In Chapter 10, we learned that, if a cloned DNA fragment is used as a probe of genomic DNA that has been cut with a restriction enzyme, then the probe will bind to one or more genomic fragments. For example, if the restriction enzyme used does not cut within the chromosomal region encompassed by the cloned fragment, then the probe should bind to one fragment flanked by restriction sites on each side. Since the DNA of chromosomes within a species is generally homologous, it might be expected that a constant-sized genomic fragment will be bound in all individuals. However, when probes are used in this way, the bound fragments are often found to be of different sizes in different individuals. The explanation is that a given restriction site is not always found in all individuals. The absence of a site is usually caused by a single nucleotide difference that is most likely biologically neutral. Hence, for example, if a probe binds a 2-kb fragment in individual A of a haploid species and it binds a fragment of 2.3 kb in individual B, the reason is usually that one of the sites that flanked the 2-kb fragment is missing in B, and the next site is 0.3 kb away, making the hybridized fragment 2.3 kb in size.

The presence and absence of the restriction site can be treated as two alleles that can be thought of as + and − alleles. The presence of the + in some individuals in the population and the absence (−) in others generates a restriction fragment length polymorphism, or RFLP. (In the case just discussed, there was a dimorphism—two “morphs,” one short and one long.) Geneticists were surprised to discover that RFLPs are quite common in populations and that a large proportion of probes will detect one. RFLPs are identified by a rather hit-or-miss method of hybridizing panels of randomly cloned genomic fragments to genomic restriction digests of several different individuals in a family or a population. Because RFLPs are a relatively common type of variation in nature, this method succeeds in finding RFLPs in most cases.

The significance of RFLPs is threefold. First, if an individual is heterozygous for two morphs of an RFLP, this heterozygous “locus” can be used as a marker in chromosomal mapping. Although at first the locus of the RFLP is not necessarily known, as more and more RFLPs are found, they can be mapped in relation to gene loci and in relation to other RFLP loci, and their positions gradually saturate the genetic map. The RFLPs are not biologically significant in most cases, but they can be used to map interesting genes and act as positions from which these genes can be cloned by positional cloning.

Second, in an extension of mapping analysis, RFLP alleles (morphs) can be used as diagnostic tools. For example, in a family with a record of a certain disease, if it can be established that the people who have the disease also carry a specific allele of an RFLP, then this fact suggests not only that the RFLP locus is linked to the disease gene locus, but furthermore that the specific RFLP allele is in cis arrangement with the disease allele. Hence the RFLP allele becomes a diagnostic marker for the disease, and this information can be used in genetic counseling.

Third, RFLPs can be used to measure genetic divergence between different populations or related species. The restriction-site difference is effectively a DNA difference, so a measure of the total number of RFLP differences represents a measure of genetic difference. Hence RFLPs are important in studies of evolution.

RFLP mapping is often performed on a defined set of strains or individuals that become “standards” for mapping that species. For example, in the fungus Neurospora, two wild-type strains, Oak Ridge and Mauriceville, are known to show many RFLP differences, so these strains have become standards used in RFLP mapping. The RFLPs can be mapped relative to one another or to genes of known phenotypic expression. For example, let ad stand for an allele for adenine requirement, and 1 and 2 stand for RFLP loci with either the Oak Ridge (OR) or Mauriceville (M) “alleles.” A cross can be made of the type ad . 1^OR . 2^OR × ad⁺ . 1^M . 2^M Progeny are tested for all three loci. Adenine requirement is tested by inoculating strains on medium lacking adenine, and the RFLP alleles are tested by probing with the relevant probes. Recombinant frequencies are calculated in the usual way. Most mutants in Neurospora have been induced in Oak Ridge wild-type strains, so it is a simple matter to map the mutant alleles to RFLPs simply by crossing the mutant Oak Ridge strain to the wild-type Mauriceville strain. An example of mapping a phenotypic mutant by using RFLP markers is shown in Figure 11-2.

Similar standard strains have been established in other organisms. An analogous approach has been used in human genome mapping by collecting DNA from a defined set of individuals in 61 families with an average of eight children per family and making this DNA available throughout the world to provide a standard for RFLP mapping.

Figure 11-3 shows an example of linkage of a human disease allele to an RFLP locus and the potential for using this information in diagnostics. Because of the close linkage, future generations of persons showing the RFLP morph 1 can be predicted to have a high chance of inheriting the disease allele D. This sort of predictive power can be used in prenatal diagnoses of the genotypes of fetuses, with the use of amniocentesis or chorionic villus sampling (considered later in this chapter).

It is worth comparing the process of making a restriction map (restriction mapping, pages 327–329) with the process of RFLP mapping. Restriction maps are based on physical analysis of DNA, whereas RFLP maps are based on recombination analysis of matings. Note also that restriction mapping is based on restriction sites with no variation, whereas RFLP mapping is based on restriction-site variation between homologous chromosomes. Most restriction maps are short-range (fine-scale) maps, although long-range maps can be constructed with rare-cutting restriction enzymes. In contrast, RFLP mapping generally produces long-range (coarse-scale) maps. RFLP mapping of whole genomes will be covered in detail in Chapter 12.