Introduction

Many characteristics of the leaf, such as water potential, osmotic adjustment, cell membrane stability, cuticular wax characteristics, and epidermal conductance, can affect drought resistance of plants. Among these, increasing attention has focused on the importance of cuticular wax on the aerial surfaces of plants because it is the outermost waterproof barrier and provides primary protection against environmental stresses (Post-Beittenmiller 1996). A cuticle layer covers most aerial organs of vascular plants and forms the outermost contacting zone between the plant and the environment (Kerstiens 1996a). Among the multiple functions of the cuticle (Barnes et al. 1996; Barthlott and Neinhuis 1997; Eigenbrode 1996; Jenks et al. 1994), the pivotal one is to provide a diffusion barrier against the uncontrolled loss or uptake of water and gases (Kerstiens 1996b, 2006).

A polyester matrix (cutin) is the major constituent of the cuticle that is overlaid or embedded with a long-chain hydrocarbon (cuticular wax) (Kunst and Samuels 2003; Riederer and Schreiber 2001; Vogg et al. 2004). Cuticular waxes are complex mixtures of primarily very-long-chain (VLC, >C18) fatty acids, hydrocarbons, alcohols, aldehydes, ketones, esters, triterpenes, sterols, and flavonoids (Kolattukudy 1980; Walton 1990). The proportions of the major constituents vary among plant species. Cutin, suberin, and wax are all derived from fatty acid precursors and their physical, functional, and biosynthetic relationships strongly suggest that wax production may be co-regulated with cutin and suberin production (James et al. 1999). An increase in cuticular permeability may be associated with a decrease (Jenks et al. 1994; Sturaro et al. 2005) or an increase (Chen et al. 2003) in cuticular thickness or coverage, or with a decrease (Chen et al. 2003; Jenks et al. 1994; Sturaro et al. 2005; Vogg et al. 2004) or increase (Aharoni et al. 2004) in wax coverage, or with a change in cuticle polymer (Xiao et al. 2004) or wax (Aharoni et al. 2004; Vogg et al. 2004) composition.

Plant cuticular wax biosynthesis and its loading to the plant surface are complicated but actively regulated processes (Broun et al. 2004; Jenks et al. 2002). Mutants with reduced wax accumulation or altered wax composition are commonly characterized by a bright green phenotype that can be detected visually (Aarts et al. 1995). In Arabidopsis, 120 cuticular wax mutants representing a total of 31 recessive mutant loci have been identified, although the dominant wax gene mutations have not been reported (Jenks et al. 2002). Wax-deficient mutants have been identified in other plants, including maize, sorghum, barley, and rape (Kunst and Samuels 2003). Studies of the eceriferum (cer) mutants in Arabidopsis and glossy (gl) mutants in maize led to the identification and isolation of a number of wax-related genes. To date, many genes involved in wax biosynthesis or its regulation have been identified in plants by molecular genetic approaches. Among these genes, CER1, CER2, CER6/CUT1, KCS1 (3-ketoacyl-CoA synthase), FIDDLEHEAD (FDH), and WAX2 from Arabidopsis, GL1 and GL8 from maize, and WXP1 from Medicago encode wax synthesis-related enzymes or proteins involved in the transport of wax compounds (Aarts et al. 1995; Chen et al. 2003; Fiebig et al. 2000; Hansen et al. 1997; Millar et al. 1999; Negruk et al. 1996; Pruitt et al. 2000; St-Pierre et al. 1998; Todd et al. 1999; Xia et al. 1996, 1997; Xu et al. 1997; Zhang et al. 2005), whereas CER3, GL2, GL15, and WIN1/SHN1 encode regulatory proteins (Aharoni et al. 2004; Broun et al. 2004; Hannoufa et al. 1996; Moose and Sisco 1996; Tacke et al. 1995). Mutation in most of these genes showed altered wax accumulation (Jenks et al. 2002). Co-suppression of some of the genes in Arabidopsis resulted in reduced wax on stems (Millar et al. 1999; Todd et al. 1999), and over-expression of some of these genes in the Arabidopsis mutants complemented the mutant phenotypes (Fiebig et al. 2000; Hannoufa et al. 1996). However, only a few reports discuss the effect of over-expression of these genes in the wild-type background. Over-expression of the condensing enzyme gene CER6/CUT1 under the control of the CaMV35S promoter failed to promote wax deposition (Millar et al. 1999), whereas under the control of the epidermis-specific CER6 promoter, CER6/CUT1 over-expression led to increased wax load in stems of Arabidopsis (Hooker et al. 2002). The only report on increased wax accumulation in leaf tissues of Arabidopsis was on the over-expression of AP2/EREBP transcriptional activator (Aharoni et al. 2004; Broun et al. 2004). Over-expression of WXP1 under the control of the CaMV35S promoter led to increased cuticular wax loading on the leaf surfaces, reduced water loss, and enhanced drought tolerance of transgenic alfalfa (Zhang et al. 2005). Transgenic expression of WXP1 or of its paralog WXP2 in Arabidopsis also leads to increased wax deposition and enhanced drought tolerance (Zhang et al. 2007).

Rice is one of the most important cereals and has become a model plant of monocot species for functional genomics studies. However, very few genes involved in leaf wax accumulation and drought resistance have been identified in rice. Identification of a T-DNA insertion mutant of the Wax-deficient anther1 (Wda1) gene in rice revealed a significant defect in the biosynthesis of VLC fatty acids in both layers (Jung et al. 2006). Recently, a cuticle-deficient rice mutant with T-DNA inserted in the Wax Crystal-sparse Leaf1 (WSL1) gene was identified (Yu et al. 2008). This mutant displayed pleiotropic phenotypes including reduced growth, leaf fusion, sparse crystals, enhanced sensitivity to drought, and low fertility. In this study, we report a systematic analysis of the Glossy 1-like gene family in rice (OsGL1) for sequence variation and expression profiles under normal and stress conditions. Further, we provide evidence for the function of this gene family in cuticular wax accumulation and drought resistance by checking the phenotypes of the over-expression and mutant plants of the OsGL1-2 gene.

Materials and methods

Identification and sequence analysis of GL1-homologous genes in rice

The reported GL1 protein sequence was used as a query to do BLAST search against the TIGR database (http://tigrblast.tigr.org/euk-blast/index.cgi?project=osa1) to identify putative OsGL1-like genes in the rice genome. The reported GL1 protein also was used to search as queries against the TAIR database (http://www.arabidopsis.org/index.jsp) and the NCBI (http://www.ncbi.nlm.nih.gov/blast/) to find a putative GL1 homolog in Arabidopsis and other species, respectively. All putative protein sequences were manually checked for a common fatty acid hydroxylase (FAH) domain (accession no. PF04116) using the PFAM program (http://www.sanger.ac.uk/Software/Pfam/) to confirm these sequences as members of GL1-homologous gene family.

The basic information of the protein sequences such as molecular weight and theoretical isoelectric point (PI) were obtained by the EXPASY PROTOPARAM tool (http://www.expasy.org/tools/protoparam.html). Protein sequences of all species were analyzed by the MEME program (http://www.meme.sdsc.edu/meme/meme.html) to predict the potential motifs. The full-length cDNAs or predicted coding sequences and the genomic sequences of GL1-like genes in rice were compared to predict the position of exons and introns by using Gene Structure Display Server (http://gsds.cbi.pku.edu.cn/). The positions of domains in different GL1-like genes were also detected using this program.

Multiple sequence alignment of the GL1-like gene family was performed by ClustalX 1.81 with default settings (Thompson et al. 1997). The phylogenetic tree was constructed using PHYLIP software. The neighbor joining method was used to calculate the phylogenetic tree with 1,000 bootstrap replicates. The GENEDOC (Nicholas et al. 1997) and TreeView (Page 1996) programs were used for editing alignment and drawing the phylogenetic tree, respectively.

Plant growth and treatments

For detecting the transcript levels of OsGL1 genes, rice plants of Minghui 63 (Oryza sativa L. spp. indica) were grown in the greenhouse with a 14-h light/10-h dark cycle. Two-week-old seedlings were treated with abiotic and chemical stresses. Abiotic treatments were conducted according to Saijo et al. (2000). Drought stress was applied by exposing intact plants in the air without water supply and plant leaves were sampled at 0, 6, 24, and 48 h after treatment. For cold stress, seedlings were transferred into a growth chamber at 4°C and sampled at 0, 2, 4, 8, 12, and 24 h. The seedlings were submersed in a 200-mM NaCl solution for salt stress and sampled at 0, 6, 24, and 48 h after treatment. Chemical treatment was conducted by spraying leaves with 0.1 mM abscisic acid (ABA) and sampled at 0, 1, 3, and 12 h.

Transcript level analysis of GL1-like gene family in rice

The expression profile of the OsGL1-like gene family was pre-checked by looking up the gene chip expression profiling data of rice for more than 30 different tissues and organs of an elite hybrid rice parent Minghui 63 (Wang et al., unpublished data). In this study, 13 representative tissues and organs of Minghui 63 were selected for reverse transcriptase-polymerase chain reaction (RT-PCR) analysis to confirm the expression profiles. Total RNA was extracted by using TRIzol reagent (Invitrogen), according to the manufacturer’s instructions. Before reverse transcription, total RNA was treated with amplification-grade DNase I (Invitrogen) for 15 min to degrade possibly contaminated residual genomic DNA. The first-strand cDNA was synthesized by Super Script II reverse transcriptase (Invitrogen) according to the manufacturer’s instructions. The PCR was performed in a volume of 50 μl with rTaq DNA polymerase (Takara) and about 1/20 of the first-strand cDNA template generated from 1 μg total RNA. PCR was performed with the following cycling profile: 94°C for 3 min, 25–30 cycles at 94°C for 40 s, 55–60°C for 40 s, and 72°C for 1 min. Rice Actin1 gene (accession no. X16280) was used as an internal control for RT-PCR. For each OsGL1-like gene, a pair of primers with an amplicon of 350–600 bp was used for RT-PCR (Supplementary Table 1). All RT-PCRs were repeated three times with independently reverse-transcribed templates.

Real-time PCR was performed in an optical 96-well plate with an ABI PRISM 7500 instrument (Applied Biosystems). The primers for real-time PCR were designed by Primer Expression Version 2.0 (Applied Biosystems; Supplementary Table 1). Each reaction contained 12.5 μl of 2× SYBR® Premix Ex Tag™ (Takara), 0.5 μl of 50× Rox Reference Dye II (Takara), 4 μl of cDNA samples and 200 nM gene-specific primers in a final volume of 25 μl. The thermal cycles used were as follows: 95°C for 10 s, 45 cycles of 95°C for 5 s, and 60°C for 34 s. Dissociation curve analysis was performed as follows: 95°C for 15 s, 60°C for 1 min, and 95°C for 15 s. The rice Actin1 gene was used as an endogenous control. The relative expression levels were determined as described previously (Liang et al. 2006).

Generation of over-expression and mutant plants of OsGL1-2 gene

Based on the full-length cDNA sequences of GL1-like genes of rice in Knowledge-based Oryza Molecular biological Encyclopedia (KOME) database (http://cdna01.dna.affrc.go.jp/cDNA/), the full-length cDNA of OsGL1-2 was isolated from Minghui 63 by gene-specific primers (Supplementary Table 1) and cloned into pCAMBIA1301S under the control of Cauliflower mosaic virus 35S promoter. The construct was introduced into Agrobacterium tumefaciens strain EHA105 by electroporation and transformed into rice Zhonghua11 (Oryza sativa L. subsp. japonica) following the standard rice transformation protocol (Hiei et al. 1994). Expression of the transgene was checked by northern blot analysis. Total RNA (15 μg) was loaded in each lane of 1.2% agarose gels with formaldehyde for the northern blot analysis. Hybridization was performed with 32P-labeled gene-specific cDNA and results were detected by autoradiography.

To obtain rice mutants of GL1-like genes, the genomic sequences of OsGL1 genes were used to search our rice mutant database (RMD, http://RMD.ncpgr.cn/). Currently, one mutant is available for the OsGL1-2 gene. Genomic DNA was extracted from the young leaves of the osgl1-2 mutant plants, and PCR was performed to genotype the mutant plants by 30 cycles of 94°C for 30 s, 55°C for 30 s, and 72°C for 1.30 min with the primers A (gene-specific forward primer), B (gene-specific reverse primer), and C (NTLB5, T-DNA border specific primer) (Supplementary Table 1).

Phenotyping and measurements of transgenic and mutant plants

Transgenic seeds of T1 families were germinated on MS medium containing 50 mg/l hygromycin for stress testing. At the same time mutant and wild-type (WT) seeds were also germinated. Drought resistance at the reproductive stage was evaluated under drought stress conditions in the refined paddy field (sand:paddy soil = 1:3) facilitated with a movable rain-off shelter. For this experiment, positive transgenic and mutant lines (16 plants each in two rows) were planted along with the WT as a control to follow the randomized complete block design with three replications.

For chlorophyll leaching assay, the third leaf from the top was sampled from each tiller at heading stage plants and the leaf was cut into segments (about 3 cm) and immersed in 30 ml of 80% ethanol at room temperature (gently agitating in the dark). At 1, 3, 6, and 9 h, 3 ml of aliquot were taken out for chlorophyll quantification and the aliquot was poured back to the same tube after measurement. The chlorophyll concentration was quantified using a UNICO 7202B spectrophotometer at wavelengths of 663 and 645 nm by using the standard method (Lolle et al. 1997).

Scanning electron microscopy (SEM) was used to study both abaxial and adaxial surfaces of matured leaves of OsGL1-2 over-expression, mutant, and WT plants. Fragments of flag leaves were fixed in 5% glutaraldehyde and mounted on stubs. Samples were coated with 15–20 Ǻ grain-size gold particles for 15 min by using an EICO IB.5 ION coater. Coated samples were transferred to a JEOL JSM-3690LV scanning electron microscope for examination. For transmission electron microscopy (TEM) analysis, leaf segments (1 × 5 mm) were cut from the leaf (at 3-leaf stage) of OsGL1-2 over-expression, mutant, and WT plants. Samples were processed as described previously (Sturaro et al. 2005). Ultrathin sections were prepared using a Leica UC6 ultramicrotome. Sections were viewed in an H-7650 transmission electron microscope (Hitachi, Japan) operated at 80.0 kV in the public laboratory of Huazhong Agricultural University, PR China.

Wax extraction and quantification

Six leaf blades (each about 10 cm in length) from each plant were immersed in 30 ml chloroform for 30 s at room temperature. The same leaf blades were then re-extracted with chloroform at 60°C for 20 s, and the two chloroform extracts containing wax were pooled. Chloroform was evaporated by a nitrogen evaporator (Dry N-EVAP™111, Organomation Associates Inc. USA). Five microgram of n-tetracosane (C24) was added to each sample as an internal standard. After the samples were transferred to GC vials, 10 μl N,O-bis(trimethylsilyl)-trifluoroacetamide (BSTFA) and 10 μl pyridine were added and the samples were heated at 70°C for 1 h. BSTFA and pyridine were then removed under the stream of nitrogen. With 100 μl chloroform added, the samples were loaded for gas chromatography–mass spectrometry (GC–MS) analysis.

Wax components were separated using a 30 m × 0.32 mm HP-1 capillary column and helium as the carrying gas with constant flow of 2 ml/m in a GC–MS-QP2010 (Plus) gas chromatograph (SHIMADZU, Japan). GC was carried out with on-column injection temperature at 250°C, oven temperature of 2 min at 50°C, increasing at 40°C/min to 200°C, 2 min at 200°C, increasing at 3°C/min to 280°C, 20 min at 280°C. Wax composition was determined by comparing peak retention times with those of reference standards, and by a GC–MS analysis of representative samples. Wax loads were estimated by quantifying the areas of major peaks in comparison with the internal standard. Wax load per unit leaf area was calculated based on the area of leaves used for wax extraction.

Results

Identification of GL1 homologous genes in rice

Using the reported GL1 protein sequence as a query to search against the annotated rice database, 11 putative GL1-like genes were found in the rice genome. These genes, designed OsGL1-1 ~ OsGL1-11, are distributed on chromosomes 2, 4, 6, 7, 10, and 11 (Table 1). In the KOME database, full-length cDNAs are available for all these genes except OsGL1-7. Almost all the rice GL1-like protein sequences contain the FAH domain (PF04116) except OsGL1-7 (Fig. 1a). Comparison of the full-length cDNAs or predicted coding sequences with the genomic sequences of OsGL1 genes revealed large variation in the number of exons among these genes (Fig. 1a; Table 1). We also noticed that the coding region of the FAH domain is interrupted by introns for the all genes except OsGL1-9, in which the entire FAH domain is located in the second exon. According to the PFAM database (http://pfam.sanger.ac.uk/family?acc=PF04116), the members of the FAH superfamily contain two copies of the HXXHH motif. Alignment of the FAH domain sequences suggested all the OsGL1 proteins have two copies of HXXHH motif and an additional His-rich motif (HXALH) (Supplementary Figure 1). By using the multiple EM for motifs elicitation (MEME) software, six putative motifs were identified in the GL1-like protein sequences (Fig. 1b; Table 2), and three of them (motif 1, 2, and 5) contain the His-rich residues. Although the putative functions of these motifs remain unknown, the motif composition patterns are useful clues for deducing the functions of the GL-like proteins. For example, a typical GL-like protein contains all six motifs and its FAH domain contains motifs 1, 2, 4, and 5. Four rice GL-like proteins (OsGL1-8, OsGL1-9, OsGL1-10, and OsGL1-11) only have a nontypical FAH domain with only two His-rich motifs. The sequence of OsGL1-7 is different because it has only two motifs (motif 3 and 6), which constitute the C-terminal region of a typical GL-like protein.

Table 1 General information of OsGL1- like genes in rice
Full size table
Fig. 1
figure 1

Genomic and structural organization GL1-like genes in rice. a Exon–intron structure of OsGL1-like genes. The line indicates introns, and rectangles indicate exons (untranslated regions [UTR] in gray, open reading frames [ORF] in white or black, and the FAH domains in black). b Six putative motifs in GL1-like gene family identified by MEME. Numbered boxes represent different putative motifs (annotation are listed in Table 2)

Full size image
Table 2 Annotation of putative motifs of GL1-like proteins identified by MEME
Full size table

Phylogenetic analysis of GL1-like gene family

For phylogenetic analysis of the GL1-like gene family in plants, all available GL1-like sequences in rice, Arabidopsis and other plant species (such as maize, Medicago, Vitis vinifera, and Senecio odorus) were picked up based on BLASTP search using GL1 as the query. A total of 28 GL1 homologs were used to construct the phylogenetic tree. The results showed that these sequences can be classified into three groups: group I containing the GL1 protein, group II containing CER1 protein, and group III containing SUR2 and hydroxylase/desaturase protein (Fig. 2). Group I and group II each contain GL-like sequences from different species, but group III consists of genes only from rice. OsGL1-1, -2, and -3 were classified into group I and they have high sequence similarity (>60%) to the GL1 protein. OsGL1-4, -5, -6, and -7 were classified into group II and they have low sequence similarity (30% ~ 40%). OsGL1-8, -9, -10, and -11 were classified into group III and had very low sequence similarity to GL1. We also noticed that most of the GL-like sequences from monocot species are closely related in the tree. The phylogenetic result may provide clues for further deducing the putative functions of unknown GL-like sequences of rice.

Fig. 2
figure 2

Phylogenetic relationship of GL1-like gene family from diverse species based on protein sequences. The number at branching sites indicated the posterior probability values for nodal support. The unrooted tree was constructed using PHYLIP software to follow the neighbor joining method. The abbreviations of diverse species: AT, Arabidopsis thaliana; Mt, Medicago truncatula; Os, Oryza sativa; S. odorus, Senecio odorus; V. vinifera, Vitis vinifera; Z. mays, Zea mays

Full size image

Expression levels of OsGL1-like family genes in different tissues and organs

RT-PCR was performed to detect the expression levels of OsGL1 genes in 13 tissues or organs (Fig. 3). OsGL1-1 had a relatively high expression level in germinating seeds (72 h after imbibitions), embryo and radical (after germination), and leaf (4–5 cm young panicle). For the OsGL1-2 gene, a relatively high expression was detected in germinating seeds (72 h after imbibitions), radical (48 h after emergence in dark), and leaf (4–5 cm young panicle). Relatively high expression was detected in germinating seeds (72 h after imbibitions) and stamens (1 day before flowering) for the OsGL1-3 gene. The expression of OsGL1-5 was detected in panicle (young panicle at stage 3), stamen (1 day before flowering), and hull (1 day before flowering). The OsGL1-6 gene showed high expression in germinating seeds (72 h after imbibitions) and shoot (seedling with two tillers). The genes OsGL1-8, -9, -10, and -11 were expressed in most of the tissues and organs investigated. On the other hand, the genes OsGL1-4 and OsGL1-7 had no or very faint expression in all the samples. According to the expression profiles, the genes in the OsGL1 family can be classified into two groups: tissue- or organ-specific expressed genes (OsGL1-1–7) that were expressed in very few tissue/organs and constitutively expressed genes (OsGL1-8–11) that were expressed in most of the tissue/organs. We searched the center of rice expression profiling (CREP) database that was established based on gene chip expression profiles of more than 30 tissues and organs (Wang et al., unpublished data) to check the relative expression levels of the OsGL1 gene family (Supplementary Figure 2). In general, the results from the gene chip analysis matched well with the RT-PCR results.

Fig. 3
figure 3

Semi-quantitative RT-PCR of OsGL1 genes in 13 different tissues and organs Tissues or organs: 1, seed (72 h after imbibition); 2, embryo and radicle (after germination); 3, radicle (48 h after emergence, dark); 4, callus (screening stage); 5, sheath (4–5-cm young panicle); 6, stem (heading stage); 7, panicle (young panicle at stage); 8, flag leaf (5 day before heading); 9, leaf (4–5-cm young panicle); 10, stamen (1 day before flowering); 11, root (seedling with two tillers); 12, shoot (seedling with two tillers); 13, hull (1 day before flowering). Numbers in brackets indicate the number of PCR cycles

Full size image

Responsiveness of OsGL1 genes to abiotic stresses

To elucidate the roles of OsGL1 gene family in stress resistance, we first investigated the transcript levels of OsGL1 genes in rice plants under drought, cold, and salt stresses. Preliminary analysis suggested that six OsGL1 genes were responsive to drought or salt stress in microarray analysis (Zhou et al. 2007). These six genes were selected to test their responsiveness to drought, salt, cold, and ABA (see “Materials and methods” for details) in rice Minghui 63 by real-time PCR analysis. All six OsGL1 genes were responsive to more than one stresses (Fig. 4). Among them, three genes (OsGL1-1, -2 and -6) were induced by drought. The expression levels of OsGL1-1 and OsGL1-2 were gradually increased throughout the time course, whereas the OsGL1-6 showed strong induction only at 2 day after treatment. OsGL1-1, -3, -5, and -6 were induced by salt stress, and the expression levels of these genes except OsGL1-6 continued to increase throughout the time course. The expression level of OsGL1-6 peaked at 6 h after treatment and then decreased. With cold treatment, OsGL1-1 and OsGL1-4 showed strong induction at both 12 and 24 h after treatment, whereas OsGL1-6 showed high expression at 12 h but dramatically decreased expression at 24 h. Four genes (OsGL1-2, -3, -4, and -5) were induced by ABA treatment, and the highest induction was at 12 h after the treatment for all four genes.

Fig. 4
figure 4

Real-time PCR analysis of six OsGL1 genes. The x-axes indicate the time course of stress treatments and y-axes are scales of relative expression level. D, drought; S, salt; C, cold; ABA, abscisic acid. A threshold of 2-fold was used to determine the stress responsiveness of genes

Full size image

Altered drought resistance and cuticular permeability in OsGL1-2 over-expression and mutant rice plants

For functional analysis of GL1-like genes in rice, we searched the T-DNA insertion mutant database (http://rmd.ncpgr.cn/) and three mutants (04Z11ML82, 04Z11ML83, and 05Z11AS53) were found only for the gene OsGL1-2 based on the flanking sequence tags of the mutants. Based on PCR analysis with two genomic primers flanking the insertion sites and a T-DNA vector primer (Fig. 5a), the insertion site was confirmed to be right for the mutant 04Z11ML82 (Fig. 5b) but not for 04Z11ML83 and 05Z11AS53. RT-PCR analysis showed that the transcript of OsGL1-2 gene was broken down in the 04Z11ML82 mutant (Fig. 5c). To study the function of the OsGL1-2 gene, the full-length cDNA of OsGL1-2 under the control of CaMV 35S promoter was transformed into japonica cultivar Zhonghua 11 for examining the phenotypes of wax accumulation and drought resistance along with the T-DNA mutant. Among 23 independent transgenic plants generated, nine transgenic plants showed high expression level of the transgene (Fig. 5d).

Fig. 5
figure 5

Altered drought resistance of the OsGL1-2 mutant. a Schematic gene structure of OsGL1-2 with a T-DNA insertion. b PCR for ten mutant plants. A, forward primer generated from the upstream region of OsGL1-2 gene; B, reverse primer generated from the coding region of the OsGL1-2 gene; C, reverse primer from the T-DNA region. H, homozygous; h, heterozygous; WT, wild-type; CK, control. Lanes 1, 2, and 7 are heterozygous; lanes 3, 6, 8, 9, and 10 are homozygous; lanes 4 and 5 are WT; lane 11 is a control. c Semi-quantitative RT-PCR of the OsGL1-2 transcript level in the osgl1-2 mutant compared to the WT. The expression level of Actin1 was used as an internal control. d Northern blot analysis to check the positive OsGL1-2 over-expression plants. CK, WT. e Appearance of mutant plants (two rows in middle side) under drought stress at reproductive stage compared to WT (both sides). f Mutant leaves became rolled (right side) compared to the non-rolled leaves of WT (left side) after drought stress

Full size image

The most important physiologic function of the cuticle is to protect the plant from water loss. Therefore, we first carried out drought resistance testing for over-expression and mutant plants. Three independent over-expression transgenic families (S8, S9, and S15) and homozygous mutant plants were selected for drought resistance testing in the field (see “Materials and methods” for details). During the process of stress at the reproductive stage, the mutant showed obviously earlier leaf-rolling than WT (Fig. 5e–f). However, we did not observe obvious difference in leaf-rolling time between the over-expression and WT plants. We then checked the cuticular permeability of the over-expression and mutant plants by using the chlorophyll leaching method. Results showed that chlorophyll leaching from the over-expression plant leaves was significantly (t test, P < 0.01) slower than that from the WT leaves, whereas chlorophyll leaching from the mutant leaves was significantly (t test, P < 0.01) faster (Fig. 6a), indicating a decrease and increase of cuticular permeability in the over-expression and mutant plants, respectively. The drought resistance phenotype was also evaluated by measuring water loss in detached leaves over a period of time. Compared to WT, the detached leaves of the OsGL1-2 over-expression plants lost water slowly and the mutant lost water rapidly (Fig. 6b). Although the difference between the over-expression plants and WT or between the mutant and WT was not statistically significant, the difference between the over-expression plants and the mutant reached very significant level (t test, P < 0.01). These results suggest that the OsGL1-2 gene has a significant role in controlling the wax layer and thus protecting leaves from water loss in rice.

Fig. 6
figure 6

Altered cuticular permeability of the OsGL1-2 mutant and over-expression plants. a Chlorophyll leaching assays with matured leaves of OsGL1-2, mutant, and WT, immersed in 80% ethanol for different time intervals. The results are derived from three independent experiments and depicted with standard error of the mean from each time point. One asterisk indicates significant at P < 0.05 level and two asterisks indicate significant at P < 0.01 level by the t test. b Water loss rate of detached leaves of the over-expression, mutant and WT plants. The x-axis is the scale for different time points and the y-axis is percent of free water loss from leaves. Data is shown by mean ± SE with three replicates

Full size image

Altered wax accumulation in the OsGL1-2 over-expression and mutant rice plants

We further checked the wax accumulation on the leaf surfaces of the over-expression and homozygous mutant plants along with the WT by SEM and TEM. Both adaxial and abaxial surfaces of the matured leaves of the over-expression plants were covered with a dense layer of wax crystals under SEM (Fig. 7a, b). Compared to the over-expression plant, fewer wax crystals were observed on both of the surfaces in the mutant (Fig. 7c, d) and WT (Fig. 7e, f). Ultrastructural analysis of the leaf cuticle by TEM indicated that the leaf cuticle membrane of the over-expression plant appears to be divided into an outermost and translucent layer (cuticle layer) and an innermost and opaque layer (reticulated cuticle layer) (Fig. 7g). Compared to WT (Fig. 7i), the cuticle membrane thickness of cuticle layer was increased the over-expression plants (Fig. 7g) and reduced in the mutant (Fig. 7h).

Fig. 7
figure 7

Electron microscopic analysis of the OsGL1-2-overexpression plant, mutant, and WT. a adaxial surface of the OsGL1-2-overexpression plant (OX), b abaxial surface of OX c adaxial surface of the mutant, d abaxial surface of the mutant, e adaxial surface of WT, f abaxial surface of WT. The scale of SEM images in a-f is 5 μm with images taken at ×5,000. g–i images of leaf epidermal cell sections for the OX (g), mutant (h), and WT (i) plants (images were taken at ×21,000). CUT, cuticle; CW, cell wall

Full size image

Wax quantification analysis revealed significant difference in cuticular wax content among the over-expression, mutant and WT plants (Fig. 8). The total proportions of alkanes and fatty acids were significantly higher (t test, P < 0.01) in the over-expression plant leaves than in WT or mutant leaves (Fig. 8a). The total proportions of aldehydes and fatty acids in the mutant leaves were significantly less (P < 0.05) than that in WT (Fig. 8a). Contents of other wax constituents (alkanes and alcohol) were also reduced, but not statistically significant compared to WT. Significant decrease of individual wax constituent was detected for aldehyde C22 (P < 0.05) and fatty acid precursors C18 (P < 0.05) and C20 (P < 0.01) in the mutant compared to WT (Fig. 8b). In general, all the wax constituents detected showed increased levels in the over-expression plants and decreased levels in the mutant compared to WT, though some of them did not reach significant levels.

Fig. 8
figure 8

Quantification of cuticular wax composition in leaves of the OX, mutant and WT plants. a The coverage of total wax constituents including alkane, aldehyde, alcohol, and fatty acid. b The coverage of representative individual precursors of cuticular wax. Values are given as mean ± SE with three replicates. One and two asterisks indicate significance at P < 0.05 and P < 0.01 level, respectively, by t test between the mutant (or overexpression line) and WT

Full size image

Discussion

A common feature of GL1-like proteins is the presence of eight conserved His-rich motifs in the tripartite domain H–X2–4–H, H–X2–3–H–H, (H/Q)–X2–3–H–H (where X stands for any amino acid), which form a di-iron-binding site essential for catalytic activity in a large family of integral membrane enzymes, such as acyl desaturases, alkyl hydroxylases, epoxydases, acetylenases, methyl oxidases, ketolases, and decarbonylases, activities found in prokaryotes and eukaryotes (Shanklin and Cahoon 1998). The positions of the first two His-rich motifs are close to the third motif. One reason for some variation in the spacing between the first two and the third His-rich motifs among protein families is that the structure of this protein part, rather than the spacing between the motifs, is important for the function of the proteins (Aarts et al. 1995). All the proteins with His-rich motifs described so far were found to contain long hydrophobic domains between the first two and the third His-rich motifs (Shanklin et al. 1994). These domains are able to span a membrane twice (Aarts et al. 1995). Our analysis (Supplementary Figure 1) suggested that all the members from different species were from the same ancestor with His-rich motifs (Chen et al. 2003; Kurata et al. 2003).

GL1 shows high sequence similarity with OsGL1-1 (76%), OsGL1-2 (62%), and OsGL1-3 (62%), suggesting that the OsGL1-1 ~ 3 may be homologs of GL1. OsGL1-4, OsGL1-5, and OsGL1-7 are very similar to CER1 (AT1g02205) (Fig. 2) indicating that they may be CER1 homologues in rice. The best-characterized cuticle-related gene in grasses that shows considerable homology (34–35%) to the barley CER1 is maize Glossy 1 (GL1; Hansen et al. 1997; Sturaro et al. 2005). Glossy 1 is the maize homologue of Arabidopsis CER1. The degree of similarity between the putative GL1 and Arabidopsis (AT5G57800) WAX2 proteins is 62%. Our result also shows that WAX2 proteins have close relation to OsGL1-1 ~ 3 proteins (Fig. 2). The WAX2 proteins have a 32% similarity to CER1 and contain certain regions with the homology to sterol desaturases and short-chain dehydrogenases/reductases (Aharoni et al. 2004). It was suggested that these enzymes play a metabolic role in both cutin and wax synthesis and therefore point to a link between cutin and wax metabolism.

We investigated the transcription levels of all OsGL1 genes in 13 different tissues or organs. The results revealed diverse expression patterns of the GL1-like genes in rice. Most of the gene expressed in leaf (4–5-cm panicle stage) but not in root (Fig. 3). Sturaro et al. (2005) also detected a similar expression pattern of the GL1-like genes in maize. OsGL1-1 ~ 3 showed high expression levels in seeds. However, Xu et al. (1997) reported a broad expression pattern of Glossy 8 in maize roots and seed. Kunst and Samuels (2003) reported the close relationship between Glossy 1 and Glossy 8. Most of the OsGL1-like genes were expressed in different panicle stages in rice, suggesting that some of these genes might be involved in the wax biosynthesis in panicle.

As pointed out by Jenks et al. (2002), alternation of wax accumulation in crop leaf tissues is an important physiologic process because leaves are the primary photosynthetic organs, comprise the primary biomass of most agronomical crops, and are often severely affected by environmental stress. To start elucidating the molecular events underlying stress-induced wax biosynthesis, mRNA steady-state levels of wax-related genes were assessed under drought, cold, and salt conditions. These analyses pointed out that in maize and Arabidopsis only a few structural genes involved in key steps of the wax biosynthetic pathway are regulated under stress (Gattuso et al. 2007). Transcript level changes of six OsGL1 genes in different stress treatments clearly suggest that different OsGL1 genes have different responses to stresses. All six genes were induced by more than one stress. In this study most of the drought- or salt stress-inducible OsGL1 genes were also induced by ABA (Fig. 4). The application of ABA to plants partially mimics the effect of stress conditions (Nambara and Marion-Poll 2005). These results indicate that OsGL1 genes may be regulated by substantially diverse regulatory systems under different stresses.

Genetic and mutant studies have suggested that wax accumulation is a potential drought adaptation trait (Jefferson 1994). The osgl1-2 mutant showed rapid water loss and decreased epidermal permeability compared to the OsGL1-2 over-expression and WT plants, suggesting that the mutant plants may be susceptible to drought. The faster leaf-rolling phenotype of the mutant compared to WT under drought stress in the field conditions further confirmed this speculation. In this drought testing, no obvious difference in leaf-rolling time was observed between the over-expression and WT plants, which may due to the acute drought stress (accompanied with high temperature) in that season that had caused a very narrow window of time to observe the difference in the field. Another piece of evidence is that OsGL1-2 over-expression and mutant plants were altered in the cuticular membrane as indicated by the chlorophyll leaching experiment. The OsGL1-2-overexpresion plants showed a significant decrease of chlorophyll leaching and the osgl1-2 mutant plants showed a significant increase of chlorophyll leaching. This result indicates that over-expression of OsGL1-2 genes decreased the cuticular permeability and perhaps can cause drought tolerance of the plants. Chen et al. (2003) reported similar results in the wax2 mutant of Arabidopsis.

Mutation of GL1 causes dramatic alternations in the crystallization patterns of cuticular waxes (Bianchi et al. 1985; Chen et al. 2003; Jenks et al. 1995; Lorenzoni and Salamini 1975). The crystallization patterns of cuticular waxes on matured leaves of the osgl1-2 mutant were significantly different and reduced compared to OsGL1-2 over-expression and WT plants (Fig. 7). The content of total wax and some of the wax constituents were significantly decreased in the mutant leaves compared to WT, indicating that mutation of this gene may block the elongation–decarboxylation pathway I (ED-I) or reduce accumulation of alkanes, aldehydes, and alcohols. Maddaloni et al. (1990) reported that gl1 mutation either blocks an early step in ED-I or interferes with the supply of precursors to ED-I in maize. This mutant has a dramatic reduction of alkanes, aldehydes, and alcohols, but the level of esters is similar to WT. The gl1 mutant, however, does not show an accumulation of metabolic intermediates due to a specific block in the pathway. Instead, there seems to be an increased flux of precursors into an alternative branch pathway, resulting in more complex changes in wax composition (Lemieux et al. 1994; von Wettstein-Knowles 1979).

According to the biosynthesis model of cuticular wax, the early precursors are short acyl chains activated by a soluble plastidic acyl carrier protein (ACP). The acyl chains are elongated by plastidic fatty acid synthetase (FAS) complex which condenses acetal groups from malonyl-ACP onto a growing chain (Ohlrogge and Jaworski 1997). Once C16 and C18 acyl-ACPs are synthesized in the plastids, acyl-ACP thioesterase cleaves the ACP and releases free C16 and C18 fatty acids (palmitic and stearic acids) into the cytoplasm, where they are activated by acyl-coenzyme A (CoA) synthetase via condensation with CoA. The C16 and C18 acyl-CoAs are recruited for conversion into waxes mainly via a membrane-associated pathway where they are modified by a series of cytoplasmic enzyme complexes. Our data showed that the fatty acid precursors C18 and C20 were significantly decreased in the osgl1-2 mutant compared to WT and the OsGL1-2-overexpression plants. This result suggests that OsGL1-2 may play a critical role in the biosynthesis of cuticular wax of leaf in rice.

In conclusion, the GL1-like genes in rice were systematically analyzed for their sequence and expression diversification in this study. Over-expression of one of these genes, OsGL1-2, resulted in an increased cuticular wax accumulation on the leaf surface of the transgenic rice, and the mutant of this gene showed reduced cuticular wax synthesis and significantly increased sensitivity to drought stress. Genetic modification of this gene may have great potential for improving drought resistance of rice.