One of the main objectives of cereal breeding is to expand genetic variability within cultivated species. Wild species, related to cultivated crops, are an important source of variability. Inter-specific hybridization can be used to introgress genetic variability from wild species into crops and to produce new species with valuable agronomic traits. An example of this is the hexaploid tritordeum, an amphiploid obtained by crossing Triticum turgidum L. (Thell) (2n = 4x = 28) with Hordeum chilense (Roem. et Schult.) (H^chH^ch, 2n = 2x = 14). Primary tritordeums exhibit enormous genetic variability for many valuable agronomic and quality traits. For example, the grain and flour from tritordeum has similar functional properties to bread wheat [1], but with higher pigment content [2, 3]. Most of this genetic variability can be attributed to H. chilense, a wild relative of cultivated barley (H. vulgare L.) that occurs exclusively in Chile and Argentina which is highly polymorphic both morphologically and biochemically. The variability for important agronomic traits, such as endosperm storage proteins [4, 5], carotenoid content [6] and resistance to biotic stresses [7], linked to its high crossability, makes H. chilense a suitable candidate as a source of genetic variability for the transfer of useful genes to wheat by wide crossing. However, in the process of hybridization, undesirable traits such as rachis brittleness and non-free threshing characters, present in wild barley, are also transferred to the hybrid, limiting its use as an alternative cereal crop.

Many genetic systems have been proposed as responsible for the free-threshing character in hexaploid wheat. MacKey [8] proposed a polygenic system distributed throughout the three genomes that counteracts glume tenacity and rachis brittleness. A second system is related to the major gene or gene complex Q [9, 10] located in the long arm of the chromosome 5A which governs the free-threshing character and square spike phenotype. In addition, the Q gene pleiotropically influences many other characters determinant for domestication such as rachis fragility [9, 11], glume shape and tenacity [10, 12], spike length [10, 13], plant height [10, 13, 14] and spike emergence time [13]. Other genes which influence the free-threshing habit include the Tg locus, located on chromosome 2D [11, 15] that codes for tenacious glumes and is thought to inhibit the expression of the Q gene.

Simons et al. [16] have cloned and characterized the Q gene in wheat and showed that it has a high homology to members of the APETALA2 (AP2) family of transcription factors. This gene family is characterized by two plant-specific DNA binding motifs referred to as AP2 domains. The AP2 genes form a large multigene family, and play multiple roles during the plant life cycle being key regulators of many developmental processes such as floral organ identity determination or control of leaf epidermal cell identity [17]. Many studies in Arabidopsis have revealed that the AP2 gene is implicated in the establishment of floral meristem identity [18, 19], floral organ identity [20–22] and the temporal and spatial regulation of flower homeotic gene expression [23].

In the present work, we have cloned and characterised an AP2-like gene from accessions of H. chilense and H. vulgare, wild and domesticated barley respectively, and compared these with other homoeologous genes, including the Q gene from wheat. The pattern of expression of the AP2-like gene in floral organs and other tissues in barley, wheat and amphiploid tritordeum was also studied. The results are relevant to understanding the role of the AP2-like gene in the spike morphology of barley and wheat and in hybrids obtained from their crossing and for modification of the expression of AP2-like genes to modify the spike characteristics of cereals for breeding purposes. In addition, the results provide insight into how important agronomic genes such as AP2 are regulated in cereal hybrids.

Alignments of the sequenced cDNA and DNA of the H. chilense lines H11 and H208, and H. vulgare line H106 have shown that the internal structure of exons and introns of the AP2-like gene described in this work (Figure 1) is the same as that reported for the Q gene in wheat [GenBank: AY702956] [16], with 10 exons and 9 introns. BLAST searches revealed that the sequenced genes belong to the AP2 family of transcription factors which includes the floral homeotic gene AP2 [22] and AINTEGUMENTA (ANT) [24, 25] involved in lateral organ development by controlling cell number and growth [26, 27]. AP2-like genes are distinguished by having two plant-specific DNA binding motifs called AP2 domains and function as key developmental regulators in reproductive and vegetative organs [17]. Additionally, the AP2-like genes possess one microRNA172 binding site in the 3' region (Figure 1). The microRNA miR172 with 21-nucleotide non-coding RNA was reported to down-regulate several Arabidopsis genes in the AP2 subfamily [28, 29]. This miR172 and its target sequence are highly conserved in all the genotypes aligned in Figure 2, with only a single change in the sequence of the T. aestivum cv Chinese spring [GenBank: AY702956].

The conserved DNA binding motifs, referred to as AP2 domains R1 and R2, consist of 60 and 61 amino acids respectively and the alignment with other AP2-like proteins confirmed that these domains are highly conserved, even between non-related species such as H. chilense and A. thaliana (Figure 2). Jofuku et al. [22] isolated and characterized the AP2 gene from A. thaliana [GenBank: U12546] and reported a 53% of identity between the two AP2 domains R1 and R2 in the central core of the AP2 polypeptide. They also described the presence of an 18-amino acids conserved core region in the two AP2 domains with a 72.2% of identity between them. This region is theoretically capable of forming amphipathic α-helical structures that may participate in protein-protein interactions. This α-helical structure could also participate in DNA binding, perhaps through the interaction of its hydrophobic face with the major groove of the DNA, as described for other proteins with similar α-helical structure [30]. In the study reported here, the AP2 R1 domain from H. chilense line H11 has 95% of identity with the corresponding amino acid sequence of that from A. thaliana described by Jofuku et al.[22], while the conserved core region is identical. On the other hand, when comparing the AP2 R2 domain of the two species we found 83.6% of identity in the full domain and 72.2% in the core region. Jofuku et al. [22] also reported the presence of a highly basic 10-amino acids domain adjacent to the AP2 R1 domain that included a putative nuclear localization sequence KKSR [31, 32] which suggested that the AP2 may function in the nucleus. The same domain of 10 amino acids is present in the sequences reported here with only a change in the second position of the nuclear localization sequence (threonine instead serine) which is conserved in all the Triticeae sequences of the AP2-like proteins aligned in Figure 2. Tang et al. [33] described the amino acid sequences of the three motifs found in the AP2-like protein in rice and compared it with other AP2-like proteins from different species. We have found similar structures in the predicted AP2-like proteins of wild (H11 and H208) and cultivated (H106) barley (Figure 2).

Previous experiments involving the cytogenetic analysis of aneuploids have located the Q gene in the long arm of chromosome 5A of the wheat genome [9, 34]. The H. chilense genome has been demonstrated to be collinear to other Triticeae genomes including those of bread wheat and H. vulgare [35] so that the AP2-like gene was predicted to be on chromosome 5H^ch of H. chilense. To confirm this, the AP2-like gene from H. chilense was amplified by PCR in H. chilense (H1), T. aestivum cv. Chinese Spring, a 5H^ch addition line of T. aestivum cv. Chinese spring, an amphiploid tritordeum (HT22), and two tritordeum lines HT374 and HT382 carrying chromosome substitutions (Figure 3). The results confirmed that the barley AP2-like gene is located on chromosome 5H^ch and that the HT374 substitution line (5D/5H^ch) effectively lacks this chromosome.

Expression of the AP2-like gene was detected in all tissues studied. The transcription level peaked in the early stages of spike development and gradually decreased with spike maturation, being very similar in all the genotypes tested when the spike was fully emerged. These results are similar to those reported by Simons et al. [16] who observed higher expression of the Q gene in developing spikes, with a peak at the first stages of spike growth in wheat genotypes. Simons et al. [16] also reported lower expression in non-floral organs such as leaves and roots. Jofuku et al. [22] studied the transcription of AP2 in A. thaliana obtaining expression in floral organs (sepals, petals, stamens, carpels, developing ovules and inflorescence meristems) and also in non-floral organs (stems and leaves).

We characterised the expression of the AP2-like gene in wild (H. chilense genotype H1) and cultivated barley (H. vulgare genotype H106) (Figure 4) and found a higher level of transcription in wild barley in roots, stems and developing spikes. Simons et al. [16] reported that the transcription level of the Q allele, contained in bread and durum wheat cultivars, was consistently higher than that of q allele, contained in wild wheats, and this was related with differences in spike morphology. They also described that the single amino acid difference found in their predicted proteins could provide higher efficiency in homodimer formation in the Q allele with respect to that of the q allele. They suggested that the Q protein homodimer complex recognizes a region on its own promoter, enhancing the expression of the Q allele and leading to higher levels of the Q protein, and this was related to phenotypic differences in spike morphology between cultivated and wild wheats. Despite this, differences between Q and q can be compensated by a gene dosage effect, with 2.5 doses of q being equal to 1 dose of Q [10].

Our results with wild and cultivated barley showed that H. chilense (wild) had higher AP2-like gene expression levels than H. vulgare (cultivated). According to the model proposed by Simons et al., [16], in the case of barley, the homodimer formation should be higher in H. chilense than that of H. vulgare, resulting in higher transcript levels of the AP2-like gene in H. chilense. As for wheat, differences in expression levels for the AP2-like gene in wild and cultivated barley could be responsible for phenotypic differences in spike morphology. However, in barley, the line showing higher expression levels of the AP2-like gene is H. chilense which is not the cultivated phenotype. Hence, the wheat model does not entirely fit to barley, either because the mechanism in barley is different or because the mechanism is more complex than that reported by Simons et al. [16]. In the case of wheat, the Q and q alleles differ only by one amino acid while the AP2-like genes from wild and cultivated barley differ by 17 amino acids. Therefore, differences in the spike morphology between wild and cultivated barley may not only be due to AP2-like gene expression level, but additional elements must be added to the model to explain better the contribution of the AP2-like gene to spike morphology in barley.

We have also characterised the expression of the AP2-like genes corresponding to wheat and barley genomes in tritordeum amphiploids (HT22, HT374 and HT382). The comparison of the expression as wheat/barley AP2 ratio when both genes are expressed together in the tritordeum amphiploid (HT22) and when they are expressed separately in their parental lines, H. chilense (H1) and T. durum (T22), showed that the relative expression of the wheat AP2-like gene in different tissues was very constant in the amphiploid and in its durum wheat parental (T22). However, the level of transcription of the barley AP2-like gene was 3.5 times lower in the amphiploid than in its barley parental (H1). This could be explained by the presence of a mechanism of regulation for the barley AP2-like gene that makes it differently expressed in the amphiploid and in barley. One of the consequences of this down-regulation of the barley AP2-like gene may be the aspect of the spike in the tritordeum amphiploid, which is more similar to that of durum wheat than that of wild barley. However, many important agronomic features of the spike in tritordeum, such as fragile rachis and non-free threshing habit are similar to H. chilense.

Two tritordeum substitution lines (HT374 and HT382) with free-threshing habit [36] showed only expression of the wheat AP2-like gene. This was predictable for line HT374, which has a 5D/5H^ch chromosome substitution and consequently lacks the barley AP2-like gene, but not for line HT382 (double substitution 1D/1H^ch and 2D/2H^ch) which carries the 5H^ch chromosome and therefore the barley AP2-like gene. Thus in this case, expression of the barley AP2-like gene was affected also by substitution of chromosomes 1H^ch and/or 2H^ch. Atienza et al. [36] proposed that the presence of a homoeologous q locus on chromosome 5H^ch would be responsible for the non-free-threshing habit in tritordeum, and consequently the substitution of 5D/5H^ch in the line HT374 was responsible of the free-threshing habit. It was also suggested that the absence of chromosome 2H^ch conferred the free-threshing habit in tritordeum HT382, because of the absence of a homoeologous Tg locus from H. chilense that codes for tenacious glumes, in spite of the presence of q H^ch locus on chromosome 5H^ch [36]. Kerber et al. [15] and Jantasuriyarat et al. [11] have proposed that the action of Tg during flower development directly or indirectly interferes with the Q gene. The results reported from our work support this hypothesis as it appears that the absence of chromosome 2H^ch in tritordeum HT382 affects on the expression of the barley AP2-like gene, reducing its transcription to low levels.

The AP2-like gene from wild and cultivated barley has been characterised in this work. The AP2-like genes contain two plant-specific DNA binding motifs called AP2 domains as does the Q gene of wheat. The results confirmed that the barley AP2-like gene is located on chromosome 5H^ch. The expression of the AP2-like genes were studied in wheat, barley and tritordeum amphiploids showing that the level of transcription of the barley AP2-like gene in tritordeum was lower than in its barley parental. The chromosome substitutions 1D/1H^ch and 2D/2H^ch influence the expression of the barley AP2 in tritordeum. The results are of interest in understanding the role of the AP2-like gene in the spike morphology of cereals and in understanding the regulation of this gene in barley × wheat amphiploids. In addition, this information may be used in breeding programs for regulating the expression of AP2-like genes to modify spike characteristics and to obtain free-threshing plants.