Abstract
Maize is one of the most important food crops and a key model for genetics and developmental biology. A genetically anchored and high-quality draft genome sequence of maize inbred B73 has been obtained to serve as a reference sequence. To facilitate evolutionary studies in maize and its close relatives, much like the Oryza Map Alignment Project (OMAP) (www.OMAP.org) bacterial artificial chromosome (BAC) resource did for the rice community, we constructed BAC libraries for maize inbred lines Zheng58, Chang7-2, and Mo17 and maize wild relatives Zea mays ssp. parviglumis and Tripsacum dactyloides. Furthermore, to extend functional genomic studies to maize and sorghum, we also constructed binary BAC (BIBAC) libraries for the maize inbred B73 and the sorghum landrace Nengsi-1. The BAC/BIBAC vectors facilitate transfer of large intact DNA inserts from BAC clones to the BIBAC vector and functional complementation of large DNA fragments. These seven Zea Map Alignment Project (ZMAP) BAC/BIBAC libraries have average insert sizes ranging from 92 to 148 kb, organellar DNA from 0.17 to 2.3%, empty vector rates between 0.35 and 5.56%, and genome equivalents of 4.7- to 8.4-fold. The usefulness of the Parviglumis and Tripsacum BAC libraries was demonstrated by mapping clones to the reference genome. Novel genes and alleles present in these ZMAP libraries can now be used for functional complementation studies and positional or homology-based cloning of genes for translational genomics.
MAIZE is one of the most important crops worldwide, already producing more tons of grain than any other plant species, and with continuing expansion of planted acreage in the developed and developing world. It is used as a staple food crop, but also for animal feed, biofuel, and various industrial by-products, such as starches, oil, sugars, sweeteners, beverages, and adhesives. It is also a key model species for research in genetics, epigenetics, genomics, development, physiology, and evolution (Bennetzen and Hake 2009; Schnable et al. 2009; Walbot 2009). Maize was domesticated ∼10,000 years ago from an annual wild teosinte, Zea mays ssp. parviglumis, that originated in the balsas river lowland in Mexico (Doebley et al. 2006; van Heerwaarden et al. 2011). One feature of contemporary maize is that it is probably the most genetically diverse of all crop species. Analyses of haplotypes; sequences around the bz, z1C1, and adh1 loci; and sequences of large genomic regions all showed striking structural diversity in maize inbred lines (Fu and Dooner 2002; Song et al. 2002; Song and Messing 2003; Jung et al. 2004; Brunner et al. 2005; Wang and Dooner 2006; Gore et al. 2009; Chia et al. 2012). It also has been shown that the two ancestral progenitors of maize and the progenitor of sorghum split ∼11.9 MYA and the progenitors of maize hybridized ∼4.8 MYA to form allotetraploid maize (Swigonova et al. 2004). Hybridization resulted in broken and fused chromosomes with the reduced set of today’s 10 chromosomes (Wei et al. 2007). Divergence of the two homologous chromosomal regions continued through loss of duplicated genes and uneven expansion through retrotransposition (Messing et al. 2004; Bruggmann et al. 2006; Woodhouse et al. 2010). A very recent study has indicated the nature of the major rearrangements that reduced ancestral Zea chromosome numbers from 20 to 10, including which 10 centromeres were retained and which 10 were lost (Wang and Bennetzen 2012).
Bacterial artificial chromosome (BAC) libraries are important tools for structural, functional, and comparative genomics studies of eukaryotic genomes. Hundreds of BAC libraries for crop plants have been constructed to date, such as those for rice and wild rice (Wang et al. 1995; Ammiraju et al. 2006), wheat (Cenci et al. 2003; Nilmalgoda et al. 2003; Janda et al. 2006), maize (O’Sullivan et al. 2001; Tomkins et al. 2002; Yim et al. 2002), soybean (Wu et al. 2004), sorghum (Woo et al. 1994), and pearl millet (Allouis et al. 2001). These BAC libraries have been used for developing sequence-ready physical maps (Chen et al. 2002; Wei et al. 2009), genome sequencing (International Rice Genome Sequencing Project 2005; Schnable et al. 2009), positional cloning of genes of agronomic importance (Song et al. 1995; Mao et al. 2000; Li et al. 2011), and comparative analysis of genome structures (Chen et al. 1997; Messing and Llaca 1998; Ammiraju et al. 2008; Lu et al. 2009; Wang and Dooner 2012). BAC end sequencing has also been used to survey genome structure (Messing et al. 2004), to assist in (and/or confirm) genome sequence assemblies (Bennetzen et al. 2012), and to develop SSR and SNP markers (Qi et al. 2001, 2004). BAC end sequences (BESs) embedded in physical maps have been used to detect structural differences between the target genomes and the reference sequence (Kim et al. 2007; Hurwitz et al. 2010; Lin et al. 2012; Deng et al. 2013). We have flagged many putative assembly inversion errors of the Nipponbare reference sequence by aligning physical maps of rice ZH11 and 93-11 to the rice Nipponbare reference sequence through BESs (Lin et al. 2012) (Pan et al., unpublished results) and confirmed five representatives by further detailed analysis (Deng et al. 2013).
A binary BAC (BIBAC) vector was developed from the BAC vector, and BIBAC libraries or the like [transformation-competent artificial chromosome (TAC)] can be used not only for general purposes as BAC libraries but also for Agrobacterium-mediated plant transformation of large genomic DNA fragments (Hamilton et al. 1996; Hamilton 1997; Liu et al. 1999). Many BIBAC libraries have been constructed to date, such as those for rice (Liu et al. 2002; Tao et al. 2002), wheat (Liu et al. 2000), soybean (Meksem et al. 2000), sunflower (Feng et al. 2006), cotton (Lee et al. 2013), and ginseng (Zhai et al. 2013). Recently, we modified a pair of BAC and BIBAC vectors coordinately to make the construction of BAC/BIBAC libraries easier and to allow simple large intact DNA insert exchange between BAC and BIBAC vectors (Shi et al. 2011).
Three BAC libraries containing genomic DNA from maize inbred B73 have been constructed, each containing DNA digested with a different restriction enzyme (Yim et al. 2002), and they have been fingerprinted by two complementary methods (Nelson et al. 2005). A large number of these clones have also been end sequenced (Messing et al. 2004). A BAC physical map for B73 was constructed (Wei et al. 2009) and genetically anchored, allowing the generation of a BAC-based draft sequence of the maize B73 genome (Schnable et al. 2009). However, to functionally annotate the maize genome and make better use of the maize reference sequence in future studies, additional related genomic resources of a broader phylogenetic diversity are needed. As an example, BAC libraries for 12 species of genus Oryza (one cultivar and 11 wild species) achieved a better understanding of the diversity of the genus and serve as a resource for new traits in cultivars (Wing et al. 2005; Ammiraju et al. 2006). This project is also known as the Oryza Map Alignment Project (OMAP) (http://www.omap.org). Here, we report the construction and quality analysis of BAC libraries for three elite inbred lines (Zheng58, Chang7-2, and Mo17) and two wild relatives of maize (Z. mays ssp. parviglumis and Tripsacum dactyloides) and BIBAC libraries for maize B73 and sorghum Nengsi-1. To ascertain that the BAC/BIBAC resources are optimal, the inbreds and accessions were selected after input from a group of maize geneticists and evolutionary biologists. Zheng58 and Chang7-2 are the respective male and female parents of Zhengdan958, the most widely planted hybrid maize in China. They are predicted to be rich in beneficial alleles and thus are appropriate subjects for gene cloning, heterosis, and genomic studies (Lai et al. 2010; Ding et al. 2011; Jiao et al. 2012). Maize inbred Mo17 represents the genetic background of many current elite inbreds in the U.S. and world seed industry. Mapping populations (B73 × Mo17) have been used to construct an integrated physical and genetic map of the maize genome and to clone maize QTL (Fu et al. 2006; Barbazuk and Schnable 2011; Eichten et al. 2011). Although a Mo17 BAC library has been constructed in the industry, it is not publicly available and does not have deep coverage (Brunner et al. 2005). Z. mays ssp. parviglumis is the most likely wild progenitor of maize and is a main subject for maize domestication and evolution studies (Doebley 2004; Swanson-Wagner et al. 2012). T. dactyloides with a large genome size of 3730 Mb (Arumuganathan and Earle 1991; Bennett and Leitch 2011) is a member of the Tribe Maydeae, closely related to maize, and is phylogenetically well defined and genetically tractable. It represents an untapped genetic resource for crop improvement and contains many useful agronomic traits like disease and pest resistance and harsh-environment adaptation (Hoisington et al. 1999; Chia et al. 2012). It is also a natural model system to study parthenogenesis, an important trait for crop improvement (Hoisington et al. 1999; Bantin et al. 2001). Tripsacum can hybridize with maize and the hybrids can grow well (Harlan and de Wet 1977). B73 has served as a maize genetics and genomics platform worldwide and is rich in genetics and genomics resources such as genetic maps, BAC libraries, a physical map, and a high-quality draft reference sequence (Yim et al. 2002; Fu et al. 2006; Schnable et al. 2009; Wei et al. 2009). However, it lacked a BIBAC library to facilitate functional genomics studies. Sorghum is also the staple food for people in semi-arid tropic areas, especially in Sub-Saharan Africa (Dicko et al. 2006). It also is superior as a biofuel crop over maize because of lower input cost (Calvino and Messing 2012). Both maize and sorghum are C4 plants with high photosynthetic efficiency. Partly because of its smaller genome size (730 Mb), sorghum became a model for research related to maize (Bennetzen and Freeling 1993), including C4 and drought tolerance research (Paterson et al. 2009; Sasaki and Antonio 2009). The high output of grain, biomass, and sugars from sorghum can provide new resources for biofuels (Byrt et al. 2011; Vermerris 2011), in particular from sweet sorghum (Calvino and Messing 2012). BAC libraries for sorghum are available (Woo et al. 1994; Lin et al. 1999). A high-quality draft genome sequence of sorghum bicolor BT × 623 has been obtained by a whole-genome shotgun sequencing approach (Paterson et al. 2009). Therefore, a BIBAC library for sorghum has great potential. The sorghum landrace Nengsi-1 used here is an important variety in China and has many elite features for both energy and fodder use. The wide selection of maize and related germplasms represented in these libraries will be critical for researchers interested in comparative and functional genomics and for cloning of agriculturally important genes because of the tremendous structural and allelic variation available in these BAC/BIBAC libraries. They should also be of interest to breeders for use in engineering quantitative traits. Furthermore, variation in both gene and transposable element content will provide a wealth of information to evolutionary biologists. This selection took into consideration the fact that BAC libraries were already available for outgroup species at intermediate evolutionary distances between maize and rice: Coix lacryma-jobi, also a member of the Tribe Maydeae (Meng et al. 2010); Sorghum, a member of the closely related Tribe Andropogonea (Woo et al. 1994; Lin et al. 1999); and two other members of the Tribe Paniceae, Pennisetum (pearl millet) (Allouis et al. 2001) and Setaria italica (foxtail millet) (Bennetzen et al. 2012).
The seven BAC/BIBAC libraries are now called the Zea Map Alignment Project (ZMAP) libraries, following the designation of Oryza BAC libraries as the OMAP collection (Wing et al. 2005; Ammiraju et al. 2006). These libraries all have high quality and comprise a total of 940,032 clones that have been stored in 384-well plates in −80° freezers in the Genome Resource Laboratory of Huazhong Agricultural University (http://GResource.hzau.edu.cn). Copies of the Parviglumis and Tripsacum BAC libraries have also been stored at Arizona Genomics Institute (http://www.genome.arizona.edu). Whole libraries, filters, and clones are publicly available through the above Web sites.
Materials and Methods
Plant materials
Seeds of Z. mays ssp. parviglumis (PI 384062, lot no. 83ncso01) and T. dactyloides (PI 595898, lot no. 96ncfo01) were obtained from the U.S. Department of Agriculture germplasm bank at Ames, Iowa (http://www.ars-grin.gov/), where they are reproduced from the originally released line, not modified by introgression of other segments. Seeds of maize inbreds Zheng58 and Chang7-2 are available and maintained at Beijing Agricultural University, currently by Jinsheng Lai. Seeds of maize Mo17 with an introgressed Rf3 gene were maintained by Yonglian Zheng. Seeds of maize inbred B73 were from the same source as that used for the generation of the B73 reference sequence (Tomkins et al. 2002; Wei et al. 2007) and were obtained from the U.S. Department of Agriculture germplasm bank at Ames, Iowa (see: http://maizegdb.org/sequencing_project.php). Seeds of sorghum landrace Nengsi-1 were provided by Ruiheng Du of Hebei Academy of Agriculture and Forestry Sciences, Shijiazhuang, China. Young leaves of Z. mays ssp. parviglumis, Tripsacum, Zheng58, Chang7-2, and Nengsi-1 and young ears of Mo17 and B73 were used as DNA sources for BAC/BIBAC library construction.
BAC/BIBAC vector preparation
The BAC and BIBAC vectors pIndigoBAC536-S and BIBAC-S were prepared with HindIII (Fermentas) and BamHI (Fermentas) from the high-copy composite vectors pHZAUBAC1 and pHZAUBIBAC1 (Shi et al. 2011), respectively. Vector plasmid DNA was isolated using the plasmid midkit (QIAGEN, Valencia, CA). The digested products were dephosphorylated with Calf Intestinal Alkaline Phosphatase (New England Biolabs, Beverly, MA), self-ligated overnight, and separated on a CHEF gel at 6 V/cm and 14° in 0.5× TBE buffer with a linear ramp time from 1 to 40 sec for 16.5 hr. The linearized and dephosphorylated pIndigoBAC536-S and BIBAC-S vectors were excised and electroeluted from the gel as previously described (Luo and Wing 2003).
BAC/BIBAC library construction
BAC libraries were constructed with HindIII, as previously described (Luo and Wing 2003; Ammiraju et al. 2006; Luo et al. 2006; Song et al. 2011). BIBAC libraries were constructed with BamHI, using the same procedure as for BAC library construction except that 20 μg/ml kanamycin was used for selection. DH10B T1-resistant cells (Invitrogen, Carlsbad, CA) were used for both BAC and BIBAC library construction. Individual BAC/BIBAC colonies were picked and the entire collection was stored in 384-well microtiter plates at −80°. For the B73 BIBAC library, clone pools were made for each 384-well plate to facilitate PCR screening of targeted clones.
Insert size analysis
BAC/BIBAC clones were randomly selected from each library for insert sizing. BAC/BIBAC plasmids were isolated as described by Luo and Wing (2003), digested with I-SceI, and separated on 1% agarose CHEF (CHEF-DRIII apparatus; Bio-Rad, Hercules, CA) gels at 6 V/cm and 14° in 0.5 × TBE buffer with a linear ramp time from 5 to 15 sec for 16 hr.
Stability analysis of sorghum BIBAC clones in Escherichia coli and Agrobacterium
Stability tests of sorghum BIBAC clones were conducted following the previously described method for testing maize B73 BIBAC clones (Shi et al. 2011). To test the stability of sorghum BIBAC clones in E. coli, BIBAC clones maintained in the DH10B T1-resistant host cells were cultured for different times and subjected to insert size analysis, as described above. To test the stability of sorghum BIBAC clones in Agrobacterium, we used an indirect method because preparing enough large plasmid DNA directly from Agrobacterium is difficult. The BIBAC plasmid DNA was transferred into Agrobacterium strain EHA105, extracted after being cultured for different times, and transferred back into E. coli DH10B T1-resistant cells. Then the plasmid DNA was reextracted from the E. coli culture after overnight growth and subjected to insert size analysis.
Probe preparation
Twelve single-copy gene probes (Supporting Information, Table S1) were selected according to the maize B73 reference sequence via MaizeGDB (http://www.maizegdb.org/). They represented all 10 chromosomes of B73: 1 from each chromosome except 2 from chr1 and 2 from chr3. PCR primers were designed using Primer5. Probes adh1, adh2, and ps1 were generated by RT-PCR amplification from B73 mRNA. The other gene probes were amplified by PCR, using B73 genomic DNA as the template. All the PCR products were purified from an agarose gel with the MinElute Gel Extraction Kit (QIAGEN). Three barley chloroplast probes (ndhA, rbcL, and psbA) (obtained from J. Mullet, Texas A&M University, College Station, TX) were used to estimate the content of chloroplast DNA and 4 rice mitochondrial probes (atpA, cob, atp9, and coxE) (obtained from T. Sasaki, Institute of Society for Techno-innovation of Agriculture, Forestry and Fisheries, Tsukuba, Ibaraki, Japan) were used to estimate the content of mitochondrial DNA, as previously described (Luo et al. 2001; Ammiraju et al. 2006).
BAC/BIBAC library screening
The 12 maize gene probes were used to screen the Parviglumis and Tripsacum BAC libraries, whereas the chloroplast and mitochondrial probes were used to screen a subset of each of all seven BAC/BIBAC libraries. High-density colony filters for the Parviglumis and Tripsacum BAC libraries were prepared with Nybond-N+ nylon membranes (GE Healthcare), using a Q-bot (Genetix) following previously described methods (Luo et al. 2001; Ammiraju et al. 2006; Luo et al. 2006). Each 22.5 × 22.5-cm filter holds 18,432 independent clones from 48 × 384-well plates arrayed in a 4 × 4 double-spotted pattern in six fields. Colony filters with half of the above densities for the other five BAC/BIBAC libraries were prepared with Nybond-N+ nylon membranes, using a semiautomatic machine (Beckman Coulter Biomek 2000). The filters were prehybridized in prewarmed hybridization solution (2× SSC) at 55° for 3 hr in a hybridization oven at 30 rpm. The probes were labeled with [α−32P]dCTP (Amersham Biosciences, Piscataway, NJ) using a DecaprimeII random prime labeling kit (Ambion), denatured at 95° for 5 min, and added to the hybridization solution for hybridization at 55° overnight. For the 2 maize probes adh1 (probe 11) and adh2 (probe 12), single probes were used for each hybridization experiment. For the other 10 maize gene probes (probes 1–10), pools of 4 probes were used for each hybridization experiment (Table S2). For the 3 chloroplast probes and the 4 mitochondrial probes, probe pools were also used. After hybridization, the filters were processed by washing twice with 1.5× SSC, 0.1% SDS at 55° for 10 min each in the hybridization oven at 30 rpm; then once with 1.5× SSC, 0.1% SDS; and then once with 1× SSC, 0.1% SDS at 55° for 10 min each in a shaker at 100 rpm. The filters were tested for signal strength with a Geiger counter to determine whether more washes were needed. After washing, the filters were dried, marked by dotting eight positions with aliquots of hybridization solution containing labeled probes, and exposed to phosphorimager screens at room temperature overnight. The screens were scanned with the Typhoon scanner (GE Healthcare). All putative positive clones of the maize gene probes were located, rearrayed onto colony filters, and rehybridized with individual probes labeled with digoxin for confirmation. Probe labeling with digoxin and colony hybridization with the digoxin-labeled probes followed the commercial protocol in the DIG High Prime DNA Labeling and Detection Smart Kit I (Roche).
BAC/BIBAC end sequencing, fingerprinting, and contig assembly
Selected BAC and BIBAC clones were sequenced at both ends with an ABI 3730 DNA Analyzer, using BigDye v3.1 (Applied Biosystems, Foster City, CA) as previously described (Wang et al. 2013). The two primers BACf (5′ aacgacggccagtgaattg 3′) and BACr (5′ gataacaatttcacacagg 3′), common to all BAC/BIBAC libraries constructed with our BAC/BIBAC vectors (Shi et al. 2011), were used as forward and reverse sequencing primers, respectively. Sequences were base-called using Phred software (Ewing and Green 1998; Ewing et al. 1998). The vector and low-quality sequences (Phred value <16) were trimmed using the Applied Biosystems Sequence Scanner v1.0 Software. Reads <50 bp in length were removed. BAC/BIBAC clones were fingerprinted using the SNaPshot method, as previously described (Luo et al. 2003; Kim et al. 2007; Wang et al. 2013), and assembled into contigs with FPC v9.4 at a fixed tolerance value of 7 and a cutoff score of 1e−10. Those containing two or more overlapping clones were counted as contigs.
Anchoring BAC/BIBAC clones to the reference sequences through BESs
Repeat Masker version 3.3.0 (http://www.repeatmasker.org) was used to mask end sequences of the maize B73 and sorghum Nengsi-1 BIBAC clones with the maize and sorghum databases, respectively. The end sequences of the Parviglumis and Tripsacum BAC clones (unmasked) and the maize B73 BIBAC clones (masked and unmasked) were aligned to the maize B73 genome sequence, and the end sequences of the sorghum Nengsi-1 (masked and unmasked) were aligned to the sorghum genome sequence with MegaBLAST, using the ncbi-BLAST 2.27+ (Camacho et al. 2009). The parameters of the program were set to –evalue 1e-100 –outfmt 6 for Parviglumis, Tripsacum, and sorghum Nengsi-1 and to –evalue 1e-150 –outfmt 6 for maize B73. The putative mapped sites of the BAC/BIBAC clones on the reference genome sequences were determined using Zhao's (Zhao et al. 2004) and Dalrymple's (Dalrymple et al. 2007) methods with the criteria specified.
Anchoring B73 BIBAC clones to the B73 reference sequence through fingerprints
The putative mapped regions of some representative B73 BIBAC clones on the B73 reference sequence were cut in silico with the restriction enzymes BamHI, EcoRI, XbaI, XhoI, and HaeIII that were used in the SNaPshot method for BAC/BIBAC fingerprinting. The real fingerprints of the B73 BIBAC clones and the in silico fingerprints of the putative mapped regions on the B73 reference sequence were compared by FPCv9.4 (Soderlund et al. 1997). We calculated background (GC) and significant quality (SQ) to measure their relationships using the algorithms described below.
Background coincidence score (GC):
GC reflects the coincidence score of a clone’s real fingerprint with any random in silico fingerprint:

In the formula, N is the number of clones involved, so we have a cohort of clones.
is the minimum coincidence score of the real fingerprints of the clone i with the in silico fingerprints corresponding to other clones.
Significant quality (SQ):
SQ is a quality reflecting the difference between the best coincidence score () and the background coincidence score:

In this formula, is the minimum coincidence score of the real fingerprints of the clone i with the in silico fingerprints corresponding to this clone. The secondary coincidence score (
) is the secondary minimum coincidence score of the real fingerprints of the clone i with the in silico fingerprints corresponding to this clone. If a clone was anchored to a single site on the reference sequence (only in silico fingerprints of a single region on the reference are available), then one takes
instead of
.
Results
BAC library construction and characterization
We constructed BAC libraries for three elite inbreds of maize (Zheng58, Chang7-2, and Mo17); one maize progenitor, Z. mays ssp. parviglumis; and one species from the closest genus to Zea within the tribe maydeae, T. dactyloides. All five BAC libraries were constructed with genomic DNA fragments prepared by HindIII partial digestion and two size selections and in the pIndigoBAC536-S vector prepared from pHZAUBAC1 (Shi et al. 2011). BAC clones were arrayed into 384-well microtiter plates and stored in −80° freezers. The numbers of BAC clones for Z. mays ssp. parviglumis, Tripsacum, Zheng58, Chang7-2, and Mo17 are 100,608, 128,256, 148,992, 145,920, and 172,416, respectively (Table 1). Each BAC library was constructed from more than one ligation, having different insert sizes and transformation efficiencies.
The average insert sizes and empty vector rates were estimated by digesting randomly selected 238–412 clones from each library with I-SceI and separating the digested clones on 1% agarose CHEF gels. Figure 1 shows the DNA analysis results of 40 randomly picked clones from the Parviglumis BAC library. Because the vector was engineered to contain two 18-bp I-SceI recognition sites flanking the cloning site and I-SceI sites are rare in genomic DNA, all BAC clones released a single genomic DNA insert. Insert sizes ranged from 20 to 257 kb, with the majority >115 kb. The average insert sizes of these BAC libraries ranged between 122 and 148 kb. The rates of empty vector clones ranged from 0.72 to 5.56% (Table 1). The insert size distributions of each BAC library are shown in Figure 2. Some libraries do not show typical Poisson patterns, probably due to the use of different ligations. To determine the organellar DNA content, BAC libraries were screened with three chloroplast probes and four mitochondrial probes. Chloroplast DNA content of the BAC libraries ranged between 0.16 and 0.50% and mitochondrial DNA content ranged from 0 to 0.10%. However, it has been reported that maize contains numerous dynamic nuclear–mitochondrial sequences (NUMTs) and nuclear–plastid sequences (NUPTs) (Lough et al. 2008; Roark et al. 2010). So screening out the mitochondrial and plastid sequences will also eliminate the NUMTs and NUPTs. The calculated genome coverages of the BAC libraries for Z. mays ssp. parviglumis, Tripsacum, Zheng58, Chang7-2, and Mo17 are 5.5, 4.7, 8.4, 8.0, and 7.9, respectively (Table 1).
DNA analysis of random BAC clones from the Parviglumis BAC library by pulsed-field gel electrophoresis. DNA samples were digested with I-SceI and separated on a 1% agarose CHEF gel. The 6.38-kb common band is the vector pIndigoBac536-S I-SceI fragment. Markers used are MidRange PFG marker I (New England Biolabs).
Insert size distributions of the seven ZMAP BAC and BIBAC libraries. In each histogram, the x-axis represents insert size range (kilobases) and the y-axis represents the number of clones within a particular insert size range. (A) Parviglumis BAC library: the average insert size is 148 kb based on 387 clones. (B) Tripsacum BAC library: the average insert size is 139 kb based on 412 clones. (C) Zheng58 BAC library: the average insert size is144 kb based on 348 clones. (D) Chang7-2 BAC library: the average insert size is 139 kb based on 349 clones. (E) Mo17 BAC library: the average insert size is 122 kb based on 238 clones. (F) B73 BIBAC library: the average insert size is 92 kb based on 282 clones. (G) Sorghum BIBAC library: the average insert size is 111 kb based on 256 clones.
BIBAC library construction and characterization
We also constructed two BIBAC libraries, one for maize B73 and one for sorghum Nengsi-1. Both BIBAC libraries were constructed with genomic DNA fragments prepared by BamHI partial digestion and two size selections into the BIBAC-S vector prepared from pHZAUBIBAC1 (Shi et al. 2011). Multiple ligations were also used for the BIBAC libraries. As for BAC clones, BIBAC clones were arrayed into 384-well microtiter plates and stored in −80° freezers. Clone pools per each 384-well microtiter plate of the maize B73 BIBAC library were made for PCR screening. The numbers of BIBAC clones for maize B73 and sorghum Nengsi-1 are 192,000 and 51,840, respectively (Table 1). DNA samples of 282 and 256 clones, randomly selected from the maize B73 and sorghum Nengsi-1 BIBAC libraries, respectively, were analyzed with I-SceI. All BIBAC clones released an intact genomic DNA insert band (Figure 3). Insert sizes ranged from 30 to 190 kb. The average insert size, empty vector rate, and chloroplast and mitochondrial DNA content were 92 kb, 0.35%, 0.15%, and 0.09%, respectively, for the maize B73 BIBAC library and 111 kb, 2.29%, 2.20%, and 0.10%, respectively, for the sorghum Nengsi-1 BIBAC library (Table 1). The insert size distribution of each BIBAC library is shown in Figure 2. We did not use very large genomic DNA fragment fractions for BIBAC library construction on purpose because previously it was reported that large BIBACs were not stable in Agrobacterium (Song et al. 2003). The calculated genome coverages of the maize B73 and sorghum Nengsi-1 BIBAC libraries are 7.6 and 7.3, respectively (Table 1).
DNA analysis of random BIBAC clones from the sorghum Nengsi-1 BIBAC library by pulsed-field gel electrophoresis. DNA samples were digested with I-SceI and separated on a 1% agarose CHEF gel. The 23.2-kb common band is the vector BIBAC-S I-SceI fragment. Markers used are MidRange PFG marker I (New England Biolabs).
Utility demonstration for the Parviglumis and Tripsacum BAC libraries
To demonstrate the potential usefulness of the Parviglumis and Tripsacum BAC libraries, we chose 12 single-copy genes as probes from the 10 maize chromosomes according to the B73 reference sequence and screened the Parviglumis and Tripsacum BAC libraries. First, we screened the two libraries with 32P-labeled probes and then rehybridized the clones collected from the first round of screening with digoxin- labeled probes. A total of 259 clones hybridized in the two libraries with the 12 32P-labeled maize probes and the identities of 205 (79%) clones were confirmed with the 12 digoxin-labeled probes (Table 2). All 12 32P-labeled maize gene probes hybridized to clones in both Parviglumis and Tripsacum BAC libraries, and 96 of the 111 Parviglumis clones (86%) and 109 of the 148 Tripsacum clones (74%) were confirmed with digoxin-labeled probes. The average confirmed positive clones per probe were 8.0 and 9.1 for Parviglumis and Tripsacum BAC libraries, respectively, even higher than predicted by their respective 5.5 and 4.7 genome equivalencies. The 54 clones that hybridized with 32P-labeled probes but not with digoxin-labeled probes could be false positive clones or, more likely, missed positive clones due to the weak hybridization signals of the digoxin-labeled heterologous probes and inefficient digoxin labeling of the probes ba1 and ps1 in Tripsacum colony hybridization (see below).
We SNaPshot fingerprinted and sequenced both ends of all 259 clones from the first-round screening. A total of 238 (92%) of the BAC clones resulted in qualified fingerprints and 149 BAC clones were assembled into contigs. Of the 149 BAC clones, 130 were from the 205 confirmed positive BAC clones and 19 were from the 54 BAC clones that were not confirmed in the second hybridization. These 19 BAC clones were considered as missed positive clones by the second hybridization probably due to the reasons mentioned above. Together, 64 Parviglumis BAC clones and 85 Tripsacum BAC clones were assembled into 20 Parviglumis BAC contigs and 27 Tripsacum BAC contigs, respectively (Table 3). The average clones per contig were 3.1 and 3.2 for Parviglumis and Tripsacum BAC libraries, respectively. Positive clones for some maize probes remained singletons (e.g., those of both BAC libraries for vp1 and those of the Parviglumis BAC library for adh2) or formed more than one contig (e.g., those of both BAC libraries for ps1, why1, gst1, and hrg1). This is probably due to the technical limitations of contig assembly, heterozygosity, and/or, most likely according to the much higher positive rates than predicted, duplication of these genes in these genomes relative to maize (see discussion below).
We used the fingerprints of all 238 Parviglumis and Tripsacum BAC clones together for contig assembly and found that the clones that were assembled into each contig were from the same library except that one gst1 clone from the Parviglumis BAC library was misassembled into the Tripsacum adh1 contig, which contains five Tripsacum clones. This result indicated that a good level of diversity exists between Z. mays ssp. parviglumis and Tripsacum at all of the homologous regions to the 12 maize genes.
We anchored the Parviglumis and Tripsacum positive clones and contigs to the maize B73 high-quality draft reference sequence through BESs (Table 4 and Table S3). From the 224 positive BAC clones (205 + 19), we obtained a total of 431 qualified BESs (98% success). The 103 Parviglumis positive BAC clones generated 202 BESs with an average read length of 654 bp and the 121 Tripsacum positive BAC clones generated 229 BESs with an average read length of 599 bp. For Parviglumis BAC clones, 102 of the 103 positive BAC clones possess BESs with hits to the B73 reference sequence, 94 with paired BES hits and 8 with single-end BES hits. Of the 202 BESs, 196 hit to the B73 reference sequence: 36 hit single sites and 160 hit multiple sites. Using criteria that the aligned paired BESs should locate vis-à-vis to the same chromosome with a distance between 35 and 500 kb, we could anchor 70 clones to the B73 reference sequence, 60 to multiple sites and 10 to single sites. Thirty-four clones were anchored to the expected sites for the maize gene probes. Of the 34 clones, 18 clones were from 9 contigs and 16 clones were singletons (Table 4 and Table S3). For Tripsacum BAC clones, 104 of 121 positive BAC clones possessed BESs with hits to the B73 reference sequence: 46 with paired BES hits and 58 with single-end BES hits. Of the 229 BESs, 150 hit the B73 reference sequence; 54 hit single sites, and 96 hit multiple sites. Twenty-two clones could be anchored to the B73 reference sequence, 15 to multiple and 7 to single sites. Seven clones, 5 from 3 contigs and 2 singletons, were anchored to the expected sites for the maize gene probes (Table 4 and Table S3). The placement of Parviglumis and Tripsacum BAC clones and contigs to the expected maize B73 sites indicated that large regions surrounding the corresponding sites were generally colinear between Z. mays ssp. parviglumis and maize and between Tripsacum and maize. Fewer BESs, clones, and contigs of Tripsacum were aligned to the maize B73 reference sequence than those of Z. mays ssp. parviglumis, consistent with the phylogenetic distance of Tripsacum to maize.
Stability of sorghum Nengsi-1 BIBAC clones in E. coli and Agrobacterium
We previously demonstrated that maize B73 BIBAC clones of different sizes from small to 160 kb were all stable in both E. coli and Agrobacterium (Shi et al. 2011). In this study, we tested the stability of sorghum Nengsi-1 BIBAC clones and found that the BIBAC clones of different sizes were all stable in E. coli (data not shown). When 10 sorghum BIBAC clones with different insert sizes from 58 to 160 kb were tested for their stability in Agrobacterium EHA105 via an indirect method (Shi et al. 2011), 9 were found to be stable after 48 and 96 hr of culture (subcultured after 48 hr). Figure 4 shows the results for 8 clones. The 48- and 96-hr culture samples of the Agrobacterium EHA105 colonies showed the same I-SceI digestion patterns as the original E. coli clones. Only one clone with an insert size of 131 kb was not stable. Three types of smaller insert fragments were found in both the 48- and 96-hr culture samples of this clone (data not shown).
Stability of sorghum Nengsi-1 BIBAC clones in Agrobacterium. BIBAC clones with different insert sizes from 58 to 160 kb were selected randomly from the sorghum Nengsi-1 BIBAC library. Plasmid DNA of each BIBAC clone was transferred into Agrobacterium EHA105 cells. Single Agrobacterium EHA105 transformed colonies were cultured at 28° for 48 and 96 hr (subcultured after 48 hr). BIBAC DNA of each culture was prepared via an indirect method (Shi et al. 2011) and analyzed with I-SceI. The molecular weight marker was the MidRange PFG marker I (NEB).
Mapping maize B73 and sorghum Nengsi-1 BIBAC clones to reference sequences
High-quality draft reference sequences for both maize and sorghum are available. Especially, our maize B73 BIBAC library was constructed using the same source of seeds as that used for B73 sequencing. To assign sequence information to our BIBAC libraries, we tried to map random maize B73 and sorghum Nengsi-1 BIBACs to the maize B73 and sorghum reference sequences, through BESs (Table 5), respectively.
We sequenced 1152 random BIBACs at both ends for the maize B73 BIBAC library and obtained a total of 2235 qualified sequences (97.0% success) with an average read length of 679 bp. A total of 1087 clones generated paired BESs and 61 clones generated single-end BESs. Of the 2235 BESs, 2064 BESs (92.3%) were aligned to the maize B73 reference sequence; 371 (18%) hit single sites and 1693 (82%) hit multiple sites. After repeat sequences were masked, 491 BESs were aligned to the maize B73 reference sequence; 293 hit single sites and 198 hit multiple sites. Using criteria that the aligned paired BESs should locate vis-à-vis to the same chromosome with a distance between 35 and 180 kb, we could anchor 852 clones (74.2% of the clones with successful BESs) to the B73 reference sequence, 625 (73.3%) to multiple sites and 227 (26.6%) to single sites.
We also sequenced 144 random BIBACs at both ends for the sorghum Nengsi-1 BIBAC library and obtained a total of 251 qualified sequences (87.2% success) with an average read length of 562 bp; 118 clones generated paired BESs and 15 clones generated single-end BESs. Of the 251 BESs, 195 BESs (77.4%) were aligned to the sorghum reference sequence; 62 (31.8%) hit single sites and 133 (68.2%) hit multiple sites. After repeat sequences were masked, 167 BESs were aligned to the sorghum reference sequence; 52 hit single sites and 115 hit multiple sites. Using criteria that the aligned paired BESs should locate vis-à-vis to the same chromosome with a distance between 35 and 500 kb, the same as those used for anchoring of Parviglumis and Tripsacum BACs, we anchored 66 sorghum Nengsi-1 BIBACs (49.2% of the clones with successful BESs) to the sorghum reference sequence, 44 (66.7%) to multiple sites and 22 (33.3%) to single sites. The percentage of sorghum Nengsi-1 BIBAC clones anchored to the sorghum reference sequence (49.2%) was significantly lower than that of maize B73 BIBAC clones to the B73 reference sequence (74.2%) even though lower stringency was used for anchoring of the sorghum BIBACs. Because the sorghum Btx623 has been generated with a higher genome coverage than the maize B73 genome, paired reads of different insert libraries, and an average high-quality long read of 735 bp, it is unlikely that the difference is due to gaps in the genome assembly. Therefore, it appears that the sequence of sorghum Nengsi-1 has diverged from the sorghum Btx623 reference sequence more than expected.
To test whether the multisite anchored clones could be further refined and to evaluate single-site anchored clones, we fingerprinted 48 B73 BIBAC clones and obtained qualified fingerprints for 36 clones, 9 single-site anchored clones and 27 multisites anchored clones. Because the maize B73 physical map was constructed with BAC fingerprints using earlier methods and different restriction enzymes (Wei et al. 2007, 2009) than the SNaPshot method (Luo et al. 2003), we could not merge our fingerprints into a common maize B73 BAC physical map. To circumvent this problem, we digested the target regions of the B73 reference sequence with the same five enzymes in silico that were used in the SNaPshot method and compared our B73 BIBAC fingerprints with the in silico fingerprints, using the software FPCv9.4 at the cutoff value of 1e-12 and tolerance value of 3 bp. SQ is defined as a value that reflects the confidence level of the match between the BIBAC fingerprints and in silico fingerprints under the setting cutoff value of 1e-12; the algorithm of SQ is described in Materials and Methods. If SQ > 2, it indicates that the result is significant. If SQ < 2, the result would not reach a significant threshold. When SQ is a negative number, the results are not usable.
Indeed, eight of the nine fingerprints from single-site anchored clones did match in silico fingerprints of the target regions of the B73 reference sequence under the FPC cutoff value. The average SQ of these eight clones is 5.084, and the FPC coincidence score ranges from 4.00e-22 to 6.47e-17. This result indicates that the single-site anchored clones determined by paired BESs have a high confidence level.
Of the 27 multisite anchored clones, fingerprints of each of 10 clones (37%) matched to in silico fingerprints of one site with SQ value >2. Hence, the 10 clones that were anchored to multiple sites of reference sequence with BESs could be further resolved into single sites of reference sequence with fingerprints at high confidence. For the 17 remaining multisite anchored clones, fingerprints from 10 clones (37% of the multisite anchored clones analyzed) matched to in silico fingerprints with SQ values between 0 and 2, whereas 2 clones (7%) had negative SQs and 5 clones (19%) did not match under the FPC cutoff value at all.
Discussion
Construction of the ZMAP BAC/BIBAC libraries
Large genome comparative projects, such as for Drosophila (Song et al. 2011) and for Oryza (Wing et al. 2005; Ammiraju et al. 2006) (www.OMAP.org), have served as foundations for evolutionary and functional genomics studies. For a similar purpose, we constructed BAC libraries for the three inbred lines Zheng58, Chang7-2, and Mo17 and the wild species Z. mays ssp. parviglumis and T. dactyloides, in addition to BIBAC libraries for maize inbred B73 and sorghum landrace Nengsi-1. The seven BAC/BIBAC libraries were called ZMAP BAC/BIBAC libraries following the OMAP nomenclature. The BAC libraries for Z. mays ssp. parviglumis, Tripsacum, Zheng58, Chang7-2, and Mo17 and the BIBAC libraries for B73 and Nengsi-1 have average insert sizes of 148, 139, 144, 139, 122, 92, and 111 kb; average organellar DNA of 0.31, 0.26, 0.55, 0.58, 0.17, 0.24, and 2.3%; empty vector rates of 0.77, 0.72, 1.14, 0.85, 5.56, 0.35, and 2.29%; and genome equivalents of 5.5, 4.7, 8.4, 8.0, 7.9, 7.6, and 7.3, respectively (Table 1). The BAC libraries for Zheng58, Chang7-2, and Mo17 are the first public BAC resources for these three elite maize inbreds; the Parviglumis and Tripsacum BAC libraries are the first BAC libraries for close wild relatives of maize; and the two BIBAC libraries are the first BIBAC resources for maize and sorghum. These new BAC/BIBAC libraries, together with the already available BAC libraries and genome map resources of other maize inbred lines, sorghum, pearl millet, Setaria and rice, provide a wide range of trait diversity and evolutionary distances. These powerful new resources will complement use of the maize reference sequence.
Coverage estimations
Genome coverage for a BAC/BIBAC library is usually estimated mathematically from the average insert size, total number of clones, and size of the corresponding genome (Luo et al. 2001; Wu et al. 2004; Ammiraju et al. 2006; Song et al. 2011; Lee et al. 2013). Hybridization and contig analysis were also used as complementary approaches to estimate genome coverage (Ammiraju et al. 2006; Song et al. 2011). When the probe sequences were from the same genome for which the BAC/BIBAC library was constructed or from a close genome, the average hybridization results were usually close to the mathematic results (Luo et al. 2001; Lee et al. 2013). However, compared to mathematic estimation, hybridization and contig-based estimation can be affected by many uncounted factors. In hybridization-based estimation, technically, inefficient probe labeling, improper hybridization conditions, filter qualities, etc., could all result in underestimation (Ammiraju et al. 2006), and at the sequence level, sequence diversity of the heterologous probes could also result in underestimation. In contig-based estimation, in addition to factors affecting the hybridization-based estimation, an unknown multicopy number and/or heterozygosity of a locus could also affect the outcome. Furthermore, low-quality fingerprints, too many or too few fingerprint bands due to nested retrotransposons, and uneven distribution of restriction sites of the enzymes for fingerprinting or not enough overlap in BAC clones also affect contig assemblies. We screened the Parviglumis and Tripsacum BAC libraries with 12 maize “single-copy” gene probes and obtained an average of 8.0 and 9.1 confirmed positive clones per probe, respectively. Still, some clones may have remained undetected as reported in our previous work (Ammiraju et al. 2006). The average hybridization-based coverage of 12 maize probes of the BAC libraries for Z. mays ssp. parviglumis and Tripsacum (8.0 and 9.1) is higher than the mathematically estimated coverage (5.5 and 4.7). Copy number variation in gene families corresponding to some maize probes could exist in Z. mays ssp. parviglumis and Tripsacum. For instance, ps1- and du1-homologous clones were overrepresented two- to threefold in the Parviglumis library, whereas tb1 was fivefold overrepresented in Tripsacum BACs. This suggested that these genes contain lineage or species-specific amplifications, a fact that was confirmed by the observation of six homologous contigs possibly containing five to six different tb1 alleles in Tripsacum.
We fingerprinted and end-sequenced the 259 BAC clones collected from the first round of screening and could assemble fingerprints of only 149 clones into a total of 47 contigs: 64 Parviglumis BAC clones and 85 Tripsacum BAC clones into 20 and 27 contigs, with average number of clones per contig totaling 3.1 and 3.2, respectively. The average clones per contig, if used, could underestimate the coverage due to the missed clones in hybridization screening and failed assembly of some homologous contigs or, especially, singletons. For example, of the 34 Parviglumis BAC clones that were anchored to the expected sites of the B73 reference sequence through BESs, 16 were singletons; of the 7 Tripsacum BAC clones that were anchored to the expected sites of the B73 reference sequence through BESs, 2 were singletons. Some homologous contigs and singletons were anchored to the same sites of the B73 reference sequence through BESs (Table 4). These contigs and singletons potentially could be manually merged, but more sequence information is required.
Utilization of the ZMAP BAC/BIBAC resources
The ZMAP BAC/BIBAC libraries provide a platform for maize research in several areas. They are particularly useful for positional or homologous cloning of genes or alleles desired for maize or even cereal crop improvement, not only because they were derived from the elite lines and untapped maize relatives that stacked or maintained many beneficial genes or alleles, but also because they were all cloned in our new BAC/BIBAC vectors (Shi et al. 2011). Our BIBAC vector is stable in Agrobacterium. Interesting BIBAC clones can be directly used for complementary experiments. For interesting BAC clones, the intact large inserts can be easily exchanged to the BIBAC vector. Our BAC and BIBAC vectors were both engineered to have two I-SceI sites flanking the cloning site in an inverted orientation and therefore add two important features. First, the I-SceI site is an 18-bp sequence, and its presence in any organism will be rare (e.g., only two sites were found in the maize B73 reference sequence). Therefore, intact large inserts can be released from the BAC and BIBAC clones. Second, the I-SceI site is nonpalindromic. The same noncomplementary 3′ protruding ATAA ends of the BAC and BIBAC vectors produced by I-SceI digestion suppress the self- and interligations. Because the BAC and BIBAC vectors contain chloramphenicol and kanamycin resistance genes for selection, respectively, intact large BAC inserts can be directly recloned into the BIBAC vector by digesting BAC clones with I-SceI, adding the BIBAC vector prepared with I-SceI to the I-SceI-digested BAC products without the purification of large insert DNA, ligating, and transforming and selecting the transformants on kanamycin-containing medium (Shi et al. 2011).
Comparative genome analysis among closely related species is a powerful approach to explore genome structure and function (Lin et al. 2012). However, to do this, it is necessary to have high-quality whole-genome or target region sequences of the investigated species.
BAC sequencing has been extensively used in comparative studies (Chen et al. 1997; Messing and Llaca 1998; Tikhonov et al. 1999; Bennetzen and Ma 2003; Ilic et al. 2003; Clark et al. 2004; Lai et al. 2004; Swigonova et al. 2004; Ammiraju et al. 2008; Lu et al. 2009; Wang and Dooner 2012). The ZMAP BAC/BIBAC libraries provide an important new set of sequencing templates for such studies in the Zea lineage. To see whether the orthologous loci of maize can be identified in Z. mays ssp. parviglumis and Tripsacum, we screened the Parviglumis and Tripsacum BAC libraries with 12 maize B73 gene probes. Positive clones were obtained for every probe. Many clones can be anchored to the expected sites for the probes on the B73 reference sequence by BESs. These results demonstrated the high utility of the two maize relative libraries.
The ZMAP BAC/BIBAC libraries provide essential tools for construction of physical maps. BAC physical map-based genome sequencing and whole-genome shotgun (WGS) sequencing are the two main approaches to obtain genome sequences. Lewin et al. (2009) reported that a good physical map is needed to obtain a high-quality genome sequence. We evaluated the two rice reference sequences, the Nipponbare reference sequence obtained with the physical map-based approach (Irgsp 2005) and the 93-11 reference sequence obtained with the WGS approach (Yu et al. 2005), and found that the Nipponbare reference sequence has high quality whereas the 93-11 reference sequence, although it also used long Sanger reads as did the Nipponbare reference sequence, has a much lower quality (Pan et al., unpublished results). Many large genomes were sequenced with the short next-generation sequencing (NGS) reads. However, without physical maps as frames for assembly, these genome sequences will stay fragmented. We recently developed a new method for simultaneous physical mapping and sequencing of complex genomes using the same set of NGS reads of BAC/BIBAC pools that can largely reduce the cost (Pan et al., unpublished results), making the physical map-based genome sequencing approach and BAC/BIBAC resources more attractive.
Genetic transformation of plants with large insert BIBAC/TAC clones is a desirable approach in gene functional analysis and molecular breeding because large insert BIBAC/TAC clones can contain complete large genes with regulatory sequences, gene clusters, or multiple genes (Shi et al. 2011; Lee et al. 2013; Zhai et al. 2013). Although the biolistic bombardment method has been successfully used to deliver 45 kb of a sorghum BAC comprising a cluster of kafirin genes into the maize genome that were tissue specifically expressed (Song et al. 2004) and 83 kb of an Arabidopsis BIBAC into tobacco (Chang et al. 2011), the Agrobacterium-mediated transformation method is still considered the best method to deliver BIBACs/TACs to different plants (Hamilton et al. 1996; Liu et al. 1999, 2002; Qu et al. 2003; Shi et al. 2011; Lee et al. 2013), because of fewer instances of multiple (e.g., concatamerized)-vector uptake and fewer cases of vector/insert rearrangement. A large DNA fragment including a herbicide resistance marker gene, a 30-kb yeast genomic fragment, and a 35S-lox-cre recombination cassette was successfully transferred from a BIBAC clone into the maize Hi II line through Agrobacterium (Vega et al. 2008). However, for Agrobacterium-mediated transformation, the stable maintenance of BIBAC clones in Agrobacterium is critical. We showed that the maize B73 BIBAC clones were stable in Agrobacterium (Shi et al. 2011). In fact, we have successfully transferred many large maize DNA fragments from maize B73 BIBAC clones into rice through Agrobacterium-mediated transformation (our unpublished data). In this study, we also demonstrated that the sorghum Nengsi-1 BIBAC clones were essentially stable in Agrobacterium. For the 10 sorghum Nengsi-1 BIBAC clones tested, 9 were stable in Agrobacterium.
The maize B73 seeds that we used in the BIBAC library construction were from the same source as those used in B73 genome sequencing. The current version of the B73 reference sequence still contains many physical and sequence gaps and assembly errors (Wei et al. 2009). The B73 BIBAC library could potentially be used to improve the B73 reference sequence.
Acknowledgments
We thank the U.S. Department of Agriculture germplasm bank at Ames, Iowa for the seeds of B73, Zea mays ssp. parviglumis and Tripsacum dactyloides; Xianmin Diao of the Institute of Crop Science, Chinese Academy of Agricultural Sciences for the sorghum seeds and leaves. We also thank Fusheng Wei, Xiang Song, José L. Goicoechea, Yeisoo Yu, Jianwei Zhang, and other members of the Arizona Genomics Institute and Yonglong Pan and Xiaoming Wang of Huazhong Agricultural University for technical assistance. This work was supported by the National Basic Research Program (the “973” project, 2009CB118404), the National Programs for High Technology Research and Development (863 Project: 2012AA10A305), and the Chinese 111 Project B07041.
Footnotes
Communicating editor: J. A. Birchler
- Received July 27, 2013.
- Accepted September 4, 2013.
- Copyright © 2013 by the Genetics Society of America