Introduction

Polyphenol oxidase (PPO) is the major cause of enzymatic discoloration in Asian noodles and other wheat-based end products (Baik et al. 1994; Morris et al. 2000). This enzyme catalyzes the oxidation of phenols to form quinones in the presence of molecular oxygen (Mayer and Harel 1979). The resulting quinones can then polymerize to form high-molecular weight black or brown pigments. Darkening and discoloration affect consumer acceptance of wheat products, especially yellow alkaline and white salted noodles (Morris et al. 2000).

The structure and organization of genes encoding PPOs have been intensively characterized in a diverse range of higher plants (Sherman et al. 1995). Most angiosperm PPOs appear to be arranged in multigene families, although the structure of the PPO genes seems to differ between monocots and dicots. Dicotyledonous species such as tomato (Newman et al. 1993), potato (Thygesen et al. 1995), and red clover (Sullivan et al. 2004) contain intronless genes, while monocots like pineapple (Zhou et al. 2003) and banana (Gooding et al. 2001) carry genes with intervening sequences. In wheat, Sun et al. (2005) reported wheat PPO sequences from genomic DNA harboring two introns each. Thus, it appears that insertion of introns into PPO genes occurred after the divergence of monocots and dicots.

It has been hypothesized that genes encoding wheat PPOs may have evolved by gene duplication into a multigene family (Jukanti et al. 2004). However, complex genomes like allopolyploid wheat species make the detection of paralogous loci more difficult. Efforts have been made to identify genes related to PPO activity in both durum (Simeone et al. 2002; Watanabe et al. 2004) and bread wheat (Anderson and Morris 2003; Demeke et al. 2001; Demeke and Morris 2002; Jimenez and Dubcovsky 1999). Southern blot analyses have suggested that wheat contains more than one PPO sequence (Demeke and Morris 2002). Nevertheless, most of the genes identified so far have been associated with orthologs, which are located on homoeologous chromosomes, rather than paralogs (Jimenez and Dubcovsky 1999; Demeke and Morris 2002). A recent study based on sequence data suggested that wheat PPO genes are arranged in a multigene family consisting of two distinct phylogenetic groups with three members each (Jukanti et al. 2004). Genes from one group are expressed in developing kernels while genes of the other group are expressed in plant tissues other than seeds (Jukanti et al. 2006). Sequence similarity, as well as functional diversity, supports the hypothesis that these two groups represent paralogous loci. However, there is no molecular genetic or phylogenetic evidence to support the hypothesis that members within either the seed or non-seed gene clusters arose by gene duplication. Most genes in hexaploid wheat exist as triplicate sets of single-copy homoeologs from the A, B, and D genomes, and therefore represent orthologous rather than paralogous loci.

The objective of this study was to identify and describe novel PPO sequences from diploid, tetraploid, and hexaploid wheat species to gain a better understanding of the PPO gene family structure and organization. We were particularly interested in the identification and characterization of PPO genes associated with PPOs found in developing kernels.

Materials and methods

Genomic DNA was extracted from leaf tissue according to the method of Murray and Thompson (1980) using a single plant per accession. Plant material consisted of four diploids: Triticum urartu Thüm. ex Gandilyan, Triticum monococcum L., Aegilops speltoides Tausch, and Aegilops tauschii Coss.; two tetraploids: Triticum turgidum ssp. durum (Körn. ex Asch. and Graebn.) Thell. cultivar Langdon and T. turgidum ssp. dicoccoides (Desf.) Husn.; and two hexaploid wheat cultivars: ID377s and Klasic (Table 1). The analyses also included publicly available DNA sequences of T. aestivum retrieved from GenBank (Table 1). PPO gene sequences were amplified by PCR, and products were cloned into a pCR 2.1 cloning vector (TA Cloning kit, Invitrogen, Carlsbad, CA, USA). PCR reactions consisted of 50 μl containing 5 μl of 10 × Taq buffer, 250 μM of each deoxynucleoside triphosphate, 1.5 U Taq DNA polymerase, 10 pmol of each primer, 40.55 μl of sterile water, and 2 μl of the DNA. The reactions were completed in 41 cycles of denaturation at 94°C for 30 s, primer annealing at 68°C for 60 s, and DNA elongation at 72°C for 120 s. Sequencing reactions were carried out with the Big Dye Terminator Version 3.1 Cycle Sequencing Kit and sequenced using an Applied Biosystems 3100 Genetic Analyzer (Perkin-Elmer Applied Biosystems Division, Foster City, CA, USA).

Table 1 Summary of plant material used in this study
Full size table

All sequences were aligned with ClustalW, version 1.83 (Thompson et al. 1997). The splicing sites of introns were identified by the alignment of genomic DNA to EST/cDNA sequences using BLAST search. Gene trees were generated by maximum likelihood and neighbor-joining algorithms as implemented in PAUP* 4.0 beta 10 (Swofford 2002) and bootstrap was performed with 1,000 replicates using the program MEGA version 2.1 (Kumar et al. 2004, http://www.megasoftware.net/index.html). A sequence variability profile of PPO was analyzed using the Sequence Variability Analysis Program (SVARAP, http://ifr48.free.fr/recherche/labo/rickettsies/svarap/svarap.htm). The variability along multiple nucleotide sequence alignments was calculated as the proportion of sequences that differ from the consensus sequence at a given site (Khamis et al. 2003; Colson et al. 2006). Transposable elements (TEs) were searched by comparison with the Triticeae Repeat Sequence Database (TREP, http://wheat.pw.usda.gov/ITMI/Repeats/index.shtml; Wicker et al. 2002). The secondary structure of TEs was predicted using the MFOLD software package (http://www.bioinfo.rpi.edu/applications/mfold/dna/form1.cgi; Zuker 2003). All sequences obtained here were deposited in GenBank under accession numbers DQ889690–DQ889710 (Table 1).

Results

First, we characterized the nucleotide variability of nearly the entire PPO coding region (∼1,400 bp, excluding introns and gaps) on the basis of the alignment of five seed-specific PPO sequences (GenBank accessions AY515506 and BT009357, and B. Beecher, unpublished data). The mean variability for sliding windows of 50 nucleotides revealed two highly polymorphic areas with mean values up to 12.8% for the first region (∼650–1,000 bp), and 17.2% for the second region (∼1,050–1,650 bp). Moreover, they were flanked by conserved sequences with mean variability ranging from 0.8 to 2.8% (Fig. 1 a). Conserved areas were used further for primer design (Table 2).

Fig. 1
figure 1

a Sequence variability of seed-specific PPO genes (excluding introns and gaps). The y-axis indicates the mean variability per window of 50 nucleotides and the x-axis specifies the positions of nucleotides. The mean variability across nearly the entire coding region was derived from five PPO sequences (∼1,400 bp). Mean variability of partial PPOs (in bold) were derived from 21 sequences (∼500 bp). Arrows indicate primer binding sites. b Structure of the partial PPO gene region analyzed in this study (shaded) showing the intron insertion site

Full size image
Table 2 Primers used in this study to amplify partial PPO sequences
Full size table

A total of nine primer combinations allowed us to amplify and sequence one of the highly variable regions, between nucleotide positions 700 and 1,200 (Fig. 1a). As a result, we obtained a total of 21 distinct PPO sequences of ∼600–900 bp long from diploid (seven), tetraploid (four), and hexaploid (ten) wheat accessions (Table 1). The analysis of sequence variability based on partial PPOs revealed a pattern of nucleotide polymorphism very similar to that observed in the analysis of the five PPO sequences (Fig. 1a). Furthermore, 78% (77) of the variable sites were parsimony informative (non-singletons). Therefore, by comparing DNA segments from a highly polymorphic and phylogenetic informative region of the gene, we were able to infer phylogenetic trees that were very similar to those based on the entire coding sequence.

Structural diversity

In addition to nucleotide sequence polymorphism, the amplified region presented an intron presence–absence polymorphism with partial PPOs consisting of two different structures: a single exon and an exon containing a single intron (intervening sequence). The alignment of sequences with an intron–exon organization revealed five intron sequence types of variable length (98–360 bp long) and nucleotide composition (Fig. 2). They were designated as a through e (Fig. 2). Intron sequences a, b, and c were unique, whereas introns d and e occurred in more than one taxon (Fig. 3). Interestingly, all introns showed a conserved insertion site within the aligned region. This was predicted at position 819/820 (the start codon as the first position) of a known mRNA (GenBank accession number BT009357) (Fig. 1b). Moreover, all introns fit into the class of non-canonical introns characterized by a GC donor dinucleotide (the two bases of the 5′-terminus) and an AG acceptor dinucleotide (last two bases of the 3′-terminus) (Fig. 2).

Fig. 2
figure 2

Sequence alignment of introns a through d. The 5′GC and 3′AC dinucleotides at their termini are indicated with boxes. Terminal inverted repeats and TA recognition sites of the Stowaway element of intron sequence a are underlined and in bold

Full size image
Fig. 3
figure 3

Phylogenetic analysis of the PPO gene family. The tree was derived by neighbor-joining distance analysis using 28 sequences of about 440 nucleotides (excluding gaps and introns). Bootstrap values over 60% are indicated. The seed specific PPO subtree (A) was re-grafted to a different scale and appended to the main backbone tree as indicated by a broken horizontal line. B Shows the branch of non-seed PPO sequences. Genes are labeled by species name, accession identifier, and GenBank accession number (in parentheses). Intron presence is indicated by the corresponding number in a square box. Scale bar indicates the number of nucleotide substitutions per site

Full size image

Intron length polymorphisms resulted from several small length differences and one large insertion (206 bp) in the T. aestivum cv. Klasic intron a (GenBank accession DQ889708, see Fig. 3). BLAST searches using this fragment against the TREP database suggested two Miniature Inverted-repeat Transposable Elements (MITEs) from the Stowaway family. It is interesting that one 115-bp long element appeared to be nested within a 101-bp element (Fig. 4). While, the outer element showed a 7-bp terminal inverted repeat (TIR, 5′CTCCCTC3’) and the 2-bp long TA recognition target site characteristic of Stowaway-type elements. The inner element lacked the 3′ TA target site, and the predicted DNA secondary structure showed three mismatches at the TIRs (Fig. 4). Nevertheless, it is predicted to fold into a double hairpin-like structure (Fig. 4). The remaining introns showed no evidence of Stowaway elements, however, those from wheat cultivars ID3775s (DQ889709) (intron c) and T. monococcum cv. DV92 (DQ889710) (intron b) harbored a 5-nt consensus sequence, 5’TCCCT’3, which was identical to that of the 5′TIR of the Stowaway element observed in intron a (Fig. 2).

Fig. 4
figure 4

Secondary structure of the double Stowaway element of intron sequence 1 from T. aestivum cv. Klasic with ΔG = −43.8 kcal/mol. Terminal inverted repeats of the inner and outer element are marked by boxes

Full size image

Gene duplication

Evidence of duplicate sequences within a single genome (paralogous loci) was found in T. monococcum cv. DV92. We amplified and sequenced three distinct PPO fragments from this accession, two of the fragments lacked introns (DQ889695, DQ889701) and the other one (DQ889710) had a 152-bp intron within the target region (Fig. 3). In addition, several partial PPOs were amplified from T. aestivum cultivars ID377s (five) and Klasic (five), and T. turgidum cv. Langdon (three). It is noteworthy that the number of copies per individual genotype was higher than the expected number of orthologs (homoeologs) based on their genome constitution, that is, two orthologs from the tetraploid durum wheat cv. Langdon (AB) and three from the hexaploid bread wheat cultivars ID377s and Klasic (ABD), thus suggesting paralogous loci.

Evolutionary relationships

Phylogenetic analyses were based on a data set of 28 PPO sequences ∼440-bp in length (excluding introns and gaps). This included seven sequences retrieved from GenBank. The latter group consisted of four seed specific PPO genes (BT009357, AY515506, AY596270, and AY596268) and three non-seed specific PPOs (AY596266, AY596267, and AF507945), which were used to root the tree (Fig. 3). Similar tree topologies resulted from maximum likelihood and neighbor joining analyses, although only the latter tree is presented (Fig. 3).

Phylogenetic inferences indicated four major clusters (bootstrap values ≥80%) within the seed specific PPO subclass (clusters I–IV; Fig. 3). Clusters I–III were more closely related to each other than they were to cluster IV. The analyses further suggested two minor groups within cluster I (α and β). Although paralogous–orthologous relationships between these two minor groups cannot be determined from the present data set, it is interesting that conserved sequences appear to characterize different genomes within α or β subgroups.

The paralogous sequences of T. monococcum cv. DV92 (DQ889695, DQ889701, and DQ889710) were allocated to three different clusters and suggested orthologous relationships within, and paralogous relationships between clusters I, III, and IV (Fig. 3). Interestingly, PPO fragments within cluster II, which were highly conserved (99% nucleotide similarity), had a different structure. That is, T. turgidum cv. Langdon (DQ889702) and Ae. speltoides TA2779 (DQ889703) had no introns, while the remaining four sequences carried a 98-bp intron with two highly similar sequence types (introns d and e, Figs. 2, 3).

Sequence divergence within Group IV, as indicated by branch length on the phylogenetic tree, together with intron nucleotide and structural diversity, further suggested three putative subgroups (α, β, and γ, Fig. 3). Yet, the evolutionary relationships among members of cluster IV are also supported by their introns (Fig. 5).

Fig. 5
figure 5

Neighbor-joining tree of intron sequences (excluding gaps). Sequences are labeled as described in the text, followed by species name, accession identifier, and GenBank accession number (in parentheses). Scale bar indicates the number of nucleotide substitutions per site

Full size image

Dicussion

In the present study, we amplified highly discriminative PPO sequences from diploid, tetraploid, and hexaploid wheat species, and provided molecular evidence to support the hypothesis of a PPO multigene family structure and organization. Nucleotide and structural diversity, together with well-resolved phylogenetic trees, allowed us to describe putative orthologous–paralogous relationships among members of this gene family.

Since most sequence data available to date is based on cDNA and ESTs, the structure of wheat PPO genes remains poorly characterized. Recently, Sun et al. (2005) amplified partial PPOs from genomic DNA and reported the occurrence of two introns in Chinese wheat PPOs. This study (Sun et al. 2005) detected intron size variation between genotypes with high and low PPO activity, but no evidence of intron presence–absence polymorphism was documented. Our results indicated structural diversity within the target region and identified two different gene structures: a single exon and an exon containing an intron. Preliminary results in our lab suggest that intronless fragments may represent single-intron genes (B. Beecher, unpublished data). While, fragments carrying an intron most likely represent genes with two introns. All intervening sequences observed in the present study shared the same (putative homologous) insertion position, which was also the same as that of the second intron (intron 2) of Chinese wheat PPOs (Sun et al. 2005). Homology with intron 2 was also supported by sequence similarity and splice consensus dinucleotides. The intron 2 of Chinese wheat cultivars was identical to that of T. aestivum cv. ID377s (DQ889709) (intron c) and 88% similar to that of T. monococcum cv. DV92 (DQ889710) (intron b). Moreover, all introns from this study as well as intron 2 (Sun et al. 2005) fit into the GC-AG non-canonical introns.

The GC-AG splice site junctions have been previously reported in wheat Wknox1 genes (Morimoto et al. 2005), Arabidopsis thaliana (Brown et al. 1996) and Oryza sativa (Sparks and Brendel 2005). This alternative isoform is a variant of the wild type GT-AG and accounts for only a small fraction (<2.6%) of all introns in Arabidopsis and rice genomes (Sparks and Brendel 2005). It has been suggested that the mechanism of splicing involving GC-AG introns is the same as that of GT-AG introns (Zhu et al. 2003). Nonetheless, it will be interesting to determine whether a T to a C mutation at the 5′ dinucleotides (GT→GC) is relevant to splicing regulation of PPO genes.

Stowaway elements accounted for the major difference between the five introns. MITEs of the Stowaway family have been extensively characterized in plants and are ubiquitous in cereal grass genomes (Bureau and Wessler 1994a; Feschotte et al. 2002). However, elements with a double hairpin-like structure like that of T. aestivum cv. Klasic (DQ889708) (intron a) have only been described in Heteranthelium piliferum of the tribe Triticeae (Petersen and Seberg 2000). While no further evidence of MITEs was found in the present study, the consensus sequence 5’TCCCT’3 observed in T. monococcum cv. DV92 (DQ889710) (intron b) and T. aestivum cv. ID3775s (DQ889709) (intron c) (Fig. 2) suggest putative footprints left after excision of a Stowaway-type element. Based on evolutionary relationships, as indicated in Fig. 3, it is quite possible that the footprints observed in introns b and c are remnants from excision of the double T. aestivum cv. Klasic (intron a) element. MITE excision events appear to be rare. Nevertheless, phylogenetic evidence for excision of Stowaway MITEs has been previously documented in the tribe Triticeae (Petersen and Seberg 2000).

Although, Stowaway and other MITEs have been found to contain regulatory sequences involved in transcription initiation and polyadenylation of plant genes (Bureau and Wessler 1994a, b), their contribution to gene expression is uncertain. Based on the present results, it will be interesting to determine whether gain (or loss) of Stowaway-type elements within cluster IV sequences is associated with patterns of gene expression and ultimately, with differences in PPO activity.

Multigene families are thought to arise from duplication and divergence of an ancestral gene. Here we present molecular evidence of duplicate PPO sequences (paralogous loci) from the A genome of T. monococcum cv. DV92 and provide support for a minimum of three family members within the PPOs expressed in developing kernels. Furthermore, these results revealed the presence of putative orthologs for each of these three paralogs indicating their occurrence in other Triticum/Aegilops genomes (Fig. 3). For instance, the two copies of T. aestivum cv. Klasic of cluster III (putative orthologs to T. monococcum cv. DV92, accession DQ889701) must be PPO sequences from two different genomes of hexaploid wheat (Fig. 3). These results are consistent with previous assertions that PPO genes are located in homoeologous chromosomes of the A, B, and D genomes of wheat (Jimenez and Dubcovsky 1999; Demeke and Morris 2002; Watanabe et al. 2004).

In addition to the three paralogous loci of T. monococcum cv. DV92, phylogenetic inferences revealed a fourth group suggesting an additional (putative) paralogous PPO locus (cluster II). An intriguing characteristic of this group was the occurrence of intron absence–presence polymorphisms among its members. It may be possible that intronless sequences arose by recombination with a processed pseudo-gene or a reverse-transcribed cDNA copy of a processed mRNA (Frugoli et al. 1998). Yet, whether sequences within cluster II diverged by speciation (orthologous) or by gene duplication (paralogous) needs to be investigated further.

Sequence divergence together with structural polymorphism further suggested three minor groups within cluster IV. Subgroups α and β are probably paralogs as divergence between T. aestivum cv. Klasic (DQ889708) and T. monococcum cv. DV92 (DQ889710) is similar to that between the paralogs T. monococcum cv. DV92 (DQ889701) and T. monococcum cv. DV92 (DQ889695). Intron size variation as well as MITEs insertion (or excision) events within Cluster IV may also support this hypothesis. It is tempting to speculate that sequence divergence between this and the remaining gene clusters also reflects functional divergence.

A previous study reported three distinct wheat PPO sequences expressed in developing kernels (Jukanti et al. 2004), but no direct evidence of orthologous–paralogous relationships was documented. The present results suggest that the seed specific PPO genes of Jukanti et al. (2004) (GenBank accessions AY596268, AY596270, AY596269) stand for at least two putative paralogous loci. Genes AY596268 and AY596270 fell into cluster IV, while AY596269 was closely related (99% similarity, 370 bp) to sequences of cluster II (data not shown). Consistent with previous data (Sun et al. 2005), the sequence similarity between AY596268 and T. monococcum cv. DV92 (DQ889710) (Fig. 3) suggest a common ancestral genome sequence. Using Chinese Spring nullisomic–tetrasomic lines, Sun et al. (2005) mapped the PPO gene AY596268 to the long arm of chromosome 2A.

The present study documents and describes novel genomic PPO sequences from diploid, tetraploid, and hexaploid wheat species, and provides a molecular genetic and evolutionary framework for these genes.