Gene Mapping Techniques

Jean-Louis Guenet

Institut Pasteur, 75724 Paris Cedex 15, France

Very accurate gene mapping is essential in both man and laboratory mammals (1- 3). Several techniques have been used over the last 50 years to localize mammalian genes on the chromosomes of a given species. This chapter reviews these techniques, with special emphasis on the most recent ones that represent a true break- through in formal genetics.

CLASSICAL GENE MAPPING TECHNIQUES AND THEIR LIMITATIONS

When two genes are linked they have a tendency to cosegregate during successive generations. The closer the linkage, the more absolute is the cosegregation. This is the fundamental principle of gene mapping, which has been successfully applied to all species, including plants, over many years. In mammals such as humans and mice, the continued discovery of marker genes scattered throughout the genome has facilitated the mapping of new genes so that we now possess for these two species, particularly the mouse, linkage maps that are far more detailed than those existing for other mammals. In the mouse, special matings can be set up, with appropriate stocks, to test for possible autosomal linkage after two successive reproductive rounds: In general, cross-back crosses are used, cross-intercrosses being reserved for studies in which the viability or the fertility of the homozygous mutant under study is impaired. In humans investigations concerning linkage are based on pedigree analysis. In other words, for both species it is essential to define as a starting point a situation where two genes are heterozygous and either in repulsion A + / + B or in coupling AB/ + +, then to look for changes in this configuration after a reproductive cycle (forms in coupling giving rise to forms in repulsion and vice versa), and finally to count the percentage or frequency of these recombination events. The smaller the frequency, the tighter is the linkage. At the population level other approaches have been used to test for possible linkage. One such method used in human populations is the lod score, which assesses abnormally high frequencies of association between genes within populations hav- 87 88 GENE MAPPING TECHNIQUES ing common ancestors. In the mouse the use of congenic strains, and especially recombinant inbred strains (RIS), has proved very useful. The use of RIS for the detection of linkage has been extensively reviewed by Taylor (4), who has considered each strain as the equivalent of an F2 individual with a unique reassociation of parental characters. Here, however, the new genotype is permanent since the strain is inbred. Although such classic techniques have been very useful over the years they share several drawbacks and have in addition two intrinsic major limitations. The drawbacks include their time-consuming nature, particularly in human studies; their expense, since several hundred mice may have to be bred to test for a hy- pothetical linkage or to ensure the necessary precision (since by definition as many recombinants as possible must be scored if tight linkage is to be detected or estab- lished with accuracy); and the fact that many mutant genes cause reductions in viability or are not fully expressed in all individuals bearing them. This is particularly likely to happen in the mouse and may prevent the determination of the true segregation ratio, and thus the proportion of recombinants in linkage crosses. The two main limitations arise because: (a) The mapping methodology only allows for linkage detection with already known marker genes. By definition, linkage cannot be detected if the gene under investigation is "far away" from all other known genes, (b) The genes coding for invariant products, which by definition have no allelic forms, cannot be mapped using these methods. It is mainly for these reasons that nonsexual techniques have been developed by mammalian geneticists over the last twenty years. These methods use somatic cells of different species grown in vitro and artificially hybridized. Although fundamen- tally different, these new techniques are complementary to the classic ones and should be considered as additions rather than alternatives to the classic mapping technologies.

GENE MAPPING WITH SOMATIC CELL HYBRIDS

Our knowledge of the mammalian genetic map (and especially the human map) has increased considerably during the past 15 to 20 years as a result of a number of very important advances in cell genetics. Among these were the discovery of human-rodent and rodent-rodent cell hybridization and the study of the segregation of different chromosomes occurring in such hybrids; the development of chromosome staining techniques (the so-called banding techniques) which permit unambig- uous identification of all human, hamster, and mouse chromosomes at the metaphase stage; and the discovery of several hundreds of biochemical markers in the different species which can be identified and characterized as electrophoretic variants (the so-called electromorphs or allozymes, when the particular protein is an enzyme). At present, every human chromosome and every mouse chromosome has at least one biochemical marker and often many more. These technical advances have per- GENE MAPPING TECHNIQUES 89 mitted tremendous progress in mapping of both human (mostly) and mouse genes. For rapid chromosomal assignment of human genes, a panel consisting of a col- lection of cloned human-rodent cell-hybrids containing specific combinations of human chromosomes can be used. The genes coding for proteins that are specific to the human species can be identified by their product, using special biochemical assays. In parallel, it is also possible to identify individual human chromosomes among the chromosomal set of the cell hybrids. Thus genes coding for unique human markers can be mapped to the individual human chromosome that matches its particular pattern in the hybrid panel. In some instances only one chromosome of a given species (humans in general) is present in addition to the normal complement of the partner species of the hybrid, so that a nonambiguous localization of the gene can be made almost immediately. In order to increase such gene assignments, the following efforts have been made: (a) To characterize more gene products physically, chemically, and immunologi- cally in order to distinguish clearly human from rodent products. Monoclonal anti- bodies, electrophoresis, and isoelectrofocalization techniques and study of thermal stability of proteins have been of cardinal importance in this field, (b) To devise methods for inducing gene activities which are not normally expressed in cell hybrids, (c) To construct additional hybrid panels containing new combinations of human, mouse, or Chinese hamster chromosomes, either as intact chromosomes or partially rearranged with each other, (d) To develop sets of deletion hybrids each containing only one chromosome of a given species, with more or less extensive ter- minal deletion. More than 400 genes have been localized using these somatic cell hybrid techniques, and about 40 new genes are mapped each year. It is, however, impossible with these techniques to localize genes coding for monomorphic gene products (those which exhibit no variation between the partners of the cell hybrids); they are also characterized by a certain lack of precision in gene localization due to the lim- ited number of chromosomal breakages available. This results in most assignments being less precise than those obtained using the classic techniques. Finally it must be pointed out that genes whose products are not expressed in vitro cannot be mapped using these techniques.

THE USE OF RECOMBINANT DNA PROBES IN GENE MAPPING

With the recent development of molecular DNA technology, new techniques have become available for mapping genes. Again, these new techniques are complementary to those already discussed. They involve the use of DNA probes. A DNA probe consists of a small DNA segment, containing a specific sequence of the genome, which is used for hybridization at the molecular level with genomic DNA. To be easily kept and prepared in large quantities for hybridization, the DNA sequence of the probe is incorporated within an autoreplicative structure, either a bacteriophage or a plasmid, prior to hybridization. To be easily recognizable (the 90 GENE MAPPING TECHNIQUES primary role of a probe) the DNA sequence is labeled, usually with a radioactive isotope. Probes containing more or less extensive parts of specific genes are able to hybridize with the homologous segment of the chromosome carrying the particular gene if used under appropriate technical conditions. Thus if a radioactive probe is hybridized with a sample of DNA prepared from interspecific cell hybrids segregat- ing for a given set of chromosomes it is possible to identify, at least roughly, where the locus coding for this gene is located by matching the pattern of positive hybrid- izations with the pattern of chromosome presence or absence in the panel (Fig. 1). These techniques represent an important advance in formal genetics because they allow gene detection without requiring their expression. Mapping of the sequences of several important structural genes has been performed using heavily labeled cDNA probes and restricted (cut in small segments) genomic DNA from several sources: human-mouse hybrid cell lines have shown the localization of the a globin gene on human chromosome 16 and the assignment of the 3 globin gene complex to human chromosome 11; human-chinese hamster hybrid cell lines have allowed the assignment of the insulin gene to human chromosome 11 (5): mouse-chinese hamster hybrid cell lines have allowed the assignment of the cluster of genes coding for 3 interferon to mouse chromosome 4 (6). In situ molecular hybridization constitutes another promising development in gene mapping. When DNA probes are labeled with 3H with a high specific activity, rather than with 32P, and hybridized to well spread mitotic metaphase preparations of somatic chromosomes, attachment of the probe to homologous regions on the chromosomes occurs (5). Again it is possible to determine where a given sequence is located on the chromosomes of a given species after autoradiography and compu- tation of the grain count. These hybridization techniques have proven very useful in formal mammalian genetics. However, they have a few disadvantages: a gene cannot be mapped unless a specific radioactive probe has been made available; in the genome of a species there are frequently sequences partially matching the probe (e.g., pseudogenes, genes coding for isoforms) which makes the recognition of a given sequence difficult; and

Probe

FIG. 1. Detection of syntenic relationships with cellular hybrids. The probe is localized on chromosome n°10. GENE MAPPING TECHNIQUES 91 since no use is made of sexual reproduction the process of meiotic recombination does not operate and thus no precise assignment is possible for a given gene.

DNA PROBES AS GENETIC MARKERS

With the increasing number of studies being carried out on the structure of genomic DNA it has become quite clear that polymorphism at the level of DNA is much more intense than it is at the level of transcribed and translated gene products. This can easily be explained by what geneticists call silent mutations, such as changes at the level of DNA resulting, for example, in the substitution of an ACA for an ACG codon, both coding for the amino acid phenylalanine. Moreover, most DNA polymorphism goes undetected simply because the DNA is not translated into proteins. According to a recent theoretical estimate no more than 1% to 3% of the genomic DNA is actually translated. If on the other hand one assumes (which is almost cer- tainly true) that the coding sequences are randomly spread over the entire genome it is possible to use polymorphism at the DNA level as a source of markers. This has been achieved by the use of so-called restriction endonucleases. These particular enzymes have the property of cutting the DNA in a nonrandom manner. They cut when a particular short sequence of 4 to 8 nucleotides is detected on the DNA strand; each restriction endonuclease recognizes a specific sequence of nucleotides. It is thus possible with a given enzyme to cut an entire genome into segments of various sizes (a few kilobase pairs in general); this dissection of the genomic DNA into small pieces can be made on different samples with two or more enzymes. If two individuals differ by just one base pair in a restriction site (a mutation) the size of the restriction fragment will be different. If these two individuals belong to two different species, such as mouse and man, the restriction fragment length will almost always be very different. This is the so-called restriction fragment length polymorphism (or RFLP). After DNA restriction, the DNA fragments can be separated according to their size using gel electrophoresis. It is possible to recognize a particular segment of the genome hybridizing to a radiolabeled probe either directly on the electrophoretic gel or more frequently on a transfer replica of it. The tandem probe-restriction fragment can be considered as a marker which is normally unique for the genome since the probability of finding two sequences giving a restriction fragment of the same size and having a similar affinity for a given probe is very small indeed. With the so-called restriction polymorphism it is possible to recognize an individual homozygous for a RFLP from another who is heterozygous. After electrophoresis of restricted DNA from a homozygous individual all fragments hybridizing with a radiolabeled probe will be of the same size, producing a single line. Two classes of labeled fragments will be recognized on DNA from a heterozygous animal, giving two lines. Another advantage of RFLP is that the sequence which is used for the recognition of a particular fragment does not necessarily code for a protein. It is not even obliga- tory for the sequence to be expressed; any "anonymous" segment of DNA, pro- 92 GENE MAPPING TECHNIQUES vided that it is present in the genome under the form of a single copy, will do. About 1% of the recombinant phages in the human genomic library contain only unique sequences. Chromosomal localization of such unique sequences is then possible using the techniques already reported (e.g., hybrid cells, in situ chromosome hybridization). It is also possible using conventional markers in conjunction with normal sexual reproduction. Together with Robert and co-workers (7) we have mapped the gene coding for the light chain of myosin very close to the mouse Idh-1 locus (chromosome 1) using sexual reproduction (and thus meiotic recombination) and two different mouse species: Mus musculus domesticus (the normal laboratory mouse) and Mus spretus (a line derived from a wild population of mice). These two lines, which have been separated for millions of years, have lost mutual sex appeal. They no longer interbreed and behave like two different species. In the laboratory, however, interbreeding is possible. This allows a tremendous amount of polymorphism to segregate since the genetic divergence has produced many translated or untranslated changes at the DNA level. What we did originally, using a known a actin probe, can now be generalized to any fragment, even an "anonymous" one. Once the DNA fragment contained in the probe has been mapped to a given chromosome the information is permanent and can, for instance, be used as a marker for other fragments. Data are cumulative and fresh information is now accumulating at an expo- nential rate. Avner and co-workers have used a similar approach for the study of a particular part of the genome. After preparing a DNA library from an almost pure preparation of mouse X chromosomes sorted by flow cytometry (8), X-chromosome-specific DNA probes have been localized using these probes at the molecular level in conjunction with studies on animal phenotype using classical marker genes for X chromosomes such as Tabby (Ta), jimpy (jp), a galactosidase (a gal), and hy- poxanthine phosphoribosyl transferase (HPRT). All these probes have been linearly arranged and can now be used to map further sequences. It is obvious that the system is autocatalytic. It can also be generalized to any chromosomes of a given species so that having a marker every milli-Morgan is now a reasonable perspective. These molecular approaches to the mapping of mammalian genes are universally applicable. For example, no understanding of the biochemical nature of a given inherited disease is required for mapping its genetic determinant. They open new and very promising perspectives in terms of genetic counseling, as most common genetic diseases will probably soon become detectable in carriers and pregnancies at risk identifiable. In the mouse, conventional mapping techniques, although still in use even in the more advanced laboratories, are going to be irreversibly transformed.

REFERENCES

1. Cooper DN, Schmidtke J. DNA restriction fragment length polymorphisms and heterozygosity in the human genome. Hum Genet 1984;66:1-16. 2. Green MC. Gene mapping. In: The mouse in biomedical research. New York: Academic Press, 1981:105-17. GENE MAPPING TECHNIQUES 93

3. Puck TT, Kao F-T. Somatic cell genetics and its application to medicine. Annu Rev Genet 1982; 16:225-71. 4. Taylor BA. Recombinant inbred strains: use in gene mapping. In: MORSE MC, III, ed. Origins of inbred mice. New York: Academic Press, 1978:423-38. 5. Malcolm S, Barton P, Murphy C, Ferguson-Smith MA. Chromosomal localization of a single copy gene by in situ hybridization—human beta globin genes on the short arm of chromosome 11. Ann Hum Genet 1981 ;45:135-41. 6. Naylor SL, Gray PW, Lalley PA. Mouse immune interferon (IFN-gamma) gene is on chromosome 10. Somatic Cell Genet 1984;10:531-4. 7. Robert B, Barton P, Minty Y, et al. Investigation of genetic linkage between myosin and actin genes using an interspecific mouse backcross. Nature 1985;314:181—3. 8. Baron B, Metezeau P, Kelly F, et al. Flow cytometry isolation and improved visualization of sorted mouse chromosomes. Exp Cell Res 1984;152:220-30. 9. Taylor X, Bailey X. Proc Nad Acad Sci (USA).

DISCUSSION

Dr. Devilliers-Thiery: Do you have any idea how polymorphic genes are? In other words what is the percentage of homology in terms of nucleotide sequences between two strains of mice? Dr. Guenet: Between two human beings one base per thousand per chain is different, so you probably have 1,000 bases that are different from your neighbors'. Dr. Devilliers-Thiery: To be able to map a variant gene, you have to have polymorphism. Have you ever encountered a situation in which you had no polymorphism? Dr. Guenet: No. There are so many restriction enzymes that the probability of not finding polymorphism within an interspecific cross is very close to zero. You must bear in mind that as well as the huge number of restriction enzymes you can use methylation, either indepen- dently or in addition, to change the restriction site. This gives you a tremendous advantage compared with formal genetics. Dr. Devilliers-Thiery: How close is the genetic map of the mouse to that of the human? In other words, when you have mapped the mouse what are the chances that the human map will be the same in terms of linkage for invariant proteins? If you know that in the mouse genome gene A is next to gene B which is next to gene C, what are the chances that the same order will also be found in the human genome? Dr. Guenet: Invariant proteins have a tremendous degree of homology. Thus a probe obtained from the mouse will have a very good chance of matching the homologous human gene. It is another matter for variant proteins, of course. As far as order is concerned, Taylor and Bailey (9) recently reported that there are sequential homologies between the mouse map and the human map. For example, chromosome 4 on the mouse map is almost a replicate of the branch q of human chromosome 1. So if you find your receptor to be in chromosome 4 in the mouse I would recommend that you look in branch q of human chromosome 1. There are exceptions of course, but there are a large number of homologies. Dr. Campagnoni: Can you map genes within a chromosome by taking somatic cell hybrids that contain, for example, one human chromosome in a background of mouse chromosomes? Dr. Guenet: Yes, this has been done. Attempts were made to map myosin with this tech- nique, but the results were incorrect because several rearrangements caused confusion. When multiple rearrangements occur the segments of the human chromosome may become so small that they are no longer easy to identify. Dr. Sotelo: You talk about hybridization in situ. If I understood you correctly, you have a 94 GENE MAPPING TECHNIQUES cDNA probe and you use a single chromosome from one cell and try to find where the cDNAs hybridize. Is it necessary that the gene is expressed? Dr. Guenet: No. The big advantage of in situ hybridization and of all nuclear techniques is that the gene does not have to be expressed. For example, the gene for phenylalanine hy- droxylase is not expressed in vitro but, nevertheless, you can detect where this gene is located. This is not the problem with these techniques. The problem is the large amount of background "noise". You have to count the grains and make the map by a process of addition.