INTRODUCTION

TOP ABSTRACT INTRODUCTION RESULTS DISCUSSION MATERIALS AND METHODS LITERATURE CITED

Maize (Zea mays) has a relatively large genome size of about 2,300 to 2,700 Mb (Arumuganathan and Earle, 1991). The genome size of many plant species differs as a result of variable amounts of repetitive DNA. Repetitive sequences make up a significant portion of the maize genome, estimated at approximately 50% to 73% (Bennetzen et al., 1998; Meyers et al., 2001; Walbot and Petrov, 2001). In recent years, rapid progress has been made in the rice (Oryza sativa) and Arabidopsis genome projects, making these organisms models for plant genome research (Bevan et al., 1998; Mozo et al., 1999; Yuan et al., 2000). Because of its large genome size, duplication of genomic regions (Gaut, 2001), low percentage of single-copy DNA, and high content of retroelements (Meyers et al., 2001), maize is a challenging target for genome analysis. However, the development of genetic and physical map resources is making maize genomics tractable.

Good genetic maps are an invaluable resource in many aspects of genome research because they enable the mapping of QTL or genes without any prior knowledge beyond phenotypic effects. They also provide a framework for anchoring the physical map. Davis et al. (1999) constructed a maize linkage map with 1,736 markers, including genomic and cDNA clones, isozymes, and expressed sequence tagged sites. A high-resolution genetic linkage map of maize with 978 simple-sequence repeat markers was recently constructed (Sharopova et al., 2002). The intermated B73 × Mo17 (IBM) population used for the simple-sequence repeat map has been used for additional mapping to produce a >1,800-marker map that is serving as the standard for ongoing genomics efforts (see http://www.maizemap/org). To facilitate an understanding of the genome sequence, the gene content, and the structure and function of the maize genome, it is necessary to construct integrated genetic and physical maps at all levels of resolution.

Successful construction of an integrated genetic and physical map in maize relies on the availability of deep-coverage large-insert genomic libraries. Yeast artificial chromosome (YAC) libraries were widely used in constructing physical maps in human (Cohen et al., 1993) and in many other plant species, including Arabidopsis (Grill and Somerville, 1991), tomato (Lycopersicon esculentum; Martin et al., 1992), and rice (Kurata et al., 1994). A maize YAC library has been constructed and is publicly available for genome research (Edwards et al., 1992). However, the maize YAC library is limited in its general use because of its high frequency of chimerism, low copy number, and low stability of clones (Haldi et al., 1994; Chumakov et al., 1995). Bacterial artificial chromosomes (BACs) have instead become the more popular choice for large-insert genomic libraries for structural genome research in plants, including Arabidopsis (Choi et al., 1995; Mozo et al., 1998), rice (Wang et al., 1995), sorghum (Sorghum bicolor; Woo et al., 1994), tomato (Hamilton, 1997), potato (Solanum tuberosum; Song et al., 2000), soybean (Glycine max; Tomkins et al., 1999), and wheat (Triticum aestivum; Lijavetzky et al., 1999). Compared with a YAC library, the BAC system has a low frequency of chimerism and a high stability of clones and is easy to manipulate.

In the current study, we report the characterization of three maize BAC libraries: the HindIII library made at the Clemson University Genomics Institute (South Carolina), and the EcoRI and the MboI libraries made at the Children's Hospital Oakland Research Institute (Oakland, CA). These libraries are being used to construct a physical map in maize and to anchor the genetic map (Cone et al., 2002). A set of complex repeat probes were hybridized to six high-density BAC filters from each of the three BAC libraries, to provide information on chromosome organization and on organellar DNA content. In addition, a second set of probes containing 90 maize RFLP core markers was screened against the HindIII library filters. These core markers function as bin delimiters and provide a framework for anchoring the BAC contigs to the IBM genetic map. The results are integrated with maize BAC fingerprinting data from the Arizona Genomics Institute (http://www.genome.arizona.edu/fpc/maize) to anchor BAC contigs to the IBM genetic map. The goal of producing a comprehensive, integrated genetic and physical map (see http://www.maizemap.org/iMapDB/iMap.html) is to facilitate map-based positional cloning, application of comparative mapping across grass species, and large-scale genome sequencing based on a minimum tiling path.

	RESULTS

TOP ABSTRACT INTRODUCTION RESULTS DISCUSSION MATERIALS AND METHODS LITERATURE CITED

BAC Library Screening with Complex Probes

Use of three libraries to make a physical map should minimize the underrepresentation of certain genomic regions arising from the use of a particular restriction enzyme (Frijters et al., 1997). To characterize and compare the quality of the libraries, high-density filter sets from all three libraries were constructed. Each filter set contained six filters, and the total of 18 filters collectively contained 331,776 BAC clones with an overall average insert size of 154 kb. The HindIII has an average insert size of 136 kb, which is equivalent to 6× haploid genome coverage (Tomkins et al., 2002). The EcoRI has an average insert size of 160 kb, and the MboI has an average insert size of 167 kb, each resulting in 7× haploid genome coverage (K. Osoegawa and J. Messing, unpublished data). These BAC filters were evaluated with a set of complex probes, which were aimed to provide information on chromosome organization and organellar DNA content. Probes in this set include the 185-bp knob repeat, ribosomal DNA, two telomere-associated repeat sequences, four centromere repeat sequences, mitochondrial DNA, , and a chloroplast DNA cocktail. All of the raw data for the high-density BAC filter hybridization are available from MaizeDB (http://www.agron.missouri.edu/bacs.html).

The percentages of positive clones in the three BAC libraries that were hybridized with the set of complex probes are shown in Figure 1A. In total, four centromeric repeat probes, CentA-ltr-11 (CentA-1), CentA-int-7 (CentA-2), CentC, and Cent4, hit 8,409 BAC clones (2.53%) of 331,776 clones tested. This number includes some duplicate hits because of BACs that cross-hybridized between four centromeric probes. Hybridization results of centromere repeat probes showed similar results for the percentage of BACs hit in all three BAC libraries. The Cent4 probe is a maize chromosome 4-specific centromere sequence. The results show that Cent4 hybridized to only 70 BAC clones (0.02%) in all three sets of filters, which is much lower than the number of hits for the other three centromeric repeat probes. The Cent4 sequence has been reported to share high homologies to CentC and the knob repeat and also to possess six copies of the common telomeric motif CCTAAA (Page et al., 2001), but our results show that BAC clones hybridized with Cent4 did not match those that hybridized with CentC and telomere probes. However, under low-stringency hybridization, Cent4 shows positive hybridization signals, which coincide with the ones to which the knob repeat probe hybridizes (results not shown). The coincident location of these sequence elements may suggest a common functional or structural role. By way of contrast, the number of positives for telomere repeat sequences was 4-fold higher in the EcoRI library. The hybridization results of centromeric and telomeric repeat probes show several contigs with multiple BAC hits, which can be assigned to each maize chromosome with further study. The contig data provide valuable information that will be a key resource for future sequence and structural analysis for those chromosomal regions.

View larger version (21K): [in this window] [in a new window]   Figure 1.   The HindIII, EcoRI, and MboI maize BAC libraries were hybridized with 12 complex probes. A, Eight repetitive DNA elements were used to identify the distribution of the repeat elements and chromosomal architecture. A set of probes including the four centromere repeat probes (CentA-1, CentA-2, Cent4, and CentC), two telomere-associated repeat sequences (Telo-1 and Telo-2), ribosomal DNA (pZMRI), and knob repeat (185 bp). B, Three organellar DNA, a multifragment chloroplast DNA probe and two different mitochondrial genome probes (Mito-A619 and Mito-Mo17, Mito-Mo17 were only hybridized to the maize HindIII BAC library), were hybridized to measure the extent of organellar contamination, and one -DNA probe was also tested.

The rDNA probe showed an even greater difference in hybridization results, the EcoRI and MboI positives were increased 7- and 16-fold, respectively, compared with the HindIII library. The rDNA of maize contains approximately 10,000 tandem repeats of a unit of 9.1 kb each. In a standard rDNA repeat, there are no HindIII recognition sites and only a single EcoRI site (McMullen et al., 1986). In contrast, there are numerous MboI sites within the repeat. The increased number of hits in MboI positives can be explained by rDNA being cloned with higher frequencies into the MboI library because of an increased number of the clonable restriction recognition sites, compared with the HindIII and EcoRI libraries.

The mitochondrial probe hybridization was performed using two different maize mitochondrial genomic probes (mito-A619 and mito-Mo17). The results show mito-A619 and mito-Mo17 hybridized to 535 (0.5%) and 335 (0.3%) BAC clones, respectively, from HindIII filters. Fauron et al. (1995) reported physical maps of three mitochondrial genomes from different maize cytotypes. These three maps show differences in organization of the sequences attributable to rearrangements and size differences attributable to the location and amount of repeat sequences. The differences in mitochondrial genome size and sequences among different maize cytotypes could explain the gap in hybridization results using two different mitochondrial genomes. In addition, nDNA contamination in the mito-A619 probe could account for the difference in total hits.

Figure 1B shows that a very low percentage of the HindIII, EcoRI, and MboI libraries consisted of chloroplast and mitochondrial clones, <0.68% and <1.46%, respectively. In total, 3,570 (1.08%) BAC clones of 331,776 clones tested were derived from organellar DNA. In all three libraries, fewer than 1,247 (0.5%) clones appeared to contain chloroplast DNA. This is substantially lower than the previously reported chloroplast contamination in sorghum and tomato BAC libraries, which were 10.5% and 1.1%, respectively (Budiman et al., 2000; Klein et al., 2000). BAC filters were hybridized with -DNA to check the extent of random contamination during the lab procedures. Only 29 BAC clones positive to -DNA were detected, indicating an exceptionally low percentage of -DNA contamination.

BAC Library Screening with RFLP Core Markers

To anchor the bin boundaries and to begin to provide a framework for anchoring the genetic and physical maps, probes representing the 90 core RFLP markers were used to screen the HindIII (six filters, representing 6× haploid genome equivalents) library. The core markers are evenly spaced in the genome and are the boundary markers for the chromosomal bin divisions on several maize genetic maps, including the IBM (Gardiner et al., 1993; Davis et al., 1999; see http://www.maizemap.org). All of the data for the high-density BAC filter hybridization with core markers are available from MaizeDB (http://www.agron.missouri.edu/bacs.html). Figure 2 shows the relationship between the probe copy number and the number of HindIII BACs identified by each probe. The graph of the best linear fit demonstrates that the actual number of hits is very close to the expected number of hits based on their copy number. Correlation analysis on copy number of the probe and number of BACs hit by each probe shows a strong relationship (r = 0.910, P = 4.05E²⁶) between these two variables. RFLP probe copy number was determined by the number of hybridized restriction fragments on Southern image. However, it is possible that two homoeologous bands from well-conserved genomic regions could have the same flanking restriction sites and therefore migrate simultaneously.

View larger version (28K): [in this window] [in a new window]   Figure 2.   Relationship between probe copy number and the number of BACs identified per probe. Dot denotes number of positive HindIII BAC clones for individual core markers. The best linear fit, Y = 1.74 + 5.99X, is given by the solid line, whereas the dotted line is the expected pattern.

About one-fourth of the core markers are single copy in the genome, and the remainder are present in two or more copies. Twenty-two of the core probes including 20 probes reported by Tomkins et al. (2002) and other two probes, p-asg8 (GenBank accession nos. G10762 and G10763; identified 14 BACs) and p-umc245 (GenBank accession nos. G13169 and G13170; identified 11 BACs) that represent single-copy sequences in the maize genome, identified an average of 7.4 ± 3.3 positive clones with a range of three to 15 positive clones. Because the library represented 6× genome equivalents, each single-copy sequence should be represented six times, if all sequences are equally represented in the library. A Student's t test was performed to compare the mean of actual positives, 7.4, versus expected number of 6.0. The statistical test (P = 0.120) indicates that the difference between the mean and the expected number is not statistically significant. The variation in the number of positive signals identified by the single-copy probes may be indicative of the effects of preferential cloning obtained from the use of the HindIII restriction enzyme or the effects of instability of certain sequences in the BAC vector.

The remaining core markers represent sequences that are present in the genome in more than one copy. Forty-nine markers with two to three copies gave three to 28 positive BAC clones. At least three positive clones were identified for all of the markers analyzed (Tables I and II). This hybridization result confirms that the HindIII library collectively provides good representative coverage of 90 sequences that are distributed throughout the maize genome.

Hybridization images of p-php20581 and p-umc64 show significant intensity differences even between positives, and thus, the scanned images of these two probes were scored with high stringency (high-density filter reader threshold level set at 0.8, other core markers were scored using the default threshold of 0.5) to score the positives only with strong signals. BAC data of these two probes scored at a high-stringency setting were integrated into fingerprint contig (FPC). The positive hits from the hybridization filters of the other four probes with exceptionally high numbers of positive BACs, p-bnl6.32, p-bnl7.08, p-umc108, and p-umc124, were not scored because there was no intensity difference between positives and because each probe showed more than 300 positives, which suggests possible existence of repeat elements within the probe sequences (data not shown).

After the scoring of hybridized filter images, sequences of all 90 RFLP clones were retrieved from the GenBank database and were used as queries in a BLAST search against all maize sequences. These analyses demonstrated that only eight probes contain multiple regions homologous to either known genes with repeat elements or a transposon. Six out of eight probes had many more positive signals on hybridized filters than expected by chance. Core markers, p-php20581, p-umc108, and p-bnl7.08, showed significant homology to a 22-kD zein sequence (AF031569). These three markers also showed significant sequence similarities to the chloroplast ATPase -subunit gene (X03396) and the -tubulin gene (AJ420857). Meyers et al. (2001) previously reported the 22-kD zein region as a genomic region containing repeat elements. The other three core markers also show significant homology to multiple known genes (Table III).

View this table: [in this window] [in a new window]   Table III.   Blast results from query of six core RFLP marker sequences with higher than expected total positive BACs against the GenBank database Significant hits (E < 1 × 105)a are listed.

Validation by Dot-Blot Hybridization and Southern Analysis

To further validate the BAC-addressing method and hybridization results, two types of DNA hybridization experiments were performed on selected clones. Dot-blot hybridization was performed using four hybridization probes, p-umc161, p-csu164, Cent4, and pZmRI (rDNA). Among the BAC clones scored as positive for these probes, two to four clones were randomly selected for the dot-blot analysis. Figure 3 shows all of the clones hybridized with the expected probe except for the negative control. This indicates that the addressing process was robust.

View larger version (31K): [in this window] [in a new window]   Figure 3.   Dot blot of positively scored BAC DNA probed with four RFLP markers (A, umc161; B, Cent4; C, csu164; and D, pZmRI). BAC clones are shown as follows: A1, 101F15; A2, negative control; A3, 244B02; B1, 18H07; B2, 14J07; B3, 132L19; B4, 132A03; C1, 124K13; C2, 124E13; C3, 200N03; C4, 200I12; D1, 163O18; D2, 165E02; D3, 233F11; and D4, 266C14. BAC clone 124E13 was used for negative control for p-umc161 dot blot; this clone was not scored as positive for hybridization with this probe.

A second type of validation was focused on analyzing BACs identified by two probes, which were thought to be single copy in the genome, p-tub1 and p-tub4. These probes were selected because they hybridized to a higher number of BACs than expectations (15 for p-tub1 and 10 for p-tub4) and because the BACs identified for each probe assemble into multiple contigs, instead of a single contig (WebFPC, October 24, 2001 version at 8× coverage; http://www.genome.arizona.edu/fpc/maize). To test the robustness of the BAC hybridization and contig assembly, several BACs associated with p-tub1 in five contigs and several others associated with p-tub4 in the other three contigs were digested with HindIII, and the fragments were then separated on agarose gels, blotted, and hybridized with the cognate probes. Figure 4A shows the ethidium bromide-stained gels and the corresponding blots.

View larger version (122K): [in this window] [in a new window]   Figure 4.   Analysis of 26 BAC clones digested with HindIII. A, Ethidium bromide-stained agarose gel. B, Southern blot of the gel in A after hybridization with radioactively labeled p-tub1 and p-tub4 probes. BAC clones are shown as follows from lane 1 through 27: 188N16, 158C08, 49G01, 121N08, 21E04, 207P12, 344O08, 391H16, 32E08, 11P17, 129G10, 220J21, 155I03, 2K20, 190E08, 296P19, 187D08, 8K16, 172P22, 20F09, 339L08, 169O03, 232P15, 252D07, 32E08, 11P17, and molecular marker (/HindIII). All of the clones except 11P17 (from EcoRI library) were from the HindIII library.

All of the BACs associated with p-tub1 (lanes 1-19), except one, share a common hybridizing band (Fig. 4B). The exceptional clone, 11P17 (lane 10), does not appear to be contiguous with the other clones based on three observations: 11P17 DNA hybridized poorly to the p-tub1 probe on the Southern analysis, the clone belonged to the contig by fingerprinting but was not detected when the BAC filters were screened by hybridization with p-tub1, and a recent version of WebFPC BAC assembly (May 3, 2002) placed 11P17 in another contig.

Figure 4B shows that four of the p-tub1 hybridizing clones (lanes 6, 7, 8, and 11) have an additional hybridizing band, which suggests a possible complication that may require further investigation to assemble a robust physical map of a recently duplicated genome such as that of maize. One possibility is that a tandem (or at least proximal) duplication of the tub1 gene exists and that all of the BACs (except 11P17) do truly represent one contiguous stretch of a single chromosome, which contains two p-tub1 loci. However, additional information will be needed before we can rule out the possibility that two unlinked, perhaps homoeologous, loci exist that have retained a sufficient level of DNA sequence conservation so that most restriction fragments are still the same, and therefore FPC groups them into a single contig.

Six of the BAC clones associated with p-tub4 (lanes 20-26) shared a hybridizing band of the same size. These Southern-blot results further validate our hybridization results and BAC-addressing procedure.

Contig Assembly and Anchoring to the Genetic Map

FPC (Soderlund et al., 2000) is a program that assembles clones into contigs based on fingerprints and markers. WebFPC (Soderlund et al., 2002) is a Java program that displays the results of FPC on the Web. Of the 90 RFLP probes hybridized, 82 probes hybridized to less than 50 BACs and thus were integrated into FPC contig assembly and contributed to anchor contigs to the genetic map (see Arizona Genomics Institute Web site, http://www.genome.arizona.edu/fpc/maize/). Four probes (p-bnl7.49, p-csu32, p-umc49, and p-umc169) hybridized with more than 50 BAC clones but have high copy numbers (8) based on Southern blots with genomic DNA, meaning large numbers of positives are legitimate (Table II). These data were not integrated into FPC to avoid ambiguity in assembly. The other four probes, p-bnl6.32, p-bnl7.08, p-umc108, and p-umc124, containing repetitive elements with over 300 positive BACs were also excluded for the FPC assembly.

Thirteen single-copy probes hit a single contig and thus could be unambiguously assigned to the IBM genetic map. Single-copy probes should associate with one contig, but nine single-copy RFLP markers hit two or more contigs. Further evaluation of the data will sort out these ambiguities, and manual editing of the FPC could merge these multiple contigs into one larger contig. Several probes with two contig hits have been merged into a single contig during the last few FPC updates (data not shown).

Probes hybridizing to two or more copies in the genome should hit multiple contigs from duplicate sequences in the genome. Among 34 duplicate-copy probes, nine probes hit two contigs. BAC contigs for maize core probe p-asg24 are shown in Figure 5. RFLP hybridization data of p-asg24 with maize genomic DNA indicates that this probe recognized two loci in the genome. This is consistent with the contig assembly data, which also indicates that there are two copies of p-asg24. The FPC display also shows that both contigs were not only hit by p-asg24 but also share an overgo CL743_1. This result shows robustness of hybridization results and thus supports the data generated by overgo hybridization and the contigs assembled by fingerprinting. However, further evaluation of the data is needed to anchor the contigs to their respective genetic positions. Twenty-one of 34 duplicate-copy probes hit three or more contigs. Four of the duplicate-copy probes hit only a single contig, indicating that one of the loci for these probes may not be represented in the HindIII BAC library or was not represented on this subset of filters. The possibility that these contigs could represent assemblies of well-conserved duplicated genomic regions will require further investigation.

View larger version (44K): [in this window] [in a new window]   Figure 5.   RFLP hybridization data for p-asg24 with maize genomic DNA indicates that this probe recognizes two loci in the genome. This is consistent with the contig assembly data, which also indicates that there are two copies of p-asg24. In contig 36 (A) and contig 231 (B), BAC clones hit by p-asg24 are highlighted in green.

Finally, we also used a probe to screen the HindIII library for BAC clones containing the R1 and B1 loci on chromosomes 10 and 2, respectively. A total of 16 clones were obtained. These clones were subjected to Southern-blot analysis (not shown) exhibiting clones with two different restriction patterns, one that corresponded 100% to WebFPC contig 318 (R1) and the other 100% to contig 65 (B1). From this example and the asg24, it appears that FPC can distinguish duplicate factors in the allotetraploid maize genome.

	DISCUSSION

TOP ABSTRACT INTRODUCTION RESULTS DISCUSSION MATERIALS AND METHODS LITERATURE CITED

Maize genetic research has a long history of traditional breeding and mutant mapping studies. Recent progress in maize high-resolution molecular genetic mapping (Davis et al., 1999; Sharopova et al., 2002) is helping to facilitate current molecular genomic research. It has also facilitated construction of comparative genetic maps among closely related grass species (Ahn et al., 1993; Zwick et al., 1998; Wilson et al., 1999; Tikhonov et al., 2000; Gaut, 2001), and molecular dissection of quantitative trait loci (Thornsberry et al., 2001).

The three BAC libraries reported in this paper have a total genome coverage of 26-fold, providing an accessible tool for genomic research. Tomkins et al. (2002) reported construction and characterization of the HindIII BAC library. Their results show that the HindIII BAC library is of good quality and allows >99% probability of recovery of any specific sequence of interest. Our hybridization results demonstrate that, despite its good quality, the HindIII library has a bias and that the bias is overcome by using libraries made by several restriction enzymes. The results for telomere and pZmRI (rDNA) probes suggest that each library has a different bias because of the use of a specific restriction enzyme for the construction of the BAC library. The lower percentage of rDNA clones in the HindIII library compared with the EcoRI and the MboI libraries suggests that other regions of the maize genome devoid of HindIII sites could also be underrepresented in the HindIII library. These libraries will provide enough coverage for future genome research, and they complement each other to minimize the underrepresentation of certain genomic regions caused by the use of a particular restriction enzyme for BAC library construction (Frijters et al., 1997). Our hybridization results are significant in that BAC clones have been identified that correspond to key chromosome architectural features such as centromeres, telomeres, and the rDNA. Further research will be necessary to anchor these BAC contigs to their specific chromosomes.

Hybridization of BAC filters with chloroplast, mitochondrial, and -DNA probes demonstrated that all three maize BAC libraries exhibit a low percentage of contamination with these sequences. The suitability of a BAC library for positional cloning, physical mapping, and the ultimate goal of whole-maize genome sequencing using a minimum tiling path depends on the ability to recover clones from specific regions by screening (Lijavetzky et al., 1999). Data from the 90 RFLP core marker hybridization show at least three positive BAC clones per probe, which confirms adequate genome coverage of the maize HindIII BAC library for the most of markers. However, this library can show biased representation because certain genomic regions lack HindIII restriction sites. The extensive level of genome coverage and low level of non-nuclear maize DNA contamination demonstrate the high quality of the three BAC libraries. The collective depth of these BAC libraries appears to be sufficient for retrieval of virtually any maize sequence.

The primary purpose of the core marker hybridizations is to provide a framework for genome assembly and sequencing. The data generally indicate that the 6× subset of HindIII clones was sufficient to anchor maize contig before manual editing. After identifying the library addresses of each positive signal, results were integrated into FPC analysis. Results from the hybridization of 71 single- and low-copy number probes generally confirmed BAC contigs assembled by fingerprinting and positively demonstrated the utility of single-copy sequences, such as single-copy RFLP and gene probes, as anchor points for tying the physical map to the genetic map. Ambiguities from paralogous and/or homoeologous sequences may be overcome by putting more genetically mapped markers onto the physical map, by screening BAC pools using locus-specific PCR primers, by manual editing of the FPC analysis, and perhaps by finer resolution fingerprinting approaches. Hybridization to additional subsets of the BAC libraries might also aid in assigning paralogous sequences to their correct genetic location. The three BAC libraries can currently be ordered from http://www.genome.clemson.edu/orders (the b library is the HindIII; the c library refers both EcoRI and MboI), and the hybridization data are publicly available from http://www.agron.missouri.edu/bacs.html. These public resources will be valuable for maize genome research, positional cloning, and comparative research between cereal plants.

	MATERIALS AND METHODS

TOP ABSTRACT INTRODUCTION RESULTS DISCUSSION MATERIALS AND METHODS LITERATURE CITED

Maize (Zea mays) Core RFLP and Complex Probes

Maize core RFLP probes used in this study included both cDNA and genomic DNA clones. They included agr (Mycogen Plant Sciences, Des Moines, IA), asg (Asgrow Seed, Galena, MD), bnl (Brookhaven National Laboratory, Upton, NY), csu (Chris Baysdorfer, California State University, Hayward), npi (Native Plants [Salt Lake City] and Pioneer-Hi-Bred International, Des Moines, IA), php (Pioneer-Hi-Bred International), and umc (University of Missouri, Columbia) clones and several known genes (Davis et al., 1999). Detailed information on these probes is available (see MaizeDB; http://www.agron.missouri.edu/probes.html). GenBank accession numbers for the clones are given in Tables I and II.

Repetitive probes were obtained from various sources. The three chloroplast-specific clones (pBHP20, pBPH134, and pBHE319; GenBank accession no. NC001666) were provided by Rod Wing (University of Arizona). These three chloroplast genes are evenly distributed around the 133-kb barley chloroplast genome. The 185-bp repeat (GenBank accession no. M35408) was from M.D.M. (McMullen et al., 1986). Telomere probes p-PMTY7SC (Telo-1; GenBank accession no. U39641) and p-PMTY9ER (Telo-2; GenBank accession no. U39642) were provided by J.G. Both telomeric subclones contained CA-rich regions with sporadic occurrences of the telomere repeat (Gardiner et al., 1996). Four centromere repeat clones, CentA-1, CentA-2, CentC, and Cent4, were provided by James Birchler (University of Missouri) and Kelly Dawe (University of Georgia, Athens). CentA is a medium-copy number-dispersed retrotransposon (GenBank accession no. AF082532; Ananiev et al., 1998), CentC is a tandem repeat element (GenBank accession nos. AF078922 and AF078923; Dong et al., 1998), and Cent4 is a centromere 4-specific sequence (GenBank accession no. AF242891; Page et al., 2001). Mitochondrial genome DNAs isolated from maize inbred lines A619 and Mo17 were provided by Kathleen Newton (University of Missouri).

BAC Library Screening

The HindIII library made at the Clemson University Genomics Institute has an average insert size of 136 kb with a genome coverage of 13.5× (Tomkins et al., 2002). For this study, only six high-density BAC filters, which provide genome coverage of 6×, were used. The EcoRI and the MboI libraries made at the Children's Hospital Oakland Research Institute have an average insert size of 160 and 167 kb, respectively, with a genome coverage of about 7×. High-density BAC filters were gridded in double spots using a 4 × 4 pattern with six fields per nylon (Hybond NT) filter. Each field consists of 16 × 24 boxes, and within each box, there are eight independent clones in duplication. This allows each filter to represent 18,432 independent maize BAC clones.

BAC hybridization screening was performed using six filters equivalent to 6× (HindIII) and 7× (EcoRI and MboI each) haploid genome coverage. Colony filters were processed and hybridized using standard protocol (Sambrook et al., 1989). Films were digitized on a large-bed ScanMaker 6400XL (Microtek Inc., Hsinchu, Taiwan), and the images were imported into the semi-automated high-density filter reader 1.0 (Incogen Inc., Clemson, SC) for the BAC scoring and addressing. Hybridization of the MboI library was performed at the Waksman Institute under the same conditions.

Clone Verification by Dot-Blot Hybridization and Southern-Blot Analyses

BAC clones giving positive signals upon hybridization were picked individually from the 384-well microtiter plates of the maize BAC library and inoculated into a starter culture of 5 mL of Luria-Bertani medium containing 12.5 µg mL¹ chloramphenicol for initial culture. After 8 h of incubation at 37°C, the starter culture was diluted into new Luria-Bertani medium, 1:1,000 (v/v), and grown overnight at 37°C. BAC DNA was isolated using the plasmid midi kit (Qiagen USA, Valencia, CA). Five hundred nanograms of DNA was transferred to an Immobilon NY+ nylon membrane (Millipore, Bedford, MA), and the DNA was fixed by UV cross-linking. Prehybridization and hybridization were done at 65°C in 50 mM Tris, 10 mM EDTA, 5× SSC, 1× Denhardts solution, 0.2% (w/v) SDS, and 100 µg mL¹ denatured salmon sperm DNA. After hybridization, the membranes were washed three times with low-stringency wash (2× SSC and 0.5% [w/v] SDS) and a final high-stringency wash (0.1× SSC and 0.1% [w/v] SDS) was done. Both high- and low-stringency washes were done at 65°C. The blots were exposed to x-ray films at 80°C overnight.

Restriction digestions were performed using HindIII for Southern-blot analysis. BAC DNA was electrophoresed in 0.8% (w/v) agarose gels at 50 V for a distance of about 10 cm. Gels were denatured for 30 min in 0.4 N NaOH and 0.6 M NaCl, followed by a 30-min neutralization in 0.5 M Tris, pH 7.5, and 1.5 M NaCl. DNA was transferred to Immobilon NY+ nylon membrane, and the rest of procedures were done based on the procedure described by Davis et al. (1999).

Homology Searching

All maize core RFLP markers were queried against public sequence databases using the BLASTN search engine. A sequence was classified as homologous to other sequences if a BLAST probability value (E-value) less than 10⁵.