From mutation mapping to phenotype cloning.

Current methods of detecting known mutations in genes are efficient and have in many instances been incorporated into the armamentarium of diagnostic medicine. Many genetic diseases, however, including autosomal dominant diseases with high mutation rates and genetically lethal X chromosome-linked diseases, result from heterogeneous and often private mutations in the respective genes. Methods of identifying new mutations in genes are still technically demanding and lack sensitivity. This has limited their utility in phenotype-genotype correlations and in clinical medicine. Short of sequencing, single-stranded conformational polymorphism and heteroduplex analysis are the most commonly used nucleic acid-based methods to map mutations in genes. They rely on detecting differences in electrophoretic mobility of gene segments carrying mutant versus normal sequences. Screening for mutations by selective cleavage of mismatched base pairs in heteroduplexes has also been attempted. Development of methods using chemical reactivity of mismatches with carbodiimide had limited success in part because the modification was not readily detected. Improved detection of mismatches using chemical cleavage was reported more recently (1) but sensitivity in detecting single base mismatches was still suboptimal (2). Early attempts to develop enzymatic methods for mutation detection using S1 nuclease to cleave singlestranded DNA in mismatched heteroduplexes (3) also lacked sensitivity in detecting single base-pair changes. The above methods recognize mismatches by reacting preferentially with single-stranded DNA and may be confounded by transient separation of strands in double-stranded DNA (breathing), particularly in regions rich in ANT base pairs (4). The development of reagents with direct reactivity toward mismatches and with minimal reactivity toward transient single strands would have theoretical advantage. This has been attempted with DNA repair enzymes which recognize modified bases in mismatched DNA (5,6). In this issue of the Proceedings Youil et al. (7) describe an advance in mapping of mutations, using T4 endonuclease VII for the enzymatic detection of mismatches in heteroduplexes comprising strands representing both normal and mutant alleles (7). The main function of endonuclease VII, produced by gene 49 of bacteriophage T4, appears to be resolution of branched DNA structures (8). Its other activities, including selective cleavage of double-stranded DNA containing mismatches, have also been carefully characterized (9). Youil et al. report that digestion with T4 endonuclease VII detected 17 out of 18 single base-pair mutations by cleavage of at least one strand in the heteroduplex; a set of A-A/T-T mismatched heteroduplexes was the only mutation not cleaved in either strand. Some nonspecific nuclease activity of endonuclease VII was noted, although it did not confound mapping of mutations. T4 endonuclease VII has great potential in mutation screening, but new and exciting genome applications of resolvases and other efficient mismatch detection enzymes loom on the horizon as well. Recent advances in human genetics have been nothing short of spectacular in identifying genes responsible for monogenic traits. In particular, the isolation of genes of unknown function by deriving information about their position in the genome ("positional cloning") has been a powerful new approach to identifying genes associated with monogenic conditions (10). Many human traits, however, including most common diseases, have a complex etiology presumably reflecting interactions between one or several genes and the environment. Identifying genes associated with complex phenotypes by established positional cloning methods becomes progressively more difficult as the number of contributing genes increases and as the effect of each individual gene to the phenotype decreases. In addition, the task can be greatly complicated if the effect of each gene is modified by the genetic background in which it operates. Current methods in molecular genetics are still inefficient ih identifying genes in these situations, especially when the pathophysiology is poorly understood and candidate genes cannot be readily identified. New approaches to directly identify genes without requiring knowledge of their function or position in the genome would be a crucial advancement toward understanding the genetics of complex traits. Several recent papers, including the one by Youil et al. (7), represent technological progress toward development of methods with just that capability. The concept of identical-by-descent (i.b.d.) sequences was developed by early geneticists and its use in mapping genesfor instance, by sib-pair analysis-was recognized (11). The conceptual leap from relying on knowing either the gene's function and/or its genome position for its isolation to the direct experimental isolation or mapping of an unknown gene based on its phenotype alone was first made convincingly by Sanda and Ford (12). They presented three arguments in support of a novel approach of identifying genes directly by isolating i.b.d. sequences. (i) The genomic sequences from two unrelated individuals are different (13). (ii) Given low mutation rates, i.b.d. segments in genomes of two related individuals are identical in sequence. (iii) Through segregation and recombination of chromosomes, the i.b.d. segments become progressively fewer and shorter with increasing numbers of meiosis separating two relatives. The i.b.d. sequences should therefore contain genes contributing to a phenotype which is shared by distant relatives. i.b.d. sequences could in principle be isolated in the laboratory by mixing, denaturing, and annealing genomic DNA from affected and distantly related individuals and enriching for heteroduplexes which contain no mismatches while discarding homoduplexes and heteroduplexes with internal mismatches (given that the genomic DNA is split in fragments long enough to contain sequence differences when inherited from different ancestors). Sanda and Ford attempted this approach to gene identification by modifying mismatched heteroduplexes with carbodiimide. Following cloning of heteroduplexes into phage, only unmodified phage-i.e., thq ones containing i.b.d. sequences without mismatches-would replicate in bacteria defective in repair enzymes. The principle of the approach was sound but the technology at that time was not sufficiently advanced to allow efficient isolation of i.b.d. sequences. The prospects for direct isolation of genes associated with a phenotype but of unknown function and genome position were revived by two recent papers describing major methodological advances in the field. Brown and associates (14) described the technique of genome mismatch scan-