A mutational analysis of the ABA1 gene of Arabidopsis thaliana highlights the involvement of ABA in vegetative development

Abstract

Much of the literature on the phytohormone abscisic acid (ABA) describes it as a mediator in triggering plant responses to environmental stimuli, as well as a growth inhibitor. ABA-deficient mutants, however, display a stunted phenotype even under well-watered conditions and high relative humidity, which suggests that growth promotion may also be one of the roles of endogenous ABA. Zeaxanthin epoxidase, the product of the ABA1 gene of Arabidopsis thaliana, catalyses the epoxidation of zeaxanthin to antheraxanthin and violaxanthin, generating the epoxycarotenoid precursor of the ABA biosynthetic pathway. This paper gives a description of the molecular and phenotypic characterization of a large series of mutant alleles of the ABA1 gene, which cause different degrees of ABA deficiency, four of them previously isolated (aba1-1, aba1-3, aba1-4, and aba1-6) and the remaining five novel (sañ1-1, sañ1-2, sañ1-3, sañ1-4, and sre3). Molecular analysis of these alleles provides insights into the domains in which they compromise zeaxanthin epoxidase function. The size of the leaves, inflorescences, and flowers of these mutants is reduced, and their rosettes have lower fresh and dry weights than their wild types, as a result of a diminished cell size. Low concentrations of exogenous ABA increase the fresh weight of mutant and wild-type plants, as well as the dry weight of the mutants. The leaves of aba1 mutants are abnormally shaped and fail to develop clearly distinct spongy and palisade mesophyll layers. Taken together, these phenotypic traits indicate, as suggested by previous authors, that ABA acts as a growth promoter during vegetative development. The abnormal shape and internal structure of the leaves of aba1 mutants suggests, in addition, a role for ABA in organogenesis.

ABA1, abscisic acid, Arabidopsis thaliana, growth promotion, vegetative development, zeaxanthin epoxidase

Introduction

The plant hormone abscisic acid (ABA) was first named as such (Addicott et al., 1968) a few years after its isolation from cotton fruits (Ohkuma et al., 1963) as an abscission-accelerating factor, and from sycamore leaves (Cornforth et al., 1965) during a search for endogenous substances that induce dormancy. Although the role of ABA in fruit abscission and dormancy of woody plants remains unclear (Schwartz et al., 2003), a vast body of evidence supports the implication of this ubiquitous hormone in essential plant processes such as seed maturation, desiccation, dormancy, and germination. In addition, ABA has long been known to be involved in the responsiveness of plants to various environmental stresses, particularly drought and salinity. ABA limits water loss through the regulation of stomatal aperture, and also affects the patterns of expression of more than 1000 genes (Skriver and Mundy, 1990; Hoth et al., 2002; Seki et al., 2002; Leonhardt et al., 2004).

Genetic approaches have helped to shed light on ABA biosynthesis and signal transduction (Finkelstein and Gibson, 2002). For instance, many mutants affected in the synthesis of this hormone have been isolated in Arabidopsis thaliana (reviewed in Schwartz et al., 2003), Nicotiana plumbaginifolia (Marin et al., 1996), Zea mays (Tan et al., 1997), Lycopersicon esculentum (Sagi et al., 1999), and Oryza sativa (Wurtzel et al., 2001). ABA-deficient mutants have been found in genetic screenings as diverse as those carried out for precocious germination (Tan et al., 1997), altered photochemical quenching (Niyogi et al., 1998), insensitivity to the inhibition of seedling growth caused by glucose (Laby et al., 2000; Rook et al., 2001), paclobutrazol (an inhibitor of gibberellin biosynthesis)-resistant germination (Koornneef et al., 1982; Leon-Kloosterziel et al., 1996; Nambara et al., 1998), and salt-resistant germination (Saleki et al., 1993; Quesada et al., 2000; González-Guzmán et al., 2002), among others.

In spite of the attention that ABA has attracted, there are many questions still to be answered on its biology. These include the putative existence of an alternative biosynthetic pathway (Taylor et al., 2000) and the identification of the key enzymes contributing to the regulation of ABA levels (Tan et al., 1997; Frey et al., 1999; Xiong and Zhu, 2003; JM Barrero, P Piqueras, PL Rodríguez, MR Ponce, JL Micol, unpublished results). Another controversial subject concerns the fact that a substantial part of the literature on ABA focuses on its role as a growth inhibitor under stress situations. These adverse conditions cause a diminished plant growth that correlates with an increase in endogenous ABA levels (Zeevaart and Creelman, 1988; Koornneef and Karssen, 1994; Rock and Quatrano, 1995; Leung and Giraudat, 1998; Qin and Zeevaart, 1999; Larkindale and Knight, 2002). In fact, ABA has been considered by many as an inhibitor of shoot growth (Bradford, 1983; Trewavas and Jones, 1991; Munns and Cramer, 1996). Paradoxically, ABA has been described as a growth promoter rather than an inhibitor in different tissues of several plant species when exposed to salinity, cold, heat, drought, and soil compaction (Saab et al., 1990; Sharp et al., 1994; Mulholland et al., 1996a, b, 2003; Hussain et al., 1999; Cramer, 2002; Sharp, 2002; Makela et al., 2003). In fact, ABA-deficient mutants are often smaller than their corresponding wild types (Koornneef et al., 1982; Quarrie, 1987). Some other observations suggest a role for ABA as a growth promoter under non-stress conditions (Bradford, 1983; Sharp et al., 2000; Chen et al., 2003; LeNoble et al., 2004). Indeed, some authors have proposed that ABA can be a growth inhibitor under stress conditions but a growth promoter in the absence of stress (Cheng et al., 2002; Sharp and LeNoble, 2002).

Previously, Arabidopsis thaliana mutants able to germinate on 250 mM NaCl agar medium were looked for. Such mutants fell into five complementation groups, which were named SALOBREÑO1 to SALOBREÑO5 (SAÑ1 to SAÑ5; Quesada et al., 2000). In another independent screening, four additional mutants, sre1 to sre4 (for salt resistant; González-Guzmán et al., 2002), which were able to germinate on 200 mM NaCl, were isolated. Germination assays in paclobutrazol have shown that the sañ mutants are affected in their synthesis of ABA, the only exception being sañ5, which was revealed to be a null allele of the ABI4 gene (Quesada et al., 2000). In the present work, the genetic and molecular analysis of nine mutants whose ABA biosynthesis is affected, including four sañ1 allelic mutants and the sre3 mutant, all of which carry alleles of the ABA1 gene, which encodes the zeaxanthin epoxidase enzyme, is described.

Here, it is demonstrated that leaf, inflorescence, and flower size, root length, and fresh and dry weights of aba1 mutants are reduced, and that such reductions in size and mass are due to a diminished cell size. Low concentrations of exogenous ABA increased the fresh weight of mutant and wild-type plants, as well as the dry weight of the mutants. In addition, it is shown that not only size but also leaf shape and internal architecture are abnormal in the aba1 mutants. The phenotypes of the mutants studied here clearly indicate that ABA acts as a growth promoter during vegetative development. Moreover, the abnormal shape and internal structure of the leaves of aba1 mutants suggests a role for ABA in organogenesis.

Materials and methods

Plant material and growth conditions

Arabidopsis thaliana (L.) Heynh. sañ1 and sre3 mutants were isolated as described in Quesada et al. (2000) and González-Guzmán et al. (2002), respectively. The sañ1 mutants were isolated from M₂ seed populations derived from fast neutron bombardment (sañ1-2, sañ1-3, and sañ1-4) or T-DNA mutagenesis (sañ1-1), respectively supplied by Lehle Seeds (Round Rock, TX, USA) and the Nottingham Arabidopsis Stock Centre (NASC), UK. The sre3 mutant was isolated from a T₄ seed population derived from T-DNA lines constructed in the laboratory of D Weigel using the pSKI15 vector, which was provided by the Arabidopsis Biological Resource Center (ABRC), Ohio, USA. The ethyl methanesulphonate (EMS)-induced aba1-1, aba1-3, aba1-4 (Koornneef et al., 1982), and aba1-6 (Niyogi et al., 1998) alleles were provided by the NASC. The aba2-14 (sañ3-2) and aba3-101 (sañ4-1) mutations were induced by fast neutron bombardment in a Col-0 genetic background, and aba3-102 (sañ4-2) by T-DNA mutagenesis in a Ws-2 background (Quesada et al., 2000). The aba2-14 allele carries a large deletion of 951 bp affecting the ABA2 gene (González-Guzmán et al., 2002). Complementation tests showed the aba3-2 mutation to be allelic to aba3-101 and aba3-102, which carry reorganizations that, until now, had not been completely characterized at a molecular level. The aba3-2 mutant was provided by the NASC (accession number N158). All the mutations used in this work are recessive.

Plant cultures were performed, as described in Ponce et al. (1998), under constant fluorescent light (7000 lx) at 20 ±1 °C and 60–70% relative humidity. Non-sterile cultures were performed in pots containing a 2:2:1 mixture of perlite, vermiculite, and sphagnum moss. For sterile cultures, seeds were sown in a water suspension, using a Pasteur pipette, in 150 mm Petri dishes filled with 100 ml of agar culture medium, at a density of 50 regularly spaced seeds per plate. Once inoculated, the Petri dishes were sealed with Micropore Scotch 3M surgical tape, which prevented contamination, but allowed gaseous exchange. Growth was allowed to proceed in Conviron TC16 tissue culture chambers.

Germination assays

Salt-tolerant germination was tested by sowing mutant and wild-type seeds on agar medium supplemented, or not, with 250 mM NaCl, as described in Quesada et al. (2000). For paclobutrazol (Cultar 25% m/v; Zeneca Agrochemicals, UK) germination assays, seeds were sown on filter paper saturated with the appropriate concentration of this substance. Only those seedlings that displayed green and fully expanded cotyledons 2 weeks after sowing were considered as salt resistant.

Water status measurements

For relative water content (RWC) measurements, sterile cultures were performed as described above. Samples of 10 leaves, excised from the third and fourth node of five rosettes, were collected from 21-d-old plants and their fresh weight (FW) was immediately obtained. Leaves were then floated for 12 h on deionized water in a Conviron TC16 tissue culture chamber and their turgid weight (TW) determined. Finally, leaves were dried for 12 h at 70 °C and weighed (DW). The RWC was calculated as [(FW−DW)/(TW−DW)]×100 (Hewlett and Kramer, 1962). Each measurement was made in triplicate.

ABA treatments

Fifty seeds of each genotype were sown on Petri dishes containing agar medium supplemented, or not, with 10 nM or 50 nM ABA (Sigma-Aldrich A1049). Ten rosettes of each genotype were collected 20 d after sowing and immediately weighed to determine fresh weight, or weighed after 24 h at 50 °C for the dry weight. To determine cell sizes, plants were grown on non-supplemented or 50 nM ABA medium and their third node leaves collected and treated with ethanol and chloral hydrate, as previously described in Candela et al. (1999). Cleared leaves were mounted on slides, and interference contrast images were taken using a Leica DMRB microscope equipped with a Nikon DXM1200 digital camera. Whole rosette and leaf pictures were taken in a Leica MZ6 stereomicroscope.

Genetic analysis

For complementation tests, homozygous mutants were intercrossed and the resulting F₁ seeds were sown on media supplemented with NaCl or paclobutrazol (Quesada et al., 2000). Linkage analysis was performed on mapping populations of homozygous F₂ plants as described in Ponce et al. (1999).

Sequencing

Synthetic oligonucleotides were designed, bought from Perkin-Elmer Applied Biosystems UK and used as primers in PCR amplifications of overlapping segments of the alleles of the At5g67030 gene, whose wild-type sequence was already available. Genomic DNA was extracted and PCR amplified, and the amplification products were sequenced using ABI PRISM BigDye Terminator Cycle Sequencing Ready Reaction kits and an ABI PRISM 3100 Genetic Analyser as described in Pérez-Pérez et al. (2004).

Complementation of the aba1-101 allele

The At5g67030 gene is contained within the K8A10 TAC clone. A BamHI 8.5 kb restriction fragment encompassing the At5g67030 gene was subcloned from the TAC clone into the pBIN19 vector. The resulting construct, which was named pBIN19-At5g67030, was transferred by electroporation to the disarmed Agrobacterium tumefaciens C58C1 strain, harbouring the pGV2260 helper plasmid (Deblaere et al., 1985). Arabidopsis thaliana aba1-101 homozygous plants were transformed by the floral dip method (Clough and Bent, 1998). Seeds were harvested and plated on kanamycin selection medium to identify transgenic plants. Complementation of the ABA-deficient phenotype in aba1-101/aba1-101 control plants and T₂aba1-101/aba1-101 transgenic plants, containing the pBIN19-At5g67030 construct, was assayed by scoring germination in agar medium supplemented either with 150 mM NaCl or 1 μM paclobutrazol (Duchefa P0922).

Results

Map-based cloning of the SAÑ1 gene

In a previous work (Quesada et al., 2000), four mutant lines of Arabidopsis thaliana displaying salt-tolerant germination, which were found to be allelic and named sañ1-1 to sañ1-4, were isolated. These mutations mapped at chromosome 5, in a genomic interval flanked by the simple sequence length polymorphism (SSLP) (Bell and Ecker, 1994) MBK5 and the telomere. Five novel SSLP markers were developed within this interval (Fig. 1A), by selecting regions rich in dinucleotide repeats, based on the available genomic sequences of BAC (bacterial artificial chromosome) clones (http://mips.gsf.de/proj/thal/), whose polymorphism was tested by determining their sequences in Ler and Col-0. These markers were used to screen for recombinants in an F₂ mapping population of 878 plants, derived from several crosses involving wild-type and sañ1 mutant plants. The informative recombinants found allowed a candidate interval of 250 kb, encompassing four TACs (transformation-competent artificial chromosomes), to be defined (Liu et al., 1999): MSN2, MUD21, K8A10, and K21H1. One of the candidate genes in the above-mentioned interval was ABA1 (At5g67030, within K8A10), the Arabidopsis thaliana orthologue of the Nicotiana plumbaginifolia ABA2 gene (Marin et al., 1996), which encodes the zeaxanthin epoxidase enzyme (Audran et al., 2001; Xiong et al., 2002). The aba1 mutants (Koornneef et al., 1982) are known to suffer continuous water loss, due to impaired stomatal closure, and to display reduced seed dormancy and salt-tolerant germination. Taken together, the genetic mapping data and the phenotypic traits shared by aba1 and sañ1 mutants provided the prompt to cross them, and the latter were found to be alleles of the ABA1 gene (Table 1). In an independent search for salt-tolerant mutants, sre3, which had a phenotype quite similar to that of the previously described aba1 mutants, was isolated. Additional complementation analyses showed that sre3 was also an allele of ABA1 (Table 1).

Fig. 1.

Molecular characterization of aba1 mutants. (A) Map-based strategy followed to determine that sañ1 mutations are alleles of the ABA1 gene. The SSLP markers used for linkage analysis and the number of informative recombinants identified (in parenthesis) are indicated. The lists of candidate BAC clones and genes were assembled from the information found in the Arabidopsis thaliana database (http://www.arabidopsis.org). (B) Structure of the ABA1 gene with the mutations studied in this work are indicated. Exons are indicated by boxes, and introns by lines between boxes. The segments of the gene encoding the domains identified by Marin et al. (1996) as being conserved among zeaxanthin epoxidases are indicated: an ADP-binding site (I), an unknown domain (II), and a flavin-adenine dinucleotide (FAD)-binding site (III). (C) Predicted amino acid sequence of the zeaxanthin epoxidase of Arabidopsis thaliana, with the effects of the mutations studied in this work indicated. Continuous, dashed, and dotted lines, respectively, indicate the N-terminal chloroplast transit peptide (as identified by the TargetP v1.0 program; Emanuelsson et al., 2000, http://www.cbs.dtu.dk/services/TargetP/), the monooxygenase domain, and the forkhead-associated (FHA) domain, all of which were described by Xiong et al. (2002), and agree with the present in silico analyses. Asterisks indicate stop codons. The 30 bp deletion carried by aba1-102 is indicated by an open box. Residues shaded in black and grey indicate the identities and similarities, respectively, found after the alignment of the sequence of the Arabidopsis thaliana zeaxanthin epoxidase (accession number BAB08942) with those of Nicotiana plumbaginifolia (Q40412), Capsicum annuum (Q96375), Lycopersicon esculentum (P93236), and Prunus armeniaca (O81360). The sequences were aligned using the program CLUSTAL-X 1.5b (Thompson et al., 1997) and shaded with BOX SHADE 3.21 (Hoffman and Baron, http://www.ch.embnet.org/software/BOX_form.html).

Table 1.

Complementation analyses of the sañ1 and sre3 mutants

Crosses	F1 seeds sown	F1 seeds germinated
		100 μM paclobutrazol	150 mM NaCl
Col-0	27	0	–
aba1-1/aba1-1	42	34	–
sañ1-3/sañ1-3×aba1-1/aba1-1	22	20	–
sre3/sre3×aba1-1/aba1-1	95	–	91
sre3/sre3×sañ1-2/sañ1-2	15	15	–
sre3/sre3×aba2-1/aba2-1	77	–	0
sre3/sre3×aba3-2/aba3-2	99	–	0

F₁ seeds were scored for germination 15 d after sowing.

Molecular characterization of aba1 alleles

The At5g67030 transcription unit in homozygous plants was sequenced for the sre3, sañ1-1, sañ1-2, sañ1-3, and sañ1-4 mutations, which were renamed, respectively, as aba1-101, aba1-102, aba1-103, aba1-104, and aba1-105. This allele numbering was chosen in order to avoid confusion with previously isolated aba1 alleles, of which at least two series already exist (Koornneef et al., 1982; and personal communication of A Marion-Poll to PLR). The ABA1 gene was also sequenced in aba1-1, aba1-3, aba1-4, and aba1-6 homozygous plants, which had been isolated by previous authors (Table 2; Koornneef et al., 1982; Niyogi et al., 1998).

Table 2.

Mutations identified in aba1 alleles

Allele	Genetic background	Mutagen	DNA change			Protein change
				Type of mutation	Affected regiona
sre3 (aba1-101)	Col-0	T-DNA	Deletion	ΔT396	Frameshift
sañ1-1 (aba1-102)	Ws-2	T-DNA	Deletion	Δ30 bp, 259–288	Shortened protein with 657 aa
sañ1-2 (aba1-103)	Col-0	Fast neutrons	Inversion	1.85 kb in the 3′ region	Shortened protein with 430–440 aa
sañ1-3 (aba1-104)	Col-0	Fast neutrons	Deletion	Δ0.65 kb in the 5′ region	No protein
sañ1-4 (aba1-105)	Ler	Fast neutrons	Deletion	Δ5 bp, 1712–1716	Frameshift
aba1-1	Ler	EMS	Transition	G2139→A	Trp489®stop
aba1-3	Ler	EMS	Transition	C374→T	Pro125→Ala
aba1-4	Ler	EMS	Transition	G1520→A	Trp363®stop
aba1-6	Col-0	EMS	Transition	G478→A	Gly160→Ser

a

Δ indicates a deletion. Numbering refers to genomic DNA, starting from the initiation codon.

The transcription unit of the ABA1 (At5g67030) gene includes 16 exons and 15 introns (Fig. 1B) and encodes a predicted protein of 667 amino acids (Audran et al., 2001; Xiong et al., 2002). The EMS-induced aba1-1 allele presents a G→A base change at position 2139 (starting from the initiation codon), introducing a stop codon in the transcript and truncating the protein, which is 488 amino acids long. This allele has been sequenced by two other research groups (Audran et al., 2001; Xiong et al., 2002). The EMS-induced aba1-3 allele suffered a C→T transition that changes proline to alanine in amino acid position 125. The EMS-induced aba1-4 allele carried a G→A transition, which introduces a stop codon after amino acid 362. The aba1-6 allele showed a change of glycine to serine in amino acid 160, since nucleotide 478 had suffered a G→A transition. The aba1-101 (sre3) mutant carried the deletion of a T in position 396, which caused a frameshift, generating a stop codon after amino acid 131. On the other hand, aba1-102 (sañ1-1), an untagged T-DNA-induced allele, presented a deletion of 30 bp that eliminates amino acids 87–96. Both aba1-103 (sañ1-2) and aba1-104 (sañ1-3), which were induced by fast neutron bombardment, carried rearrangements whose structural characterization required Southern analyses (data not shown). The aba1-103 mutation is a 1.85 kb inversion affecting the 3′ region of the ABA1 gene, which might still produce a peptide of 430–440 amino acids. In aba1-104, a 0.65 kb deletion was found affecting the 5′ region of the gene, which abolishes transcription, as was later shown by means of RT-PCR analyses (data not shown). In the EMS-induced aba1-105 (sañ1-4) allele, a deletion of 5 bp caused a frameshift that in turn introduces a stop codon after amino acid 399 (Table 2; Fig. 1B, C). Only two other alleles of the ABA1 gene, aba1-5 (Audran et al., 2001) and los6 (Xiong et al., 2002), had been sequenced previously.

In parallel with the molecular characterization of sañ1 and sre3 mutants, a transgene-mediated complementation experiment was performed. A genomic fragment containing the At5g67030 gene (see Materials and methods) was cloned in the pBIN19 vector, and the resulting construct was used to transform aba1-101/aba1-101 (sre3/sre3) plants, whose phenotypic rescue was obtained. Seven transformant lines were obtained, whose germination rate in 1 μM paclobutrazol (an inhibitor of gibberellin synthesis; Koornneef et al., 1982) or 150 mM NaCl media was lower than 40%, while aba1-101/aba1-101 plants reached a rate higher than 80% (Fig. 2). In non-supplemented control media, the germination rates were in all cases higher than 95% (data not shown). The fresh weight of the aba1-101 mutant and two of these transgenic lines was determined 20 d after sowing and found to be 33.0, 78.8, and 97.0%, respectively, of the wild type.

Fig. 2.

Seed germination analysis of aba1-101/aba1-101 plants transformed with the At5g67030 gene. Germination percentages were scored 6 d after sowing, on 200 seeds of Col-0 and aba1-101/aba1-101 plants, together with the T₂ progeny of seven aba1-101/aba1-101 transgenic lines carrying the pBIN19-At5g67030 construct. The photographs were taken 21 d after sowing on soil not supplemented with NaCl or paclobutrazol.

The morphological phenotype of aba1 mutants

The aba1 mutants studied here developed cotyledons that first displayed a purple pigmentation before gradually becoming green. The rosettes of aba1 mutants are smaller than those of their wild types, and their vegetative leaves are reduced in size, with their margins curled downwards and occasionally lobed (Figs 3, 4). Such leaf morphology, in particular that of aba1-102, is reminiscent of that of as1 (asymmetric leaves1) and as2 mutants, whose vegetative leaves are round, lobed, and with margins rolled under (Byrne et al., 2000; Semiarti et al., 2001). It is known that the AS1 and AS2 genes repress KNOX (knotted1-like homeobox) genes, because the loss-of-function as1 and as2 mutations cause the ectopic expression of KNOX genes in leaves (Lincoln et al., 1994; Long et al., 1996; Byrne et al., 2000; Semiarti et al., 2001). In spite of the striking similarity of the leaf phenotypes of as and aba1 mutants, ectopic expression of the KNAT1, KNAT2, and STM genes was not found, as tested by RT-PCR amplifications of RNA extracted from the leaves of aba1/aba1 plants (data not shown).

Fig. 3.

Rosettes of the aba1 mutants studied in this work. Pictures were taken 21 d after sowing in Petri dishes. The scale bar represents 5 mm. The genetic background of aba1-103 (E), aba1-104 (F), aba1-101 (H), and aba1-6 (L) mutants is Col-0 (B), that of aba1-102 (D) is Ws-2 (A), and that of aba1-105 (G), aba1-1 (I), aba1-3 (J), and aba1-4 (K) is Ler (C).

Fig. 4.

Some other morphological traits of aba1 mutants. Transverse sections of third-node vegetative leaves are shown from the Ws-2 wild-type (A, C, E) and the aba1-102 mutant (B, D, F). Sections were made at the central (A–D) or basal (E, F) regions of the lamina. Photographs of transects shown in C and D were taken midway between the mid-vein and leaf margin. (G) Flowers of the aba1-102 and aba1-104 mutants, and the Col-0 wild type. Scale bars represent 100 μm (A–F) and 1 mm (G). (H) Root length in aba1 mutants collected 21 d after sowing.

The internal tissues of Arabidopsis thaliana and other dicotyledonous species manifest dorsoventrality, the palisade layer lying below the adaxial epidermis and consisting of tightly packed elongated cells, below which the spongy mesophyll consists of more rounded, loosely packed cells with frequent intercellular air spaces (Bowman, 1993). At the histological level, the leaves of aba1 mutants display a slightly disorganized mesophyll (Fig. 4B, D, F), making it difficult to distinguish the spongy from the palisade layers. Another aberration found in the aba1 mutants is the absence of the typical abaxial cells surrounding the mid-vein (Fig. 4F). The petioles of aba1 rosette leaves are thick and turn yellowish during vegetative development.

All the mutants studied here display early flowering, bolting about 4 d before their corresponding wild types. Both flower size and inflorescence height are reduced by all aba1 mutant alleles (Fig. 4G). The administration of ABA by spraying greatly improved the efficiency of crosses and significantly increased the yield of seeds. Root length was also found to be 47.8% lower in the aba1-1/aba1-1 plants, 25.8% in aba1-101/aba1-101, and 43.1% in aba1-102/aba1-102, compared with their respective wild types (Fig. 4H).

Phenotypic strength of aba1 alleles

ABA deficiency in aba1 mutants was demonstrated long ago by Koornneef et al. (1982), who analysed, by means of HPLC, the ABA content in aba1-1 and aba1-3 homozygotes (initially named aba1 and aba3, respectively). In addition, the ABA content of aba1-101 homozygous plants, which was found to be under the detection level of the kit, was tested by means of a conventional ELISA test kit. On the contrary, the ABA levels of wild-type plants, as well as those of the aba1-101 transgenic lines carrying a wild-type ABA1 transgene (data not shown), were found to be clearly detectable. Hence, it was assumed that the ABA content of the other six aba1 alleles under study would be lower that those of their corresponding wild types.

The strongest morphological phenotypes were those of the aba1-102, aba1-103, aba1-104, aba1-101, and aba1-6 mutants, and those of aba1-105 and aba1-3 the weakest (Fig. 3). All the aba1 mutants studied here display a wilty phenotype similar to that of ABA-deficient mutants previously described (Koornneef et al., 1982), which is far from obvious when the plants are permanently kept within a growth chamber equipped with humidity control. The germination rates of aba1 mutant seeds sown on Petri dishes containing agar medium supplemented with 100 μM paclobutrazol were quantified (Fig. 5A). In addition, the aba1 mutants were grown in sealed Petri dishes at 60–70% relative humidity in a growth chamber and their water loss measured after opening the plates in a 30% relative humidity environment (Fig. 5B). The Ws-2, Ler, and Col-0 wild-type accessions behaved similarly with regard to water loss and to germination on paclobutrazol-supplemented media. It was found that the aba1-4, aba1-102, aba1-103, and aba1-104 lines display the highest, and aba1-1 and aba1-6 the lowest water losses and germination percentages in paclobutrazol. The present results indicate that the strongest aba1 alleles cause severe effects on both plant morphology and physiological responses.

Fig. 5.

Germination rate in paclobutrazol (A), water loss (B) and relative water content (RWC) (C) of aba1 mutants. (A) Absolute germination values are the mean of two independent experiments and correspond to the germination percentages in 100 μM paclobutrazol. To obtain relative values absolute values were divided by germination percentages obtained in water, and the result was multiplied by 100. Seeds were considered germinated when green and expanded cotyledons were visible 15 d after sowing. (B) The data correspond to the mean weight of 20 rosettes of each genotype. To study water loss, plants were taken out of the growth chamber and kept 15 h at 30% relative humidity and 25 °C. The control plants were kept in the growth chamber. (C) The RWC was measured from vegetative leaves as indicated in the Materials and methods.

In addition, the water status of the aba mutants and their corresponding wild types were studied (see Materials and methods). As shown in Fig. 5C, the RWC of the wild-type lines was very similar, and similar to that of the corresponding mutants. A small reduction in RWC was observed only for the mutants obtained in a Col-0 genetic background.

Two previous structural analyses of the zeaxanthin epoxidase of Arabidopsis thaliana have been published. The first (Marin et al., 1996) described three domains conserved among zeaxanthin epoxidases: a nucleotide (ADP)-binding site, a domain of unknown function, and a flavin-adenine dinucleotide (FAD)-binding site (Fig. 1B). Perturbation of the last domain, which is probably essential for the epoxidation reaction catalysed by zeaxanthin epoxidase (Buch et al., 1995), seems to cause the strong salt-tolerant phenotype of aba1-4, which underlines the importance of this domain for the activity of the enzyme. The nucleotide-binding domain is also assumed to be important for the activity of the protein, which is consistent with the effects of the 30 bp deletion carried by aba1-102, the mutant with the strongest wilty phenotype of those studied here. The aba1-105, aba1-1, aba1-3, and aba1-6 mutations cause phenotypes weaker than that of aba1-102 and damage zeaxanthin epoxidase interdomain regions, which are therefore likely to be less critical for the activity of the enzyme. On the other hand, Xiong et al. (2002) described three different domains: an N-terminal chloroplast transit peptide, a large central mono-oxygenase domain, and an FHA (forkhead-associated) phosphopeptide binding motif, which has an unclear function (Fig. 1C). The mono-oxygenase domain described by Xiong et al. (2002) includes the FAD-binding site described by Marin et al. (1996), whose importance is highlighted by the phenotypic effects of the aba1-4 mutation, which are stronger than those of aba1-105. On the other hand, the FHA site does not seem to make much of a contribution to zeaxanthin epoxidase activity, as indicated by the relatively weak phenotypes of aba1-1 and aba1-105.

Effects of ABA on plant size and weight

Previous authors have suggested that ABA has some sort of positive effect on vegetative growth, since the mutants affected in its synthesis have a smaller size than wild-type plants (Koornneef et al., 1982; Bradford, 1983; Cheng et al., 2002). The aba1 mutants studied in this work showed a significant reduction in size, which is noticeable in rosettes, leaves, inflorescences, and flowers (Figs 3, 4). To quantify the growth inhibition suffered by the mutants, which was found lower than that of their corresponding wild types, the fresh and dry weights of several aba1 mutants were determined (Fig. 6). The aba1-102, aba1-1, and aba1-101 mutants, together with their respective wild types, Ws-2, Ler, and Col-0, were sown in media supplemented with low concentrations of ABA. First it was determined that 1 μM ABA reduces seed germination and inhibits vegetative growth in the wild-type lines, and then the effects on the mutants of 10 and 50 nM ABA-supplemented media were studied. The fresh and dry weights of 30 rosettes (roots were removed) of each line were determined 20 d after sowing (Fig. 6). An increase in the fresh weight of plants grown in the presence of ABA was found, especially in the mutants. Exogenous ABA also caused a great increase in the dry weight of the mutants, but not in that of the wild types.

Fig. 6.

Effects of exogenous ABA on plant weight. Each column represents the mean of three measurements (10 rosettes each). Plants grown in Petri dishes were collected 20 d after sowing and immediately weighed (A), or dried for 24 h at 50 °C and then weighed (B).

The size of third node leaves of mutant and wild-type plants grown on media supplemented, or not, with 50 nM ABA, was determined. As regards the plants grown in non-supplemented media, the leaves of aba1 mutants were smaller than those of the wild type. Such differences in size were much less pronounced in the plants grown on ABA-supplemented media (Fig. 7A–D). In addition, the mesophyll cells of aba1 mutant leaves were found to be much smaller than those of the wild type. In both aba1 mutants and wild-type plants grown on 50 nM ABA medium, the volume of mesophyll cells increased (Fig. 7E–H).

Fig. 7.

Effects of ABA on leaf and mesophyll cell size. (A–D) Third-node vegetative leaves were collected from plants grown in Petri dishes containing non-supplemented medium (left) or 50 nM ABA-supplemented medium (right). Pictures were taken 21 d after sowing. (E–H) Interference contrast photographs of mesophyll cells of the central region of third-node leaves, which were collected from plants grown on non-supplemented medium (E, G) or on 50 nM ABA-supplemented medium (F, H). The scale bars represent 2 mm (A–D) and 40 μm (E–H).

Double-mutant analysis

Double-mutant plants were obtained by intercrossing homozygous aba1 (aba1-102, aba1-103, and aba1-104), aba2 (aba2-13 and aba2-14), and aba3 (aba3-101 and aba3-102) putatively null mutants. As regards morphology, wiltiness and germination rates on 200 mM NaCl agar medium, the aba1 aba2, aba1 aba3, and aba2 aba3 double mutants were indistinguishable from their parental single mutants (data not shown).

Discussion

ABA is involved in vegetative growth and development

The semi-dwarf phenotype of ABA-deficient mutants has been attributed to their inability to retain water, as a result of impaired stomatal closure. However, ABA-deficient mutants grow less than the wild type even in near-saturation humidity conditions, as it has been shown in tomato (Sharp et al., 2000) and Arabidopsis thaliana (Cheng et al., 2002; Cramer, 2002). Moreover, ABA is known to promote primary root elongation in maize (Spollen et al., 2000), and leaf expansion in maize and in Arabidopsis (Sharp, 2002; LeNoble et al., 2004), by restricting ethylene production. Chen et al. (2003), however, demonstrated that the administration of silver thiosulphate, an inhibitor of ethylene activity, does not rescue the semi-dwarf phenotype of the tomato flacca ABA-deficient mutant. In addition, the situation in Arabidopsis thaliana appears to be different from that in maize, since ABA and ethylene act synergistically in Arabidopsis thaliana to inhibit root growth, while chemical or genetic disruption of ethylene signalling results in the reduced sensitivity of root growth to its inhibition by ABA (Beaudoin et al., 2000; Ghassemian et al., 2000).

In this work, it is shown that a series of aba1 mutants displays a reduced size, which is apparent in the leaves, roots, inflorescences, and flowers, and that aba1 mutant rosettes have lower fresh and dry weights than their wild types. It is also shown that low concentrations of exogenous ABA increase the fresh weight of mutant and wild-type plants, as well as the dry weight of the mutants. These growth-promoting effects of ABA suggest that this plant hormone is involved in developmental events other than the hydric balance mediated by stomatal closure. In fact, no significant differences in the RWC of the mutants studied here and their corresponding wild types were found, the only exception being the small reduction observed in the mutants obtained in a Col-0 genetic background when compared with Col-0 (Fig. 5C). The fact that the dry weight of wild-type plants did not change after ABA treatment suggests that this hormone acts as a growth promoter only below a given concentration threshold, which is easily exceeded in non-ABA-deficient plants treated with exogenous ABA.

Not only size, but also shape is abnormal in the aba1 mutants, as observed in the morphology of vegetative leaves, which were studied at whole organ and ultrastructural levels. Leaves of these mutants are reduced in size, their margins curl downwards and display a disorganized mesophyll, traits that are particularly conspicuous in plants carrying strong aba1 alleles. This observation suggests a role for ABA in leaf organogenesis. Failure to develop distinct spongy and palisade mesophyll layers has been described in other Arabidopsis thaliana mutants and transgenic lines, such as those constitutively over-expressing the CYCD3;1 gene, which encodes the D-type cyclin CYCD3, reducing the proportion of cells in the G₁ phase of the cell cycle (Dewitte and Murray, 2003). Recently, LeNoble et al. (2004) showed that normal endogenous ABA levels are necessary for correct leaf expansion.

When the morphogenetic effects of exogenous ABA were analysed at cell level, it was observed that cells of the internal leaf tissues reached a larger size in the presence of low concentrations of exogenous ABA, both in aba1-mutant and wild-type plants. These results reveal a role for ABA in cell expansion, a function that might explain, at least partially, the semi-dwarf phenotype of ABA-deficient mutants, and which suggests that this hormone acts as a growth promoter under some circumstances (Cheng et al., 2002; Cramer, 2002; Sharp and LeNoble, 2002; Makela et al., 2003). The proposal of a role for ABA as a growth promoter is supported, in addition, by the demonstration that ABA accumulation is required for the maintenance of primary root elongation at low water potentials (Spollen et al., 2000; Sharp and LeNoble, 2002), and by the observation that ABA concentration peaks half-way through seed development, when cell division ceases and active cell expansion takes place (Black, 1991; Rock and Quatrano, 1995).

Zeaxanthin epoxidase (Duckham et al., 1991; Rock and Zeevaart, 1991) catalyses the epoxidation of the ABA precursors, zeaxanthin and antheraxanthin. These xanthophylls play a photoprotective role, as they are associated with photosystem II of the photosynthetic apparatus, preventing the production of free radicals caused by an excess of light (Nigoyi et al., 1998; Pogson et al., 1998). The npq2 (non-photochemical quenching2) mutations (Nigoyi et al., 1998), which were isolated in a search for mutants with altered non-photochemical quenching, were found to be alleles of the ABA1 gene. The kinetics of induction and relaxation, but not the extent of non-photochemical quenching, are affected in the npq2 mutants (Nigoyi et al., 1998). Indeed, the npq2 mutants accumulate zeaxanthin and do not exhibit reduced photoprotection, which, on the contrary, increases slightly, somewhat diminishing photosynthesis efficiency under sub-saturating light (Niyogi et al., 1998). However, vegetative growth is restored by ABA supplementation in the aba1 mutants, which reach normal size despite the fact that their zeaxanthin levels remain high. Furthermore, it was found that aba1 mutants grown in the dark display reduced hypocotyl elongation, which is promoted by exogenous ABA, conditions in which no effects of light, photosynthesis, or nonphotochemical quenching can be presumed (JM Barrero, PL Rodríguez, MR Ponce, JL Micol, unpublished results). These observations suggest that size differences between aba1 mutants and their wild types are caused mainly by ABA deficiency, and cannot be explained by a decrease in photosynthesis efficiency.

Effects of aba1 alleles on the biosynthesis of ABA

The analysis of the phenotypic strength of the mutant alleles presented here sheds light on the nature of zeaxanthin epoxidase domains that are crucial for its activity. The present analysis shows that the nucleotide-binding domain and FAD-binding domain are necessary for the complete functioning of the enzyme, since aba1-102 and aba1-4 alleles are affected in these regions and display very strong phenotypes. The inter-domain and carboxy-terminal regions of the protein seem to be less important for its enzymatic activity. The extreme phenotype caused by the aba1-3 allele might indicate the existence of a novel functionally important domain in the protein, since its mutation causes a single substitution (proline to alanine) in a conserved region (Fig. 1B, C).

If plants have a single ABA biosynthesis pathway and if it is assumed that ABA is essential for plant life, null mutations abolishing ABA production should be lethal. However, although several of the aba1 mutants described in this work probably carry null alleles of the ABA1 gene, as the present molecular analyses indicate, they are still viable and grow (10–15 cm tall), producing 50–200 seeds per plant. Furthermore, when mutants carrying the presumably null or extremely hypomorphic aba1, aba2, and aba3 alleles were intercrossed, the severity of their phenotypes did not increase in the double mutants obtained, indicating that obstruction of two steps of the ABA biosynthetic pathway has the same phenotypic effects as obstruction of only one. These results suggest the existence of an alternative, less important ABA biosynthesis pathway. In fact, the aba2 mutant of Nicotiana plumbaginifolia, with impaired zeaxanthin epoxidation, is still able to synthesize ABA to 30% of the level found in the wild type (Marin et al., 1996), and the aba1-4 mutant of Arabidopsis thaliana, which presents a strongly ABA-deficient phenotype, is also capable of synthesizing small quantities of violaxanthin and neoxanthin (Rock and Zeevaart, 1991). Moreover, it has been shown that in homozygotes for the aba2-11 null allele, small quantities of ABA can still be detected (González-Guzmán et al., 2002). Finally, transgenic plants carrying a transgene producing an anti-ABA antibody display a much more severe phenotype than aba mutants (Phillips et al., 1997). Taken together, these results support the existence of an additional minor pathway for the synthesis of ABA, which might be less specific or act in a redundant way to produce small quantities of ABA (Cutler and Krochko, 1999; Audran et al., 2001).

The authors wish to thank V Quesada and P Robles for comments on the manuscript, and JM Serrano and S Gerber for technical assistance. This work was supported by research grants BIO2000-1082 (to JLM) and BIO2002-03090 (to PLR) from the Ministerio de Ciencia y Tecnología of Spain. MGG, JMB, and PP were respectively supported by fellowships from the Ministerio de Educación y Cultura of Spain, the Ministerio de Ciencia y Tecnología of Spain, and the Conselleria de Cultura, Educació i Ciència of the Generalitat Valenciana.