Access to the Maize Genome: An Integrated Physical and Genetic Map

STRATEGIES TO ACCESS A COMPLEX CROP GENOME

Crop plant research is poised to make revolutionary strides including the following: cloning target genes based on their function and/or their position in the genome; documenting all genes and their interplay; defining and exploring all the existing genetic diversity in a species; and using functional information and syntenic relationships of genes in closely related species to extrapolate gene function in crop plants. The challenge, however, is to develop a set of comprehensive and systematic resources to facilitate these research endeavors. Genomic resources in maize (Zea mays) will undergird sequencing of the maize genome and will complement and contribute to research in the cereals, other grasses, and other crop plants.

For maize, developing genomics tools means facing the daunting realities of size and complexity. At approximately 2,500 megabases, the maize genome is comparable in size to that of humans, and the complexity is likely to be greater because of the abundance of multiple families of repetitive elements. Thus, gaining access to the maize genome is best tackled by following Goethe's advice to seek entry to the whole by going to its parts:

Willst du ins Unendliche schreiten,

Geh nur im Endlichen nach allen Seiten.

—Johann Wolfgang von Goethe

To efficiently take a genome apart and put it back together requires a combination of genetic and physical mapping. In principle, genetic mapping serves to subdivide and order the genome by crossing over and recombining parts, whereas physical mapping allows ordering of genomic fragments (parts) by determining the overlap among them. Accordingly, the components of the public effort to develop resources to access the complete maize genome are to produce a high-resolution genetic map densely populated with markers; to produce, fingerprint, and assemble a deep-coverage library of bacterial artificial chromosomes (BACs) into physical map segments; and through molecular markers to integrate the genetic and physical maps. The tools that are being built in maize will provide a scaffold upon which to hang the sequence and the gene constitution of the genome and will link the sequence to the collected efforts of the maize genetics community over the past century.

THE GENETIC MAP

A core resource essential for developing an integrated genetic/physical map for maize is a densely marked, high-resolution genetic map. We have constructed a >1850 marker map for the intermated B73/Mo17 (IBM) population (Davis et al., 2001). The parents of the population, B73 and Mo17, represent the two major heterotic groups of U.S. maize germplasm. The IBM population consists of 304 recombinant inbred lines that underwent four rounds of random mating at the F₂ stage. The additional meioses result in a 3-fold expansion of the genetic map (Liu et al., 1996). The combination of a large number of lines and the map expansion generate a map resource with approximately 17 times the order resolution power of the prior maize map standard (UMC 98 genetic map, Davis et al., 1999). The IBM map is populated with >1,000 RFLP and >850 simple sequence repeat (SSR) markers.

In addition to the map per se, numerous related resources are provided to the maize researcher. Seeds of the IBM lines can be obtained from the Maize Genetics Cooperation Stock Center (Urbana, IL). A collateral resource resulting from map development is numerous additional genetic markers. These markers include novel single copy RFLP probes, markers for Mutator insertion sites, and new SSR primer sets. Links to obtaining clones for hybridization probes, screening images for RFLP probes, primer and screening information for 1,800 maize SSR loci, and mapscore data for the IBM population are available from the Maize Genome Database (MaizeDB, http://www.agron.missouri.edu/). Twelve hundred SSR loci were developed by our group (Sharopova et al., 2001). Comparative maps for SSR loci not on the IBM map have also been constructed and displayed.

We have undertaken a number of initiatives promoting the use of the IBM map as a community resource. A subset of 94 IBM lines has been identified for general community use. We distribute DNA of the 94 lines in microtitre plate format for individual investigators to map their genes much as radiation hybrid panel DNA is available for mammalian gene mapping. We have implemented Web entry for submission of map score data, from which the resulting locus positions are returned to the investigator. These features are intended to provide the maize community at large with resources to contribute to generating comprehensive information on maize gene map position.

THE PHYSICAL MAP

As a first step in producing a physical map, we constructed three genomic DNA libraries in BACs using DNA from the inbred line B73. B73 was selected because it is one of the parents of the genetic mapping population; thus, markers mapped on the IBM population and used to screen the BAC libraries could provide anchors for connecting the genetic and physical maps. To ensure deep coverage and minimal gaps in sequence representation in the libraries, libraries were made using three different restriction enzymes, and each library contained sufficient numbers of clones to provide severalfold coverage of the haploid genome. Details of the three libraries are shown in TableI. Together, the libraries represent 27-fold genome coverage.

To assemble the BACs into contigs, the clones are fingerprinted by digesting with HindIII. The fingerprinting involves fractionating the fragments on agarose gels, scanning the gel images, digitizing them with IMAGE software (Sanger Center, UK), and analyzing for contig formation using Fingerprint Contig (FPC) software (Soderlund et al., 2000). Contigs are generated automatically with a cutoff value of 10⁻¹². Once all fingerprint data have been collected, the contigs will be edited manually to ensure accuracy and consistency.

Accuracy and speed of contig assembly can be enhanced by screening the BAC libraries with molecular markers. Resulting BAC addresses provide important anchor points for assembling individual BACs into contigs. A variety of markers is being used to screen the BAC libraries.

INTEGRATING THE PHYSICAL AND GENETIC MAPS

Current estimates indicate that at least 50% of the maize genome consists of complex arrays of retrotransposon-like elements and that the majority of these repetitive elements represent a small number of related families (San Miguel et al., 1996). There are no prior attempts to assemble a physical map for any eukaryotic organism with this structure. The size of the maize genome, 2,500 Mb, also contributes to the difficulty of assembly. With the first of 83,000 of the 450,000 BAC fingerprints completed, FPC assembled the BACs into 13,000 contigs and 9,300 singletons (Clemson University Genomics Institute [CUGI] Web site, August 2001). Assuming 450,000 BACs, 150 kb in length, representing 27× in coverage, and 80% overlap; fingerprinting alone is expected to result in approximately 2,000 apparent contigs (Lander and Waterman, 1988). Complexity in assembly caused by repetitive elements may cause that number to increase.

It is clear the task at hand is to place large numbers of genetically mapped anchor points against the BACs to both coalesce contigs and order contigs on the chromosome framework. To tie the assembled BAC contigs to the genetic map, the BACs must be screened with genetically mapped markers. The 90 core RFLP markers define the bin boundaries on the genetic map and set a framework for the integrated map. Many of the AFLP and MITE markers detected by screening the BAC pools are polymorphic in the IBM population and serve as genetic anchors. In addition, the sorghum markers are useful anchoring tools because they have all been mapped in sorghum, most have been mapped in maize, and many cross-hybridize to DNA from other cereals such as rice (Oryza sativa), sugarcane (Saccharum officinarum), and Pennisetum glaucum. Thus, these markers will not only help connect the BAC contigs to the maize map, but because of the colinearity of cereal genomes, they will also facilitate creation of comparative maps.

Of the 10,000 ESTs that will be used to screen the BAC libraries, we anticipate anchoring 1,000 contigs by EST and SSR markers currently on the IBM map. Many of these locations were determined by mapping SSRs that were derived from the public EST sequences. We are initiating development of 2,000 single-nucleotide polymorphism markers derived from the ESTs. Once these global anchoring approaches draw the majority of contigs to the genetic framework, directed sequencing of BACs and BAC ends will be used to derive the source sequence for development of additional single-nucleotide polymorphism markers to attempt to bring the physical map to closure.

At MaizeDB, a BLAST server has been implemented to allow users to compare sequences of interest to all public maize sequences (including those in the unigene set) to return a map location if known. The output of such a query is a bin location, BLAST score, and database links for more information, e.g. MaizeDB link to map details; CUGI link to BAC contigs; GenBank; and ZmDB (Zea mays database at Iowa State University) links for sequence and clone information.

Maps can be viewed at MaizeDB in three ways. For any single map, chromosome-specific views are available. For simultaneously looking at two genetic maps, a comparative map viewer has been developed by enhancing GIOT software obtained from the Rice Genome Project (Japan). This viewer shows side-by-side comparisons of similar regions of selected versions of maize maps and can be expanded to include comparisons of maize to sorghum and rice. For comparing the maize genetic map to the developing physical map—as marker data are obtained for the BAC contigs—the GIOT viewer can be adapted to allow queries by locus, marker name, or linkage group to display the associated physical contigs.

The genomic resources being developed for maize will serve as a skeleton upon which to hang the sequence and the gene constitution of the genome, and will link the sequence to a century of accumulated genetic studies by members of the maize genetics community. Genomic and genetic knowledge in other cereals, other grasses, and other plant species will complement the resources in maize and will be enhanced in concert with them.

MaizeDB can be accessed at http://www.agron.missouri.edu/, where maps, probes, primers, screening images, and lab protocols are presented.