Introduction

In bread wheat, the processing quality is largely determined by the composition of a group of seed storage proteins (named glutenin subunits) that are deposited in the wheat grains (Lawrence and Shepherd 1981; Shewry et al. 1995). The high molecular weight (HMW) glutenin subunits are encoded by the Glu-1-1 (x type) and Glu-1-2 (y type) genes contained in the Glu-1 locus (Lawrence and Shepherd 1981; Payne 1987). Since 1970 s, a large body of evidence from genetic, biochemical and transgenic studies has shown conclusively that HMW glutenin subunits are the major determinants of the processing quality of wheat grains (Shewry and Halford 2002). In addition to wheat, orthologous HMW glutenin subunits have also been found in Aegilops and rye species (William et al. 1993; Wan et al. 2000, 2002; De Bustos et al. 2001; Liu et al. 2003). In barley, the D-hordeins are structurally related to HMW glutenin subunits (Halford et al. 1992).

Extensive comparisons of HMW glutenin subunit genes have shown that the genomic space occupied by the x or y type genes may be below 10 kb (Anderson et al. 2002). Within the 2.5-kb region upstream of the coding sequence of well characterized HMW glutenin subunit genes (e.g., 1Ax2*, 1Ay, 1Bx7, 1Dx5 and 1Dy10), there are usually a conserved matrix attachment region (MAR) and the regulatory sequences required for gene transcription (Anderson et al. 2002; Rampitsch et al. 2000). Several types of miniature inverted-repeat transposable elements (MITEs) have been found within the MAR elements although the relative positions of the MITE insertions differ among different subunit genes. Waffle was inserted in the MAR element in 1Ax2* and 1Dx5, whereas Stowaway-Ta3 was found present in the MAR element of 1Ay (Anderson et al. 2002). In another investigation, an uncharacterized MITE sequence was found in the 5′ flanking region of 1Bx20 from a durum wheat variety (Anderson et al. 1998). Compared to the 5′ flanking region, the non-coding sequence immediately after the 3′ end of the open reading frame (ORF) of a HMW glutenin subunit gene is usually short (400–500 bp) (Anderson et al. 2002). At the protein level, the primary structures of x and y types of HMW glutenin subunits are similar and composed of a signal peptide (removed from the protein in mature seeds), an N-terminal domain, a central repetitive domain and a C-terminal domain (Shewry et al. 1995). In either x or y type of HMW glutenin subunits, the cysteine residues are usually conserved in both numbers and positions (Shewry et al. 1995). However, in the recently characterized 1Bx20 subunit, the last two of the three cysteine residues in the N-terminal domain, are replaced by tyrosine residues (Shewry et al. 2003). Both the conserved cysteine residues and the size of the repetitive domains contribute to the high order structure of HMW glutenin subunits. The former is involved in the formation of inter- or intramolecular disulphide bonds; the latter may promote intermolecular interactions through hydrogen bonding (Shewry et al. 2002). In this respect, it would be very interesting to identify more 1Bx20 like subunits and to study the effect of the reduction in conserved cysteine residues on their high order structures and their function in the end use qualities of wheat grains.

Compared to above studies, fewer investigations have been conducted on molecular evolution of HMW glutenin subunit genes. The primary structure of a HMW glutenin subunit may originally be formed by the triplication of an ancestral domain and the subsequent acquisition of a repetitive domain (Kreis et al. 1985). The later events leading to the formation of various Glu-1 loci (such as Glu-A1, Glu-B1 and Glu-D1) and the multiple alleles of a given locus have not been specifically addressed in past literatures. There is clear evidence for the evolution of novel allelic subunit through change in the number of the conserved cysteine residues. For example, the 1Dx5 is a novel subunit conferring good processing properties in bread wheat varieties because of the presence of a cysteine residue in its repetitive domain (Shewry et al. 1995; Anderson et al. 1989). Based on phylogenetic analysis using a conserved region (241–243 bp) located immediately upstream of the start codon of HMW glutenin subunit genes, researchers have estimated that the Glu gene duplication event (i.e., the differentiation of x and y types of HMW glutenin subunit genes) occurred 7.2–10.0 MYA (Allaby et al. 1999). The origin of A, B and D genomes (and hence the differentiation of the Glu-A1, Glu-B1 and Glu-D1 loci) may be dated 5.0–6.9 MYA (Allaby et al. 1999). The Glu-B1-1 alleles from cultivated wheats might be divided into two subgroups that diverged 1.4–2.0 MYA (Allaby et al. 1999). 1Bx7, 1Bx17 and several alleles amplified from an archaeological wheat sample were contained in one subgroup. However, the identities of the alleles contained in the other subgroup were not clear owing to the lack of sufficient sequence information.

The objectives of the studies reported in this paper are to characterize novel HMW glutenin subunit gene alleles and to investigate the evolutionary biology of this important group of genes. In the sections below, we describe molecular characterization of 1Bx14 and its gene and further insights into the evolutionary biology of Glu-B1-1 alleles based on our results and those published previously.

Materials and methods

Plant materials

The plant materials used in this study included hexaploid wheat varieties (Xiaoyan 54, Chinese Spring, Bobwhite, L88-6 and L86-69), tetraploid wheats (Triticum turgidum ssp. dicoccoides, T. turgidum ssp. dicoccon, T. turgidum ssp. turgidum, two accessions for each subspecies), and S genome containing Aegilops species (Ae. speltoides, six accessions; Ae. searsii, seven accessions; Ae. longissima, three accessions; Ae. bicornis, two accessions). L88-6 and L86-69 were both derived from the crosses between the mutant lines of Olympic and Gabo (Lawrence et al. 1988; Reddy and Appels 1993; Barro et al. 1997). The composition of HMW glutenin subunits in the two lines was 1Ax1, 1Bx17, 1By18, 1Dx5 and 1Dy10 (Reddy and Appels 1993; Barro et al. 1997).

SDS-PAGE and N-terminal protein sequencing

High molecular weight glutenin subunits were preferentially extracted from seed materials and were separated using SDS-PAGE as described elsewhere (Wan et al. 2000). The four HMW glutenin subunits in Chinese Spring (1Bx7, 1By8, 1Dx2, 1Dy12) were used as electrophoretic mobility standards in SDS-PAGE analysis. Using SDS-PAGE, the composition of HMW glutenin subunits in Xiaoyan 54 had previously been found to be 1Ax1, 1Bx14, 1By15, 1Dx2 and 1Dx12 (Zhang et al. 2002). To verify this finding we analyzed HMW glutenin subunits of Xiaoyan 54 by N-terminal protein sequencing as described previously (Wan et al. 2002).

Cloning and bacterial expression of the complete ORF of 1Bx14

Genomic DNA was extracted from the etiolated seedlings of Xiaoyan 54 as described previously (Wan et al. 2002). For amplifying the complete coding sequence of 1Bx14 from Xiaoyan 54 using genomic PCR, a pair of degenerate primers (P1 and P2, Supplementary Fig. S1 and Table S1) was designed according to the nucleotide sequences conserved in the 5′ or 3′ ends of the ORFs of published HMW glutenin subunit genes (Liu et al. 2003). The cycling parameters for the genomic PCR were the same as those reported previously (Liu et al. 2003). The amplified products included four fragments ranging from approximately 1.8–2.5 kb. Three fragments (whose size was about 1.9, 2.4 and 2.5 kb, respectively) were separately purified and cloned into pGEM-T Easy vector (Promega). By restriction enzyme digestion mapping and partial DNA sequencing, the 1.9 and 2.5 kb fragments were found to represent the ORFs of 1Dy12 and 1Dx2, respectively. The 5′ and 3′ sequences of the 2.4 kb fragment (in four independent plasmid clones) were highly similar to those of 1Bx20. This fragment was deduced to represent the ORF of 1Bx14 in Xiaoyan 54 and was completely sequenced. Potential mistakes brought about by genomic PCR using degenerate primers (P1 and P2) were corrected by additional PCR experiments amplifying the sequences flanking the 5′ or 3′ ends of the 2.4 kb fragment (see below).

To confirm the cloned 2.4-kb fragment as the coding sequence of 1Bx14, a set of bacterial expression experiments was conducted. The 2.4-kb fragment was reamplified using primers P3 and P4 (Supplementary Fig. S1 and Table S1) in order to remove the coding sequence for the signal peptide and to introduce restriction enzyme sites for subsequent cloning work. The reamplified fragment was cloned into the bacterial expression vector pET-30a (Invitrogen). The resultant expression construct pET-1Bx14 was induced to express the mature protein of 1Bx14 in bacterial cells as detailed in a previous publication (Wan et al. 2002). The electrophoretic mobility of the 1Bx14 protein produced in the bacterial cells was compared to that of 1Bx14 extracted from the seeds of Xiaoyan 54 using SDS-PAGE.

Cloning and sequencing the 5′ and 3′ flanking sequences of 1Bx14 ORF

The 5′ flanking sequence of 1Bx14 ORF was amplified using genomic PCR (as described above) with primers P5 and P1Bx14/20 (Supplementary Fig. S1 and Table S1). P5 was designed based on a sequence element that was strictly conserved in the 5′ flanking regions of 1Ax2*, 1Bx7, 1Bx20 and 1Dx5, whereas P1Bx14/20 was derived from the sequence coding for the six amino acid residues (ITVSPG) in the N-terminal domains of 1Bx14 and 1Bx20. The 3′ flanking sequence of 1Bx14 ORF was amplified by genomic PCR using primers P1Bx14 and P6 (Supplementary Fig. S1 and Table S1). P1Bx14 was derived from the sequence encoding the seven amino acid residues (AMCRLEG) in the C-terminal domain of 1Bx14. P6 was designed based on a sequence element strictly conserved in the 3′ flanking regions of 1Ax2*, 1Bx7 and 1Dx5. Taken together, the nucleotide sequence of 1Bx14 determined in this study was 4,021 bp (GenBank accession AY367771). During above experiments, the desired PCR fragments were purified, cloned in the pGEM-T Easy vector, and were sequenced from both strands. The final nucleotide sequences for the 5′ or 3′ regions of 1Bx14 were each constructed based on the sequencing results of three independent clones.

Molecular analysis of the MITE insertion in the 5′ flanking region of 1Bx14

To determine the copy numbers of the Tripper element in bread wheat, Southern hybridization experiments were conducted using genomic DNA samples of Xiaoyan 54 and Chinese Spring. Genomic DNA samples were digested with either HindIII or NsiI, separated in agarose gels, transferred onto nylon membrane, and hybridized using a Tripper specific probe (Sambrook et al. 1989). The 32P-labeled probe was prepared using the DNA fragment of Tripper (238 bp, amplified by PCR using primers P7 and P8, Supplementary Fig. S1 and Table S1) and the RadPrime DNA Labeling System (Invitrogen). To assess the influence of Tripper insertion on the transcription directed by the 5′ flanking region of 1Bx14, two expression constructs (pM1Bx14PR-GUS, p1Bx14PR-GUS) were prepared. A DNA fragment containing Tripper (−1,140~+3, relative to the start codon) was amplified by PCR with primers P9 and P10 (Supplementary Fig. S1 and Table S1). The amplified fragment, after digestion with KpnI and NcoI, was cloned into the pJIT166 vector (http://www.pgreen.ac.uk) that had previously been digested with the same enzymes. The resulted construct pM1Bx14PR-GUS was used to investigate the presence of Tripper insertion on the expression of the GUS marker gene. A DNA fragment lacking Tripper (−868~+3, relative to the start codon) was amplified using primers P10 and P11 (Supplementary Fig. S1 and Table S1). The resulted fragment was cloned into pJIT166 (as described above), giving rise to p1Bx14PR-GUS that was used to assess the absence of Tripper on the expression of the GUS marker gene. The two expression constructs were tested in a transient expression assay as described previously (Oñate et al. 1999). Briefly, gold particles coated with the DNA of the expression constructs were delivered into the endospermic tissues (20 tissues per bombardment) extruded out from the developing seeds of the bread wheat variety Bobwhite at 12–14 days after flowering using the PDS-1000/HE system (BIO-RAD). The bombardment was repeated three times for each construct. The results of the bombardment experiments were calculated as the mean numbers of GUS spots per endospermic tissue segment plus standard deviations.

Detection of potential 1Bx14 and 1Bx20 like alleles in Aegilops species and tetraploid wheat subspecies

The existence of potential 1Bx14 and 1Bx20 like alleles in four diploid, S genome containing Aegilops species and three tetraploid wheat subspecies was investigated using genomic PCR with primers P5 and P1Bx14/20 (Supplementary Fig. S1 and Table S1). For successful amplifications, the desired fragments were cloned and sequenced (as described above). The nucleotide sequences of the cloned fragments were constructed using sequence information derived from at least three independent clones.

Reinvestigation of nucleotide sequence of 1Bx17

While carrying out the studies in this paper, we found that the nucleotide sequence for a part of the 5′ flanking region (965 bp upstream of the start codon) of the 1Bx17 allele (from L86-69, designated here as 1Bx17-86) reported previously (Reddy and Appels 1993) differed, unexpectedly, in several locations from those of the homologous regions in 1Ax2*, 1Bx7, 1Bx20, and 1Dx5 (Supplementary Fig. S2). This prompted us to reinvestigate the nucleotide sequence of 1Bx17. Using the primers P5 and P1Bx17 (Supplementary Table S1), we amplified a DNA fragment of about 1.25 kb from L88-6, which would cover a part of the 5′ flanking region of 1Bx17 and the entire sequence encoding the N-terminal domain of 1Bx17 protein. This fragment was subsequently cloned and sequenced. Compared to 1Bx17-86, the sequence amplified from L88-6 (constructed from three independent clones, designated as 1Bx17-88) was more similar to its orthologous sequences in 1Ax2*, 1Bx7, 1Bx20, and 1Dx5 (Supplementary Fig. S2). So for the DNA or protein alignments that involved the 5′ flanking region of 1Bx17 or the protein sequence of the N-terminal domain of 1Bx17 in this paper, we employed the sequences derived from our own investigations using L88-6 (Supplementary Figs. S2 and S3).

DNA and protein sequence analyses and evolutionary investigations

For multiple alignments of DNA or protein sequences, the ClustalW program (Thompson et al. 1994) was generally used. For maximizing the similarities among the repetitive domains of 1Bx7, 1Bx14, 1Bx17 and 1Bx20, some manual adjustment to the multiple alignment was required. For predicting potential secondary structure of the Tripper element, the MFOLD program (http://bioweb.pasteur.fr/seqanal/interfaces/mfold-simple.html) was employed with default options. To investigate the phylogenetic relationship of 1Bx14 and 1Bx20 with previously characterized Glu-1-1 alleles (represented by 1Ax2*, 1Bx7, 1Bx17 and 1Dx5), a multiple alignment was created with homologous nucleotide sequences (the 5′ flanking sequences plus the ones encoding the N-terminal domains) using the ClustalW program. This alignment file was converted to mega format at the MEGA website (Version 2, http://www.oup-usa.org/sc/0195135857) for building phylogenetic trees using neighbor joining, minimal evolution or parsimony programs (Nei and Kumar 2000).

For inferring the ancestral amino acid sequences of the N-terminal domains of Glu-1-1 subunits, the computer program ANCESTOR (Zhang and Nei 1997) was employed. For this purpose, a phylogenetic tree was constructed with the amino acid sequences of the N-terminal domains of 1Ax1, 1Ax2*, 1Bx7, 1Bx14, 1Bx17, 1Bx20, 1Dx2 and 1Dx5 (and the corresponding region of barley D-hordein as an outgroup) using the neighbor joining method (in the MEGA website). The topology of the phylogenetic tree and the aligned amino acid sequences were then used as input information for computing the ancestral amino acid sequences at the different nodes of the phylogenetic tree.

To estimate the divergence time between 1Bx7 and 1Bx14 allelic lineages, the genomic sequence (5′ flanking sequence plus the one encoding the signal peptide and the N-terminal domain) of 1Bx14 (1,468 bp) was aligned to its homologous sequences in 1Bx7 (1,269 bp) or 1Bx17 (1,268 bp). The divergence time was calculated as described previously (Sanderson 1998). We used the average nucleotide substitution rate of 6.5×10−9 per site per year calculated for barley ADH genes (Gaut et al. 1996). This substitution rate has recently been used successfully for estimating the divergence time between two low molecular glutenin subunit genes in T. monococcum (Wicker et al. 2003).

For generating the alignments described in above analyses, appropriate DNA (or protein) sequences of 1Ax1, 1Ax2*, 1Bx7, 1Bx17, 1Bx20, 1Dx2, 1Dx5 and the barley D-hordein gene were retrieved from the GenBank. The EMBL accession numbers for 1Ax1, 1Ax2*, 1Bx7, 1Bx20, 1Dx2, 1Dx5 and the barley D-hordein gene are X61009, M22208, X13927, AJ437000, X03346, X12928, and AY268139, respectively. The accession number of the 1Bx17 protein sequence (Reddy and Appels 1993) is JC2099. The nucleotide sequence for the 5′ flanking region of 1Bx20 was derived from a previous publication (Anderson et al. 1998).

Results

N-terminal protein sequencing of 1Bx14 and homoeologous subunits in Xiaoyan 54

In the bread wheat variety Xiaoyan 54, the complement of expressed HMW glutenin subunits has previously been identified to be 1Ax1, 1Bx14, 1By15, 1Dx2 and 1Dy12 based on electrophoretic mobility comparisons (Zhang et al. 2002, Fig. 1a). We tried to confirm the expression of 1Ax1, 1Bx14 and 1By15 in Xiaoyan 54 by direct protein sequencing. The 18 residues obtained for 1Ax1 (EGEASGQLQCERELQEHS) were indeed identical to those in the published 1Ax1 subunit. The 20 residues obtained for 1Bx14 (EGEASGQLQCERELRKRELE) were identical to those in the previously reported 1Bx20 subunit. The 15 residues found for 1By15 (EGEASRQLQCERELQ) were identical to those in 1By9, 1Dy10 and 1Dy12.

Fig. 1a–c
figure 1

Characterization of HMW glutenin subunits expressed in the hexaploid wheat variety Xiaoyan 54 by SDS-PAGE and molecular cloning of the complete coding region of the Glu-B1-1 allele 1Bx14. a SDS-PAGE analysis showed that five HMW glutenin subunits (1Ax1, 1Dx2, 1Bx14, 1By15 and 1Dy12) were expressed in Xiaoyan 54 (lane 1) as compared to the four subunits (1Dx2, 1Bx7, 1By8 and 1Dy12) expressed in the hexaploid wheat variety Chinese Spring (lane 2). b Amplification of the complete coding regions of HMW glutenin subunit genes in Xiaoyan 54 via genomic PCR using degenerate primers (P1 and P2). By cloning and sequencing analysis, the three fragments whose size was above 1.9 kb were found to represent the ORFs of 1Dx2, 1Bx14 and 1Dy12, respectively (lane 2). The fragment marked by an asterisk was not characterized because its size may be below that of a functional HMW glutenin subunit gene ORF. The DNA markers (kb) were contained in lane 1. c Bacterial expression of 1Bx14 coding sequence. The expression of 1Bx14 mature protein (lane 2, indicated by an arrowhead) was detected in the IPTG induced bacterial culture (lane 2). In contrast, overexpression of 1Bx14 mature protein was not observed in the control bacterial culture that was not induced by IPTG (lane 1). The bacterially expressed 1Bx14 mature protein showed an electrophoretic mobility identical to that of 1Bx14 subunit extracted from the seeds of Xiaoyan 54 (lane 3)

Full size image

Characterization of 1Bx14 coding sequence and the primary structure of deduced 1Bx14 protein

Among the several molecularly characterized x type HMW glutenin subunits, 1Bx 20 is unusual in that its N-terminal domain contains only one conserved cysteine residue (Shewry et al. 2003). The finding on the N-terminal sequence of 1Bx14 in the protein sequencing experiment prompted us to further investigate if this subunit would be similar to 1Bx20 in possessing only one conserved cysteine residue in its N-terminal domain. We amplified the complete ORF of the 1Bx14 gene using primers P1 and P2 (Fig. 1b). The size of 1Bx14 ORF was 2,391 bp (including the six nucleotides coding for the tandem stop codons at the end of the ORF). When expressed in bacterial cells, the cloned 1Bx14 coding sequence yielded a polypeptide showing an electrophoretic mobility identical to that of 1Bx14 subunit extracted from the seeds (Fig. 1c), indicating that the cloned sequence was an accurate representation of the 1Bx14 ORF. The amino acid sequence deduced from 1Bx14 ORF contained 795 amino acid residues (Fig. 2). The primary structure of 1Bx14 was identical to that of the HMW glutenin subunits characterized previously and composed of a signal peptide, an N-terminal domain, a central repetitive domain and a C-terminal domain (Fig. 2, Table 1). The first 20 residues of the deduced 1Bx14 mature protein were identical to those determined by direct protein sequencing (Fig. 2). Amino acid sequence comparison showed that 1Bx14 was most closely related to that of 1Bx20. The two sequences differed in five positions (one in the N-terminal domain, three in the repetitive domain, and one in the C-terminal domain) involving the substitutions of single amino acid residues (Fig. 2). Interestingly, in the N-terminal domains of 1Bx14 and 1Bx20, the last two of the three cysteine residues, which were conserved in 1Bx7 and 1Bx17 (and all other Glu-1-1 subunits characterized so far), were replaced by tyrosine residues (Fig. 2).

Fig. 2
figure 2

A comparison of the amino acid sequences of four Glu-B1-1 subunits (1Bx14, 1Bx20, 1Bx7 and 1Bx17). This comparison shows that the four subunits possess an identical primary structure composed of signal peptide (underlined region), N-terminal domain (represented by bold letters), C-terminal domain (represented by italicized letters), and the central repetitive domain situated in between the N and C-terminal domains. The filled arrowhead indicates the cysteine residue that is strictly conserved among the four subunits. The empty arrowheads mark the two locations where the cysteine residues that are conserved in 1Bx7 and 1Bx17 are replaced by tyrosine residues in 1Bx14 and 1Bx20. The boxed regions indicate the five differences (caused by substitutions of single residues) between the amino acid sequences of 1Bx14 and 1Bx20. In the 1Bx17 sequence determined previously (Reddy and Appels 1993), the two locations marked by asterisks were occupied by “T” and “A”, respectively. According to the results of our own investigation on 1Bx17 in this work (Supplementary, Fig. S3B), the residues in the two marked locations have now been changed into “A” and “P”, respectively. The first 20 residues of the deduced 1Bx14 mature protein (represented by italicized, bold letters) are identical to those determined by direct protein sequencing

Full size image
Table 1 Some properties of the mature protein of 1Bx14 compared to those of previously reported 1Ax, 1Bx and 1Dx subunits
Full size table

Structural features of the 5′ flanking region of 1Bx14

The 5′ flanking region (−1,147 bp relative to the start codon) of 1Bx14 was obtained by PCR amplification using the primers P5 and P1Bx14/20 and was aligned to homologous regions in 1Ax2*, 1Dx5, 1Bx7, 1Bx17 and 1Bx20 (Fig. 3). This alignment showed that the 5′ flanking region of 1Bx14 was similar to those of 1Bx7, 1Bx17 and 1Bx20 (higher than 90% identities). Among the compared regions, a typical MITE insertion (Fig. 3, sequence represented by bold letters) was found in 1Bx14 sequence. This MITE insertion was shared by 1Bx20 but not 1Ax2*, 1Dx5, 1Bx7 and 1Bx17 (Fig. 3). The waffle insertions previously identified in the 5′ flanking regions of 1Ax2* and 1Dx5 (Anderson et al. 2002) were located more upstream than the MITE insertions seen in the 5′ flanking region of 1Bx14 and 1Bx20, and were not shown in Fig. 3. A tandem duplication of a sequence element consisted of 54 nucleotides (Fig. 3, underlined) was found in all four Glu-B1-1 alleles. A part of this duplicated sequence element was also present in the 5′ flanking regions of 1Ax2* and 1Dx5, but it was not duplicated (Fig. 3). The enhancer element (Fig. 3, sequence in brackets, Thomas and Flavell 1990) and the TATA box (Fig. 3, marked by asterisks) were conserved in the 5′ flanking regions of all compared alleles.

Fig. 3
figure 3

A comparison of the 5′ flanking sequences of four Glu-B1-1 alleles (1Bx7, 1Bx17, 1Bx14, 1Bx20) with homologous sequences from representative Glu-A1-1 (1Ax2*) and Glu-D1-1 (1Dx5) alleles. The MITE elements (Tripper) present in the 5′ flanking regions of 1Bx14 and 1Bx20 are represented by bold letters. The target site duplications caused by the Tripper element are boxed. The underlined sequence element (54 bp) is tandemly duplicated in all four Glu-B1-1 alleles. However, only a part of this element is present in the 5′ flanking regions of 1Ax2* and 1Dx5 and it is not tandemly duplicated. The region in brackets is the enhancer element conferring seed specific expression of HMW glutenin subunit genes. The TATA box is marked by asterisks

Full size image

The nucleotide sequences of the MITE insertions in the 5′ flanking regions of 1Bx14 and 1Bx20 were identical. The MITE had 14 bp terminal inverted repeat (TIR, 5′-CAGTGGCGGAGCTT-3′, Fig. 3). The insertion of the MITE produced 8 bp target site duplication (TSD, 5′-CATAGGAG-3′, Fig. 3, boxed regions). Nucleotide sequence comparisons suggested that the MITE associated with the 5′ flanking regions of 1Bx14 and 1Bx20 was not related to any other MITEs identified previously and was therefore given a new name Tripper. Southern hybridization analysis showed that Tripper existed as multiple copies in the genome of hexaploid bread wheat varieties (Fig. 4a). Computer modeling indicated that the nucleotide sequence of Tripper could potentially form a stable secondary structure composed of stems and loops (data not shown). The location of the Tripper element and its potential to form secondary structure led us to test if Tripper could affect the transcription directed by the 5′ flanking region of 1Bx14. Interestingly, we found that the presence or absence of Tripper in the 5′ flanking region of 1Bx14 did not significantly affect the expression of the GUS marker gene (Fig. 4b) in transient expression assays mediated by particle bombardment.

Fig. 4a, b
figure 4

Molecular analysis of the Tripper element present in the 5′ flanking region of 1Bx14. a Southern hybridization analysis, using NsiI (lanes 1 and 2) or HindIII (lanes 3 and 4) digested genomic DNA samples of Xiaoyan 54 (lanes 1 and 3) and Chinese Spring (lanes 2 and 4), showed that Tripper existed as multiple copies in the hexaploid genome of bread wheat. The size of the DNA markers (kb) is shown on the right side of the graph. b Influence of Tripper on the transcription (expression) of the GUS gene directed by the 5′ flanking region of 1Bx14 assessed using particle bombardment mediated transient assay. The 5′ flanking region of 1Bx14 in the expression construct pM1Bx14PR-GUS contained the Tripper element, whereas the Tripper element was deleted from 5′ flanking region of 1Bx14 in the construct p1Bx14PR-GUS via PCR mutagenesis. Based on the average numbers of GUS spots per endospermic tissue segment, the two expression constructs did not differ significantly in terms of GUS gene expression, indicating that the presence or absence of Tripper may not substantially affect the transcription directed by the 5′ flanking region of 1Bx14

Full size image

Absence of 1Bx14 and 1Bx20 like alleles in diploid Aegilops species containing various types of S genomes

Owing to the unusual properties in the promoter regions of 1Bx14 and 1Bx20 and in the amino acid sequences of the two subunits, it was interesting to investigate if 1Bx14 and 1Bx20 like alleles would be present in the ancestral species donating the B genome. Because the precise identity of the ancestral species donating the B genome is still not known, we tried to amplify 1Bx14 and 1Bx20 like alleles from four Aegilops species containing various S genomes, to which the B genome of tetraploid and hexaploid wheats may be related. Using the PCR primers P5 and P1Bx14/20, the anticipated fragment from the S genome Aegilops species would be about 1.35 kb (including 5′ flanking region, the coding sequence for the signal peptide, and the sequence encoding the first 45 amino acid residues of the N-terminal domain). In repeated PCR experiments, the expected 1.35 kb fragment was not detected in any of the six accessions of Ae. speltoides, seven accessions of Ae. searsii, three accessions of Ae. longissima, and two accessions of Ae. bicornis. In stead, a 1.1-kb fragment was found in the majority of the Aegilops accessions. The nucleotide sequence of this 1.1 kb fragment was highly similar to the 5′ flanking region plus the coding sequences for the signal peptide and the first 45 amino acid residues of the N-terminal domain in 1Bx7 or 1Bx17 rather than to the corresponding regions in 1Bx14 or 1Bx20. These results indicated that 1Bx14 and 1Bx20 like alleles might not be present in the S genome containing Aegilops species. In contrast, using the same PCR conditions 1Bx14 and 1Bx20 like alleles were readily detected in two (T. turgidum ssp. dicoccon and T. turgidum ssp. turgidum) of the three tetraploid subspecies (Supplementary Fig. S4).

Evolutionary analyses of Glu-B1-1 alleles

The results described above on comparative analyses of the amino acid sequences and the 5′ flanking regions of 1Bx14 and its alleles (1Bx7, 1Bx17, 1Bx20) suggested that 1Bx14 and 1Bx20 might represent a novel group of HMW glutenin subunit gene alleles. To study the evolutionary biology of 1Bx14 and 1Bx20, we asked three questions. (1) What was the phylogenetic relationship of 1Bx14 and 1Bx20 with previously characterized Glu-1-1 alleles? (2) Was the N-terminal domain possessing three conserved cysteine residues (exemplified by 1Bx7, 1Bx17, and homoeologous 1Ax and 1Dx subunits) more ancestral than those containing only one conserved cysteine residue (represented by 1Bx14 and 1Bx20), or vice versa? (3) What was the divergence time between 1Bx14 and 1Bx20 type of alleles and the 1Bx7 and 1Bx17 type of alleles?

To approach the first question, we conducted phylogenetic analysis of 1Ax2*, 1Bx7, 1Bx14, 1Bx17, 1Bx20 and 1Dx5 using the 5′ flanking sequences plus the sequences encoding the signal peptides and N-terminal domains. The selection of these sequences for phylogenetic analysis was based on the following reasoning. First, with the exception of the silenced alleles, HMW glutenin subunit genes are specifically and highly expressed in the endospermic tissue of developing seeds, indicating that the cis-elements that control tissue specificity and expression level of different HMW glutenin subunit genes are well conserved in the 5′ flanking regions. Second, the coding sequences for the signal peptides and N-terminal domains are also relatively conserved among different HMW glutenin subunit genes, probably owing to the important roles of the signal peptides (in targeting the newly synthesized subunits into protein bodies) and N-terminal domains (in maintaining the high order structure of the subunits). Third, the conservations in the 5′ flanking sequences and the sequences encoding the signal peptides and N-terminal domains suggest that these regions are subject to progressive changes during the evolution of HMW glutenin subunit genes. They are therefore phylogenetically informative. Fourth, the coding sequences for the repetitive domains are not suitable for phylogenetic investigations because they contain repetitive motifs that interfere with the correct alignment of the sequences to be compared. The phylogenetic tree thus constructed had two clades, one composed of Glu-B1-1 alleles and the other of Glu-A1-1 and Glu-D1-1 alleles (represented by 1Ax2* and 1Dx5, respectively) (Fig. 5). In the former clade, there were clearly two branches: one composed of 1Bx7 and 1Bx17 and the other of 1Bx14 and 1Bx20 (Fig. 5). Furthermore, the division of the two 1Bx branches was supported by high bootstrap values (Fig. 5), indicative of strong statistic support for the existence of two Glu-B1-1 allelic lineages.

Fig. 5
figure 5

Phylogenetic relationship of 1Bx14 and 1Bx20 with previously characterized Glu-B1-1 (1Bx7, 1Bx17), Glu-A1-1 (represented by 1Ax2*) and Glu-D1-1 (represented by 1Dx5) alleles. The rootless phylogenetic tree was constructed based on a multiple alignment of the 5′ flanking sequences plus the sequences encoding the signal peptides and N-terminal domains of the six Glu-1-1 alleles and the barley D-hordein gene (used here as an outgroup). The bootstrap values were calculated based on 500 replications

Full size image

For investigating the second question, we attempted to infer the ancestral amino acid sequence for the N-terminal domain of the x type HMW glutenin subunits (immediately before the differentiation of the two 1Bx allelic lineages) using the computer program ANCESTOR (Zhang and Nei 1997). To this end, a phylogenetic tree was constructed using the amino acid sequences of the signal peptides and N-terminal domains of 1Ax1, 1Ax2*, 1Bx7, 1Bx14, 1Bx17, 1Bx20, 1Dx2 and 1Dx5 and the corresponding regions of barley D-hordein as an outgroup (Fig. 6a). The results showed clearly that, immediately before the separation of the two allelic lineages of Glu-B1-1 subunits (in node 14, Fig. 6a and b), the configuration of the N-terminal domain was the one possessing three conserved cysteine residues, thus indicating that the N-terminal domain that contained three conserved cysteine residues was likely to be more ancestral.

Fig. 6a, b
figure 6

Inference of ancestral amino acid sequences for the N-terminal domains of x type HMW glutenin subunits using the ANCESTOR program (Zhang and Nei 1997). a A phylogenetic tree generated using the amino acid sequences of the signal peptides plus N-terminal domains of eight x type subunits and the barley D-hordein (used here as an outgroup). The numbers in brackets indicate nodal positions. b Prediction of ancestral amino acid sequences at nodes 10–16. Periods indicate amino acid residues identical to the ones in the 1Dx5 sequence. Hyphens indicate deletions of single or multiple amino acid residues. The ancestral amino acid sequence inferred for Node 14 (underlined) contains the three cysteine residues (indicated by arrowheads) conserved in the N-terminal domains of the majority of the HMW glutenin subunits characterized so far, indicating that the N-terminal domain containing three conserved cysteine residues is likely to be more ancestral to the one possessing one conserved cysteine residue

Full size image

The differentiation of the two Glu-B1-1 allelic lineages may occur either during the evolution history of tetraploid wheat or in the ancestral species denoting the B genome. If the former scenario was true, then the divergence time for the two Glu-B1-1 allelic lineages would be about 0.5 MYA because tetraploid wheat was formed approximately 0.5 MYA. By aligning the genomic sequences of 1Bx7 and 1Bx14 (or those of 1Bx17 and 1Bx14) and calculating the numbers of total nucleotide substitutions, the divergence time for 1Bx7 and 1Bx14 types of alleles was estimated to be 0.46±0.15 MYA (Table 2). This indicated that the divergence time of the two Glu-B1-1 allelic lineages might coincide with the timing of the tetraploidization event that led to the formation of tetraploid wheat. The relatively short divergence time between the two Glu-B1-1 allelic lineages was against the alternative scenario that the origin of the two lineages could be traced back to the ancestral species denoting the B genome. Moreover, the results from our PCR experiments on the four S genome containing Aegilops species indicated that the B genome ancestral species, although expressing 1Bx7 and 1Bx17 like alleles, might not encode 1Bx14 and 1Bx20 like alleles.

Table 2 Estimation of divergence time between 1Bx7 and 1Bx14 allelic lineages
Full size table

Discussion

In the work described in this paper, we reported, for the first time, molecular information on the nucleotide sequence of the HMW glutenin subunit allele 1Bx14 and the amino acid sequence of its protein product. Using high fidelity genomics PCR, the nucleotide sequences covering the coding region as well as the 5′ and 3′ flanking regions of 1Bx14 were obtained. Each nucleotide sequence was constructed using sequencing information derived from multiple independent clones. Potential errors brought about by the use of degenerate primers (e.g., P1 and P2) were corrected by amplifying and sequencing additional DNA fragments containing the regions from which the degenerate primers were originally derived. Using bacterial expression experiment, the correctness of the amplified coding region sequence of 1Bx14 was confirmed. Finally, the first 20 residues of the mature protein of 1Bx14 deduced from the cloned coding region were found to match exactly with those determined by direct protein sequencing. Taken together, it can be concluded that the molecular information we generated for 1Bx14 and its protein product is reliable and is hence suitable for investigating structural differentiations and evolution of Glu-B1-1 alleles in bread wheat and related tetraploid wheat species.

It has long been shown that, in both tetraploid and hexaploid wheats, the Glu-B1 locus encodes more x (and y) alleles than the Glu-A1 and Glu-D1 loci (Payne et al. 1984). Prior to our study, molecular information on the gene structure of Glu-B1-1 alleles was only available for 1Bx7, 1Bx17 and 1Bx20. The results reported here suggest that 1Bx14 was similar, but not identical, to 1Bx20. The insertion of the Tripper element in the 5′ flanking region and the possession of a novel ORF that encodes HMW glutenin subunit with fewer conserved cysteine residues are unique features of 1Bx14 and 1Bx20. However, the amino acid sequences of 1Bx14 and 1Bx20 subunits differed in five locations (involving the substitutions of single amino acid residues). This may have been the cause for the difference in the electrophoretic mobilities of the two subunits in SDS-PAGE analysis (Payne et al. 1984). It is also likely that the two subunits may differ functionally because past studies have demonstrated that allelic HMW glutenin subunits that are more than 95% identical in amino acid sequences (e.g., 1Dx2 and 1Dx5) possess drastic differences in their effects on the processing qualities of bread wheat varieties (Shewry et al. 1995; Shewry and Halford 2002). In view of above discussions, it is possible that there may be both similarities and differences in the function of 1Bx14 and 1Bx20. Because both subunits contain fewer conserved cysteine residues, their high order structures may be more similar to each other than to those of 1Bx7 or 1Bx17 that contain the whole complement of conserved cysteine residues. The 1Bx20 subunit has been postulated to confer poor processing properties in bread wheat based on in vitro incorporation assays (Shewry et al. 2003). It will be important to examine whether or not the expression of 1Bx14 is also associated with poor processing properties in bread wheat in future investigations. We noticed that, in past literatures, 1Bx14 and 1Bx20 were deduced to contain different numbers of cysteine residues in their proteins based on the separation of reduced or reduced and alkylated protein samples through reverse phase high performance liquid chromatography (RP-HPLC, 37). However, estimation of cysteine residues in protein samples by RP-HPLC may not be precise. For example, amino acid analysis showed that 1Bx20 and 1Bx13 contained similar numbers of cysteine residues, yet the two subunits differed substantially in their behavior during RP-HPLC separations (Margiotta et al. 1993).

Central to the evolution of HMW glutenin subunit genes are the duplication event that gives rise to x and y types of genes, the divergence of various Glu-1 loci, and the differentiation of multiple alleles for a given Glu-1 locus. In past investigations (Allaby et al. 1999), a time frame for the above evolutionary events was given, and some evidence on the differentiation of Glu-B1-1 alleles was uncovered. The availability of the knowledge on 1Bx14 and its deduced protein permitted us to conduct a better analysis of the evolutionary relationships among Glu-B1-1 alleles characterized so far. Based on our work, several new insights into the evolution of Glu-B1-1 alleles were produced. First, our analysis demonstrated that four molecularly characterized Glu-B1-1 alleles (1Bx7, 1Bx14, 1Bx17, 1Bx20) possessed clear differences in their 5′ flanking regions (in terms of Tripper insertion) and their amino acid sequences (with respect to the numbers of the conserved cysteine residues in the N-terminal domain). Second, our phylogenetic analysis showed that the four Glu-B1-1 alleles could be classified into two allelic lineages with strong statistical support. The results on the inference of the more ancestral amino acid sequence of the N-terminal domain suggest that the lineage represented by 1Bx7 and 1Bx17 is more ancient than the one represented by 1Bx14 and 1Bx20. Third, by using longer stretches of nucleotide sequences that were judged to be phylogenetically informative, we estimated that the 1Bx7 and 1Bx14 allelic lineages probably diverged 0.46±0.15 MYA. Because the divergence of the two allelic lineages coincided with the timing of the formation of the tetraploid wheat, we hypothesize that the tetraploidization event might have some connection with the differentiation of the two Glu-B1-1 lineages (see below). The two Glu-B1-1 lineages uncovered previously (Allaby et al. 1999) diverged 1.4–2.0 MYA. Because this divergence time is substantially earlier than the time at which the tetraploidization event occurred, the two Glu-B1-1 lineages reported by previous investigators (Allaby et al. 1999) might have existed in the ancestral species denoting the B genome. Considering the fact that the Glu-B1 locus possesses a significantly greater diversity of alleles than the Glu-A1 and Glu-D1 loci in tetraploid and hexaploid wheats, the differentiation of Glu-B1-1 alleles at different evolutionary stages (i.e., in the ancestral species, in the tetraploid wheat, or after the formation of the hexaploid wheat) is possible. However, further studies, involving the characterization of more Glu-B1-1 alleles with diverse molecular structures, are needed to verify this possibility.

Tripper, a novel MITE inserted in the 5′ flanking regions of 1Bx14 and 1Bx20, was found to exist as multiple copies in the genome of hexaploid wheat. It is interesting to find that the nucleotide sequence of the Tripper element in 1Bx20 (from tetraploid wheat) was identical to that in 1Bx14 (from hexaploid wheat). Considering that the positions of the Tripper insertion in 1Bx14 and 1Bx20 were also identical, it is likely that the Tripper insertion in 1Bx14 and 1Bx20 took place before the split of the two genes (i.e., the two insertions did not occurred independently). In past literatures, MITE insertions have been found in numerous plant genes, either in 5′ flanking regions, introns, or 3′ untranslated regions (Wessler 1998). More recently, evidence supporting active MITE transposition has been found in rice, a distant relative of wheat (Jiang et al. 2003; Kikuchi et al. 2003; Nakazaki et al. 2003). MITE insertion can have three potential consequences on genes. First, the insertion may be lethal, which mutates the normal function of the inserted gene. For example, the insertion of the mPing MITE in rice Rurm1 gene caused a slender mutation of the glume (Kikuchi et al. 2003). Second, the inserted element may form a part of the gene structure. For example, the MITE Ditto-Os2 may provide the TATA box for the transcription of a rice gene homologous to maize knotted-1 (Wessler 1998). Third, MITE insertion modifies gene expression pattern or the biochemical function of the product of the inserted gene. In this case, MITE insertion would contribute directly to the natural evolution of a functional allele with new property. An interesting example in the literature is that the evolution of a new phosphate transporter gene allele might be linked to a MITE insertion in the 5′ flanking region (Rausch et al. 2001). It has been proposed that “genomic shocks” (caused by biotic and abiotic stresses) may enhance the activities of transposons (McClintock 1984). In this respect, it is important to find that tissue culture and γ-radiation stresses stimulated MITE transposition and that polyploidization activated the transcription of a retrotransposon (Kikuchi et al. 2003; Kashkush et al. 2003). Based on above discussions (and the results described in this paper), it is tempting for us to speculate that the evolution of the 1Bx14 lineage from 1Bx7 like alleles might be linked to Tripper insertions in the 5′ flanking regions of 1Bx14 like alleles, and that this insertional event might be triggered by tetraploidization during the formation of tetraploid wheat. In our transient expression assays, Tripper insertion did not affect the transcription directed by the promoter region of 1Bx14, indicating that the evolution of the 1Bx14 like subunits may not be linked to an alteration in the expression level of their coding genes.

In conclusion, we have characterized 1Bx14 and its coding and promoter sequences. Comparative analysis of 1Bx14 and other HMW glutenin subunit genes has provided new insights into structural differentiation and evolution of Glu-B1-1 alleles. 1Bx14, together with 1Bx20, constitute a novel subclass of HMW glutenin subunits with fewer conserved cysteine residues in their proteins. Their genes represent an allelic lineage distinct from the one containing 1Bx7 and 1Bx17. The precise mechanisms causing the divergence between 1Bx7 and 1Bx14 allelic lineages are currently unknown. But they may be linked to the polyploidization event and the dynamics of MITE insertions, both of which have profoundly affected the constitutions and activities of the genomes of grass species (Jiang and Wessler 2001; Ozkan et al. 2001; Feschotte et al. 2002; Kashkush et al. 2002).