An Embryo-Defective Mutant of Arabidopsis Disrupted in the Final Step of Biotin Synthesis

Abstract

Auxotrophic mutants have played an important role in the genetic dissection of biosynthetic pathways in microorganisms. Equivalent mutants have been more difficult to identify in plants. The bio1 auxotroph of Arabidopsis thaliana was shown previously to be defective in the synthesis of the biotin precursor 7,8-diaminopelargonic acid. A second biotin auxotroph of A. thaliana has now been identified. Arrested embryos from this bio2 mutant are defective in the final step of biotin synthesis, the conversion of dethiobiotin to biotin. This enzymatic reaction, catalyzed by the bioB product (biotin synthase) in Escherichia coli, has been studied extensively in plants and bacteria because it involves the unusual addition of sulfur to form a thiophene ring. Three lines of evidence indicate that bio2 is defective in biotin synthase production: mutant embryos are rescued by biotin but not dethiobiotin, the mutant allele maps to the same chromosomal location as the cloned biotin synthase gene, and gel-blot hybridizations and polymerase chain reaction amplifications revealed that homozygous mutant plants contain a deletion spanning the entire BIO2-coding region. Here we describe how the isolation and characterization of this null allele have provided valuable insights into biotin synthesis, auxotrophy, and gene redundancy in plants.

Biotin is an essential vitamin that facilitates CO₂ transfer during carboxylation, decarboxylation, and transcarboxylation reactions (Dakshinamurti and Bhagavan, 1985; Knowles, 1989). Bacteria, plants, and some fungi are capable of synthesizing biotin directly from chemical intermediates. Other organisms must obtain biotin from their surrounding environment. The biosynthetic pathway for biotin was first elucidated in Escherichia coli and Bacillus subtilis through biochemical studies involving the analysis of auxotrophic mutants (Eisenberg, 1973; Pai, 1975). Biotin operons from several prokaryotes have now been sequenced and analyzed in detail (Otsuka et al., 1988; Cronan, 1989; Gloeckler et al., 1990; Bower et al., 1996). Enzymes required for biotin synthesis in E. coli and Bacillus sphaericus have been purified and their activities characterized in vitro (Izumi et al., 1981; Ploux and Marquet, 1992; Ploux et al., 1992; Alexeev et al., 1994; Huang et al., 1995). Related genes required for biotin synthesis have also been identified in a variety of other microorganisms (Zhang et al., 1994; Fleischmann et al., 1995; Bult et al., 1996). The entire biosynthetic pathway has therefore been the subject of extensive investigation.

The common pathway for biotin synthesis in plants and microorganisms is summarized in Figure 1. Although most steps in the pathway have been clearly documented in the literature, questions remain concerning the immediate precursor of pimeloyl CoA in different microorganisms, the role of specific proteins, cofactors, and intermediates associated with the final step of the pathway, and the nature of the sulfur donor required for biotin synthesis (Baxter et al., 1992; Marquet et al., 1993; Ifuku et al., 1994; Birch et al., 1995; Bower et al., 1996; Sanyal et al., 1996). The conversion of dethiobiotin into biotin, which is catalyzed in part by the enzyme biotin synthase, is perhaps the most fascinating and complex part of the pathway because it involves the addition of sulfur to form a thiophene ring. This reaction may also represent a rate-limiting step for biotin synthesis (Baldet et al., 1993b).

Figure 1.

Biotin biosynthetic pathway in plants and microorganisms. Letters correspond to cloned bio genes of E. coli. Numbers correspond to mutant bio genes of Arabidopsis. The immediate precursor of pimeloyl CoA appears to be pimelic acid in many but not all organisms.

The identity of the sulfur donor in biotin synthesis has long been questioned. Candidate molecules have included Cys, Met, and S-adenosylmethionine. Recent studies have provided contradictory evidence. Birch et al. (1995) reported that Cys may be the sulfur donor based on their ability to label biotin with [³⁵S]Cys in a cell-free extract of an E. coli strain designed to overexpress the cloned biotin synthase gene. Sanyal et al. (1996) could not detect a similar incorporation of label into biotin using a highly purified preparation of biotin synthase. The purified E. coli enzyme appears to be a dimer that contains two iron atoms and two acid-labile sulfur atoms per monomer (Sanyal et al., 1994). Several additional proteins and cofactors appear to be required for efficient biotin synthesis in cell-free extracts. These include flavodoxin, flavodoxin reductase, NADPH, and perhaps several compounds of low molecular weight present in crude extracts (Ifuku et al., 1994; Birch et al., 1995; Sanyal et al., 1996). A potential intermediate in this final step of the pathway (9-mercaptodethiobiotin) has also been identified, synthesized in the laboratory, and examined for its role in biotin synthesis (Baxter et al., 1992; Baldet et al., 1993b; Marquet et al., 1993).

Several different strategies have been used to study biotin synthesis and utilization in plants. One approach has been to analyze biotinylated proteins (Nikolau et al., 1985). Four biotin-dependent carboxylases have been examined to date (Harwood, 1988; Wurtele and Nikolau, 1990, 1992; Alban et al., 1993). Another biotinylated protein identified in pea seeds may play a role in sequestering biotin late in embryogenesis for subsequent use during germination (Duval et al., 1993, 1994). A holocarboxylase synthetase responsible for biotin attachment to plant proteins has also been identified (Tissot et al., 1996, 1997). Other studies have dealt with biotin content and the relative amounts of free and protein-bound biotin in plant cells (Baldet et al., 1993a). Biotin synthesis has been examined in part by noting the fate of radioactive precursors and identifying known chemical intermediates in plant tissues (Baldet et al., 1993b). These biochemical studies have for the most part confirmed the similarity of the prokaryotic and eukaryotic pathways for biotin synthesis, although questions remain concerning details of the initial and final steps.

The intracellular location of biotin synthesis also remains to be defined. Some studies suggest that biotin synthesis occurs primarily in the chloroplast, whereas other results are more consistent with a cytosolic or mitochondrial localization (Shellhammer, 1991; Baldet et al., 1993a; Patton et al., 1996a; Weaver et al., 1996). These studies have been complicated by the extremely small amounts of biotin produced by most plant cells. Molecular characterization of the biosynthetic pathway has dealt primarily with the biotin synthase gene. Three laboratories have recently cloned the cDNA corresponding to this gene from Arabidopsis (Baldet and Ruffet, 1996; Patton et al., 1996a; Weaver et al., 1996).

An alternative approach used to study biotin synthesis in plants has been the isolation and characterization of auxotrophic mutants. All biotin auxotrophs identified to date have been obtained from collections of embryo-defective mutants of Arabidopsis (Meinke, 1994). The bio1 mutant was the first plant auxotroph shown to result in embryo lethality (Schneider et al., 1989). Mutant bio1 embryos remain pale throughout development, typically arrest between the heart and cotyledon stages of embryogenesis, contain severely reduced levels of biotin, and can be rescued by biotin, dethiobiotin, or 7,8-diaminopelargonic acid (Shellhammer and Meinke, 1990). The initial model that this mutant was defective in the conversion of 7-keto-8-aminopelargonic acid to 7,8-diaminopelargonic acid (Shellhammer, 1991) has recently been confirmed by demonstrating that mutant plants can be rescued by introduction of a functional copy of the bioA gene of E. coli through Agrobacterium-mediated transformation (Patton et al., 1996b).

We describe in this report the isolation of a second biotin auxotroph of Arabidopsis disrupted at a different step in the biosynthetic pathway, the conversion of dethiobiotin to biotin. Once again, the vitamin requirement of this bio2 mutant was identified by comparing the growth of mutant embryos on minimal and enriched media. We demonstrate here that the bio2 mutation corresponds to a deletion of the entire genomic coding region for biotin synthase in Arabidopsis. This observation has allowed us to establish the cellular and developmental consequences of a complete loss of biotin synthesis in plants. The isolation of a second biotin auxotroph and the failure to identify other types of nutritional mutants among embryonic lethals also raises fundamental questions concerning the scarcity of plant auxotrophs. We conclude that although alternative methods may be devised to identify additional auxotrophs in the future, the number of mutants with defined nutritional requirements that can be identified in plants will continue to be limited by widespread redundancy of essential genes and biochemical pathways.

MATERIALS AND METHODS

Mutant Isolation and Plant Maintenance

The bio2 mutation was induced by chemical mutagenesis of Arabidopsis thaliana ecotype Columbia. The mutagenesis was performed by R. Dinkins (University of British Columbia, Vancouver; University of Kentucky, Lexington) as part of a screen to identify high-chlorophyll fluorescence mutants. Mature seeds were treated for 30 min with 0.3% (v/v) ethyl methanesulfonate, followed by rinsing for 2 h with distilled water. Progeny M₂ seeds were harvested from pools of 5 to 10 M₁ plants. Some of the resulting M₂ plants were scored for embryo-defective mutations by screening siliques for the presence of 25% abnormal seeds. A high frequency of mutant phenotypes was observed in the M₂ generation; albino seedlings were found in 30% of the pools examined. The bio2 mutant family, originally named M₂-266–3G and then emb49, was saved because it segregated for an embryo-defective mutation with a consistent phenotype. Heterozygous plants were subsequently grown in 16-/8-h light/dark cycles and identified by screening immature siliques for defective seeds (Heath et al., 1986; Meinke, 1994). Isolation and characterization of the bio1 auxotroph, known originally as mutant 122G-E, was described by Meinke (1985), Schneider et al. (1989), and Shellhammer and Meinke (1990). The bio1–1 allele used here was isolated following ethyl methanesulfonate seed mutagenesis of ecotype Columbia. A second allele (bio1–2) was recently identified by R. Fischer and colleagues (University of California, Berkeley) in the Feldmann collection of embryo-defective mutants generated by Agrobacterium-mediated seed transformation of ecotype Wassilewskija (Feldmann, 1991; Yadegari et al., 1994).

Genetic Analysis of Mutant Plants

Segregation ratios were calculated by determining the locations of normal and aborted seeds within heterozygous siliques (Meinke, 1994). The distribution of aborted seeds was used to examine gametophytic expression of the mutant gene (Meinke, 1982, 1985). Terminal phenotypes of arrested embryos were determined by examining mutant seeds under a dissecting microscope (Meinke, 1994). Early defects were identified in seeds cleared with Hoyer's solution (7.5 g of gum arabic, 100 g of chloral hydrate, and 5 mL of glycerol in 30 mL of water) and viewed with a compound microscope equipped with Nomarski optics (model BH-2, Olympus). Mapping of the bio2 (emb49) locus with visible markers and genetic complementation tests with linked mutants were performed as described by Franzmann et al. (1995). Two approaches were used to eliminate unlinked mutations in the parental bio2 family. The first involved selection of the most normal plants found in each generation following self-pollination. The second approach involved outcrossing to wild-type Columbia plants. Five outcross generations have been produced to date. Both approaches improved the vegetative appearance of heterozygous plants but failed to eliminate the nonbolting phenotype of rescued homozygotes.

Screen for Auxotrophic Mutants

Arrested embryos from 31 embryo-defective mutants were tested for their response in culture on minimal and enriched media. The methods used were similar to those described by Baus et al. (1986). Transition mutants were chosen because their embryos arrested early in development but were still amenable to manipulation in culture. The following mutants were tested: emb19, emb20–2, emb34, emb49, emb52, emb67, emb69, emb81, emb83, emb86–2, emb90, emb94, emb106–1, emb106–2, emb111, emb131, emb149, emb150, emb151, emb154, emb156–2, emb213, emb222, emb224, emb228, emb234, emb236, emb245, emb246, emb253, and emb279. Mutant seeds and isolated embryos from heterozygous siliques at a cotyledon stage of development were placed in culture. Segments of normal cotyledons were included as controls. Minimal medium contained 0.8% Phytagar (Life Technologies), inorganic salts described by Murashige and Skoog (1962), 3% Suc, 500 μm inositol, 5 μm thiamine hydrochloride, 0.1 mg/L 1-NAA, and 1 mg/L 6-benzylaminopurine, adjusted to pH 5.7. Thiamine was included because the corresponding auxotrophs are known to be seedling lethals.

Vitamin plates were supplemented with 5 μm each of p-amino-benzoic acid, nicotinamide, calcium pantothenate, pyridoxine hydrochloride, biotin, and riboflavin-5′-P, and 25 μm choline chloride. Enriched plates were supplemented with the vitamins noted above, 50 μm each of 20 l-amino acids, and 50 μm each of five nucleosides. Organic supplements were sterilized with a 0.2-μm syringe filter (Gelman, Ann Arbor, MI) and then added to autoclaved media. Chemicals used in media preparation were obtained from Sigma and Fisher Scientific. For initial screens, 20 to 40 seeds and embryos from each mutant line were tested on each type of medium. Growth responses were noted after 2 to 4 weeks in culture. The biotin requirement of bio2 embryos was confirmed using media supplemented with 1 to 10 μm biotin or dethiobiotin. Several rescued plantlets with roots were transferred to soil and watered daily with 1 mm biotin as described by Schneider et al. (1989). Mature seeds from rescued heterozygotes were germinated on media without phytohormones. Rescued seedlings from biotin plates were transplanted to soil approximately 2 weeks after germination.

Mapping the Cloned BIO2 Gene

The BIO2 gene was mapped using the RI lines of Lister and Dean (1993), and an SSLP was located in the first intron of the gene. The Landsberg allele has 49 copies of a “CT” repeat in this region; the Columbia allele has only 15 copies. The following primer pair flanking this polymorphism was used to amplify DNA from RI plants: 5′-GGAGTAGAGATGAAATCAAGTC-3′ and 5′-GAACATGTCTATGAACCTGAGC-3′. Bulked F₈ seeds were obtained from the Arabidopsis Biological Resource Center (The Ohio State University, Columbus). DNA was isolated from 10 to 20 seedlings from each of 94 lines grown in liquid cultures as described by Reiter et al. (1992) and amplified using standard PCR conditions. The resulting segregation data were compared with data for 157 other markers provided by J. Ecker (University of Pennsylvania, Philadelphia) using Map Manager, a computer program developed by K. Manly and R. Cudmore (Roswell Park Cancer Institute, Buffalo, NY).

PCR Analysis of the bio2 Allele

The molecular basis of the bio2 mutation was examined by comparing DNA amplified from leaves and callus of homozygous mutant plants rescued in culture and wild-type control plants. DNA was isolated from lyophilized tissue as described by Reiter et al. (1992). DNA amplifications were carried out in 100-μL reactions containing 100 ng of genomic DNA, 0.15 μm of each primer, 1.5 mm MgCl₂, 10 mm Tris-HCl (pH 8.3), 50 mm KCl, 200 μm of each deoxyribonucleoside triphosphate, and 2.5 units of AmpliTaq DNA polymerase (Perkin-Elmer). The standard thermal profile included an initial denaturation for 5 min at 95°C followed by 30 cycles at 94°C for 30 s, 50°C for 30 s, and 72°C for 3 min. Three gene-specific primer pairs that span the BIO2 locus were used: 5′-ACGCCGTTATCATGTGAAG-3′ and 5′-TCCATGGAAGAGGAGGTC-3′ (exon 1 to exon 2), 5′-CTAACTTTCTGGGCTCTCAC-3′ and 5′-GGAGATATTCCTTCTGGTC-3′ (exon 3 to exon 4), and 5′-GGCTTTGAGAACCGTTTGTG-3′ and 5′-GAGAAATCTAGACATCTTCG-3′ (exon 6). Other gene-specific primer pairs used as controls were: 5′-GCGTGACCATCAAGA- CTAAT-3′ and 5′-AAAAATGGCAACACTTTGAC-3′ (alcohol dehydrogenase; ADH), 5′-GTTCTCTTCTGTGTCATC-3′ and 5′-TCCCCAGGTAAAGACGTC-3′ (5′-phos- phoribosyl-5-amino-imidazole synthetase), 5′-GTTATCAAGGTGGGAAGA-3′ and 5′-CCAATAGATGACGG- AAGA-3′ (2-keto-3-arabinoheptulosonate 7-P synthase), 5′-CGTAGATGATGCGTCCAG-3′ and 5′-TAAGCATAGGTCCCAATA-3′ (phosphoribosyl-anthranilate transferase), 5′-CAGATGAGAGTGCCTAAA-3′ and 5′-ACCTTTTCCTTTCGCCTC-3′ (Gln synthetase), 5′-GGCGAT-TCTCCGTTACAG-3′ and 5′-CATCATCAGCTCGTC-AAC-3′ (β-tubulin 4), and 5′-ATTCCTTAACGCCGGAA-TATTCGG-3′ and 5′-CTTAACATATTGGAATGGGA-GCTC-3′ (Phe ammonia-lyase).

DNA and RNA Gel-Blot Analysis

For Southern blots, 1 μg of genomic DNA from wild-type and rescued bio2/bio2 plants was digested with 20 units of EcoRI for 2 h, electrophoretically separated in 0.75% agarose, and alkaline blotted to Hybond N⁺ (Amersham). Blots were probed sequentially with a full-length BIO2 cDNA (Patton et al., 1996a), a control cDNA corresponding to a single-copy gene of Arabidopsis, and a BIO2 λ genomic clone containing a 22-kb insert with the BIO2 coding region located on a 5-kb EcoRI fragment near the middle of the clone. Probes labeled by the random-priming method (Feinberg and Vogelstein, 1983) were purified using NucTrap push columns (Stratagene). Genomic DNA isolation, blot hybridization (65°C), and washes were carried out as described by Reiter et al. (1992). Tissue harvested for RNA isolation was frozen in liquid nitrogen and stored at −80°C. Total RNA was isolated using the phenol extraction method (Lagrimini et al., 1987) and quantified by measuring UV A₂₆₀. Samples containing 2 μg of total RNA were separated on formaldehyde electrophoresis gels and blotted to Genescreen-Plus membranes (NEN). Blots were hybridized to radioactively labeled probes and washed as described by Ausubel et al. (1987).

Co-Segregation of the bio2 Deletion and Nonbolting Phenotype

Segregating populations of BIO2/BIO2, bio2/BIO2, and bio2/bio2 plants were generated by either planting mature seeds obtained from rescued heterozygotes directly on soil or germinating these seeds on culture medium containing 10 μm biotin and then transplanting the seedlings to soil. All plants grown in soil were supplemented daily with 1 mm biotin. Rosette leaves were harvested from each plant after 2 weeks of growth. Plants were scored several weeks later for the presence of a primary bolt. DNA was isolated from frozen leaf samples stored in a microfuge tube. Samples were ground in liquid nitrogen, briefly vortexed with 100 μL of warmed 2× CTAB extraction buffer (2% cetyl-trimethyl-ammonium bromide, 2% PVP, 200 mm Tris, pH 7.5, 1.4 m NaCl, and 20 mm EDTA), and incubated at 65°C for 10 min. Samples were then extracted with 1 volume of chloroform:isoamyl alcohol (24:1) and nucleic acids were precipitated by adding 0.6 volume of isopropanol to the aqueous phase. Pellets were washed with 70% ethanol, dried in a vacuum, and resuspended in 50 μL of TE buffer (10 mm Tris and 1 mm EDTA) at pH 7.5. Duplicate PCR reactions containing 1-μL aliquots of DNA were used to identify homozygous mutant plants lacking the BIO2 gene. Two primer pairs described above were used: the BIO2 exon 6 pair and the β-tubulin 4 pair as a positive control. Each plant was then assigned a genotype (bio2/bio2 or BIO2/-) to compare with the nonbolting phenotype.

RESULTS

The bio2 Auxotroph Is an Embryo-Defective Mutant

The bio2 auxotroph was originally identified as an embryo-defective mutant following EMS seed mutagenesis. The pattern of inheritance indicated that a single recessive mutation was responsible for the observed defects in seed development. The mutant was named emb49 before the biotin requirement of arrested embryos was discovered (Meinke, 1994). The bio2 allele was mapped by crossing heterozygotes with plants homozygous for recessive visible markers and scoring F₂ plants for marker phenotypes and aborted seeds following self-pollination. The bio2 locus was assigned to position 67 cM on chromosome 2 of the classical genetic map. This location is shown on the updated genetic map available through the World Wide Web (http://mutant.lse.okstate.edu/). The bio1 locus described previously maps to position 74 cM on chromosome 5 (Patton et al., 1991; Franzmann et al., 1995). Only a single bio2 allele has been identified to date. Five other emb mutants located within 5 cM of the bio2 locus were shown through complementation tests to be defective in different genes. The original bio2 mutant, which produced deformed rosettes and developed more slowly than normal, was outcrossed to wild-type plants for several generations to remove unlinked mutations. Embryo defects and biotin responses characteristic of the parental line remained constant through successive generations.

Embryo-defective mutants have traditionally been grouped into phenotypic classes based on the size and shape of mutant embryos at maturity (Muller, 1963; Meinke, 1994). The bio2 auxotroph was originally classified as a transition mutant. Arrested embryos remain white or pale yellow-green throughout development, often reach a globular or heart shape prior to desiccation, and consistently fail to germinate at maturity. Arrested bio2 embryos are typically smaller and blocked earlier in development than arrested bio1 embryos. Terminal phenotypes of these two mutants are compared in Figure 2. When aborted seeds from selfed bio1 and bio2 heterozygotes grown under identical conditions were compared, most of the bio2 embryos examined arrested before the torpedo stage of development, whereas every bio1 embryo examined reached a more advanced cotyledon stage. The bio2 mutation therefore appears to disrupt embryogenesis to a greater extent than the bio1 mutation.

Figure 2.

Terminal phenotypes of bio1 and bio2 embryos obtained from heterozygous siliques. A, Examples of phenotypic classes observed. Embryos ranged from the globular (A) to cotyledon (E–G) stages. B, Distribution of phenotypic classes (A–G) in random samples of 125 arrested embryos from each mutant. Class X embryos were too small to be visible under a dissecting microscope. Note the different distribution of embryo phenotypes found in bio1 and bio2 mutant seeds.

Examination of cleared mutant seeds with Nomarski optics revealed a variety of early defects in cell division patterns and a characteristic elongation of the mutant embryo proper. Examples of these developmental abnormalities are shown in Figure 3. The earliest defect observed in bio2 seeds occurred when the embryo proper was composed of a single cell (Fig. 3A). Related but less-severe abnormalities were found in bio1 mutant seeds. These defects were not caused by another linked mutation because they disappeared in heterozygous plants watered with biotin. The unusual pattern of morphogenesis observed in some bio2 embryos (Fig. 3C) demonstrates that disrupting a general metabolic function such as biotin synthesis can have developmental consequences that may resemble defects in pattern formation. This supports the conclusion that interesting defects in cell division patterns are not necessarily the result of mutations in genes that play a direct role in the regulation of morphogenesis (Meinke, 1996).

Figure 3.

Early defects in development of bio2 seeds revealed after treatment with Hoyer's solution and examination with Nomarski optics. A and C, Immature mutant seeds. Note unusual elongation of apical cell (A) and embryo proper (C) in mutant seeds. B and D, Wild-type seeds at the proembryo (B) and late-globular (D) stages. Scale bars = 20 μm.

Homozygous Mutant Embryos Are Rescued by Biotin

The biotin requirement of mutant embryos was identified during a search for additional auxotrophs among existing embryo-defective mutants. The approach used was similar to that reported for the bio1 mutant (Baus et al., 1986; Schneider et al., 1989). Aborted seeds and isolated embryos from 31 mutants arrested at the globular to cotyledon stages were plated on basal and enriched media supplemented with vitamins and amino acids. The transition class was chosen for analysis because mutant embryos arrested during a period of rapid growth, which might be expected for an auxotroph, and were sufficiently large to remove from aborted seeds. One mutant (bio2) consistently responded only on enriched medium. Biotin was identified as the essential nutrient after mutant embryos were plated on different mixtures of vitamins. The failure of mutant embryos to respond on basal media was consistently observed in repeated experiments. The results of several independent tests are presented in Table I. The positive response of bio2 embryos in the presence of biotin was reproducible but less pronounced than previously reported for bio1. This was likely due to the small size of bio2 embryos at the time of culture. Mutant embryos arrested early in development are typically more difficult to rescue in culture than those arrested later in development.

Table I.

Response of mutant embryos in culture

Mutant	Mediuma	Embryos Cultured	Embryo Response after 7 d in Culture
			Percent greenb	Embryos measuredd	Average lengthe	Maximal lengthe
				no.	mm
bio2	Minimal	139	0	10	0.6	0.8
	Dethiobiotin	181	2c	41	0.5	0.8
	Biotin	143	37	40	1.5	4.0
bio1	Minimal	6	0	6	0.8	1.0
	Dethiobiotin	22	100	22	3.3	5.0
	Biotin	22	100	22	4.1	7.0

^aEnriched media contained 1 or 5 μm biotin or dethiobiotin. 

^bPercentage of cultured embryos that became green after 7 d. This was the best early indication of a positive response in culture. Many small bio2 embryos that failed to respond on biotin were likely damaged at the time of culture. 

^cThese embryos later turned white and stopped growing. 

^dNumber of embryos measured with an ocular micrometer after 7 d in culture. 

^eAverage lengths were calculated for all embryos that enlarged in culture. Maximal lengths represent the largest embryo found on each medium. Initial lengths at the time of culture averaged 0.2 mm (bio2) and 0.4 mm (bio1). 

A more significant difference between bio1 and bio2 mutants was found in the response of arrested embryos to dethiobiotin, the immediate precursor of biotin in plants and microorganisms. These results are summarized in Table I and Figure 4A. Arrested bio2 embryos were not rescued by dethiobiotin, whereas bio1 embryos included as positive controls on the same plate were rescued. The failure of bio2 embryos to grow on dethiobiotin was therefore not caused by culture conditions but, rather, by the inability of mutant embryos to convert this intermediate into biotin. Even high concentrations of dethiobiotin (10 μm) were unable to promote sustained growth of bio2 embryos in culture. These results are consistent with the conclusion that bio2 is defective in the final step of biotin synthesis in plants, the conversion of dethiobiotin into biotin.

Figure 4.

Response of immature mutant embryos (A) and mature rescued seeds (B) in culture. A, Arrested embryos were removed from immature siliques of heterozygous plants, plated on basal and enriched media, and observed after several weeks in culture. The top row of plates contained bio2 embryos on the left half of each plate and bio1 embryos on the right half. The left plate is a basal medium; the right plate contains 1 μm dethiobiotin. The bottom row of plates contains plants derived from bio2 embryos rescued on 1 μm biotin. Note that bio2 embryos were rescued only in the presence of biotin. B, Mature seeds from heterozygous (bio2/BIO2) plants grown in pots watered with biotin were plated on basal medium (right) and 1 μm biotin (left). Each plate received seeds from a single silique. Segregating mutant seedlings appeared pale in the absence of biotin (right) but normal in the presence of biotin (left).

Most bio2 plantlets rescued on biotin appeared normal but were difficult to transplant to soil because roots were not well established. An alternate method was therefore devised to generate large numbers of homozygous mutant plants for analysis. Pots containing a mixture of bio2/BIO2 and BIO2/BIO2 seedlings were watered daily with 1 mm biotin. This treatment rescued many of the mutant seeds produced by heterozygous (bio2/BIO2) plants later in development. Several plants in these pots were identified as heterozygotes based on the presence of a few pale seeds. Mature siliques from these rescued heterozygotes were then harvested and the resulting seeds were germinated on plates in the presence and absence of biotin. Examples of such plates are shown in Figure 4B. Approximately 25% of the seedlings grown on a basal medium became pale shortly after germination and failed to produce a viable rosette. In contrast, all of the seedlings grown in the presence of biotin were phenotypically normal. These results indicate that many homozygous mutant seeds produced by heterozygous plants watered with biotin were indeed rescued, survived desiccation, and germinated to produce normal seedlings.

The bio2 Mutant Contains a Deletion Spanning the Biotin Synthase Gene

The failure of bio2 arrested embryos to grow in the presence of dethiobiotin suggested that the mutation disrupted the gene coding for biotin synthase. The Arabidopsis cDNA for biotin synthase, which was named BIO2 before the corresponding mutant was identified, has been cloned and characterized in several laboratories (Baldet and Ruffet, 1996; Patton et al., 1996a; Weaver et al., 1996). The structure of this gene is summarized in Figure 5A. If bio2 is indeed defective in biotin synthase, then the mutation should map to the same chromosome location as the cloned gene. This prediction was tested by mapping the cloned gene on RI lines of Arabidopsis developed by Lister and Dean (1993). The estimated map location was then compared with the known position of the bio2 locus at 67 cM on chromosome 2. The similarity of BIO2 sequences obtained from the Landsberg and Columbia ecotypes used to generate the RI lines made it difficult to identify a restriction fragment-length polymorphism for mapping purposes. The most striking difference was an SSLP located within the first intron of the BIO2 gene. The location of this repeat is shown in Figure 5A. DNA samples from 94 RI lines were then screened for polymorphic PCR products generated with primers flanking this SSLP and the resulting data were analyzed with the Map Manager program. The results placed the BIO2 gene between nga168 and g4514, at position 79 cM, on the RI map of chromosome 2. This corresponds to position 69 cM on the classical map when expansion associated with RI populations is taken into account. The cloned gene and mutant allele therefore map to the same chromosome location.

Figure 5.

Molecular characterization of the bio2 locus. A, Structure of the wild-type (BIO2) gene. Thin lines represent introns; thick bars correspond to exons. Horizontal arrows note the locations of primers used to confirm the presence of a deletion in mutant plants. The vertical arrow denotes a region within the first intron that contains a “CT” repeat detectable as an SSLP between Landsberg erecta and Columbia ecotypes. B, Analysis of PCR products generated using template DNA from wild-type (BIO2/BIO2) and mutant (bio2/bio2) plants. “BIO2” lanes contain PCR products generated using primers spanning the BIO2 locus. Locations of these primer pairs are designated with arrows in A. “Control” lanes contain products generated using primers specific for seven different single-copy genes in Arabidopsis. Lanes with identical numbers used the same primer pairs. Reference lanes contain a 1-kb ladder. Note that DNA isolated from mutant plants yielded PCR products in all cases tested except with BIO2 primers. C, Gel blot of genomic DNA isolated from wild-type (BIO2) and mutant (bio2) plants and digested with EcoRI. The same blot was probed sequentially with the full-length BIO2 cDNA (1); a random single-copy gene used as a positive control (2); and a λ clone containing the BIO2 gene located within a 22-kb insert (3). There was no hybridization between the BIO2 cDNA probe and DNA isolated from mutant plants. Three bands marked with arrows (13 kb total) are missing from the blot of bio2 DNA probed with the λ BIO2 genomic clone.

PCR primers covering the BIO2 locus were then used to compare DNA isolated from wild-type and rescued bio2 plants. As shown in Figure 5B, the expected BIO2 products were recovered from wild-type plants but not from mutant plants. PCR products generated using primers within other single-copy (control) genes were recovered equally from both mutant and wild-type plants. These results suggested that bio2 plants contained a deletion that spanned the entire coding region of the biotin synthase gene. Results of Southern blots of genomic DNA isolated from wild-type and bio2 mutant plants and probed with full-length BIO2 cDNA and genomic clones were consistent with this interpretation. A typical DNA gel blot is shown in Figure 5C. These results suggested that BIO2 is a single-copy gene and that mutant plants are missing at least 13 kb of DNA surrounding the BIO2 locus. In addition, RNA gel blots revealed the presence of a transcript complementary to the BIO2 cDNA probe in wild-type plants but not in rescued mutant plants (data not shown). Taken together, these results indicate that the bio2 mutation is associated with a deletion that spans the entire coding region of the biotin synthase gene and probably extends into adjacent regions of the genome. Although large deletions are not common following EMS mutagenesis, they have been reported to occur in other organisms when high doses of mutagen were involved (Ashburner, 1989).

Mapping the distribution of aborted seeds in siliques of selfed heterozygotes has been used previously to identify mutations that interfere with both embryogenesis and pollen-tube growth (Meinke, 1982, 1985). In cases where gametophytic expression of the target EMB gene is required for normal pollen-tube growth, aborted seeds are rarely found at the base of heterozygous siliques because mutant pollen tubes are at a competitive growth disadvantage. Results of segregation tests involving bio2 and bio1 heterozygotes are summarized in Table II. The deletion associated with the bio2 mutation does not appear to interfere with pollen development or pollen-tube growth because mutant seeds were distributed randomly along the length of heterozygous siliques and segregation ratios were not significantly different from those expected for a single recessive mutation. Pollen grains obtained from selfed bio2 heterozygotes also appeared normal. Thus, even the complete loss of biotin synthesis in mutant pollen grains does not appear to interfere with normal pollen-tube growth. Furthermore, the deletion appears to cover a small segment of the genome because heterozygotes do not exhibit the pollen abortion characteristic of large deletions and chromosome aberrations.

Table II.

Distribution of aborted seeds in siliques from bio1 and bio2 heterozygotes

Result	bio1	bio2
Total siliques screened	40	40
Total no. of seeds	2086	2149
Percentage mutant seedsa	25.6	24.2
Percentage in top halfb	48.8	52.3

^aSegregation ratio of aborted seeds. Results are not significantly different (at P = 0.05) from the expected 25% mutant seeds. 

^bPercentage of total mutant seeds located in the top half of heterozygous siliques. Results are not significantly different (at P = 0.05) from the expected random distribution of mutant seeds (50% top half). 

Mutant Plants Do Not Flower in the Presence of Biotin

Mutant plants rescued in the presence of biotin exhibited one additional defect that could not be separated through recombination. This defect became apparent as large numbers of rescued seedlings were grown in pots supplemented with biotin. Because mutant plants were expected to appear normal in the presence of biotin, a PCR assay was devised to identify rescued mutants based on the absence of BIO2 sequences in bio2/bio2 plants. Leaf samples from segregating populations of bio2/bio2, bio2/BIO2, and BIO2/BIO2 plants grown in the presence of biotin were analyzed with PCR primers designed to detect the presence of the BIO2 gene. Plants that failed to amplify the expected BIO2 band in duplicate tests were scored as bio2/bio2 homozygotes. Results of this experiment are summarized in Table III. Plants were grown from seeds that were germinated either directly on soil or first in culture and then transplanted to pots supplemented with biotin.

Table III.

Co-segregation of the bio2 mutant allele and nonbolting phenotype

Genotype	Phenotype	Plants
		Soila	Platesb	Total
		no.
BIO2/-	Bolting	62	110	172
BIO2/-	Nonbolting	0	0	0
bio2/bio2	Bolting	0	0	0
bio2/bio2	Nonbolting	6	19	25

Mature seeds from rescued heterozygotes were germinated in the presence of biotin, and rescued homozygous mutant plants were identified by PCR as described in the text. The nonbolting phenotype was scored at maturity.

^aSeeds were planted directly on soil. 

^bSeeds were germinated in culture and the resulting seedlings were transplanted to soil. 

All rescued bio2 homozygotes identified by PCR failed to bolt after extended growth under either long days or continuous light. Conversely, all plants that flowered contained at least one copy of the BIO2 gene. Mutant rosettes produced dark green leaves that were slightly reduced in size but otherwise appeared normal. The low ratio of mutant plants identified in these populations (12.7 instead of 25.0%) probably resulted from experimental selection against defective seeds and weak seedlings that corresponded to mutant seeds incompletely rescued by the biotin provided to parent plants. The failure of mutant plants to bolt was likely not caused by inadequate transport of biotin to the shoot apex because rescued bio1 homozygotes flower without difficulty in the presence of biotin. Rescued bio2 heterozygotes watered with biotin also produce nearly 100% phenotypically normal seeds. A more likely explanation is that a second gene covered by the deletion is required for bolting. Since none of the known late-flowering mutants has been mapped to this precise region of the genome, despite extensive screens in several different laboratories, the loss of more than one gene adjacent to the BIO2 locus may be required to produce the nonbolting phenotype.

DISCUSSION

Embryo-defective mutants of Arabidopsis represent a valuable source of genes with essential functions during plant growth and development (Meinke, 1995). More than 500 mutants disrupted in embryogenesis have been identified following chemical and insertional mutagenesis (Jur-gens et al., 1994; Meinke, 1994; Yadegari et al., 1994; Devic et al., 1996; Lindsey et al., 1996). Several mutants defective in known cellular functions have been examined in detail (Shevell et al., 1994; Springer et al., 1995; Traas et al., 1995; Lukowitz et al., 1996; Tsugeki et al., 1996). The molecular basis of abnormal development in many other mutants remains to be determined. The present study was designed to expand the search for auxotrophs among existing collections of emb mutants. The assumption as outlined by Langridge (1958) and confirmed by Schneider et al. (1989) was that some auxotrophs escape detection at the seedling stage because they arrest early in development. The discovery reported here of another emb mutant altered in biotin synthesis, when combined with previous work on two different bio1 mutant alleles, raises a fundamental question concerning the nature of plant auxotrophs in relation to plant embryogenesis. Although screens for auxotrophic emb mutants have not yet approached saturation, the prevalence of biotin auxotrophs and the absence of other auxotrophs identified among more than 60 emb mutants examined to date is intriguing. What is peculiar about the biotin pathway in Arabidopsis that has allowed these auxotrophs to be identified at a moderate frequency among embryo-defective mutants, while auxotrophs disrupted in other biosynthetic pathways continue to escape detection?

Isolation of the bio1 auxotroph initially demonstrated that biotin serves an essential function during plant growth and development (Schneider et al., 1989). Analysis of this mutant allowed the disrupted step to be identified and suggested that the biotin pathway might be conserved between plants and bacteria (Shellhammer and Meinke, 1990; Patton et al., 1996b). Further interpretation of the mutant phenotype was limited by the availability of a single mutant allele of unknown strength. This made it difficult to determine whether mutant embryos survived to the cotyledon stage because they obtained biotin from surrounding maternal tissues, retained some activity of the altered protein, or expressed a duplicate gene or biosynthetic pathway. This has been a common problem when trying to predict which auxotrophs should arrest during embryogenesis and which might survive to the seedling stage. Analysis of the bio2 deletion mutant has finally resolved this question with respect to biotin by establishing the null phenotype and demonstrating the consequence of a complete loss of biotin synthesis during embryo development. Mutant bio2 embryos typically arrest at the globular stage, apparently because maternal tissues cannot supply enough biotin to support the rapid cell division and increased lipid biosynthesis associated with later stages of development. Further growth of bio1–1 embryos probably results from residual activity of a hypomorphic allele. The identification of a second bio1 allele that arrests slightly earlier in embryogenesis (R. Fischer and N. Ohad, University of California, Berkley, personal communication) is consistent with this model. The somewhat variable phenotype of bio2 embryos most likely reflects minor differences in maternal biotin available from heterozygous siliques produced under different conditions.

We propose that biotin auxotrophs have been easy to identify among embryo-defective mutants because the nutrient deficiency cannot be corrected by duplicate genes or maternal sources of biotin and because mutant embryos arrest at a stage of development when they can be effectively rescued in culture. This model predicts that auxotrophs defective in other steps of the biotin pathway should also result in embryonic lethality, provided the corresponding genes are not duplicated, and that mutant embryos should arrest between the globular and cotyledon stages. It should therefore be possible to complete a genetic dissection of the entire biotin pathway in plants by analyzing collections of embryo-defective mutants. The bio2 mutant should be particularly useful in future attempts to identify additional cofactors required for biotin synthase activity (Birch et al., 1995; Sanyal et al., 1996), resolve the significance of the putative intermediate 9-mercaptodethiobiotin in biotin synthesis (Baxter et al., 1992; Baldet et al., 1993b), and determine whether ecotypes of Arabidopsis reported to have temperature-sensitive growth defects that can be rescued with supplemental biotin (Langridge and Griffing, 1959; Langridge, 1965) contain deleterious mutations within the BIO2 gene.

Several complementary approaches have been used in the past to identify mutants defective in amino acid, vitamin, and nucleoside biosynthesis in higher plants (Schneider et al., 1989). The initial approach was to screen populations of M₂ seedlings for growth defects that could be reversed with supplemental nutrients. In Arabidopsis, this strategy led almost exclusively to the isolation of thiamine auxotrophs (Li and Redei, 1969; Redei and Acedo, 1976), several of which have subsequently been examined in detail (Komeda et al., 1988; Ribeiro et al., 1996). Attention then shifted to the isolation of auxotrophic cell lines in culture, using strategies employed successfully with microorganisms (Blonstein, 1986). These efforts resulted in the isolation of a number of auxotrophs, particularly among members of the Solanaceae (Negrutiu et al., 1992), but plant regeneration and transmission of the phenotype through successive generations often proved to be difficult. A third approach has been to apply microbial selection strategies to M₂ seedlings of Arabidopsis. This approach has been particularly useful for studying Trp biosynthesis (Rose and Last, 1994). A common assumption when auxotrophs have not been recovered at the seedling stage has been that the desired mutants arrested early in development. Results presented here challenge this assumption because they raise the possibility that biotin mutants may be the only auxotrophs that can be readily identified among embryo defectives of Arabidopsis.

Several alternative models for this apparent scarcity of plant auxotrophs must be considered. One model postulates that additional auxotrophs are present among embryo-defective mutants, but maternal supplies of the missing nutrient are quickly depleted and the embryo arrests very early in development. Such auxotrophs could have escaped detection in previous screens because the small mutant seeds are difficult to manipulate in culture. Arrested embryos lacking a nutrient with many essential functions in growth and development might also be more difficult to rescue and identify in culture than arrested embryos from biotin auxotrophs, which should be disrupted only in a small number of biotin-dependent enzymes. An alternative model, which includes the missing auxotrophs among gametophytic lethals, cannot be eliminated based on available evidence but may be difficult to test without exposing large numbers of flowers to supplemental nutrients. In some cases, supplements containing the required nutrient may still fail to rescue the corresponding mutant because transport and utilization of the essential component are disrupted. The most probable explanation for the scarcity of plant auxotrophs appears to be the widespread existence of multigene families. Even in plants with small genomes, many proteins with essential cellular and developmental functions are encoded by duplicated genes (McGrath et al., 1993; Pickett and Meeks-Wagner, 1995). Results of mapping large numbers of EMB genes in Arabidopsis suggest that the number of target genes that can mutate to give an embryo-defective phenotype is far less than the total number of genes expressed at this stage of the life cycle (Franzmann et al., 1995). Molecular data supporting the concept of gene redundancy have also been obtained for a number of biosynthetic pathways (Rose and Last, 1994; Tada et al., 1994; Bogdanova et al., 1995). Additional examples of gene duplications in amino acid and vitamin biosynthesis are likely to be uncovered with continued advances in the Arabidopsis genome project. Molecular genetic dissection of biosynthetic pathways in plants may therefore need to rely increasingly on the use of reverse genetics (McKinney et al., 1995), gene trapping with reporter fusions (Sundaresan et al., 1995), and homology searches of genome databases (Fleischmann et al., 1995) to identify desired target genes, analyze potential chemical intermediates, and characterize the phenotypic consequences of loss-of-function mutations.

Abbreviations:

cM

centiMorgan(s)

EMS

ethyl methanesulfonate

RI

recombinant inbred

SSLP

simple sequence-length polymorphism