Cereal asparagine synthetase genes

Abstract Asparagine synthetase catalyses the transfer of an amino group from glutamine to aspartate to form glutamate and asparagine. The accumulation of free (nonprotein) asparagine in crops has implications for food safety because free asparagine is the precursor for acrylamide, a carcinogenic contaminant that forms during high‐temperature cooking and processing. Here we review publicly available genome data for asparagine synthetase genes from species of the Pooideae subfamily, including bread wheat and related wheat species (Triticum and Aegilops spp.), barley (Hordeum vulgare) and rye (Secale cereale) of the Triticeae tribe. Also from the Pooideae subfamily: brachypodium (Brachypodium dIstachyon) of the Brachypodiae tribe. More diverse species are also included, comprising sorghum (Sorghum bicolor) and maize (Zea mays) of the Panicoideae subfamily and rice (Oryza sativa) of the Ehrhartoideae subfamily. The asparagine synthetase gene families of the Triticeae species each comprise five genes per genome, with the genes assigned to four groups: 1, 2, 3 (subdivided into 3.1 and 3.2) and 4. Each species has a single gene per genome in each group, except that some bread wheat varieties (genomes AABBDD) and emmer wheat (Triticum dicoccoides; genomes AABB) lack a group 2 gene in the B genome. This raises questions about the ancestry of cultivated pasta wheat and the B genome donor of bread wheat, suggesting that the hybridisation event that gave rise to hexaploid bread wheat occurred more than once. In phylogenetic analyses, genes from the other species cluster with the Triticeae genes, but brachypodium, sorghum and maize lack a group 2 gene, while rice has only two genes, one group 3 and one group 4. This means that TaASN2, the most highly expressed asparagine synthetase gene in wheat grain, has no equivalent in maize, rice, sorghum or brachypodium. An evolutionary pathway is proposed in which a series of gene duplications gave rise to the five genes found in modern Triticeae species.


| INTRODUCTION
Asparagine is an important nitrogen transport and storage molecule in many plant species (Lea, Sodek, Parry, Shewry, & Halford, 2007). In most plant tissues it is a major component of the free (soluble, nonprotein) amino acid pool and, together with glutamine and glutamate, it accounts for up to 70% of free amino acid content in wheat grain (Curtis et al., 2009). It reaches high concentrations during seed germination and in response to different biotic and abiotic stresses (Lea et al., 2007). It is also, of course, one of the amino acids used to make proteins.
Interest in the synthesis, accumulation and breakdown of asparagine in crop plants has been stimulated in recent years because of the discovery that free asparagine is the precursor for acrylamide formation. Acrylamide forms during the frying, baking, roasting, toasting and high-temperature processing of grains, tubers, storage roots, beans and other crop products (Mottram, Wedzicha, & Dodson, 2002;Stadler et al., 2002;Zyzak et al., 2003; reviewed by . It is classed as a group 2a carcinogen by the International Agency for Research on Cancer (IARC, 1994) and in 2015 the European Food Safety Authority (EFSA) Expert Panel on Contaminants in the Food Chain (CONTAM) concluded that the margins of exposure to dietary acrylamide indicated "a concern for neoplastic effects" (CONTAM Panel, 2015). Subsequently (April 2018), Commission Regulation (EU) 2017/2158 came into force across the European Union, introducing compulsory risk management measures that apply to all food businesses (European Commission, 2017).
The development of crop varieties with reduced acrylamideforming potential will require greater knowledge and understanding of the genetic control of asparagine metabolism. In wheat grain, free asparagine accumulates to very high levels in response to sulphur deficiency (Curtis et al., 2009;Curtis, Powers, Wang, & Halford, 2018;Granvogl, Wieser, Koehler, Von Tucher, & Schieberle, 2007;Muttucumaru et al., 2006) and poor disease control Martinek et al., 2009), while nitrogen fertilisation, in contrast to sulphur, also promotes free asparagine accumulation (Claus et al., 2006;Martinek et al., 2009). There are also substantial differences in the free asparagine concentration of grain from different wheat and rye varieties and genotypes (Curtis et al., 2010;Curtis, Powers, et al., 2018;Postles, Powers, Elmore, Mottram, & Halford, 2013). A complex network has been drawn up to describe the genes, enzymes, metabolites and environmental factors that affect asparagine metabolism (Curtis, Bo, Tucker, & Halford, 2018), but most attention to date has been focussed on the enzyme at the heart of that network, glutamine-dependent asparagine synthetase, which catalyses the transfer of an amino group from glutamine to aspartate to form asparagine and glutamate (Gaufichon, Reisdorf-Crena, Rothstein, Chardona, & Suzuki, 2010;Xu et al., 2018).
Of these, TaASN3 is the most highly expressed during early grain development, but TaASN1 and TaASN2 are the most highly expressed by mid-development, with expression of TaASN2 much higher than TaASN1 (Curtis et al., 2019;Gao et al., 2016). Indeed, TaASN2 expression in the grain at this stage is much higher than that of any of the other genes in any tissue, while its expression in other tissues is almost undetectable (Gao et al., 2016). Its expression in the embryo is much higher than in the endosperm (Curtis et al., 2019;Gao et al., 2016), with its expression in the embryo almost certainly the main determinant of free asparagine levels in the grain as a whole (Curtis et al., 2019). The A genome TaASN2 gene has been shown to be much more highly expressed than the D genome version (Curtis et al., 2019), while the same study detected no expression of the B genome gene, consistent with the observation that the B genome gene is absent in some genotypes (Xu et al., 2018).
Here we review publicly available genome data for wheat and other cereals, including bread wheat (T. aestivum), its close relatives and more distantly related species such as maize (Zea mays), sorghum (Sorghum bicolor) and rice (Oryza sativa). Phylogenetic analyses were performed to show differences in the organisation of the gene family across the cereal species and a pathway constructed to describe how the cereal asparagine synthetase gene family has evolved.

| IDENTIFYING AND COMPARING CEREAL ASPARAGINE SYNTHETASE GENES FROM PUBLICLY AVAILABLE GENOME DATA
The EnsemblPlants database (http://plants.ensembl.org/index.html) was used to provide genome information for each species, searching for asparagine synthetase genes based on annotation or by BLASTn or BLASTp searches (Altschul, Gish, Miller, Myers, & Lipman, 1990) using wheat asparagine synthetase gene nucleotide or derived amino acid sequences. The DNA sequences were then downloaded and viewed using Geneious (Geneious 10.1.3; Kearse et al., 2012), and nucleotide sequences upstream of the translation start codon and downstream of the translation stop codon were discarded. The nucleotide sequences were then aligned using Geneious Alignment on its default settings, with a cost matrix of 65% similarity using a global alignment with free end gaps, and the similarities between the genes were quantified using the nucleotide alignment matrix that was generated. All cDNA alignments were confirmed using MUSCLE alignments within the Geneious package (Kearse et al., 2012). Similarity values were generated between the asparagine synthetase genes of a single species and in comparison to the wheat genes.
The alignments were used to build trees, using Geneious tree builder software (Kearse et al., 2012), to visualise the relationship between the genes. The Jukes-Cantor model was used for genetic distance, and the tree was built via a neighbour-joining method. The Jukes-Cantor model is a simple substitution model which assumes that all bases have the same equilibrium base frequency and that nucleotide substitutions occur at equal rates (Jukes & Cantor, 1969).
The neighbour-joining method was employed as it allowed for unrooted trees to be generated quickly without assuming a molecular clock (Saitou & Nei, 1987). A resampling method was also employed. Bootstrapping (Felsenstein, 1985) was performed with 100 replicates, creating consensus trees, which were then used to examine the relationship of the genes across the cereal species. The consensus tree allowed for an estimation of the support for each clade, and a 50% threshold was applied in a majority rule consensus tree. The length of the branches in the consensus tree corresponded to the average over all trees containing the clade, with the length of the tip branches calculated by averaging over all trees. The arabidopsis (Arabidopsis thaliana) gene, AtASN1 (Lam, Peng, & Coruzzi, 1994), was used as an outgroup to anchor the trees. The scale bars on the tree represented the length of the branches and were expressed in units of substitutions per site of the sequence alignment.

| Bread wheat (T. aestivum)
The Triticeae are a tribe within the Pooideae subfamily of the family Poaceae, comprising cultivated and wild wheat, rye and barley species. The asparagine synthetase gene family of bread wheat (T. aestivum; hexaploid, genomes AABBDD) has been described in some detail. It comprises five genes per genome (Gao et al., 2016;Xu et al., 2018), with single copies of TaASN1 on chromosomes 5A, 5B and 5D, single copies of TaASN2 on chromosomes 3A, 3B (missing in some varieties) and 3D, two copies of TaASN3 (TaASN3.1 and TaASN3.2) on chromosomes 1A, 1B and 1D and single copies of TaASN4 on chromosomes 4A, 4B and 4D. These are annotated as TaASN1-TaASN5 in some datasets, but the high sequence identity, similar intron/exon pattern and close chromosomal proximity show that TaASN3.1 and TaASN3.2 are paralogues resulting from a relatively recent gene duplication (Xu et al., 2018).
A similarity matrix was generated using nucleotide sequence data for TaASN1 Figure 1.
The analysis confirmed a previous observation (Xu et al., 2018) that variety Chinese Spring lacks a B genome TaASN2 gene. Analysis of variety Spark and a doubled haploid line, SR3, from a Spark × Rialto mapping population also found no evidence of expression of a B genome TaASN2 gene (Curtis et al., 2019), suggesting that the gene might be missing in those genotypes as well. However, the variety Cadenza genome data, also available from EnsemblPlants, do contain a B genome TaASN2 gene, so the absence of the B genome TaASN2 gene is not universal in modern bread wheat genotypes.
dicoccoides, and commonly as emmer or hulled wheat, is a tetraploid (genomes AABB). It is widely regarded as the wild progenitor of cultivated Triticum durum (pasta wheat) and Triticum dicoccum (also known as T. turgidum subsp. durum and T. turgidum subsp. dicoccum), the main difference between the wild and cultivated species being that the ripe seed heads of the wild wheat shatter to spread the seed, whereas the seed heads of the cultivated species do not. Emmer wheat is also considered to be one of the progenitors of bread wheat, with bread wheat forming from a hybridisation event between T. dicoccoides (AABB genomes) and Aegilops tauschii (DD) around 8,000 years ago (Matsuoka, 2011). Along with einkorn wheat (Triticum urartu), emmer wheat was one of the first crops to be domesticated and it was widely cultivated in the ancient world.
The T. dicoccoides accession used for the genome assembly was Zavitan, because of the availability of pre-existing genetic data for this genotype (WEWSeq v.1.0, INSDC Assembly; Avni et al., 2017). Nine asparagine synthetase genes were identified in the database, of which genes with Ensembl reference numbers TRIDC5AG025640 and TRIDC5BG026790 were located on chromosomes 5A and 5B, respectively, TRIDC3AG009140 was located on chromosome 3A, with no homeologue on chromosome 3B, TRIDC4AG015200 and TRIDC4BG033760 were located on chromosomes 4A and 4B, respectively, while TRIDC1AG056280 and TRIDC1AG062090 genes were located on chromosome 1A, with TRIDC1BG064720 and TRIDC1BG071210 on chromosome 1B.
These genes fall into the same groups as the TaASN genes from the A and B genomes (Table S1, Similarity Matrix 1.2), and this was confirmed by phylogenetic analysis (Figure 1). We therefore annotate TRIDC5AG025640 and TRIDC5BG026790 as TdiASN1, TRIDC3A G009140 as TdiASN2, TRIDC1AG056280 and TRIDC1BG064720 as TdiASN3.1, TRIDC1AG062090 and TRIDC1BG071210 as TdiASN3.2 and TRIDC4AG015200 and TRIDC4BG033760 as TdiASN4 (Table 2; using the Tdi prefix to distinguish the T. dicoccoides genes from the T. durum genes described below). The fact that the B genome TdiASN2 gene is missing is somewhat unexpected given that, as stated above, some hexaploid wheat varieties do have this gene.
When the bread wheat genes were added into the alignment matrix (Table S1, Similarity Matrix 1.3), the shared identity reflected the close evolutionary history between the two species and the evolutionary constraints on asparagine synthetase function. For example, there was 100% identity between TdiASN2 (TRIDC3AG009140) and the chromosome 3A version of TaASN2, and 96.5% identity with the chromosome 3D TaASN2.
It is widely grown in the Middle East and southern Europe (Tedone, Ali, & de Mastro, 2017 Tables 2-4

| Goat grass (Aegilops tauschii)
Aegilops tauschii, commonly called Tausch's goat grass or rough-spike hard grass, is the diploid progenitor of the D genome of bread wheat (McFadden & Sears, 1946). The genome assembly uses the subspecies  (Table S1, Similarity Matrix 1.8). These genes can therefore be annotated as AetASN1 and AetASN2, respectively (

| Einkorn wheat (T. urartu)
Triticum urartu, commonly known as red wild einkorn (German einkorn, meaning single grain) wheat, is the diploid progenitor of the A genome of cultivated wheat (Dvořák, Terlizzi, Zhang, & Resta, 1993), and its genome was analysed initially to provide information on the evolution of hexaploid bread wheat (T. aestivum) and tetraploid durum wheat (T. durum). The accession G1812 was used for the genome analysis, but to date no chromosomes have been assembled from the data, with scaffolds presented instead (ASM34745v1, INSDC Assembly; Ling et al., 2013), so chromosomal positioning information was unavailable.
Initial assessment of the asparagine synthetase genes showed a surprisingly high number of 8 (  Figure 1). That leaves TRIUR3_27580 as the TaASN3.1 homologue; it only showed 81.6-84.6% identity with TaASN3.1, which was lower than expected, but it did cluster with TaASN3.1, TdiASN3.1 and TduASN3.1 in the phylogenetic analysis (Figure 1).

| Barley (Hordeum vulgare)
Barley was domesticated alongside wheat in the Fertile Crescent over 10,000 years ago (Badr et al., 2000). It is a diploid close relative of wheat and, like wheat, its haploid genome comprises seven chromo- Five HvASN genes were identified from the genome data (Table 3 and  Again, this is consistent with HvASN2 being expressed grain-specifically, but to date there are no data to support this.

| Rye (Secale cereale)
Rye is a diploid member of the Triticeae which diverged from the ancestors of wheat relatively recently. Genome data were obtained through the whole-genome sequencing of the winter rye inbred line, Lo7 (Bauer et al., 2017). The data in that resource are presented as scaffolds, with no chromosome assemblies. Five asparagine synthetase genes were identified, one each on scaffolds 370516, 174491, 3445 and 527072, with a fifth located across three scaffolds (1245, 81708 and 36651).
From the initial BLAST search, the scaffold 370516 gene was annotated as ScASN1, the scaffold 174491 gene as ScASN2, the scaffold 3445 and the tri-scaffold gene as ScASN3 and the scaffold 527072 gene as ScASN4. When aligned with each other (Table S1, Matrix 1.13), the highest similarity was seen between the two ScASN3 genes, at 92.6%. This was followed by the ScASN1 and ScASN2 genes at 83.5% similarity, with the ScASN4 gene showing the lowest similarity to the other genes at 63.5-69.3%.
The annotations were confirmed when the bread wheat genes were added to the alignment matrix (

| Comparison and phylogenetic analysis of all the Triticeae genes
A similarity matrix was constructed for all of the genes from the Triticeae species (Table S1, Matrix 1.15) and a phylogenetic tree of all the genes was constructed (Figure 1). This confirmed that the asparagine synthetase gene families were the same in all the species, with five genes per genome, assigned to four groups.

| BRACHYPODIUM (BRACHYPODIUM DISTACHYON)
Brachypodium is a model species for the cereals, and particularly the Pooideae subfamily, although it is usually classified in its own tribe of Brachypodiae. Its common name is purple false brome, but in science circles nowadays it is commonly just called brachypodium and we will continue to use that name. It is a diploid, with a small genome of approximately  showed 84.4 and 84.7% identity, respectively, with the A and D genome TaASN2 genes.
The brachypodium data suggest that the relatively recent gene duplication event that gave rise to separate ASN1 and ASN2 genes, and the even more recent event that gave rise to the ASN3.1 and ASN3.2 genes did not occur in that species, and therefore presumably occurred in the Triticeae after brachypodium diverged from that line. This was confirmed by phylogenetic analysis (Figure 2). Brachypodium therefore has BdASN1, BdASN3 and BdASN4 genes (Table 3 and Figure 2).
Transcriptomic data for developing brachypodium grain (Davidson et al., 2012) shows BdASN3 to be the most highly expressed asparagine synthetase gene during early development, with BdASN1 the most highly expressed in the embryo by mid-development, with expression in the embryo much higher than in the endosperm. This pattern of expression is similar to that of bread wheat (Curtis et al., 2019;Gao et al., 2016) with the exception, of course, that the most highly expressed gene in wheat grain, TaASN2, does not have an equivalent in brachypodium.

| THE PANICOIDEAE: SORGHUM AND MAIZE
Sorghum (S. bicolor) is a widely grown cereal crop, able to grow under harsher conditions than most other cereal crops, including its close relative, maize (Z. mays). Its grain is used to make flat breads that form the staple food of many cultures, and it can also be used to make bioethanol. It is diploid, with a haploid chromosome number of 10.
The sorghum genome data utilised were from the  (Table 4). Within the group, SORBI_3001G406800 and SORBI_3005G003200 showed the highest nucleotide sequence identity at 70.9%, with less than 65% identity seen between SORBI_3010G110000 and the other two genes (  Tables 2-4 and S1. The branch labels show the consensus support in the clade from bootstrapping, shown as a percentage. The scale bar represents the length of the branches, expressed in units of substitutions per site of the sequence alignment. The arabidopsis (Arabidopsis thaliana) gene, AtASN1, was used as an outgroup to anchor the tree. Note that the maize and sorghum group 1 genes actually cluster with the group 4 genes, but are closely related to both groups and are assigned to group 1 based on similarity matrices (Table S1, Similarity Matrices 2.4 and 2.6) copies of ASN3 did not occur in the Panicoideae (Figure 2). The analysis also showed the sorghum group 1 gene, SbASN1, to have diverged from the group 4 genes much less than the brachypodium and Triticeae group 1 and 2 genes; indeed, the phylogenetic analysis placed it marginally in the group 4 cluster (Figure 2), although we continue to regard it as a group 1 gene based on the similarity matrix.
Despite some differences in gene positioning, therefore, the sorghum and maize genes showed a high level of similarity.
SbASN3 has been shown to be the most highly expressed asparagine synthetase gene overall in sorghum seeds, but SbASN1 is the most highly expressed in the embryo (Davidson et al., 2012;Makita et al., 2015). Indeed, while SbASN3 and SbASN4 are expressed throughout the plant, SbASN1 is embryo-specific (Davidson et al., 2012), meaning that it has an even narrower tissue-specific expression than TaASN2 in bread wheat. The group 3 gene of maize (annotated as ZmASN1) is also the most highly expressed asparagine synthetase gene in the grain of maize Zhan et al., 2015), but it is also highly expressed in leaves (Baute et al., 2015), seedlings (Chang et al., 2012) and 5-day-old caryopses (Pang et al., 2019). One of the group 4 genes (annotated as ZmASN3) is also highly expressed in 5-day-old caryopses (Pang et al., 2019). The group 1 gene (annotated as ZmASN2) was found to be expressed at very low levels in all of these studies, but so far data are not available on its expression later on in seed development. indica and japonica are both diploids (AA genome).
Two ASN genes were identified in reference genomes for both indica and japonica subspecies (Table 4) Yu et al., 2002), one located on chromosome 3 (BGIOSGA010942 and Os03g0291500 for indica and japonica, respectively), which has previously been annotated as OsASN1 by the Rice Annotation Project (Sakai et al., 2013), and the other on chromosome 6 (BGIOSGA021489 and Os06g0265000 for indica and japonica, respectively), previously annotated as OsASN2 (Sakai et al., 2013). The BGIOSGA021489 (indica) and Os06g0265000 (  Of the two genes that rice does have, the group 3 gene (annotated as OsASN2) has been shown to be the more highly expressed in seeds, shoots and the root tips of developing seedlings (Davidson et al., 2012;Reynoso et al., 2018;Sakai et al., 2011;Zhang et al., 2014), while the group 4 gene (annotated as OsASN1) shows greater expression in anthers and carpels (Zhang et al., 2014). No data are available on expression in developing seeds.

| ASSIGNMENT OF ALL THE ASPARAGINE SYNTHETASE GENES TO GROUPS
Confirmation of the assignment of the different genes to groups 1-4, with group 3 subdivided into 3.1 and 3.2, was achieved by compiling a phylogenetic tree of all the genes. The groups were numbered to fit the annotation of the wheat genes, with TaASN1 in group 1 and so on. The tree is shown in Figure 3. The analysis confirmed that the family structure of five genes, in groups 1, 2, 3 (separated into 3.1 and 3.2) and 4, is unique to the Triticeae tribe, with brachypodium, maize and sorghum lacking a group 2 gene and having only one group 3 gene TaASN1 and TaASN2 were named as such because they were the first asparagine synthetase genes to be identified in wheat (Wang, Liu, Sun, & Zhang, 2005). The same is true for the maize and rice genes, Note: The maize genes are annotated ZmASN1 to ZmASN4 as previously (Todd et al., 2008), but are assigned to groups based on those identified in the other species. Maize also contains a number of partial genes that are not included. The rice genes, which have been annotated previously as OsASN1 and OsASN2 (Sakai et al., 2013), are assigned to groups 4 and 3, respectively.
but the maize gene annotated as ZmASN1 is a Group 3 gene and ZmASN2 is a group 1 gene, while ZmASN3 and ZmASN4 are both group 4 genes, with no genes in group 2 (Figures 2 and 3). For rice, OsASN1 is a group 4 gene while OsASN2 is a group 3 gene, with no genes in groups 1 or 2. Genes of the other species that were identified had not been annotated previously, and we have assigned names based on the groupings.
We propose that the evolutionary development of the gene family began with an initial gene duplication that gave rise to the ancestors of the group 3 and group 4 genes ( Figure 4) Figures 2 and 3).
We used arabidopsis gene AtASN1 (Lam et al., 1994) to anchor the trees generated in our analyses. Arabidopsis actually has three asparagine synthetase genes (AtASN1, AtASN2 and AtASN3) (Lam, Hsieh, & Coruzzi, 1998). Gaufichon et al. (2010) proposed that asparagine synthetase genes could be divided into two classes, I and II, with AtASN1 in class I and AtASN2 together with AtASN3 in class II. Duff (2015) proposed a third class, class III, based on a phylogenetic analysis of asparagine synthetase genes from both monocotyledonous and dicotyledonous species in which a cluster of cereal genes emerged as a separate clade. That class corresponds to group 3 in our analysis, while class II roughly corresponds to our group 4. Rice is shown as having class II and III genes (groups 3 and 4 in our analysis) but no group 1. However, the barley genes grouped differently to our analysis, with three of them shown as group 1. This may be because nucleotide sequence data for many more cereal genes are now available. It is now clear that the cereal gene family has evolved considerably since the divergence of monocots and dicots and does not fit well within a broader classification that attempts to include all of the plant genes. For example, in our analyses it proved unhelpful to include more than one arabidopsis gene in the phylogenetic trees because the arabidopsis genes did not cluster with the cereal groups (not shown).  Figure 2 the maize and sorghum group 1 genes actually cluster with the group 4 genes, but are closely related to both groups and are assigned to group 1 based on similarity matrices (Table S1, Similarity Matrices 2.4 and 2.6) Duff (2015) proposed that the three classes of asparagine synthetases emerged before the divergence of monocots and dicots, but that many species lost some genes and saw duplications in others. If this model were correct it would mean that rice once had an OsASN1 gene but lost it. However, this conclusion is arrived at again by trying to fit the cereal genes into a broader classification that no longer looks convincing to us. We propose that the evolution of plant asparagine synthetase genes into the gene families seen today occurred after the separation of monocots and dicots, and that the model shown in Figure 4 is a better fit for the cereal data now available.
The presence of a B genome ASN2 gene in some hexaploid wheat (T. aestivum) varieties (genomes AABBDD) and in pasta wheat (T. durum; genomes AABB) but the absence of a B genome ASN2 gene in emmer wheat (T. dicoccoides; genomes AABB) raises questions about bread and pasta wheat ancestry. If emmer wheat is the donor of the A and B genomes of bread and cultivated pasta wheat, then all hexaploid wheat varieties and pasta wheat might be expected to have inherited the B genome ASN2 deletion. In our view, the most likely explanation is that some emmer wheat genotypes have a B genome ASN2 gene while some lack it, and that the hybridisation event that produced bread wheat occurred more than once, involving B genome donors with and without a B genome ASN2 gene. Another possible but in our view less likely explanation is that the B genome TdiASN2 gene was present when the hybridisation event that produced hexaploid wheat occurred, but that emmer wheat and some hexaploid wheat genotypes subsequently lost it. A third possibility is that some bread and pasta wheat genotypes regained a B genome ASN2 gene through an introgression from a related species, such as a wild wheat or rye.
Another important question, given the link between free asparagine concentrations, acrylamide formation during baking and processing, and food safety, is the implications that differences in the asparagine synthetase gene family have for free asparagine concentrations in the grain of the different species. In bread wheat, TaASN2 is expressed specifically in the grain, with much higher expression in the embryo than the endosperm (Curtis et al., 2019;Gao et al., 2016), and its expression in the grain at mid-development is far higher than the expression of any of the genes in any other tissue. Biochemical analyses have shown little difference in activity of the ASN1 and ASN2 enzymes (Xu et al., 2018), making the TaASN2 gene almost certainly the most important in determining the amount of free asparagine that accumulates in wheat grain. The A genome TaASN2 is much more highly expressed than the D genome gene (Curtis et al., 2019), but the relative expression of the B genome gene when it is present is not yet known.
Rice lacks homologues of both TaASN1 and TaASN2, but little information is available on the concentration of free asparagine in rice grain. The fact that it only has two genes may mean that there is less scope for genetic interventions to reduce free asparagine accumulation: T-DNA insertion mutants and CRISPR-Cas edited lines lacking a F I G U R E 4 Diagram representing the evolution of asparagine synthetase genes in the Triticeae, Brachypodiae, Panicoideae and Ehrhartoideae functional version of the gene that we annotate as OsASN4, for example, showed effects on plant height, root length, and tiller number compared with wild type (Luo et al., 2019). Note that this gene was annotated as OsASN1 by the authors of that study. Maize also lacks a group 2 gene, but it does contain a group 1 gene. Expression analyses of the maize gene family have shown the group 1 and group 3 genes (previously annotated as ZmASN2 and ZmASN1, respectively) to be expressed in the seed, but expression in other tissues is higher (Todd et al., 2008). In other words, the high asparagine synthetase gene expression in the grain of wheat, accounted for mainly by TaASN2, has not been observed in maize. In addition, biochemical analyses have shown the reactions catalysed by the wheat asparagine synthetases, ASN1 and ASN2, to proceed much more rapidly than has been reported for the maize enzymes (Duff et al., 2011;Xu et al., 2018).
However, a study that measured free asparagine concentrations in a range of cereal species found maize to be lower than rye but higher than barley, pasta wheat or bread wheat (Žili c et al., 2017).
Clearly, more research is required on the role of different asparagine synthetase genes in controlling the concentration of free asparagine in maize and rice grain, and more data are required generally on the concentrations of free asparagine in different cereal grains. Eighteen years after the discovery of acrylamide in food and the identification of free asparagine as its precursor, the data on free asparagine concentrations in these major food crops are surprisingly sparse.
These data will be required along with much more data on and improved understanding of the tissue-specific and developmental expression patterns of different gene family members in the various species before relationships can be drawn between gene family structure, expression patterns and relative free asparagine concentrations.