A horse whole-genome-radiation hybrid panel: chromosome 1 and 10 preliminary maps

In recent years there has been increasing interest in mapping thehorse genome, particularly to identify disease and performance-enhancing genes. Although a number of horse mapping tools havebeen developed and have proved very useful (genetic maps, so-matic cell hybrid panels, and a BAC library), the benefits ofwhole-genome–radiation hybrid (WG-RH) mapping have not beenavailable. Its advantages over genetic linkage mapping are: (i) theresolving power is not lost in regions of the genome with a lowrecombination rate, because it does not rely on meiotic recombi-nation events; (ii) it is especially useful in animals such as thehorse with relatively long generation times and single births; and(iii) genetic and physical maps can be integrated, as both poly-morphic and non-polymorphic markers can be placed on the samemap. The WG-RH panel constructed in this study is the first suchpanel to be reported for the horse, and its preliminary character-ization demonstrates its usefulness for horse genome mapping.The panel was constructed by the fusion of horse embryonicendothelial primary lung cells (male) to the established hamsterfibroblast cell line A23 (Westerveld et al. 1971) by using themethod described in McCarthy et al. (1997). A series of fusions,involving irradiation (3000 rads) of donor cells prior to fusion withequal numbers of recipient cells, generated ∼160 hybrids in total.From these 160 hybrids, 94 were selected at random and screenedby using 20 widely distributed markers (representing 17 chromo-somes). Any hybrids that did not produce an amplification productwith these markers were screened by FISH to determine whetherthey contained horse DNA from other chromosomes or regions ofchromosomes (not represented by the 20 markers); hybrids foundto be negative by the FISH screen were replaced at random fromthe remaining unused hybrids, and the same PCR and FISH screen-ing procedure was repeated until 94 hybrids were assembled.These 94 hybrids were used for preliminary characterization as amapping panel. It is expected that the majority of the hybrids(>95%) characterized here will be represented in the TM99 panelof 94 hybrids that is being grown on a large scale by ResearchGenetics Inc. (Huntsville, Ala. 35801).The amount of horse DNA retained by the hybrids in the panelwas evaluated by examining the retention of the 20 widely dis-tributed horse markers. On average each marker was retained in27.8% of the hybrids (ranging from 10.6% to 71.3%; Table 1),implying that the panel as a whole retains the equivalent of ap-proximately 26 horse genomes. These retention frequencies com-pare well with those found for the human and mouse RH panels,which have been used successfully for creating whole-genomemaps (Gyapay et al. 1996; McCarthy et al. 1997).The mapping ability of the panel was assessed by producingRH maps for the two horse chromosomes with the most markersavailable, and comparing these with the latest genetic maps (Swin-burne et al. 2000a). In total, 39 markers on Chromosome (Chr) 1and 15 markers on Chr 10 were analyzed (Table 1). The averageretention of markers was 15.4% (ranging from 5.3% to 29.8%) onChr 1, and was higher, 25.4% (ranging from 16.0% to 44.7%) onChr 10. An increase in the retention frequency on smaller chro-mosomes was also noted in human and mouse RH panels (Gyapayet al. 1996; McCarthy et al. 1997). For each chromosome, linkagegroups with at least 4-LOD units support were identified; four suchgroups were found on Chr 1 and two were detected on Chr 10(Figs. 1 and 2). Within each of these linkage groups, frameworkmarkers were ordered with 3-LOD units support, and most of thenon-framework markers were ordered with 2-LOD units support.Some non-framework markers had lower statistical support fortheir order (shown in italics, Figs. 1 and 2). Similarly, the relativeorder of some linkage groups was suggested by linkage analysisbut had low statistical support [Chr 1 (groups B and C) and Chr 10(groups A and B)]. The relative order of the other RH linkagegroups was determined by comparison with the genetic map andFISH localizations.The genetic map can give a misleading impression of markerdensity because genetic distances may be small owing to regionshaving a low meiotic recombination rate. In such regions the mark-ers may actually be physically far apart and, therefore, at a lowerdensity than predicted by the genetic map. Since RH panels requirea high density of markers, this might account for the low statisticalsupport for the ordering of some markers and linkage groups. Alow density of available markers could also explain why two mark-ers (HLM5, ICA22) could not be placed on the RH map. HLM5 hasalready been shown to be 24 cM from its nearest neighboringmarker on the genetic map. Three other markers could not beplaced on the RH map (UM004, HTG12, and HMS15), but wereable to be placed on the genetic map in a region equivalent to RHlinkage group D; this discrepancy between the two maps should beresolved as both maps are characterized further.This study has demonstrated the ability of the horse WG-RHpanel to produce an accurate genome map as there is good agree-ment of the genetic and physical maps with the RH maps for Chrs1 and 10 (Figs. 1 and 2). Only a few differences between the mapswere observed, and experiences with human mapping suggest thatsuch differences are common and are resolved as more markers areincorporated into the maps (Walter et al. 1994). The WG-RH panelwas able to order some markers that co-segregate on the geneticlinkage map, i.e., Chr 1 markers LEX39 and ICA18, and ICA41 andICA32, and Chr 10 markers LEX8 and COR015, and NVHEQ18