Hexaploid bread wheat (Triticum aestivum L. em. Thell) is one of the world's most important crop plants and displays a very low level of intraspecific polymorphism. We report the development of highly polymorphic microsatellite markers using procedures optimized for the large wheat genome. The isolation of microsatellite-containing clones from hypomethylated regions of the wheat genome increased the proportion of useful markers almost twofold. The majority (80%) of primer sets developed are genome-specific and detect only a single locus in one of the three genomes of bread wheat (A, B, or D). Only 20% of the markers detect more than one locus. A total of 279 loci amplified by 230 primer sets were placed onto a genetic framework map composed of RFLPs previously mapped in the reference population of the International Triticeae Mapping Initiative (ITMI) Opata 85 × W7984. Sixty-five microsatellites were mapped at a LOD >2.5, and 214 microsatellites were assigned to the most likely intervals. Ninety-three loci were mapped to the A genome, 115 to the B genome, and 71 to the D genome. The markers are randomly distributed along the linkage map, with clustering in several centromeric regions.
WHEAT (Triticum aestivum L. em. Thell.) is one of the most important food crops in the world, and understanding its genetics and genome organization using molecular markers is of great value for genetic and plant breeding purposes. It is an allohexaploid (2n = 6x = 42) with the three genomes A, B, and D and has an extremely large genome of 16 × 109 bp/1C (Bennett and Smith 1976) with more than 80% repetitive DNA. Detailed RFLP (restriction fragment length polymorphism) linkage maps (Chaoet al. 1989; Devos and Gale 1993; Xieet al. 1993; Nelson et al. 1995a,b,c; Van Deynzeet al. 1995; Marinoet al. 1996) and physical maps (Gillet al. 1993; Kotaet al. 1993; Hohmannet al. 1994; Ogiharaet al. 1994; Delaney et al. 1995a,b; Mickelson-Young et al. 1995; Gillet al. 1996) have been published for all seven homoeologous groups.
Although the progress in building wheat genetic maps has been steady, the use of RFLP markers in gene mapping has been slow because of the very limited level of polymorphism in wheat (Chaoet al. 1989; Kam-Morganet al. 1989; Liuet al. 1990; Cadalenet al. 1997). Because of this limited polymorphism, gene and genome mapping has required the use of populations derived from wide crosses. However, mapping many agronomically important genes or QTL (quantitative trait loci), amajor goal in plant breeding, requires informative markers in an intraspecific context. This is particularly true for marker-assisted selection. RFLPs detected with single-copy genomic and cDNA clones are extremely powerful for comparative mapping approaches (Ahnet al. 1993; Mooreet al. 1995; Shermanet al. 1995; Yuet al. 1996). They are only of limited use for intraspecific molecular analysis of agronomic traits, however, because usually <10% of all RFLP loci are polymorphic in wheat.
The genomes of all eukaryotes contain a class of sequences, termed microsatellites (Litt and Luty 1989) or simple sequenced repeats (SSRs) (Tautzet al. 1986). Microsatellites with tandem repeats of a basic motif of <6 bp have emerged as an important source of ubiquitous genetic markers for many eukaryotic genomes (Wanget al. 1994). The analysis of microsatellites is based on the polymerase chain reaction (PCR), which is much easier to perform than RFLP analysis and is highly amenable to automation. In plants, it has been demonstrated that microsatellites are highly informative, locus-specific markers in many species (Condit and Hubbell 1991; Akkayaet al. 1992; Lagercrantzet al. 1993; Senior and Heun 1993; Wu and Tanksley 1993; Bell and Ecker 1994; Saghai-Maroofet al. 1994; Rongwenet al. 1995; Liuet al. 1996; Mörchenet al. 1996; Provanet al. 1996; Szewc-McFaddenet al. 1996; Taramino and Tingey 1996; Smulderset al. 1997). Because they are multiallelic, microsatellites have high potential for use in evolutionary studies (Schloettereret al. 1991; Buchananet al. 1994) and studies regarding genetic relationships.
Microsatellites show a much higher level of polymorphism and informativeness in hexaploid bread wheat than any other marker system (Plaschkeet al. 1995; Röderet al. 1995; Maet al. 1996; Bryanet al. 1997). However, due to the large genome size, the development of microsatellite markers in wheat is extremely time-consuming and expensive. Only 30% of all primer pairs developed from microsatellite sequences are functional and suitable for genetic analysis (Röderet al. 1995; Bryanet al. 1997). The majority of such markers are inherited in a codominant manner and, in most cases, they are chromosome-specific. This is a useful feature in a hexaploid genome. In this article, we present the development of 230 polymorphic primer sets and a genetic map of the wheat genome containing 279 microsatellites covering the seven homoeologous chromosome groups.
MATERIALS AND METHODS
Plant material and DNA extraction: The variety Chinese Spring was used as the DNA source for the development of wheat microsatellites. Mapping was performed on 70 recombinant inbred (RI) lines from the International Triticeae Mapping Initiative (ITMI) population. This population was derived by single seed descent (F8) from the cross of W-7984, an amphi-hexaploid wheat synthesized from Triticum tauschii (DD) and the T. durum (AABB) variety Altar 84, with the Mexican wheat variety Opata 85 from CIMMYT (Centro Internacional de Mejoramiento de Maiz y Trigo). The plant material was described in Van Deynze et al. (1995), and seeds were kindly provided by M. Sorrells, Cornell University. DNA was extracted from whole seeds as described in Plaschkeet al. 1995.
Microsatellite marker development: For microsatellite isolation, various phage λ libraries were constructed by cloning Chinese Spring genomic DNA. After digestion with the restriction enzyme AcsI, DNA was cloned into the EcoRI site of the vector Lambda Zap II (Stratagene, La Jolla, CA) or, alternatively, after digestion with MboI or Sau3A, into the BamHI site of the vector Lambda Zap express (Stratagene) according to the manufacturer's instruction. Initially, total genomic DNA was completely digested and used without size selection. Later, genomic wheat DNA (500 μg) was predigested with the methylation-sensitive restriction enzyme PstI. PstI-digested DNA was separated on preparative agarose gels, and the size range of 2–5 kb was excised and isolated using the GeneClean kit (Dianova). The size-selected DNA was further digested with MboI and cloned as described above. Unamplified libraries were plated and phage filters were probed with synthetic polymers of GA and GT (Pharmacia, Piscataway, NJ) and then washed to a stringency of 0.5× SSC, 0.1% sodium dodecyl sulfate (SDS)at 65° (Röderet al. 1995). Positive plaques were purified and converted into plasmids by in vivo excision. Plasmid clones were reconfirmed by colony hybridization and sequenced according to standard procedures using automated laser fluorescence (ALF) DNA sequencers (Pharmacia). Primer pairs flanking the microsatellite motifs were designed using the program Primer 0.5, which was kindly provided by E. Lander (Massachusetts Institute of Technology). The program Primer 0.5 allows checking for known repetitive sequences and exclusion of these sequences in the designated primers. For this purpose a data file was created consisting of published repetitive wheat sequences and of sequences of microsatellite markers that had resulted in a smear after PCR amplification. This data file was routinely used to check for repeated sequences when new primer pairs were developed. One primer was always labeled with fluorescein. If it was not possible to design both primers simultaneously, one fluorescein-labeled primer was designed close to the microsatellite, and further sequence information was obtained in another sequencing reaction using that primer.
A list of all pribmer sequences and mapped microsatellites, including the microsatellite motif, annealing temperatures (Tm), and allele sizes in the parent lines are presented in the appendix.
Polymerase chain reaction and fragment analysis: PCR reactions were performed in a volume of 25 μl in Perkin-Elmer (Norwalk, CT) thermocyclers. The reaction mixture contained 250 nm of each primer, 0.2 mm of each deoxynucleotide, 1.5 mm MgCl2, 1 unit Taq polymerase, and 50–100 ng of template DNA. The mapping reactions were set up using a pipetting robot (Biomek 1000; Beckman, Fullerton, CA). After 3 min at 94°, 45 cycles were performed with 1 min at 94°, 1 min at either 50, 55, or 60° (depending on the individual microsatellite), 2 min at 72°, and a final extension step of 10 min at 72°.
Fragment analysis was carried out on automated laser fluorescence (ALF) sequencers (Pharmacia) using short gel cassettes. Denaturing gels (0.35 mm thick) with 6% polyacrylamide were prepared using SequaGel XR (Biozym). The gels were run in 1× TBE buffer [0.09 m Tris-borate (pH 8.3) and 2 mm EDTA] with 600 V, 50 mA, and 50 W with 2 mW laser power and a sampling interval of 0.84 sec. The gels were reused four to five times. In each lane, fragments with known sizes were included as standards. Fragment sizes were calculated using the computer program Fragment Manager Version 1.2 (Pharmacia) by comparison with the internal size standards.
Approximately 30 microsatellites were mapped using conventional sequencing gels and visualization by silver staining as described by Sourdille et al. (1998).
Genetic mapping: The microsatellites were integrated into a framework map composed of 302 RFLP markers. The data for the RFLP markers were kindly provided by C. Nelson and M. Sorrells (Cornell University) and are based on previously published RFLP maps (Nelson et al. 1995a,b,c; Van Deynzeet al. 1995; Marinoet al. 1996). As far as possible, the RFLP framework was constructed at a LOD of 3.0, andthe microsatellite markers were assigned to chromosomes using the “PLACE” command of the computer program MAPMAKER 2.0 (Landeret al. 1987). Marker position within the respective chromosome was determined with the “TRY” and “RIPPLE” commands. Centimorgan units were calculated using the Kosambi mapping function (Kosambi 1944). In a few ambiguous cases, additional nulli-tetrasomic analysis of the microsatellite markers was performed as described previously (Röderet al. 1995). Mapped wheat microsatellite loci were designated Xgwm for “Gatersleben wheat microsatellite.”
Marker development: Efficacy of microsatellite isolation: Microsatellite-containing clones were purified from various genomic phage λ libraries containing small inserts (see materials and methods). Primer pairs could be designed for ~54% of the sequenced clones containing GA or GT microsatellites based on hybridization of the plasmid clones. It was not possible to design two primers for the other 46% because of the following reasons: First, 36% of the clones did not contain microsatellite arrays in the sequenced region (usually 400–500 bp from either side). This was due to the fact that a number of clones were much larger than the sequenced region or contained multiple inserts. Second, for 4% of the microsatellites it was not possible to design both primers because the microsatellite was too close to one of the cloning sites. Finally, 6% of the clones contained repeated DNA regions close to the microsatellite site that were detected with the program Primer 0.5.
Functionality of primer pairs: As previously reported (Röderet al. 1995; Bryanet al. 1997), only ~30% of the primer pairs designed from wheat microsatellite sequences yield functional microsatellite markers. Functionality is defined as amplification of a fragment of the same size as the sequence of the respective clone. Nonfunctional primer pairs amplified either a smear (large numbers of fragments), nothing, or fragments of the wrong size. Fragments with unexpected sizes were usually monomorphic.
Effects of different libraries: The AcsI and MboI libraries yielded a large number of primer pairs that produced a smear after PCR amplification. We assumed that, due to the large genome size of wheat, such a smear was created from microsatellites harbored in repeated DNA. We have investigated this by predigesting wheat DNA with the methylation-sensitive restriction enzyme PstI. This enzyme is known to cut preferentially in single-copy DNA of many plant species. By predigestion with PstI and subsequent isolation of the fragments in the size range of 2–5 kb before digestion with a 4-bp restriction enzyme (MboI or Sau3A) and cloning, it was possible to increase the success rate of functional primers from 31 to 67% (Table 1). Using this procedure, the number of primer pairs yielding a smear was reduced significantly. Interestingly, this increase in effectiveness was only obtained by predigestion with PstI. The use of EcoRII, another CNG methylation-sensitive restriction enzyme, did not produce this increase in effectiveness. In total, 1380 clones were sequenced, and primer pairs were designed for 720 clones. A total of 294 primer pairs (41%) yielded a discrete fragment of the expected fragment size.
Number and polymorphism of amplified PCR fragments: Eighty percent of the primer pairs amplifying a fragment of the expected size detected polymorphism between Opata 85 and the synthetic wheat W7984, the parents of the RI lines. Of these, ~40% exclusively amplified the expected fragment, 40% amplified mostly one or, in a few cases, several additional monomorphic fragments, and 20% amplified one or several additional polymorphic fragments. Therefore, only one site could be mapped for 80% of the markers, and two or more sites were mappable for 20% of the markers.
The wheat microsatellite map: Map construction: The polymorphic microsatellites were integrated into a framework RFLP map of all chromosomes. Only those markers that could be ordered at a LOD score of >2.5 were directly included in the RFLP framework. All other markers were assigned to the most likely interval according to Nelson et al. (1995a,b,c). The linkage map is shown in Figure 1. In total, 230 primer sets amplified 279 microsatellites, 65 of which were mapped at a LOD score >2.5 and 214 of which were assigned to intervals on the RFLP map.
In two cases, independently isolated microsatellites appeared to be duplicates that cosegregated and consisted of identical or almost identical sequences. This was the case for Xgwm213 and Xgwm335 on chromosome 5B and for Xgwm269 and Xgwm565 on chromosome 5D.
The centromeres were positioned according to previously published RFLP maps (Nelson et al. 1995a,b,c; Van Deynzeet al. 1995; Marinoet al. 1996). In cases where microsatellites mapped in the centromeric region, their chromosomal arm locations were determined by analysis with the respective ditelosomic lines of Chinese Spring.
Compared to the previously published RFLP maps, three changes were made in the framework. These were suggested by new results of nulli-tetrasomic analyses of RFLP markers in the respective chromosomal regions (J. C. Nelson, personal communication). The end of the 2AS linkage group from Xbcd348.1 to Xcdo447 was moved to the end of the 2BS linkage group, the end of the 3AL linkage group ranging from Xabc172.2 to Xbcd451 was moved to the end of the 3DS linkage group, and the 4AL linkage group from Xbcd129 to Xbcd1975 was moved to the end of the 7DS linkage group. These changes were corroborated by nulli-tetrasomic analysis of the microsatellites mapping to the respective chromosomal regions: Xgwm210-2B mapped to chromosome 2B, Xgwm114-3D to 3D, and Xgwm635-7D to chromosome 7D.
The original RFLP framework map was extended by microsatellites mapping outside the outermost RFLP locus on the ends of the 2AS, 5AS, 5AL, 5DS, 6BS, 7AS, 7BS, and 7BL linkage groups.
Genome specificity of microsatellite markers: Only 37 of 230 primer sets produced more than one mappable locus. The majority of 193 microsatellite markers constitute genome-specific markers. The highest number of loci was detected by Xgwm666 with five sites, all mapping to the A genome. The primer sets that amplified two or more loci mapped to homoeologous as well as to nonhomoeologous sites. In nine cases, microsatellites mapped to two homoeologous sites, and in four cases they mapped to three homoeologous sites (Figure 1). Xgwm165 mapped to chromosome arms 4AS, 4BL, and 4DL, thus marking the known chromosome 4A pericentric inversion (Nelsonet al. 1995c).
The B genome contains the highest number of microsatellites, 115, the A genome 93, and the D genome only 71. Low numbers of microsatellite markers were found in chromosomes 1A, 4A, 6A, 1D, 4D, 6D, and 7D. Along the individual linkage groups, the mapped markers were evenly distributed with no significant clustering except in the centromeric regions of some chromosomes.
We present here the first genetic map of the wheat genome based on microsatellites. The development of wheat microsatellites is a tedious task. Primer pairs can be developed for only 54% of the sequenced plasmid clones containing microsatellites. Also, using short insert libraries developed from digestion with 4-bp recognition restriction enzymes, the percentage of useful primer pairs that amplify a polymorphic fragment of the expected size is in the range of 30%. Thus, on average, one out of six purified microsatellite-containing clones yields a functional primer pair. From these data, it is obvious that the development of wheat microsatellites is a tedious process that requires optimization. One possible way to increase the rate of microsatellite-containing clones for which primer pairs can be designed might be the use of libraries that are enriched for microsatellites and/or are size-selected for clones below an insert size of 1000 bp. However, a disadvantage of such enrichment procedures, which is associated with smaller inserts, is the increased frequency of microsatellites too close to one of the cloning sites. Furthermore, enriched libraries carry a considerable risk of obtaining duplicate clones.
We found that an effective way to increase the efficiency of functional primer pairs is to use the undermethylated fraction of the wheat genome as a source for microsatellite isolation. As has been shown for the isolation of single-copy RFLP clones from plants with large genomes, predigestion with the CNG methylation sensitive restriction enzyme PstI creates a fraction that is highly enriched for low- and single-copy DNA. Using this DNA fraction as a source for microsatellite clones, it was possible to reduce the number of microsatellite clones derived from repeated DNA and thus effectively double the number of functional microsatellites isolated from the wheat genome. Interestingly, the use of the similarly CNG methylation-sensitive enzyme EcoRII did not yield this increase in effectiveness. At the moment, it is not clear why such differences between CNG methylation-sensitive restriction enzymes exist.
The identification and mapping of 279 microsatellites amplified with 230 primer sets demonstrates that wheat microsatellites are mainly genome-specific and that microsatellite primer sets usually amplify only a singlelocus from one of the three genomes. Wheat microsatellite primer sets were successfully used for the amplification of DNA from wild progenitors or relatives of bread wheat T. monococcum, T. boeoticum, T. urartu (V. Korzun and M.-H. Tixier, unpublished data), T. dicoccoides (Fahimaet al. 1998), T. durum, and T. aethiopicum (Plaschkeet al. 1995). This indicates that microsatellite sequence diversity between the genomes is much higher than between each genome and its diploid and tetraploid ancestors. Only 20% of all primer sets amplify more than a single locus. Of these, approximately one-half amplify orthologous loci. The other one-half amplify loci from nonhomoeologous regions in the wheat genome. One possible explanation for this is that microsatellite markers can be derived from moderately repeated DNA sequences, provided that their primer sequences are sufficiently specific to amplify only a single or very few loci. It is known that a large portion of the Gramineae genomes is composed of ancestral transposable elements such as inactive retrotransposons. If a microsatellite marker resides within such a moderately repetitive element, nonorthologous loci could be amplified.
Of 279 microsatellites, 65 could be integrated into the RFLP framework with a LOD >2.5, whereas 214 microsatellites were assigned to intervals. In the previously published RFLP maps of wheat also <50% of the RFLP markers were mapped with a LOD >3.0 (Nelson et al. 1995a,b,c; Van Deynzeet al. 1995; Marinoet al. 1996). One reason for the occurrence of low LOD scores in the mapping population may be, besides very close distances of the markers, a considerable amount of residual heterozygosity in the recombinant inbred (RI) lines. For mapping of the RFLPs and the microsatellites, different generations of RIs were used, which might lead to different levels of heterozygosity in the same RI lines. Furthermore, for the mapping of microsatellites, only 70 plants were used, although the RFLP framework is composed of data for 114 plants. This results in a reduced amount of mapping information for the microsatellite markers related to the RFLPs.
Microsatellites in hexaploid wheat are fairly evenly distributed along the linkage groups. We have not observed a significant clustering of such markers, with the exception of several centromeric regions on chromosomes 2A, 3A, 3B, 4B, 5B, and 6B. Thus, microsatellites are useful for complete coverage of the wheat genome in the same way as RFLP markers. Data from physical mapping of microsatellites on deletion stocks of group 2 chromosomes (Röderet al. 1998) confirm that microsatellites are not physically clustered in specific regions of the wheat chromosomes. This situation is similar to the results found for other Gramineae and is clearly different from their chromosomal location in sugar beet and tomato. In these two species, microsatellites are heavily clustered around the centromeres (Schmidt and Heslop-Harrison 1996; T. Areshchenkova and M. W. Ganal, unpublished results).
Of the 279 microsatellites, 93 mapped to the A genome, 115 to the B genome, and 71 to the D genome. The percentage of markers assigned to the respective genomes and chromosomes is in good agreement with the numbers obtained for RFLP markers (Marinoet al. 1996) and thus reflects mainly the amount of polymorphism within the different genomes in the ITMI mapping population, rather than an unequal distribution of microsatellites. In order to increase the number of A or D genome microsatellites, they could be isolated from T. monococcum or T. tauschii. Preliminary data suggest that by using the diploid ancestors as a source for microsatellite isolation, it is possible to specifically enrich for microsatellites from the D genome (M. S. Röder, unpublished results).
Most of the published molecular maps of wheat include only a few mutant loci and agronomically important genes. The main reason for this is that the use of RFLPs and isozyme markers for mapping has been inefficient because of a low level of allelic variation (<10%) among cultivated varieties (Chaoet al. 1989; Kam-Morganet al. 1989). In addition, RFLP assays require large quantities of DNA and are technically demanding and laborious, and the most common detection method uses radioisotopes. In contrast, microsatellites are abundant, highly polymorphic, evenly distributed over the genome, and require only small amounts of genomic DNA for analysis. Therefore, they are highly suitable as genetic markers in wheat for mapping agronomically important genes. Furthermore, the analysis of microsatellites can easily be automated and applied to large plant numbers, as has been shown for microsatellite analysis in the human genome (Mansfieldet al. 1994).
The map presented here provides a good starting point for the production of a saturated map of the wheat genome based on microsatellites. Microsatellites provide readily detectable markers for agronomically important genes and quantitatively inherited traits and facilitate their handling in segregating breeding populations. Examples for this are the use of microsatellites for molecular mapping of known genes of bread wheat, including the dwarfing genes Rht8 (Korzunet al. 1998) and Rht12 (Korzunet al. 1997) in chromosome arms 2DS and 5AL and the major vernalization genes Vrn1, Vrn2, and Vrn3 (V. Korzun, unpublished data) in chromosome arms 5AL, 5BL, and 5DL, respectively.
We thank Angelika Flieger and Susanne König for excellent technical assistance. This work was supported by the Deutsche Forschungsgemeinschaft (Ro-1055/1-2).
APPENDIX Description of wheat microsatellite primer sets and loci
The primer sequences described in this article are available for public research only. Requests for commercial use of the primer pairs should be directed to the corresponding author.
Communicating editor: J. A. Birchler
- Received February 2, 1998.
- Accepted April 24, 1998.
- Copyright © 1998 by the Genetics Society of America