The current rate of yield gain in crops is insufficient to meet the predicted demands. Capturing the yield boost from heterosis is one of the few technologies that offers rapid gain. Hybrids are widely used for cereals, maize and rice, but it has been a challenge to develop a viable hybrid system for bread wheat due to the wheat genome complexity, which is both large and hexaploid. Wheat is our most widely grown crop providing 20% of the calories for humans. Here, we describe the identification of Ms1, a gene proposed for use in large-scale, low-cost production of male-sterile (ms) female lines necessary for hybrid wheat seed production. We show that Ms1 completely restores fertility to ms1d, and encodes a glycosylphosphatidylinositol-anchored lipid transfer protein, necessary for pollen exine development. This represents a key step towards developing a robust hybridization platform in wheat.

Wheat flower opens when self-pollination fails. The unfertilized ovary pushes the palea and lemma apart by increasing radial size associated with cell enlargement and suppression of pericarp degradation, therefore facilitating cross-pollination.

Summary Nuclear male‐sterile mutants with non‐conditional, recessive and strictly monogenic inheritance are useful for both hybrid and conventional breeding systems, and have long been a research focus for many crops. In allohexaploid wheat, however, genic redundancy results in rarity of such mutants, with the ethyl methanesulfonate‐induced mutant ms5 among the few reported to date. Here, we identify TaMs5 as a glycosylphosphatidylinositol‐anchored lipid transfer protein required for normal pollen exine development, and by transgenic complementation demonstrate that TaMs5‐A restores fertility to ms5. We show ms5 locates to a centromere‐proximal interval and has a sterility inheritance pattern modulated by TaMs5‐D but not TaMs5‐B. We describe two allelic forms of TaMs5‐D, one of which is non‐functional and confers mono‐factorial inheritance of sterility. The second form is functional but shows incomplete dominance. Consistent with reduced functionality, transcript abundance in developing anthers was found to be lower for TaMs5‐D than TaMs5‐A. At the 3B homoeolocus, we found only non‐functional alleles among 178 diverse hexaploid and tetraploid wheats that include landraces and Triticum dicoccoides. Apparent ubiquity of non‐functional TaMs5‐B alleles suggests loss‐of‐function arose early in wheat evolution and, therefore, at most knockout of two homoeoloci is required for sterility. This work provides genetic information, resources and tools required for successful implementation of ms5 sterility in breeding systems for bread and durum wheats.