Although soluble sugar levels affect many aspects of plant development and physiology, little is known about the mechanisms by which plants respond to sugar. Here we report the isolation of 13 sugar-insensitive (sis) mutants of Arabidopsis that, unlike wild-type plants, are able to form expanded cotyledons and true leaves when germinated on media containing high concentrations of glucose or sucrose. The sis4 and sis5 mutants are allelic to the ABA-biosynthesis mutant aba2 and the ABA-insensitive mutant abi4, respectively. In addition to being insensitive to glucose and sucrose, the sis4/aba2 and sis5/abi4 mutants also display decreased sensitivity to the inhibitory effects of mannose on early seedling development. Mutations in the ABI5 gene, but not mutations in the ABI1, ABI2 or ABI3 genes, also lead to weak glucose- and mannose-insensitive phenotypes. Wild-type and mutant plants show similar responses to the effects of exogenous sugar on chlorophyll and anthocyanin accumulation, indicating that the mutants are not defective in all sugar responses. These results indicate that defects in ABA metabolism and some, but not all, defects in ABA response can also alter response to exogenous sugar.
In recent years, genetic approaches to identify the factors involved in controlling plant sugar responses have been initiated. One approach has been to screen for Arabidopsis mutants with altered expression of a particular sugar-regulated gene. This approach has led to the identification of mutants that are defective in sugar-regulated expression of the genes encoding patatin ( Martin et al. 1997 ), beta-amylase ( Mita et al. 1997a,b ) and plastocyanin ( Dijkwel et al. 1997 ). A second type of screen for sugar response mutants is based on the finding that high concentrations of exogenous sugars inhibit seed germination and early seedling development of Arabidopsis ( Gibson et al. 1996 ). Nineteen sugar-insensitive (sis) mutants have been isolated by screening mutagenized Arabidopsis for the ability to develop relatively normal shoot systems on high concentrations of exogenous glucose or sucrose ( Gibson et al. 1996 ). Similar screens have been used to identify glucose-insensitive (gin) mutants ( Zhou et al. 1998 ), carbohydrate-insensitive (cai) mutants ( Boxall et al. 1996 ) and sucrose-insensitive growth (sig) mutants ( Kortstee et al. 1998 ). A pleiotropic regulatory locus (prl1) mutant, with increased sensitivity not only to sugar, but also to ethylene, abscisic acid (ABA), auxin, cytokinin and cold stress, has also been isolated ( Németh et al. 1998 ; Salchert et al. 1998 ). The PRL1 gene encodes a protein that interacts with SNF1-like proteins in a yeast two-hybrid screen ( Bhalerao et al. 1999 ). The fact that PRL1 interacts with SNF1 is interesting because SNF1-like protein kinases are thought to play an important role in plant sugar response pathways ( Halford & Hardie, 1998). Transgenic Arabidopsis that over-express the Arabidopsis hexokinase genes AtHXK1 or AtHXK2 show increased sensitivity to sugar, whereas plants expressing hexokinase in an antisense orientation show a reduced sensitivity to sugar ( Jang et al. 1997 ).
Here we report the identification and characterization of the sis4 and sis5 mutants, which are allelic to the aba2 and abi4 mutants, respectively. Mutations in the ABI5 gene also lead to a weak sis phenotype, whereas mutations in the ABI1, ABI2 and ABI3 genes have little to no effect on sugar sensitivity. While the extent to which the connection between ABA and sugar response is direct or indirect cannot be determined at this time, these results indicate that the connection does exhibit a substantial degree of specificity, as mutations in only specific components of the ABA response pathway(s) confer a sis phenotype.
High sugar concentrations inhibit seedling development
The majority of seeds grown on minimal Arabidopsis media ( Kranz & Kirchheim, 1987) containing 0.3 m sucrose ( Fig. 1) or 0.3 m glucose (data not shown) germinate but fail to develop true leaves and have purple/white cotyledons that show little expansion. In contrast, most of these seedlings develop significant root systems, indicating that roots are less susceptible than shoots to the inhibitory effects of high sugar concentrations. In comparison, seedlings grown on low, non-inhibitory (e.g. 0.028 m) concentrations of glucose ( Fig. 1) or sucrose (data not shown) develop both extensive shoot and root systems. The effects of 0.3 m glucose or sucrose on seedling development do not appear to result solely from the high osmotic potential of the media, as wild-type seedlings grown on 0.3 m sorbitol develop true leaves and expanded cotyledons ( Fig. 1). In addition, approximately 0.4 m concentrations of sorbitol are required to exert the same inhibitory effect as 0.3 m glucose or sucrose ( Fig. 2). As equi-molar concentrations of different chemicals may exert different degrees of osmotic stress, the osmolality of different sucrose, glucose and sorbitol solutions was determined. Osmolality provides a measure of the effect of a particular solute on the ordering of water molecules. These experiments indicate that 0.3 m solutions of sucrose, glucose or sorbitol all have values of between 300 and 340 mOsm kg−1, while that for 0.4 m sorbitol is about 420 mOsm kg−1. Therefore, 0.4 m sorbitol may be expected to exert more osmotic stress than, for instance, 0.3 m glucose.
Isolation of sugar-insensitive (sis) mutants
The finding that glucose and sucrose can act as negative regulators of seedling development made possible an efficient screen for identifying mutants defective in sugar sensing or response. In brief, approximately 60 000 M2 seeds derived from ethylmethane sulphonate-mutagenized Arabidopsis var. Columbia were sown on solid minimal media containing 0.3 m sucrose. After 12–14 days, seedlings that had undergone cotyledon expansion and the formation of true leaves were transferred to soil, grown to maturity and re-assayed in the next generation. Thirteen sis mutants were isolated during the course of this screen. Here we describe the characterization of the nine mutants that comprise the sis4 and sis5 complementation groups. The sis4 and sis5 mutations are recessive, and each segregates as a single locus. The sis4 complementation group consists of two mutants, sis4-1 and sis4-2, that arose from independent mutagenic events and are likely to represent distinct alleles. The sis5 complementation group consists of seven mutants. Sequencing of the defective gene from these mutants (described in the following section) has revealed that they define four alleles, which have been designated sis5-1 to sis5-4.
Unlike wild-type plants, sis4 and sis5 mutants are able to form expanded cotyledons and true leaves when grown on media containing high (0.3–0.33 m) concentrations of sucrose ( Figs 1 and 2). The sis4 and sis5 mutants are also less sensitive to the inhibitory effects of high (0.3 m) glucose concentrations ( Fig. 2). The sis4 and sis5 mutants were also assayed for tolerance to high concentrations of sorbitol. When assayed on medium supplemented with 0.272 m sorbitol plus 0.028 m glucose (for a total supplement concentration of 0.3 m), differences between sis4, sis5 and wild-type plants were difficult to detect, largely due to the fact that development of wild-type plants is only slightly affected at this sorbitol concentration ( Fig. 2). Therefore, wild-type and mutant plants were assayed on media supplemented with 0.4 m sorbitol plus 0.028 m glucose. When grown on this media, wild-type plants show a significantly decreased ability to form expanded cotyledons and true leaves. In contrast, the sis4 and sis5 mutants are relatively insensitive to the inhibitory effects of this sorbitol concentration ( Fig. 2), demonstrating that these mutants have a significant osmo-tolerant phenotype. The basis for the osmo-tolerant phenotype of the sis4 and sis5 mutants is currently not understood. One possibility is that their sugar-insensitive phenotype might cause them to accumulate higher levels of intracellular sugars than wild-type plants, which could protect them against osmotic stress.
The mutants were also tested for sensitivity to mannose, a glucose analogue. Even low concentrations of mannose have been postulated to affect the expression of several plant genes ( Graham et al. 1994 ; Sheen, 1994) and to inhibit seed germination and early seedling development ( Pego et al. 1999 ) through a hexokinase-mediated pathway. The sis4 mutants show very slight decreases in mannose sensitivity. In contrast, three of the sis5 mutants (sis5-1, sis5-3 and sis5-4) exhibit large decreases in mannose sensitivity and the sis5-2 mutant shows a slight decrease in mannose sensitivity ( Fig. 2). As the concentration of mannose used in these experiments was quite low (1.7 m m), the mannose-insensitive phenotypes of the mutants are unlikely to be due to their osmo-tolerant phenotypes.
Sis4 and sis5 mutants are allelic to aba2 and abi4
Low concentrations of exogenous sugars allow wild-type Arabidopsis seeds to germinate on media containing ABA ( Finkelstein & Lynch, 2000; Garciarrubio et al. 1997 ). As the sis mutations cause decreased sensitivity to the inhibitory effects of sugars on early seedling development, it was of interest to determine whether the sis mutations also affect the ability of sugars to promote germination in the presence of ABA. Towards this end, sis mutant and wild-type seeds were sown on ABA-containing media, in the presence and absence of 0.044 m sucrose. While wild-type, sis4 and sis5 seeds all germinate at high frequency on media containing both ABA and sucrose (data not shown), sis5 mutant seeds also germinate on media containing ABA without exogenous sugar ( Table 1). Therefore, the sis5 mutants have an ABA-insensitive (abi) germination phenotype. This finding raised the possibility that the sis4 and sis5 mutants might exhibit other phenotypes associated with defects in ABA metabolism or response. For instance, ABA biosynthesis (aba), abi1 and abi3, but not abi2, mutants are able to germinate in the presence of gibberellin biosynthesis inhibitors, such as paclobutrazol and uniconazol ( Koornneef et al. 1982 ; Léon-Kloosterziel et al. 1996 ; Nambara et al. 1991 ). In addition, aba, abi1 and abi2, but not abi3, abi4 or abi5 mutants have wilty phenotypes ( Finkelstein, 1994; Koornneef et al. 1982 ; Koornneef et al. 1984 ; Léon-Kloosterziel et al. 1996 ). As shown in Table 1, sis4 and sis5 seeds exhibit decreased sensitivity to paclobutrazol. The sis4 mutants, but not the sis5 mutants, also have a wilty phenotype (data not shown). These experiments indicate that the sis4 mutants are phenotypically similar to aba mutants, whereas the sis5 mutants are phenotypically similar to abi3, abi4 and abi5 mutants.
Table 1. Germination on ABA and paclobutrazol
Percentage germination on
To aid in identifying the SIS4 and SIS5 genes, the chromosomal locations of the sis4 and sis5 mutations were determined. The results of these experiments indicate that the sis4 locus maps approximately 8 cM from the SSLP marker nga280 ( Bell & Ecker, 1994), near the position of a previously identified ABA biosynthesis mutant, aba2 (Léon-Kloosterziel et al. 1996). The sis5 locus maps approximately 4 cM from the SSLP marker nga168 ( Bell & Ecker, 1994), near the position of two previously identified ABA-insensitive mutants, abi4 and abi5 ( Finkelstein, 1994). To determine whether the sis4 mutants are allelic to aba2, complementation experiments were performed. The sis4-2 mutant was crossed to the sis4-1 and aba2-1 mutants. As shown in Table 2, the resulting F1 plants were able to form shoot systems on high concentrations of exogenous sugar, indicating that the sis4-2 mutation is unable to complement either the sis4-1 or aba2-1 mutations. These experiments indicate that sis4-1 and sis4-2 are allelic to each other as well as to aba2-1. Therefore, sis4-1 and sis4-2 represent new alleles of the aba2 locus, and have been renamed aba2-3 and aba2-4, respectively. As the sis5 locus maps near both the abi4 and abi5 loci, it was of interest to determine whether the sis5 mutations lie in the ABI4 or ABI5 genes. As the ABI4 gene had been cloned and sequenced prior to the onset of this work ( Finkelstein et al. 1998 ), an efficient means of determining whether the sis5 mutations lie in the ABI4 gene was to isolate and sequence the ABI4 gene from each of the sis5 mutants. These experiments indicate that all seven of the sis5 mutants carry defective ABI4 genes that, together, define four new abi4 alleles ( Table 3). Therefore, the sis5 mutants have been renamed abi4-101 to abi4-104.
Table 2. aba2 complementation analysis
Percentage normal shoot development
aArabidopsis minimal medium with 0.3 m glucose; bArabidopsis minimal medium with 0.3 m sucrose.
Aba and abi mutants exhibit varying degrees of sugar insensitivity
The finding that the sis4 and sis5 mutants are allelic to aba2 and abi4, respectively, raised the possibility that other aba and abi mutants may have sis phenotypes. Three aba loci have been characterized ( Koornneef et al. 1982 ; Léon-Kloosterziel et al. 1996 ). When grown on media containing 0.27 m glucose, 48% of aba1-1 seedlings and 99% of aba3-2 seedlings, but only 2% of wild-type Ler-0 seedlings, formed expanded cotyledons. These results indicate that the aba1-1 and aba3-2 mutants have a sis phenotype. To determine whether mutations in any of the five known abi loci ( Finkelstein, 1994; Koornneef et al. 1984 ) confer a sis phenotype, seeds from mutant and wild-type plants were sown on media containing high concentrations of glucose. Slightly different concentrations (0.27 or 0.3 m) of glucose were used in these experiments because plants of the Columbia (Col) ecotype are more resistant to high sugar concentrations than plants of the Wassilewskija (Ws) or Landsberg erecta (Ler-0) ecotypes (Laby and Gibson, unpublished results). These experiments indicate that the abi4-1 mutant is very resistant, while the abi5-1 mutant is slightly resistant, to the effects of high concentrations of glucose. In contrast, the abi1-1, abi2-1 and abi3-1 mutations confer little to no resistance to high concentrations of glucose ( Fig. 3). The same abi mutants were also examined for a mannose-insensitive phenotype. These experiments indicate that the abi1-1, abi2-1 and abi3-1 mutations confer little to no resistance to mannose, while the abi4-1 and abi5-1 mutations confer slight or moderate increases in mannose resistance ( Fig. 3).
Sis mutants accumulate wild-type chlorophyll and anthocyanin levels
Exogenous sugars have been shown to cause chlorophyll levels to decrease ( Krapp et al. 1991 ) and anthocyanin levels to increase ( Tsukaya et al. 1991 ). In addition, several mutants thought to be defective in sugar sensing/response are defective in chlorophyll and/or anthocyanin accumulation ( Dai et al. 1999 ; Dijkwel et al. 1997 ; Jang et al. 1997 ; Mita et al. 1997a,b ; Németh et al. 1998 ; Zhou et al. 1998 ). Therefore, chlorophyll levels were measured in wild-type, abi4 and aba2 seedlings grown in the presence of 0.03 m or 0.15 m sucrose. As shown in Fig. 4(a), the abi4 and aba2 mutants show wild-type chlorophyll accumulation in response to sucrose. All the plant lines tested, including the wild-type plants, have higher chlorophyll concentrations on 0.15 m sucrose than on 0.03 m sucrose. Results from other groups found that chlorophyll levels decreased with increasing sugar concentration ( Krapp et al. 1991 ). Possible explanations for the different effect of sugar on chlorophyll accumulation observed in this study include differences in the ages of the plants being studied (relatively young plants were analysed in this study), in the method of supplying sugar to the plants (through the roots versus through cut petioles or stems), in the sugar concentrations and exposure times used, and in the species being examined. The fact that young seedlings were assayed may be particularly relevant, as young seedlings represent a large metabolic sink and may be able to effectively metabolize even 0.15 m sucrose, so that the intracellular sugar concentration of seedlings growing on 0.15 m sucrose may be lower than that of older plants growing on the same medium. As shown in Fig. 4(b), the abi4 and aba2 mutants tested also have wild-type responses to the effects of sucrose on anthocyanin accumulation. These findings are significant because they indicate that the abi4 and aba2 mutations affect only a subset of sugar responses and are therefore unlikely to affect sugar responses via an effect on sugar transport.
Characterization of the sis4 and sis5 mutants has shown that they are allelic to the ABA biosynthesis mutant aba2 and the ABA-insensitive mutant abi4, respectively. These results indicate that altered responses to sugar can arise from mutations that affect ABA response or synthesis. These results also suggest that some of the sugar response mutants isolated by other research groups, such as some of the gin ( Zhou et al. 1998 ), cai ( Boxall et al. 1996 ) or sig ( Kortstee et al. 1998 ) mutants, may be defective in ABA response or metabolism, although there have been no published reports to that effect. Precedence for a relationship between phytohormone and sugar responses comes from several sources. For example, the gin1 mutant of Arabidopsis, which displays a glucose-insensitive phenotype, is also defective in its response to ethylene ( Zhou et al. 1998 ). Similarly, the prl1 mutant of Arabidopsis, which shows increased sensitivity to the effects of exogenous sugars, exhibits increased sensitivity to ethylene, ABA, cytokinin and auxin ( Németh et al. 1998 ; Salchert et al. 1998 ). Interestingly, low concentrations of exogenous sugars have also been reported to allow germination of wild-type seeds on media containing ABA. However, whether this effect is due to sugar acting as a nutrient or as a signalling molecule remains unclear ( Finkelstein & Lynch, 2000; Garciarrubio et al. 1997 ). While these results and those of other studies indicate the existence of connections between phytohormone responses and sugar responses, how direct these connections are remains to be determined ( Gibson & Graham, 1999). Further characterization of the sis mutants and other mutants with defective responses to sugar and to phytohormones should help to address these issues.
Characterization of mutants carrying defects in all the known ABI genes revealed that abi4 mutations confer significant resistance to the inhibitory effects of high concentrations of exogenous glucose, sucrose and mannose on cotyledon expansion and formation of true leaves. In contrast, the abi5-1 mutation has very marginal effects on glucose and mannose resistance, while the abi1-1, abi2-1 and abi3-1 mutations confer little to no glucose or mannose resistance. These differences in glucose sensitivity are unlikely simply to reflect differences in the severity of the mutant alleles tested, as the abi1-1, abi2-1 and abi3-1 mutations confer greater resistance to ABA than the abi4-1 mutation ( Finkelstein, 1994). In addition, the fact that all seven of the ABA-insensitive sis mutants analysed during this study contain mutations in the ABI4 gene, together with the recent finding that two sugar response mutants (sun6-1 and sun6-2) isolated by an independent group ( van Oosten et al. 1997 ) also contain mutations in the ABI4 gene ( Huijser et al. 2000 ), suggests that mutations in the ABI4 gene are particularly likely to cause defects in sugar response.
The ABI4 gene encodes a protein with significant sequence similarity to AP2 domain-containing transcriptional regulators ( Finkelstein et al. 1998 ). Previous phenotypic analyses of abi mutants have suggested that ABI1 and ABI2 function predominantly during vegetative growth, while ABI3, ABI4 and ABI5 are primarily involved in seed maturation. For instance, abi1 and abi2 mutants, but not abi3, abi4 or abi5 mutants, display wilty phenotypes because they are deficient in the ability to close their stomata in response to water stress ( Finkelstein, 1994; Koornneef et al. 1984 ). Conversely, abi3, abi4 and abi5 mutants, but not abi1 or abi2 mutants, are defective in accumulation of late-embryogenesis-abundant genes during seed development ( Finkelstein, 1993; Finkelstein, 1994; Finkelstein & Somerville, 1990). Double mutant analyses suggest that plants contain at least two pathways for ABA response, with ABI1 and ABI2 acting in one pathway and ABI3, ABI4 and ABI5 acting in a second signal transduction pathway ( Finkelstein, 1994; Finkelstein & Somerville, 1990). The results of this study show a significant difference between the abi3-1, abi4 and abi5-1 mutants in their response to glucose and mannose. A possible explanation for this difference is that ABI4 might act upstream of ABI3 and ABI5, and so might affect a greater number of processes. Alternatively, ABI4 might define a branch of an ABA response pathway that is involved in sugar response.
The fact that abi4 mutations confer a strong sis phenotype, while abi5 mutations confer only a marginal sis phenotype and abi1, abi2 and abi3 mutations confer little to no sis phenotype, suggests that the connection between sugar and ABA response is relatively specific. This finding argues against the possibility that defects in ABA response or synthesis might have general effects on physiology that could affect sugar response. For instance, the fact that ABA is synthesized upon seed imbibition ( Debeaujon & Koornneef, 2000) means that mutations in ABA synthesis or response are likely to have a broad effect on the physiology and progression of seed germination and early seedling development. In theory, any mutation that alters ABA response or synthesis might then also alter plant response to sugar via an indirect effect on seedling physiology or metabolism rather than via a direct effect on sugar sensing or signalling. However, the fact that only specific abi mutants are sugar-insensitive argues against, but does not rule out, this possibility. While the connection between ABA and sugar response does appear to exhibit a significant degree of specificity, determination of how direct or indirect the connection is between ABA and sugar response pathways will probably require significantly greater knowledge of both response pathways than is currently available.
While the molecular mechanism(s) by which abi4 mutations affect sugar responses are currently unclear, the available evidence indicates that the osmo-tolerant phenotype of the mutants is not sufficient to explain their sugar-insensitive phenotype. First, not all mutations that lead to an osmo-tolerant phenotype also cause a sugar-insensitive phenotype. For instance, 40% of abi2-1 seeds, but less than 1% of Ler-0 wild-type seeds, form seedlings with expanded cotyledons when grown on media containing 0.4 m sorbitol (Gibson, unpublished results). These results demonstrate that the abi2-1 mutant has a strong osmo-tolerant phenotype. However, results presented here show that the abi2-1 mutant does not have a sugar-insensitive phenotype. Therefore, osmo- and sugar-tolerance are genetically separable. In addition, the abi4 mutants are insensitive to the inhibitory effects of concentrations of exogenous mannose that are too low (typically 1.5–3 m m) to exert an osmotic stress. Therefore, the mannose-insensitive phenotype of the mutants cannot be explained by their osmo-tolerant phenotype.
The basis for the mannose-insensitive phenotype of the mutants remains to be determined. Mannose is a glucose analogue that, like glucose, has been postulated to inhibit seed germination through a hexokinase-mediated signal transduction pathway. This model is based on the finding that mannoheptulose allows wild-type seeds to germinate on otherwise inhibitory concentrations of mannose ( Pego et al. 1999 ). As mannoheptulose is an inhibitor of hexokinases ( Coore & Randle, 1963), this finding indicates that the inhibitory effects of mannose on seed germination require hexokinase activity. Hexokinases have been implicated in plant sugar responses ( Dai et al. 1999 ; Graham et al. 1994 ; Jang & Sheen, 1994; Jang et al. 1997 ), although the exact role of hexokinases in sugar responses remains controversial ( Halford et al. 1999 ). Therefore, one possibility is that mannose is very effective in promoting hexokinase-mediated sugar responses, so that even low concentrations of mannose may cause an inhibitory effect by inappropriate and/or excessive stimulation of sugar response pathway(s). If this model is correct, than abi4 mutations may alleviate mannose inhibition by decreasing plant responses to sugar. However, alternative possibilities cannot be ruled out at this time. For instance, the mannose-6-phosphate generated by hexokinase-mediated phosphorylation of mannose may either be inhibitory or may be metabolized to form an inhibitory compound. If this model is correct, then abi4 mutations may alleviate mannose inhibition by decreasing plant sensitivity to mannose-6-phosphate or to a metabolic product of mannose-6-phosphate.
The effects of abi4 and aba mutations on sugar response analysed as part of this work are limited to seed germination and early seedling development. Several possible explanations exist for the lack of known effects of abi4 and aba mutations on sugar responses later in plant development. One trivial possibility is that such effects do exist but have not yet been identified. Alternatively, the action of ABA on later sugar responses might be via a redundant pathway, so that the effects of abi4 and aba mutations on those responses are masked. Another possibility is that the effects of abi4 and aba mutations on sugar response are limited to seed germination and early seedling development. In this regard, it is of interest to note that ABI4 is thought to be involved primarily in seed maturation ( Finkelstein, 1994). Further characterization of abi4 and aba mutants, particularly at later developmental stages, will be required to distinguish between the above possibilities.
Plant material, growth conditions and media
The aba1-1, aba2-1, aba3-2, abi2-1, abi3-1, abi4-1 and abi5-1 mutants of Arabidopsis thaliana were obtained from the Arabidopsis Biological Resource Center at Ohio State University. The abi1-1 mutant was obtained from Dr Ruth Finkelstein (University of California at Santa Barbara, CA, USA). The M2 seeds from which the abi4-101, abi4-103 and abi4-104 mutants (formerly sis5-1, sis5-3 and sis5-4, respectively) were isolated were obtained from Lehle Seeds (Tucson, AZ, USA). These M2 seeds were harvested from M1 plants generated by EMS-induced mutagenesis of Columbia seeds. The M2 seeds from which the abi4-102, aba2-3 and aba2-4 mutants (formerly sis5-2, sis4-1 and sis4-2, respectively) were isolated, as well as the Columbia (Col), Landsberg erecta (Ler-0) and Wassilewskija (Ws) lines, were obtained from Dr Chris Somerville (Carnegie Institute, Palo Alto, CA, USA). These M2 seeds were harvested from M1 plants generated by EMS-induced mutagenesis of Columbia seeds. The sis5-1/abi4-101, sis5-3/abi4-103 and sis5-4/abi4-104 mutants carry the gl1-1 mutation. Unless otherwise specified, plants were grown under continuous cool white fluorescent light at approximately 21°C. Minimal Arabidopsis media was prepared as described by Kranz & Kirchheim 1987) .
Sugar-insensitive mutant screen and assays
To identify sugar-insensitive mutants, M2 seeds derived from EMS-mutagenized populations were sown on Petri plates with solid minimal Arabidopsis media ( Kranz & Kirchheim, 1987) supplemented with 0.3 m sucrose. Approximately 1000 seeds were sown per 100 mm Petri plate. The plates were incubated at 21°C under approximately 80–120 μE m−2 sec−1 continuous fluorescent light. After 2 weeks, seedlings that had formed relatively normal shoot systems (seedlings that had expanded cotyledons and true leaves) were transplanted to soil and grown to maturity. Seeds were harvested from these putative mutants and re-screened to distinguish between wild-type plants that had escaped the selection in the first round of screening and true mutants. To assay sugar sensitivity, seeds were surface-sterilized and sown on Petri plates with solid minimal Arabidopsis media containing the indicated additives and grown under 80–120 μE m−2 sec−1 continuous fluorescent light at 21°C for the time periods indicated (usually 2 weeks). Typically, 50–100 seeds were scored per assay. Seeds/seedlings were scored for seed germination, cotyledon expansion and true leaf formation. Seed germination is defined as the emergence of any part of the seedling from the seed coat. Cotyledon expansion is defined as the development of cotyledons that are significantly greater in size than those found on seedlings that have just emerged from the seed coat. True leaf formation is defined as the formation of true leaves that are readily visible without magnification.
Aliquots (50 μl) of the indicated solutions were assayed using a Micro Osmometer Model 5004 from Precision Systems Inc. (Sudbury, MA, USA), according to the manufacturer's directions.
Abscisic acid and paclobutrazol sensitivity assays
To measure ABA sensitivity, seeds were plated on minimal media with 5 μm ABA. After 10 days at 21°C under 90–110 μE m−2 sec−1 continuous light, the percentage of seeds that had germinated (defined as the emergence of any part of the seedling from the seed coat) was determined. To measure paclobutrazol sensitivity, seeds were plated on minimal medium with 35 mg l−1 paclobutrazol. The plates were incubated in the dark at 4°C for 3 days and then at 21°C and 90–100 μE m−2 sec−1 continuous light for an additional 3 days prior to determining the percentage of seeds that had germinated.
Identifying ABI4 mutations
Oligonucleotides complementary to the ABI4 gene were used to amplify DNA from the abi4 mutant lines via PCR. The resulting DNA fragments were separated on an agarose gel and purified using the Qiaex 2 Gel Purification Kit (Qiagen Inc., Santa Clarita, CA, USA) prior to being submitted to a commercial facility for DNA sequencing (Lone Star Labs Inc., Houston, TX, USA).
Chlorophyll and anthocyanin assays
To measure chlorophyll levels, seeds were surface-sterilized and sown on solid minimal media supplemented with either 0.03 m or 0.15 m sucrose. After 2 weeks of growth under 60–80 μE m−2 sec−1 continuous light, total chlorophyll levels were measured as described by Wintermans & de Mots, (1965). In brief, total shoot systems from three plants per assay were weighed and then extracted several times with 96% ethanol at 80°C. The OD654 values for the combined extracts from each sample were measured and chlorophyll levels determined using the equation: OD654× extraction volume (ml) × 1/39.8 × 1/weight of tissue sample (g) = mg chlorophyll/g fresh weight of tissue. To measure anthocyanin levels, seeds were surface-sterilized and then plated on solid minimal medium with 0.03 m sucrose. After 2 weeks, the seedlings were transferred to solid minimal media with 0.03 m or 0.18 m sucrose and grown for an additional week under 60–80 μE m−2 sec−1 continuous light. Anthocyanin levels were measured as described previously ( Rabino & Mancinelli, 1986). In brief, whole shoot systems from three plants per assay were weighed and then extracted with 99:1 methanol:HCl (v/v) at 4°C. The OD530 and OD657 for each sample were measured and relative anthocyanin levels determined using the equation: OD530– (0.25 × OD657) × extraction volume (ml) × 1/weight of tissue sample (g) = relative units of anthocyanin/g fresh weight of tissue. Results were scaled relative to the anthocyanin levels found in wild-type seedlings grown on 0.18 m sucrose.
We thank Dr Chris Somerville for helpful suggestions regarding experiments, Drs Janet Braam and Bonnie Bartel for helpful suggestions regarding the manuscript, Dr Ruth Finkelstein for sharing unpublished data and for generously providing us with seeds from abi1-1 plants, Dr Chengdong Zhang for technical assistance and Dr Sjef Smeekens for sharing unpublished data. This work was supported in part by the United States Department of Energy, Energy Biosciences Program Grant DOE DE-FG03-98ER20300.