Introduction

Among the various functions of the cuticle, a structure which covers the aerial surface of terrestrial plants, are the prevention of non-stomatal water loss, the inhibition of organ fusion during development (Sieber et al. 2000; Raven and Edwards 2004), protection from UV radiation damage (Barnes et al. 1996) and the imposition of a physical barrier against infection by bacterial and fungal pathogens (Jenks et al. 1994; Riederer 2006). The two major components of the cuticle are the hydroxyl and hydroxyl-epoxy fatty acid polyester cutin, and cuticular wax, which is a complex mixture of straight chain C20–C60 aliphatics. A typical cuticle has an outer coating of cuticular wax (a mixture of long-chain lipids (Walton 1990), a thick electron-translucent middle layer (the “cuticle proper”) and an electron-dense cuticular layer (Taiz and Zeiger 1998). Both cutin and cuticular wax are synthesized in the epidermal cells. In Arabidopsis thaliana, the synthesis of cuticular wax derives from the elongation of saturated C16 and C18 fatty acids, which is catalysed by a multi-enzyme fatty acid elongase (FAE) complex (Joubes et al. 2008; Kunst and Samuels 2009). The resulting precursors are subsequently modified into a range of aldehydes, alkanes, secondary alcohols and ketones via the alkane pathway or into primary alcohols and wax esters via the primary alcohol pathway (Samuels et al. 2008).

Many of the wax-related genes cloned so far in A. thaliana appear to be involved in the synthesis of long-chain fatty acid wax precursors. FATB, for example, has a role in the supply of saturated fatty acids for the synthesis of VLCFAs in the plastid (Bonaventure et al. 2003). LACS2 and LACS1/CER8 are primarily responsible for the CoA esterification of fatty acids en route to wax synthesis (Schnurr et al. 2004; Lu et al. 2009). CER6 (Fiebig et al. 2000) and CER10 (Zheng et al. 2005) are both targeted to the endoplasmic reticulum (ER), where they show, respectively, β-ketoacyl-CoA synthase and enoyl-CoA reductase activity. AtKCR1 is a homologue of a yeast enzyme which catalyses an early step in the elongation of VLCFAs, and functions as a β-ketoacyl-CoA reductase (Beaudoin et al. 2009). PAS2, which encodes a β-hydroxyacyl-CoA dehydratase, is considered essential for normal plant development (Bach et al. 2008). Each of the four latter enzymes is an important component of the FAE complex (Kunst and Samuels 2009). The fatty acyl-CoA reductase CER4 catalyses the production of primary alcohols from VLCFA acyl-CoA (Rowland et al. 2006) and can also act as the substrate for the synthesis of alkyl esters by wax synthase WSD1 (Li et al. 2008). Meanwhile, MAH1, a mid-chain alkane hydroxylase, catalyses the oxidization of alkanes to form secondary alcohols and ketones (Greer et al. 2007).

Mutants of the A. thaliana genes CER3, WAX2, YRE and FLP1 (multiple alleles, hereafter this gene will be referred to as WAX2) are all associated with major loss in the amount of wax produced (Ariizumi et al. 2003; Chen et al. 2003; Kurata et al. 2003; Rowland et al. 2007), but how they participate in wax synthesis remains unclear. In the wax2 mutant, the extent of the reduction is close to 80%, and except for the C30 primary alcohols, the presence of all the other components (alkanes, ketones, aldehydes and secondary alcohols) is markedly reduced (Chen et al. 2003). However, in the maize gl1 mutant (GL1 is the assumed orthologue of WAX2), the dramatic reduction in the presence of aldehydes and alcohols has suggested that GL1 is essential for the elongation process during the synthesis of cuticular wax. Unlike the wax2 phenotype, however, that of gl1 does not involve any post-genital fusion or the formation of a somewhat thinner cuticle membrane (Chen et al. 2003; Sturaro et al. 2005). WAX2 may be regulated at the transcriptional level by the 3′-5′ exoribonuclease CER7 (Hooker et al. 2007).

Here, we describe the characteristics of wsl2, a leaf wax-deficient EMS mutant in rice. The leaves of this mutant produce markedly less cuticular wax than the wild type, and the cuticle membrane is substantially altered in form. The mutated gene has been isolated, and shown to be a homologue of WAX2/GL1, which are known to play an important role in the synthesis of cuticular wax.

Materials and methods

Plant materials and growth conditions

A rice cuticle wax-deficient mutant (wsl2) was identified among a mutagenised population derived by treating the japonica landrace Lijiangxintuanheigu (LTH) with 0.6% w/v EMS. The mutant was crossed with the cv. Nanjing11 to construct an F2 mapping population. A set of 2,017 mutant type F2 progeny was selected at the four-leaf stage to take forward for fine genetic mapping. For a drought test, 2 week-old seedlings were grown under 28°C, 14 h photoperiod. Grain of LTH, Nanjing11 and the standard japonica cultivar Nipponbare and Kitaake were obtained from the Chinese National Key Facility for Crop Gene Resources and Genetic Improvement.

Scanning and transmission electron microscopy

Leaves of 6 week-old seedlings and the lemmas of wild-type and wsl2 plants were air dried (Jenks et al. 1992). Wild-type and wsl2 anthers were fixed for 24 h in 2.5% (v/v) glutaraldehyde (Sigma), rinsed three times for 10 min in distilled water, dehydrated through an ethanol series, fixed in 1% (w/v) osmium tetroxide (Alfa Aesar) for 2 h, rinsed three times for 30 min in 0.1 M sodium phosphate buffer, dehydrated through an acetone series, exchanged for three times with isoamyl acetate (Sigma), and subjected to critical point drying with CO2. The material was mounted on aluminium stubs and sputter-coated with gold palladium in six 30 s bursts (E-1010 sputter-coater, Hitachi, Tokyo, Japan) in preparation for SEM (Hitachi S-3400N). For TEM, leaf specimens were collected from the middle of the blade between the mid-vein and the leaf margin, fixed in 1% (v/v) glutaraldehyde and 1% (w/v) osmium tetroxide for 1 h at room temperature, dehydrated through an ethanol series, infiltrated and embedded in London Resin White (London Reson Co. Ltd.). A series of 80 nm sections was cut using an Ultracut E ultramicrotome (Leica, Wetzlar, Germany) and attached to a formvar-coated copper grid (Electron Microscopy Sciences). The sections were air-dried, treated with 2% (v/v) uranyl acetate for 5 min at room temperature and then stained with lead citrate for 3 min viewing with a TEM (Hitachi H-7500).

Chlorophyll efflux and the rate of water loss

Leaf blades of 12 week-old plants were cut into 8 cm lengths and immersed in 80% ethanol. A series of 100 μl aliquots was taken at 1, 2, 4, 6, 8, 10 and 24 h, and subjected to spectrophotometry (absorption measured at 647 and 664 nm) to quantify the amount of chlorophyll leached. The concentration (μM) of chlorophyll in the leachate was given by 7.93 × A664 + 19.53 × A647 following Lolle et al. (1997). Leaf water loss rate was assessed from detached leaves of 12 week-old plants. The leaves were initially soaked in water in the dark for 1 h, after which excess water was removed and the leaves returned to the dark. The leaves were thereafter weighed every 30 min (Chen et al. 2003). The experiment comprised three replicates.

Cuticular wax analysis

Cuticular wax was extracted from the leaf and sheath of 12 week-old plants by a 30 s immersion in chloroform at 60°C (Bianchi et al. 1979; Haas et al. 2001). As an internal standard, 20 μg n-tetracosane was added to each sample. The solvent was removed by heating to 40°C under nitrogen. The extracted monomers were treated with 100 μl bis-N, N-(trimethylsilyl) trifluoroacetamide and 100 μl pyridine for 1 h at 70°C to transform hydroxyl containing compounds into their corresponding trimethylsilyl derivatives. The composition of the sample was then analysed using a capillary gas chromatograph equipped with an HP-1MS column (30 m length, inner diameter 0.32 mm, film thickness 0.25 μm) and attached to a mass spectrometer (GCMS-QP2010, Kyoto, Japan). The GC–MS protocol was as follows: injection at 250, 50°C for 2 min, ramped to 200 at 20°C min−1, 2 min at 200°C, ramped to 320 at 2°C min−1, 14 min at 320°C with He supplied at 1.2 ml min−1 as the carrier gas. A flame ionization detector was used for quantitative analyses. The quantity of cuticular wax was expressed per unit leaf surface area, with leaf areas measured using a LI-3000C Portable Area Meter (LI-COR Biosciences).

Fine mapping and isolation of WSL2

The genomic DNA of the 2,017 F2 progeny showing the mutant phenotype was subjected to microsatellite analysis (McCouch et al. 2002). A number of de novo indel markers were also developed based on sequence comparisons between the genomic sequences in the critical regions of cv. Nipponbare and the indica cv. 9311. Two pairs of PCR primers (cF1/cR1 and cF2/cR2, sequences given in Table S1) were designed to detect WSL2 transcription. The 8,331 bp genomic DNA fragment containing the entire WSL2 coding region and its immediate up- and downstream sequence was amplified from LTH genomic DNA using the primer pair WSL2F1/WSL2R1 (sequences given in Table S1). A CloneEZ™ PCR Cloning kit (GenScript Co., Nanjing, China) was used to insert the resulting fragment into SmaI digested pCAMBIA1305 to generate the transformation plasmid pWSL2, which was then introduced into Agrobacterium tumefaciens strain EHA105 by electroporation, and from there into the wsl2 mutant, following methods described by Hiei et al. (1994). The presence of the transgene was confirmed using the wsl2-dCAPS assay (primer sequences given in Table S1).

RNA interference suppression of WSL2

The construct pCUbi1390-△FAD2 (ubiqutin promoter and a FAD2 intron inserted into pCAMBIA1390) was used as an RNAi vector (Stoutjesdijk et al. 2002; Wu et al. 2007). Both the anti-sense and sense orientation of a 303 bp fragment (WSL2-A and WSL2-S) from the Os09g0426800 3′-UTR was PCR amplified (primer pairs siWSL2F1/siWSL2R1 or siWSL2F2/siWSL2R2, see Table S1), and separately cloned into the SacI or SnaBI sites of pCUbi1390-△FAD2 using a CloneEZ™ Enzyme kit (GenScript Co.), to form the transformation construct pUbi-dsRNAiWSL2. The binary plasmid was introduced into A. tumefaciens strain EHA105 by electroporation and transformed into cv. Nipponbare, following Hiei et al. (1994).

Bioinformatics

Sequences matching that of WSL2 were identified using the BLAST algorithm and aligned with BioEdit 7.0 software (http://www.mbio.ncsu.edu/bioedit/bioedit.html). A rooted phylogenetic tree was constructed using MEGA v4.0 software (http://www.megasoftware.net/). The robustness of the phylogeny was tested by bootstrapping (1,000 replicates). Transmembrane analysis was performed using TMHMM Server v2.0 (www.cbs.dtu.dk/services/TMHMM/).

RT-PCR, quantitative PCR and promoter-GUS analysis

Total RNA was extracted from the roots, leaves, leaf sheaths, stems and young panicles of 3 month-old cv. Nipponbare wild-type plants, and RNAi transformants, using a RNA prep pure Plant kit (Tiangen Co., Beijing, China), and was reverse transcribed using a SuperScript II kit (TaKaRa), according to the manufacturers’ protocols. An RT-PCR was then conducted based on primer pair cF2/cR2 (targeting WSL2) along with the primer pair Actin1F/Actin1R (sequences given in Table S1) targeting the reference gene ACTIN1. A quantitative PCR was based on total RNA extracted as above from the leaves of 2 month-old plants. The PCR targeted the following genes: OsFATB1 (Os06g0143400), OsLACS1 (Os05g0132100), OsCER6 (Os06g0598800), OsKCR1 (Os04g0483500), OsPAS2 (Os04g0271200), OsCER10 (Os01g0150000), OsCER4 (Os04g0354600), OsCER7 (Os02g0550700), OsMAH1 (Os03g0140100) and WSL2 (Os09g0426800). The rice ubiquitin gene (Os03g0234200) was chosen as a reference gene. The sequences of the relevant primers are given in Table S1. The PCR was based on an SYBR premix Ex TaqTM kit (TaKaRa), following the manufacturer’s protocol. The 2−△△CTmethod was used to calculate relative changes in gene expression (Livak and Schmittgen 2001).

A promoter GUS assay was based on the 3,391 bp of sequence upstream of WSL2, which was amplified by primer pair ProF1/ProR1 (sequences given in Table S1), and the resulting amplicon was cloned into the SmaI site of pCAMBIA1305 using a CloneEZ™ PCR Cloning kit (GenScript Co.). The resulting plasmid pWSL2promoter-35S-GUS was digested by BglII (TaKaRa), and the 35S promoter sequence deleted. The final transformation construct (pWSL2promoter-GUS) was introduced into cv. Kitaake by A. tumefaciens-mediated transformation, as above. GUS activity was assayed following Jefferson (1987). The images were obtained using a Leica MZ16 microscope with DFC420 camera (Leica).

Subcellular localization

A 1.86 kb WSL2 cDNA fragment, amplified by the primer pair gfpF1(XhoI)/gfpR1(SpeI) (sequences given in Table S1), was cloned into the XhoI–SpeI site of the PA7-GFP vector (Sohlenkamp et al. 2002), and this construct was co-transformed into A. thaliana and rice protoplasts using the marker plasmid mCherry-HDEL which contains an ER retrieval tetrapeptide (Chiu et al. 1996; Nelson et al. 2007). The transformed protoplasts were then incubated in the dark at 28°C for 16 h, before being monitored for GFP expression, and observed using a Nikon Eclipse TE2000-U microscope (Nikon Co. Tokyo, Japan).

Results

wsl2 is a cuticular wax mutant

M3-688 was a selection from a screen of 4,000 M3 lines derived from EMS-treated LTH. Its leaves are light green in appearance, its mature height is somewhat less than that of the wild type, and its self-fertility is around 74%, even though its pollen after iodine staining appeared normal (Fig. S1a–c). The leaves of the mutant displayed a thin film wetting, unlike water droplets forming beads by wild-type leaves. This trait was used as a phenotypic marker during the subsequent map-based cloning exercise (Fig. 1a, b). SEM analysis revealed that substantially fewer cuticular wax crystals were deposited on the surface of the mutant’s leaf blade and sheath compared with the wild type (Fig. 1c–h), whereas the outer epidermis waxes on the anther and lemma appeared unaffected (Fig. S1 d–g). The mutant was designated wsl2 (wax crystal-sparse leaf2), following the previously identified wax crystal-sparse leaf mutant wsl1 (Yu et al. 2008).

Fig. 1
figure 1

Phenotype of the rice wild type and the wsl2 mutant. The behaviour of water droplets on the leaves of wild type (a) and wsl2 (b). Cuticular wax formed by wild type (c–e) and wsl2 (f–h), as visualized by SEM, on the adaxial (c, f) and the abaxial surface (d, g) of the leaf blade. e, h Outer epidermis of the leaf sheath. cp cuticular papillae. Bars = 10 cm (a, b), 5 μm (c–h)

Full size image

wsl2 displays altered cuticle permeability

Cuticle permeability is strongly influenced by the quantity of cuticular wax present. The chlorophyll leaching assay showed that chlorophyll was more readily extracted from wsl2 leaves than from wild-type ones (Fig. 2a), and the water loss rate from detached leaves was faster in the mutant than in the wild type (Fig. 2b). When the plants were drought stressed, only 5 out of 40 of the mutant seedlings recovered, whereas 36 out of 40 wild-type seedlings survived (Fig. 2c–e). Thus, the wsl2 cuticle is more permeable than the wild-type one.

Fig. 2
figure 2

Physiological characterization of rice wild-type and wsl2 mutant. a Chlorophyll leaching assays with leave of wild-type (WT) and the wsl2 mutant. b Rate of water loss from detached leaves of WT and wsl2. The results were obtained from three replicates and depicted with standard error of the mean from each time point. c–e Drought treatment with the 2 week-old seedlings in the air for 0 h (c), 10 h (d), seedling recovery 7 days after re-watering (e)

Full size image

wsl2 forms a thicker cuticle membrane

Some A. thaliana wax-related mutants display alterations in their cuticle membranes (Chen et al. 2003; Lee et al. 2009; Lu et al. 2009). TEM analysis of three-leaf seedlings showed that the wsl2 cuticle membrane was somewhat thicker than that of the wild type (Fig. 3a, b). In older plants, this difference became magnified, with the cuticle membrane thickness ranging from 300 to 350 nm in the mutant, but only from 110 to 150 nm in the wild type (Fig. 3c–d). Rather than the compact cuticle proper of the wild type, the mutant cuticle proper had a fluffy appearance and was intermingled with unidentified shadow-stained materials, resulting in the mutant cuticle proper being more fragile and more easily detached from the cuticular layer during sample preparation (Fig. 3d).

Fig. 3
figure 3

TEM analysis of leaf cuticle membranes. The leaf cuticle membrane of a wild-type young leaf (a) and adult leaf (c) comprises the outer electron-translucent cuticle proper and the inner opaque cuticular layer, especially visible in the adult leaf. In the wsl2 mutant, the young leaf (b) and the adult leaf (d) cuticle membrane appears thickened and disorganized, especially in the adult leaf. The zone of detachment occurring during sample preparation is indicated by a white arrowhead in (d). CW cell wall. Bars = 200 nm

Full size image

Composition of the cuticular wax in wsl2

The per area wax content on the wsl2 leaf blade was 1.68 μg/cm2, equivalent to ~20% of wild-type content. The most obvious loss of individual wax components were for fatty acids (0.96 μg/cm2 in wsl2 vs. 2.86 μg/cm2 in the wild type), aldehydes (0.13 vs. 2.60 μg/cm2), and primary alcohols (0.06 μg vs. 2.00 μg/cm2). Relative component contents were also altered. The VLCFA precursors represented 34.3% of the overall wax content in the wild type, but 57.4% in the mutant, while the relative contribution of aldehydes fell from 31.2 to 7.9%, and of primary alcohols from 23.9 to 3.4% (Table 1). The leaf sheath wax behaved similarly (Table 1). Each of the individual constituents (C22–C32 fatty acids, C26–C32 primary alcohols, C28–C34 aldehydes and C23–C33 alkanes) was reduced in the mutant compared with the wild type (Fig. 4a, b). Particularly striking was the change in both C30–C32 aldehydes and C30 primary alcohol, which were just 4.8 and 2.2% on the wsl2 leaf. All the results suggested that, in wsl2, the wax constituents decrease both in acyl reduction pathway and alkane-forming pathway.

Table 1 Leaf blade and leaf sheath cuticular waxes in the wild-type and the wsl2 mutant
Full size table
Fig. 4
figure 4

Cuticular wax composition in the leaf blade (a) and the leaf sheath (b). Bars indicate the mean and standard error from three independent assays. Statistical significance of differences between wild type and wsl2 means indicated by * (P < 0.05) or ** (P < 0.01)

Full size image

Mapping and isolation of wsl2

The mutant phenotype segregated in the F2 mapping population consistently with a 3:1 ratio (P > 0.05, n = 205) indicating that the mutant phenotype is under the control of a single gene. A map based on a moderate number of progeny placed the WSL2 locus on the long arm of chromosome 9 in an interval flanked by the microsatellites RM3700 and RM2437. The fine-mapping population narrowed the interval to 30 kb, flanked by the de novo indel markers In 9-1 and In 9-5, both of which are present on BAC clone OJ1299_A11 (Fig. 5a). This interval contains four recognizable open reading frames, one of which (Os09g0426800) encodes a protein homologous to WAX2/GL1, and was thus a strong candidate for WSL2. The sequence consists of ten exons and nine introns, producing a predicted transcript of length 2,366 bp (AK060786), which is composed of an 1,860 bp coding region, a 158 bp 5′-UTR and a 348 bp 3′-UTR (Fig. 5b). When the 30 kb interval was resequenced in the mutant line, a single base mutation (G to A) was discovered in Os09g0426800. The effect of the variant was to lengthen the transcript by 85 bp, since the sixth intron was no longer spliced, and this resulted in the formation of a premature TAA stop codon (Fig. 5b–d). When the 8.33 kb genomic WSL2 DNA fragment (including the 2.12 kb upstream of the ATG start and the 2.27 kb region downstream of the stop codon, was introduced into the wsl2 mutant, the cuticular wax of the transgenic plants resembled that of the wild type (Fig. 5e, f). This result confirms that the single base change in WSL2 caused the wax crystal sparse leaf phenotype.

Fig. 5
figure 5

Positional cloning of WSL2. a Genetic and physical map, showing the identity of the marker loci and the number of recombinants identified between adjacent markers. b The structure of WSL2. The mutant allele contains a G to A transition at the first base of the sixth intron. Black boxes indicate the coding sequence, white boxes the 5′- and 3′-UTRs, and lines between boxes introns. The start codon (ATG) and the stop codon (TGA) are both shown. c The mutation incorporates the sixth intron into the transcript, introducing a premature TAA termination codon. d The mutant transcript as detected by RT-PCR is longer than the corresponding wild-type one. e–f Genetic complementation of wsl2. e A dCAPS marker based on the mutant′s G to A transition identifies transgenic plants. f SEM analysis of the leaf of the wild type (left), the wsl2 mutant (centre) and a transgenic plant C19, a wsl2 individual transformed with an 8.33 kb DNA fragment composed of 2.12 kb of sequence upstream of the WSL2 ATG start codon, the WSL2 coding sequence and 2.27 kb downstream of the WSL2 stop codon (right). Bars = 5 μm

Full size image

RNAi knock-down of wild-type WSL2

When the construct containing both the antisense and sense version of a 303 bp 3′UTR fragment of WSL2 driven by the ubiqutin promoter (Fig. 6a) was introduced into cv. Nipponbare, RT-PCR analysis revealed that WSL2 expression was disrupted in six of the 21 primary transgenics investigated (Fig. 6b). The leaves of these plants displayed the mutant phenotype (Fig. 6c, d). SEM analysis of the three RNAi plants R4, R8 and R14 showed that they all had the wax crystal-sparse leaf phenotype (Fig. 6e–h). GC–MS analysis of the cuticular wax present on R4 leaves revealed a two- to fourfold reduction in the content of C24–C30 fatty acids, C28–C34 aldehydes and C26–C32 primary alcohols (Fig. 6i, Table S2). However, unlike in cer6 mutants (Millar et al. 1999; Fiebig et al. 2000), no shorter product accumulation was found either in wsl2 or in WSL2 knock-down transformants.

Fig. 6
figure 6

RNAi knock-down phenotype. a Structure of the RNAi construct. b Expression of WSL2, as assessed by RT-PCR in six transformed plants. c, d Behaviour of water droplets on the leaf of wild-type and transformant R4. e–h SEM analysis of wild type and transformants R4, R8 and R14. i Constituents of the leaf cuticular wax of wild-type and RNAi knock-down plants. Bars indicate the mean and standard deviation error from three independent assays. Statistical significance of differences between wild-type and transgenic plants indicated by *(P < 0.05) or **(P < 0.01). Bars = 10 cm (c, d), 5 μm (e–h)

Full size image

WSL2 is a member of WAX2-like protein family

The WSL2 protein comprises 619 residues with a predicted molecular mass of 69.7 kDa and pI of 9.30. The protein includes four transmembrane domains, as well as a functional domain in both its N-terminal and C-terminal regions. The N-terminal region also contains three His-rich motifs and shares partial homology with a fatty acid hydroxylase superfamily involved in cholesterol synthesis and plant cuticular wax synthesis (Arthington et al. 1991). The Wax2_C domain has a conserved LEGW sequence motif with high similarity to short-chain dehydrogenases (Chen et al. 2003). WSL2 shares 85% peptide identity with GL1, and 64 and 37% identity with WAX2 and CER1, respectively (Aarts et al. 1997; Chen et al. 2003; Sturaro et al. 2005). Five WSL2 homologues in rice share 27–69% amino acid identity with WSL2 (Fig. S2a). All 13 proteins showing a substantial level of peptide similarity were classifiable into two clades. The first includes WSL2, GL1, OsGL1-2 (Islam et al. 2009), WAX2, and two sorghum (04g005330, 10g025920) and two maize (LOC100280351, LOC100192547) proteins of unknown function, while the second groups CER1, WDA1 (Jung et al. 2006) and two rice proteins (Os02g0621300, Os04g0512200) of unknown function (Fig. S2b; Table S3).

Expression and intracellular localization of WSL2

RT-PCR analysis of 3 month-old wild-type and mutant plants confirmed that the WSL2 (or wsl2) sequence was strongly expressed in the panicle, leaf and leaf sheath, only weakly in the stem, and not at all in the root (Fig. 7a). The expression of wsl2 appeared to be enhanced in the leaf of the mutant. As expected, the transcript wsl2 transcript was longer than the WSL2 one (Fig. 5b, Fig. 7a). GUS expression, driven by the 3.39 kb fragment upstream of the wsl2 ATG start, was strong in the glume, leaf blade and sheath, weak in the stem, and undetectable in the root (Fig. 7b–g), fully consistent with the RT-PCR result. Of the ten rice homologues of A. thaliana wax synthesis-related genes, four (OsFATB1, OsLACS1, OsCER6, OsKCR1) were more highly expressed in the mutant than in the wild type (Fig. 7h).

Fig. 7
figure 7

Expression of WSL2. a RT-PCR profiles derived from the root (R), stem (S), leaf blade (L), leaf sheath (LS) and panicle (P) of wild-type and the wsl2 mutant. b–g Expression, as detected by a GUS assay in a transgenic plant carrying the WSL2 promoter fused to GUS, in a seedling (b), a stem (c), a leaf sheath (d), a glume (e), a leaf blade (f), and a leaf blade shown under higher magnification (g). h Relative expression of a set of genes associated with wax synthesis in the wild-type and the wsl2 mutant. Bars indicate the mean and standard error derived from three independent assays. Bars = 1 mm (b–f), 0.2 mm (g)

Full size image

Most of the wax synthesis-related enzymes in A. thaliana are located at the ER (Xu et al. 2002; Zheng et al. 2005; Rowland et al. 2006; Greer et al. 2007). The SignalP 3.0 Server (www.cbs.dtu.dk/services/) predicted the presence of an ER-signalling peptide at the 5′-terminus of the WSL2 protein. The intracellular localization of WSL2 was explored using a GFP fusion construct transiently expressed in both A. thaliana and rice leaf protoplasts. The resulting GFP signal co-localized with the ER marker mCherry-HDEL (Nelson et al. 2007; Wang et al. 2010) both in A. thaliana (Fig. 8a–d) and rice (Fig. 8e–h) protoplasts, supporting the suggestion that WSL2 is targeted to the ER.

Fig. 8
figure 8

Intracellular localization of WSL2 in protoplasts of A. thaliana (a–d) and rice (e–h). a, e WSL2-GFP on its own. b, f ER marker mCherry-HDEL. c Merged image of a and b. g Merged image of e and f. d, h DIC images. Bars = 5 μm (a–d), 4.4 μm (e–h)

Full size image

Discussion

The three wax-related genes WSL2, WAX2 and GL1 share a substantial level of homology at the peptide level (85% between WSL2 and GL1, 64% between WSL2 and WAX2). In the wsl2 mutant, the amount of wax present on the surface of both the leaf blade and leaf sheath is considerably reduced compared with the wild type, with most of the components generated both by the acyl reduction and alkane-forming pathways being affected. The A. thaliana wax2 mutant has a similar phenotype, with the presence of alkanes, ketones, aldehydes and secondary alcohols being greatly diminished, although that of the C30 primary alcohols on the stem surface is enhanced (Chen et al. 2003). Compared with its wild type, the maize gl1 mutant produces some 73% less wax on young leaves, largely due to a loss of aldehydes and primary alcohols (Sturaro et al. 2005). The cuticular wax formed on the rice leaf consists mainly of fatty acids, primary alcohols and aldehydes, but in A. thaliana, the cuticular wax on the stem is composed of a mixture of alkanes and ketones (of various lengths), with alcohols, aldehydes and fatty acids accounting for less than 25% of the total (Chen et al. 2003). The contrasting composition of cuticular wax between rice and A. thaliana, together with the different effect on C30 primary alcohol content caused by the loss of function of wsl2 and wax2, suggests that these two genes do not play exactly the same role in the synthesis of cuticular wax.

In all three mutants (wsl2, wax2 and gl1), both the morphology of the cuticle membrane and the composition of the cuticular wax are altered from the wild type (Chen et al. 2003; Sturaro et al. 2005). The wsl2 leaf cuticle membrane was thickened, in addition to produce a clearly visible demarcation between the cuticle proper and the cuticular layer; in wax2 the cuticle proper appears thickened, but there was no evidence for a bilayered structure (Chen et al. 2003). A further feature of the wsl2 mutant is the fact that the cuticle proper becomes easily detached from the cuticular layer during sample preparation. As a result, most of TEM samples lacked the cuticle proper, which created a false impression of the thinness of the cuticle membrane. However, in the gl1 mutant, the cuticle proper was more or less absent, reducing cuticle membrane thickness by about 50% (Sturaro et al. 2005). The extraction of chlorophyll from wax2 and wsl2 leaves is much easier than from those of gl1, which behave identically to wild-type leaves (Chen et al. 2003; Sturaro et al. 2005).

In A. thaliana, a defective cutin structure is often associated with organ fusion and loss of pollen fertility (Lolle et al. 1998; Pruitt et al. 2000; Wellesen et al. 2001). For example, the wax2 phenotype includes various abnormalities in trichome development and postgenital organ fusion (Chen et al. 2003; Kurata et al. 2003). However, as in gl1 (Sturaro et al. 2005), postgenital organ fusion does not occur in wsl2. This difference may reflect the closer phylogenetic relationship existing between the two monocotyledonous species maize and rice, than between either of them and the dicotyledonous species A. thaliana. The wsl2 mutant developed normal wax coating on its anthers, and enjoyed a high degree of pollen fertility (although a somewhat reduced rate of grain set), while wax2 suffers from severe pollen sterility, at least under low relative humidity growing conditions (Chen et al. 2003). The cause of the compromised self fertility of wsl2 is unclear.

WSL2, WAX2 and GL1 each feature two putative conserved domains, namely a fatty acid hydroxylase superfamily domain in the N-terminal region and a wax2 C-terminal domain. In addition, they also contain three conserved His-rich motifs. Almost all wax synthesis-related proteins have been localised to the ER (Samuels et al. 2008; Kamigaki et al. 2009), and the pattern of WSL2 expression is consistent with this (note, however, Qin et al. (2011) reported that the gene product of Osgl1-1, which appears to be the same protein as WSL2, is deposited in the nucleus, cytoplasm, and plasma membrane). According to the Kunst and Samuels (2009) model, long-chain (C16–C18) fatty acyl groups are initially synthesized in the plastid, from where they are exported to the ER and extended to VLCFAs by the FAE complex, and then further modified to form the various aliphatic wax components present in the ER. Wax components can be generated either via the alcohol-forming (acyl reduction) pathway (producing primary alcohols and wax esters) or the alkane-forming (decarbonylation) pathway (aldehydes, alkanes, secondary alcohols, and ketones). The elevated level of C30 primary alcohols present in the wax2 wax was suggested to reflect the encoding by WAX2 of an enzyme involved in the reduction of aldehyde-generating acyl-CoAs (Chen et al. 2003; Kurata et al. 2003). Compared with the wild type, almost all the components of the wsl2 wax were reduced in amount. As the defect set in as early as during C22 fatty acid synthesis, the wsl2 mutation clearly exerted its effect at the initiation stage of the VLCFA elongation process.

The expression of some wax synthesis-related genes differed between the wild-type and the wsl2 mutant. Apart from the expected reduction in WSL2 expression, there was an increase of two- fivefold in OsFATB1, OsLACS1, OsCER6 and OsCER4 (Fig. 7h). In A. thaliana, FATB1 and LACS1 are both thought to participate in the synthesis of <C16 fatty acids precursors (Bonaventure et al. 2003; Lu et al. 2009). A comparison of the wax produced by wsl2 and wild type showed a similar level of C16 fatty acids, indicating that there was no direct effect of enhanced OsFATB1 and/or OsLACS1 expression on short-chain precursor synthesis. The A. thaliana FAE complex consists of CER6, AtKCR1, PAS2 and CER10, and in the absence of CER6 expression stem wax accumulation is abolished (Millar et al. 1999). In wsl2, the co-ordinated expression of OsCER6 and wsl2 could mean that wsl2 is a regulator of the rice FAE complex. The wax-deficient phenotype of the wsl2 mutant is reminiscent of that of various FAE complex gene mutants (Zheng et al. 2005; Bach et al. 2008; Beaudoin et al. 2009), which suggests that wsl2 has a similar role in VLCFA elongation. However, yeast two-hybrid experiments have not so far succeeded in demonstrating any direct interaction between WSL2 and any of OsCER6, OsKCR1, OsPAS2 or OsCER10 (data not shown), so the control of WSL2 over the expression of FAE complex genes may rather be indirect. Unexpectedly, OsCER4 are up-regulated in the wsl2 mutant. In Arabidopsis, CER4 is responsible for producing primary alcohols (Rowland et al. 2006), but despite of OsCER4 increased expression, the wsl2 mutation still blocks accumulation of fatty alcohols. The reason why this kind of apparent paradox happens in wsl2 need to be further explored in future experiment.

Here, we have shown that WSL2 encodes a protein homologous to A. thaliana WAX2 and maize GL1, and is important for the synthesis of cuticular wax and the formation of the cuticle membrane. When the expression of WSL2 is abolished, the synthesis of almost all the constituents of the wax was decreased. Because shorter product accumulation was found neither in wsl2 nor in WSL2 knock-down transformants as in cer6 mutants (Millar et al. 1999; Fiebig et al. 2000), WSL2 likely affects utilization of the elongated products in some way; however, its precise function and roles in wax synthesis clearly needs further study.