LMW-GS as important grain storage proteins in B. distachyon
Based on the solubility in a series of solvents, plant proteins are traditionally classified into water-soluble proteins (albumins), saline-soluble proteins (globulins) and alcohol-soluble proteins (prolamins) . Different plant species generally have one predominant protein type in the grain endosperm. Laudencia-Chingcuanco and Vensel  reported that globulins were the main seed storage proteins in B. distachyon. But in their work, only 7 major protein bands were detected by SDS-PAGE and mass spectrometry, of which 6 were identified as globulins. Glutenins and gliadins were not extracted and analyzed. Recently, LMW-GS like proteins were identified by SDS-PAGE, and their masses were determined by MALDI-TOF in B. distachyon. These suggest that Brachypodium grains may contain LMW-GS like proteins.
Our results demonstrated that Brachypodium grains had similar electrophoretic compositions with common wheat in LMW glutenin subunits, In particular, B. distachyon appeared to have more abundant LMW-C type subunits than wheat (Figure 1). A total of 18 LMW-GS genes isolated from B. distachyon displayed highly homologous with those from Triticum and Aegilops species. The presence of LMW-GS in B. distachyon was further confirmed by Western blotting, Southern blotting and MALDI-TOF-MS (Table 4, Figure 2 and Additional file 2, Additional file 3, Additional file 4, Additional file 5, Additional file 6 and Additional file 7). Glutenins play important roles in the plant life cycle . Therefore, the primary role of LMW-GS in B. distachyon is probably involved in providing energy and nutrition for seed germination.
Allelic variations and gene organization at Glu-3 in B. distachyon
High homology of LMW-GS genes between Brachypodium and common wheat and related species was found in this work (Figure 4
5), including similar gene sizes and structural characteristics. This suggests that a highly conserved Glu-3 locus is present in B. distachyon. Of the 18 LMW-GS genes isolated from B. distachyon, extensive allelic variations were detected, including 34 and 9 SNPs in 5 typical LMW-m and LMW-i type genes, respectively, and 6 deletions present in the LMW-i type gene HQ220190. Particularly, both HQ220202 and HQ220204 had an extra cysteine, locating at the same position as GQ870250 and GQ870241 from Ae. markgrafii and Ae. umbellulata, respectively  (Figure 4). Thus, they probably represent an ancient type among the LMW-GS gene family. On the other hand, as in previous reports , HQ220206 only had 7 cysteine residues and a cysteine in the C-terminal II domain was changed into tyrosine because of a G→A transition. Therefore, the abundant SNP and InDel variations present in the LMW-GS genes of B. distachyon could result in different biochemical properties of their deduced protein subunits such as higher surface hydrophobicity as revealed by RP-HPLC (Figure 3).
The LMW-m and LMW-i genes isolated from B. distachyon were used to blast at the Brachypodium genome project websites (http://www.phytozome.net and http://www.brachypodium.org/). The blast results only returned one locus, Bradi3g17070, which had some sequence similarity with the LMW-GS genes cloned in this work. However, Bradi3g17070 was more similar to gliadin or avenin-like seed proteins. This suggests that the data of the Glu-3 locus encoding for LMW-GS in B. distachyon is not present in the genome sequence database.
So far, although considerable work was carried out, the precise gene organizations at the Glu-3 loci in wheat and related species are still unclear. LMW-GS can be devided to LMW-s, LMW-m, and LMW-i types according to the first amino acid residue of the mature protein, serine, methionine, or isoleucine, respectively . LMW-GS are encoded by multiple gene family at the Glu-3 loci of A, B, D chromosomes of common wheat . Some other related genomes, e.g. S, C, M, N and U also contain highly homologous Glu-3 loci [8, 9, 32]. According to previous reports, the copy numbers of the LMW-GS genes in common wheat were estimated to be 10–15 [33, 34] or 35–40 [35–37]. Wicker et al.  found that two LMW-i type genes from the A genome of Triticum monococcum that were located more than 150 kbp apart. This could facilitate the occurrence of illegitimate recombination events within the LMW-GS genes and result in novel allelic variations, such as chimeric genes . Both homologous and illegitimate recombination may occur at the Glu-3 and Glu-1 loci [9, 39], resulting in the formation of novel allelic genes. In the current study, our results demonstrated that LMW-GS encoded by the Glu-3 locus in Brachypodium also display the properties of a complex gene family as those in Triticum and related species. The numbers of copies of LMW-GS genes in B. distachyon are probably less than that in T. aestivum according to the Southern blotting analysis but similar mechanisms for generating allelic variations at the Glu-3 locus might be present in Brachypodium. Frequent SNP and InDel variations, duplications and inversions of one and more repeats, by unequal crossing over, slippage or illegitimate recombination [9, 40], could result in the allelic variations observed at Glu-3 in B. distachyon.
Phylogenetic evolutionary relationships of B. distachyon with Triticum and related species as revealed by Glu-3 loci
B. distachyon. has been shown to be much more closely related to wheat, barley and rice than to sorghum, rye or maize [19, 20]. Recent studies have shown that B. distachyon is closely related to the tribe Triticeae and Ae. tauschii, the donor of D genome of hexaploid wheat [18, 41]. The analysis of bacterial artificial chromosome (BAC) end sequences (BES) of Brachypodium genome also revealed a closer relationship between Brachypodium and Triticeae than Brachypodium and rice or maize . In the present work, phylogenetic tree based on the LMW-m type genes (Figure 5) indicated that Brachypodium is more closely related to Aegilops than to wheat, especially much closer to Ae. markgrafii (CC), Ae. umbellulata (UU), Ae. uniaristata (NN) and Ae. tauschii (DD). On the other hand, results revealed by the LMW-i type genes demonstrated that B. distachyon was closer to hexaploid common wheat and Ae. markgrafii than to Ae. uniaristata and Ae. comosa. It has been argued that the LMW-m type genes could be the progenitor of LMW-i and LMW-s type genes . The HQ220202 and HQ220204 from B. distachyon and GQ870241 from the U genome of Ae. umbellulata were clustered into a clade while GQ870250 from the C genome of Ae. markgrafii had higher similarity with GQ870241. These 4 LMW-GS genes all had 9 cysteine residues located at a highly conserved position (Figure 4), indicating that B. distachyon is more closely related to Ae. markgrafii and Ae. umbellulata. Our results further supported the recent report that the C and U genomes appear to be closely related .
Evolution of Glu-1 and Gli-1 loci in B. distachyon
Our work demonstrated that B. distachyon grains had similar compositions of albumins, globulins, and LMW-GS with these of wheat, but with fewer gliadins and HMW-GS. This suggests that, although the Glu-3 loci are highly conserved among Brachypodium, Aegilops, Triticum and other related cereal species, the Glu-1 and Gli-1 loci in B. distachyon could undergo dramatic divergence during its evolutionary process. A recent report has shown that a single copy of HMW glutenin gene with a premature stop codon was found in Brachypodium, and its structure was considered to be different from the wheat HMW glutenin gene . However, we identified a HMW glutenin subunit with a lower expression level in Bd21 by a proteomic approach  and its encoding gene has been recently cloned in our lab (data not shown). This suggests that the Glu-1 is also conserved in Brachypodium.
In common wheat, LMW-GS are encoded by the genes at the orthologous Glu-A3, Glu-B3 and Glu-D3 loci on the short arms of group 1 chromosomes (1AS, 1BS and 1DS), which are closely linked with the Gli-A1, Gli-B1 and Gli-D1 loci encoding gliadins . The physical relationships of Glu-3 and Gli-1 genes showed that gliadin or gliadin-like genes can distribute between two typical LMW-GS genes  and different types of LMW-GS can locate separately [9, 38]. This can potentially result in formation of different chimeric genes between gliadin and the LMW-GS genes due to crossing over and illegitimate recombination between or in the Gli-1 and Glu-3 loci in wheat and related species [9, 46]. Therefore, some modified LMW-GS may also be present in Brachypodium as reported in wheat .
According to our results, B. distachyon grains had very few gliadins when separated by the same extraction method as for wheat. Larré et al. also only found a few putative avenin-like proteins in B. distachyon. We speculate that most of the gliadins in B. distachyon may have evolved into LMW-GS, most likely the LMW-C subunits. This may explain why Brachypodium contained abundant LMW-GS especially the LMW-C type subunits than wheat. Recent reports strongly support this scenario: some modified LMW-GS in wheat were identified by 2-DE and MALDI-TOF-MS, which might belong to modified α/β- and γ-gliadins . It is also possible that most of the gliadin genes were pseudogenes and thus silent in mature grains due to premature stop codons as those in wheat, rice, maize and other cereals [47, 48].
Both globulin and glutenin genes are specifically expressed in seed developing tissues. Particularly, glutenin genes have a higher expression level than globulin genes [43, 49]. Since fewer HMW-GS and no gliadins are expressed in the grains, the LMW glutenin subunits with higher expression level could be important grain storage protein in Brachypodium.