Key message

Specific combinations of LFY and WAPO1 natural alleles maximize spikelet number per spike in wheat. Spikelet number per spike (SNS) is an important yield component in wheat that determines the maximum number of grains that can be formed in a wheat spike. In wheat, loss-of-function mutations in LEAFY (LFY) or its interacting protein WHEAT ORTHOLOG OF APO1 (WAPO1) significantly reduce SNS by reducing the rate of formation of spikelet meristems. In previous studies, we identified a natural amino acid change in WAPO1 (C47F) that significantly increases SNS in hexaploid wheat. In this study, we searched for natural variants in LFY that were associated with differences in SNS and detected significant effects in the LFY-B region in a nested association mapping population. We generated a large mapping population and confirmed that the LFY-B polymorphism R80S is linked with the differences in SNS, suggesting that LFY-B is the likely causal gene. A haplotype analysis revealed two amino acid changes P34L and R80S, which were both enriched during wheat domestication and breeding suggesting positive selection. We also explored the interactions between the LFY and WAPO1 natural variants for SNS using biparental populations and identified significant interaction, in which the positive effect of the 80S and 34L alleles from LFY-B was only detected in the WAPO-A1 47F background but not in the 47C background. Based on these results, we propose that the allele combination WAPO-A1-47F/LFY-B 34L 80S can be used in wheat breeding programs to maximize SNS and increase grain yield potential in wheat.

Background

Wheat (Triticum ssp.) is an important food source for humans in many regions around the world. However, the ability to understand and modify gene function for crop improvement is hindered by the lack of available genomic resources. TILLING is a powerful reverse genetics approach that combines chemical mutagenesis with a high-throughput screen for mutations. Wheat is specially well-suited for TILLING due to the high mutation densities tolerated by polyploids, which allow for very efficient screens. Despite this, few TILLING populations are currently available. In addition, current TILLING screening protocols require high-throughput genotyping platforms, limiting their use.

Results

We developed mutant populations of pasta and common wheat and organized them for TILLING. To simplify and decrease costs, we developed a non-denaturing polyacrylamide gel set-up that uses ethidium bromide to detect fragments generated by crude celery juice extract digestion of heteroduplexes. This detection method had similar sensitivity as traditional LI-COR screens, suggesting that it represents a valid alternative. We developed genome-specific primers to circumvent the presence of multiple homoeologous copies of our target genes. Each mutant library was characterized by TILLING multiple genes, revealing high mutation densities in both the hexaploid (~1/38 kb) and tetraploid (~1/51 kb) populations for 50% GC targets. These mutation frequencies predict that screening 1,536 lines for an effective target region of 1.3 kb with 50% GC content will result in ~52 hexaploid and ~39 tetraploid mutant alleles. This implies a high probability of obtaining knock-out alleles (P = 0.91 for hexaploid, P = 0.84 for tetraploid), in addition to multiple missense mutations. In total, we identified over 275 novel alleles in eleven targeted gene/genome combinations in hexaploid and tetraploid wheat and have validated the presence of a subset of them in our seed stock.

Conclusion

We have generated reverse genetics TILLING resources for pasta and bread wheat and achieved a high mutation density in both populations. We also developed a modified screening method that will lower barriers to adopt this promising technology. We hope that the use of this reverse genetics resource will enable more researchers to pursue wheat functional genomics and provide novel allelic diversity for wheat improvement.

Comprehensive reverse genetic resources, which have been key to understanding gene function in diploid model organisms, are missing in many polyploid crops. Young polyploid species such as wheat, which was domesticated less than 10,000 y ago, have high levels of sequence identity among subgenomes that mask the effects of recessive alleles. Such redundancy reduces the probability of selection of favorable mutations during natural or human selection, but also allows wheat to tolerate high densities of induced mutations. Here we exploited this property to sequence and catalog more than 10 million mutations in the protein-coding regions of 2,735 mutant lines of tetraploid and hexaploid wheat. We detected, on average, 2,705 and 5,351 mutations per tetraploid and hexaploid line, respectively, which resulted in 35-40 mutations per kb in each population. With these mutation densities, we identified an average of 23-24 missense and truncation alleles per gene, with at least one truncation or deleterious missense mutation in more than 90% of the captured wheat genes per population. This public collection of mutant seed stocks and sequence data enables rapid identification of mutations in the different copies of the wheat genes, which can be combined to uncover previously hidden variation. Polyploidy is a central phenomenon in plant evolution, and many crop species have undergone recent genome duplication events. Therefore, the general strategy and methods developed herein can benefit other polyploid crops.