetrant phenotypes that result in common diseases -early coronary disease in consequence of a defective LDL-receptor gene is an example - is whether attenuated alleles, less penetrant but more common, might exist in the population. With respect to colon cancer, might there exist attenuated alleles of the polyposis gene that cause not hundreds or thousands of polyps in the colon but only a few? If so, might such alleles be common in the population and account for some significant proportion of colon cancers? Exploration of the first part of this hypothesis led Mark Leppert in our group to examine a Utah pedigree having a heterogeneous polyp syndrome (several members have more than 100 polyps, but many others have only a few upon examination). Linkage studies have now confirmed that the syndrome in this pedigree indeed maps to the polyposis locus. The first part of the hypothesis is, therefore, established: attenuated alleles do exist in the gene responsible for polyposis. Determination of their frequency in the population remains to be established. The primary goal of gene mapping is to identify the gene causing a disorder, i.e., to clone it, sequence it, and learn how its molecular lesions result in disease. Dramatic successes in these steps with the genes responsible for retinoblastoma, Duchenne muscular dystrophy, and chronic granulomatous disease have now been followed by the identification of the cystic fibrosis gene. Peter O'Connell in our group has been pursuing the same paradigm to identify the neurofibromatosis gene. Linkage mapping (Barker et al. 1987; O'Connell et al. 1989c) had indicated a map location on the proximal part of the long arm of chromosome 17. Recently, comparative gene mapping opened a surprising possibility: a mouse oncogene identified by retroviral insertion was mapped to mouse chromosome 11, known to possess conserved syntenic relationships with a number of loci on human chromosome 17. Mapping studies have now shown that the human homologue of this gene lies between the sites of different mutations (translocation breakpoints), about 60 kb apart, that originally appeared on chromosome 17 in the karyotypes of two individuals with neurofibromatosis. Sequencing studies must subsequently determine whether this gene, or other candidates from the region, are the sites of mutations leading to neurofibromatosis. That several such candidate genes have now emerged, however, is a dramatic reflection of the rate of progress and the poten-
1989 Allen Award Address
tial power of gene mapping as a crucial step toward identifying unknown genes. Linkage analysis on the basis of DNA markers has proven its worth. Many disease genes have been mapped in laboratories around the world, and the first few that have been cloned and sequenced are providing insights into human pathophysiology. I am very pleased to have had the opportunity to be a part of the work that is making this progress possible.
References Barker D, Schafer M, White R (1984) Restriction sites containing CpG show a higher frequency of polymorphism in human DNA. Cell 36:131-138 Barker D, Wright E, Nguyen K, Cannon L, Fain P, Goldgar D, Bishop DT, et al (1987) Gene for von Recklinghausen neurofibromatosis is in the pericentromeric region of chromosome 17. Science 236:1100-1102 Bell GI, Selby JJ, Rutter WJ (1982) The highly polymorphic region near the insulin gene is composed of singletandemly repeated sequences. Nature 295:31-35 Drayna D, White R (1985) The genetic linkage map of the human X chromosome. Science 230:753-758 Jeffreys AJ (1979) DNA sequence variants in the Ggamma, Agamma, delta, and 0-globin genes of man. Cell 18:1-10 Julier C, Nakamura Y, Lathrop M, O'Connell P, Leppert M, Mohandas T, Lalouel J-M, et al (1990) A primary map of 24 loci on human chromosome 16. Genomics 6:419-427 Kan YW, Dozy AM (1978) Polymorphism of DNA sequence adjacent to the human 3-globin structural gene: relationship to sickle mutation. Proc Natl Acad Sci USA 75: 5631-5635 Lathrop GM, Lalouel JM, Julier C, Ott J (1985) Multilocus linkage analysis in humans: detection of linkage and estimation of recombination. Am J Hum Genet 37:482-498 Lathrop GM, O'Connell P, Leppert M, Nakamura Y, Farrall
895 M, Tsui L-C, Lalouel J-M, et al (1989) Twenty-five loci form a continuous linkage map of markers for human chromosome 7. Genomics 5:866-873 Nakamura Y, Lathrop M, Bragg T, Leppert M, O'Connell P, Jones C, Lalouel J-M, et al (1988a) An extended genetic linkage map of markers for human chromosome 10. Genomics 3:389-392 Nakamura Y, Lathrop M, O'Connell P, Leppert M, Barker D, Wright E, Skolnick M, et al (1988b) A mapped set of DNA markers for human chromosome 17. Genomics 2: 302-309 Nakamura Y, Leppert M, O'Connell P, Wolff R, Holm T, Culver M, Martin C, et al (1987) Variable number of tandem repeat (VNTR) markers for human gene mapping. Science 235:1616-1622 O'Connell P, Lathrop GM, Law M, Leppert M, Nakamura Y, Hoff M, Kumlin E, et al (1987) A primary genetic linkage map for human chromosome 12. Genomics 1:93-102 O'Connell P, Lathrop GM, Nakamura Y, Leppert M, Ardinger R. Murray J, Lalouel J-M, et al (1989a) Twenty-eight loci form a continuous linkage map of markers for human chromosome 1. Genomics 4:12-20 O'Connell P, Lathrop GM, Nakamura Y, Leppert M, Lalouel J-M, White R (1989b) Twenty loci form a continuous linkage map of markers for human chromosome 2. Genomics 5:738-745 O'Connell P, Leach RJ; Ledbetter DH, Cawthon RM, Culver M, EldridgeJR, Frej A-K, et al (1989c) Fine structure DNA mapping studies of the chromosomal region harboring the genetic defect in neurofibromatosis type 1. Am J Hum Genet 44:51-57 White R, Leppert M, Bishop DT, Barker D, BerkowitzJ, Brown C, Callahan P, et al (1985) Construction of linkage maps with DNA markers for human chromosomes. Nature 313:101-105 Wyman A, White R (1980) A highly polymorphic locus in human DNA. Proc Natl Acad Sci USA 77:6754-6758
