| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Dept. of Agronomy, Purdue University, 1150 Lilly Hall, 915 W. State St., West Lafayette, IN 47906
* Corresponding author (cweil{at}purdue.edu).
 |
ABSTRACT |
|---|
 TOPNOTES ABSTRACT INTRODUCTIONThe Maize TILLING ProjectREFERENCES |
|---|
|
INTRODUCTION |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONThe Maize TILLING ProjectREFERENCES |
|---|
In crop species, these traits are often commercially important, leading to improvements in disease or stress resistance or in nutritional quality. Even in cases of quantitative traits, it is the accumulation of several point mutations that often lead to the desired result. Thus, as genome sequences are completed, the ability to detect and characterize these mutations and then work with the mutant lines is now more important than ever. The complete sequence of the maize genome will become available in about two years. However, large segments of the sequence are available now and allow us to begin the next great challenge, understanding the functions and interactions of all the genes.
TILLING is a high throughput, reverse genetics technique for identifying single base changes in specific gene targets across a mutagenized population (McCallum et al., 2000; Till et al., 2003, 2006a; Henikoff et al., 2004; Comai and Henikoff, 2006). Detection of point mutations is based on high throughput PCR and mismatch detection. The mutations are induced by mutagenesis with chemicals such as ethyl nitrosourea and sodium azide. In maize, we have used EMS, which causes G to A transitions randomly throughout the genome. Mismatched bases form when DNA from a point mutant for a specific gene of interest is PCR amplified as part of a pool of DNAs in which the rest of the DNAs are not mutant for that target gene (Fig. 1 ). PCR primers are differentially end-labeled with fluorescent dyes. When mutant and nonmutant molecules amplified with these primers reanneal with one another, the heteroduplex molecules are cleaved specifically at the mismatch by an S1 family mismatch nuclease such as CEL1 (Till et al., 2004a). The result is two molecules of less than full length, complementary in size, each labeled with a different fluorescent tag. These products can then be resolved easily on slab gel or capillary DNA analyzers, the mutant individuals identified and the mutations sequenced. The two-dimensional readout of the slab gel format (e.g., Li-COR 4300, Lincoln, NE) provides an excellent combination of useful diagnostics for reaction quality, accurate size estimations and throughput. Eight-fold pools are constructed from 8 by 8 arrays of mutant DNAs (Fig. 2 ), and two-dimensional pooling allows us to identify mutant individuals in a single gel run without the need to deconvolute pools.
|
|
Projects in Arabidopsis, C. elegans, Drosophila, rat, and zebrafish have all confirmed the power of TILLING for functional genomics (Henikoff and Comai, 2003; Henikoff et al., 2004). Among crop plants, TILLING is now also used in maize, barley (Hordeum vulgare L.), and wheat (Triticum aestivum L.), and pilot studies are being completed in tomato (Solanum Lycopersicum L.), rice (Oryza sativa L.), and soybean [Glycine max (L.) Merr.], among others (Caldwell et al., 2004; Till et al., 2004b; Slade et al., 2005). In addition, a modification of the basic technique, called "EcoTILLING," can show rapidly and inexpensively the natural allelic variation at a locus among accessions of a species (Comai et al., 2004).
The complete sequence of the maize genome will become available in about two years. However, large segments of the sequence are available now and allow us to begin the next great challenge, understanding the functions and interactions of all the genes.
A central element of functional genomics is linking allelic variation, induced or otherwise, to phenotypes, and TILLING provides a way to screen through a large population of mutants. In maize, analyses of these mutations are enhanced by genetic and cytogenetic tools that have been developed over decades (perhaps the most extensive for any crop plant). By combining those tools with a sequenced genome and the analytical and functional genomics tools that have been developed in model systems such as Arabidopsis and rice, maize will be one of the most powerful plant biology systems available as well as economically and agriculturally important.
One of the potential concerns for functional genomics in an ancient tetraploid genome like that of maize is that, in some cases, double and even triple maize mutants will be required before phenotypes can be analyzed. Reverse genetic alleles will be extremely important in this effort, especially chemically induced TILLING alleles because point mutations are much less constrained to particular locations in the genome. Once identified, mutant alleles in duplicate factors can be combined readily and analyzed for phenotype, particularly if they are all in the same genetic background, (recently demonstrated for TILLING alleles of the wheat waxy genes [Slade et al., 2005]). In these cases, the multiple mutants will also be much more effective than single mutants in array-based gene expression and interaction studies.
|
The Maize TILLING Project |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONThe Maize TILLING ProjectREFERENCES |
|---|
47 kb of protein coding exon). In addition, the Maize TILLING Project is in the process of TILLING 17 other genes, and completing primer testing for another 25. These data contribute to a genome-wide understanding of gene function. More importantly for that understanding, once mutant lines are identified, seed samples for growing those lines are made available for further study. The data are kept confidential for a period of 6 mo after the results are returned to the laboratory requesting the TILLING. After 180 d, the data are made available to the community via the project website (http://genome.purdue.edu/maizetilling; verified 12 Dec. 2006) and MaizeGDB (http://maizegdb.org; verified 12 Dec. 2006), and the sequences of the target (from both the B73 and the W22 inbreds) are submitted to GenBank. Another important component in TILLING is the informatics at both the beginning and the end of the process. These have been developed and are administered through the ProWeb group housed at the Fred Hutchinson Cancer Research Center (Seattle, WA). A sequence of interest is entered by the user into the web-based program CODDLe (Codons Optimized to Detect Deleterious Lesions) (McCallum et al., 2000) along with a gene model, which the program will proofread to verify that it is correct. CODDLe then scans the input sequence for all possible G to A transitions on both strands (the lesion induced by EMS). In addition, it uses the BLOCKS program (Henikoff et al., 2000) to identify any known sequence motifs in the input sequence that might represent important conserved amino acids; alternatively, the user can input sequence alignments and blocks of conservation they have identified. CODDLe then identifies the effects that all the transition mutations would have on the predicted protein, then identifies the region of the input gene that is most likely to produce the highest number of severely deleterious mutations. The user is shown the result and asked if primers should be designed for TILLING this region or whether another region should be screened. Several sets of primers are designed using Primer3 (Rozen and Skaletsky, 2000) a publicly available program that has been preset with parameters specific for TILLING. The results are again presented to the user, who selects the primers and can then enter the order.
The duplicated nature of the maize genome, the wide variety of inbred lines used by the maize community, and the sometimes patchwork methods with which sequences can be assembled in silico has required that an additional but important step be added to TILLING at this stage of the process. -Unlabeled primers for each target gene are first tested on genomic DNA of both B73 and W22 (our TILLING inbreds) to be certain that they will produce amplicons of the expected size and quality. This prescreen prevents expensive and wasteful attempts to analyze genes that would not "TILL" successfully. The two TILLING inbred lines can differ substantially: B73 is the inbred chosen for genome sequencing, and W22 is an inbred with extensive seed trait and transposon mutagenesis resources that the community can leverage. As a result, testing the performance of the primers on both inbreds has also been a valuable prescreen. This prescreening has not dramatically affected our throughput (Fig. 3 ). More recently, we have posted our PCR protocols on the project website so that users can prescreen the primer sets they identify in CODDLe (often available to them even less expensively through in-house oligonucleotide synthesis services) and, by the time the order is submitted, success is virtually assured. With the advent of prescreening primers, >95% of TILLING orders pass through to the next stages on the first or second design attempts. We also DNA sequence the amplicons we recover in prescreening to establish a baseline against which to compare any mutations we identify. Once the B73 sequence is completed this may become less necessary for B73 but will remain important for any other inbreds.
|
TILLING screens are regarded as complete only when alleles that either truncate or severely damage the protein are returned to the user. Ideally, an allelic series is returned that includes such alleles, but if not the target is screened at no extra charge each time new DNA templates become available until such mutations are found. As mentioned earlier, data returned to users are kept confidential for a period of 180 d before being posted to the project website, GenBank, and MaizeGDB.
Mutant Populations: In Search of the "Mother Lode"
TILLING requires large mutant populations of maize and we have made these by EMS-mutagenizing pollen from our two inbred lines. These populations prove to be extremely valuable both as reverse and as forward genetics material. TILLING itself represents an extremely effective and accurate way to gauge how well the mutageneses have worked by measuring the induced mutation density in a treated population and doing so at the level of the DNA rather than phenotypically. All mutations are heterozygous at this stage. The assay is performed on DNA collected while the M1 population are still seedlings, rather than having to wait until M2 phenotypes (e.g., chlorosis, kernel defects, or embryo lethality) can be evaluated.
A breakdown of what is contained in such a TILLING population and how it is calculated is a useful exercise, and we present it here in text rather than table form for that reason. The best populations for TILLING on a large scale will be those that approach the limits of the mutational load that plants can tolerate and still remain fertile. Our first TILLING population (POPULATION A) included 2370 unique B73 mutant lines donated by Nathan Springer, now at the University of Minnesota, and 1276 W22 lines mutagenized at Purdue University. Both mutageneses, albeit at different locations, were performed using the method published by Neuffer (1992), treating pollen with 0.0625% EMS in paraffin oil for 1 h. The B73 mutant lines identified an average of 0.93 mutations per 1 kb screened every 1000 families, and the W22 lines identified 2.10 mutations per kilobase per 1000 families. There are numerous anecdotal reports about differences between inbred lines in their response to EMS pollen treatment; however, a systematic survey of which inbreds respond most effectively, using controlled conditions for all treatments, has never been performed. We will be doing just such a survey among the Maize Diversity Lines in 2007.
Using these empirical data, POPULATION A contains an estimated 1.1 x 107 mutations. Twenty-six (17.7%) of the sequenced mutations we have identified in this population either truncate the gene product (3.4%) or are predicted to be severely damaging using the SIFT program (14.3%), which considers the severity of an amino acid substitution and how conserved the amino acid is at that position among other, similar proteins. If 20% of the maize genome is taken to encode genes, this represents 2.20 x 106 mutations in the "gene space." The maize genome is estimated to contain 35 000 to 41 000 genes (S. Tingey and R. Martienssen, pers. comm., 2005) and a survey of >3700 maize proteins in GenBank estimates that an "average" maize gene contains 945 bp of protein coding exon, thus 7 to 8% of the gene space (33 750 kb) appears to be coding exon. We can therefore estimate 164 700 of the mutations in POPULATION A are in protein coding exon and that there are 29 152 truncating or severely damaging alleles, or 0.71 to 0.83 damaging alleles in every gene. It is important to remember that these estimates of damaging allele frequency are based on informatics that evaluate sequence conservation. What fraction of nonsilent mutations will produce phenotypes, either on their own or in combination with other alleles, remains a crucial open question. In addition, mutations can prove to be phenotypically interesting even though they do not occur within stretches of highly conserved amino acids (Mason-Gamer et al., 1998).
Two additional populations have been made, one of W22 (at Purdue) and one of B73 (at Iowa State University by An-Ping Hsia and Pat Schnable). Preliminary TILLING tests on samples of these populations indicate they each have mutation densities roughly two to four times that of POPULATION A, or approximately 100 000 truncating or severe alleles. We are in the process of optimizing the mutagenesis protocol still further, with the goal of developing populations in which screens of 3000 plants can yield >20 alleles for any given gene. Our preliminary trials have indicated that such plants can be made and remain fertile. In contrast to species that self-fertilize on their own, the logistics and timelines for doing this are complex for a plant that requires hand-pollination like maize, particularly because there are significant numbers of lethal and sterile mutations present in the M1 generation. However, the benefit of such an effort is enormous because these populations provide both reverse genetics power to begin characterizing genes of unknown function as well as a renewable, common-use resource for phenotype-based ("forward genetics") screens. We invite the community to come walk through our fields each summer to look for interesting mutations, and we post kernel, ear, and plant phenotypes to the MaizeGDB EMS phenotype database. Seed for all these lines are available to the community at no charge. In addition to our own efforts to make EMS "Mother Lode" maize populations, EMS pollen mutagenesis is often done on smaller scales by other labs as part of specific projects. We regularly seek collaborations with all these efforts, and, if we can access M1 plant tissue for DNA together with seed we can increase for distribution, we are happy to evaluate DNAs from anyone wishing to provide them.
The best EMS-mutagenized maize populations obtained using pollen treatment have been reported to have mutation densities as high as one visible, recessive mutation (and therefore nonsilent and damaging to the gene product) per gene per 1000 families (Neuffer et al., 1997). This figure reflects how many genes there are for which one mutation can produce a phenotype, that is, nonduplicated genes or genes for which one member of a family has a prominent role. Until the genome sequence is complete, the size of this fraction is difficult to pinpoint, even with high Cot fractions and methyl filtration techniques. Presumably, a larger number of mutations are induced in still other genes for which no phenotype results even though the mutations are nonsilent. However, we can get an estimate of what one visible mutation per gene per 1000 families represents, even if it is a clear underestimate. If one mutation per 1000 families is taken as 17.7% of the total mutations in the gene space, as described above, then these lines would carry a minimum of 5.65 (1/0.177) total mutations in the protein coding exons of each gene per 1000 families or 17 alleles per gene across a population of 3000. If we then multiply that estimate out by the estimated number of genes in maize, such a population would contain a total of 595 000 to 697 000 new alleles. Individual plants would be carrying 198 to 232 detectable, damaging lesions and over five times that many missense mutations that may produce either subtle missense mutations or conservative changes within protein coding sequence alone.
Not surprisingly, empirical data from the >100 genes we have screened to date indicate that we detect <2% (155) of the nearly 13 620 possible nonsilent mutations that could have occurred in these particular sequences by EMS treatment. These data are consistent with TILLING data from other organisms as well. TILLING population sizes are relatively small (as opposed to millions of individuals), and the enormous numbers of possible transition mutations are randomly induced. All of these calculations suggest that the maize genome is capable of carrying a higher mutational load than current methods induce, particularly for reverse genetics applications.
Getting such populations will take considerable effort, but the benefit considerably outweighs the cost. One of the key factors in establishing mutagenesis protocols has been the balance between useful mutations, sterility or lethality of the M1, and pollen death during the EMS treatment. It would seem that, as long as one is willing to perform a large number of pollen treatments, and plant a large number of M1 seed (many of which will not produce an M2), then whatever viable and fertile M2 lines result would be extremely valuable for reverse genetics. These lines can, essentially, be immortalized with one or two rounds of intermating members within a family and careful storage of the seed.
This is clearly a monumental amount of work, but it is important to remember that, eventually, assignment and verification of biological function for genes inevitably requires mutants. As mentioned previously, the most useful analysis of those genes will require more than just knockout mutants, and such analysis in maize will often require combining mutations. The various transposon-oriented and deletion-oriented projects currently underway are extremely valuable and will provide a lot of useful material, but they will not be sufficient on their own.
EcoTILLING
The immediate goal of Maize TILLING is identi-fication and delivery to the community of an allelic series for individual genes. Alleles with varying levels of gene function can be induced, as with EMS, but they also occur naturally and these variants can sometimes prove even more informative than induced variation (Gazzani et al., 2003). This is also likely to be true in maize, where the sequence diversity just between different inbred lines rivals the diversity between humans and chimps (Buckler et al., 2001; Liu et al., 2003). A variation on TILLING, called EcoTILLING, was devised to assess single nucleotide polymorphism among Arabidopsis ecotypes (Comai et al., 2004) and has since proven to be a powerful, accurate, and inexpensive SNP discovery tool in both humans and plants (Gilchrist et al., 2006; Till et al., 2006b). The methodology is similar to TILLING as described above for mutagenized populations except that the templates for the PCR contain two templates, one a reference genome and the other a cultivar, accession, ecotype, etc., to be tested. SNPs are detected as mismatched bases and cleaved. Using the Li-COR-based system as an example and duplicates of each sample for confirmation, 48 different inbred lines can be screened simultaneously in comparison to the reference (ideally the sequenced genome of the inbred B73). The Maize Diversity Lines (Liu et al., 2003) are known to reflect >80% of the diversity in maize germplasm worldwide. By EcoTILLING these lines using targets already submitted for EMS TILLING, we can begin to accumulate diversity data and allelic variation for other, nonessential genes that may be under selection (Wright et al., 2005; Yamasaki et al., 2005). EcoTILLING provides a rapid and inexpensive—once a target is submitted for EMS TILLING most of the cost is already incurred—way to inform the community about how much natural diversity may be available to them for their gene(s) of interest as they investigate gene function. The gene models, already a part of the TILLING request, identify which polymorphisms are likely to be in exons (Fig. 4
). Genes from specific inbreds of interest can then be resequenced to determine the nature of the polymorphisms and their effects.
|
For now, EcoTILLING serves a useful role in assessing maize diversity at different loci because the genes submitted for TILLING represent a broad cross-section of the genome. Recent studies have demonstrated that genes likely to have been involved in the domestication of maize, and therefore under strong positive selection, show much less diversity than the genome average (Tenaillon et al., 2001; Matsuoka et al., 2002; Jaenicke-Despres et al., 2003; Clark et al., 2005). However, these "domestication genes" were chosen for testing precisely because of the likelihood they would have been selected by early farmers (e.g., seed architecture, starch quality, etc.). Whether less overtly agronomic genes might have been less intentionally involved in domestication remains an interesting and open question.
TILLING and Resequencing
It is likely that, within the next decade, resequencing entire eukaryotic genomes is going to become a surprisingly routine exercise. For bacterial genomes this has already come to pass. It is thus quite reasonable to imagine, should the mutant populations described above get made, that sequencing the entire genome of each mutant line and creating a database of that information would be a long-standing and heavily used resource. The alternative to such a collection would be the capacity to alter DNA sequences at will and replace them into any maize genome at their appropriate location. This, too, may be nearer than we think but, for now at least, routine, inexpensive maize transformation and homologous gene targeting are not available at the necessary levels. In particular, transformation of the sequenced inbred, B73, has proven difficult. As the two technologies of sequencing and homologous targeting of transgenes race forward, resequencing appears to be the stronger of the two possibilities in the near term. We would suggest that now is the time to begin making and evaluating mutant populations to resequence, and that TILLING is the best way to evaluate those populations.
|
ACKNOWLEDGMENTS |
|---|
|
NOTES |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONThe Maize TILLING ProjectREFERENCES |
|---|
Received for publication October 25, 2006.
|
REFERENCES |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONThe Maize TILLING ProjectREFERENCES |
|---|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| The SCI Journals | Agronomy Journal | Vadose Zone Journal | |||
| Journal of Plant Registrations | Soil Science Society of America Journal | ||||
| Journal of Natural Resources and Life Sciences Education |
Journal of Environmental Quality |
||||
