Wheat (Triticum aestivum L. and T. turgidum L. ssp. durum) Kernel Hardness: I. Current View on the Role of Puroindolines and Polar Lipids


  • Anneleen Pauly,

    Corresponding author
    • Laboratory of Food Chemistry and Biochemistry and Leuven Food Science and Nutrition Research Centre (LFoRCe), KU Leuven, B-3001 Leuven, Belgium
    Search for more papers by this author
  • Bram Pareyt,

    1. Laboratory of Food Chemistry and Biochemistry and Leuven Food Science and Nutrition Research Centre (LFoRCe), KU Leuven, B-3001 Leuven, Belgium
    Search for more papers by this author
  • Ellen Fierens,

    1. Laboratory of Food Chemistry and Biochemistry and Leuven Food Science and Nutrition Research Centre (LFoRCe), KU Leuven, B-3001 Leuven, Belgium
    Search for more papers by this author
  • Jan A. Delcour

    1. Laboratory of Food Chemistry and Biochemistry and Leuven Food Science and Nutrition Research Centre (LFoRCe), KU Leuven, B-3001 Leuven, Belgium
    Search for more papers by this author

Direct inquiries to author Pauly (E-mail: anneleen.pauly@biw.kuleuven.be).


Wheat hardness has major consequences for the entire wheat supply chain from breeders and millers over manufacturers to, finally, consumers of wheat-based products. Indeed, differences in hardness among Triticum aestivum L. or between T. aestivum L. and T. turgidum L. ssp. durum wheat cultivars determine not only their milling properties, but also the properties of flour or semolina endosperm particles, their preferential use in cereal-based applications, and the quality of the latter. Although the mechanism causing differences in wheat hardness has been subject of research more than once, it is still not completely understood. It is widely accepted that differences in wheat hardness originate from differences in the interaction between the starch granules and the endosperm protein matrix in the kernel. This interaction seems impacted by the presence of either puroindoline a and/or b, polar lipids on the starch granule surface, or by a combination of both. We focus here on wheat hardness and its relation to the presence of puroindolines and polar lipids. More in particular, the structure, properties, and genetics of puroindolines and their interactions with polar lipids are critically discussed as is their possible role in wheat hardness. We also address future research needs as well as the presence of puroindoline-type proteins in other cereals.

Introduction: The Importance of Wheat Hardness

Wheat ranks 3rd behind maize (corn) and rice only in terms of worldwide production (FAOSTAT 2010). In the diet of over 35% of the world's population, it is an important source of starch and protein. It owes its popularity to its ability to grow under a wide range of soil, climate, and environmental conditions as well as to the unique properties of its storage proteins.

Wheat hardness, defined as the force needed to crush the kernels, is, next to its protein content, a major quality trait. Bread wheat (Triticum aestivum L.) endosperm texture ranges from very soft to hard, whereas durum wheat (T. turgidum L. ssp. durum) has the hardest kernels of all wheat cultivars (cvs.). The “soft” and “hard” wheat terminology for T. aestivum cvs. is used in North America. Also in North America, the term durum wheat refers to T. turgidum ssp. durum. In contrast, in some European countries, the terms “soft” and “hard” wheat refer to T. aestivum and T. turgidum ssp. durum cvs., respectively (Delcour and Hoseney 2010). Here, we use the North American terminology.

Endosperm texture is of interest to every link in the wheat supply chain. The farmer generally makes more profit with harder textured wheat (Turnbull and Rahman 2002; USDA-ERS 2012), due to the higher protein content of the latter as discussed in an accompanying paper (Pauly and others 2013). To the miller, endosperm texture in a pronounced way affects wheat tempering (Delcour and Hoseney 2010), energy requirements, and technical settings of the mill (Posner 2000) as well flour yield. Soft wheat generally yields flour with smaller average particle size and lower levels of damaged starch (and thus lower water absorption) than hard wheat. Durum wheat is usually milled into semolina. Semolina has larger average particle size, up to 500 μm (Dalbon and others 1996), than flour derived from T. aestivum cvs., average 25 to 35 μm for soft and 50 to 70 μm for hard wheat cvs. (Posner 2009). Differences in endosperm texture impact flour quality and, thus, the preferential use of any given flour. In general, soft wheat flour is used for cake and cookie, hard wheat flour for bread, and durum wheat semolina for pasta. The impact of wheat hardness on end-product quality is discussed in an accompanying paper (Pauly and others 2013).

For the purposes of this paper, it is important to distinguish wheat hardness from vitreousness. The latter term describes the glassy appearance of some wheat kernels. While Morris and Beecher (2012) recently provided evidence for a genetic component in vitreousness, it can occur in all cvs., especially when grown with high nitrogen fertilization levels and at high temperature. The molecular basis for vitreousness is not known. Vitreous kernels generally contain more protein and their interior is more tightly packed than that of other kernels. The absence of air spaces accounts for the vitreous appearance (Delcour and Hoseney 2010).

The Determinants: Puroindolines and Wheat Polar Lipids

Wheat endosperm hardness is strongly genetically controlled, although moisture content and environmental conditions can impact it as well (Turnbull and Rahman 2002). Symes (1965) demonstrated the simple inheritance of grain hardness, with one major and several minor genes responsible for differences in hardness. The major gene is located at the distal end of the short arm of chromosome 5D and was named Hardness (Ha) locus, with the soft phenotype as the dominant trait (Doekes and Belderok 1976).

The biochemical basis of wheat hardness remains largely unknown. Based on micropenetrometry measurements, Barlow and others (1973) demonstrated that gluten isolated from wheat cvs. with varying hardness has similar hardness readings. In the same way, starches isolated from wheat cvs. with varying hardness also have similar hardness. Several authors (Barlow and others 1973; Hoseney and Seib 1973; Simmonds and others 1973) suggested that the difference in endosperm hardness between hard and soft wheat results from differences in adhesion between starch and gluten proteins. Evidence for differences in adhesion was based on scanning electron microscopy observations that more gluten adheres to the surface of starch granules isolated from hard wheat cvs. than to that of starch isolated from soft wheat cvs. (Barlow and others 1973; Hoseney and Seib 1973). Simmonds and others (1973) suggested that specific compounds at the starch granule–protein matrix interface can interfere with adhesion between the two. Later, Greenwell and Schofield (1986) discovered that endosperm hardness is strongly related to the occurrence on the surface of water-washed starch granules of a group of proteins, which they called friabilin. High levels of friabilin were associated with water-washed soft wheat starch granules, and much less friabilin was present at water-washed hard wheat starch granules. No friabilin was detected on water-washed durum wheat starch granules. Initially, friabilin was believed to act as “nonstick” protein preventing strong adhesion between starch and gluten proteins (Greenwell and Schofield 1986). However, it is now clear that the mechanism determining wheat hardness is rather complex. In the beginning of the 1990s, several authors found friabilin to encompass a mixture of proteins (Jolly and others 1993; Morris and others 1994; Rahman and others 1994; Oda and Schofield 1997). N-terminal amino acid sequencing revealed the basic cysteine-rich proteins puroindoline a (PINA) and b (PINB) to be major components of friabilin. The name “puroindoline” is derived from the Greek word “πυρος” (“puros”), which means “wheat,” and from “indoline,” referring to their unique tryptophan-rich (Trp-rich) domain (see “Structure” section). Other components are Grain Softness Protein-1 and some α-amylase inhibitors (Morris and others 1994; Oda and Schofield 1997).

The level of polar lipids on the surface of water-washed starch granules seems to follow the same pattern as that of friabilin, in that high and low levels of polar lipids are present on the starch granule surface in soft and hard wheat, respectively (Greenblatt and others 1995). This suggests the involvement of wheat polar lipids in the mechanism dictating kernel texture. The lipids associated with the starch granule surface mainly originate from the amyloplast bilayer lipid membrane surrounding the amyloplast in which starch is synthesized and stored during kernel development. During seed desiccation, the amyloplast lipid membrane is degraded (Tan and Morrison 1979). Starch granules in mature wheat kernels are, hence, amyloplast remnants on which remainders of the desiccated membrane are found (Barlow and others 1973). According to Feiz and others (2009b), the composition of bound polar lipids, namely lipids extracted with a polar solvent after extracting lipids with a nonpolar solvent, in wheat resembles that of amyloplast membranes from potato tubers in that it contains higher levels of monogalactosyldiacylglycerols (MGDG), digalactosyldiacylglycerols (DGDG), and phosphatidylcholine (PC) than other lipid membranes (Fishwick and Wright 1980).

Evolution of the Ha Locus in Wheat

During wheat evolution, several genomic rearrangements resulted in the presence of a D-genome (which carries the Ha locus) in T. aestivum, but not in T. turgidum ssp. durum. The origin of wheat and the major hybridization events along its evolution probably occurred in the Middle Eastern region about 2.5 to 6 million years ago (Wrigley 2009; Matsuoka 2011). The tetraploid T. turgidum ssp. durum (AABB) gradually derived from natural hybridization of the diploids T. urartu (AA) and the ancestral Aegilops speltoides (BB) (Chantret and others 2005; Matsuoka 2011). After the polyploidization process, the Ha locus was eliminated from both the A and B genomes (Chantret and others 2005). The hexaploid T. aestivum (AABBDD) originated from an additional polyploidization event between tetraploid T. turgidum ssp. dicoccum (AABB) and diploid A. tauschii (DD; Wrigley 2009), which restored the Ha locus in T. aestivum (Chantret and others 2005).

From the above, it seems that PINA and PINB lie at the molecular basis of wheat hardness. In what follows, the genetics of puroindolines (PINs), their structure, occurrence, and possible biological function are discussed.

Wheat Puroindolines: Genetics, Structure, Occurrence, and Biological Function


The Puroindoline a (Pina-D1) and Puroindoline b (Pinb-D1) genes encoding PIN proteins are part of the Ha locus on the short arm of chromosome 5D (Jolly and others 1993; Sourdille and others 1996; Ragupathy and Cloutier 2008). Their coding regions contain 450 base pairs (bp) without introns. The 2 genes are approximately 70% identical in the coding regions, but only about 50% in the 3’-untranslated region (Gautier and others 1994). Next to the Pina-D1 and Pinb-D1 genes, the Ha locus contains the Gsp-1 gene (Turnbull and others 2003b). Tranquilli and others (2002) demonstrated that deletions or allelic variations of the Gsp-1 gene do not cause significant differences in grain hardness. In contrast to Pina-D1 and Pinb-D1, which only reside at the D-genome, Gsp-1 is present on all 3 wheat genomes (A, B, and D) (Morris 2002). The Pina-D1, Pinb-D1, and Gsp-1 genes are located in a 60-kbp DNA fragment of the Ha locus of A. taushii, the D-genome donor of T. aestivum (see “Evolution of the Ha locus in wheat” section). The Pina-D1 gene is situated between the Pinb-D1 and Gsp-1 genes at a distance of maximum 30 kbp from either gene (Turnbull and others 2003b). Because T. turgidum L. ssp. durum lacks the D-genome, PINs are not expressed, resulting in the typical very hard endosperm texture (Morris 2002).

The presence of both PIN wild-type sequences, designated as Pina-D1a and Pinb-D1a, results in soft endosperm texture. A mutation in either Pina-D1 or Pinb-D1 yields hard endosperm (Table 1). The most common hardness-associated mutation in Pina-D1 is a null mutation (Pina-D1b). It results in the absence of Pina transcripts and, as a result, of PINA proteins in the kernel (Giroux and Morris 1998). A deletion of thousands of bp, including the first 21 bp of Pina-D1, has been found to be responsible for the Pina-D1b mutation (Ragupathy and Cloutier 2008). Giroux and Morris (1997) were the first to report a single nucleotide mutation in Pinb-D1 leading to a glycine-to-serine change at position 46 in PINB (G46S; Pinb-D1b; Figure 1). Since then, many other Pinb-D1 mutations have been discovered in wheat. Lillemo and Morris (2000) found 2 frequently occurring point mutations in Pinb-D1 in Northern European wheat cvs. The first one involves a leucine-to-proline alteration at position 60 (L60P; Pinb-D1c; Figure 1), while the latter has a tryptophan instead of an arginine at position 44 (W44R; Pinb-D1d; Figure 1). Morris and others (2001) described the occurrence of 3 new Pinb alleles in North American wheat, all resulting in an early stop codon: nonsense mutations of tryptophan at position 39 and 44 (respectively, Pinb-D1e and Pinb-D1f) and of cysteine at position 56 (Pinb-D1g; Figure 1). More recently, a whole range of other Pin alleles have been reported, many of them in Chinese cvs. (Xia and others 2005; Chen and others 2006; He and others 2007). The presence/absence and sequence polymorphism of Pina-D1 and Pinb-D1 genes has been excellently reviewed by Morris and Bhave (2008).

Table 1. Wheat (Triticum aestivum L.) kernel texture, puroindoline a (PINA), and b (PINB) allele designations and molecular changes at DNA and protein levels, with the point mutations at DNA level of PINB indicated in bold and underlined. Only the 7 most frequently occurring mutations in European and North American wheat cultivars are listed. For a complete overview of mutations in PINA or PINB, the reader is referred to Morris and Bhave (2008)
 Allele designationMolecular change
Kernel texturePINAPINBDNA levelProtein level
  1. a

    Ragupathy and Cloutier (2008).

  2. b

    Giroux and Morris (1998).

  3. c

    Giroux and Morris (1997).

  4. d

    Lillemo and Morris (2000).

  5. e

    Morris and others (2001).

HardPina-D1bPinb-D1aDeletion of first 21 base pairsaNo PINA presentb
HardPina-D1aPinb-D1bGGC → AGCGlycine to serine at position 46 (G46S)c
HardPina-D1aPinb-D1cCTG → CCGLeucine to proline at position 60 (L60P)d
HardPina-D1aPinb-D1dTGG → AGGTryptophan to arginine at position 44 (W44R)d
HardPina-D1aPinb-D1eTGG → TGATryptophan to early stop codon at position 39e
HardPina-D1aPinb-D1fTGG → TGATryptophan to early stop codon at position 44e
HardPina-D1aPinb-D1gTGC → TGACysteine to early stop codon at position 56e
Figure 1.

The 6 most common mutations in PINB in European and North American wheat cvs. The tryptophan-rich domain is underlined. Point mutations resulting in an early stop codon (at positions 39, 44, and 56) result in a mature protein with about 1/3 of the length of wild-type PINB or degradation of the mRNA before it is translated by the nonsense-mediated mRNA decay pathway (Amrani and others 2006). In the W44R mutation, a tryptophan residue is replaced by a charged arginine residue, while in the G46S mutation, a more polar serine residue replaces a glycine residue. The L60P mutation, relatively far from the tryptophan-rich domain, has a leucine residue replaced by a proline residue. The latter confers rigidity to the protein backbone. Amino acids are represented by their letter abbreviations.

Studies using genetically modified crops and wheat breeding programs have extended our knowledge on the effect of Pina-D1 and Pinb-D1 on endosperm hardness. Beecher and others (2002a) expressed the wild-type Pinb-D1a sequence in the hard wheat cv. Hi-Line containing the G46S mutation (Pinb-D1b). The resulting transgenic kernels were softer, showed increased PIN levels, and had lower damaged starch levels after milling than the parent (Hi-Line hard wheat) kernels. In a similar study by Hogg and others (2004), the same hard wheat was complemented with wild-type Pina-D1a, Pinb-D1a, or both Pina-D1a and Pinb-D1a. All transformed wheat lines were of reduced grain hardness, with the most pronounced decrease when both functional Pina-D1a and Pinb-D1b had been expressed. Swan and others (2006) and Wanjugi and others (2007) reported similar findings and added that in order to have soft endosperm texture, the total PIN level is less important than the presence of both functional PINA and PINB. Xia and others (2008) demonstrated that silencing of Pina-D1a results in significantly harder kernels. Gazza and others (2008) showed that the introduction of Pina-D1a and Pinb-D1a in durum wheat reduces the single kernel characterization system (SKCS) hardness value by 60%. More recently, Morris and others (2011) transferred Pin genes from the soft wheat cv. Chinese Spring through ph1b-mediated homoeologous recombination into durum wheat. The SKCS hardness values of the developed soft durum and its milling behavior, damaged starch level, and particle size distribution after milling were similar to those of soft wheat. In contrast to the soft durum wheat produced by Gazza and others (2008), these durum lines are stable and can be used in crossing and transferring the soft endosperm texture to other durum wheat cvs.

Since rice (Oryza sativa L.) and maize (Zea mays L.) contain no PINs or homologous proteins (see “Other Cereals” section) and generally have hard texture, these crops are well suited to study the impact of Pina-D1 and Pinb-D1 genes. Krishnamurthy and Giroux (2001) demonstrated that expression of Pina-D1a or Pinb-D1a reduces grain hardness of rice seeds, which, when milling, decreases both the levels of damaged starch and the average particle size. The observation of a softer rice seed texture after Pinb-D1a gene expression was confirmed by Wada and others (2010). Similar results were reported for transgenic maize following the expression of Pina-D1a and Pinb-D1a (Zhang and others 2009).


PINA (120 amino acids) and PINB (119 amino acids) have similar molecular mass (about 13 kDa). Both are basic proteins with a calculated isoelectric point between 10.5 and 11, and they exhibit about 55% amino acid homology (Gautier and others 1994; Branlard and others 2003). Comparison of the deduced primary structure of PINs with the primary structure of mature PINA and the N-terminal sequence of the mature PINB allowed concluding that PINs are synthesized as preproproteins containing a signal peptide (for protein-targeting) and both a N-terminal and a C-terminal cleavable polypeptide (Gautier and others 1994). The microheterogeneity observed between both PINs at the N- and C-termini can result from slightly different post-translational processing pathways (Gautier and others 1994; Day and others 2006).

Both PINs possess a Trp-rich domain, which is unique among proteins. In PINA, 5 tryptophan residues are located in this domain (WRWWKWWK; positions 38 to 45), which, in PINB, is truncated and contains only 3 tryptophan residues (WPTKWWK; positions 39 to 45; Blochet and others 1993; Gautier and others 1994). This hydrophobic region confers strong affinity for polar lipids to PINs (Kooijman and others 1997; Douliez and others 2000). Section “Interactions between PINs and polar lipids” discusses the lipid-binding capacity of PINs in more detail. Infrared (IR) studies demonstrated that PINA and PINB have a similar secondary structure. At pH 7, they consist of approximately 30% α-helices, 30% β-sheets, and 40% unordered structures (Le Bihan and others 1996). This similarity in secondary structure has been confirmed by Kooijman and others (1997) based on far-UV circular dichroism (CD) experiments.

Both PINA and PINB contain 10 cysteine residues (Figure 2A), which form 5 intramolecular disulfide bonds. Eight of these residues form a specific pattern known as the “eight-cysteine motif” (8CM; including a CC and CXC sequence, with X = arginine for both PINs). The 2 additional cysteine residues are located at the beginning and the end of the Trp-rich domain. The 8CM is widely distributed among plant proteins. Proteins having this conserved motif show a wide range of functions in storage, plant defense, enzyme inhibition, lipid transfer, and cell wall structure. It is thought that the 4 disulfide bonds confer stability to the three-dimensional (3D) structure (Jose-Estanyol and others 2004). Le Bihan and others (1996) demonstrated that the disulfide bonds are indeed necessary to stabilize the α-helical structure and are essential to maintain the native structure and solubility of PINA and PINB. However, although PINs contain 5 disulfide bonds, their conformation changes slightly with pH. Indeed, at pH 4, the α-helix level is about 8% higher than at pH 7 for both PINs. At pH 10, the PINB conformation is similar to that at pH 7, while PINA again shows more α-helices at the higher pH. Kooijman and others (1997) suggested that the Trp-rich domain of PINs also stabilizes the structure.

Figure 2.

Schematic representation of models of the 3D structure of PINs based on the folding pattern of nonspecific lipid transfer proteins (ns-LTPs; A, B; Marion and others 1994) and 2S storage proteins of dicotyledonous plants (C; Lesage and others 2011). In both models, the structure of PINs consists of 4 α-helices (indicated with math formula in A) connected by flexible loops. In the model based on the structure of ns-LTPs (B; Shewry, personal communication), all 10 cysteine residues are connected by disulfide bonds, while in the model based on the folding pattern of 2S storage proteins (C), only 4 cysteine residues are present as disulfide bonds. The unique tryptophan-rich domain (TrpRD) is located at the exterior of the protein, between the first and second α-helix (Le Bihan and others 1996; Douliez and others 2000).

All 8CM proteins likely share a similar tertiary structure (based on the 3D structures known so far) containing 4 α-helices and variable loops (Jose-Estanyol and others 2004). The 3D structure of PINs has not been unraveled. A first model was put forward based on the folding pattern of the 8CM containing nonspecific lipid transfer proteins (ns-LTPs; Figure 2B; Marion and others 1994). These small proteins (about 9 kDa) share not only the 8CM but also several properties with PINs. First, Marion and others (1994) found about 30% amino acid homology between PINs and wheat ns-LTPs. The least homology was found in the region corresponding to the Trp-rich domain of PINs. This is logical since ns-LTPs do not contain this region. In addition, Fourier transform IR spectroscopy demonstrated similar levels of α-helices in PINs and ns-LTPs (Marion and others 1994). For 8CM proteins, it has been proposed that the variable loops connecting the α-helices are the functional regions of the proteins (Jose-Estanyol and others 2004). This also seems true for PINs, since the Trp-rich domain of PINs is located in a loop between the first and second α-helix at the outside of the protein in the 3D model based on ns-LTPs (Marion and others 1994; Kooijman and others 1997; Douliez and others 2000). Like ns-LTPs and other 8CM proteins, PINs possess a conserved tyrosine residue (positions 23 and 24 for PINA and PINB, respectively) in the first α-helix, which may be functionally important (Poteete and others 1992; Le Bihan and others 1996).

Recently, Lesage and others (2011) proposed a different 3D structure model of PINs using iterative threading assembly refinement. More homologies with 2S storage proteins of dicotyledonous plant species than with ns-LTPs were revealed (Figure 2C). These 2S storage proteins also contain the 8CM. The predictions resulted in better C-scores than the earlier model based on wheat ns-LTPs. In the recent 2S storage protein-based model, not all 10 cysteine residues are involved in disulfide bonds: only 4 of them readily form disulfide bonds (C20/C55 and C56/C104). The 6 remaining cysteine residues are predicted to be located too far from each other to form stable disulfide bonds. Table 2 summarizes structural similarities of and dissimilarities between PINs, ns-LTPs, and 2S storage proteins of dicotyledonous plants, while Figure 3 presents the amino acid alignment of these proteins. The cysteine residues of the 8CM are the only conserved residues among the 6 proteins. However, still some other amino acid residues with similar characteristics occur in the sequences. This may indicate that they are required for proper folding and/or functioning. In several proteins, conserved amino acid residues are functionally more important than nonconserved residues (Poteete and others 1992).

Table 2. Structural similarities and dissimilarities between puroindoline (PINs), nonspecific lipid transfer proteins (ns-LTPs), and 2S storage proteins of dicots
 PINsns-LTPs2S storage proteins
  1. a

    Le Bihan and others (1996) and Kooijman and others (1997).

  2. b

    Charvolin and others (1999) and Douliez and others (2000).

  3. c

    Lesage and others (2011); the templates used for predicting the 3D structure of PINs were 1psy (a 2S albumin storage protein of Ricinus communis), 1s6d (a methionine-rich albumin from sunflower seed), and 1w2q (a 2S storage protein type of peanut seeds). The 3D structures of these proteins are found in Pantoja-Uceda and others (2003), Pantoja-Uceda and others (2004), and Lehmann and others (2006), respectively.

  4. d

    Eight-cysteine motif.

Molecular weight (kDa)ca. 127–912–15
Secondary structureRich (ca. 30%) in α-helices (based on CD and IR spectroscopy studiesa)4 α-helices (based on 3D structureb)5 α-helices (based on 3D structurec)
 8CMd + 2 additional cysteine residues8CMd8CMd
 ?4 disulfide bonds4 disulfide bonds
Other structural features?Hydrophobic cavity/
 Tryptophan-rich domain//
 1 polypeptide chain1 polypeptide chain2 polypeptide chains connected by 2 disulfide bonds
Figure 3.

Sequence alignment (ClustalW2) of PINA, PINB, wheat ns-LTP, and selected 2S storage proteins of dicotyledonous plants as based on Lesage and others (2011). Abbreviations refer to the structures in the RCSB Protein Data Bank (RCSB PDB) (Berman and others 2000). Amino acids are represented by their letter abbreviations. The unique tryptophan-rich domain of PINs is underlined. The conserved 8-cysteine motif (8CM) is indicated in boxes and the symbols above indicate which cysteine residues are connected by disulfide bonds. The cysteine residues in PINA and PINB at the beginning and the end of the tryptophan-rich domain form a disulfide bond and are indicated with a gray star. The α-helices are shaded: the exact locations of the α-helices in wheat ns-LTP and 2S storage proteins are based on their 3D structure (Charvolin and others 1999; Douliez and others 2000), while the predicted locations of the α-helices in PINA and PINB are based on Le Bihan and others (1996).

Neither of the above 2 models contains β-sheets. This contradicts the IR spectroscopy (Marion and others 1994; Le Bihan and others 1996) and far-UV CD (Kooijman and others 1997) data, which revealed the presence of approximately 30% β-sheets. However, Le Bihan and others (1996) suggested that some protein regions that, based on IR spectroscopy, are interpreted to be β-sheets are not observed as such by nuclear magnetic resonance (NMR). The same phenomenon has been described for wheat ns-LTPs (Désormeaux and others 1992). In the case of far-UV CD of proteins containing clusters of aromatic amino acid residues such as tryptophan, such residues can significantly perturb the spectrum, making correct secondary structure prediction difficult (Woody 1994; Sreerama and others 1999). This probably causes the discrepancy between secondary structure predictions and the proposed 3D models.

In both models, most uncertainty exists about the region of the Trp-rich domain of PINs, because a similar domain has not yet been found in other proteins (Douliez and others 2000; Lesage and others 2011). Jing and others (2003) determined the structure of a fragment of PINA (puroA; FPVTWRWWKWWKG) that is slightly larger than its Trp-rich domain. Their NMR data suggest that puroA is rather unstructured in aqueous solutions, but forms an amphipathic, helical structure when present in sodium dodecyl sulfate micelles. Although it is not known how representative the structure of puroA is for the full-length PINA, Kooijman and others (1997) with far-UV CD observed a slight increase in α-helix level of PINA upon binding of the synthetic molecule alkylphosphocholine, which has a structure very similar to that of lyso-PC (LPC). Random coil-to-α-helix conformational changes upon interaction with membranes have been observed for many proteins. The formation of α-helices at membrane surfaces considerably reduces the free energy and is thus an important driving force for membrane binding (Seelig 2004). In contrast to what is the case for PINA, PINB did not show an increase in α-helix level (as calculated from far-UV CD spectra) when similar amounts of phosphocholine were added (Kooijman and others 1997), presumably because of lower levels of tryptophan and charged amino acid residues in its Trp-rich domain.

Although PINs exhibit several features of membrane proteins such as high levels of α-helices and the need to use detergents to extract them (Blochet and others 1993), neither of the proteins contains a typical transmembrane region as predicted from their amino acid sequences (Hofmann and Stoffel 1993).

Le Bihan and others (1996) observed that PINA, but not PINB, forms aggregates under acidic and high ionic strength conditions and at low temperatures, an observation recently confirmed by Clifton and others (2011). PINA spontaneously forms monodisperse prolate ellipsoidal micelles in solution. Such single micelle consists of 38 PINA molecules and is stable over a wide pH and temperature range. The self-assembly of PINA micelles in solution is likely driven by intermolecular hydrophobic forces between tryptophan residues in the Trp-rich domain of the proteins. Although PINB shows more than 50% amino acid homology with PINA and both proteins have been suggested to have similar 3D structures, no such self-assembly into micelles has been observed. This probably results from the truncated PINB Trp-rich domain with less tryptophan residues, which makes this region less hydrophobic, resulting in weaker intermolecular hydrophobic interactions (Clifton and others 2011).


Spatial distribution

Several studies have focused on the (sub)cellular location of PINs. There is no consensus on whether the spatial distribution of PINA and PINB in the kernel is the same (Capparelli and others 2005; Feiz and others 2009b) or differs (Dubreil and others 1998). However, elucidation of the location of both PINA and PINB would help to understand their role in endosperm hardness. In most studies, PINs were found to occur in the endosperm and/or the aleurone layer, but not in roots and hypocotyls of seedlings (Dubreil and others 1998). In soft wheat endosperm, PINA and possibly PINB occur in the protein matrix and around starch granules surrounded by the protein matrix (Dubreil and others 1998). PINB also occurs in the aleurone layer in small inclusion bodies (Dubreil and others 1998). According to Capparelli and others (2005), both PINs are colocalized in the starchy endosperm, but occur to a greater extent in the aleurone layer and its intercellular space. Recently, Feiz and others (2009b) confirmed their localization at the starch granule surface using immunofluorescent studies. That PINs are present in the starchy endosperm is not very surprising since they are the main components of friabilin, which is associated with water-washed starch granules (Greenwell and Schofield 1986). Keeping in mind that the difference in hardness between hard and soft wheat cvs. probably results from different adhesion phenomena between starch and gluten proteins (Barlow and others 1973), the starch surface granule seems the most probable site where PINs exert their effect on wheat hardness. Section “The Mechanism Determining Wheat Hardness” reviews the binding of PINs to the starch granule surface. Still, the occurrence of PINA and PINB in the aleurone seems to be consistent with their presumed antimicrobial properties (see “Biological function” section). The aleurone is a peripheral cell layer of the wheat kernel, which is first encountered by pathogens during grain development (Capparelli and others 2005). However, since both PINs possess signal sequences both at their N- and C-termini, which are possibly involved in protein targeting, one has argued that PINs may be synthetized in the aleurone and subsequently transported to the starchy endosperm (Gautier and others 1994).

However, Wiley and others (2007) showed that Pin genes are exclusively expressed in the endosperm. Lesage and others (2011) found PINs in protein bodies during endosperm development. In mature seeds, PINs were detected in the protein matrix and in some cases at the starch granule surface, but not in the amyloplast membrane. That PINs are localized in protein bodies suggests the same intracellular targeting as that of prolamins. The latter are synthesized on ribosomes bound to the endoplasmatic reticulum and then transported to the vacuoles where they are deposited and form protein bodies (Shewry and Halford 2002; Tosi 2012). PINs contain a signal peptide similar to that of prolamins. It consists of 20 amino acids in total, with an internal stretch of at least 6 hydrophobic amino acids preceded by a short positively charged sequence (Tosi 2012). Finally, Lesage and others (2011) and Meziani and others (2012) did not detect any PINs in the aleurone layer, in contrast to what was found in the aforementioned studies (Dubreil and others 1998; Capparelli and others 2005).

Temporal distribution

Although there are different views on the location of PINs, there is a consensus on their occurrence during grain development. The small variations between different studies are likely explained by varying growth conditions, making comparisons between number of days after flowering (DAF) difficult. Gautier and others (1994) reported on accumulation of Pina-D1 and Pinb-D1 transcripts between 8 and 12 DAF followed by a significant increase between 15 and 18 DAF. The highest level of transcripts was reached between 26 and 33 DAF, followed by a fast decrease. Similar results were obtained by Hogg and others (2004). Amoroso and others (2004) found 2 different expression patterns: in the first, transcript levels peaked between 19 and 26 DAF and decreased between 26 and 33 DAF, while, in the second, transcript levels remained constant between 13 and 33 DAF. For most cvs., the second pattern described the expression pattern of Pina-D1 transcripts, whereas Pinb-D1 transcripts followed either of the 2 patterns with the same frequency.

Figure 4 shows a schematic overview of PIN protein accumulation in relation to grain development together with the accumulation of wheat's main constituents, starch and gluten protein. Starch and gluten show similar patterns of accumulation, with their synthesis initiating at about 10 DAF, accumulating rapidly until about 21 DAF, slowing down by 35 DAF, and being completed by 42 DAF (Shewry and others 2012). Turnbull and others (2003a) detected low levels of PINA and PINB in a soft cv. at the end of endosperm cellularization (respectively, at 10 and 15 DAF). Both PINA and PINB levels increase during the grain filling and maturation stages (until about 40 DAF), after which their increase slows down (Dubreil and others 1998; Turnbull and others 2003a). In a hard PINA null cv., PINB follows the same trend (increase from 15 until 32 DAF), but lower levels are obtained in the end (Turnbull and others 2003a). Similar trends, but lower levels, were reported by Lesage and others (2011). PINs are degraded during seed germination (Dubreil and others 1998).

Figure 4.

Evolution of PIN levels during grain development of a soft wheat cultivar as based on Dubreil and others (1998), Turnbull and others (2003a), and Lesage and others (2011), and the accumulation of the endosperm main constituents starch and gluten (Shewry and others 2012). Bechtel and others (1996) and Turnbull and others (2003a) described that soft and hard cultivars are distinguishable from 15 d after flowering (DAF) onward.

Biological function

Although PINs presumably form the biochemical basis of wheat hardness, it seems logical that the reason for their presence in the wheat kernel has nothing to do with their effects on kernel texture. The biological function of PINs remains largely unknown. A role in plant defense has been suggested, based on the observation that purified PINs display in vitro antifungal (Dubreil and others 1998) and antibacterial (Capparelli and others 2005; Capparelli and others 2006) properties. PINs share properties with other antimicrobial peptides, such as a low molecular weight, a compact structure with disulfide bonds, and a (relatively) high tryptophan level (Schibli and others 2002; Chan and others 2006). PINB has higher activity against a range of fungi than PINA (Dubreil and others 1998), but this may be fungal strain-dependent. Both PINs show similar activity against bacteria and seem to act in synergy (Capparelli and others 2005). The Trp-rich domain of PINs has been suggested to be responsible for their antibacterial and antifungal activities (Jing and others 2003; Philips and others 2011; Miao and others 2012), since it shows high affinity for polar lipids (see “Interactions between PINs and polar lipids” section), which are the building blocks of bacterial and fungal membranes. PINs are thought to interact with these membrane lipids and to disturb the regular membrane bilayer structure. The Trp-rich domain contains relatively high levels of tryptophan and arginine, which are common residues in potent antimicrobial peptides (Chan and others 2006). A synthetic peptide corresponding to the Trp-rich domain of PINA has a much higher antibacterial activity than an analogous peptide corresponding to the Trp-rich domain of PINB, which likely can be explained in terms of the number of tryptophan and positively charged residues (Jing and others 2003; Philips and others 2011). Evrard and others (2008) showed that, for the mature full-length PINA, 2 tryptophan residues in the Trp-rich domain (on position 41 and 44) are essential for interacting with the membrane of Saccharomyces cerevisiae, while, in the Trp-rich domain of PINB, where lysine residues seem to be more important, none of the tryptophan residues is necessary for interaction. This, combined with the fact that the antimicrobial activity of the short synthetic peptides is not in line with that of the mature, full-length proteins PINA and PINB (Capparelli and others 2005), likely indicates that also other regions and/or protein folding are important for their antimicrobial properties. Taken together, the antimicrobial activity of PINs is related to their ability to interact with cellular membranes, but the underlying mechanism still needs to be further unraveled. In this light, Charnet and others (2003) proposed that PINs exert their antimicrobial activity by forming ion channels in membranes. Other more indirect evidence for the antimicrobial activity of PINs is that transgenic crops expressing both PINs show higher resistance against plant pathogens than their nontransgenic counterparts. This has been demonstrated for rice (Krishnamurthy and others 2001), apple (Faize and others 2004), and tetraploid (Luo and others 2008) and hexaploid (Kim and others 2012) wheat.

Additionally, very recently, Lesage and others (2012) proposed a role for PINs in the storage protein folding machinery. Comparison of the proteomes of 2 wheat near-isogenic lines with different Pina genotype and, hence, different hardness, revealed that the hard line contains more stress-related and folding proteins during grain development. Furthermore, endosperm development is completed earlier in the hard than in the soft line. This, combined with their earlier findings that PINs and storage proteins follow the same targeting in the grain (see “Occurrence” section; Lesage and others 2011), Lesage and others (2012) suggested that the higher oxidative stress when lower levels of PINs are present results in higher storage protein aggregation, which in turn results in harder wheat kernels.

Wheat Polar Lipids

Wheat kernels typically contain 2.5% to 3.3% lipids, of which 30% to 36% are found in the germ, 25% to 29% in the aleurone, and 35% to 45% in the endosperm (Morrison 1978; Hargin and Morrison 1980). Wheat contains the same typical plant lipid classes that other cereals also do (Morrison 1978). Nonpolar lipids are primarily present in the germ and aleurone, while almost all polar lipids occur in the endosperm as remnants of amyloplast and other membranes (Hargin and Morrison 1980). The main glycolipid classes are DGDG and MGDG, while major phospholipid classes are PC, phosphatidylethano-lamine (PE), LPC, lysophosphatidylethanolamine (LPE), and N-acylphosphatidylethanolamine (NAPE; Hargin and Morrison 1980; Finnie and others 2009). The lyso-polar lipids LPC and LPE are predominantly present as starch internal lipids (Finnie and others 2009). As a result of milling, the polar lipids in the endosperm end up in flour, together with a portion of the germ and aleurone nonpolar lipids (Morrison and Hargin 1981). The research area of plant lipid genetics and biosynthesis is very complex (Ohlrogge and others 1991; Ohlrogge and Browse 1995; Dörmann and Benning 2002) and out of the scope of this review, but it is worthwhile to mention that the polar lipid levels are regulated by 2 genes (Fpl-1 and Fpl-2), of which at least 1 (Fpl-2) seems to be allelic to or closely linked to the Ha locus (Morrison and others 1989). For more information on wheat lipids and their classification, the interested reader is referred to Chung and others (2009) and Pareyt and others (2011).

Interactions between PINs and Polar Lipids

PINs display strong affinity toward polar lipids, and their biological (plant defense) and technological (wheat hardness) functions are likely related to this lipid-binding capacity. The role of PINs in plant defense seems to result from their possibility to interact with bacterial membranes (see “Biological function” section). Their role in wheat hardness is largely related to their ability to stabilize the amyloplast membrane during grain development (see “The Mechanism Determining Wheat Hardness” section).

Several studies emphasized the importance of the Trp-rich domain in lipid binding. Its tryptophan residues allow intrinsic fluorescence measurements that are valuable for investigating the binding affinity of PINs for lipids and their specificity. Kooijman and others (1997) and Le Guernevé and others (1998) showed that the conformation of the Trp-rich domain of PINs changes upon lipid binding. The tryptophan residues become involved in hydrophobic interactions with the alkyl chains of the lipids. Since the truncated Trp-rich domain of PINB contains only 3 tryptophan residues, its hydrophobic interactions are weaker than those of PINA, which contains 5 such residues. A minimum length of the aliphatic chain of the lipid is required for proper binding (Husband and others 1994; Kooijman and others 1997). Next to hydrophobic interactions, there are electrostatic interactions between positively charged residues in the Trp-rich domain (R, K) and the charged head groups of polar lipids. Therefore, PINs can interact more strongly with anionic phospholipids than with neutral polar lipids such as galactolipids. Since PINB has a higher net positive charge than PINA, stronger electrostatic forces between lipids and PINB may exist than between lipids and PINA. In the case of zwitterionic lipids, such as PC and PE, a higher electrostatic repulsion occurs between the positively charged head group of the lipid and the positively charged residues in the Trp-rich domain, resulting in lower affinity of PINB than of PINA for zwitterionic lipids. However, high ionic strengths can shield positive charges on lipids as well as on amino acids in the Trp-rich domain and increase the affinity of PINs for zwitterionic lipids (Dubreil and others 1997; Kooijman and others 1997; Le Guernevé and others 1998; Bottier and others 2008). Furthermore, interaction of PINs with glycolipids can be sterically hindered when the hexose residues (mainly galactose) of the glycolipids interact with each other (Dubreil and others 1997). Lipid binding may further be enhanced by a small hydrophobic phenylalanine-rich region in PINs. In the 3D structure model based on that of ns-LTPs, this region is located close to the functionally important Trp-rich domain (Kooijman and others 1997). Although PINs have high affinity for polar lipids, there is no agreement on the requirements of lipid organization for lipid binding. According to Kooijman and others (1997), the organization of lipids into micelles is not essential, but the authors did not exclude that micelle formation would facilitate lipid binding. Husband and others (1994), on the contrary, stated that PINs can only bind with lipids present as micelles. However, it is not clear whether, in the case of micelles, actual binding occurs or whether PINs instead become included within the micelles. Kooijman and others (1997) showed that lipids bind to PINs in a cooperative manner. Binding lipids makes PINs more hydrophobic. This enhances subsequent binding of additional lipids. Dubreil and others (1997) even found that PINs only interact with highly aggregated lipid structures, such as uni- and multilamellar vesicles.

Impact of amino acid substitutions on the lipid-binding properties of PINB

Wheat hardness results from deletion of Pina-D1 or a mutation in Pinb-D1, resulting in the absence of PINA or an amino acid substitution in the mature PINB protein, respectively (see “Genetics” section; Table 1 and Figure 1). Point mutations resulting in an early stop codon (Pinb-D1e, Pinb-D1f, and Pinb-D1g) result in proteins with an amino acid chain length of about one-third of the wild-type PINB. It is thus logical that the functionality is (almost) completely lost in these short-PINB proteins. However, generation of truncated proteins does not often occur in vivo. Many organisms use the nonsense-mediated mRNA decay pathway to degrade mRNA before it is translated into short, nonfunctional proteins (Amrani and others 2006). For other mutations, it is striking that a single amino acid substitution (Pinb-D1b: G46S; Pinb-D1c: L60P; Pinb-D1d: W44R) tremendously impacts wheat kernel texture. The G46S (from a glycine to a more polar serine residue) and W44R (from a tryptophan to a basic arginine residue) mutations in wild-type PINB cause no changes in the global folding conformation as shown with Fourier transform IR (Clifton and others 2007). Differences in lipid-binding capacity have been suggested to result from small changes in conformation of the Trp-rich domain, which are not detectable with Fourier transform IR. Clifton and others (2007) indeed found that the G46S and W44R mutants penetrate significantly less deeply into a lipid layer.

The G46S change impacts the polarity due to the additional OH-group, although this change in polarity is rather low considering the hydrophobicity of the complete Trp-rich domain. Clifton and others (2007), therefore, suggested that the major effect of this mutation is the loss of conformational freedom in the Trp-rich domain, since the larger side chain of serine reduces its rotation possibilities. Also, additional hydrogen bonds between serine and other polar amino acids are possible, again decreasing the mobility of the Trp-rich domain and, hence, the degree of penetration in lipid monolayers. Furthermore, the glycine residue under consideration is conserved among PINA, PINB, and wheat ns-LTPs (Figure 3). This may indicate that it is required for proper PINB functionality (Poteete and others 1992).

For the W44R mutation, one can expect that the impact of replacing a tryptophan by an arginine is more drastic than that of introducing a serine residue (as in the G46S mutation). Indeed, Shiffer and others (1992) demonstrated that tryptophan residues are important to position and translocate transmembrane proteins and that they can play a role in lipid–protein interactions. However, the importance of tryptophan residues in the PINB Trp-rich domain is not yet clear, as they do not seem to be essential for interaction of PINB with Saccharomyces cerevisiae plasma membrane lipids (Evrard and others 2008). Apart from this, the introduction of a basic arginine residue, which is positively charged over a wide pH range, to replace a nonpolar tryptophan residue can significantly alter the polarity of the Trp-rich domain. Arginine can also form hydrogen bonds with polar amino acids. Taken together, it is likely that the arginine residue interacts strongly with anionic head groups of charged polar lipids, which may interfere with the orientation of the remaining tryptophan residues, as suggested by Clifton and others (2007).

The L60P substitution is located relatively far from the Trp-rich domain and, according to Le Bihan and others (1996), in the second α-helix. However, proline is unique in that its side chain is covalently bound to the preceding peptide-bond nitrogen, which makes hydrogen bonding with other amino acid residues impossible. The 5-membered pyrrolidine ring also confers rigidity and a bend to the protein backbone and restricts the conformational space available to the preceding residue. MacArthur and Thornton (1991) highlighted the unique role of proline residues in determining conformation, protein structure, and sometimes even function. Figure 3 shows that the L60 residue is conserved among PINs, wheat ns-LTPs, and selected 2S storage proteins, which points to its functional importance. In contrast to what is the case for G46S and W44R mutants, no data are available on the impact of this mutation on the secondary structure of PINB and on lipid-binding properties.

The 3 aforementioned point mutations are the most common ones in European wheat cvs. Worldwide, several other point mutations in PINA and PINB have been reported, as of 2008, the Pinb alleles went up to “ab” (Morris and Bhave 2008), some occurring at a relatively long distance from the Trp-rich domain (for example, Pinb-D1w: S115I). Whatever the case may be, further research is needed to understand the impact of single amino acid substitutions on the lipid-binding properties and, hence, wheat hardness.

The Mechanism Determining Wheat Hardness

The initial hypothesis for the role of friabilin in determining wheat hardness, suggesting a role as “nonstick” protein between starch and gluten, is nowadays considered as too simplistic. However, the idea of differences in adhesion between starch and gluten in wheat cvs. with different endosperm hardness still stands. Barlow and others (1973) reported that the starch granule surface is the likely place of difference between soft and hard wheat cvs. Later, a correlation was found between the presence of friabilin and endosperm hardness (Greenwell and Schofield 1986). Since PINs are the major constituents of friabilin, it is thought that they interact in one way or another with starch granules. Individual, but yet cooperative, roles for PINA and PINB in binding to the starch granules have been hypothesized (Amoroso and others 2004; Capparelli and others 2005; Swan and others 2006). Using transgenic wheat lines overexpressing PINA and/or PINB, Wanjugi and others (2007) concluded that both functional PINA and PINB are required to obtain high levels of starch granule-associated PINs. In that case, soft texture is obtained. When either high levels of (wild-type) PINA but not (wild-type) PINB are expressed, or vice versa, intermediate levels of PINs associate to starch and the wheat kernels are of intermediate hardness. However, no consensus exists on whether one of the PINs would limit binding of the other to the starch granule surface. In this context, Capparelli and others (2003) proposed a primary role for PINA in binding to the starch granule surface. They observed that Italian hard wheat cvs. having the Pina-D1b and Pinb-D1a alleles (no PINA present) have lower PIN levels (and almost no PINs bound to the starch granules) than hard wheat cvs. with the Pina-D1a and Pinb-D1b alleles (G46S mutation in PINB). However, in the case of Pina-D1b mutation, no PINA protein is expressed, while with the Pinb-D1b mutation PINB is expressed, but probably with a different functionality (decreased lipid-binding capacity or lipid penetration behavior; “Interactions between PINs and polar lipids” section) due to the mutation. It is likely that the latter mutant PINB protein still exerts some effect on wheat hardness. Additionally, PINA might be the limiting factor in binding PINs to the starch granule surface (Amoroso and others 2004; Gazza and others 2005), and, therefore, its presence might be more important than the presence of PINB to obtain high levels of starch-bound PINs. Only one study has shown that PINB restricts binding of PINA to starch granules (Swan and others 2006). Transgenic wheat lines expressing additional PINB contained higher levels of both starch granule-associated PINs and were much softer than when additional PINA was expressed. Only the level of starch granule-associated PINA was increased in the latter plants. The findings by Feiz and others (2009b) concur with the above. They observed that, in transgenic wheat, most overexpressed PINB, but not PINA, is associated with starch. A final reasoning would be that interactions between PINA and PINB occur in the wheat kernel, as demonstrated in vitro (Ziemann and others 2008). Notwithstanding the above, there is general agreement that both functional PINA and PINB are essential for soft texture.

Next to PINs, polar lipids presumably codetermine wheat hardness. Greenblatt and others (1995) were the first to report a relationship between levels of bound glyco- and phospholipids on water-washed wheat starch and friabilin, and thus endosperm hardness. Feiz and others (2009b) later confirmed these results using transgenic cvs. overexpressing PINs. Addition of Pina-D1 to a Pina-D1b/Pinb-D1a (PINA null) genotype, or Pinb-D1 to a Pina-D1a/Pinb-D1e (PINB null) genotype, resulted in higher bound polar lipid levels and softer texture than addition of the other (already present) PIN protein, indicating that both are necessary for soft texture. Furthermore, Greenblatt and others (1995) showed that following removal of starch-bound polar lipids, with a propan-2-ol/water (90:10) mixture, friabilin becomes extractable with a Tris-buffered salt solution, whereas only a very small amount of friabilin can be extracted with the same buffer when no lipids are removed beforehand. They hypothesized that most friabilin components associate with starch through polar lipid-mediated hydrophobic and ionic interactions. In vitro starch-binding experiments indeed showed that defatting starch granules reduces their capacity to bind proteins extracted from soft wheat flour using the nonionic detergent Triton X-114. The Triton X-114 extract contained PINs and other proteins (such as storage proteins and the enzyme granule-bound starch synthase I) with affinity for defatted starch granules (Bako and others 2007). Oda and Schofield (1997) postulated the lipid-mediated association of PINs with starch granules to be either direct, involving lipid “bridges” between the starch granule surface and the friabilin proteins, or indirect, when PINs undergo a conformational change due to the presence of polar lipids, hence allowing them to bind to the starch granule. PINs acquire a slightly higher α-helix level when binding polar lipids (see “Structure” section). It is also possible that polar lipids bind proteins already adhering to starch granules (Oda and Schofield 1997). However, the precise mechanism is unknown. Extraction of polar lipids from the starch granule surface with organic solvents of different polarity, such as chloroform and isopropanol:water (90:10, v/v), did not remove PINs from the granule surface (unpublished results). It thus seems likely that not only lipid bridges are involved in the interaction between starch and PINs. Indeed, removal of polar lipids, that is the bridges, would then lead to removal of PINs. Furthermore, it seems unlikely that conformational changes due to lipids remain when lipids are removed. Possibly, more than one mechanism is involved in PIN-lipid–starch interactions.

In general, because of its lipid-binding capacity (see “Interactions between PINs and polar lipids” section), the Trp-rich domain is thought to be the functionally important region of PINs. By creating new Pin alleles in the soft wheat cv. Alpowa, Feiz and others (2009a) showed the importance of the PINB Trp-rich domain in modulating wheat endosperm texture. Only mutations in the Trp-rich domain of PINB increased wheat hardness significantly, while the hardness of seeds having a similar mutation in PINA was hardly affected. This finding supports the idea that the conformation of PINB is more sensitive to mutations than that of PINA. Furthermore, Wall and others (2010) identified the Trp-rich domain of PINs as the starch-binding region. In situ tryptic digestion and mass spectrometry of the starch-associated proteins revealed a peptide corresponding to the Trp-rich domain of PINB at the starch granule surface, while no such peptide from PINA was found. Instead, 2 peptide fragments of PINA were found in the supernatant after tryptic treatment. This may indicate that PINB is more critical in binding PINs to starch granules (Swan and others 2006) than PINA and that PINA is not directly bound to the surface of the starch granules. However, the observed differences between PINA and PINB can also be due to differences in hydrolysis pattern of trypsin toward the proteins. Although this study determined the Trp-rich domain as the likely place for starch binding, no information is available about the involvement of polar lipids in this interaction.

Bechtel and others (1996) freeze dried, air dried, or oven dried immature (15 to 28 DAF) kernels from soft and hard wheat cvs. While the freeze-dried kernels had similar SKCS hardness values, the air- or oven-dried kernels showed different endosperm hardness. It seems plausible that the latter drying processes have a different impact on the amyloplast membrane. To most authors (Lillemo and Morris 2000; Feiz and others 2009b), it is clear that PINs play a role during seed desiccation and maturation. Recently, Kim and others (2012b), using near-isogenic lines, demonstrated that soft cvs. contain more phospho- and glycolipids than the corresponding hard cvs. during seed development, with the largest differences in polar lipids found in mature seeds. Indeed, earlier, Feiz and others (2009b) found an increase in bound polar lipids in mature seeds when PINs were overexpressed. Therefore, it is hypothesized that PINs determine wheat hardness by stabilizing the amyloplast membrane during grain desiccation, thereby preventing total breakdown of the lipid membrane when the wheat kernel ripens (Lillemo and Morris 2000; Feiz and others 2009b; Kim and others 2012b). In this view, a thin barrier of membrane remnants separates the starch granules from the gluten in the endosperm. However, in hard wheat, mutated PINs with a decreased lipid-binding capacity (see “Interactions between PINs and polar lipids” section) cannot stabilize the amyloplast membrane during seed desiccation, resulting in a closer contact and stronger adhesion between starch granules and the protein matrix.

Proteomic analysis of isolated amyloplasts from immature wheat (cv. Butte; Pina-D1b) 10 DAF revealed the presence of PINB as an amyloplast membrane-bound protein (Balmer and others 2006). However, in an analogous study (Andon and others 2002), neither PINA nor PINB were detected in wheat amyloplast (membranes) of the hard wheat cvs. Savannah (Pinb-D1d). This cv. contains a mutated PINB protein, possibly with reduced lipid-binding capacity. Together with the fact that one study (Swan and others 2006) proposed PINB to be the factor limiting binding PINs to the starch granule surface, it is possible that in this cv. PINA is prevented from binding to the starch granule. Also, immunolocalization studies did not clearly indicate whether or not PINs are amyloplast membrane-associated proteins during grain development (see “Occurrence” section). During wheat endosperm development, Lesage and others (2011) localized PINs in protein bodies, while, in the mature kernel, Dubreil and others (1998) located PINs at the starch granule surface, but only when surrounded by the protein matrix. Possibly, PINs or a fraction of PINs are only found at the starch granule surface when the protein bodies are fused to form a continuous matrix, which occurs at the end of the grain-filling stage (Bechtel and others 1982). However, differences in wheat hardness are already earlier detectable, starting from 15 DAF (Bechtel and others 1996; Turnbull and others 2003a). This may indicate that PINs stabilize the amyloplast membrane only at the later stages of grain development and that they impact wheat hardness in the earlier stages (starting from 15 DAF) by a different mechanism. In this regard, Lesage and others (2012) suggested that PINs affect the protein matrix formation during wheat endosperm development.

Other Cereals

Genes encoding PIN-like proteins occur in cereals related to wheat (such as barley, oat, and rye), but not in maize, rice, and sorghum. When present, these gene sequences are highly conserved in these cereals (Tanchak and others 1998; Darlington and others 2000; Gautier and others 2000). Barley (Hordeum vulgare L.) contains the PIN homologues hordoindoline a and b, which have been mapped on the short arm of chromosome 5H (Beecher and others 2001). The barley Ha locus has been shown to be associated with variations in endosperm texture (Beecher and others 2002b). Later, Takahashi and others (2010) showed that, when Hordoindoline b has a mutation resulting in an early stop codon, harder kernels are obtained. However, it is not clear whether Hordoindoline polymorphism affects barley hardness, since Turuspekov and others (2008) later found that differences in barley hardness could also be attributed to different seed size. In rye (Secale cereale L.), the PIN homologues are called secaloindoline a and b. Until now, no relationship between polymorphism of the Secaloindoline genes in rye and grain hardness has been reported because rye cvs. have generally soft endosperm texture with very low variation in kernel hardness (Simeone and Lafiandra 2005).


Wheat hardness is a major trait for classifying cvs. worldwide as it influences many aspects of the supply chain such as milling and end-use quality. Nowadays, it is widely accepted that PINs form the genetic basis of wheat hardness, with mutations in either of the genes Pina-D1 or Pinb-D1, or the absence of Pina-D1, resulting in increased kernel hardness. The way endosperm hardness is established, however, remains unclear. Because the levels of polar lipids at the starch granule surface are related to PIN levels, and because PINs show high affinity in vitro for these lipids, it has been suggested that polar lipids (amyloplast membrane remnants) are also involved in establishing wheat hardness. It has been suggested that PINs exert their effect on kernel texture by stabilizing the amyloplast membrane during seed maturation and desiccation. In hard wheat, such stabilization does not or only to a smaller extent occur, resulting in a more direct contact between starch and the gluten protein matrix and a harder texture. However, it is remarkable that even small differences in amino acid sequence (point mutations) have an enormous impact on wheat hardness. Differences in lipid-binding capacity for the mutant PINs are suggested. However, whether this is a direct effect or induced by conformational changes is not yet known. Unraveling the 3D structure of PINs would provide more information about the mechanism determining wheat hardness and PINs’ lipid-binding capacity.




eight-cysteine motif


base pairs


circular dichroism




days after flowering




glycine-to-serine change at position 46 in PINB


grain softness protein-1

Ha locus

Hardness locus




leucine-to-proline change at position 60 in PINB










nuclear magnetic resonance


nonspecific lipid transfer protein






puroindoline a


puroindoline b






sodium dodecyl sulfate


scanning electron microscopy


single kernel characterization system




tryptophan-to-arginine change at position 44 in PINB.


We kindly acknowledge Professor Peter Shewry and Dr. Frederic Beaudoin (Rothamsted Research, Harpenden, U.K.) for providing the figure of the PIN structure. This work is part of the Methusalem programme “Food for the future” (2007–2014). Bram Pareyt wishes to acknowledge the Research Foundation—Flanders (FWO—Vlaanderen, Brussels, Belgium) for a position as postdoctoral researcher. Jan A. Delcour is W.K. Kellogg Chair in Cereal Science and Nutrition at the KU Leuven.