W3 Is a New Wax Locus That Is Essential for Biosynthesis of β-Diketone, Development of Glaucousness, and Reduction of Cuticle Permeability in Common Wheat

The cuticle plays important roles in plant development, growth and defense against biotic and abiotic attacks. Crystallized epicuticular wax, the outermost layer of cuticle, is visible as white-bluish glaucousness. In crops like barley and wheat, glaucousness is trait of adaption to the dry and hot cultivation conditions, and hentriacontane-14,16-dione (β-diketone) and its hydroxy derivatives are the major and unique components of cuticular wax in the upper parts of adult plants. But their biosynthetic pathway and physiological role largely remain unknown. In the present research, we identified a novel wax mutant in wheat cultivar Bobwhite. The mutation is not allelic to the known wax production gene loci W1 and W2, and designated as W3 accordingly. Genetic analysis localized W3 on chromosome arm 2BS. The w3 mutation reduced 99% of β-diketones, which account for 63.3% of the total wax load of the wild-type. W3 is necessary for β-diketone synthesis, but has a different effect on β-diketone hydroxylation because the hydroxy-β-diketones to β-diketone ratio increased 11-fold in the w3 mutant. Loss of β-diketones caused failure to form glaucousness and significant increase of cuticle permeability in terms of water loss and chlorophyll efflux in the w3 mutant. Transcription of 23 cuticle genes from five functional groups was altered in the w3 mutant, 19 down-regulated and four up-regulated, suggesting a possibility that W3 encodes a transcription regulator coordinating expression of cuticle genes. Biosynthesis of β-diketones in wheat and their implications in glaucousness formation and drought and heat tolerance were discussed. Key Message W3 is essential for β-diketone biosynthesis but suppresses its hydroxylation. Loss-of-function mutation w3 significantly increased cuticle permeability in terms of water loss and chlorophyll efflux.


Introduction
During transition from water to land colonization, plants developed an array of mechanisms for adaptation to the desiccation environment. One of these mechanisms is deposition of cuticle, a hydrophobic coat, to cover the aerial organ surfaces. In addition to function in protecting water loss, cuticle, as the interface between the sessile plants and their environment, plays important roles in plant defense against heat, UV radiation, pathogen and insect attacks [1]. Cuticle consists of the fundamental framework of cutin and intracuticular wax inserted in it and epicuticular wax overlaid on them. Cutin is the cell wall-bounded ester polymer of hydroxy fatty acids [2][3][4], and waxes are the very long chain fatty acids (VLCFAs) and their derivatives including alcohols, aldehydes, alkanes, ketones and wax esters [5]. Variation in epicuticular wax composition causes changes in plant appearance: glaucous or nonglaucous. Glaucousness, the bluish-white look, is the visible form of densely arrayed wax crystals.
In wax biosynthesis, de novo C 16 and C 18 fatty acids, the precursors, are first elongated into VLCFAs (C 20 to C 34 ) by fatty acyl elongase (FAE) complex. The VLCFAs can be converted to primary alcohols and further esterified with the C 16 fatty acids into wax esters via the acyl reduction pathway [6][7][8] and converted to alkanes, aldehydes, secondary alcohols and ketones through decarbonylation pathway [9][10][11]. All the wax species are synthesized in epidermal cells, transported extracellular and deposited to cuticle. Wax synthesis and export are under tight developmental and environmental regulations. Molecular identification of the genes coding for the wax biosynthetic enzymes, wax transporters and wax regulators in the model plant Arabidopsis provided genetic and molecular supports to the VLCFA and associated pathways (reviewed in [12][13][14]. The VLCFA and associated pathways are conserved in higher plants, and many of their components of these pathways have also been identified in the grass genomes. Maize wax genes GLOSSY1 (GL1) [15,16], GL2 [17], GL4 [18] and GL8 [19,20] are homologous to ECER-IFERUM3 (CER3), CER2, CER6 and β-keto acyl-CoA reductase 1 (KCR1) of Arabidopsis, respectively. A T-DNA insertion mutant in a CER6 homolog reduced leaf wax crystal density in rice [21]. Maize AP2 transcription factor GL15 involves in the transition from juvenile to adult leaf identity including wax composition [22], and HD-ZIP IV family transcription factor OCL1 regulates cuticular wax biosynthesis by activation of a fatty acyl-CoA reductase (FAR) and a putative wax transporter [23,24].
In addition to the VLCFA and associated pathways, a diverse of plants, including barley and wheat (Triticum L.), employ another parallel wax biosynthetic pathway for biosynthesis of hentriacontane-14,16-dione (also known as β-diketone) and its derivatives. In wheat, the VLCFA pathways are active in the vegetative stage, but the β-diketone pathway predominates in the reproductive stage [25]. It is believed that β-diketone is synthesized using C 14 and C 16 fatty acids as precursors [26]. Wax profiling of barley mutants suggests that cer-q, cer-c and cer-u mutants respectively define the initial condensing reaction, subsequent chain extensions and hydroxylation [27], but little is known about the genetic components of this pathway. In common wheat (T. aestivum L., genomes AABBDD), two wax inhibitors and two wax production genes underlie the glaucousness variations: Iw1 on chromosome arm 2BS [28][29][30][31], Iw2 on 2DS [30][31][32][33], W1 on 2BS [30,34] and W2 on 2DS [30]. In durum wheat (T. turgidum subsp. durum, genomes AABB), a third wax inhibitor Iw3 is located in the distal region of chromosome arm 1BS [35,36]. We previously characterized a set of wheat near isogenic lines (NILs) differing at W1, W2, Iw1 and Iw2 loci and durum NILs differing at the Iw3 locus, and found that loss of both functional alleles of the W genes or presence of either Iw gene causes depletion of β-diketones and nonglaucous phenotype. While elimination of β-diketones is compensated by increase of aldehydes and primary alcohols in the Iw1 and Iw2 NILs [37], Iw3 inhibits β-diketone, reduces primary alcohols but increase aldehyde and alkanes in the glume wax [36]. Although expression of a FAR1 homolog was maintained at high level in the Iw NILs, no match was found between expression of 72 wax genes and β-diketone distribution pattern [36,37]. All this suggests that β-diketones are synthesized through a not-yet-identified pathway.
We recently identified a nonglaucous (NG) mutant in Bobwhite (BW) wheat. Compared to the wild type BW, the mutant lost 63% of total wax load and almost all of the β-diketones, and showed a significant increase of water loss. Genetic and molecular analyses indicated that the mutant gene is not allelic to either W1 or W2 and expression of numerous cuticle genes of five different pathways was significantly down regulated. Here we report the results and their implications in Triticeae cuticle biosynthesis and drought tolerance.

Genetic Analysis of the Wax Mutant
During intercrossing BW transgenic plants to combine the different RNAi transgenes, a NG plant was found in a small T 1 population of the transgenic line #056 in the spring of 2010. It dried out prematurely ( Fig 1A). We recovered two NG plants, NG1 and NG2, from the T 2 populations derived from the glaucous T 1 plants of the same pot. These NG plants had much lower level of glaucousness in leaves, sheaths, peduncle, and spikes as compared to the wild type BW ( Fig 1B). The mutant plants died prematurely again in the late spring of 2011 although they were well watered ( Fig 1C) probably due to the high temperature.
We started genetic analysis of this presumable NG mutant by crossing it with the wild type BW. The F 1 plants were glaucous but their glaucousness level was intermediate between BW and the NG mutant (Fig 2A). A population of 361 F 2 individuals segregated into 265 glaucous and 96 nonglaucous, which fits the 3 (glaucous) to 1 (nonglaucous) ratio (P = 0.38812; Table 1), indicating that a single-gene mutation underlies the nonglaucous phenotype. To test if the mutation was due to transgene insertion, we screened 38 nonglaucous F 2 individuals derived from the cross between NG1 and BW for the transformation selection marker bar gene and RNAi transgene constructs [38] by targeting at the Ubq-bar and gus-nos junctions, respectively. Result showed that eight plants were negative for both the Ubq-bar and gus-nos junctions, and 7 additional plants were negative for the gus-nos junction. Furthermore, NG2 contained neither Ubq-bar nor gus-nos (S1 Fig). These data indicate that one bar locus was not linked with the RNAi locus in NG1 and that the nonglaucous mutation is independent of the transgene insertions.
To test if the nonglaucous mutation is allelic to the known wax production genes in wheat, we crossed S-615 w1w2 near isogenic line (NIL), which is nonglaucous due to loss-of-function at the functionally redundant W1 and W2 loci [37], with BW and the NG2 mutant line. The F 2 population derived from the cross between S615-w1w2 and BW segregated into 133 glaucous and 11 nonglaucous plants, well-fitting the 15 to 1 ratio (P = 0.49112; Table 1) and indicating that BW carries both W1 and W2 functional alleles. The F 1 hybrid plants between S615-w1w2 and NG2 carry glaucousness at low intensity (Fig 2B), indicating that the nonglaucous mutation in BW is not allelic to either W1 or W2, but its wild type allele in S615-w1w2 complements to the functional alleles of the W1 and W2 loci carried by BW. According to the recommended rules for gene symbolization in wheat [39], we designated this new wax gene locus as W3 and the mutant as w3. A population of 270 F 2 plants derived from the cross between the w3 mutant and the w1w2 NIL segregated into 141 glaucous and 129 nonglaucous, which fits a 1 to 1 (P = 0.46477) or 9 to 7 ratio (P = 0.18217; Table 1) and is more similar to the two-locus segregation ratios than a three-locus segregation. These results corroborate that W3 is not homologous to the known wax gene loci W1 and W2 and suggested that W3 is most probably linked with the W1 or W2 locus. To identify the chromosomal location of the W3 locus, we developed a mapping population from a cross between the w3 mutant and Chinese Spring (CS)-T. turgidum subsp. dicoccoides 2B substitution line (CS-TDIC 2B) and genotyped the 82 NG F 2 individuals (168 gametes) using 2BS and 2DS SSR markers for detection of marker-W3 linkage. The result showed that W3 is linked with the 2BS SSR markers and 5.5 cM proximal to the marker loci Xwmc770 and Xgwm148 (Fig 3A).
We also crossed the w3 mutant with CS 2BS deletion lines (2BS-1, -2, -3, -5, -10 and -14). All these deletion lines and nullisomic 2B-tetrasomic 2D (N2B-T2D), in which lack of one pair of chromosome 2B is compensated by two pairs of chromosome 2D, are nonglaucous due to missing the W1 locus. The F 1 hybrid plants between the w3 mutant and the 2BS deletion lines were nonglaucous similar to N2B-T2D, or carried nearly undetectable glaucousness ( Fig 3B). No glaucous individuals were observed in their F 2 populations (Table 1). This result corroborates that the W3 locus is located in the distal region of chromosome arm 2BS.
Furthermore, we genotyped BW and the w3 mutant using SSR markers located in the distal regions of the 42 wheat chromosome arms on wheat genetic maps. All 42 SSR markers were positive and detected monomorphism (S1 Table). This indicated BW and the w3 mutant shares the same genetic background.

Wax Morphology
At stage F10.5.1, wheat plants are flowering, and glaucousness is fully developed. We inspected the cuticle surfaces of flag leaf sheaths of BW and w3 mutant under a scanning electron microscope (SEM) at this developmental stage. The electron micrographs showed clear differences between BW and w3 mutant (Fig 4). In BW, cuticle surface was covered with thick meshwork of long wax crystal tubes ( Fig 4A). In contrast, short wax bodies were appressed to the leaf cuticle surface of w3 mutant at low density ( Fig 4B). This result indicated that the W3 locus has a major effect on cuticle, epicuticular wax in particular, development. Considering that the great contrast in wax crystal morphology and density between BW and w3 mutant was found in the flag leaf sheath, we used this tissue for measuring cuticle permeability, profiling composition of the cuticular waxes and quantifying transcription of cuticle genes.

Cuticle Permeability
We measured the cuticle permeability in two experiments using the flag-leaf sheath: water loss in air and chlorophyll leaching in 80% ethanol. In the first experiment, significantly higher water loss rate was observed in the w3 mutant 1 h after detachment (P < 0.03303) and incremented throughout the time course ( Fig 5A). We inspected stomata, and no significant difference was found in stomata density between BW and w3 mutant,~60 stomata in one view field of 10x20 magnification under microscope. The stomata were closed within one hour after detachment. This suggests that the difference in water loss rate is attributed to cuticle permeability. In the second experiment, w3 mutant showed significantly higher chlorophyll efflux rate from the sixth to eighth hour of treatment (P < 0.02491; Fig 5B). Both lines were similar in total chlorophyll content after 48-hour extraction (P = 0.80435). This corroborates the conclusion from the water loss experiment that change of wax crystal morphology and density led to increase of cuticle permeability in w3 mutant.
Difference in water-loss rate was also found in spikes in relation to glaucousness. Six hours after detachment, top awns of the w3 mutant started to dry ( Fig 6B). Twelve hours after detachment, spike of w3 mutant completely discolored and dried out; awns of w1w2 double recessive line were twisted and some glumes discolored, but BW and F 1 between w3 mutant and w1w2   stayed in normal shape ( Fig 6C). This indicates that W3 played an important role in preventing non-stomatal transpiration.
In addition to β-diketones, wax ester content, especially octadecanoic acid octadecyl ester (C 36 ), was significantly decreased in w3 mutant (P = 0.00002; Figs 7 and 8). Contrary to diketones and wax esters, total aldehyde was significantly increased in w3 mutant (P = 0.01026; Fig  7). With regard to specific carbon length homologues of aldehydes, significant increase of C 28 (P = 0.00309) and C 32 (P = 0.00008) homologues and significant reduction of C 26 (P = 0.00485) and C 30 (P = 0.00008) homologues was observed in w3 mutant (Fig 8). Although no significant difference was found between BW and w3 mutant in the total primary alcohol and alkane content (Fig 7), changes were detected in some homologues of these wax species: C 20 (P = 0.01215) alcohol was reduced in w3 mutant, but C 22 , C 26 and C 28 alcohols (P < 0.00036) and C 33 alkane (P = 0.02579) were significantly increased in w3 mutant (Fig 8).

Transcription of Cuticle Genes
To gain insights into the W3-dependent genetic pathways, we profiled expression of 72 wheat wax candidate genes belonging to six categories [37] in the flag-leaf sheath of BW and w3-BW by qPCR. Of these 72 wax genes, five are responsible for cutin biosynthesis, 17 for fatty acyl elongation, 17 for VLCFA reduction and wax esterification, 21 for VLCFA decarbonylation, seven for wax transport and five for wax regulation (S2 Table). Result showed that expression of 19 genes was significantly down regulated and four up regulated (Fig 9). Of these 19 downregulated genes, nine genes were down regulated more than twofold. These include five CER1 members, FAR5, KCR2, KCS-3 and LTP (Fig 9). Of the five down-regulated CER1 genes, CER1-1 showed greatest fold change, i.e. 3.4-fold down-regulation. Expression of four CER3 genes   Table. doi:10.1371/journal.pone.0140524.g009 reduced only~1.5-fold though they may also be involved in decarbonylation via interaction with CER1 gene products. By contrast, expression of CER4-3, CER4-12, MAH1-5 and WSD1 increased~1.5-fold (Fig 9).

Discussion
Glaucousness is one of the most eye-catching traits and has been used as a morphological marker in wheat genetic studies for~80 years (reviewed in [30]). Its adaptive value in improving crop tolerance to drought and heat was recognized in 1980s [40,41], and the climate change resumed interest in this trait more recently [36,37,42,43]. Nevertheless, basic research of glaucousness in wheat lagged behind compared to other cereal crops, such as barley, a distant relative of wheat. One of the major reasons for this is lack of nonglaucous mutants in polyploid wheat. In this study, we identified a new β-diketone-deficit wax mutation w3 and investigated the function of β-diketone as a major wax component in drought tolerance. Molecular characterization of w3 mutant also shed light on diketone biosynthesis.

Wax Loci
In barley, a diploid species, 79 wax loci have been identified by 1,580 induced eceriferum mutants [44], of which 53 mutant loci were localized to the all 7 chromosomes [45]. Chromosome 2H, which is homoeologous to wheat chromosomes 2A, 2B and 2D, carries 11 wax loci, of which cer-c, cer-q, cer-u, cer-v and cer-zt are located on the short arm and cer-g, cer-n, cer-yb and gsh5 (cer-s) on the long arm. In contrast, only four wax loci, W1, W2, Iw1 and Iw2 have been identified in hexaploid wheat with W1 and Iw1 located on 2BS and W2 and Iw2 on 2DS [30]. In polyploid wheat, mutation rate are 5-to 10-fold higher than diploid species due to genetic redundancy [46,47], but the mutant phenotypes are often masked by the wild type alleles of the homoeologous loci. The W1 and W2 genes are functionally redundant in forming glaucousness, but they differ in chromosomal locations [30] and expression profiles [37], suggesting a possibility that they are paralogous and their orthologs in the homoeologous chromosomes are silenced, deleted, neo-or sub-functionalized, which is an important feature of polyploid evolution [48]. In the present research, we showed W3 is located on chromosome arm 2BS and complements with W1 and W2 for β-diketone production. Different from W1 and W2, homoeologs of W3 in the A and D genomes of polyploid wheat lost their function in glaucousness formation. This makes W3 more tractable for genetic manipulation and molecular study of wheat wax.
In addition to BW, a single functional W3 locus is present in CS-TDIC 2B (Table 1), probably also in CS. But it was previously not detected in the CS aneuploids, including nullisomics, nullitetrasomics (NT) and deletion lines. This failure was because W3 is linked with W1. Deletion of either one causes the nonglaucous phenotype and prevents identification of another wax locus supposing that CS does not carry the W2 allele. We examined the 42 NT lines and found that only N2B-T2A and N2B-T2D were nonglaucous. We also examined all eight homozygous deletion lines available for chromosome arm 2BS (2BS-1, -2, -3, -5, -6, -10,-12 and -14) and found that all of them were nonglaucous. These results confirmed that W1 is located in the distal region of the chromosome arm [30], which prevents to identify additional wax loci on same chromosome arm although several wax loci located on homoeologous chromosome arm 2HS of barley [45]. No segregation in the F 2 populations derived from the crosses between the w3 mutant and 2BS deletion lines suggests that both W1 and W3 are located in the distal bin. The w3 mutants may already exist in other wheat cultivars besides BW, but they are assumed to be the w1 mutant. Therefore, allelism test is essential for identifying new wax gene loci in addition to marker linkage mapping. We are mapping the W1 and W3 loci, and the closely linked molecular markers will provide more detailed information on the organization of these wax loci on 2BS.
The w3 mutation is independent of transgene insertions (S1 Fig), suggesting that it could have been induced by tissue culture process during transformation. It is well known that in vitro tissue culture generates genomic stress, reprograms cell development and causes genetic and epigenetic changes [49]. Some of these changes inherit as morphological or physiological mutants. Spontaneous mutation of other origin, however, cannot be excluded because we recently identified several other morphological mutants in BW, which did not go through tissue culture.

W3 Effect on Biosynthesis of β-Diketones
Like the wax genes W1 and W2 [37], W3 is essential for biosynthesis of β-diketones, but had little effect on biosynthesis of alkanes. This supports the proposal that β-diketones are synthesized through a genetic pathway different from VLCFA synthesis and associated alkaneforming decarbonylation pathways although both β-diketones and alkanes are odd-carbon waxes [26]. Two biosynthetic pathways have been proposed for β-diketone biosynthesis: condensation of two C 16 β-ketoesters via a "biological Claisen reaction" [50] and a modified elongation reaction on a selectively protected β-keto acid precursor [51,52]. Both proposals lack molecular evidence although the latter was based on biochemical analysis of the substrate and inhibitor specificity [26], wax metabolite profiling and genetic analysis of the barley cer-c, cer-q and cer-u mutants [51]. In the both proposed models, a decarboxylative reaction was included. The cer-q and cer-c mutations impair the keto-protected chain elongation and lead to C 13 and C 15 alkan-2-ol formation [51,52]. These secondary alcohols, products of β-keto fatty acid decarbonylation, were not detected either in w1w2 [37] or the w3 mutant in this research, suggesting that these wheat wax production genes act in upstream of the pathway or β-diketone is synthesized through the "biological Claisen reaction" in wheat [50]. qPCR assays showed that transcription of three FAE genes, five CER1 and four CER3 homologs was significantly downregulated in w3 mutant (Fig 9; S2 Table). In Arabidopsis, CER1 and CER3 are involved in VLCFA decarbonylation [11]. Because alkane content remained at the same level in w3 mutant as in the wild type, these decarbonylation genes were not involved in alkane synthesis, but probably participated in the β-diketone-forming decarboxylative reaction. Considering that decarbonylation is the last step of β-diketone synthesis, transcription of these CER1 and CER3 homologs is possibly regulated either directly by W3 or by a substrate-mediated feed-forward loop.
The hydroxy β-diketone isomers are derived from β-diketone by hydroxylation [27], hence their changes in w3 mutant are expected to be proportional. Contrary to this expectation, w3 mutant had a much higher HO-D/β-D ratio compared to BW. This result suggests that hydroxylation reaction enzymes in the BW background are less sensitive to depletion of the substrate, i.e. β-diketone, or W3 promotes synthesis of β-diketone but suppresses its hydroxylation. In agreement with the latter hypothesis, expression of MAH1-7, a homolog of MAH1 that is responsible for midchain alkane hydroxylation in Arabidopsis [10], increased 1.5-fold in the w3 mutant (Fig 9; S2 Table). In another research, we found that interaction between wax production genes W1 and W2 is required for biosynthesis of hydroxy-β-diketones and that up-regulation of MAH1-8 paralleled with β-diketone hydroxylation in the W1W2 double dominant genotype [37]. It would be interesting to see if W3 interacts with W1 and W2 by combining their loss-of-function mutations in the same genetic background and measuring their effect on the hydroxy β-diketone isomers.
In addition to FAE, decarbonylation and hydroxylase genes, expression of three cutin biosynthetic genes, FAR5 and the putative wax transporter LTP was also down regulated in the w3 mutant (Fig 9; S2 Table). It is hard to imagine how a single-gene mutation affects expression of genes functioning in five cuticle pathways. In the model plant Arabidopsis, mutation in wax regulator SHN of the AP2/EREBP family altered property of both wax and cutin and expression of genes in multiple pathways [53][54][55]. We measured expression of five known wax regulator genes, but none of them changed significantly (S2 Table). Accordingly, one possibility would be that W3 encodes a wax regulator that orchestrates transcription of the biosynthetic and transporter genes and those acting at the upstream of β-diketone pathway. This may explain the mutant effect on other wax species, such as wax esters, C 33 alkane, C 22 , C 26 and C 28 alcohol, and aldehydes (Fig 8). An alternative and less likely scenario would be that W3 targets on the initial step of diketone biosynthesis and the expression of these 20 genes are regulated by substrate-mediated feed-forward loops or secondary responses as discussed earlier. In either case, our results suggest a possibility that β-diketone pathway in Triticeae uses some CER1 and CER3 homologs for decarbonylation and MAH1 homologs for hydroxylation.

β-Diketones and Glaucousness
Cuticular waxes from the upper parts of adult Triticeae plants can be divided into two types according to their major constituents: primary alcohol-rich and β-diketone-rich wax. The diploid wheat T. monococcum and T. urartu belong to the former and polyploid wheat T. turgidum and T. timopheevii to the latter type [56,57]. Analyses of NG mutants of barley [27] and wheat [28,36,37,58,59] indicated that β-diketones play an important role in glaucousness development. Wax profiling of the wheat wax gene NILs identified three types of waxes from the flag leaf sheaths: β-diketone-rich wax from the glaucous NILs carrying W1w2, w1W2 or W1W2, alkane-rich wax from the nonglaucous NIL w1w2, and alcohol-rich wax from the nonglaucous NILs carrying the wax inhibitor Iw1 or Iw2 [37]. In BW, β-diketone and hydroxy-β-diketones account for 63% of the total wax load. In the w3 mutant, β-diketones reduced~100-fold. At the same time, wax crystal tubes were almost invisible in the flag-leaf sheath (Fig 4B). This confirms the role of β-diketones in glaucousness determination.
Next to β-diketones, alkanes, mainly C 29 and C 31 homologues, constitute the second major wax component, accounting for 34% of the total wax in BW. Although only significant net change was found in C 33 alkane (Fig 8), proportion of the whole alkanes was increased to 90% due to depletion of the β-diketones. Significant differences were also observed between BW and w3 mutant in aldehydes, primary alcohols, and wax esters, but they had very low abundance, less than 3% of total wax in BW, suggesting that they are less likely to contribute to the mutant phenotype.

β-Diketones and Abiotic Stress
Early physiological studies in cereal crops showed that glaucousness increased grain yield by >7% in barley [60] and wheat [40,41]. More recent experiments performed in Mexico, where no rainfall was received during the growing cycle, showed glaucousness significantly contributed to grain yield even in irrigated condition [42].
Our recent research using the W1, W2, Iw1 and Iw2 NILs indicated that glaucousness significantly reduced cuticle permeability expressed as water loss and chlorophyll efflux. Difference in cuticle permeability was also found among the glaucous NILs, which differed in the hydroxy-β-diketones, and suggested a role of these hydroxy isoforms in drought tolerance [37]. In the present research, we showed that the w3 mutation impairs the biosynthesis of β-diketone and further deplete the hydroxy forms. At the same time, the w3 mutation significantly increased water loss and chlorophyll leaching (Fig 5). In the spike, w3 mutant showed even faster water loss than the w1w2 double recessive plant (Fig 6). One explanation is that the w3 mutant lost 99% of β-diketones (Fig 7), but they reduced 92% in the w1w2 NIL [37]. In addition to cuticular waxes, cutin proper plays an even greater role in protecting the non-controllable water loss as recently demonstrated in Arabidopsis [61] and barley [62]. In the w3 mutant, three of the five cutin genes assayed showed~1.5-fold down-regulation (Fig 9; S2 Table). Transmission electron microscopic inspection of cutin organization in BW and w3 mutant may provide more insights into the effect of the w3 mutation on cutin foundation organization.
Another important function of glaucousness is to reflect extra irradiation [1]. In this respect, β-diketones greatly contribute to heat tolerance. The glaucousness can reduced the photosynthetic temperature by up to 0.7°C in the drought-stressed field [41]. We observed that wellwatered w3 mutant plants died prematurely in the later spring seasons of 2010 and 2011 in a temperature poorly-controlled greenhouse room (Fig 1C). This did not occur in the fall and winter seasons, suggesting that loss of β-diketones in w3 mutant increased its susceptibility to the high temperature. Preliminary result from field experiment in 2014 showed that w3 mutation significantly reduced 1000-kernel weight (Wanlong Li and Karl Glover, unpublished data). Detailed physiological experiments need to be conducted using growth chambers with well-controlled temperatures and relative humidity to shed light on how the w3 mutation affects the thermal characteristics of adult wheat plants.

Plant Materials and Growing Conditions
Transgenic plant #056 was generated in BW by bombardment transformation of RNAi construct pVGS193 (Bi C, Trick HN, Gill BS and Li W, unpublished), which was developed using pANDA, a Gateway-based RNAi vector [38]. Other plant materials are listed in S3 Table with their accession numbers, wax genotypes, wax phenotypes and sources of seeds. The crosses were made by manual emasculation and pollination. All the plants were grown in 4"x4" pots containing Sunshine 1 potting mix #3 (Sun Gro Horticulture, Agawam, MA, USA) supplemented with Multicote 1 8 Controlled-Release Fertilizer (Haifa, Altamonte, FL, USA) in a greenhouse. Temperature in the greenhouse was set as 22°at day and 17°at night, and day length was 16 h.
We used MAPMAKER3.0 [64] for determining the order of marker and gene loci and the Kosambi mapping function for converting the recombination into genetic distance in terms of centi-Morgan (cM) [65]. An LOD score of 3.0 is applied to all marker loci.

Microscopy
For stoma counting and aperture observations, both sides of flag leaf sheaths were coated with 10% cellulose acetate dissolved in acetone using a paint brush. Air-dried for 5 min, the cellulose film was carefully peeled and the imprinted slides were observed under a light microscope at a magnification of 10x20. The fresh cuticle samples of flag leaf sheaths were sputtered with gold powder using the CrC-150 Sputtering System and inspected using a Hitachi S-3400N SEM (Hitachi, Tokyo, Japan).

Cuticle Trait Measurements
At Feekes' stage 10.5.1 (F10.5.1) when wheat plants flower, flag leaf sheathes were detached and immersed in 30 ml 80% ethanol in a 50-ml falcon tube. The capped tube was agitated at 50 rpm on an orbital shaker. A 150-μl aliquot was removed at a two-hour time interval for measuring the absorbance at 648 nm and 664 nm on Synergy 2 Multi-Mode Microplate Reader (Biotek, Winooski, VT, USA). The sample was returned to the same tube after measurement. Six replicates of extraction were conducted each from an individual of the same line. Two measurements were performed for each extraction and their values were averaged for subsequent calculation. The sheaths were also subjected to air-drying at room temperature with 45% relative humidity. Their weight was measured at a one-hour time interval for 12 consecutive hours to evaluate the water loss rate using an AB54-S/FACT analytical balance (Mettler Toledo, Columbus, OH, USA) with an accuracy of ±0. 0001 g.

Wax Extraction and Metabolite Analysis
One flag leaf sheath was detached at stage F10.5.1 and immersed in 10 ml HPLC-grade chloroform containing 2 μg tetracosane as an internal reference, which is not present in plant wax, and gently agitated for 1 min, the tissue was rinsed in another tube containing 5 ml chloroform. Two extracts were merged and purified by filtration with Iso-Disc PTFE-13-2 filter (Sigma, St Louis, MO, USA). For each line, five biological replicates were extracted separately. Purified wax extract was dried under a nitrogen stream. Wax silylation, gas chromatography-mass spectrometry (GC-MS) profiling, and substance identification were performed at the W.M. Keck Metabolomics Research Laboratory of Iowa State University (Ames, IA, USA) on a feefor-service basis. In brief, the dried wax extract was dissolved in 100 μl of BSTFA+TCMS (N, O-Bis(trimethylsilyl)trifluoroacetamide with 1% trimethylchlorosilane) and derivatized at 80°C for 30 minutes. The solution is dried under a stream of nitrogen and the residue is reconstituted in 50 μl chloroform and subjected to GC-MS analysis. GC-MS analysis was performed with an Agilent 6890 GC interfaced to a 5973 mass spectrometer. The HP-5ms column (30 m x 0.25 mm i.d. coated with a 0.25 μm film, Agilent Technologies) was used, and temperature gradient was programmed from 150 to 320°C at 5°C/min with helium flow rate at 1.0 ml/min. Operating parameters for MS were set to 70 eV of ionization voltage and 280°C of interface temperature.

Quantitative RT-PCR
Of the 72 quantitative RT-PCR (qPCR) primer pairs used, 64 were adopted from Zhang et al [37] and eight pairs from Kosma et al. [66] together with their gene name designations. At stage F9.0 when the flag leaf is fully emerged from the whorl and flag leaf sheath is rapidly elongating, the flag leaf sheath was collected and immediately frozen in liquid nitrogen and used for RNA isolation using Trizol reagent (Thermo Fisher Scientific Inc., Waltham, MA, USA) following the manufacturer's instruction. Four biological replicates were used. After evaluating RNA integrity in agarose gel and quantification with Nanodrop ND-1000 (Thermo Scientific, Waltham, MA, USA), 1 μg total RNA was used for reverse transcription using QuantiTect Reverse Transcription Kit (Qiagen, Valencia, CA, USA). Approximately 5 ng cDNA was used as template for qPCR, which was performed on ABI 7900HT High-Throughput Real-Time Thermocycler (Thermo Fisher Scientific Inc.) using the iTaq™ SYBR 1 Green Supermix with ROX (Bio-Rad, Hercules, CA, USA). Two technical replicates were included for each biological replicate. Melt curve analysis was conducted to determine the amplified specificity of PCR products. TaRPII36 was used as the internal reference [37] and relative transcript abundance in the RNA samples was quantified using the 2 -ΔΔCt method as described by [67].

Data Analysis and Statistics
Wax load, components, chlorophyll efflux, water loss, and wax gene transcription were measured from four or more biological replicates. The means, standard deviations and P values were estimated using Microsoft 1 Excel functions. Student's t-tests were performed to evaluate statistical significance of the differences between BW and the NG mutant. Pearson's chisquared test was calculated for the deviation of F 2 segregation from the expected ratios. The cut-off for statistical significance was set to P-value 0.05.