The wild diploid grass Ae. tauschii, hexaploid wheat (T. aestivum L.) cv. Chinese Spring (CS) and wheat nullisomic–tetrasomic lines were grown at the experimental farm of Northwest A&F University, Yangling, China. Eight-week old leaves at the seedling stage and flag leaves, leaf sheaths, peduncles, glumes, and anthers (160 days old) at the flowering stage were used for analysis of wax composition and micromorphology. Simultaneously, materials were collected for analysis of tissue-specific expression patterns of Ae.tFARs. The seedling leaves and flag leaves were used for quantitative real-time PCR (qRT-PCR) and cuticular wax analysis in wild-type wheat and nullisomic–tetrasomic lines.

Wax load and chemical composition were determined in Ae. tauschii and hexaploid wheat. Various organs including seedling leaves, flag leaves, leaf sheaths, peduncles, glumes, and anthers from Ae. tauschii and leaves from wheat were harvested and immersed in chloroform for 60 s to extract waxes. For analysis by GC-MS and GC-FID, dried wax samples were derivatized with equal amounts of pyridine (Alfa Aesar) and BSTFA [bis-N,O-(trimethylsilyl)trifluoroacetamide] (Fluka) for 40 min at 70°C in a small GC vials. Then, samples were dried under nitrogen gas before being re-dissolved in chloroform (Wang et al., 2015c). n-tetracosane (C24) was added as an internal standard. Wax compounds were quantified by gas chromatography equipped with mass spectrometric detector (GC-MS) (GCMS-QP2010, SHIMADZU, Japan). Each wax component load was calculated based on peak area of each compound and that of the C24 internal standard. The sample surface areas were determined using ImageJ software.

To observe the micromorphology of wax crystals, various organs including seedling leaves, flag leaves, leaf sheaths, peduncles, glumes, and anthers were gently flattened and dried in oven. Sample segments were carefully mounted onto scanning electron microscopy (SEM) stubs, and then coated with superfine platinum powder in an ion sputter coater (Hitachi E-1045, Japan). A Hitachi S-4800 field emission SEM was used to view the surfaces of coated samples at an accelerating voltage of 5 kV (Wang et al., 2016).

Total RNA was extracted from seedling leaves of wild Ae. tauschii using Trizol extraction reagent (TaKaRa, Japan). Then, 1 μg RNA was used for first-strand cDNA synthesis by PrimeScript^® reverse transcriptase (TaKaRa) with primer oligo (dT)18. To amplify the putative Ae.tFARs, PCR was performed with HS PrimeSTAR^® DNA polymerase (TaKaRa) and the following cycling conditions: initial denaturation at 94°C for 1 min, followed by 35 cycles, 98°C for 10 s, 58°C for 30 s, 72°C for 2 min. Gene-specific primers Ae.tFARx were used to amplify the full-length coding sequences (Supplementary Table 1). The amplified DNA fragments were separated and purified using the DNA purification kit (TIANGEN, China). These corresponding fragments were cloned into pMD-18T vector (TaKaRa) and sequenced. Subsequent construction of expression vectors was made using this clone as template.

The entire coding sequences of Ae.tFAR1, Ae.tFAR2, Ae.tFAR3, Ae.tFAR4, and Ae.tFAR6 were amplified using primers shown in Supplementary Table 1. The amplified DNA fragments were double digested and cloned into the corresponding sites of the yeast (Saccharomyces cerevisiae) expression vector pYES3 (Invitrogen) under the control of GAL1 promoter to generate pYES3-Ae.tFARx. A vector of p416 MET25-FLAG3:Sur4-F262A/K266L was used to obtain the yeast mutant INVSur4# (a mutation in ketoacyl-CoA synthase, KCS), which allows for the production of fatty acids with longer chain lengths in yeast (Denic and Weissman, 2007). The p416 MET25-FLAG3:Sur4-F262A/K266L was co-transformed with pYES3-Ae.tFARx into wild-type INVSc1 strain (MATa his3-D1 leu2 trp1-289 ura3-52). The corresponding empty vectors were used as positive control as detailed in Supplementary Table 2.

The yeast transformation was carried out as described by Gietz and Woods (2002). Yeast cells transformed with different vector combinations were cultured on synthetic complete (SC) selection medium lacking tryptophan and uracil with 2% (w/v) glucose for 48 h, induced by 2% (w/v) galactose for 12 h and followed by incubation in 0.1 M potassium phosphate and 2% (w/v) glucose for 24 h at 28°C. For extraction of lipids from yeast cells, 40 mL of liquid cultures were collected, re-suspended in 20% KOH/50% ethanol, and extracted three times with hexane at 70°C for 5 min as described by Wang et al. (2011).

For analysis of fatty alcohols, lipid mixtures from yeast were separated and purified by thin layer chromatography (TLC) plate (5 cm × 10 cm, silica gel, HSGF254, 0.15–0.20 mm) using chloroform as the mobile phase. Different lipids on the plates were then stained with primuline and observed by UV light. The fatty alcohol bands were scratched off and were further analyzed by GC-MS and GC-FID.

Transcript levels were determined by qRT-PCR. Various tissues including seedling leaves, flag leaves, leaf sheaths, peduncles, glumes, and anthers from Ae. tauschii and seedling leaves and flag leaves from wheat were used for total RNA extractions. The RNA samples were then used for cDNA synthesis by PrimeScript^® reverse transcriptase following standard protocols. The qRT-PCR was performed using SYBR^® Premix Ex Taq^TM (TaKaRa) with gene-specific primers (Supplementary Table 1). Ae. tauschii GAPDH gene (GenBank accession no. KD518701) and wheat Actin gene (GenBank accession no. DQ115883) were used as internal expressed controls. Primers of GAPDH and Actin were designed to amplify a 177- and 118-bp fragment, respectively. Real-time PCR was carried out in 25 μl reaction volume using a CFX96 real-time PCR detection system (BIO-RAD, United States) with the following program: 1 cycle of 30 s at 95°C, followed by 40 cycles of 5 s at 95°C and 30 s at 60°C.

Multiple sequence alignments were performed with DNAMAN program using amino acid sequences of FARs from different species. A phylogenetic tree was created with the MEGA software (version 5.1) based on a Maximum Likelihood method (1,000 replicates). TMHMM Server (version 2.0) was used to predict potential transmembrane domains.

To investigate genes involved in wax biosynthesis in Ae. tauschii, we first analyzed the wax load and composition from various organs, including seedling leaves, flag leaves, leaf sheaths, peduncles, glumes, and anthers by GC-MS and GC-FID. The results showed that the waxes from different organs consisted of the same chemical classes, including fatty acids, primary alcohols, esters, aldehydes, alkanes, hentriacontane-14, 16-dione (β-diketone), and OH-β-diketone (Table 1), however, the relative amount of each chemical class was quite different depending on organs. Alkanes and β-diketones were the dominant chemical species in panicle and glume, respectively, while fatty acids and aldehydes were major chemical classes in anthers. Primary alcohol was found to be the richest chemical class in leaves, accounting for 84% of total wax load in seedling leaves and 76% in flag leaves, whereas only small amounts of alkanes and aldehydes were detected in leaves (Figure 1 and Table 1). The high level of primary alcohol was also observed in leaf sheaths, being 57% of total wax load (Table 1). The chain lengths of primary alcohols ranged from C20 to C32, among which C26 was the most prevalent one in seedling leaf, flag leaf, and leaf sheath (Figures 1A–C). Intermediate levels of primary alcohols, β-diketone and alkanes were observed. Chain length distributions of primary alcohols in peduncles and glumes varied from C20 to C32 and the dominant chain length was C28, which is different from that in leaves (Figures 1D,E). These results suggest that primary alcohols are the key wax components and vary considerably depending on organs in leaves of Ae. tauschii.

Since wax morphology is influenced by wax composition (Zhang et al., 2013), we investigated the morphology of wax crystals deposited in various organs by SEM. In seedling leaves, a dense array of wax platelets was found covering both adaxial and abaxial surfaces (Figures 2A–D), while in flag leaves, wax crystal density of both sides was different, with thicker platelet-shaped wax crystals being on adaxial surface (Figures 2E,F) and scattered platelet-shaped crystals being on abaxial surface (Figures 2G,H). Similarly, platelet-shaped crystals were found in leaf sheaths, peduncles, and glumes, but distributed sparsely (Figures 2I–N). In anthers, wave-shaped wax crystals were found (Figures 2O,P). These observations are consistent with studies showing that platelet-shaped wax crystals are associated with the deposition of primary alcohols in wheat leaf blades (Hallam and Chambers, 1970; Gülz, 1994).

Previous studies have shown that formation of primary alcohols was catalyzed by FAR in the acyl-reduction pathway (Kunst and Samuels, 2003; Samuels et al., 2008). Since high level of primary alcohols are present in Ae. tauschii leaves, FAR may play a significant role in the accumulation of leaf cuticular wax in Ae. tauschii. We thus continued to search for the genes encoding FARs in Ae. tauschii.

Using the amino acid sequence of Arabidopsis CER4/FAR3 (GenBank accession no. NP567936) as a query, a BLAST search was performed in the Ae. tauschii genome database (NCBI BioProject ID: PRJNA182898). In total, 10 FAR-like genes (designated as Ae.tFAR1 to Ae.tFAR10) were obtained, with amino acid sequences being 24–47% and 40–100% identical with Arabidopsis CER4 and wheat FARs over the entire length, respectively (Supplementary Tables 3, 5).

Tissue-specific expression patterns of these putative Ae.tFAR genes were determined by qRT-PCR using RNA derived from various organs, including seedling leaves, flag leaves, leaf sheaths, peduncles, glumes, and anthers (Figure 3A). The results showed that Ae.tFAR3 was the most expressed gene in seedling leaves and flag leaves, while Ae.tFAR5 was largely restricted to seedling leaves and anthers. Ae.tFAR4 was found to be widely expressed in all aerial organs, whereas Ae.tFAR7 to Ae.tFAR10 genes were expressed only at later developmental stages of Ae. tauschii. Ae.tFAR1, Ae.tFAR2, and Ae.tFAR6 transcripts were detected in all organs, but at low levels.

In order to identify FARs associated with biosynthesis of leaf cuticular wax, FAR genes expressed in leaves of Ae. tauschii were selected for further analysis. Of six genes (Ae.tFAR1–6) expressed in leaves, Ae.tFAR5 was 100% identical with a known alcohol-forming FAR (TaFAR5) in hexaploid wheat (Wang et al., 2015c). Thus, we subsequently focused on the other five FAR-like genes (Ae.tFAR1–4 and Ae.tFAR6) in Ae. tauschii. A phylogenetic analysis using neighbor-joining methods showed that the five Ae.tFARs were highly related to FARs from other species and belonged to the same monocot clade as wheat FARs (Figure 3B). These five Ae.tFARs were further analyzed to determine whether they are involved in the biosynthesis of primary alcohols in the leaf blades of Ae. tauschii.

To isolate the five FAR candidates, reverse transcription-PCR (RT-PCR) was performed with sets of specific primers and templates that were derived from leaf blades of Ae. tauschii. Sequence analysis revealed that Ae.tFAR1, Ae.tFAR2, Ae.tFAR3, Ae.tFAR4, and Ae.tFAR6 cDNA contained open reading frames of 1,497, 1,470, 1,497, 1,524, and 1500bp, encoding a polypeptide of 498, 489, 498, 507, and 499 amino acids, respectively. The molecular masses of Ae.tFAR1, Ae.tFAR2, Ae.tFAR3, Ae.tFAR4, and Ae.tFAR6 protein were predicted to be 57.0, 55.8, 55.7, 57.3, and 56.7 kDa, respectively.

A predictive protein analysis indicated that the five Ae.tFARs contain two distinct FAR protein domains: a Rossmann-fold NADB binding domain (FAR-N_SDR_e) linked with a FAR domain (FAR_C) (Figure 4A), in accordance with FAR proteins from other plant species, such as TaFAR1, TaFAR2, TaFAR3, and TaFAR4 in wheat (Wang et al., 2015b, 2016), and AtFAR5 and AtFAR8 in Arabidopsis (Chacón et al., 2013). Proteins with these two domains are believed to function as FARs, acting on medium- and long-chain fatty acids. Furthermore, the Rossmann-fold domain had two classical conserved motifs, a NAD(P)H binding site motif (TGXXGXXG) and an active site motif (YXXXK) (Figure 4B). Analysis by TMHMM program revealed that Ae.tFAR3 and Ae.tFAR4 contain one transmembrane helix between residues 391 to 413 and 411 to 433, respectively. This feature was found to be similar with that in jojoba FAR, which also contains membrane-spanning domains (Metz et al., 2000). However, no transmembrane domain was found in Ae.tFAR1, Ae.tFAR2, and Ae.tFAR6, which is similar to the alcohol-forming Arabidopsis FAR3/CER4 (Rowland et al., 2006, Figure 4B). Based on these analyses, we hypothesized that these five Ae.tFAR proteins might have the reductive activity of FAR and be involved in primary alcohol biosynthesis in leaf blades of Ae. tauschii.

To prove that Ae.tFAR1, Ae.tFAR2, Ae.tFAR3, Ae.tFAR4, and Ae.tFAR6 are FARs responsible for the production of primary alcohols, we expressed the five Ae.tFAR genes in a yeast system. The open reading frames of the five Ae.tFAR genes were cloned into the yeast expression vector pYES3 downstream of the GAL1 promoter. Since C20 to C32 primary alcohols are components of leaf cuticular wax in Ae. tauschii, the presence of same chain lengths of VLCFA substrates in yeast is required. However, wild-type yeast can not produce VLCFAs of longer than C28, which may hamper us from correctly assessing the catalytic capacities of Ae.tFARs (Oh et al., 1997; Rowland et al., 2006; Domergue et al., 2010). To circumvent this problem, we used a yeast mutant INVSur#4 by introducing the vector p416 MET25-FLAG3:Sur4-F262A/K266L into wild-type yeast. This yeast mutant was capable of producing ≥C28 fatty acids (Denic and Weissman, 2007; Bernard et al., 2012, Supplementary Table 2). When Ae.tFARs were expressed in INVSur4#, all transgenic yeast cells produced novel chemical compounds (Figure 5), which were identified as primary alcohols ranging from C16 to C20 in yeast cells expressing Ae.tFAR1 and Ae.tFAR2 (Figures 5B,C), from C22 to C30 in yeast cells expressing Ae.tFAR3 and Ae.tFAR6 (Figures 5D,F), and from C22 to C26 in yeast cells expressing Ae.tFAR4 (Figure 5E). In contrast, yeast cells hosting empty vectors accumulated no primary alcohols (Figure 5A). It is noted that all these primary alcohols produced by Ae.tFARs in yeast were consistent with those detected in the leaf blades of Ae. tauschii in our study (Figure 5G).

Further analysis revealed that each Ae.tFAR showed distinct chain length preferences. Expression of Ae.tFAR1 resulted in large quantities of long-chain C16 alcohol (97.6%), and very small amounts of C18 (2.3%) and C20 (0.1%) alcohols. In contrast, expression of Ae.tFAR2 mainly produced C18:0-OH (99.2%), with traces of C16:0-OH (0.4%) and C20:0-OH (0.3%) being detected. It appeared that Ae.tFAR4 preferred producing C24:0-OH (86.8%) to C26:0-OH (10.4%) and C22:0-OH (2.8%). Ae.tFAR6 was primarily responsible for generating C28:0-OH (91.3%). For Ae.tFAR3, the major product was C26:0-OH (62.2%), followed by C28 (34.1%), C24 (1.8%), C30 (1.0%), and C22 (0.8%) alcohols (Table 2). These results confirmed that the five Ae.tFARs from Ae. tauschii are FAR, being able to produce long and very-long-chain primary alcohols. Notably, Ae.tFAR3 mainly produced C26 primary alcohol, which is the richest alcohol species in the leaf blades of Ae. tauschii.

In order to further confirm that the five Ae.tFARs are able to generate primary alcohols, we investigated the functions of Ae.tFAR paralogous genes in wheat. Ae.tFAR amino acid sequences were used as queries to search the wheat genome database (Ensembl Plants). Nine sequences in wheat D genome were found to be identical with Ae.tFAR amino sequences (Supplementary Table 3), while no sequence was found to match with Ae.tFAR6 amino acid sequence. The paralogous genes of Ae.tFAR1, Ae.tFAR2, Ae.tFAR3, Ae.tFAR4, Ae.tFAR5, Ae.tFAR7, Ae.tFAR8, Ae.tFAR9, and Ae.tFAR10 are located on wheat chromosome 3D, 7D, 5D, 4D, 4D, 7D, 1D, 3D, and 4D, respectively. We then examined the relative expression levels of these Ae.tFAR paralogous genes in seedling and flag leaves of wheat. The results showed that the Ae.tFAR3 and Ae.tFAR4 paralogs, Traes_5DL_5597A11EC.1 and Traes_4DL_9480F40CF.1, were highly expressed in wheat seedling leaf and flag leaves, respectively (Figure 6), highlighting the importance of Ae.tFAR3 and Ae.tFAR4 in the primary alcohol biosynthesis in the leaf blades.

The wheat nullisomic–tetrasomic substitution lines were used to further investigate Ae.tFAR3 and Ae.tFAR4 paralogs in hexaploid wheat. cDNA was isolated from a set of seven nullisomic–tetrasomic substitution lines of wheat cv. CS with the loss of one D-chromosome and then was used as template for the amplification of Ae.tFAR3 and Ae.tFAR4 paralogs. As shown in Figure 7A, no amplification product was detected in N5DT5B and N4DT4A, which lacked the whole of chromosome 5D and 4D, respectively. In contrast, PCR products were obtained from all the other six nullisomic–tetrasomic substitution lines. Our results further confirm that Ae.tFAR3 and Ae.tFAR4 paralogs (Traes_5DL_5597A11EC.1 and Traes_4DL_9480F40CF.1) are located on wheat chromosome 5D and 4D. The substitution lines N5DT5B and N4DT4A lacking Ae.tFAR3 and Ae.tFAR4 paralogs, respectively, were then further analyzed. The wax mixtures from seedling and flag leaves of these substitution lines lacking of corresponding Ae.tFAR3 and Ae.tFAR4 paralogs showed a significant reduction in primary alcohols. Compared with wild-type wheat, the N5DT5B substitution line lacking of Ae.tFAR3 paralog (Traes_5DL_5597A11EC.1) exhibited a large decrease in the amount of C26:0-OH and C28:0-OH primary alcohols in seedling leaf (Figure 7B), while N4DT4A with the loss of Ae.tFAR4 paralog (Traes_4DL_9480F40CF.1) showed a notable decrease in C24:0-OH and C26:0-OH primary alcohols in seedling and flag leaves (Figures 7B,C). Altogether, these results might hint that Ae.tFAR3 and Ae.tFAR4 encode active FARs, catalyzing the biosynthesis of primary alcohols in the leaf blades of Ae. tauschii.