Bacterial Artificial Chromosome-Based Physical Map of the Rice Genome Constructed by Restriction Fingerprint Analysis

Corresponding author: Hong-Bin Zhang, Department of Soil and Crop Sciences and Crop Biotechnology Center, 2123 TAMUS, Texas A&M University, College Station, TX 77843-2123., hbz7049{at}pop.tamu.edu (E-mail)

Communicating editor: Z-B. ZENG

	ABSTRACT

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Genome-wide physical mapping with bacteria-based large-insert clones (e.g., BACs, PACs, and PBCs) promises to revolutionize genomics of large, complex genomes. To accelerate rice and other grass species genome research, we developed a genome-wide BAC-based map of the rice genome. The map consists of 298 BAC contigs and covers 419 Mb of the 430-Mb rice genome. Subsequent analysis indicated that the contigs constituting the map are accurate and reliable. Particularly important to proficiency were (1) a high-resolution, high-throughput DNA sequencing gel-based electrophoretic method for BAC fingerprinting, (2) the use of several complementary large-insert BAC libraries, and (3) computer-aided contig assembly. It has been demonstrated that the fingerprinting method is not significantly influenced by repeated sequences, genome size, and genome complexity. Use of several complementary libraries developed with different restriction enzymes minimized the "gaps" in the physical map. In contrast to previous estimates, a clonal coverage of 6.0–8.0 genome equivalents seems to be sufficient for development of a genome-wide physical map of 95% genome coverage. This study indicates that genome-wide BAC-based physical maps can be developed quickly and economically for a variety of plant and animal species by restriction fingerprint analysis via DNA sequencing gel-based electrophoresis.

GENOME-WIDE physical mapping using large-insert DNA clones is becoming the centerpiece of current genomics research of virtually all plant and animal species. Genome-wide physical maps provide essential platforms for large-scale genome sequencing, effective positional cloning, high-throughput expressed sequence tag (EST) physical mapping, and target DNA marker development. Bacteria-based large-insert clones, including bacterial artificial chromosomes (BACs; SHIZUYA et al. 1992 ), bacteriophage P1-derived artificial chromosomes (IAONNOU et al. 1994 ), and large-insert conventional plasmid-based clones (TAO and ZHANG 1998 ), have provided desirable resources for genomics research because of their high stability, low chimerism, and facility for large-scale DNA purification (ZHANG and WING 1997 ). To develop physical maps from bacteria-based large-insert clones, several approaches have been developed and used (for review, see ZHANG and WU 2001 ). These include hybridization-based methods such as iterative hybridization (e.g., MOZO et al. 1998 , MOZO et al. 1999 ; ZHU et al. 1999 ), restriction-based fingerprinting methods (COULSON et al. 1986 ; GREGORY et al. 1997 ; MARRA et al. 1997 , MARRA et al. 1999 ; ZHANG and WING 1997 ; DING et al. 1999 ; ZHU et al. 1999 ; HOSKINS et al. 2000 ; Y.-L. CHANG, Q. TAO, C. SCHEURING, K. MEKSEM and H.-B. ZHANG, unpublished results), and integrated BAC end sequencing, fingerprinting, and genome sequencing methods (VENTER et al. 1996 ; MAHAIRAS et al. 1999 ). Since the restriction-based fingerprinting method is not significantly affected by repeated sequences as is the iterative hybridization method and is much more rapid and economical than the integrated sequencing and fingerprinting method, it promises to provide a powerful means for rapid development of genome-wide physical maps from bacteria-based large-insert random clones.

In the restriction fingerprinting approach, the restricted fragments of clonal DNA were fractionated on either agarose gels (MARRA et al. 1997 ) or denaturing polyacrylamide DNA sequencing gels (COULSON et al. 1986 ; GREGORY et al. 1997 ; TAIT et al. 1997 ; ZHANG and WING 1997 ; TAO and ZHANG 1998 ; DING et al. 1999 ; ZHANG and WU 2001 ). In the DNA sequence electrophoresis-based restriction fingerprinting method, the restricted fragments of clones are end labeled with either a radioactive nucleotide (COULSON et al. 1986 ; ZHANG and WING 1997 ; TAO and ZHANG 1998 ) or a fluorescent dideoxynucleotide (GREGORY et al. 1997 ; TAIT et al. 1997 ; DING et al. 1999 ).

Validity of the restriction fingerprinting approach was first demonstrated by the development of genome physical maps of Saccharomyces cerevisiae (OLSON et al. 1986 ; RILES et al. 1993 ) and Caenorhabditis elegans (COULSON et al. 1986 ; HODGKIN et al. 1995 ) with cosmid or clones. Recently, BAC-based physical maps were developed for small genome species, Arabidopsis thaliana (130 Mb; MARRA et al. 1999 ; MOZO et al. 1999 ), chromosome 7 of Magnaporthe grisea (4.2 Mb; ZHU et al. 1999 ), and the major autosomes (120 Mb) of Drosophila melanogaster (HOSKINS et al. 2000 ) using integrated iterative or sequence-tagged site-based hybridization and agarose gel-based fingerprinting (MARRA et al. 1997 ) methods. However, the use of the restriction fingerprinting approach for development of genome-wide physical maps of large, complex genomes remains to be investigated. Unlike physical mapping of the small genome species, the development of global physical maps of large, complex genomes must fingerprint and analyze a large number of clones. Therefore, a high-resolution, high-throughput restriction fingerprinting method is needed to generate physical maps of large, complex genomes from large-insert random clones. The DNA sequence electrophoresis-based fingerprinting method (COULSON et al. 1986 ; GREGORY et al. 1997 ; ZHANG and WING 1997 ; TAO and ZHANG 1998 ; DING et al. 1999 ; ZHANG and WU 2001 ) is not only high in resolution (one nucleotide), which is several hundredfold higher than that of the agarose gel-based method (10–1000 bp; for review, see ZHANG and WU 2001 ), but also highly amenable to automation on automated DNA sequencers (GREGORY et al. 1997 ; DING et al. 1999 ) and to high throughput (ZHANG and WU 2001 ). Therefore, it should be suitable for genome-wide physical mapping of large, complex genomes from bacteria-based large-insert random clones. However, no genome-wide, BAC-based physical maps have been developed to date using the DNA sequence electrophoresis-based fingerprinting method. Demonstration of the feasibility and development of strategies for genome-wide physical mapping with BACs by this method will greatly enhance research of large, complex genomes. This result will also provide a basis of incorporating the newly developed capillary DNA automated sequencing technology into the fingerprinting method for genome-wide physical mapping of large, complex genomes with bacteria-based large-insert random clones.

Rice, Oryza sativa L., is considered to be a model species for genome research of monocotyledonous plant species because of its relative small genome size. It has a wealth of genetic and genomic resources and is well established in genetic transformation. Rice has a genome size of 430 Mb/1C (where 1C is the haploid genome; ARUMUGANATHAN and EARLE 1991 ) in which about 70% of the DNA is repetitive. The genome of rice is >3.5-fold larger than those of A. thaliana (LIN et al. 1999 ) and the major autosomes of D. melanogaster (HOSKINS et al. 2000 ) in size. Although a yeast artificial chromosome (YAC)-based physical map has been developed for rice by the Japan Rice Genome Program to facilitate rice genome research (SAJI et al. 2001 ), it covers only 63% of the rice genome. In addition, YACs are limited in applications for extensive genome research because they are relatively unstable and high in chimerism and their DNA is difficult to purify. Efforts are also being made to develop BAC-based physical maps for rice (http://www.genome.clemson.edu; http://rgp.dna.affrc.go.jp); however, no genome-wide, BAC-based physical maps of the rice genome have been reported to date. Furthermore, all these efforts are working with japonica rice (cv. Nipponbare), which accounts for <10% of world rice production. In this study, we developed a genome-wide BAC-based physical map of indica rice, which accounts for >90% of world rice production, from three complementary large-insert BAC libraries, and demonstrated the feasibility of and developed strategies for genome-wide physical mapping with bacteria-based large-insert random clones using the DNA sequence electrophoresis-based fingerprinting method. Contig reliability of the physical map was verified using different approaches and the results indicate that the physical map is reliable and provides a readily used framework for genomics research of monocotyledonous plants. The results of this study have provided a paradigm for rapid development of genome-wide physical maps of plant and animal genomes from bacterial clone-based, large-insert random clones.

	MATERIALS AND METHODS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

BAC libraries and DNA markers:
Three O. sativa ssp. indica cultivar Teqing BAC libraries were used to develop the BAC-based physical map of the rice genome because >90% of the world rice production is indica rice. The libraries were constructed in the HindIII site of pBeloBAC11 (KIM et al. 1996 ; ZHANG et al. 1996 ), the BamHI and EcoRI sites of pECBAC1 (FRIJTERS et al. 1997 ; H.-B. ZHANG, unpublished results), respectively, and have average insert sizes of 130, 150, and 147 kb, respectively. The vector pECBAC1 was derived from pBeloBAC11 by knocking out the EcoRI site in its chloramphenicol resistance gene, thus making the EcoRI site in the multiple cloning sites suitable for cloning. These BAC libraries are permanently maintained in 384-well microplates and publicly available at the GENEfinder Genomic Resources (formerly, the Texas A&M BAC Center) (http://hbz.tamu.edu-BAC Library-Library List).

The DNA markers were selected from the Cornell University (CAUSSE et al. 1994 ) and Japan Rice Genome Research Program (HARUSHIMA et al. 1998 ) rice genetic maps and kindly provided by S. McCouch and the Japan MAFF DNA Bank at the National Institute of Agrobiological Resources (http://bank.dna.affrc.go.jp). The random rice EST clones were kindly provided by Dupont Company (G.-H. Miao).

BAC fingerprinting and contig assembly:
BAC clones maintained in a 384-well microplate were inoculated in four 96-deep well plates containing 1 ml LB medium plus 12.5 µg/ml chloramphenicol and grown at 37° with shaking at 250 rpm overnight. BAC DNA was isolated and purified in the 96-deep well plates and then in 8- or 12-microtube strips using a modified alkaline lysis method (Q. TAO, Y.-L. CHANG, B. VINATZER and H.-B. ZHANG, unpublished results). The DNA was double-digested with HindIII and HaeIII, end labeled with [³²P]dATP using reverse transcriptase at 37° for 2 hr, and then subjected to 4.0% (w/v) polyacrylamide DNA sequencing gel electrophoresis at 85 W for 100 min. The gel was dried and autoradiographed.

The fingerprints on the autoradiographs were scanned into image files using a UMAX Mirage D-16L scanner. The image of the fingerprints was size adjusted to 1.1 MB, transferred to a computer workstation (SUN Microsystems, Utra10), and edited using the Image 3.8 of the FPC (FingerPrinted Contig) package (SULSTON et al. 1988 ; SODERLUND et al. 1997 ). The fragments ranging from 58 to 673 bases were used in contig assembly, on average, 22 bands per BAC fingerprint. The bands derived from the BAC vectors (pBeloBAC11 and pECBAC1) were manually deleted from the image files, and the clones without inserts were excluded.

The BAC contigs of the rice genome were assembled from the fingerprint database using the FPC 3.4 of the FPC package (SODERLUND et al. 1997 ) in two steps. We first assembled automated BAC contigs under highly stringent criteria (see below) to ensure that they are accurate. Then we joined automated contigs into larger contigs, using a less stringent criterion for the number of consensus bands (fewer common bands). When the fingerprints on the autoradiograph were scanned into image files, the original image size of each autoradiograph (35 x 43 cm) was 7.8 MB. To facilitate fingerprint analysis, we reduced the image size of each autoradiograph to 1.1 MB before transferring the image to the Image 3.8 of the FPC package at the computer workstation for data analysis. SODERLUND et al. 1997 recommended that tolerance 7 be suitable to build contigs from the fingerprints fractionated on polyacrylamide DNA sequencing gels. In our case, tolerance 3 was selected for contig assembly, which was equivalent to tolerance 7 for the original size of the autoradiograph image.

To select the cutoff values suitable for contig assembly, we used three DNA probes, adhA, psbA, and rbcL, that are approximately 50 kb apart on the barley chloroplast genome to screen the source rice BAC libraries and obtained 615 positive clones. We supposed that all positive clones should be assembled into a single contig if the tolerance values and cutoff scores were properly selected for contig assembly. After a series of tests according to this criterion, tolerance = 3 and cutoff = 10^-10–10^-18 were selected and used for the BAC physical map contig assembly. The other software parameters used were Diff = 0.3, MinBands = 5, Diffbury = 0.10, and Minends = 8. To achieve the best overlap, each contig was subjected to analysis at cutoff = 10^-4 and then by running "Calculation," and "Again" until the best result was obtained.

Library screening:
The rice BAC libraries or the BACs of the map contigs were double-spotted on Hybond N + membrane (Amersham, Piscataway, NJ) in a 3 x 3 format using the Biomek 2000 robotic workstation (Beckman, Fullerton, CA). The membranes were prepared following a published procedure (ZHANG et al. 1996 ). To estimate the realized genome coverage of the rice BAC libraries, the filters of the rice cv. Teqing and Lemont HindIII BAC libraries (ZHANG et al. 1996 ) were probed with 93 DNA markers selected from the rice genetic map (CAUSSE et al. 1994 ). To identify the BACs derived from chloroplast DNA, the filters prepared from the rice physical map BACs were hybridized with three chloroplast DNA probes (see above). The colony hybridization was performed as described at http://hbz.tamu.edu. In the post-hybridization, the filters were washed for three times in 0.1% SDS, 0.5x SSC at 65°, 30 min each wash.

To test the reliability of the rice map BAC contigs, the filters of the rice physical map BACs were probed with 77 markers selected from linkages 8, 11, and 12 of the existing rice genetic maps (CAUSSE et al. 1994 ; HARUSHIMA et al. 1998 ) and six random rice EST clones. Clone DNA was prepared by the conventional alkaline lysis method. The insert of each clone was released from its cloning vector by restriction enzyme digestion or PCR amplification using the DNA sequences immediately flanking the cloning site as primers. The insert DNA was purified with the GENECLEAN Kit according to its manufacturer (BIO 101, Vista, CA) and labeled with the Dig high primer labeling kit (Roche Molecular Biochemicals). The BACs on the filters were screened with row and column probe groups of the DNA markers, respectively, with nine DNA markers per probe group. The positive clones of each probe were identified by cross-hybridization between the column and row probe groups to the filters. The BAC clone filters were transferred into an appropriate amount of prewarmed Dig prehybridization buffer (5x SSC, 0.1% N-lauroylsarcosine, 0.02% SDS, and 1.0% blocking reagent) and incubated at 65° for 1 hr with gentle agitation. Then the hybridization was conducted by adding denatured Dig-labeled probes to the prewarmed hybridization buffer, mixing well, transferring the filters from the prehybridization buffer into the probe/hybridization buffer mixture, and incubating at 65° with gentle agitation overnight. The filters were washed in 2x SSC, 0.1% SDS for two times, 5 min each time, at room temperature, followed by two washes in 0.1x SSC, 0.1% SDS, 15 min each wash, at 65°. The hybridization signals were detected with the Detection Starter Kit II according to the manufacturer (Roche Molecular Biochemicals).

	RESULTS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Development of a genome-wide BAC-based physical map of the rice genome:
Bacteria-based large-insert clone libraries of truly high-genome coverage are of significance for genome-wide physical mapping by restriction fingerprint analysis. To develop a BAC-based physical map of the rice genome, we previously developed two large-insert rice BAC libraries, the Teqing HindIII and Lemont HindIII BAC libraries (ZHANG et al. 1996 ). To test the true genome coverage of the libraries, we screened the Teqing HindIII BAC library with 97 mapped DNA markers. The Teqing HindIII BAC library has a theoretical genome coverage of 98% (4.4 x genome coverage; ZHANG et al. 1996 ). Surprisingly, the result showed that only 83% of the DNA markers gave one or more positive BACs—there was a 15% difference between the theoretical and realized genome coverage. To further test the relationship between the number of clones in a BAC library and its true genome coverage, we screened the Lemont HindIII BAC library with the same set of the DNA markers. The Lemont HindIII BAC library has a theoretical genome coverage of 97% (2.6 x genome coverage; ZHANG et al. 1996 ). The result was also surprising in that 85% of the DNA markers gave one or more positive clones in at least one of these two rice HindIII BAC libraries. This result indicates that it is necessary to develop several individual source BAC libraries with different enzymes in order to develop a genome-wide physical map of a high-genome coverage. Therefore, we constructed two additional Teqing BAC libraries with BamHI and EcoRI (H.-B. ZHANG, unpublished results), respectively, to develop the genome-wide BAC-based physical map of the rice genome. The three rice cv. Teqing BAC libraries have average insert sizes of 130, 150, and 147 kb, respectively (see http://hbz.tamu.edu-BAC Library-Library List).

We used the DNA sequencing gel-based, radioactive nucleotide labeling method to generate BAC fingerprints (e.g., see Fig 1). A total of 21,087 BACs, covering 6.9 x rice haploid genomes, were fingerprinted on 380 autoradiographs. Of these clones, 3.1 x genome BACs were randomly selected from the HindIII library, 1.7 x genome BACs from the EcoRI library, and 2.1 x genome BACs from the BamHI library. The BAC fingerprints were scanned into image files, edited, and created into FPC database. The overlapping clones were assembled into contigs using the FPC program (SODERLUND et al. 1997 ). From the BAC fingerprint database, the FPC assembled 585 contigs, designated hereon as "automated contigs" (Table 1). With the FPC program, it was established that these 585 contigs encompassed 70,009 unique bands and each band, on average, represented a 6.3-kb fragment of a BAC clone. Therefore, the 585 contigs collectively cover 441 Mb in length. This collective physical length of the contigs is larger than the 430-Mb genome size of rice because most of the contigs are overlapped despite not being detected under the conditions used in the study. Of these automated contigs, the largest one (ctg13) contains 128 clones, encompassing 579 unique bands and spanning 3648 kb in length; 291 contigs contain 26 or more clones; 226 contain 10–25 clones; and 68 contain 5–9 clones. The contigs containing 4 or fewer clones were dismissed, and 1942 clones remained as singletons. We then manually analyzed every contig, extended the automated contigs with the End Extension program, and added the singletons to the contigs with the Singles Hit program of the FPC (SODERLUND et al. 1997 ). We assumed that if two contig end clones between contigs had 10 or more bands in common, they were claimed as overlapped. Only after careful comparison of the contig end clones, were suspected overlapping contigs merged to form "extended contigs." As a result, the number of contigs was reduced to 298 contigs (Table 2), encompassing 66,589 unique bands and collectively covering 419 Mb in length. The largest contig (ctg3) contains 257 clones, encompassing 972 unique bands and spanning 6.1 Mb in length. Eight hundred ninety-six clones remained as singletons, each of which consisted of four or fewer bands that were insufficient to be included in contig assembly. Both the automated contigs and extended contigs are posted at http://hbz.tamu.edu-Physical Mapping-Indica Rice Map. Fig 2 shows an example of the automated BAC contigs of the map and the distribution of the BACs from three complementary BAC libraries in the contig.

View larger version (60K): In this window In a new window Download PPT slide   Figure 1. Example of the autoradiographs of BAC fingerprints that were used for contig assembly of the rice BAC-based physical map. DNA markers ( DNA/Sau3AI) were used in the first lane and every ninth lane thereafter. The fragments of BAC DNA were labeled with [32P]dATP and the fragments of marker DNA were labeled with [33P]dATP. The fingerprints were fractionated on a 4% (w/v) denaturing polyacrylamide DNA sequencing gel. The band appearing in all BAC lanes was derived from the BAC cloning vector pBeloBAC11 (KIM et al. 1996 ), which was manually deleted during fingerprint image editing.

View larger version (25K): In this window In a new window Download PPT slide   Figure 2. Example of the BAC contigs of the rice physical map showing the distribution of the BACs from the three complementary libraries (ctg148 in Table 1). The contig includes 101 clones and has a length of 412 unique bands, being equivalent to 2595 kb. The highlighted clones in blue color were from the rice cv. Teqing EcoRI BAC library, the highlighted clones in green color from the rice cv. Teqing BamHI BAC library, and the remaining clones from the rice cv. Teqing HindIII BAC library (see http://hbz.tamu.edu). Asterisk indicates a parent clone that covers one or more clones.

The reliability of the rice BAC-based physical map:
We conducted the following experiments to test the reliability of the automated contigs of the map.

Chloroplast DNA BAC contig analysis: The chloroplast genome of rice is 140 kb in size. Therefore, all of the chloroplast DNA-derived BACs should be assembled into a single contig if the map contigs were assembled properly. We identified 615 chloroplast DNA-derived BACs from the entire database of the BACs using three chloroplast DNA probes (see MATERIALS AND METHODS) and checked their positions in the contigs. The result showed that 588 of them were in a single contig (data not shown) and 27 were as singletons. The 27 singleton BACs were excluded from their assembly into the contig because the fingerprint of each of them consisted of four or fewer bands that were insufficient to be included in the contig assembly. These 615 chloroplast DNA-derived BAC clones were from three BAC libraries, and the fingerprint data were collected from 380 autoradiographs generated by three scientists in different experiments. The assembly of all 588 chloroplast DNA derived BACs having five or more bands in each of their fingerprints into a single contig indicated that the tolerance and cutoff values were properly selected and the map contigs were properly assembled.

Screening the contig BACs with mapped DNA markers: We hypothesized that if the map contigs are "reliable," the BACs selected with a single-copy DNA marker should all be located to a single contig. To test this hypothesis, we screened the BACs of the contigs with 77 mapped DNA markers and six random EST clones. The result is shown in Table 3 and summarized in Table 4. Library screening showed that 61 of the 83 DNA markers and ESTs gave two or more positive clones, 18 gave one positive clone, and 4 gave no positive clone (Table 3 and Table 4). Note that of the 6.9 x genome coverage clones analyzed, 1.7 x, 2.1 x and 3.1 x genome coverage clones were selected from each library, respectively. The uneven numbers of clones from each library might result in 18 of the 83 markers identifying one positive clone. Overall, 79 of the 83 markers (95%) gave one or more positive clones, which is consistent with the estimate of the map contig genome coverage (97%) based on the total length of the contigs.

We then checked the positions of the BACs selected with each of the 61 markers that hybridized to two or more BACs in the 585 automated contigs. For 45 of the 61 markers, all of the clones selected with each marker were found to be members of a single contig (Table 3 and Table 4), indicating that the contigs containing these DNA markers were properly assembled. Furthermore, we investigated the clones selected by 2 or more closely linked DNA markers and found that they were located at a single contig in 28 cases (Table 1 and Fig 3). These results also agreed with the genetic maps (CAUSSE et al. 1994 ; HARUSHIMA et al. 1998 ) from which the DNA markers were selected and thus further verified the reliability of these contigs.

View larger version (20K): In this window In a new window Download PPT slide   Figure 3. Example of the contigs of the rice BAC-based map containing the positive clones of four DNA markers (ctg207 in Table 1). This contig contains 65 clones and has a length of 285 unique bands, estimated equivalent to 1796 kb. Note that four DNA markers, CDO464, RG28, C277, and C390, were located to this contig, all of which were also located at the same region of linkage group 8 of both rice genetic maps (CAUSSE et al. 1994 ; HARUSHIMA et al. 1998 ). The highlighted clones indicate the positive clones of C390 and CDO464. Asterisk indicates a parent clone that covers one or more clones.

BAC screening with the DNA markers showed that BACs identified by each of the remaining 16 markers were members of two or more contigs. For these 16 markers, it was possible that some of them actually detected two adjacent contigs that could not be linked by fingerprint analysis although further investigation is needed to establish this. The localization of the clones selected with each of the 16 DNA markers at two or more contigs could also be due to the multiple copies of the DNA markers in the rice genome, contig assembly errors or both. To answer this latter question, we investigated the copy number of the 16 markers in the rice genome by Southern hybridization. At the Japan Rice Genome Program website (http://www.dna.affrc.go.jp:84/publicdata/naturegenetics/ricegmap.html), we were able to find the restriction patterns of 7 of the 16 DNA markers. Southern hybridization patterns indicate that 5 of the 7 DNA markers are multiple copy and 2 are single copy in the rice haploid genome. It is estimated from these 7 DNA markers that 71% (5/7) of the 16 DNA markers ( x 16 = 11.4) are multiple copy in the rice genome. Therefore, it was possible to explain that those clones selected with such DNA markers were located on multiple contigs. If the 11.4 marker contigs were properly assembled, 92.5% [(45 + 11.4)/61] of the automated contigs of the rice physical map were properly assembled. Furthermore, we assumed that the association of the remaining 7.5% DNA markers with two or more contigs resulted from "misassembly" of some of the BACs selected with the markers although it was possible that they actually hybridized two adjacent overlapping contigs (see above). We studied the clones selected by single-copy markers (e.g., R1943) each of which was shown to be associated with BACs in two contigs. We found that most of the selected BACs were located on one of the two contigs and one or two on the other contig. This indicated that for the contigs that might have some errors in contig assembly, most of their BACs were properly assembled.

	DISCUSSION

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

We have successfully developed a genome-wide BAC-based physical map of indica rice from 21,078 BACs randomly selected from three complementary libraries by the DNA sequence electrophoresis-based restriction fingerprinting method. The map consists of 298 BAC contigs, which were merged from 585 automated contigs, and covers 97% of the rice genome. This may represent a slight overestimate because it is possible that some of the 298 contigs are overlapped even though the overlaps could not be detected by fingerprint analysis under the conditions used in this study. Since the method used in this study is well suited for contig assembly from large-insert random BACs derived from centromeric and rDNA regions (T. UHM, C. WU and H.-B. ZHANG, unpublished results), the contigs for these regions are included in the 298 contigs. Hybridization analysis of the chloroplast DNA BAC contig and screening of the physical map BACs with numerous DNA markers consistently indicate that the BAC contigs constituting the physical map are properly assembled. Consistence was also observed between this BAC-based physical map and the rice genetic maps (CAUSSE et al. 1994 ; HARUSHIMA et al. 1998 ; see Fig 3), which further verifies the reliability of the physical map contigs in a long range. The physical mapping result of the rice genome is strongly supported by that of the Arabidopsis genome using the approach employed in this study, in which nearly all contigs were tested to be accurate by the international Arabidopsis genome sequencing results (ARABIDOPSIS GENOME INITIATIVE 2000) and numerous mapped DNA markers (Y.-L. CHANG, Q. TAO, C. SCHEURING, K. MEKSEM and H.-B. ZHANG, unpublished results).

The BAC-based physical map of the rice genome is suitable for genomics research of rice and other grass species, including large-scale genome sequencing, effective positional cloning, high-throughput EST physical mapping, and target DNA marker development. First, although there is no published data available for comparison between the reliability of this map and those of the physical maps developed with other methods, it is possible that some errors exist in a genome-wide physical map developed with any or combined existing methods (see ZHANG and WU 2001 ). The development of the genome-wide physical map of A. thaliana using the method employed in this study (Y.-L. CHANG, Q. TAO, C. SCHEURING, K. MEKSEM and H.-B. ZHANG, unpublished results) is an indication of the powerfulness of the method for genome-wide physical mapping from large-insert random BACs. The accuracy of the A. thaliana physical map was verified by both the Arabidopsis genome sequencing results (ARABIDOPSIS GENOME INITIATIVE 2000) and numerous mapped DNA markers. Second, the rice map developed in this study has a 7.0x redundancy; i.e., about seven clones could be selected for any region of the map. To build the tiling clone path of the genome for the above research purposes, analysis of the BAC fingerprints in a target contig with an aid of computers (see below) would minimize, if not completely eliminate, the clones that were not assembled properly, if any. Third, for genome sequencing a BAC that is anchored to the region of interest is selected from its contig and sequenced. The 1–3 BACs that overlap with the sequenced BAC at each end are then selected and end sequenced. The end sequence analysis of the selected BACs against the sequenced BAC will further verify the selection of the BACs for continuous sequencing (MAHAIRAS et al. 1999 ). Fourth, the misassembled BACs, if any, in a contig of interest could be readily eliminated by refingerprinting the BACs of the contig, followed by contig reassembly. Because this experiment includes only the BACs of a target contig, it is much simpler than genome-wide physical mapping. The BACs that were previously assembled into the contig by chance (improperly) will be assembled as singletons and thus excluded, whereas the BACs that were correctly assembled will be reassembled into a single contig. Although this involves some additional work, it is manyfold simpler to develop contigs of interest from the genome-wide physical map contigs than from libraries by chromosome walking. Alternatively, the clones selected could also be verified by using the Clone–Fingerprint Map tool of the Genomic Information System (GIS) developed by this group (see below). Fifth, the BAC fingerprint database generated in this study has provided a means for chromosome walking and the construction of minimally overlapping clone tiling paths for the above research puproses via web-based tools. This is because the tiling clone path construction and chromosome walking can be directly conducted using the fingerprint database, without need of the assembled contigs by using the FPC Hitting tool (see http://hbz.tamu.edu-Physical Mapping-Indica Rice Map and MARRA et al. 1999). To facilitate the management and use of integrated physical maps of agricultural genomes, we have created a database, developed the GIS system (H. CHEN, Q. TAO, Y.-L. CHANG and H.-B. ZHANG, unpublished data), and posted the contigs of the indica rice physical map at http://hbz.tamu.edu-Physical Mapping-Indica Rice Map. Using the GIS, users can readily access the rice BAC fingerprint database and the physical map, perform chromosome walking on the rice genome, select clones and contigs of interest, and build contig tiling clone paths via WWW by using not only the FPC Hitting tool as MARRA et al. 1999 , but also four additional tools: Clone–Graphic Contig Map, Clone–Fingerprint Map, Contig No.–Graphic Contig Map, and Marker/EST–Positive Clones–Contig/PFC Hit/ Fingerprint Matches.

The indica rice BAC-based physical map has provided a readily used platform for genomics research of rice and other monocotyledonous species. Two major subspecies of O. sativa, indica rice and japonica rice, are cultivated. Although both are equally good as models for grass genome research and japonica rice cv. Nipponbare is being used in rice genome sequencing by an international rice genome sequencing consortium led by the Japan Rice Genome Program, >90% of the world rice production is indica rice. Therefore, the genome research of indica rice, the staple food of about half of the world population, is far more important than that of japonica rice for the world rice economy. Because of this, sequencing of the indica rice genome is also ongoing in several countries. Additionally, we are developing a genome-wide BAC-BIBAC-based physical map of japonica rice cv. Nipponbare using the method and strategies employed in this study (Y. LI and H.-B. ZHANG, unpublished data). The indica rice physical map reported here will provide a framework within which to perform evolutionary genomics research between the two rice subspecies and between rice and other gramineous crop plants. Studies have demonstrated that the gene content and order are highly conserved among the grass genomes (AHN and TANKSLEY 1993 ; AHN et al. 1993 ; MOORE et al. 1995 ; PATERSON et al. 1995 ; BENNETZEN et al. 1996 ; CHEN et al. 1997 ; DEVOS and GALE 1997 , DEVOS and GALE 2000 ). Therefore, the rice physical map developed in this study could also be used as a reference to expedite DNA marker development, gene identification, and gene cloning in gramineous crops with large genomes such as maize, wheat, and barley.

This rice genome BAC-based physical map represents the first report of the genome-wide physical mapping of large, complex genomes with large-insert, ordered random BACs using the DNA sequence electrophoresis-based restriction fingerprinting method. This method seems to offer a paradigm for genome-wide physical mapping of different plant and animal species of economic importance. The rice BAC-based map was developed in 1.5 scientist years. Similarly, we have developed a genome-wide BAC-BIBAC-based, integrated genetic, physical, and sequence map of the A. thaliana genome in 4 scientist months using the method and strategies of this study (Y.-L. CHANG, Q. TAO, C. SCHEURING, K. MEKSEM and H.-B. ZHANG, unpublished results). In addition, we are developing the genome-wide physical maps of soybean, chicken, wheat, and cotton from BACs and BIBACs using the method and strategies developed in this study. The physical mapping results of rice, A. thaliana, and other species have demonstrated that it is feasible to rapidly develop genome-wide physical maps of the genomes of crop plants, farm animals, and humans at a reasonable cost using the method and strategies used in this study.

This study indicates that genome-wide physical mapping by restriction fingerprint analysis is not significantly influenced by genome size, genome complexity, and/or abundance of repeated sequences. This result was further confirmed by fingerprint analysis of BACs of 14 different plant and animal species with genome sizes ranging from 120 to 23,000 Mb/1C and repetitive sequences from 10 to 95% of the genomes (our unpublished results). Use of several complementary, bacteria-based large-insert clone libraries developed with different restriction enzymes, respectively, is an efficient strategy for minimizing "gaps" in the physical map because such libraries are balanced in distribution of clones in the genome and thus are equivalent to physically sheared shotgun libraries. A similar strategy has been or is being used for the physical mapping of Arabidopsis (MARRA et al. 1999 ; Y.-L. CHANG, Q. TAO, C. SCHEURING, K. MEKSEM and H.-B. ZHANG, unpublished results), Drosophila (HOSKINS et al. 2000 ), and Neurospora crassa (KELKAR et al. 2001 ). The number of clones covering 6.0–8.0 haploid genomes seems to be sufficient for development of a genome-wide physical map of 95% genome coverage if they are truly random clones from the genome. This genome coverage of clones has been widely used for genome-wide shotgun genome sequencing (FLEISCHMANN et al. 1995 ; LIN et al. 1999 ) and confirmed by this and our Arabidopsis (Y.-L. CHANG, Q. TAO, C. SCHEURING, K. MEKSEM and H.-B. ZHANG, unpublished results) physical mapping results. A high-resolution electrophoresis system for fingerprint generation is crucial for ensuring the accuracy and reliability of contig assembly. This is especially true for genome-wide physical mapping of large, complex genomes because the data from tens of thousands of BAC fingerprints are needed to assemble the target physical map. The DNA sequence electrophoresis-based fingerprinting method has been proven to be reliable, high throughput, and economical for rapid genome-wide physical mapping of large, complex genomes with bacteria-based large-insert random clones. Furthermore, the physical mapping process could be further accelerated by a few fold by incorporating the newly developed capillary DNA automated sequencing technology into the fingerprinting approach.

	FOOTNOTES

¹ These authors contributed equally to this work.

	ACKNOWLEDGMENTS

The authors acknowledge Dr. S. McCouch at Cornell University and the Japan MAFF DNA Bank at the National Institute of Agrobiological Resources for kindly providing the DNA markers. This project was supported in part by Texas Agricultural Experiment Station (8536-203104), the Rockefeller Foundation (RF97001#555), and the Texas Higher Education Coordinating Board (999902-042).

Manuscript received December 6, 2000; Accepted for publication May 11, 2001.

	LITERATURE CITED

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

AHN, S. and S. D. TANKSLEY, 1993  Comparative linkage maps of the rice and maize genomes. Proc. Natl. Acad. Sci. USA 90:7980-7984[Abstract/Free Full Text].

AHN, S., J. A. ANDERSON, M. E. SORRELLS, and S. D. TANKSLEY, 1993  Homoeologous relationships of rice, wheat and maize chromosomes. Mol. Gen. Genet. 241:483-490[Medline].

Analysis of the genome sequence of the flowering plant Arabidopsis thaliana.. (2000) Nature 408:796-815[Medline].

ARUMUGANATHAN, K. and E. D. EARLE, 1991  Nuclear DNA content of some important plant species. Plant Mol. Biol. Reporter 9:208-218.

BENNETZEN, J. L., P. SANMIGUEL, C.-N. LIU, M. CHEN, A. TIKHONOV et al., 1996 Microcolinearity and segmental duplication in the evolution of grass genomes, pp. 1–3 in Unifying Plant Genomes edited by J. S. HESLOP-HARRISON. Company of Biologists, Ltd., Cambridge.

CAUSSE, M. A., T. M. FULTON, Y. G. CHO, A. N. AHN, and J. CHUNWONGSE et al., 1994  Saturated molecular map of the rice genome based on an interspecific backcross population. Genetics 138:1251-1274[Abstract].

CHEN, M., P. SANMIGUEL, A. C. DE OLIVEIRA, S.-S. WOO, and H.-B. ZHANG et al., 1997  Microlinearity in sh-2-homologous regions of the maize, rice, and sorghum genomes. Proc. Natl. Acad. Sci. USA 94:3431-3435[Abstract/Free Full Text].

COULSON, A., J. SULSTON, S. BRENNER, and J. KARN, 1986  Toward a physical map of the genome of the nematode Caenorhabditis elegans.. Proc. Natl. Acad. Sci. USA 83:7821-7825[Abstract/Free Full Text].

DEVOS, K. M. and M. D. GALE, 1997  Comparative genetics in the grasses. Plant Mol. Biol. 35:3-15[Medline].

DEVOS, K. M. and M. D. GALE, 2000  Genome relationships: the grass model in current research. Plant Cell 12:637-646[Abstract/Free Full Text].

DING, Y., M. D. JOHNSON, R. COLAYCO, Y. J. CHEN, and J. MELNYK et al., 1999  Contig assembly of bacterial artificial chromosome clones through multiplexed fluorescence-labeled fingerprinting. Genomics 56:237-246[Medline].

FLEISCHMANN, R. D., M. D. ADAMS, O. WHITE, R. A. CLAYTON, and E. F. KIRTNESS et al., 1995  Whole-genome random sequencing and assembly of Haemophilus influenzae Rd. Science 269:496-511[Abstract/Free Full Text].

FRIJTERS, A. C. J., Z. ZHANG, M. VAN DAMME, G.-L. WANG, and P. C. RONALD et al., 1997  Construction of a bacterial artificial chromosome library containing large EcoRI and HindIII genomic fragments of lettuce. Theor. Appl. Genet. 94:390-399.

GREGORY, S. G., G. R. HOWELL, and D. R. BENTLEY, 1997  Genome mapping by fluorescent fingerprinting. Genome Res. 7:1162-1168[Abstract/Free Full Text].

HARUSHIMA, Y., M. YANO, A. SHOMURA, M. SATO, and T. SHIMANO et al., 1998  A high-density rice genetic linkage map with 2275 markers using a single F2 population. Genetics 148:479-494[Abstract/Free Full Text].

HODGKIN, J., R. H. A. PLASTERK, and R. H. WATERSTON, 1995  The nematode Caenorhabditis elegans and its genome. Science 270:410-414[Abstract/Free Full Text].

HOSKINS, R. A., C. R. NELSON, B. P. BERMAN, T. R. LAVERTY, and R. A. GEORGE et al., 2000  A BAC-based physical map of the major autosomes of Drosophila melanogaster.. Science 287:2271-2274[Abstract/Free Full Text].

IAONNOU, I., C. T. AMEMIYA, J. GARNES, P. M. KROISEL, and H. SHIZUYA et al., 1994  A new bacteriophage P1-derived vector for propagation of large human DNA fragments. Nat. Genet. 6:84-89[Medline].

KELKAR, H. S., J. GRIFFITH, M. E. CASE, S. F. COVERT, and R. D. HALL et al., 2001  The Neurospora crassa genome: cosmid libraries sorted by chromosome. Genetics 157:979-990[Abstract/Free Full Text].

KIM, U.-J., B. W. BIRREN, T. SLEPAK, V. MANCINO, and C. BOYSEN et al., 1996  Construction and characterization of a human bacterial artificial chromosome library. Genomics 34:213-218[Medline].

LIN, X., S. KAUL, S. ROUNSLEY, T. R. SHEA, and M.-I. BENITO et al., 1999  Sequence and analysis of chromosome 2 of the plant Arabidopsis thaliana.. Nature 402:761-768[Medline].

MAHAIRAS, G. G., J. C. WALLANCE, K. SMITH, S. SWARTZELL, and T. HOLZMAN et al., 1999  Sequence-tagged connectors: a sequence approach to mapping and scanning the human genome. Proc. Natl. Acad. Sci. USA 96:9739-9744[Abstract/Free Full Text].

MARRA, M. A., T. A. KUCABA, E. D. GREEN, B. BROWNSTEIN, and R. K. WILSON et al., 1997  High throughput fingerprint analysis of large-insert clones. Genome Res. 7:1072-1084[Abstract/Free Full Text].

MARRA, M. A., T. KUCABA, M. SEKHON, L. HILLER, and R. MARTIENSSEN et al., 1999  A map for sequence analysis of the Arabidopsis thaliana genome. Nat. Genet. 22:265-270[Medline].

MOORE, G., K. M. DEVOS, Z. WANG, and M. D. GALE, 1995  Gramineouses, line up and form a circle. Curr. Biol. 5:737-739[Medline].

MOZO, T., S. FISCHER, S. MEIER-EWERT, H. LEHRACH, and T. ALTMANN, 1998  Use of the IGF BAC library for physical mapping of the Arabidopsis thaliana genome. Plant J. 16:377-384[Medline].

MOZO, T., K. DEWAR, P. DUNN, J. R. ECKER, and S. FISCHER et al., 1999  A complete BAC-based physical map of the Arabidopsis thaliana genome. Nat. Genet. 22:271-275[Medline].

OLSON, M. V., J. E. DUTCHIK, M. Y. GRAHAM, G. M. BRODEUR, and C. HELMS et al., 1986  Random-clone strategy for genomic restriction mapping in yeast. Proc. Natl. Acad. Sci. USA 83:7826-7830[Abstract/Free Full Text].

PATERSON, A. H., Y.-R. LIN, Z. LI, K. F. SCHERTZ, and J. F. DOEBLEY et al., 1995  Convergent domestication of cereal crops by independent mutations at corresponding genetic loci. Science 269:1714-1717[Abstract/Free Full Text].

RILES, L., J. E. DUTCHIK, A. BAKTHA, B. K. MCCAULEY, and E. C. THAYER et al., 1993  Physical maps of six smallest chromosomes of Saccharomyces cerevisiae at a resolution of 2.6 kilobase pairs. Genetics 134:81-150[Abstract].

SAJI, S., Y. UMEHARA, B. A. ANTONIO, H. YAMANE, and H. TANOUE et al., 2001  A physical map with yeast artificial chromosome (YAC) clones covering 63% of the 12 rice chromosomes. Genome 44:32-37[Medline].

SHIZUYA, H., B. BIRREN, U.-J. KIM, V. MANCINO, and T. SLEPAK et al., 1992  Cloning and stable maintenance of 300-kilobase-pair fragments of human DNA in Escherichia coli using an F-factor-based vector. Proc. Natl. Acad. Sci. USA 89:8794-8797[Abstract/Free Full Text].

SODERLUND, C., I. LONGDEN, and R. MOTT, 1997  FPC: a system for building contigs from restriction fingerprinted clones. Comput. Appl. Biosci. 13:523-535[Abstract/Free Full Text].

SULSTON, J., F. MALLETT, R. STADEN, R. DURBIN, and T. HORSNELL et al., 1988  Software for genome mapping by fingerprinting techniques. Comput. Appl. Biosci 4:125-132[Abstract/Free Full Text].

TAIT, E., M. C. SIMON, S. KING, A. J. BROWN, and N. A. R. GOW et al., 1997  A Candida albicans genome project: cosmid contigs, physical mapping, and gene isolation. Fungal Genet. Biol. 21:308-314[Medline].

TAO, Q. and H.-B. ZHANG, 1998  Cloning and stable maintenance of DNA fragments over 300 kb in Escherichia coli with conventional plasmid-based vectors. Nucleic Acids Res 26:4901-4909[Abstract/Free Full Text].

VENTER, J. C., H. O. SMITH, and L. HOOD, 1996  A new strategy for genome sequencing. Nature 381:364-366[Medline].

ZHANG, H.-B. and R. A. WING, 1997  Physical mapping of the rice genome with BACs. Plant Mol. Biol. 35:115-127[Medline].

ZHANG, H.-B. and C. WU, 2001  BAC as tools for genome sequencing. Plant Physiol. Biochem. 39:195-209.

ZHANG, H.-B., S. CHOI, S.-S. WOO, Z. LI, and R. A. WING, 1996  Construction and characterization of two rice bacterial artificial chromosome libraries from the parents of a permanent recombinant inbred mapping population. Mol. Breed. 2:11-24.

ZHU, H., B. P. BLACKMON, M. SASINOWSKI, and R. A. DEAN, 1999  Physical map and organization of chromosome 7 in the rice blast fungus, Magnaporthe grisea.. Genome Res. 9:739-750[Abstract/Free Full Text].

This article has been cited by other articles:

				R. M. Stupar, K. A. Beaubien, W. Jin, J. Song, M.-K. Lee, C. Wu, H.-B. Zhang, B. Han, and J. Jiang Structural Diversity and Differential Transcription of the Patatin Multicopy Gene Family During Potato Tuber Development Genetics, February 1, 2006; 172(2): 1263 - 1275. [Abstract] [Full Text] [PDF]

				Z. Xu, M. A. van den Berg, C. Scheuring, L. Covaleda, H. Lu, F. A. Santos, T. Uhm, M.-K. Lee, C. Wu, S. Liu, et al. Genome physical mapping from large-insert clones by fingerprint analysis with capillary electrophoresis: a robust physical map of Penicillium chrysogenum Nucleic Acids Res., March 14, 2005; 33(5): e50 - e50. [Abstract] [Full Text] [PDF]

				C. Wu, S. Sun, P. Nimmakayala, F. A. Santos, K. Meksem, R. Springman, K. Ding, D. A. Lightfoot, and H.-B. Zhang A BAC- and BIBAC-Based Physical Map of the Soybean Genome Genome Res., February 1, 2004; 14(2): 319 - 326. [Abstract] [Full Text] [PDF]

				C. Ren, M.-K. Lee, B. Yan, K. Ding, B. Cox, M. N. Romanov, J. A. Price, J. B. Dodgson, and H.-B. Zhang A BAC-Based Physical Map of the Chicken Genome Genome Res., December 1, 2003; 13(12): 2754 - 2758. [Abstract] [Full Text] [PDF]

				D. R. Fuhrmann, M. I. Krzywinski, R. Chiu, P. Saeedi, J. E. Schein, I. E. Bosdet, A. Chinwalla, L. W. Hillier, R. H. Waterston, J. D. McPherson, et al. Software for Automated Analysis of DNA Fingerprinting Gels Genome Res., May 1, 2003; 13(5): 940 - 953. [Abstract] [Full Text] [PDF]

				S. Gong, X. W. Yang, C. Li, and N. Heintz Highly Efficient Modification of Bacterial Artificial Chromosomes (BACs) Using Novel Shuttle Vectors Containing the R6Kgamma Origin of Replication Genome Res., December 1, 2002; 12(12): 1992 - 1998. [Abstract] [Full Text] [PDF]

				N. KURATA, K.-I. NONOMURA, and Y. HARUSHIMA Rice Genome Organization: the Centromere and Genome Interactions Ann. Bot., October 1, 2002; 90(4): 427 - 435. [Abstract] [Full Text] [PDF]

				M. Chen, G. Presting, W. B. Barbazuk, J. L. Goicoechea, B. Blackmon, G. Fang, H. Kim, D. Frisch, Y. Yu, S. Sun, et al. An Integrated Physical and Genetic Map of the Rice Genome PLANT CELL, March 1, 2002; 14(3): 537 - 545. [Abstract] [Full Text] [PDF]

				Y.-L. Chang, Q. Tao, C. Scheuring, K. Ding, K. Meksem, and H.-B. Zhang An Integrated Map of Arabidopsis thaliana for Functional Analysis of Its Genome Sequence Genetics, November 1, 2001; 159(3): 1231 - 1242. [Abstract] [Full Text] [PDF]