Construction and characterization of a new bovine bacterial artificialchromosome library with 10 genome-equivalent coverage

Wesley Warren,1 Timothy P.L. Smith,2 Caird E. Rexroad III, 2 Scott C. Fahrenkrug,2 Tanya Allison,1

Chung-Li Shu,3 Joseph Catanese,4 Pieter J. de Jong4

1Monsanto Company, Ag sector, 700 Chesterfield Parkway, St. Louis, Missouri 63198, USA2USDA-ARS U.S. Meat Animal Research Center, Clay Center, Nebraska, USA3Children’s Hospital, Oakland Research Institute, 747 52nd Street, Oakland, California 94609, USA4Children’s Hospital Oakland Research Institute, 747 52nd Street, Oakland, California 94609, USA and Parke Davis Laboratory for MolecularGenetics, 1501 Harbor Bay Parkway, Alameda, California 94502, USA.

Received: 22 September 1999 / Accepted: 27 March 2000

Genetic linkage maps of the bovine genome have recently beenconstructed (Barendse et al. 1997; Kappes et al. 1997) and used toassign loci affecting heritable traits of economic importance(Georges et al. 1995; Stone et al. 1999) to specific chromosomalsegments. Markers can thus be identified that may be useful inmarker-assisted selection (MAS) to increase the frequency of fa-vorable allele(s) in commercial populations (reviewed in Soller1994). In addition, mapping of these loci creates the opportunity toidentify gene(s) influencing a trait, through positional cloning orpositional candidate gene approaches (Georges and Andersson1996). The positional cloning approach requires the constructionof contigs that physically span large sections of chromosomes. Inthe human and mouse systems, contig construction has dependedon the availability of multiple YAC and BAC libraries that providedepth of coverage to minimize the impact of chimeric or deletedclones. BACs have proven the vector of choice for maintaininggenomic DNA inserts owing to relatively low rates of chimerism,ease of manipulation, and adaptability to automated DNA purifi-cation and sequencing procedures (Kelly et al. 1999). Moreover,physical BAC maps have been proposed as the basis for sequenc-ing the genomes of both prokaryotic and eukaryotic organisms(Venter et al. 1996).

Cai et al. (1995) reported the construction of the first bovineBAC library utilizing the pBeloBAC11 vector, with approximately6× genome coverage and an average insert size of 146 kb. Re-cently, Zhu et al. (1999) reported the construction of a bovine BAClibrary with 5× genome coverage and average insert size of 105 kb.Here, we describe the construction of a new bovine BAC library(RPCI-42) with the aim of increasing genome coverage and aver-age insert size, thus improving the probability of finding BACclones for target DNA sequences and the crucial goal of gap clo-sure between contigs. We report the characterization of this libraryfor insert size and coverage of the bovine genome.

Genomic DNA was isolated from a single Holstein male’swhite blood cells, subjected to partialEcoRI andEcoRI methylaserestriction enzyme digestion, size selected on agarose gels, andligated to the pBACe3.6 vector as described (Osoegawa et al.1998). Ligation products were transformed by electroporation intotheDH10Bstrain (BRL Life Technologies) to create two segmentsof the RPCI-42 library. Transformants were picked into 384-wellformat for growth, storage, and replication. BAC DNA was pre-pared as described (http://www.chori.org/bacpac) and subjectedto pulsed field gel electrophoresis. High-density grids representingboth segments of the library were prepared on nylon membraneswith a simple indexing system for clone identification as described

(Osoegawa et al. 1998). Separate oligonucleotides designed to becomplementary at their 38 ends with 8 bp of overlap, thus referredto as overgo-oligos, were designed for labeling by primer exten-sion in the presence of radionucleotides by using bovine cDNA orEST sequence (http://genome.wustl.edu/gsc/overgo/overgo.html).Alternatively, selected bovine gene segments, originating from ge-nomic subclones or cDNA clones, were random prime labeled(Megaprime, Amersham) according to the manufacturer’s proto-col. Pre-hybridization for oligo-based screening was 1–4 h at 60°Cin 1 mM EDTA, 7% SDS, 0.5M Na2PO4 buffer, followed byhybridization with a pool of 11 probes for 12–14 h in the samebuffer. Pre-hybridization when using the random primed probeswas 1–2 h at 65°C in 5× Denhardts solution, 0.5% SDS, followedby hybridization with pools of 10–13 probes for 16–24 h at 65°Cin 7% SDS, 0.5M Na2PO4, 1% BSA. Oligo blots were washed at60°C in the following order; once in 1× SSC, 1% SDS, 40 mM

Na2PO4 for 30 min; twice in 1.5× SSC, 0.1% SDS buffer for 20min each; and once in 0.5× SSC, 0.1% SDS buffer for 5 min.Random primed probes were rinsed three times with 1× SET, 0.5%SDS, and washed one time each for 5–10 min with 0.5× SET, 0.5%SDS, and 0.2× SET, 0.5% SDS at 65°C. Filters were sealed inplastic bags and exposed to XAR5 film, or to phosphor screens forscanning on a phosphoimager (Molecular Dynamics). Clones cor-responding to positive signals were transferred from the primarylibrary plates into secondary plates and spotted onto nylon filtersfor secondary screening with individual probes as above. Indi-vidual clones were verified to contain the appropriate target geneby amplification with gene-specific primers, designed usingPrimer3 software (Steve Rozen and Helen Skalepsky; http://www/genome.wi.mit.edu/genome-software/other/primer3.html).

The coverage of the BAC library is a function of the averageinsert size and the total number of clones. After electroporation,216,439 BAC transformants were picked and cultured in 384-wellplates. The number of clones that failed to grow was approxi-mately 1% (Table 1). Non-recombinant “empty” clones representapproximately 1% of the library (Table 1) as detected by hybrid-ization with a probe made from pUC plasmid (corresponding to the“stuffer fragment” found in the cloning site of the original cloningvector) to membranes onto which the BAC library had been grid-ded at high density. Therefore, the RPCI-42 library contains ap-proximately 212,000 total bovine genomic clones. Individualclones (158 total) were picked at random throughout both seg-ments of the library to determine the average insert size. The insertsizes ranged from 120 to 250 kbp with an average size of 164 kbp(Table 1). Given this average insert size and the number of clonesisolated, the library represents from 10 to 12 genome equivalents(assuming a genome size of 3 × 109 bp).Correspondence to:W. Warren, Email: wesley.c.warren@monsanto.co

Mammalian Genome 11, 662–663 (2000).DOI: 10.1007/s003350010126

© Springer-Verlag New York Inc. 2000

Incorporating Mouse Genome

The apparent coverage of the library was also tested by deter-mining the number of independent clones carrying specific targets,by hybridization of gene-specific probes to the grid membranes.Two types of probes were used for this hybridization, randomprime-labeled gene segments (from cDNAs or genomic subclones)and overgo-oligos. A total of 19 random primed and 11 overgo-oligo probes were utilized to make this estimate. All probes iden-tified at least one positive BAC clone. Hybridization with the 19random prime-labeled probes resulted in the identification of 170BAC clones that could be confirmed upon secondary hybridiza-tion. Hybridization-positive clones for 14 of the 19 random primedprobes were successfully verified by PCR with gene-specific prim-ers. Verification primers for the other five random prime probes,all of which represented cDNA probes, failed to amplify specificproducts from genomic DNA or any of the BAC clones, possiblyowing to large introns in the genomic sequence. For the 14 genesin which the PCR confirmation protocol was successful, 91% ofstrong primary hybridization signals corresponded to clones thatcould be confirmed. The failure to amplify the remaining 9% ofclones could be owing to BACs that have only a portion of thetarget gene, or to incorrect BAC clone identification resulting fromcross-hybridization to related sequences. For the five probes inwhich the verification protocol failed, the number of verifiedclones is estimated by using 91% as a correction factor. The av-erage number of positive clones including all data was 8.0 ± 4.1(range 4–18).

In contrast to the results with random primed probes, the 11overgo-oligo probes resulted in the identification of 64 positiveBAC clones, all of which were verified to be positive. The averagenumber of clones per probe with overgo-oligo probes was 5.8 ± 2.9(range 1–11). Since the range of errors in these measurementsoverlap, it is possible that the apparent difference in screeningresults is a simple reflection of the low number of probes of eachtype used. Alternatively, the difference could be a real reflection ofa difference in sensitivity with short, overgo-oligo probes vs. thelonger cDNA or genomic fragment probes, or could reflect theextended hybridization target for cDNA probes whose genes mayextend over several kilobases in the genomic clones. Using thedata from all confirmed positive clones, we estimate that the RPCI-42 library has 7.4 ± 3.8 BAC clones corresponding to each specificprobe.

Cai et al. (1995) reported an average of 1 BAC clone perbovine gene probe (n4 31 genes), with 25% of these probesfinding no corresponding BAC. From these parameters, it wasestimated that a given gene sequence has a 75% probability ofbeing found in this library. Zhu et al. (1999) reported primaryscreening by PCR, and hybridization yielded positive BACs for alltarget sequences evaluated, but did not report confirmation bysecondary screening. In summary, we believe the public availabil-ity of BAC library RPCI-42, in conjunction with the high-resolution genetic linkage maps, represents an important resourcefor the bovine and ovine mapping communities and will be auseful asset in construction of contigs spanning QTL intervals. Thebovine BAC library resource described in this publication can bepurchased by interested parties in academia or industry by usingthe following email address: BACPACorders@mail.cho.org

Acknowledgments.The authors acknowledge the technical assistance ofKevin Tennill, Brian Kirkpatrick, and Nengbing Tao.

References

Barendese W, Vaiman D, Kemp SJ, Sugimoto Y, Armitage SM et al.(1997) A medium-density linkage map of the bovine genome. MammGenome 8, 21–28

Cai L, Taylor JF, Wing RA, Gallagher DS, Woo SS et al (1995) Construc-tion and characterization of a bovine bacterial artificial chromosomelibrary. Genomics 29, 413–425

Georges M, Andersson L (1996) Livestock genomics comes of age. Ge-nome Res 6, 907–921

Georges M, Bielsen D, Mackinnon M, Mishra A, Okimoto R et al. (1995)Mapping quantitative trait loci controlling milk production in dairy cattleby exploiting progeny testing. Genetics 139, 907–920

Kappes SM, Keele JW, Stone RT, McGraw RA, Sonstegard TS et al.(1997) A second-generation linkage map of the bovine genome. GenomeRes 7, 235–249

Kelly JM, Field CE, Craven MB, Bocskai D, Kim U et al. (1999) Highthroughput direct end sequencing of BAC clones. Nucleic Acid Res 27,1539–1546

Osoegawa K, Woon PY, Zhao B, Frenger E, Tateno M et al. (1998) Animproved approach for construction of bacterial artificial chromosomelibraries. Genomics 51, 1–8

Soller M (1994) Marker assisted selection—an overview. Anim Biotechnol5, 193–207

Stone R, Keele JW, Shackelford SD, Kappes SM, Koohmaraie M (1999) Aprimary screen of the bovine genome for quantitative trait loci affectingcarcass and growth traits. J Anim Sci 77, 1379–1384

Venter JC, Smith HO, Hood L (1996) A new strategy for genome sequenc-ing. Nature 381, 364–366

Zhu B, Smith JA, Tracey SM, Konfortov BA, Welzel K et al. (1999) A 5×genome coverage bovine BAC library: production, characterization, anddistribution. Mamm Genome 10, 706–709

Table 1. Characteristics of bovine BAC library RPCI-42. Methods used for thecharacterization are described in Osoegawa et al. (1998).

SegmentNon-recombinantclones (%)

Emptywells(%)

Insert size(average kb)

Estimatedgenomiccoverage

1 0.6 1 165 62 1.4 1.2 163 5.9Mean 1 1.1 164 11.9 Sum

663