Integrated Mapping Project: interval mapping strategy

--Ed Coe and Shiaoman Chao

About 125 NEP variants and isozymes (see MNL 64:47) that have been mapped over the years into the "classical map" are currently being placed on the RFLP Core Map as part of an ongoing NSF-supported mapping project, by interval mapping. The target number of NEP loci to be interval mapped on that project in its next phase is approximately 250 selected loci. The principle of interval mapping (Lander and Botstein, 1986) is that segregating markers nearest to a target gene remain linked with the target and tend to become homozygous when it becomes homozygous. The stock materials required for interval mapping of NEPs are segregating families in which multiple-point polymorphism is present; in fact the extent of polymorphism in maize is such that virtually any segregating family inherently has sufficient interval marking. Interval mapping is typically done individual-by-individual, by determining association of markers and target gene in 25-30 homozygous individuals, directly giving a measure of frequency of recombination.

Expansion to high-density interval mapping, potentially locating every available visible variant, is the goal of the following plan we have developed with the help of Tim Helentjaris and Dave Weber.

Materials for interval mapping include any and all segregating (e.g., F2) progenies that may be made available by maize investigators. These may be the segregating family in which a mutant is found or from which it is being recovered following outcrossing. Progenies segregating for over 2800 characterized mutants, 800 or so of which have been located to some degree in the genome, are ready in Gerry Neuffer's materials to be used in this part of the mapping. In addition there are in this resource many, unnumbered mutants of "common" classes such as albinos and virescents, and new cases of all kinds are currently being screened from M2s of mutagenized families. A collection of unanalyzed mutants is available in the Maize Genetics Stock Center, and in the stocks of scores to hundreds of investigators. The problem is one of prioritization rather than supply. The focus of our selection will be, first, on partially located mutants of all types; second, on certain more-characterized types (e.g., biochemically or physiologically characterized traits); third, on seed and seedling expressions; and fourth, on unique growth types and others.

The strategy we now intend to use, in order to achieve large numbers of interval-mapping evaluations, is a pooled-sample approach. In place of plant-by-plant interval mapping, an approach is needed that can be more bold in the determination of location and map order than distance. The principle of our approach is that, for alleles of RFLP loci more and more closely linked with the target locus (M/m), two changes occur in their average distribution: In dominant individuals (MM and Mm) they change from a 1:1 ratio (independence) to 2:1 (complete linkage), and in recessive individuals from 1:1 to 0:1. In pooled samples, reduction or disappearance of one of two bands in the recessive class, accompanied by its enhancement in the dominant class, evidences linkage and has its own internal control.

Resolution of autoradiograms in our experience generally permits easy discrimination of 2:1 from 1:1, and dilution series indicate that decrements of a band down to 9:1 or less can be distinguished clearly from more-equal proportions or from 10:0. Among 10 pooled recessives containing 20 strands, 5% recombination will average one recombinant strand per pool (in a series of pools: 0, 1, 2... ), so this resolution becomes dependent upon the "disappearance point" for the particular band.

Our specific plan at this time, subject to refinements, is to pool 2 samples of 10 dominant individuals each and 3 samples of 10 recessive individuals each, prepare DNA from each of the 5 pooled samples, and digest with 4 restriction enzymes. The 20 samples, plus markers, can then be loaded to produce as many as 8 blots in one casting and one run, sufficient for up to 80 probings. For loci that already have been placed to chromosome arm (e.g., within 50 cM), 50-80 probings potentially could define the position of the target gene with a precision finer than 1cM (limited only by the size of population and the "disappearance point"). For loci that have not yet been placed, 60 probings followed by 2-3 increasingly focussed probings potentially can define the position within 10cM or less (genome size estimated to be 2300cM per D. Grant and W. Beavis, personal communication). We consider these estimates of precision to be conservative. Pooled-sample interval mapping, if consistently successful, could establish a new and generally applicable technique for rapid and efficient mapping of mutants.

Among the considerations to be weighed for the pooled-sample approach are (1) limitations for particular classes of mutants and (2) equivalence of tissue representation in the pool. Mutants expressed in kernels can be mapped if they will germinate, simply by growing the two classes and pooling seedlings (heterofertilization, averaging 1%, will reduce precision only to that extent). Mutants that are inviable and will not germinate (e.g., defective or germless kernels, both of which are very common) can be mapped by growing the normal class and evaluating pools for 2:1 ratio vs. 1:1 ratio; densitometry readings will be tested as a method for refining the precision of this approach. The inviable class of mutants is particularly important, because these represent losses of essential, cell-limited functions required in the development and survival of the zygote and developing endosperm, in contrast to ubiquitous housekeeping functions, which will not generally survive the gametophyte screen (the latter, however, will be represented among cDNAs). Because band ratios have internal controls through comparison of dominants with recessives, modest precautions will be sufficient to ensure that tissue from each seedling represents approximate equivalence (i.e., not so little as one-half relative to the others in the pool). It should be noted again in this context that the intent of most of our present interval-mapping tests is not so much to determine distance of a target gene from the markers as to determine order of the loci, even if only approximately.

We would be pleased to have comments or suggestions from Cooperators on this approach to mapping of NEPs.

Return to the MNL 65 On-Line Index
Return to the Maize Newsletter Index
Return to the MaizeGDB Homepage