Construction and characterization of a soybean bacterial artificial chromosome library and use of multiple complementary libraries for genome physical mapping.

Two plant-transformation-competent large-insert binary clone bacterial artificial chromosome (hereafter BIBAC) libraries were previously constructed for soybean cv. Forrest, using BamHI or HindIII. However, they are not well suited for clone-based genomic sequencing due to their larger ratio of vector to insert size (27.6 kbp:125 kbp). Therefore, we developed a larger-insert bacterial artificial chromosome (BAC) library for the genotype in a smaller vector (pECBAC1), using EcoRI. The BAC library contains 38,400 clones; about 99.1% of the clones have inserts; the average insert size is 157 kbp; and the ratio of vector to insert size is much smaller (7.5 kbp:157 kbp). Colony hybridization with probes derived from several chloroplast and mitochondrial genes showed that 0.89% and 0.45% of the clones were derived from the chloroplast and mitochondrial genomes, respectively. Considering these data, the library represents 5.4 haploid genomes of soybean. The library was hybridized with six RFLP marker probes, 5S rDNA and 18S-5.8S-25S rDNA, respectively. Each RFLP marker hybridized to about six clones, and the 5S and 18S-5.8S-25S rDNA probes collectively hybridized to 402 BACs--about 1.05% of the clones in the library. The BAC library complements the existing soybean Forrest BIBAC libraries by using different restriction enzymes and vector systems. Together, the BAC and BIBAC libraries encompass 13.2 haploid genomes, providing the most comprehensive clone resource for a single soybean genotype for public genome research. We show that the BAC library has enhanced the development of the soybean whole-genome physical map and use of three complementary BAC libraries improves genome physical mapping by fingerprint analysis of most of the clones of the library. The rDNA-containing clones were also fingerprinted to evaluate the feasibility of constructing contig maps of the rDNA regions. It was found that physical maps for the rDNA regions could not be readily constructed by fingerprint analysis, using one or two restriction enzymes. Additional data to fingerprints and/or different fingerprinting methods are needed to build contig maps for such highly tandem repetitive regions and thus, the physical map of the entire soybean genome.