Gsp-1 genes comprise a multigene family in wheat that exhibits a unique combination of sequence diversity yet conservation

Abstract

While the commercially important trait of grain hardness in wheat is largely determined by the puroindoline proteins, a number of studies suggest additional loci, and a role for the grain softness protein-1 (GSP-1), encoded by the gene Gsp-1 tightly linked to the puroindoline genes, remains ambiguous. To investigate their role individual Gsp-1 genes were cloned from several accessions of durum, diploid and hexaploid wheat, rather than sequencing of genomic DNA PCR products. Analysis of clones identified >1 restriction fragment length polymorphisms (RFLPs) in some 2n progenitors, >2 RFLP types in several durums and >3 in common wheat. DNA sequencing identified 30 different Gsp-1 haplotypes from various 2n, 4n and 6n wheats, exhibiting extensive single nucleotide polymorphisms (SNPs) and a 3-base deletion, all but three haplotypes being novel. Shared SNPs between various diploids, indicated an ancient origin of this gene family, but certain aspects were unique to durums, and existence of multigenes in at least some genomes was confirmed through single plant analysis of select accessions of Aegilops speltoides, durum and common wheat. Most interestingly, despite the great sequence diversity, the functionally important amino acids involved in lipid binding, i.e., the 10 cysteines and two tryptophans, were retained in all putative proteins. The results suggest that these genes may be functionally important, particularly in durums which lack puroindolines, and may have major roles in plant defence and only a minor influence on grain texture.

Introduction

One of the most important characteristics of wheat is kernel texture (hardness), as it dictates the end-uses of the grain. Commercially, wheat is differentiated into three distinct texture classes: the soft and hard common hexaploid wheat varieties and the ‘very hard’ durum varieties. Grain hardness has long been recognised as a largely inheritable trait (Symes, 1965), controlled by the major hardness (Ha) locus on the short arm of chromosome 5D (Law et al., 1978; Mattern et al., 1973). The hardness trait has been associated with a ∼15 kDa group of proteins, initially identified through electrophoretic separation of proteins extracted from water-washed starch from soft, hard and durum lines (Greenwell and Schofield, 1986). The protein was initially termed ‘friabilin’ due to its abundance in water-washed starch from friable soft wheats but its paucity in hard wheats and its absence in durums. It was suggested that the protein acted as a ‘non-stick’ surface between starch granules and the protein matrix (Greenwell and Schofield, 1986). Later it was shown to be heterogeneous and composed of several polypeptides, encoded by genes which were all linked to the Ha locus in a ‘minimum recombination unit’ (Jolly et al., 1993), as confirmed by later studies (Turnbull et al., 2003). The electrophoretic separation and amino acid sequencing of the various friabilin polypeptides (Jolly et al., 1993; Morris et al., 1994) showed two major, basic polypeptides, puroindoline-a (PINA) and puroindoline-b (PINB) that shared ∼60% amino acid sequence identity (Gautier et al., 1994), and a minor, related polypeptide, the grain softness protein-1 (GSP-1) (Rahman et al., 1994). Since these discoveries, the linkage of all three genes, Pina, Pinb and Gsp-1, encoding PINA, PINB and GSP-1, respectively, to the Ha locus has been firmly established, as is the occurrence of the three genes at the Ha locus only on chromosome 5D in common wheat, both Pin genes having been deleted from chromosomes 5A and 5B in durum wheat and hence in common wheat. On the other hand all GSP-1 genes are conserved, as shown by numerous linkage analyses and/or physical localisation of genes in wheats of different ploidy levels (Chantret et al., 2004, Chantret et al., 2005; Gautier et al., 1994, Gautier et al., 2000; Jolly et al., 1996; Tranquilli et al., 1999; Turnbull et al., 2003; Turner et al., 1999).

Expression of the dominant soft phenotype (Morrison et al., 1989) appears to require a functional, starch-associated friabilin (as opposed to total friabilin or puroindoline), composed of both PINA and PINB and thus requiring expression of both the wild-type alleles, Pina-D1a and Pinb-D1a. (Capparelli et al., 2003; Giroux and Morris, 1998; Hogg et al., 2003). Deletion of the Pin genes from both A and B genomes is considered responsible for the very hard texture of durum wheat (Gautier et al., 1994, Gautier et al., 2000), with a degree of softness brought back into common wheat with restoration of the Pina-D1 and Pinb-D1 through the soft grained D genome donor, Aegilops tauschii (Gautier et al., 2000; Lillemo et al., 2002). The puroindolines are able to insert into lipid bilayers (Douliez et al., 2000; Guerneve et al., 1998), which may allow them to attach to the starch granules, leading to a soft endosperm (Giroux et al., 2000; Mairon et al., 1994). The relevance of starch granule-bound friabilin to the strength of adherence between starch granules and their surrounding endosperm proteins in the hard versus soft wheats, and the implications of this with respect to their milling energies, flour particle sizes, extent of starch damage during milling, water absorption properties of their flours and their suitability for breads versus biscuits have been explained elsewhere (Beecher et al., 2002; Giroux and Morris, 1997; Jolly et al., 1993; Lillemo and Morris, 2000). Mutations in either of the Pin genes have been associated with the hard texture in common wheat, the most common being the deletion of the Pina gene, i.e., allele Pina-D1b (Giroux and Morris, 1998), or six separate point mutations in the PinbD1 gene, i.e., alleles Pinb-D1b-g, Pinb-D1b being more prevalent (Cane et al., 2004; Giroux and Morris, 1997; Lillemo and Morris, 2000; Morris et al., 2001). The Pina-D1b allele is known to result in a harder texture than the Pinb-d1b allele (Giroux et al., 2000) and certain mutations in the Pinb-D1a gene also exhibit a range of hardness values (Lillemo and Morris, 2000; Lillemo and Ringlund, 2002). A few other rare mutations have been reported in both genes (Ikeda et al., 2005; Pan et al., 2004; Ram et al., 2005; Xia et al., 2005). Recent studies on a number of accessions of Ae. tauschii (Massa et al., 2004) and synthetic wheats (Gedye et al., 2004) also identified further alleles in both Pin genes; however, none of these has been associated with grain hardness.

Despite these reports showing a strong association between the Pina and Pinb mutant alleles and grain hardness, numerous observations since the earliest days have also indicated that the Ha locus on chromosome 5D is the major, but possibly not the only, genetic factor involved in determining this trait. Symes (1965) suggested at least two additional or modifying genes and Baker and Sutherland (1991) noted significant genetic variation in texture within crosses of the same textural class. Examination of published data (Giroux et al., 2000; Morris et al., 2001; Tranquilli et al., 2002) also shows that cultivars with the same Pina or Pinb allele display a range of hardness values not accounted for by the minor effects of environmental factors (Igrejas et al., 2001). Further, although minor texture variations within the ‘very hard’ durum wheats can result from environmental effects (Williams et al., 1998), the hardness range in durums has been shown to be independent of growing conditions (Chaurand et al., 1999; Dexter and Marchylo, 2000; Turnbull and Rahman, 2002; Williams et al., 1998; G. Grimes, pers. comm.). Considering the absence of both puroindolines in Triticum turgidum, such textural variations suggest a role for other genetic factors, a suggestion supported by recent work on synthetic wheats (Gedye et al., 2004) that showed that variations in hardness could also be assigned to the durum parent in the cross. There is also considerable evidence from QTL studies that additional factors are involved, with a major (Giroux et al., 2000) or minor (Campbell et al, 1999; Lillemo and Ringlund, 2002) influence. Such regions of interest include chromosome 5A (Morrison et al., 1989); loci on 4D and 4B (Osborne et al., 2001); various other regions showing single factor (2A, 2D, 5A and 6D) or interactive (5A, 6D, 7A) effects (Sourdille et al., 1996); a significant QTL on 1B, a minor QTL on short arm of 5A in a position homologous to the Ha locus, and one on long arm of 5A (Turner et al., 2004); minor QTLs on 1A and 6D (Perretant et al., 2000); a major QTL on long arm of 5D and suggestions of interactions (Igrejas et al., 2002); involvement of other amphiphilic proteins (Amiour et al., 2003), and most recently, reports of QTLs on 1AS, 1BL and 6B related to grain softness characteristics (Breseghello et al., 2005). Considering the critical nature of grain hardness to international wheat trade and usage, including to Australia, the world's second largest exporter of wheat, it is clear that the question of genetic determinants of this property has not been addressed unequivocally yet.

These observations raise the logical question as to whether the additional loci/QTLs could involve the third component of friabilin, i.e., GSP-1. The genetic variations in grain texture noted in durum wheats (mentioned above) and the degree of similarity between PINA, PINB and GSP-1 lend significant circumstantial support to this suggestion. The Gsp-1 gene is tightly linked to the Ha locus, GSP-1 shows approximately 40% identity and 60% similarity to the PINA and PINB proteins (Chantret et al., 2004; Turner et al., 1999) and the promoter region of Gsp-1 displays the same endosperm motif (EM) as the Pin genes, involved in tissue specificity of expression (Digeon et al., 1999; Turner et al., 1999). A tryptophan (Trp)-rich domain (TRD) also occurs at a highly conserved position in all three proteins, and further, all three exhibits 10 highly conserved cysteine (Cys) residues that are suggested to form disulphide bridges that stabilise a secondary structure with a potential site for lipid binding (Bihan et al., 1996). Despite these obvious similarities, the GSP-1 protein has remained somewhat of an enigma for several reasons. Firstly, there are contradictory suggestions regarding its role in grain hardness. Certain studies indicate a minor role, if any, for all or some of the Gsp-1 genes (Giroux and Morris, 1998; Lillemo and Ringlund, 2002; Tranquilli et al., 2002), while others imply a role for the Gsp-B1 locus (Tranquilli et al., 2002), or a critical role for the Gsp-1 genes and/or QTLs (Gedye et al., 2004). In relation to gene copy numbers, results of some Southern blots appear contradictory or inconclusive, with some digests showing only three bands while others suggesting multiple copies on (at least) some genomes and up to 18 genes in common wheat (Jolly et al., 1996; Turner et al., 1999).

With respect to biochemical properties, GSP-1 shows only two Trp residues in the TRD domain, in comparison to the five in PINA and four in PINB. No point mutations in Gsp-1 such as those in Pinb have been reported, raising doubts about its involvement in grain texture, yet there are no confirmed reports of deletions of Gsp-1 (similar to the Pina-D1b allele) and no mutations affecting the 10 Cys and the two Trp residues in the TRD in the many GSP-1 sequences reported from wheat (Massa et al., 2004; Rahman et al., 1994; Turner et al., 1999), T. monococcum (Chantret et al., 2004; Tranquilli et al., 1999; Turner et al., 1999) and Ae. tauschii (Massa et al., 2004). The gene is persistent through all ploidy levels and appears to have survived the recombination event (Chantret et al., 2004, Chantret et al., 2005) that led to the loss of the nearby, tightly linked Pin genes from durum (and thus common) wheat, suggesting an important functional role for the gene and possibly selection of those genotypes wherein Gsp-1 genes escaped this event.

The present work was directed to providing more information on the Gsp-1 gene family. The experiments involved cloning, restriction fragment length polymorphism (RFLP) analysis, sequence analysis and estimation of copy numbers of Gsp-1 genes from durum land races, common wheat cultivars and the three diploid progenitors. We also propose a model for their genealogy and a new haplotype nomenclature system, and suggest a possible function for what appears to be a small but moderately complex gene family.