A novel allele of L-galactono-1,4-lactone dehydrogenase is associated with enhanced drought tolerance through affecting stomatal aperture in common wheat

In higher plants, L-galactono-1,4-lactone dehydrogenase (GLDH) plays important roles in ascorbic acid (AsA) biosynthesis and assembly of respiration complex I. Here we report three homoeologous genes (TaGLDH-A1, -B1 and -D1) encoding common wheat GLDH isozymes and a unique allelic variant (TaGLDH-A1b) associated with enhanced drought tolerance. TaGLDH-A1, -B1 and -D1 were located on chromosomes 5A, 5B and 5D, respectively, and their transcripts were found in multiple organs. The three homoeologs each conferred increased GLDH activity when ectopically expressed in tobacco. Decreasing TaGLDH expression in wheat significantly reduced GLDH activity and AsA content. TaGLDH-A1b differed from wild type allele TaGLDH-A1a by an in-frame deletion of three nucleotides. TaGLDH-A1b was biochemically less active than TaGLDH-A1a, and the total GLDH activity levels were generally lower in the cultivars carrying TaGLDH-A1b relative to those with TaGLDH-A1a. Interestingly, TaGLDH-A1b cultivars showed stronger water deficiency tolerance than TaGLDH-A1a cultivars, and TaGLDH-A1b co-segregated with decreased leaf water loss in a F2 population. Finally, TaGLDH-A1b cultivars generally exhibited smaller leaf stomatal aperture than TaGLDH-A1a varieties in control or water deficiency environments. Our work provides new information on GLDH genes and function in higher plants. TaGLDH-A1b is likely useful for further studying and improving wheat tolerance to drought stress.

L-galactono-1,4-lactone dehydrogenase (GLDH) (EC 1.3.2.3) is a mitochondrion located enzyme found in many eukaryotic organisms [10][11][12][13] . In higher plants, GLDH is essential for the synthesis of ascorbic acid (AsA), a vital and abundant antioxidant, through the D-mannose/L-galactose pathway 14 . GLDH acts in the final step of this pathway, converting L-galactono-1,4-lactone (L-GalL) into AsA 14 . GLDH has also been found required for the assembly and accumulation of plant respiratory complex I (RCI) 10,15 . Through transgenic studies in multiple plant species, GLDH activities have been shown required for the normal growth and development of plant cells and organs and their efficient response to adverse environmental factors. For example, in both tobacco and tomato, suppression of GLDH function results in decreased cell and organ growth 16,17 . In rice, inhibition of GLDH activity leads to decreased AsA content, with the transgenic plants exhibiting reductions in seed set and tiller number 18,19 . In Arabidopsis, our study revealed that partial reduction of GLDH expression through RNA interference not only lowers AsA content, but also causes a substantial decrease of foliar water loss rate and potential tolerance to drought stress 20 . Overexpression of GLDH confers enhanced tolerance to oxidative stress in tobacco cells 21 and to salt stress in tobacco plants 22 . In the studies above, changes in GLDH expression and activity levels are mainly rendered using transgenic technology. To date, there is still no published information on natural molecular variations of GLDH genes and their potential influences on agronomic traits. One study has revealed that GLDH protein and activity levels differ significantly between two common wheat cultivars with high (cv Buck Chambergo, BCH) or low (cv Cooperativa Maipún, CM) AsA content, and that drought treatment can substantially up-regulate the amounts of GLDH protein and activity in CM 13 . However, the molecular basis underlying GLDH expression difference between the two varieties is still unclear, and the functional importance of such difference in wheat drought tolerance requires further study.
The findings described by Bartoli and colleagues 13 and the observation that partial suppression of Arabidopsis GLDH activity led to enhanced drought tolerance 20 stimulated us to further study the potential involvement of GLDH in common wheat response to water deficiency stress. Towards this end, we characterized the three homoeologous genes (designated TaGLDH-A1, -B1 and -D1, respectively) encoding common wheat GLDH isozymes, tested their GLDH activity through viral vector mediated ectopic expression in Nicotiana benthamiana, analyzed their function in AsA biosynthesis using virus induced gene silencing (VIGS), and examined their molecular variations in common wheat germplasm lines. On the basis of these experiments, we comparatively analyzed two alleles of TaGLDH-A1, the major wild type (WT) allele TaGLDH-A1a and a rare allele TaGLDH-A1b, in more detail, and found that TaGLDH-A1b was associated with the enhanced drought tolerance of a number of varieties cultivated in arid regions. Our work provides new information on GLDH encoding genes and their function in a polyploid crop species, and demonstrates the association between GLDH and plant tolerance to drought stress.

Characteristics of TaGLDH-A1, -B1 and -D1 homoeologs. To investigate the genes encoding
TaGLDH-A1, -B1 and -D1 isozymes in common wheat, a bacterial artificial chromosome (BAC) library prepared for the bread wheat variety Xiaoyan 54 was screened by PCR (see Methods). A total of seven unique positive BAC clones containing TaGLDH sequence were identified. Three similar but not identical TaGLDH genomic open reading frames (ORFs) were amplified from the seven BAC clones, and completely sequenced. The genes represented by the three ORFs were designated as TaGLDH-A1, -B1 and -D1 based on their chromosomal locations (see below). The complete genomic ORF of TaGLDH-A1, -B1 and -D1 (from start to stop codons) was 5498, 5685 or 5526 bp, and each was composed of six exons and five introns (Fig. 1). This exon-intron pattern was also found for the GLDH genes in other monocot (e.g., rice and Brachypodium distachyon) and dicot (e.g., Arabidopsis thaliana) plants (Fig. 1). The ORF sequences of TaGLDH-A1, -B1 and -D1 were 97% identical, with nucleotide differences found mainly in the introns. The coding region sequences of TaGLDH-A1, -B1 and -D1, as deduced from their cDNA clones, were all 1752 bp (excluding the stop codon) and 99% identical. The deduced proteins of the three homoeologs all contained 584 residues, and were 98-99% identical. Comparison with previously characterized GLDH proteins from Arabidopsis, cauliflower, sweet potato and tobacco indicated the presence of a mitochondrial targeting signal peptide (78 residues) and conservation of the residues involved in FAD-binding and those taking part in substrate recognition and catalysis in TaGLDH-A1, -B1 and -D1 (Fig. 2). The calculated Scientific RepoRts | 6:30177 | DOI: 10.1038/srep30177 molecular mass was 66.1 kD for TaGLDH-A1, -B1 and -D1 preproteins, and 57.5 kD for their putative mature proteins (after removing the putative mitochondrial targeting signal peptide).
To find the chromosome locations of TaGLDH-A1, -B1 and -D1, PCR mapping was carried out using the amplicons yielded by primer sets PS3 and PS4 (Table S1, Figure S1). In the PCR with the genomic DNA of Xiaoyan 54 or Chinese Spring (CS) as templates, both PS3 and PS4 gave rise to three amplicons with variable size diagnostic for TaGLDH-A1, -B1 and -D1 ( Figure S2). Through examining amplicons from the nulli-tetrasomic (NT) lines of CS 23 , TaGLDH-A1, -B1 and -D1 were assigned to 5A, 5B and 5D chromosomes, respectively ( Figure S2). Consistent with this finding, three different GLDH genes (Traes_5AL_E7806F3C1, Traes_5BL_E8D36EEA8 and Traes_5DL_696F6D975) have been annotated in the draft genome sequence of CS 24 . These three genes are located on the long arms of 5A, 5B and 5D, respectively. The deduced amino acid sequences of Traes_5BL_E8D36EEA8 and Traes_5DL_696F6D975 are identical to those of TaGLDH-B1 and -D1, respectively. However, the gene model annotated for Traes_5AL_E7806F3C1 is partial because the polypeptide deduced from it contains only 400 amino acids. Nevertheless, this hypothetical polypeptide is nearly identical to the last 400 residues of TaGLDH-A1.
Expression pattern of TaGLDH in common wheat organs. As revealed by qRT-PCR (Fig. 3A), the expression level of TaGLDH was highest in the young spikelets, intermediate in the seedling leaves, flag leaves and mature spikelets, and comparatively low in the roots and stems. The relatively high expression level of TaGLDH in young spikelets was also found by semi-quantitative RT-PCR assay comparing TaGLDH transcript levels in five vegetative and reproductive organs of common wheat (Fig. 3B). TaGLDH TaGLDH-A1, -B1 and -D1 homoeologs and orthologous GLDH proteins from Arabidopsis (AtGLDH), cauliflower (BoGLDH), sweet potato (IbGLDH) and tobacco (NtGLDH). Arrow indicates the likely cleavage site of the mitochondrial targeting peptide. The underlined sequence element is required for FAD binding. The conserved cysteine residue labeled in purple is involved in regulating the enzyme activity of plant GLDH proteins, whereas the E-R pair of residues marked in blue are required for efficient substrate binding and catalysis. The glutamic acid (E) marked in red is deleted in a novel allele (TaGLDH-A1b) of TaGLDH-A1. Asterisks denote residues conserved among the compared GLDHs. The UniProtKB numbers for AtGLDH, BoGLDH, IbGLDH and NtGLDH are Q8GY16, O47881, Q9ZWJ1 and Q9FXL9, respectively. Figure S3a) were constructed and introduced into N. benthamiana plants. Two weeks after the inoculation of PEBV:GUS, abundant β -glucuronidase (GUS) staining signals were detected in the inoculated leaves ( Figure S3b), thus confirming the capacity of PEBV vector to mediate high level of ectopic expression of heterologous coding sequence in N. benthamiana. Subsequently, the inoculated leaves in the plants infected by PEBV:GUS, PEBV:A1, PEBV:B1 or PEBV:D1 were collected and examined for total GLDH enzyme activity. Additionally, the mock-inoculated leaves in the control plants were also analyzed for GLDH activity. From Fig. 4A, it is clear that GLDH activity level was significantly higher in the leaves infected by PEBV:A1, PEBV:B1 or PEBV:D1 than in those infected by PEBV:GUS, but there was no significant difference in GLDH activity level between PEBV:GUS infected leaves and the mock-inoculated controls. However, the total AsA content did not differ significantly among the examined leaf samples (Fig. 4B). TaGLDH Figure S4a) used in this analysis, BSMV:GLDHas was developed to specifically silence all three TaGLDH homoeologs in common wheat. A 431 bp sequence element strictly conserved among TaGLDH-A1, -B1 and -D1 was employed to construct BSMV:GLDHas. This element recognized only the three GLDH genes (Traes_5AL_E7806F3C1, Traes_5BL_E8D36EEA8 and Traes_5DL_696F6D975, see above) in the draft genome sequence of CS. BSMV:GFP expresses green fluorescent protein (GFP) in wheat cells, and is useful for monitoring virus spread in the inoculated seedlings. BSMV:PDSas, silencing wheat phytoene desaturase (PDS) gene and causing leaf bleaching, is valuable for checking the progress of VIGS.

Function of
Two weeks after virus inoculation, strong GFP fluorescence was observed in the leaf tissues of the seedlings infected with BSMV:GFP ( Figure S4b), and apparent leaf bleaching appeared in those infected by BSMV:PDSas ( Figure S4c), thus verifying the effectiveness of BSMV VIGS in this work. In the seedlings infected by BSMV:GLDHas, the expression level of TaGLDH was significantly reduced (by > 75%) relative to that detected for the controls (uninoculated, mock-inoculated or BSMV:GFP infected seedlings, Fig. 5A). In line with this finding, total TaGLDH activity level in BSMV:GLDHas-infected seedlings was substantially decreased (by > 70%) relative to that in the controls (mock-inoculated or infected by BSMV:GFP, Fig. 5B). Silencing TaGLDH expression caused significant reductions of total AsA (Fig. 5C) and reduced AsA (Fig. 5D) contents in the plants infected by BSMV:GLDHas, but for both parameters the reduction was only about 20%.
Past studies have shown that exogenous application of L-GalL can substantially stimulate AsA biosynthesis in plant tissues 25,26 . Thus, the small decrease of AsA content in the plants infected by BSMV:GLDHas might be due to an insufficient amount of L-GalL available for AsA biosynthesis in the cells. To investigate this possibility, we exogenously applied 15 mM L-GalL to the leaf segments excised from BSMV:GLDHas infected seedlings and the controls, followed by measuring total and reduced AsA contents. For the controls (mock-inoculated or infected with BSMV:GFP), the leaf segments treated with L-GalL exhibited dramatically increased total and reduced AsA contents (by 6.32-6.63 times) relative to those without exogenous application of the substrate ( Figure S5). For the plants infected by BSMV:GLDHas, treatment with L-GalL also resulted in increases of total and reduced AsA contents but at much lower levels (by 3.83-3.98 times, Figure S5). In the presence of 15 mM L-GalL, the total and reduced AsA contents in BSMV:GLDHas infected leaf tissues were reduced by about 42% relative to those of the controls ( Figure S5). This level of reduction was substantially higher than that found for the BSMV:GLDHas infected leaf tissues (~20%, Fig. 5C,D) in the absence of artificially supplied L-GalL. Identification and characterization of TaGLDH-A1b. Potential molecular variations in the coding region of TaGLDH-A1, -B1 or -D1 in the micro-core collection (MCC) of Chinese wheat germplasm were examined by genomic PCR with the primer sets PS3 and PS4 (Table S1, Figure S1), respectively. The 262 MCC lines, including both land races and improved varieties, represent 70% of the genetic diversity of Chinese common wheat 27 . No length variation was detected in the amplicons produced by PS3. However, in the amplicons by PS4, length variation was found for TaGLDH-A1 in four unique cultivars (i.e., Hongdongmai, Kashi 1, Kashibaipi and Tutoumai, Figure S6). The size of TaGLDH-A1 amplicons in the four lines was 452 bp, which was 3 bp shorter than those from Xiaoyan 54 and the remaining MCC lines ( Figure S6). Sequencing the full-length genomic ORF and cDNA coding region of TaGLDH-A1 in the four cultivars revealed an in-frame deletion of three nucleotides (GAG) compared to TaGLDH-A1 sequence in Xiaoyan 54. However, the four lines did not differ from Xiaoyan 54 in either the coding sequence of TaGLDH-B1 or that of TaGLDH-D1. To facilitate further characterization, the TaGLDH-A1 sequence in Xiaoyan 54 was designated as the WT allele TaGLDH-A1a, whereas that in Hongdongmai, Kashi 1, Kashibaipi and Tutoumai was named as TaGLDH-A1b. Comparison of deduced proteins showed that the glutamic acid residue at position 501 of TaGLDH-A1a was absent in TaGLDH-A1b (Fig. 2). This glutamic acid was 42 residues away from the conserved Glu-Arg pair (Fig. 2). These two residues have been suggested to reside in the active site of plant GLDH proteins, and are thus essential for optimal function of these enzymes 28 .
Potential biochemical differences between the proteins specified by TaGLDH-A1a or TaGLDH-A1b were examined. TaGLDH-A1a and TaGLDH-A1b, lacking the sequence coding for the putative mitochondrion targeting peptide, were individually expressed in Escherichia coli using the bacterial expression vector pET-30a (see Methods), with the aim to produce recombinant TaGLDH-A1a and TaGLDH-A1b proteins without the mitochondrial targeting peptide but with a histidine tag at the N-terminus. The recombinant proteins were purified, and verified by protein blot analysis with an antibody specific for the histidine tag (Fig. 6A). Subsequently, the kinetic properties of purified TaGLDH-A1a and TaGLDH-A1b were investigated using L-GalL as a substrate.
Based on their Michaelis-Menten kinetic characteristics, the mean K m and V max of TaGLDH-A1a were estimated to be 61 μ M and 11.43 μ mol. mg −1 protein.min −1 , respectively, and the corresponding values for TaGLDH-A1b were 181 μ M and 0.16 μ mol.mg −1 protein.min −1 , respectively (Table 1). These data indicated that TaGLDH-A1b had significantly reduced kinetic activities than TaGLDH-A1a towards the substrate tested. In line with the biochemical difference between TaGLDH-A1a and TaGLDH-A1b, total TaGLDH activity levels in the TaGLDH-A1b lines were generally and significantly lower than that found for Xiaoyan 54 carrying TaGLDH-A1a (Fig. 6B). Under normal growth conditions, total AsA contents in the seedling leaves of the three TaGLDH-A1b lines tended to be lower than that of Xiaoyan 54, but the differences were not statistically significant (Fig. 6C).  Association of TaGLDH-A1b with enhanced tolerance to water deficiency stress. The four cultivars carrying TaGLDH-A1b all came from the Xinjiang Uygur Autonomous Region (Xinjiang hereafter) in Northwest China 29 . The common wheat lines cultivated in Xinjiang generally have enhanced drought tolerance because this region has a strong continental climate with an annual rainfall varying from about 40 mm (in the south) to 192 mm (towards the north) 29,30 . Therefore, we tested if TaGLDH-A1b might be found in an additional set of 11 wheat cultivars from Xinjiang (Table S2) with a newly developed molecular marker CAP-A1b ( Figure S7a). The PCR fragments derived from TaGLDH-A1a (538 bp) and TaGLDH-A1b (535 bp) were digested using the restriction enzyme BsaJI, which cleaved the amplicon of TaGLDH-A1a once (yielding two fragments of 266 and 272 bp, respectively) but not that of TaGLDH-A1b ( Figure S7a). Among the 11 cultivars, three lines (Jiuchun 1, Redstar and Xinchun 3) were detected to have TaGLDH-A1b, with the remaining ones carrying TaGLDH-A1a (Table S2). Together, seven cultivars carrying TaGLDH-A1b were found among a total of 273 germplasm lines examined in this study. The frequency of TaGLDH-A1b was thus 2.56%. The seven TaGLDH-A1b lines are all known to have enhanced tolerance to abiotic stresses (Table S3), and four of them (Hongdongmai, Kashibaipi, Redstar, and Xinchun 3) have been extensively cultivated in Xinjiang 29 . One of the TaGLDH-A1b lines, Redstar, was originally from the former Soviet Union 29 , whereas the remaining ones were native to Xinjiang (Table S3).
Next, we performed three independent pot trials for comparing the drought tolerance of eight Xinjiang cultivars, four carrying TaGLDH-A1a and four with TaGLDH-A1b. The plants of the eight lines were deprived of water for 30 days followed by rewatering, with their growth behavior recorded at specific time points. At 20 days post water deprivation (DPWD), the four lines with TaGLDH-A1b were apparently greener than those carrying TaGLDH-A1a (Fig. 7A). At 30 DPWD (just before rewatering), the four TaGLDH-A1a lines exhibited severe wilting, but the wilting was much less severe in the four TaGLDH-A1b lines (Fig. 7A). At the 7th day after rewatering, the TaGLDH-A1b lines were obviously taller and had more green leaves than the ones with TaGLDH-A1a (Fig. 7A), indicating that the former lines recovered from water deficiency stress more efficiently than the latter ones. Free proline content is an important indicator of plant drought tolerance 31 . At 20 DPWD, free proline contents in the leaf tissues of TaGLDH-A1b lines were generally and significantly higher than those of TaGLDH-A1a lines (Fig. 7B). The plants of the eight cultivars, whether carrying TaGLDH-A1a or TaGLDH-A1b, all survived under the above drought treatment conditions; no difference was observed in the survival rate between the two types of lines in the three pot trials.
Lastly, we compared the degree of leaf water loss between TaGLDH-A1a and TaGLDH-A1b lines because this parameter is another valuable indicator of plant response to water deficiency stress 2,5 . Water loss behavior of the leaves was monitored at 2, 4 and 6 h post detachment, respectively, for the two groups of lines carrying TaGLDH-A1a or TaGLDH-A1b. At all three time points, the levels of water loss recorded for the TaGLDH-A1b lines were consistently and significantly lower than those found for the TaGLDH-A1a lines (Fig. 8A). Furthermore, we investigated if there would be differences in water loss behavior among the different genotypes identified in a F 2 population developed using two Xinjiang cultivars Yichun 4 (carrying TaGLDH-A1a) and Xinchun 3 (with TaGLDH-A1b) (Table S2). After genotyping 134 F 2 individuals with the DNA marker CAP-A1b ( Figure S7b), the individuals homozygous for TaGLDH-A1a or TaGLDH-A1b were 35 and 30, respectively, and those heterozygous at TaGLDH-A1 locus were 69. The levels of water loss determined for Xinchun 3 or TaGLDH-A1b/TaGLDH-A1b leaves were highly similar, both being significantly lower than those obtained for Yichun 4, TaGLDH-A1a/TaGLDH-A1a or TaGLDH-A1a/TaGLDH-A1b leaves (Fig. 8B). The water loss levels of Yichun 4, TaGLDH-A1a/TaGLDH-A1a and TaGLDH-A1a/TaGLDH-A1b differed only at 2 h post detachment (Fig. 8B).

Discussion
Although understanding and enhancing drought tolerance represent a major task in current plant biology and crop improvement research, only a few genes have so far been molecularly characterized for their involvement in wheat drought tolerance 9,32-34 Here, we investigated the function of TaGLDH homoeologs and explored the    association of a previously undescribed TaGLDH-A1 allele (TaGLDH-A1b) with enhanced drought tolerance by combining molecular, biochemical and physiological approaches. The new information obtained is discussed below.

Molecular features of TaGLDH-A1, -B1 and -D1 homoeologs. Prior to this work, little is known
about TaGLDH homoeologs and their protein products at the molecular level, although there is evidence for TaGLDH protein expression and enzyme activity in common wheat 13 . Based on the data gathered here, several suggestions can be made on the molecular features of TaGLDH homoeologs. First, TaGLDH-A1, -B1 and -D1 are located on wheat group 5 chromosomes. This finding is consistent with the annotation of three TaGLDH homoeologous genes on 5A, 5B and 5D chromosomes, respectively, by the draft genome sequence of CS 24 . Second, the exon and intron pattern in TaGLDH-A1, -B1 and -D1 is identical to that found for orthologous GLDH genes in rice, B. distachyon and Arabidopsis, suggesting the preservation of exon and intron structure in GLDH genes in higher plants. Third, the deduced primary structure and mitochondrial targeting of TaGLDH-A1, -B1 and -D1 are highly similar to those of orthologous GLDHs in Arabidopsis, cauliflower, sweet potato and tobacco, highlighting the conservation of GLDH protein structure and subcellular location in plant cells. Fourth, TaGLDH-A1, -B1 and -D1 all have GLDH activity, and may not differ substantially from each other in the level of such activity. This is supported by similar increases of total GLDH activity in the N. benthamiana plants ectopically expressing TaGLDH-A1, -B1 or -D1. Fifth, the recombinant protein of TaGLDH-A1a (lacking the putative mitochondrial targeting peptide) is biochemically active towards the substrate L-GalL, with a K m value very similar to that reported for tobacco GLDH enzyme 35 . Lastly, TaGLDH transcripts are present in multiple vegetative and reproductive organs of common wheat, with relatively high transcript abundance in the leaf and spike tissues. The observation of a higher TaGLDH transcript level in the leaves by this work agrees well with the finding of abundant TaGLDH proteins in wheat leaf tissues by a previous study 13 . However, it is interesting to note that TaGLDH transcripts are highly abundant in the spikelets, particularly in the young spikelet tissues. It is well known that the development of cereal spikelets involves intensive cell growth and differentiation and enhanced metabolic activity 36 , and that the level of reactive oxygen species in this developmental process needs to be tightly controlled especially under stress conditions 37 . Furthermore, AsA has been found to regulate flowering 20,38 . Therefore, the high TaGLDH transcript level detected in the immature spikelets by this work may reflect the function of TaGLDH in active AsA biosynthesis during common wheat spikelet development. The importance of having sufficient GLDH proteins during cereal spikelet development has recently been demonstrated by the finding that silencing rice GLDH expression led to substantially reduced seed set 19 .

Function of TaGLDH in AsA biosynthesis in common wheat.
In Arabidopsis, rice and several other plant species, the function of GLDH in AsA biosynthesis is well supported by genetic evidence obtained using defined mutants and transgenic lines with altered expression GLDH level 17,[19][20][21][22] . But in common wheat genetic evidence for the function of GLDH in AsA biosynthesis is not available before this work. From the data of our BSMV VIGS experiment, it is clear that normal TaGLDH expression is required for efficient AsA biosynthesis in common wheat seedlings. In agreement with previous studies 39-42 , our BSMV VIGS was reliable and effectively Scientific RepoRts | 6:30177 | DOI: 10.1038/srep30177 lowered the expression of TaGLDH. However, in the leaf tissues undergoing TaGLDH silencing, the reductions of total and reduced AsA contents were considerably below the decreases in TaGLDH expression and GLDH activity. The lack of correlation in between GLDH expression level, GLDH activity and AsA content has been observed in a number of plant species 13,43 , and several possibilities have been suggested to explain such a phenomenon 13,25 .
Here we found that exogenous application of 15 mM L-GalL strongly increased AsA contents in both controls and the plants infected by BSMV:GLDHas, and substantially enlarged the degree of AsA content reduction by silencing TaGLDH expression. Therefore, in common wheat leaf tissues, the function of TaGLDH in AsA biosynthesis is at least partly affected by the amount of substrate.
One concern on VIGS data is that their reliability may be affected by off target silencing 39,41 . In this work, the chance of off target silencing in the VIGS initiated by BSMV:GLDHas is likely to be very small because the fragment employed for developing BSMV:GLDHas recognized only the three TaGLDH homoeologs and no additional GLDH paralogous sequences were found in the draft genome sequence of CS. Therefore, the reductions of TaGLDH expression, TaGLDH activity and AsA contents in BSMV:GLDHas infected plants are due to specific silencing of the expression of TaGLDH-A1, -B1 and -D1. Nevertheless, a more detailed understanding of the function of TaGLDH-A1, -B1 and -D1 in AsA biosynthesis (and other processes) would require the preparation and examination of wheat mutants lacking one or more of the three homoeologs. We are now in the process of developing such mutants.
Association of TaGLDH-A1b with enhanced drought tolerance. Natural allelic variation is a fundamental mechanism in the genetic control of plant traits 44 . Mining allelic variations at the molecular level has contributed enormously to understanding and improving the genetic basis of valuable traits in higher plants 45 . The cloning of three TaGLDH homoeologs in this work enabled us to examine their molecular variations in common wheat germplasm, leading to the discovery of a novel allele TaGLDH-A1b. Compared to WT allele (TaGLDH-A1a), TaGLDH-A1b missed one glutamic acid residue, and exhibited a significant reduction in its catalytic activity towards the tested substrate (Table 1). To our knowledge, this type of allele has not been described previously for either plant or animal GLDH proteins. Further, our data suggest that there is very likely a positive association between TaGLDH-A1b and enhanced drought tolerance in common wheat. This proposition is supported by four lines of evidence. First, the seven cultivars carrying TaGLDH-A1b were all from Xinjiang that has a limited rainfall in wheat production season 29,30 . Second, our tests showed that, relative to TaGLDH-A1a lines, those with TaGLDH-A1b generally displayed better tolerance to water deficiency stress in growth trails (Fig. 7A), had a higher foliar proline content (Fig. 7B), and showed lower leaf water loss rate (Fig. 8A). We did not observe any difference in the survival rate between the cultivars carrying TaGLDH-A1a or TaGLDH-A1b probably because the drought treatment applied was not sufficiently severe. Third, TaGLDH-A1b co-segregated with lower water loss in a F 2 population (Fig. 8B). Lastly, under either normal or water deficiency conditions, TaGLDH-A1b cultivars generally had smaller stomatal aperture (Fig. 9), which is known to contribute directly to reduction of water loss and to stronger drought tolerance in water limited environments 2,5 . Interestingly, our previous study showed that partial suppression of GLDH activity also resulted in enhanced drought tolerance in Arabidopsis 20 . In both our previous study 20 and this work, partial reduction of GLDH activity did not cause a drastic decrease of AsA content, and yet drought tolerance was significantly increased. This suggests that, in both monocot and dicot plant species, GLDH activity may be partially lowered to enhance drought tolerance.
Because the frequency of TaGLDH-A1b (2.56%) was relatively low, we deduce that TaGLDH-A1b is a rare allele that has so far not gained intensive use in wheat breeding. However, TaGLDH-A1b may have at least two different geographical origins because one of its carriers (i.e., Redstar) is not native to Xinjiang 29 . There is now growing interest in characterizing rare alleles of plant genes because many of them have been found to confer positive effects on agronomic traits 44,45 . For example, a rare allele of the crtRB1 gene enhances β -carotene content in maize grains 46 . The rare alleles of rice OsPPKL1 and GS2/OsGRF4 genes are associated with larger grains and higher yield 47,48 . Considering the overwhelming importance of wheat as a major staple food and the urgent need to develop drought tolerant wheat cultivars in both China and the world 9,33,49 , it is highly worthwhile to test if TaGLDH-A1b may confer improved drought tolerance when transferred to high yielding common wheat varieties with more diverse genetic backgrounds. Aided by the CAP-A1b marker developed in this work ( Figure S7a), we have now embarked on this test.
Mechanism underlying the function of TaGLDH-A1b in enhanced drought tolerance. From the data gathered in this study (Figs 8 and 9), it is clear that decrease of stomatal aperture and reduction of leaf water loss are two physiological events closely linked with the enhanced drought tolerance by TaGLDH-A1b. However, further efforts are needed to reveal the additional physiological processes and molecular interactions involved. A previous study has shown that reducing AsA content in transgenic tobacco plants led to enhanced water stress tolerance through accumulation of a higher concentration of H 2 O 2 in the guard cells and an increased stomatal closure in water limiting environment 50 . But in common wheat water deficiency treatment, although increasing GLDH activity in certain cultivar, did not cause substantial changes in AsA content 13 . We also found that water deficiency stress did not alter AsA content considerably in the common wheat plants carrying TaGLDH-A1b ( Figure S8). Furthermore, although the cultivars carrying TaGLDH-A1b generally exhibited substantial reductions in TaGLDH activity, their foliar AsA contents did not differ significantly from those of the cultivars with TaGLDH-A1a (Fig. 6C). These findings suggest that the overall foliar AsA content may not be directly related to the smaller stomatal aperture observed for TaGLDH-A1b plants. However, there may be a difference in the guard cell AsA content between TaGLDH-A1a and TaGLDH-A1b plants. This is possible because H 2 O 2 accumulation in the guard cells has been found vital for stomata closing 51,52 , and AsA, being the most abundant water-soluble antioxidant in plant cells, plays a critical role in regulating cellular level of H 2 O 2 14,53 . Consequently, further work will be directed to compare if the guard cells of TaGLDH-A1a and TaGLDH-A1b leaves may differ in the levels of AsA and H 2 O 2 under normal or water deficiency conditions. Because it has been shown that GLDH regulates the assembly and function of plant RCI 10,15 , it will also be interesting to investigate if RCI activity may differ between TaGLDH-A1a and TaGLDH-A1b plants, and whether such difference might be involved in the enhanced drought tolerance associated with TaGLDH-A1b. Two recent studies have shown that partial reduction of RCI activity can lead to constitutive enhancement of drought tolerance in tobacco plants through reducing stomatal conductance and aperture 54,55 . Thus, it is worthy to analyze if the RCI activity of TaGLDH-A1b plants may be altered as compared to those carrying TaGLDH-A1a, and if so, its potential impact on stomatal aperture control and leaf water loss behavior.
Finally, owing to the difficulty in designing homoeolog specific primers, we were unable to compare if TaGLDH-A1a and TaGLDH-A1b may differ in their expression level by qRT-PCR. The resolving of this question is clearly important for further studying the functional difference between the two alleles. Therefore, we have started a RNA sequencing experiment, which should reveal any potential difference between the transcript levels of TaGLDH-A1a and TaGLDH-A1b. Because we observed a co-segregation between TaGLDH-A1b and reduced leaf water loss rate (Fig. 8B), and the occurrence of smaller stomatal aperture in TaGLDH-A1b plants (Fig. 9), it becomes necessary to examine if TaGLDH-A1b may co-segregate with decreased stomatal aperture. However, the investigation of this question using segregating F 2 plants is time consuming. The development of defined mutants for TaGLDH homoeologs (see above) or near isogenic lines for TaGLDH-A1a and TaGLDH-A1b should facilitate this analysis in further research.
In summary, this work has improved our understanding on plant GLDH genes and their function through characterizing TaGLDH homoeologs and discovering an association between TaGLDH-A1b and enhanced drought tolerance. The new information obtained may prompt further research on GLDH genes and their application in improving the drought tolerance and other traits of crop plants.

Methods
Plant materials, nucleic acid extraction, oligonucleotide primers and chemicals. Four sets of common wheat varieties were used in this study. The first set included Xiaoyan 54, CS and the NT lines derived from CS. The second set was the mini-core collection (MCC) of Chinese wheat germplasm composed of 262 landraces and improved varieties 27 . The third set included 11 cultivars from Xinjiang (Table S2). The fourth set consisted of 134 F 2 seedlings developed in this work through crossing two Xinjiang varieties (i.e., Yichun 4 × Xinchun 3). Unless specifically stated, wheat plants were grown in a greenhouse at 25 °C (day)/20 °C (night) and with a 16 h light/8 h dark photoperiod. A vernalization treatment (5 weeks at 4 °C) was applied to Xiaoyan 54 seedlings to promote flowering. Genomic DNA and total RNA samples were prepared from desired wheat tissues as described previously 40 . The oligonucleotide primer sets (PS1 to PS11) used in this work are listed in Table S1, and their positions are given in Figure S1.
Cloning TaGLDH homoeologs and expression analysis of TaGLDH. The BAC library of Xiaoyan 54 was screened by PCR with the conserved primer set PS1 (Table S1, Figure S1), and following the procedure reported previously 56 . TaGLDH ORFs, corresponding to TaGLDH-A1, -B1 and -D1, respectively, were amplified from the positive BAC clones using the primer set PS2 (Table S1, Figure S1). PS2 was also used to clone the cDNA coding region of TaGLDH-A1, -B1 and -D1 by RT-PCR with the cDNAs derived from Xiaoyan 54 leaf tissues as described previously 40 . Subsequently, two primer sets (PS3 and PS4, Table S1, Figure S1), both of which could yield amplicons with different lengths for TaGLDH-A1, -B1 and -D1, were developed for chromosome assignment of the three homoeologs. The assignment was facilitated using CS and associated NT lines and by fragment analysis of fluorescently labeled PCR amplicons 57 .
Because the coding sequences of TaGLDH-A1, -B1 and -D1 were nearly identical, it was difficult to design copy specific PCR primers to study the expression patterns of individual homoeologs. Therefore, two primer sets PS5 and PS6 (Table S1, Figure S1), capable of recognizing all three TaGLDH homoeologs, were designed for detecting the transcript level of TaGLDH by qRT-PCR and semi-quantitative RT-PCR, respectively. Total RNA samples were extracted from vegetative (root, stem, seedling leaf and flag leaf) and reproductive (young and mature spikelets) organs. The cDNAs derived from the different organs were used for qRT-PCR with the 26S rRNA gene (GenBank accession Z11889) as an internal reference and following the cycling conditions reported previously 40 . The semi-quantitative RT-PCR assay was performed as reported in a previous study 57 , with the wheat beta-tubulin gene 2 (GenBank accession U76745) as an internal reference gene.

Ectopic expression of TaGLDH homoeologs in N. benthamiana. The coding sequence of
TaGLDH-A1, -B1 or -D1 was each expressed in N. benthamiana using the PEBV vector. This vector has been used successfully for ectopically expressing cloned genes in N. benthamiana 58,59 . For constructing PEBV:GUS, GUS coding sequence was excised from the plasmid pJIT166 60 using NcoI and EcoRI digestions, followed by cloning into PEBV vector. The coding sequence of TaGLDH-A1, -B1 and -D1 was each cloned into PEBV vector using PCR fragment amplified with the primer sets PS7 (for A1) or PS8 (for B1 and D1) to form PEBV:A1, PEBV:B1 and PEBV:D1, respectively. The information on PS7 and PS8 is given in Table S1 and Figure S1. The inoculation of N. benthamiana plants with PEBV:GUS, PEBV:A1, PEBV:B1 or PEBV:D1 was conducted as detailed in our previous publication 59 . Two weeks after inoculation, the leaves infected by PEBV:GUS were collected for histochemical staining of GUS activity 61 , whereas those infected by PEBV:A1, PEBV:B1 or PEBV:D1 were harvested for determining total GLDH activity and AsA content (see below).
Scientific RepoRts | 6:30177 | DOI: 10.1038/srep30177 VIGS analysis. Among the three recombinant BSMVs used in this analysis ( Figure S4a), BSMV:GFP and BSMV:PDSas were developed previously 40,42 . BSMV:GLDHas was constructed using a 431 bp sequence element conserved among TaGLDH-A1, -B1 and -D1 coding sequence. This element was amplified from leaf cDNAs as a NheI fragment using the primer set PS9 (Table S1, Figure S1). It was cloned into BSMV vector and gave rise to BSMV:GLDHas. The genome wide target specificity of the 431 bp fragment was checked by searching the draft genome sequence of CS at http://plants.ensembl.org/Triticum_aestivum/Info/Index. RNA transcripts were prepared for the three recombinant BSMVs, and introduced into Xiaoyan 54 seedlings (at two leaf stage) as detailed previously 40,42 . For each virus, 60 seedlings were inoculated. The same number of seedlings was buffer-inoculated as mock controls, and another 60 seedlings were kept as uninoculated controls. Three weeks after inoculation, the fourth leaves of the plants infected by BSMV:GLDHas were collected for examining TaGLDH silencing by qRT-PCR and alterations of total GLDH activity and AsA content (see below). For testing the effect of exogenous application of L-GalL on AsA biosynthesis, the fourth leaves were collected from mock controls and the plants infected by BSMV:GFP or BSMV:PDSas. They were cut into 1 mm segments, followed by incubation in a 15 mM solution of L-GalL for 24 h at 25 °C under a light intensity of 150 μ mol/m 2 ·s. Afterwards, the segments were collected for measuring total and reduced AsA contents (see below).
Measurement of total GLDH activity and AsA content in leaf tissues. The assay of total GLDH activity in leaf tissues was carried out using a crude mitochondrial fraction that was prepared at 4 °C following a previous study 11 . Measurement of GLDH activity level in the crude mitochondrial fraction was conducted in 0.5 ml volume consisting of 50 mM Tris-HCl, pH 8.0, 60 μ M cytochrome c, 0.15% (w/v) Triton X-100 and 25 μ g sample protein. The reaction was started by adding 4 μ M L-GalL. After the reaction, the assay mixtures (150 μ l per sample) were transferred to a 96 well microtiter plate, and the absorbance at 550 nm was recorded for each well using a Thermo Multiscan Scientific Spectrophotometer (Thermo Scientific, Waltham, USA). One unit of GLDH activity was defined as the formation of 2 nm reduced cytochrome c per minute.
To determine AsA content in a leaf sample, about 40 mg foliar materials were homogenized, and suspended in 6% trichloroacetic acid. The suspension was centrifuged at 13,000 g for 5 min at 4 °C, and the supernatant was kept for measuring the contents of total and reduced AsA following the method detailed previously 62 . TaGLDH-A1b. The screening was executed by fragment analysis of fluorescently labeled PCR amplicons. Briefly, genomic DNA samples were extracted from 262 MCC lines, and used in two series of genomic PCR assays with the primer sets PS3 and PS4 (Table S1, Figure S1), respectively. The resultant amplicons were subjected to fragment analysis as described above. For tagging TaGLDH-A1b allele, a cleaved amplified polymorphic marker CAP-A1b was developed ( Figure S7a). A portion of the genomic ORF of TaGLDH-A1 (538 bp for TaGLDH-A1a and 535 bp for TaGLDH-A1b) was specifically amplified by PCR with the primer set PS11 (Table S1, Figure S1). The amplicons were digested with BsaJI, which cut the amplicons from TaGLDH-A1a but not those of TaGLDH-A1b. CAP-A1b was employed to identify TaGLDH-A1b in 11 Xinjiang common wheat lines, and in the F 2 individuals derived from Yichun 4 × Xinchun 3 (see below).

Bacterial expression and biochemical study of TaGLDH-1a and -A1b. TaGLDH-A1a and
TaGLDH-A1b sequences were amplified from the cDNAs of Xiaoyan 54 and Kashibaipi, respectively, using the primer set PS10 (Table S1, Figure S1). The resultant TaGLDH-A1a and TaGLDH-A1b fragments lacked the nucleotide sequence encoding the first 78 residues (the putative mitochondrial targeting peptide). These fragments were cloned into the plasmid vector pET-30a (Novagen, Madison, WI, USA), yielding the constructs pET-A1a and pET-A1b, respectively. Induction of TaGLDH-A1a and TaGLDH-A1b expression and purification of the two recombinant proteins by combining nickel chelate affinity chromatography and fast performance liquid chromatography followed the practices detailed previoulsy 59 . The purified TaGLDH-A1a and TaGLDH-A1b were checked by 10% SDS-PAGE and a protein blot assay with an anti-histidine antibody (Roche Diagnostic GmbH, Mannheim, Germany).
For investigating the kinetic properties of recombinant TaGLDH-A1a and TaGLDH-A1b, the assay was performed as described above for testing GLDH activity with minor modifications. Each assay was conducted in 0.2 ml volume consisting of 25 mM Tris-HCl, pH 8.0, 25 mM NaCl, 60 μ M cytochrome c, 0.167 μ g recombinant protein (TaGLDH-A1a or TaGLDH-A1b) and a specific concentration of L-GalL. A total of 11 concentrations of L-GalL, varying from 0.015625 to 5 mM, were used. The reaction was started by addition of TaGLDH-A1a or TaGLDH-A1b. After the reaction, the absorbance at 550 nm was read for each assay. The obtained readings were then employed to calculate the kinetic parameters (K m and V max ) with the SigmaPlot Enzyme Kinetics Module for Windows (SPSS Inc., Chicago, IL, USA).
Drought tolerance growth trial and determination of free proline content. The trial was conducted essentially as described by Zhang and coauthors 33 . In brief, each trail started with the seedlings at three-leaf stage, and a total of 40 uniform seedlings (5 individuals per pot) were used for each variety. The plants were placed in a growth chamber set at 23 °C (day)/20 °C (night) and with a 16 h light/8 h dark photoperiod, but without watering for 30 days. They were re-watered afterwards. The volumetric water content (VWC) of each pot was checked every two days, and the plants at the same VWC were used for comparison. The trial was repeated three times. At 20 days post water deprivation, free proline content was determined for each variety using the leaves collected from 5 individual plants in five different pots. The measurement of proline content followed a procedure detailed previoulsy 63 . The proline content assay was conducted in two of the three drought tolerance trails.
Water loss assay. The assay was carried out using wheat plants (each with 3 tillers) cultured in 1 × Hoagland solution in a growth chamber with a temperature regime of 15 °C (day)/10 °C (night) and a photoperiod of 16 h light/8 h dark. In each assay and for each cultivar, 7 fully expanded leaves from 7 plants were pooled and subjected to water loss measurement 33 . For investigating water loss behavior of the F 2 plants derived from Yichun 4 × Xinchun 3, the F 2 individuals (at tillering stage) were first genotyped with the marker CAP-A1b (see above). This divided the F 2 progenies into three groups, homozygous for TaGLDH-A1a (i.e., A1aA1a) or TaGLDH-A1b (A1bA1b) and heterozygous at TaGLDH-A1 locus (A1aA1b) ( Figure S7b). The water loss assay was then conducted for the three types of plants and their parents as described above.
Measurement of stomatal aperture. Wheat plants were cultured as described above for the water loss assay. For each of the eight wheat varieties under examination, 9 uniform plants (each with 3 tillers) were selected and divided into 3 batches (3 in each batch). The first batch of plants was transferred into double distilled water (as controls), with 6 epidermal peels immediately prepared from the 3 plants. The second and third batches of plants were transferred into 13% PEG 6000, and treated for 24 and 48 h, respectively. Subsequently, 6 epidermal peels were immediately prepared from the second and third batches of plants, respectively. The epidermal peels were stained with 2′ ,7′ -dichlorofluorescin diacetate 64 , which facilitated the examination of guard cells and stomatal aperture under a confocal laser scanning microscope (LSM710; Carl Zeiss AG, Oberkochen, Germany). Ten photos were taken for each epidermal peel, and used for measuring stomatal aperture with the ZEN lite 2012 software. This experiment was repeated three times.
Statistical analysis. Statistical analysis of the experimental data (presented as mean ± SD) was performed by ANOVA with the SPSS program (SAS Institute Inc., Cary, NC, USA).