Characterization of Aldehyde Oxidase (AO) Genes Involved in the Accumulation of Carotenoid Pigments in Wheat Grain

Aldehyde Oxidase (AO) enzyme (EC 1.2.3.1) catalyzes the final steps of carotenoid catabolism and it is a key enzyme in the abscisic acid (ABA) biosynthesis. AO isoforms are located in the cytosolic compartment of tissues in many plants, where induce the oxidation of aldehydes into carboxylic acid, and in addition, catalyze the hydroxylation of some heterocycles. The goal of the present study was to characterize the AO genes involved in the accumulation of carotenoid pigments in wheat grain, an important quantitative trait controlled by multiple genes. The cDNAs corresponding to the four AO isoforms from Arabidopsis thaliana and five AO isoforms from Brachypodium distachyon were used as query in 454 sequence assemblies data for Triticum aestivum cv. Chinese Spring (https://urgi.versailles.inra.fr/blast/blast.php) to obtain the partial or whole orthologous wheat AO sequences. Three wheat isoforms, designated AO1, AO2, and AO3 were located on the chromosome groups 2, 5, and 7, respectively, and mapped on two consensus wheat maps by SNP markers located within the AO gene sequences. To validate the possible relationships between AO3 genes and carotenoid accumulation in wheat, the expression levels of AO-A3 and AO-B3 gene were determined during the kernel maturation stage of two durum wheat cultivars, Ciccio and Svevo, characterized by a low and high carotenoid content, respectively. Different AO-A3 gene expression values were observed between the two cultivars indicating that the AO-A3 allele present in Ciccio was more active in carotenoid degradation. A gene marker was developed and can be used for marker-assisted selection in wheat breeding programs.


INTRODUCTION
Yellow pigment concentration (YPC) in wheat is a quantitative trait controlled by a complex genetic system and influenced by environmental factors (Qin et al., 2016). The yellow color of grain and flour is mainly due to carotenoid accumulation in the pericarp and endosperm. Carotenoids are precursors of the vitamin A, with high nutritional relevance for human diet (Della Penna and Pogson, 2006;Britton, 2009), and substrates for the synthesis of apocarotenoids, compounds derived from oxidative cleavage and further modifications (Wurtzel et al., 2012). Apocarotenoids include retinol (vitamin A) and the hormones abscisic acid (ABA) and strigolactones (Rosati et al., 2010).
Oxidative cleavage enzymes involved in carotenoid degradation, sequestration, and storage can also influence the accumulation of carotenoid pigments in tissues (Gonzalez et al., 2013). β-carotene and some of its derivatives can be modified by carotenoid cleavage dioxygenases (CCDs) resulting in the production of strigolactones (Alder et al., 2012). Moreover, a family of carotenoid 9-cis-epoxycarotenoid dioxygenases (NCEDs), which catalyzes the cleavage of carotenoids at specific double bonds, was shown to be active in the production of apocarotenoids (Auldridge et al., 2006). In particular, the ABA, derived from the conversion of violaxanthin by aldehyde oxidase (AO), is involved in regulating plant responses to various environmental stresses (e.g., drought and salinity stress) and long-distance signaling within the plants (Davies et al., 2005; Figure 1).
The objectives of the present study were to: (a) identify the AO gene family in wheat and map each member on two wheat high-density consensus maps; (b) characterize AO3 transcription level by qRT-PCR in kernel tissue of two durum wheat cultivars differing for carotenoid content; (c) develop AO3 markers suitable for MAS to be used in breeding programs.

Plant Materials
A recombinant inbred line (RIL) mapping population, previously developed from the cross Svevo × Ciccio and genotyped using the SNP iSelect array (Colasuonno et al., 2014), was evaluated for the abscisic aldehyde oxidase 3 (AO-A3) gene and yellow pigment content. The elite durum wheat cultivars Svevo and Ciccio were different for qualitative and quantitative traits, such as grain yield components, grain protein content, grain yellow pigments and adaptive traits. Regression analysis, carried out in the Svevo x Ciccio RIL population grown in four environments (Valenzano 2006, Foggia 2006, Valenzano 2007, and Foggia 2007 and implemented in QGene, was used to underlined the association between the abscisic aldehyde oxydase 3 (AO-A3) gene and yellow pigment content. The marker-trait association was considered significant when -log10(P) ≥ 3.0. Phenotypic variation explained by the gene marker (R 2 ) and additive effects were investigated. Positive and negative signs on the effect values indicated the contribution of Svevo and Ciccio, respectively, toward higher trait value.
The three wheat AO sequences included in the contigs 2AL_6408908, 5BL_10864426 and 7AL_4455104 were blasted against the available dataset of SNP marker sequences reported by Wang et al. (2014), and SNPs with ≥ 80% identity were considered within the AO genes. The high-density linkage maps described by Maccaferri et al. (2015) for durum wheat and by Wang et al. (2014) for common wheat were used as reference maps for determining a detailed map position of AO genes.
(http://www.softberry.com/berry.phtml?topic=fgenesh&group= programs&subgroup=gfind). Each predicted protein was considered into the next step of analysis. Further BLASTn analysis for three representative isoforms was extended to Wheat 61k GeneChip in PLEXdb database (http://www.plantgdb.org) for obtaining information on each transcription pattern variation during different wheat developmental stages (Experiment TA3).

Quantitative Real-Time PCR
In order to pick primer combinations to be used for AO3 gene expression analysis, cDNA sequences of AO-A3 (included in the contig 7AL-4455104) and AO-B3 (accession number Ta_AK331622 from NCBI database) were aligned to highlight dissimilarities between the two homoeologous genes. Specific primer pairs for each gene were designed in a region of the second exon since a number of polymorphisms were detected between the A and B genomes (Table S1).
The genetic material used for the AO expression analysis was represented by the durum wheat cultivars Ciccio and Svevo characterized by low and high values of YPC, respectively. Kernel tissues from each cultivar were harvested, frozen in liquid nitrogen and stored at −80 • C until RNA extraction. Total RNA was extracted with RNeasy Plant Mini Kit (QIAGEN R ) and checked on 1.5% denaturing agarose gel. All RNA samples were lead to the same concentration (1 µg) for subsequent treatment with DNase I recombinant (Roche Applied Science, Mannheim, Germany), in order to remove genomic DNA, and then were reverse-transcribed into double stranded cDNA with Trascriptor First Strand cDNA Synthesis Kit (Roche Applied Science, Mannheim, Germany). Data were normalized using three reference genes: Cell Division Control AAA-Superfamily of ATPases (CDC), ADP-Ribosilation Factor (ADP-RF), and RNase L Inhibitor-like protein (RLI) (Paolacci et al., 2009;Giménez et al., 2011). These genes have a stability value around 0.035 when evaluated with NormFinder software (Andersen et al., 2004).
Quantitative Real-Time PCR analyses were carried out using EVA R GREEN in the CFX96TM Real time PCR Systems (BIO-RAD). The PCR cycle was: 95 • C for 3 min followed by 40 cycles of 95 • C for 10 s and at 60 • C for 30 s. Amplification efficiency (98% to 100%) for the primer set was determined by amplification of cDNA with a series of six scalar dilution (1:5) per reaction. Each 10 µl PCR reaction contained 1 µl of a 1:5 dilution cDNA, 5 µl of EvaGreen Mix 10X (Bio-Rad), and 500 nM of each primer. All experiments were performed in Hard-Shell 96-well skirted PCR plates (HSP9601) with Microseal R "B" Adhesive Seals (MSB-1001) from Bio-Rad. Fluorescence signals were collected at each polymerization step. The specificity of the amplicons was confirmed by the presence of a single band of expected size for each primer pair in agarose gel (2% w/v), by a single peak melting curves of the PCR products and by sequencing of the amplified fragments (3,500 Genetic Analyzer, Applied Biosystems). qRT-PCR data for both genes were derived from the mean values of three independent amplification reactions carried out on five different plants harvested in the same phenotypic stage (biological replicates). All calculations and analyses were performed using CFX Manager 2.1 software (Bio-Rad Laboratories) using the C t method, which used the relative quantity (RQ) calculated with a ratio of the RQ of the target gene to the relative expression of the reference gene (including the three reference targets in each sample). Standard deviations were used to normalize values for the highest or lowest individual expression levels (CFX Manager 2.1 software user manual, Bio-Rad Laboratories). The Student's t-test was used to underline significant differences between control and treated samples for the two considered AO3 genes.
PCR products derived from the durum cvs. Ciccio and Svevo were loaded on agarose gel to confirm amplicon length, and injected onto the WAVE R system (Transgenomic, Omaha, NE, USA) to undergo check peak intensity of the amplicons. The temperature required for successful resolution of heteroduplex molecules was determined using the DHPLC Navigator TM Software (Transgenomic, Omaha, NE, USA), which considered the specific amplicon sequence and size to calculate the denaturation curve. PCR product was cleaned using ExoSAP-IT (USB, Cleveland, OH, USA) according to manufacturer's instructions, then sequenced using BigDye terminator sequencing kit on a ABI-3500 Genetic Analyzer (Applied Biosystems, Foster City, CA, USA). The detection of molecular marker was performed on DHPLC device in "mutation detection" under "Rapid DNA" run mode.

Isolation of Genomic Sequences of AO Genes in Wheat
The cDNAs corresponding to the four AO isoforms from A. thaliana (AT5G20960, AT3G43600, AT2G27150, and AT1G04580, designated, respectively O1, AO2, AO3, and AO4) and the five AO isoforms from B. distachyon (XM010230033, XM_003557870, XM_003559293, and XM_003559295, all designated AO2, and XM_003561213 (designated AO) were used as query in 454 sequence assemblies data for T. aestivum cv. Chinese Spring (https://urgi.versailles.inra.fr/blast/blast.php) to obtain the partial or whole orthologous wheat AO sequences. The search into the Chinese Spring sequence database released several AO sequence fragments included in 13 different contigs located on the wheat chromosome groups 2, 5, and 7 ( Table 1). The definitive assignment of AO sequences to the wheat A, B and D homoeologous chromosomes and the accurate map position was performed based on the best blastn hit (percentage identity) with the available dataset of SNP marker sequences reported by Wang et al. (2014) ( Table 1). The Recommended Rules for Gene Symbolization reported in the Wheat Catalog (McIntosh et al., 2005) were used for AO nomenclature. In particular, six SNPs corresponding to AO1 genes mapped on chromosome group 5 (AO-B1 and AO-D1), 12 SNPs within AO2 mapped on chromosome group 2 (AO-A2, AO-B2, and AO-D2), and 16 SNP markers mapped on chromosome group 7 (AO-A3, AO-B3, and AO-D3). Out of 34 SNPs corresponding to AO gene sequences, five and nine markers were located on the consensus durum (Maccaferri et al., 2015) and bread wheat maps (Wang et al., 2014), respectively (Figure 2).

Characterization of Wheat AO3 Gene Sequences
The genomic sequence and structure of AO-A3, AO-B3, and AO-D3 genes was investigated in the cv Chinese Spring in order to study the possible relationships between AO3 genes and carotenoid accumulation in the wheat grain (Colasuonno et al., 2014). Relying on the Softberry calculation regarding the 7AL-4455104 contig, the AO-A3 genomic sequence was 6,749 bp long with 42% GC content. The predicted gene sequence included an mRNA of 3,786 bp and a protein length of 1,262 amino acids. For AO-B3, the partial sequence included in the contig 7BL_6713884 was implemented by the complete EST sequence AK331622 (from NCBI database) and allowed to obtain a cDNA of 4,279 bp. The prediction of AO-D3 gene, considering the 7DL_3391726 contig, presented a genomic sequence of 6,215 bp with an mRNA length of 4,038 bp and a protein 1,346 amino acids. A similar intron/exons structure was predicted between the wheat AO3 genes composed by 10 exons and 9 introns sharing an identity of > 80% among homoeologues and 97% between AO-B3 and AO-D3. The Fgenesh++ gene prediction pipeline highlighted a lack of similarity for the first and last exons between AO-A3 and AO-B3/AO-D3. Only the last exon displayed high identity (100%) between AO-B3 and AO-D3 isoforms.
Furthermore, the wheat AO3 cDNA alignment showed a smaller region (81 bp) in exon 6 for AO-A3 than in exon 6 for AO-B3 and AO-D3, and a corresponding longer region in the adjacent intron. This difference suggested an alternative splicing site and the sequencing analysis of cv. Chinese Spring cDNAs confirmed this length polymorphism among genomes.
BLAST analysis using Phytozome v.7 software (http://www.phytozome.net) with B. distachyon and O. sativa genomes allowed the comparison of wheat AO3 genes with the orthologous genes located on chromosome 1 of Brachypodium (locus name: Bradi1g52740) and chromosome 7 of rice (locus name: Os07g18154). The Brachypodium AO consist in a sequence of 12,336 bp with a CDS of 4,053 bp, whereas in rice genome AO had a sequence of 12,959 bp with a CDS of 2,535 nucleotides. Comparison between wheat and Brachypodiun genomic AO sequences showed identities of about 84%, and an identity of 89% between the two CDS. A similar intron/exons structure was observed between AO3-A1 from wheat and AO from Brachypodium composed by 10 exons and nine introns, except for the first and last exons (Figure 3). An identity of 79% was found aligning the AO-A3 genomic sequence with the rice AO sequence.
Further investigation through the Plant MITE database revealed the presence of two transposon regions. The first one was located at 336 bp and belonged to MITE family DTC, the other one at 901 bp position corresponded to MITE family DTT. The promoter region sequences in the cvs Ciccio and Svevo relieved absence of significant polymorphisms. According to this, the differences on the expression levels among cultivars could be due to a different regulatory mechanisms in action inducing different genes. 1 | List of aldehyde oxidase genes in wheat with corresponding contig number, SNP markers, chromosome localization and map position on the durum (Maccaferri et al., 2015) and bread wheat (Wang et al., 2014)

Expression Profile of AO Genes in Wheat
The possible relationships between AO genes and carotenoid accumulation was investigated in the wheat grain. The AO-A3 and AO-B3 gene expression levels were determined in two durum wheat cultivars, Ciccio and Svevo, characterized by a low and high YPC, respectively. Total RNA was extracted from kernels, and quantitative real-time PCR was conducted with specific primers in order to analyze individually the two homeologous AO3 genes. High expression levels were observed for AO-A3 and AO-B3 during seed maturation stage in Ciccio, while low amounts were detected in Svevo (Figure 4). Significant different expression values (P < 0.001, t-test) were observed between the two cultivars only for AO-A3. The data suggested that the AO-A3 allele present in Ciccio (cultivar characterized by a low content of YPC) was more active into carotenoid degradation, while the Svevo (high YPC) allele was not fully involved in the carotenoid catabolism.
In order confirm our data and to understand the expression trend of wheat AO genes, a check on data available in the PLEXdb database was carried out considering the cDNAs included in the largest contigs (2AL_6408908, 5BL_10864426, and 7AL-4455104) chosen as representative of each wheat homoeologous group (Figure 5). The Wheat 61k GeneChip showed different expression pattern of AO genes during developmental stages FIGURE 2 | Schematic representation of the durum wheat linkage map and AO markers. Each linkage map derives from the durum consensus map (Maccaferri et al., 2015) and has been represented by a SNP marker every about 20 cM. SSR markers have been also inserted every about 20 cM to compare the consensus SNP map with published SSR-based maps.
Frontiers in Plant Science | www.frontiersin.org FIGURE 3 | Comparison of AO3 gene structures in rice, Brachypodium, and wheat is shown based on colored boxes highlighting conserved exons. Intron and exon sizes are shown as well as the whole gene (in brackets the total length). Rice and Brachypodium AO share the same structure with ten exons of conserved sizes and nine introns. Brachypodium and wheat AO3 show an high similarity in sequence and structure for eight exons. Black dashed line indicates the absence of intron sequence since only a cDNA sequence has been found for AO-B3. underlining how AO2 and AO1 resulted constantly expressed in all stages with maximum levels of 8.93 RMA normalization for AO2 in seedling root (phase 4). Instead, AO3 showed significantly high expression levels (values higher than the mean values ± 2 SD) especially in last phases of embryo and endosperm kernel formation (phase 9, 12, and 13) indicating a major role in the last developmental stage of wheat seeds.

Development of a Functional AO3 Marker
A considerable emphasis has been placed on the AO-A3 gene for the development of molecular markers to be used for MAS in wheat breeding. Due to the availability of SNP data corresponding to this isoform, the IWB59875 marker mapped on chromosome 7A and resulted associated to carotenoid content (Colasuonno et al., 2014) was chosen for setting up a SNP-based method suitable for wheat breeding. Before proceeding into the analysis of DNA fragments, PCR primers specific for AO-A3 gene were designed respecting the DHPLC conditions, such as the distance of 50 bp from the SNP site and a fragment size ≥ 200 and ≤ 800 bp. After that, the DHPLC technique was optimized for the run conditions on each amplicon, considering the optimal temperature (55 • C). According to the basic principle of the DHPLC technique, if the sample contained the DNA from a homozygous genotype, the chromatogram showed a single peak derived from only homoduplex molecules. The homoduplex originated from Ciccio was practically undistinguishable from that of Svevo, having the same retention time. On the contrary, when the sample was a mix of two cultivars, and a SNP was present, the chromatogram exhibited two peaks of similar area, one corresponding to the coeluting homoduplex molecules (imputable to both Ciccio or Svevo with themselves), the other to the early elution of the heteroduplex molecules (due to Ciccio DNA combined with that of Svevo). As shown in Figure 6, the DHPLC allowed detecting two chromatographic peaks corresponding to heteroduplex and homoduplex molecules derived by the presence of the T/C substitution (detected by the IWB59875 marker) in the two durum varieties. The regression analysis conducted between the AO-A3 gene marker and YPC confirmed the association with the trait with an high LOD scores in all the four environment considered and in the mean across them. The phenotypic variation explained by the gene marker (R 2 ) ranged from 18.0 to 41.0% and the effect due to the Ciccio alleles was reported for all the environment analyzed in Table 2. The asterisk signals indicate, respectively the values higher or lower than the mean values ± 2 SD.
FIGURE 6 | Optimization of DHPLC analysis for SNP detection between cvs. Ciccio and Svevo at locus IWB59875. The chromatograms correspond to the elution profiles of homoduplex molecules (cv Ciccio in black, cv Svevo in dark blue), and of homoduplexes plus heteroduplexes derived from 1:1 mixed DNA (pink line).

DISCUSSION
High levels of carotenoid pigments in wheat kernels have important positive implications for human health since they are antioxidant compounds and precursor of vitamin A. Knowing the role of the main carotenoid genes in wheat and the specific alleles present in each cultivar can allow to develop superior cultivars through marker-assisted breeding programmes.
Several studies reported many QTL for carotenoid content spread all over the wheat genome (reviewed by Colasuonno et al., 2017). Among them, two different QTL were mapped on the long arm of chromosome group 7, co-localized with phytoene synthase 1 (Psy1) and aldehyde oxidase 3 (AO3) genes, respectively (Zhang and Dubcovsky, 2008;Blanco et al., 2011;Colasuonno et al., 2014Colasuonno et al., , 2017. While the Psy1 involvement in YPC has been deeply studied, the role of AO3 gene in the carotenoid accumulation needs to be elucidated. AO isoforms are key enzymes for ABA biosynthesis (Fang et al., 2008;Zdunek-Zastocka, 2010). Plant AO family is composed by proteins with high similarity in sequences, but different subunit composition and substrate preferences. AO isoforms have been largely characterized in Arabidopsis and resulted composed by 4 isoforms (Seo et al., 2000a). While significant information exists on AO families in Arabidopsis, lettuce, tomato, peas, Brassica, rice and maize, a complete picture of this family is missing in wheat (Zdunek-Zastocka, 2010). Overall, 34 SNPs within AO wheat sequences were identified, making significant advancement on the localization of AO genes on wheat chromosome groups. Indeed, the recent availability of the high-resolution consensus map of durum (Maccaferri et al., 2015) and bread wheat maps (Wang et al., 2014) allowed to determine the precise map position. The AO3 mapping was consistent with results reported by Colasuonno et al. (2014Colasuonno et al. ( , 2017, who identified the chromosomal locations based on survey sequence from the International Wheat Genome Sequencing Consortium (http://www.wheatgenome.org/). Instead, the map positions of AO2 and AO1 genes were reported for the first time: AO2 localized on chromosome group 5 and the AO1 on chromosome group 2.
The present study considered the AO gene expression in wheat. AO isoforms were found expressed in root tissues under osmotic stress (Gallè et al., 2013). The AO2 and AO1 genes resulted expressed in all stages with maximum levels for AO2 in seedling root. While, the AO3 showed elevated expression levels during kernel maturation with a possible involvement in carotenoids pathway since they are accumulate at last developmental stage.
The first connection between carotenoid content and the AO3 genes emerged form a genetic study on QTL in a RIL population derived crossing the durum wheat cvs Ciccio and Svevo (Colasuonno et al., 2014). Previously, other studies (Zhang and Dubcovsky, 2008;Blanco et al., 2011) indicated the presence of this second QTL for carotenoids, using different genetic materials. Then again, Colasuonno et al. (2017) showed that AO-A3 gene was significant associated to carotenoid variation using GWAS and the "candidate gene" approaches.
Based on the evidence presented, we have envisaged that mutations in AO3 induce activity-loss in carotenoid catabolism from violaxanthin/neoxanthin to (ABA) and allows an accumulation of carotenoid compounds. The transcriptional level of the AO-A3 and AO-B3 genes in the cvs Ciccio and Svevo was in accordance with this hypothesis, demonstrating how AO-A3 gene resulted in significant low expression levels in Svevo, cultivar with high content of carotenoids. This could be explained by the presence of an AO-A3 allele in the cv Svevo not fully activated for carotenoid catabolism and ABA biosynthesis. The AO3 analysis need to be further examined to evaluate how the ABA concentration differs in relation to allele functionality. Indeed, in Arabidopsis and maize (Wurtzel et al., 2012;Gonzalez et al., 2013) the carotenoid degradation is important in determining total carotenoid accumulation.
In addition, the regression analysis conducted in the RIL population confirmed the association between the AO-A3 gene marker and YPC. Besides, the development of a molecular gene marker for AO-A3 through the most sensible and cost-effective technique (such as the DHPLC) demonstrated for the first time its applicability on marker assisted selection programmes (MAS). Giancaspro et al. (2016) used the same technology as tool for SNP detection based on genotyping arrays in food science.
The present work characterizes suitable AO3 genes providing a new insight into the regulation of carotenoid accumulation. Although the gene expression analysis revealed differences between the two cultivars, no polymorphisms were observed in the promoter regions, suggesting the presence of complex gene regulation mechanisms. The factors influencing pigment content are complex (Howitt and Pogson, 2006;Lachman et al., 2017). Further investigations needed in order to understand the carotenoid pigment regulation system in wheat.

AUTHOR CONTRIBUTIONS
PC, AB, and AG designed the research; IM and ML performed the research. PC and RS wrote the manuscript. All authors read and approved the final manuscript.