Cloning and Characterization of GLOSSY1, a Maize Gene Involved in Cuticle Membrane and Wax Production^,

Abstract

The cuticle covering the aerial organs of land plants plays a protective role against several biotic and abiotic stresses and, in addition, participates in a variety of plant-insect interactions. Here, we describe the molecular cloning and characterization of the maize (Zea mays) GLOSSY1 (GL1) gene, a component of the pathway leading to cuticular wax biosynthesis in seedling leaves. The genomic and cDNA sequences we isolated differ significantly in length and in most of the coding region from those previously identified. The predicted GL1 protein includes three histidine-rich domains, the landmark of a family of membrane-bound desaturases/hydroxylases, including fatty acid-modifying enzymes. GL1 expression is not restricted to the juvenile developmental stage of the maize plant, pointing to a broader function of the gene product than anticipated on the basis of the mutant phenotype. Indeed, in addition to affecting cuticular wax biosynthesis, gl1 mutations have a pleiotropic effect on epidermis development, altering trichome size and impairing cutin structure. Of the many wax biosynthetic genes identified so far, only a few from Arabidopsis (Arabidopsis thaliana) were found to be essential for normal cutin formation. Among these is WAX2, which shares 62% identity with GL1 at the protein level. In wax2-defective plants, cutin alterations induce postgenital organ fusion. This trait is not displayed by gl1 mutants, suggesting a different role of the maize and Arabidopsis cuticle in plant development.

The cuticle forms the outermost layer of the above-ground parts of most plants. The physical and chemical properties of this structure support vital functions such as prevention of nonstomatal water loss, protection against UV irradiation, and reduction of deposition of dust, pollen, and air pollutants. In addition, it plays a critical role in plant defense against bacterial and fungal pathogens and participates in a variety of plant-insect interactions (Post-Beittenmiller, 1996).

The cuticle is synthesized by the epidermal cells and consists of an outer layer of epicuticular waxes overlaying the cuticle membrane, which is composed of a network of interesterified hydroxy and epoxy-hydroxy fatty acids of mainly 16 and 18 atoms in length (cutin) interspersed by intracuticular waxes (Walton, 1990). Cuticular waxes consist primarily of complex mixtures of aliphatic molecules of mainly 16 to 34 carbon atoms in length that occur as free fatty acids, aldehydes, primary alcohols, alkanes, and esters. Their production is a biologically complex process involving a host of synthetic and transport mechanisms. The composition of cuticular waxes differs among plant species, organs, and tissues, and during development. Wax deposition on the leaf surface is also regulated by environmental signals such as light, moisture, and temperature (Kolattukudy, 1996).

Although advances have been made in the understanding of the biosynthesis of specific cutin and wax constituents, many questions pertaining to the organization and regulation of the concerned biochemical pathways remain unanswered. The availability of mutants deficient in cuticular wax accumulation in a variety of species and the isolation of the corresponding genes reveal helpful information to elucidate cuticular wax biosynthesis and to characterize molecular aspects of regulatory control (Kunst and Samuels, 2003).

In maize (Zea mays), at least 18 loci (the GLOSSY or GL loci) have been found to affect the quantity and/or the composition of cuticular waxes on the surface of seedling leaves (Neuffer et al., 1997). From genetic and biochemical analyses of maize plants carrying different gl mutations, a preliminary model predicting two distinct pathways for cuticular wax biosynthesis has been set forth (Bianchi et al., 1985). One pathway would be responsible for wax synthesis in the first five or six juvenile leaves, whereas the other would produce waxes during the whole life cycle of the maize plant. The products of these two pathways can be distinguished by their chemical composition: Approximately 80% of the juvenile waxes are very-long-chain alcohols and aldehydes, whereas approximately 70% of the waxes produced throughout the life of a maize plant consist of esters (Bianchi et al., 1985). These ontogenetic differences in wax composition lead to different phenotypes of the maize leaves; juvenile leaves of wild-type maize plants have a glaucous surface appearance, whereas all leaves appearing later in plant development have a glossy surface. Mutations impairing the juvenile wax pathway also confer a glossy appearance to the juvenile leaves. Because of their phenotypic appearance, such mutants were designated glossy. The different composition of the wax layer of epidermal cells of juvenile and adult leaves is one of the traits that defines the juvenile-to-adult phase transition in wild-type maize plants (Lawson and Poethig, 1995, and refs. therein).

Over the past years, various maize GLOSSY genes involved in cuticular wax production have been cloned (Moose and Sisco, 1994; Tacke et al., 1995; Hansen et al., 1997; Xu et al., 1997). Based on evidence presented by Xu et al. (2002), GL8 encodes as a β-ketoacyl reductase of the fatty acid elongase complex involved in wax production. GL2 is apparently involved in acyl chain elongation from C30 to C32. However, comparisons of the predicted GL2 sequence with those in protein databases revealed no similarities to any known fatty acyl synthases, as may be expected for a component of the acyl elongation pathway (Tacke et al., 1995). The GL15 locus is a developmental gene belonging to the APETALA2 family of regulatory genes involved in the transition from juvenile to adult leaves (Moose and Sisco, 1996). The glossy phenotype of gl15 mutants is secondary to the primary mutant defect supporting the precocious development of adult leaves.

Mutation at the GL1 locus causes dramatic alterations in the amount, composition, and crystallization patterns of juvenile cuticular waxes (Bianchi et al., 1985). The most conspicuous property of gl1 waxes, in addition to a large reduction of aldehydes and alcohols, concerns the predominant chain length of long-chain aldehydes and the corresponding free and esterified alcohols, which turn out to be reduced by two carbon atoms. In this context, it has been hypothesized that the GL1 locus is either required for an elongation step in cuticular wax biosynthesis or is affecting the supply of wax precursors. However, the specific role of GL1 that emerged from these chemical studies was neither precise nor definitive.

In an effort to characterize the Gl1 gene, we performed transposon-tagging experiments with the Enhancer/Suppressor mutation (En/Spm) element, which led to the tagging of the GL1 locus (Maddaloni et al., 1990). In this article, we describe the cloning and molecular characterization of this gene. The genomic and cDNA sequences we have isolated differ in most of the coding region from the putative GL1 gene and transcript previously identified by Hansen et al. (1997). The protein encoded by GL1 shows significant homology with the entire sequence of the WAX2 gene product of Arabidopsis (Arabidopsis thaliana) involved in both cutin development and cuticular wax production (Chen et al., 2003). Similarly, the gl1 mutant displays a reduction in cuticular wax deposition and an alteration in cuticle membrane and plant hair (trichome) morphology.

RESULTS

Isolation of Transposon-Tagged Alleles of the GL1 Locus

From the cross outlined in “Materials and Methods,” nine glossy seedlings with revertant nonglossy sectors were identified out of approximately 90,000 F₁ seedlings. The new alleles were designated gl1-m1 through gl1-m9. Genetic analyses of the new mutable alleles are described by Maddaloni et al. (1990). Seven alleles, gl1-m1, 2, 3, 5, 7, 8, and 9, are due to the insertion of an element behaving autonomously but, contrary to expectation, are different from Activator (Ac). The alleles gl1-m4 and -m6 carried a nonautonomous element.

Plants carrying the gl1-m5 allele were tested for the functional presence in their genome of the transposable element En/Spm. Plants heterozygous for the gl1-m5 allele and for the stable recessive reference allele gl1-ref were crossed with the En/Spm tester strains described in “Materials and Methods.” Three independent test crosses were performed. F₁ plants of each test cross were selfed, and the resulting ears were scored for F₂ kernels displaying reversions of the a1-m1 or a1-m(r) tester allele. All three groups of F₁ plants gave rise to segregating and nonsegregating ears, with a percentage of the former near to 75%, suggesting the presence, in the gl1-m5 progenitor, of two unlinked copies of an active En/Spm element. F₂ kernels of both groups of ears were grown to seedlings, and these were scored for leaf variegation due to the gl1-m5 allele. The results of these experiments are summarized in Table I. Approximately one-half of the ears that segregated for variegated kernels also segregated gl1-m5 variegated seedlings, while ears without variegated seeds did not originate variegated seedlings. This result indicates that the gl1-m5 allele is likely to be caused by the insertion of an autonomous active En/Spm element.

Table I.

Genetic test for the presence of En/Spm in the gl1-m5 strain

	Segregation
Cross	F2 Ears with Variegated Kernels		F2 Ears without Variegated Kernels
	gl1-m5	gl1-ref	gl1-m5	gl1-ref
1	4	3	0	3
2	5	4	0	4
3	5	6	0	4
Total	14	13	0	11

Phenotypic Analysis of gl1 Mutants

Morphology of epicuticular waxes of wild-type and gl1 plants was previously examined with scanning electron microscopy (SEM) by Lorenzoni and Salamini (1975). This study showed that epidermal cells of gl1 seedling leaves are almost waxless, except for stomata subsidiary cells and cells of the leaf borders. Wax extrusion on gl1 leaves is reduced in size and has a round shape in contrast to the crystalline microstructure of wild-type epicuticular wax.

The wax phenotypes of a GL1 wild-type allele, of the recessive gl1-ref allele and of the unstable gl1-m5 allele are shown in Figure 1, A to C, as revealed by visual and microscopic inspections. The gl1-m5 allele shows clear somatic instability that is visible as sectors of wild-type tissue in a mutant background (Fig. 1D). The revertant sectors can cover small or large parts of a leaf (up to one-half) or are restricted to single epidermal cells. This finding shows that GL1, like other GL genes, acts cell autonomously during juvenile leaf development (Moose and Sisco, 1994; Tacke et al., 1995).

Figure 1.

The leaf surface in wild-type and gl1 alleles. Phenotype of a wild-type allele at the GL1 locus (A) compared to a stable recessive (B) and a mutable (C) allele, as seen when under water. D, Phenotype seen on the abaxial surface of the second leaf viewed with SEM: The reverted cells appear white in a dark background. SEM analysis of wild-type (E) and mutant (F) leaf surfaces. Insets, Close-up view of upper surface showing details of trichome morphology and density (all images are at 100× enlargement). TEM analysis of wild-type (G) and mutant (H) cuticle membranes is shown. CU, Cuticle; CW, cell wall.

Epidermal cells of maize leaves are arranged in cell rows that extend longitudinally parallel with the veins and show a gradient of cell differentiation from the leaf base to the leaf tip. Selected cell rows are enriched with stomata, others with trichomes, while still others are devoid of both types of specialized epidermal cells. gl1 mutant trichomes are smaller and more closely spaced in comparison to the wild type. In the apical region of fully developed juvenile leaves, their size is about one-half that of the wild type; on the leaf margin, mean distance between mutant and wild-type trichomes is 124 ± 14 μm and 171 ± 28 μm, respectively (Fig. 1, E and F). Stomata distribution on gl1 leaves does not differ from the wild type.

Ultrastuctural analysis of the leaf cuticle with transmission electron microscopy (TEM) indicates that the wild-type cuticle membrane appears to be divided into an outermost translucent layer (the cuticle proper) and an innermost opaque layer (the reticulated cuticular layer; Fig. 1G). In the gl1 mutant, cuticle membrane thickness is clearly reduced by about 50% and the cuticle proper appears almost absent (Fig. 1H). No differences in permeability to chlorophyll were detected in gl1 leaves compared to the wild type, as determined with extraction in 80% ethanol, with or without prior removal of cuticular waxes (data not shown). gl1 plants do not show reduction in pollen fertility, probably because waxes affected by the mutation are those found on juvenile leaves.

Isolation and Characterization of the gl1-m5 Allele

The internal EcoRI/BamHI fragment of the En/Spm element was used as a hybridization probe in DNA gel-blot analyses of families segregating for gl1-m5. An 8.3-kb HindIII fragment that cosegregated with the gl1-m5 mutant phenotype was identified. Representative homozygous (lanes 4 and 5) and heterozygous (lanes 6 and 7) gl1-m5 plants with the 8.3-kb En/Spm hybridizing fragment are shown in Figure 2A. Evidence that the HindIII fragment represents an En/Spm insertion in the GL1 gene came from the absence of this fragment in plants homozygous for germinal reverted alleles derived from gl1-m5 (Fig. 2A, lanes 2 and 3). No other En/Spm hybridizing fragments from gl1-m5 plants were consistently found to be missing from these derivatives.

Figure 2.

Southern analysis of the gl1-m5 locus and other mutable alleles. Maize DNA cleaved with HindIII was subjected to Southern hybridization. The EcoRI/BamI fragment spanning the region between positions 2,518 to 4,459 of the En/Spm element and the 0.95-kb HindIII-XhoI gl1-m5-derived fragments were used as a probe in A and B, respectively. Alleles are designed as follows: (1) wild-type allele from the inbred line WF9; (2) GL1-Rev2 and (3) GL1-Rev3; revertant alleles obtained from gl1-m5; (4 and 5) homozygous gl1-m5 plants; (6 and 7) plants heterozygous for gl1-m5 and gl1-ref showing a gl1 mutable phenotype; and (8) homozygous gl1-ref plants. A, The 8.3-kb band characteristic of the gl1-m5 allele is absent in revertant derivatives of gl1-m5. B, The blot in A was stripped and rehybridized with an HindIII-XhoI probe derived from gl1-m5. The 8.3-kb HindIII fragment from the gl1-m5 allele and the 6.0-kb fragment from the GL1-Wf9 and either gl1-m5 revertant or the gl1-ref alleles are indicated. C, Allelic cross-referencing experiments. DNA isolated from plants that carried the seven unstable mutant alleles gl1-m1 (1), -m2 (2), -m3 (3), -m5 (4), -m7 (5), -m8 (6), and -m9 (7) was digested with HindIII, electrophoresed on agarose gel, and transferred to a nylon membrane. Hybridization with the 0.95-kb HindIII-XhoI fragment of λ-09 revealed polymorphism associated with the gl1-m5 allele relative to other gl1 En/Spm mutable alleles. Molecular sizes (kb) are indicated.

A size-fractionated subgenomic library of HindIII fragments from heterozygous gl1-m5 plants was constructed in the λEMBL3 vector, and the 8.3-kb HindIII fragment was isolated in the clone λ-09 using the EcoRI/BamHI probe of the En/Spm element. The restriction map of the cloned HindIII fragment indicated the presence of an En/Spm element flanked by non-En/Spm sequences. A restriction fragment (0.95-kb HindIII-XhoI) carried by the non-En/Spm sequence was used as a hybridization probe to the same DNA gel blot shown in Figure 2A. The resulting hybridization pattern is shown in Figure 2B. Homozygous gl1-m5 plants, which exhibit an unstable phenotype in the presence of the autonomous En/Spm, showed the expected 8.3-kb HindIII fragment and a low-intensity 6.0-kb fragment (lanes 4 and 5). The 6.0-kb fragment was correlated with the generation of somatic revertant sectors from gl1-m5 and thus represented the original progenitor allele. Proof that the 0.95-kb HindIII-XhoI fragment represents part of the GL1 gene came from comparing the two independent germinal revertant derivatives of gl1-m5 (lanes 2 and 3) with their mutable siblings (lanes 4–7). The homozygous revertant plants contained only a 6.0-kb HindIII hybridizing fragment, whereas their mutable siblings heterozygous for the gl1-m5 and gl1-ref alleles, contained the 8.3- and 6.0-kb fragments (lanes 6 and 7). The size difference between the restriction fragment representing gl1-m5 and its somatic and germinal revertant derivatives was consistent with the En/Spm insertion observed within the cloned 8.3-kb HindIII fragment (Fig. 2B).

Although these results indicated that λ-09 contains a DNA fragment cosegregating with the gl1-m5 allele, they did not unambiguously establish that this clone was derived from the GL1 locus. To establish whether the DNA fragment in clone λ-09 indeed represented the GL1 locus, the 0.95-kb HindIII-XhoI fragment of this clone was used as a hybridization probe in allelic cross-reference experiments. The rationale for these experiments was that if this probe derives from the GL1 locus, then it should detect RFLPs between unstable gl1 mutants and their respective wild-type alleles. Southern-blot analyses were performed on DNA from seven independent gl1 mutable alleles (each of which carried an independently derived gl1 En/Spm allele). Hybridization with the 0.95-kb HindIII-XhoI fragment from λ-09 revealed a fragment of different size according to the specific allelic state of the GL1 locus (Fig. 2C). All indications obtained from the Southern experiments strongly suggested that the probe used for hybridization was able to recognize allelic states modifying, at the molecular level, the GL1 locus. It was concluded that the 0.95-kb HindIII-XhoI DNA sequences isolated from clone λ-09 mark a specific tract of the maize genome that corresponds to the GL1 locus.

Sequence Analysis of the GL1 Locus

The HindIII-XhoI fragment derived from clone λ-09 was used as a molecular probe in hybridization experiments on a maize bacterial artificial chromosome (BAC) library derived from inbred line F₂. Twenty-three independent clones were identified, eight of which were subsequently analyzed by restriction mapping. Two adjacent HindIII fragments, one of which hybridized with the molecular probe derived from clone λ-09, could be identified in all clones considered and were used to obtain the nucleotide sequence of the entire GL1 locus together with a 2.1-kb promoter fragment.

Computer-aided analysis of the genomic sequence obtained identified the putative exons encompassing the GL1 transcript. A database search for proteins homologous to the deduced GL1 polypeptide bolstered the postulated mRNA sequence. On the basis of these data, two primers were designed to isolate the full-length coding sequence of GL1 by reverse transcription (RT)-PCR. A single 2,056-bp fragment, including a 1,866-nucleotide-long open reading frame (ORF), was amplified from RNA extracted from wild-type seedling leaves of the inbred Wf9. From a partial cDNA clone isolated from a seedling cDNA library, we deduced that the GL1 transcript contains a 240-bp-long untranslated region (UTR) at its 3′ end (data not shown). The 5′ UTR was previously found to be 185 bp long. Taken together, these data suggest an approximate length of 2,291 bp for the GL1 transcript. An in-frame stop codon was present 87 bp upstream of the ATG start codon of the main ORF. No alternative translation start sites were present, indicating that the amplified fragment included the complete coding region. Putative CAAT- and TATA-box motifs were found in the promoter sequence 200 and 146 bp upstream of the ATG start codon, respectively, while a putative polyadenylation site was present 312 bp downstream of the translation stop codon.

Alignment of cDNA and genomic sequences revealed the presence of eight noncontiguous stretches of homology, interspersed with seven intron sequences ranging in size from 95 to 2,025 bp (Fig. 3). The deduced coding sequence displays three single base differences with respect to the genomic sequence considered. This discrepancy is due to polymorphism between the two strains used for the isolation of genomic and cDNA clones, as confirmed by sequence analysis of the corresponding genomic regions of the Wf9 inbred line (data not shown).

Figure 3.

Molecular map of the GL1 locus. Exon sequences (top line) together with interspersing intron sequences (middle and bottom lines) are reported. The approximate position of insertion sites of En/Spm transposable element sequences present in gl1 mutable alleles (m1–m9) is indicated with arrowheads indicating transposable element orientation. ATG start and TGA stop codons, as well as restriction sites of the ApaI (A), BamHI (B), HindIII (H), KpnI (K), SalI (Sa), SpeI (Sp), and SstI (Ss) endonucleases are reported. DNA segments used as molecular probes in Southern analyses are indicated and marked I, III, and IV. Probe II, consisting of an SstI (intron 3)-SstI (intron 4) fragment, is not indicated.

Mapping of En/Spm Insertion Sites

The available unstable gl1 alleles were examined by Southern analysis using different fragments of the GL1 gene as probes. All unstable alleles were found to be caused by independent insertions of members of the En/Spm transposable element family. The approximate insertion site and the orientation of the elements with respect to the GL1 gene were determined for all seven unstable alleles. In four cases (gl1-m1, 2, 5, and 8), the transposable element was inserted distal to the HindIII restriction site present in the fourth intron of the GL1 locus, which has been used previously to delimit the end of the GL1 gene (Fig. 3; Hansen et al., 1997).

The identification of the transposable elements present at the different unstable gl1 alleles and the location of the insertion point in the known sequence of the GL1 gene made possible a PCR-mediated amplification of specific fragments spanning the 5′ and 3′ junctions between the gl1 and terminal transposable element sequences. GL1- and En/Spm-specific primers were employed in PCR amplification reactions (see Supplemental Tables II and III) generating amplified fragments of the expected lengths. All amplification products were sequenced to determine the precise insertion point of the transposable element present in each of the unstable gl1 alleles (Fig. 3). The seven En/Spm insertions are all placed within the genomic region encompassed by the cDNA sequence. In all cases, the characteristic target site duplication of three nucleotides for En/Spm was observed (data not shown).

Characterization of the Predicted GL1 Protein

Conceptual translation of the 1,866-nucleotide-long ORF present in the GL1 cDNA sequence gave rise to a putative polypeptide of 621 amino acids with an apparent molecular mass of 69.6 kD and a pI of 9.89. Hydropathy analysis predicted the presence of several transmembrane domains in the N-terminal region of the GL1 polypeptide, as well as of a hydrophilic C-terminal domain. Furthermore, a tripartite His-rich motif characteristic of a family of membrane-bound desaturases/hydroxylases was present in the N-terminal part.

Compared to our cDNA, the sequence previously identified as the GL1 transcript by Hansen et al. (1997) is shorter and differentially spliced. In particular, it includes the first four exons of our cDNA and two stretches of intervening sequences: four bases of the third intron and 533 bases of the fourth intron. The former base insertion changes the reading frame and leads to the suppression of the third His-rich motif in the putative polypeptide (Fig. 4).

Figure 4.

Alignment of GL1 homologous sequences. The GL1 protein sequence (top line) was aligned against the polypeptides encoded by Hansen's cDNA (GL1*), Arabidopsis At5g57800 (WAX2), Arabidopsis At1g02205 (CER1), rice AK060786.1 (EST-Os), and S. odorus L33792 (EST-So). Identical residues are boxed in black and similar residues in gray. The conserved desaturase/hydroxylase domain is marked above the GL1 sequence with a gray line. Darker gray segments identify conserved His residues within the desaturase/hydroxylase domain.

A database search for proteins homologous to GL1 with the TBLASTX algorithm revealed several sequences exhibiting high levels of similarity with the query sequence used. In particular, a putative polypeptide of 619 amino acids encoded by a cDNA from rice (Oryza sativa; AK060786) showed 84% identity over its entire coding sequence. Furthermore, significant homologies, with a 67% identity score, were found with the products of two other rice cDNAs (AK066569 and AK070469), with the WAX2 locus of Arabidopsis encoding a protein involved in cuticle synthesis (62% identity), and a partial polypeptide (L33792) derived from Senecio odorus (55% identity). The alignment of the deduced GL1 amino acid sequence and deduced protein sequences exhibiting high similarity scores is depicted in Figure 4. The highest degree of homology consistently regards the C-terminal part of the deduced proteins.

A comparison of the deduced GL1 protein sequence and the product of the Arabidopsis ECERIFERUM1 (CER1) locus, a putative aldehyde decarbonylase active in the cuticular wax biosynthesis pathway, reveals an overall identity of 35%. This similarity score was significantly lower than the degree of similarity encountered between the putative GL1 and Arabidopsis WAX2 proteins (62%). Since previous results attributed to the maize GL1 locus a role as an Arabidopsis CER1 ortholog, we investigated amino acid sequence similarities among a restricted group of GL1 homolog sequences by means of phylogenetic analysis (Fig. 5). These analyses suggested the presence of two groups of protein sequences, the former containing the CER1 protein as a founder sequence, the latter including the WAX2 sequence. Interestingly, the GL1 sequence showed a high level of homology with the members of the WAX2 group, while a second maize sequence (GenBank AY104752) was located within the CER1 group with which it shares 55% amino acid identity. Thus, phylogenetic analysis indicated that Gl1-related sequences can be divided into two subgroups, each comprising genes from at least three species: maize, rice, and Arabidopsis (Fig. 5).

Figure 5.

Phylogenesis of GL1-like sequences. The deduced amino acid sequence of the GL1 locus was aligned against 12 deduced sequences obtained from homologous loci, as available in GenBank. Neighbor-joining analysis was used to obtain a phylogenetic tree, which was bootstrapped over 1,000 cycles. Significance values above a 50% cutoff threshold are indicated near the relative branches. Two distinct groupings are boxed: CER1-related sequences (I), including rice AK066386, maize AY104752, rice AK100751, rice AK068166, Arabidopsis At1g02205 (CER1), Arabidopsis At2g37700, and Arabidopsis At1g02190; and WAX2-related sequences (II), including Arabidopsis At5g57800 (WAX2), S. odorus L33792, maize AY505017 (Gl1, this study), rice AK060786, rice AK066569, and rice AK070469. Scale bar, Similarity coefficient.

GL1 Transcription Analysis

The 3′ end of the GL1 cDNA was used as a probe in northern-blot experiments performed with total RNA extracted from different tissues of wild-type plants and from leaf tissue homozygous for the gl1-ref allele. As shown in Figure 6A, the RNA extracted from wild-type seedlings showed a transcript with an estimated size of 2,300 residues, in accordance with the expected length of the GL1 mRNA (lane 1). The accumulation of this RNA was dramatically reduced in the gl1-ref mutant (lane 2) and was completely blocked in the root where, instead, a transcript of greater size was detected (lane 3). GL1 expression was evident also in adult leaves (lane 4) and in floral organs (silks and anthers; lanes 5 and 6, respectively), suggesting that GL1 activity was not restricted to the juvenile developmental phase of the maize plant. The same pattern of hybridization was observed using the complete Gl1 cDNA as a probe (data not shown). To check the amount of RNA loading, the filters were stripped and reprobed with a maize cytosolic GAPDH clone (Fig. 6B). The GL1 transcript was further studied by RT-PCR analysis using the samples described above (Fig. 6C). The use of forward and reverse GL1 primers allowed the amplification of a fragment of the expected size from RNA samples obtained from wild-type seedlings (lane 2) and, at low abundance, from gl1-ref mutant leaf (lane 3), mature leaf (lane 5), and anther tissue (lane 7). PCR amplification with primers against cytosolic GAPDH was used to verify the integrity of the samples (Fig. 6D).

Figure 6.

Northern analysis and RT-PCR analysis of GL1 transcription. A, mRNA extracted from young wild-type leaf (1), young gl1-Ref leaf (2), root (3), old wild-type leaf (4), silk (5), and anther (6) tissue was hybridized with the 3′ end of the GL1 cDNA. B, The same blot in A was stripped and rehybridized with a GAPDH probe. C and D, The presence of GL1 (C) and GAPDH (D) transcripts was assayed by RT-PCR in young wild-type leaf (2), young gl1-ref leaf (3), root (4), old wild-type leaf (5), silk (6), and anther (7) tissue. Size markers present in lane 1 were used to deduce the sizes of amplification products as indicated on the right.

As can be seen in Figure 6C, from the root (lane 3) and silk (lane 5) extracts no RNA amplification was obtained by RT-PCR. In this respect, we identified an incomplete Gl1-related clone by screening a silk cDNA library using Gl1 as a probe (H. Hartings, R. Velasco, and M. Motto, unpublished data). This silk cDNA shows 78% identity with Gl1; northern experiments performed with the same samples as those in Figure 6A give a similar hybridization pattern but with a higher intensity in the silk extract (see Supplemental Fig. 7). This was taken as evidence that a Gl1-related gene is expressed mainly in the silk tissue and gives rise to an mRNA cross-hybridizing to the Gl1 probe.

As concerns the band in lane 3 of Figure 6A, this might be either a root-specific transcript related in sequence to Gl1 or an unspliced version of Gl1 not amplified by RT-PCR with the conditions used. However, using different combinations of Gl1-specific primers aimed at identifying the presence of intron sequences in the Gl1 transcript, we had no indication of the occurrence of an unspliced version of the Gl1 mRNA in the root extract (data not shown). Accordingly, this band is likely to be the result of unspecific cross-hybridization.

DISCUSSION

Molecular Cloning and Characterization of the GL1 Locus

To obtain molecular insights into the nature of the genetic lesion that gives rise to the gl1 phenotype, a collection of unstable gl1 mutations induced by autonomous elements of the En/Spm family was generated (Maddaloni et al., 1990). From one of these mutable alleles, a partial sequence of Gl1 was identified and molecularly cloned, as confirmed by allelic cross-referencing and northern-blot experiments, and used to recover the complete gene from a maize genomic library. GL1 is a single-copy gene, which gives rise to a transcript carrying a 1,866-bp ORF and spanning eight noncontiguous genomic stretches. The genomic region encompassed by the cDNA sequence includes all the En insertion sites found in the unstable gl1 alleles.

Together, our results indicate that we have cloned the GL1 genomic sequence and the complete coding region of its major transcript. Hansen et al. (1997), who performed similar experiments to characterize the GL1 gene, isolated a genomic and a cDNA clone that, according to our results, represent a part of the GL1 locus and a differently spliced/partial unprocessed mRNA, respectively. Our conclusions are based on several lines of evidence. First, an entire collection of seven independent En/Spm insertions was analyzed to define the position of the transposable element in the unstable gl1 alleles. In four cases, the En/Spm insertion sequence is located downstream of the coding region of the GL1 gene as described by Hansen et al. (1997). Second, the cDNA isolated by Hansen et al. is 1,585 bp long, includes only the first four exons of our cDNA, and codes for a polypeptide of 319 amino acids with homology to CER1 and related proteins limited to the N-terminal region. The C-terminal domain has no counterparts in any of the GL1 homologs identified so far. Moreover, this polypeptide includes only the first two His-rich motifs. Third, Hansen's clone exactly matches the 5′ region of our genomic sequence. Because GL1 is a single-copy gene, it is concluded that both sequences derive from the same locus. However, the former was isolated by screening a genomic library constructed with HindIII-digested DNA. It is likely that the availability of a short GL1 transcript and the concomitant molecular situation at the GL1 locus presenting a HindIII site in the fourth, relatively lengthy, intron have suggested (Hansen et al., 1997) that the HindIII site was located beyond the end of the GL1 gene. In conclusion, our data regarding the GL1 transcript, its coding region, and the distribution of En/Spm element insertion sites within the locus suggest that the polypeptide described by Hansen et al. is not sufficient to perform the function of the GL1 gene.

The GL1 Gene Encodes a WAX2-Related Protein Displaying Transmembrane Domains

The putative protein encoded by GL1 is 621 amino acids long and is related in length and sequence to those coded by a number of loci from different plant species. These polypeptide sequences display several predicted transmembrane domains in the N-terminal region and a globular domain in the C-terminal part. A common feature shared by these proteins is the presence of eight conserved His motifs in the tripartite domain H-X_2–4-H, H-X_2–3-H-H, (H/Q)-X_2–3-H-H, 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, and refs. therein). The GL1-related proteins identified by means of sequence comparisons include the WAX2 and CER1 gene products of Arabidopsis, both of which are involved in cuticular wax production. Interestingly, in the GL1, WAX2, and CER1 genes, intron positions are conserved (see Supplemental Fig. 8), suggesting a common origin from the same ancestor gene. However, GL1 shows 62% overall amino acid identity to WAX2, which rises to 78% when the C-terminal region is considered, compared to 35% homology with CER1. Therefore, GL1 is probably the WAX2 ortholog from maize. In addition, sequence comparison clearly indicates the presence of two distinct subgroups of GL1-related genes in rice, maize, and Arabidopsis. The former, including GL1 itself, is related to the WAX2 gene of Arabidopsis, while the latter comprises sequences more similar to CER1, among which is a maize expressed sequence tag (EST; ID AY104752; see Fig. 5), which could be the true CER1 ortholog of maize.

Mutations of GL1, CER1, and WAX2 cause dramatic alterations in composition and crystallization patterns of cuticular waxes (Lorenzoni and Salamini, 1975; Bianchi et al., 1985; Jenks et al., 1995; Chen et al., 2003). Cuticular wax of juvenile maize leaves is composed of primary alcohols (63%), aldehydes (20%), and esters (16%) mainly derived from C32 acyl moieties, whereas on Arabidopsis stems the main constituents are alkanes (53%) and ketones (23%), with primary alcohols and aldehydes reduced to 11% and 4%, respectively. Arabidopsis leaves are almost devoid of ketones, and alkanes represent 72% of total waxes. Because of these differences in wax composition, comparing the biochemical effect of wax gene mutations in maize and Arabidopsis is not conclusive in defining the homology in gene functions.

CER1 was suggested to be an aldehyde decarbonylase because the mutant shows an increase in aldehydes and a reduction of the products of aldehyde decarbonylation, namely, alkane, secondary alcohols, and ketones (Aarts et al., 1995). In maize, these latter compounds represent only 1% of total cuticular waxes; therefore, mutations in a putative aldehyde decarbonylase are not likely to be identified based only on visual screenings.

In wax2 mutants, total wax load is diminished by about 80% because of the reduced accumulation of all the prevalent wax constituents, including aldehydes, with the exception of C30 primary alcohols, which are increased on wax2 stems.

Wax load on gl1 juvenile leaves is reduced by 73% compared to the wild type due to a decreased accumulation of both aldehydes and primary alcohols, while the amount of esters does not change. The mutation has a pronounced effect on the synthesis of long-chain wax compounds (C32), whereas those with shorter acyl chains are less affected or even increased (Bianchi et al., 1977). The overall change in wax composition on gl1 seedlings is similar to that observed on wax2 mutant leaves. In conclusion, these biochemical data on composition of mutant waxes also support the conclusion that GL1 is more closely related to WAX2 than to CER1.

The gl1 Mutation Affects Cuticular Wax Accumulation and Other Cuticular Traits

The Arabidopsis WAX2/YRE gene described by Chen et al. (2003) and Kurata et al. (2003) has a broad role in cuticle biosynthesis. Its mutation alters both cutin morphology and wax production. In addition, this mutation affects other cuticular and plant traits, including trichome development, leaf transpiration, pollen fertility, and postgenital organ fusion during early organ development. Similarly to wax2-defective plants, gl1-ref mutants display a reduction in cuticular wax deposition and alterations in cuticle membrane structure and trichome development. However, postgenital organ fusion was not observed in the gl1-ref mutant. As a possible explanation for these discrepancies, the gl1-ref allele might condition a leaky mutation impairing only some of the GL1 functions. However, a null-transcript gl1 mutant shows normal development and the absence of organ fusion (data not shown). A second possible explanation is the different effect of the two mutations on cutin morphology, which could be of functional importance: In wax2 mutants, cuticle membrane thickness is increased, although its weight is reduced.

Alternatively, different roles on plant development may be ascribed to maize and Arabidopsis cuticles. In addition to wax2, the abnormal leaf shape1 (ale1) and lacerata (lcr) Arabidopsis mutants are altered in cuticle membrane morphology and display postgenital organ fusion (Tanaka et al., 2001; Wellesen et al., 2001). Instead, the long-chain acyl-CoA synthetase2 (lacs2) mutant shows a reduction in cutin thickness (−40%) and several phenotypic alterations normally associated with defective cutin structure but not organ fusion, suggesting that this trait is very likely linked to more drastic cutin impairments (Schnurr et al., 2004). The causal relationship between cuticle membrane defects and postgenital organ fusion in Arabidopsis is strengthened by the finding that degradation of the cutin layer in transgenic Arabidopsis plants expressing a cutinase gene leads to adhesion of different adult organs (Sieber et al., 2000).

A similar correlation is not observed in maize and other monocots. Maize mutants with adhesion-competent epidermal cells include crinkly4 (cr4) and adherent1 (ad1). The cr4 mutation has a broad effect on epidermal cell morphology, not restricted to the cuticular layer (Becraft et al., 1996). ad1 mutants exhibit alteration in cell wall structure and epicuticular wax deposition in the epidermal layer, while the cuticle membrane in adherent regions appears intact (Sinha and Lynch, 1998). By contrast, the sorghum bm-22 mutant with a severely reduced cuticle membrane displays normal plant structure and development (Jenks et al., 1994). Moreover, maize seedlings treated with an inhibitor of cutin synthesis have no visual phenotypic alterations (Lequeu et al., 2003). Collectively, these results point to a different role of maize and Arabidopsis cuticles in plant development, namely, in the prevention of postgenital organ fusion.

Expression of the GL1 Gene Is Tissue and Organ Specific and Developmentally Regulated

As far as GL1 function is concerned, analysis of the mutant phenotype indicates that GL1, like the other GLOSSY genes, is an essential component of the juvenile wax layer biosynthetic route (Bianchi et al., 1985). Young gl1 leaves have the chemical wax composition of wild-type adult leaves. It can be concluded that the regulation of GL1 expression is an integral part of the cell response to age-related events of differentiation. It was shown that in maize the juvenile-to-adult transition is under genetic control, with a series of independent mutations—CORN-GRASS1 (CG1), TEOPOD1 (TP1), TEOPOD2 (TP2), HAIRY-SHEATH-FRAYED1 (HSF1; Poethig, 1988, and refs. therein), and GL15 (Moose and Sisco, 1996)—altering the transition from juvenile to adult vegetative growth. The interaction of GL15 with TP1 and TP2 indicates that GL15 acts downstream of these genes and that it is required for their effect during epidermis development (Evans et al., 1994). The GL15 gene is a transcriptional activator of the APETALA2 family of regulatory genes (Moose and Sisco, 1996), which is expected to interact directly with the promoters of structural genes needed for the accumulation of cuticular waxes, a prediction that can be experimentally tested now that GL1, GL2, and GL8 are listed among the cloned genes contributing to wax deposition.

From the expression profile experiments, it can be argued that the regulation of the expression of the characterized GLOSSY genes is more complex than predicted only on the basis of seedling phenotypes. GL8 turns out to be expressed in different organs of the adult plant, including the roots, although to a lesser extent than in seedling leaves (Xu et al., 1997). GL1 is not expressed in the root but is detectable in anthers and, to a minor extent, also in adult leaves, similar to the GL2 transcript that, besides being characteristically produced in young leaves, is also detectable in the part of the maize shoot contributing to the development of the female inflorescence (Velasco et al., 2002). These results demonstrate that the GL1 protein, like the products of other GLOSSY genes, is not merely involved in the cuticle formation of the green part of the maize seedling. Together, these findings indicate that factors conditioning tissue juvenility may be reactivated after the formation of the lateral meristem (Uhrig et al., 1997). Alternatively, a different control mechanism affecting wax biosynthetic genes should exist in adult plant tissues.

MATERIALS AND METHODS

Plant Material

The origin and maintenance of the wx-m7 transposon stock, the gl1-ref allele used in this study, and the recovery of the gl1-mutable strains have been described previously (Maddaloni et al., 1990). Briefly, gl1 alleles were identified in the F₁ generation of a cross between a strain homozygous for the unstable allele wx-m7 of the WAXY locus and for a dominant GL1 allele (used as the male parent) and a female parent strain homozygous for the stable recessive gl1-ref allele. Both strains were in the Wf9 inbred line genetic background. The instability of the wx-m7 allele is due to the transposable element Ac (McClintock, 1962; Behrens et al., 1984). In our tagging experiments, nine unstable alleles were generated at the GL1 locus. Seven, gl1-m1, 2, 3, 5, 7, 8, and 9, were due to the insertion of an autonomous element, while gl1-m4 and gl1-m6 were caused by a nonautonomous element of a family of transposable elements different from Ac/Ds (Maddaloni et al., 1990).

The wx-m7 stock also contains copies of active autonomous En/Spm elements (Michel et al., 1995). To test for the presence in gl1-m5 in stable mutants of En/Spm elements, crosses with two tester strains were generated. The resulting F₁ plants were selfed and the F₂ seeds were checked for mutability of either of the two tester alleles, a1-m(r) or a1-m1, both due to the insertion of a defective En/Spm element (I/dSpm) and showing somatic instability only in the presence of an active En/Spm element. In its absence, the a1-m1 allele gives rise to kernels with a pale color, while a1-m1(r) is colorless. In the presence of an active En/Spm element, colored spots are produced on the pale (a1-m1) or colorless [a1m(r)] background. Crosses and selfings were according to standard procedures.

Genomic Cloning and Southern Analysis

Maize (Zea mays) DNA was isolated from leaves of flowering plants or from seedlings as described (Michel et al., 1995). Southern hybridizations and radioactive labeling of probes were according to standard procedures (Sambrook et al., 1989). A plasmid clone carrying the complete En/Spm element isolated from the wx-844 allele was provided by Dr. Alfons Gierl (Munich). An EcoRI/BamHI restriction fragment that covered the internal region of the En/Spm element between positions 2,518 to 4,459 (Pereira et al., 1986) was used as a molecular probe.

Southern analysis was performed to map the position of the En element in seven unstable gl1 alleles. For this purpose, genomic DNA from homozygous mutant plants was digested with the restriction enzymes BamHI, HindIII, ApaI, SstI, SpeI, and KpnI. Blots were probed with four PCR-derived fragments of the GL1 allele covering regions (base positions are indicated relative to the translation start site) from −2,302 to −648 (probe I), from +1,213 to +2,048 (probe II), from +2,426 to +3,581 (probe III), and from +4,131 to +4,745 (probe IV).

HindIII-digested genomic fragments from 7 to 11 kb from plants carrying the gl1-m5 allele were cloned into the λEMBL3 vector, after separation and purification from agarose gels with the QIAEX gel extraction kit (Qiagen, Valencia, CA). Nine clones hybridizing to a fragment covering the En/Spm sequence between the EcoRI sites at positions 5,836 and 8,278 were identified within approximately 100,000 recombinant phages. Restriction mapping of the clones identified a single clone, designated λ-09, carrying a HindIII insert of approximately 8.3 kb, which was chosen for further analysis.

Recombinant clones carrying a wild-type GL1 allele were recovered from a maize BAC library derived from DNA extracted from the inbred F₂, kindly provided by Dr. Keith Edwards (University of Bristol, UK).

Northern and RT-PCR Analysis

Total RNA was isolated from the following organs and tissues of the inbred Wf9: second and third leaf of wild-type and gl1-ref mutant seedlings, wild-type roots of 1-week-old seedlings, wild-type adult leaf (top leaf, surrounding the tassel), wild-type silks, and wild-type anthers (both immature and pollen-shedding, from the same tassel). Extractions were performed using TRizol (Invitrogen, Carlsbad, CA) according to the manufacturer's instructions. For northern experiments, 20 μg of total RNA samples were fractionated on denaturing gels, capillary blotted onto nylon membranes (Hybond N+; Amersham, Little Chalfont, UK), and hybridized at 45°C in Ultrahyb solution (Ambion, Austin, TX). Filters were washed at 55°C in 2× SSC/0.1% SDS (15 min), in 1× SSC/0.1%SDS (30 min), and in 0.1× SSC/0.1% SDS (30 min). RNA markers from 0.2 to 10 kb (Sigma-Aldrich, St. Louis) were used as size standards. Probes were the full-length GL1 cDNA and the 400-bp 3′ end, including the complete 3′ UTR, of a partial GL1 clone isolated from a seedling cDNA library. To check the amount of RNA loading, filters were rehybridized with a probe derived from maize cytosolic GAPDH cDNA.

For RT-PCR, 5 μg of total RNA were reverse transcribed using SuperScriptII reverse transcriptase (Invitrogen) according to manufacturer's instructions. One-twentieth of the final reaction product was amplified by PCR with the following primers: forward, 5′-ATCGAATTCACGTACGGCACAGTTGCTAGC-3′; reverse, 5′-CGCTCTAGACCACCAATTCACACTCGACG-3′.

The forward primer annealed to the region of the GL1 cDNA starting 70 bp upstream of the ATG start codon, while the reverse primer annealed to the region starting 101 bp downstream of the stop codon. To avoid formation of secondary structures, PCR reactions were performed in the presence of 10% DMSO (final concentration). The 5′ end of the forward and reverse primers included, respectively, EcoRI and XbaI restriction sites (indicated in italics in the above sequences), which were used to subclone the GL1 cDNA from leaves of the inbred Wf9 into the pBluescriptSKII vector (Stratagene, La Jolla, CA) prior to sequencing. Five independent clones were sequenced on both strands to determine the sequence of the GL1 cDNA.

DNA Sequencing

DNA sequencing was carried out with an automatic sequencer (CEQ 8000; Beckman-Coulter, Fullerton, CA). Genomic and cDNA sequences were determined on both strands.

Microscopic Inspection

SEM was used to study adaxial surfaces of primary leaves of gl1 mutant and wild-type plants grown for 2 to 3 weeks in a phytochamber at 26°C/19°C (day/night) and 40% humidity with a 16/8-h light/dark rhythm, and a light intensity of 1,900 μE m⁻² s⁻¹. Segments of the middle part of the leaf blade were fixed to a specimen holder by tissue tek and shock frozen with liquid nitrogen within a high vacuum cryo preparation stage. Samples were transferred under vacuum to a cryo preparation chamber where they were sputter coated with gold and examined on the cold stage of a Zeiss DSM 940 SEM (Carl Zeiss NTS GmbH, Oberkochen, Germany). For TEM investigation of the cuticle, small pieces (2–3 mm²) of primary leaf blades were fixed for 2 h at room temperature in 2.5% (v/v) glutaraldehyde and 2% (v/v) formaldehyde in 0.05 m phosphate buffer (PB), pH 6.8. After washing in PB, samples were postfixed for 1 h in 2% (v/v) osmium tetroxide in PB, washed again, and dehydrated through a graded series of ethanol. Samples were then infiltrated with LR White resin (Plano, Marburg, Germany) and polymerized for 48 h at 60°C. Ultrathin cross-sections were prepared and mounted on carbon-coated Formvar copper grids (200 mesh; Plano). After staining with 2% uranyl acetate for 2 h, sections were inspected with a Zeiss EM 10 TEM (Carl Zeiss NTS GmbH).

Statistical Analysis

Multiple DNA and protein sequence alignments were performed using ClustalW (Thompson et al., 1994), while phylogenetic analysis was performed according to the MEGA version 2.1 software package (Kumar et al., 2001). For tree construction based on aligned amino acid sequences, the neighbor-joining tree-building method was utilized. Bootstrap analysis (1,000 replicates) was used to assign a consensus tree at a 50% cutoff value.

Sequence data from this article have been deposited with the EMBL/GenBank data libraries under accession numbers AY505017 and AY505498.