Gibberellin Dose-Response Curves and the Characterization of Dwarf Mutants of Barley

Abstract

Dose-response curves relating gibberellin (GA) concentration to the maximal leaf-elongation rate (LER_max) defined three classes of recessive dwarf mutants in the barley (Hordeum vulgare L.) ‘Himalaya.’ The first class responded to low (10⁻⁸–10⁻⁶ m) [GA₃] (as did the wild type). These grd (GA-responsive dwarf) mutants are likely to be GA-biosynthesis mutants. The second class of mutant, gse (GA sensitivity), differed principally in GA sensitivity, requiring approximately 100-fold higher [GA₃] for both leaf elongation and α-amylase production by aleurone. This novel class may have impaired recognition between the components that are involved in GA signaling. The third class of mutant showed no effect of GA₃ on the LER_max. When further dwarfed by treatment with a GA-biosynthesis inhibitor, mutants in this class did respond to GA₃, although the LER_max never exceeded that of the untreated dwarf. These mutants, called elo (elongation), appeared to be defective in the specific processes that are required for elongation rather than in GA signaling. When sln1 (slender1) was introduced into these different genetic backgrounds, sln was epistatic to grd and gse but hypostatic to elo. Because the rapid leaf elongation typical of sln was observed in the grd and gse backgrounds, we inferred that rapid leaf elongation is the default state and suggest that GA action is mediated through the activity of the product of the Sln gene.

Dwarf mutants have proven to be valuable tools in hormone studies in a wide range of plant species. This has been particularly so for the GAs (for review, see Ross et al., 1997) and more recently for the brassinosteroids (Clouse and Sasse, 1998). The identification of hormone-biosynthetic mutants that are normalized by hormone application illustrates the importance of such hormones in determining plant stature. Therefore, when characterizing new dwarf mutants, the growth response after hormone application is an important first screen because it allows potential hormone biosynthetic mutants to be identified. However, a considerable proportion of new mutants may show either no or only partial growth response to applied hormone; and these are potentially altered in signal transduction or in processes affecting growth. For the GAs, several such classes of dwarf can be recognized (Ross et al., 1997). In addition, there are GA-signaling mutants that show either a constitutive or an enhanced GA response (Ross et al., 1997).

Current interpretations of this broad grouping of GA “response” mutants are imprecise, largely because plant growth is regulated by many factors in addition to GA. Therefore, failure to respond to applied GA does not necessarily mean a deficiency in GA signaling. For example, two pea mutants that were originally interpreted as GA-response mutants were later shown to be brassinosteroid deficient and nonresponsive (Nomura et al., 1997). It is clearly an advantage to use, when possible, an independent GA response that does not involve growth; in this respect cereal systems have proved valuable because they allow changes in leaf growth to be compared with α-amylase production by aleurone tissue (Gale and Marshall, 1973; Chandler, 1988; Lanahan and Ho, 1988).

Another problem arises because the final extent of a response may not be the most accurate measure of hormone responsiveness. Nissen (1988) analyzed data in the literature for several GA responses, including leaf elongation, and concluded that they were “almost uniformly subsensitive”; i.e. a greater-than-expected concentration range of applied GA was required for the response to go from 10% to 90% of maximal values. Weyers et al. (1987, 1995) have emphasized the importance of determining the initial or maximal rates of response to hormone application rather than the final extent, but despite the renewed interest in hormone-response mutants (largely because of studies using Arabidopsis), this approach has not been widely adopted. When a reduced rate of response to hormone application is observed, it is necessary to determine whether there has also been a change in the concentration range over which the response occurs. In this manner, a mutant that requires higher hormone concentrations than the wild type does to bring about a similar response can be distinguished from one with a reduced response capacity (Firn, 1986).

Hormone dose-response curves therefore provide data essential for characterizing GA-response mutants (Weyers et al., 1995; Swain and Olszewski, 1996). Where the responses being measured are complex (such as organ-elongation rates) and likely to integrate a number of “simpler” components, it is important that mutants be compared with the appropriate wild-type background. In the case of an induced mutation, backcrossing is required so that the possibility is reduced that independent mutational events contributed to the response being measured. In this study we describe a leaf-elongation assay for GA responsiveness that defines three classes of dwarf mutants in barley: grd (GA-responsive dwarf), gse (GA sensitivity), and elo (elongation). Genetic interactions of these mutants with the sln (slender) “constitutive GA response” mutant (Foster, 1977) suggest that GA signaling proceeds through the SLN gene product.

MATERIALS AND METHODS

Chemicals

Initial experiments used commercial preparations of GA₃ (>90%, Sigma), but to obtain saturation in dose-response experiments, very high (millimolar) concentrations were required. Considerable GA₁ was detected by GC-selected ion monitoring in the commercial GA₃ preparation, so pure preparations of GA₁ and GA₃ were kindly provided by L.N. Mander, Research School of Chemistry, Australian National University, Canberra, ACT. Samples of GA₄₄, GA₁₉, and GA₂₀ were also kindly provided by L.N. Mander. All GA solutions were prepared by dissolving powder in 1 mm potassium-phosphate buffer, pH 5.5, and diluting in this solution when necessary. The stock solution of GA₃ was 9.76 mm, and excess acidity due to GA₃ was neutralized by the dropwise addition of 20 mm KOH until the pH returned to 5.5. GA solutions were stored at −20°C. Tetcyclacis (94.8%) was kindly provided by Dr W. Rademacher (BASF, Limburgerhof, Germany) and dissolved in ethanol at 3.5 mm.

Plant Material

The dwarf mutants of barley (Hordeum vulgare) ‘Himalaya’ were isolated after mutagenesis with sodium azide as described by Zwar and Chandler (1995). From about 200 independent dwarf mutants we selected several different types for detailed study, based on GA₃ application and genetic-complementation studies. Seven mutants are described here, all of which are recessive and all of which have been through three back-crossing generations before the establishment of homozygous seed stocks; at the seedling stage they ranged in height from approximately 20% to 50% of the wild-type parent. The first class (M117, M359, and M411) showed a large response to GA₃ application (microdrops or spray), and the three mutants represent three genetic loci. Phenotypically these mutants are similar, with leaves that are shorter and darker green and stem internodes that are shorter than the wild type. The second class (M121 and M488) showed only partial growth responses to GA₃, even at high concentrations. These two mutants are phenotypically similar to those described above, and represent two alleles at a single locus. The third class (M21 and M626) did not show any growth response to GA₃, and the two mutants are at different genetic loci.

The introduction of the sln mutation into the wild type and M117 was described by Smith et al. (1996). The same procedure was followed in crossing sln1 into the other classes described here. In one class (M121 and M488) the sln phenotype is expressed as it is in the wild type and in M117. In the other class (M21 and M626) the sln phenotype is not observed during growth of the first two leaves, but homozygous (sln1sln1) plants could be identified later in their growth on the basis of abnormal stem elongation. Plants homozygous for sln1 do not set seed, so comparisons were made between phenotypically normal (Sln1−) and slender (sln1sln1) segregants in the progeny of Sln1sln1 heterozygotes in different genetic backgrounds.

Seedling Growth and Determination of LER_max and Final Blade Length

Grains were surface-sterilized as described previously (Chandler and Jacobsen, 1991) and placed embryo-side down between two sheets of autoclaved paper (3MM, Whatman) “envelopes” that were moistened with the appropriate solution and held vertically in a plastic frame placed in the solution. After the grains were stratified (4°C in the dark) for 48 h, they were placed under low-intensity fluorescent lighting at 20°C (d 0). After a further 3 d, the germinated grains in each envelope were culled for uniformity of shoot length (providing approximately 15 seedlings per sample), and the envelope was aligned with positional markers on a clear plastic sheet. The position of the tip of each leaf was marked on the plastic sheet, and the envelope was returned to the growth assembly. After an additional 1, 2, 3, and 4 d, the envelopes were again placed on the original sheets in their original position, and the new position of each leaf tip was marked. For each 24-h interval, the distance between marks was recorded and the mean length increment was determined. In most cases the maximal elongation occurred between d 4 and 5 after transfer from 4°C to 20°C. Occasionally, maximal growth occurred in either the previous or the following 24-h interval. For each genotype and treatment, the maximal value in each set of daily increments was expressed as a millimeter-per-day rate and abbreviated LER_max. In one experiment, seedlings of M117 and M411 were maintained until the growth of L₁ had ceased. The shoot was then dissected to determine the final blade length.

Construction of Dose-Response Curves

LER_max data were analyzed as a function of [GA] using individual seedling data and PEST software (Weyers et al., 1987). This program provides estimates of hormone-sensitivity parameters fitted to a modified Hill equation. The curves fitted to the data by the PEST software were plotted together with raw data points representing the means ± se of LER_max.

α-Amylase Determinations

The embryonic axes of dry grains of the wild type and M488 were removed using a dissecting blade under a dissecting microscope. There was minimal damage to the scutellum. “De-axised” grains do not produce α-amylase unless incubated with GA₃ (Jones and Armstrong, 1971), and any α-amylase produced remains within the grain. The de-axised grains were surface-sterilized, placed in paper envelopes, and incubated under conditions identical to those for intact grains (see above). α-Amylase activity was determined as previously described (Chandler and Jacobsen, 1991) on duplicate samples of five grains each.

RESULTS

Growth of L₁

Elongation of L₁ initially involved only the blade, but between d 5 and 7 both blade and sheath were elongating (Fig. 1). At later stages, elongation of the leaf involved only the sheath. The rate of leaf elongation was far from uniform, and LER_max occurred just before blade growth began to slow (just before sheath growth commenced; Fig. 1, inset). Maximal rates of sheath elongation did not exceed those of the blade (data not shown).

Figure 1.

Growth of L₁ of wild-type barley. Grains were surface-sterilized, placed in moist paper envelopes, stratified, and incubated as described in Methods. At the indicated times the mean L₁ lengths of 10 seedlings were determined, as well as the mean lengths of the blade and sheath. Where not visible, error (se) bars lie within the symbols. Inset, Mean elongation rate of the blade of L₁ in the previous 24 h was plotted as a function of days of growth. The LER_max value for this set of data is indicated.

L₁ of the dwarf mutants was always smaller than that of the wild type, and LER_max values were considerably lower (7–15 mm d⁻¹, compared with 37 mm d⁻¹ for the wild type; see below). However, the pattern of L₁ growth of the mutants was similar to that of the wild type; blade elongation preceded that of the sheath, and LER_max occurred just before the transition from blade growth to sheath growth (data not shown).

Effects of GA₃ on LER_max

Grains of the wild type and of the dwarf mutants were germinated in a range of concentrations of GA₃, and the LER_max was determined. The resulting dose-response curves (Fig. 2) show an LER_max for the wild type of about 37 mm d⁻¹ at low [GA₃] (<10⁻⁸ m), increasing to about 67 mm d⁻¹ at high [GA₃] (>10⁻⁶ m). Among the dwarf mutants, three response classes were identified that differed in the effect of GA₃ on LER_max: the first class (M117, M359, and M411) responded to GA₃ over the same concentration range (10⁻⁸–10⁻⁶ m) as the wild type; the second class (M121 and M488) responded over a much higher and wider concentration range (10⁻⁶–10⁻³ m); and the final class (M21 and M626) showed no response to [GA₃] as high as 10⁻³ m. Growth rates on the lowest concentrations of GA₃ were not significantly different from rates on control medium without GA₃ (compare Figs. 2, 6, and 7).

Figure 2.

Dose-response curves relating LER_max of L₁ to [GA₃]. Grains of the indicated lines were surface-sterilized, placed in paper envelopes moistened with the appropriate [GA₃], stratified, incubated in low light, and LER_max (mean ± se) of seedling L₁ was determined as described in Methods. Curves in the top six panels were fitted using PEST software (Weyers et al., 1987). Note differences in the range of GA₃ concentrations in different panels.

Figure 6.

LER_max of L₁ of elo mutants and a grd mutant growing with or without tetcyclacis and GA₃. Grains of M21 (elo1), M626 (elo2), and M117 (grd1) were surface-sterilized, placed in moist paper envelopes containing, where appropriate, 2 μm tetcyclacis or 2 μm tetcyclacis plus 10 μm GA₃, stratified, and incubated under low light; and the mean LER_max of seedling L₁ was determined as described in Methods. A replicate consisted of 10 seedlings, and there were three replicates for each genotype and treatment. Within a genotype, letters (a, b, or c) indicate significance (P < 0.05) for the differences between the means of each treatment.

Figure 7.

LER_max of L₁ of Sln1− and sln1sln1 segregants in different genetic backgrounds. Grains from stocks in which the sln1 allele was segregating in different genetic backgrounds (WT, wild type; grd1, M117; gse1, M121; and elo1; M21) were surface-sterilized, placed in moist paper envelopes, stratified, and incubated in low light, and the LER_max (means ± se) of seedling L₁ was determined as described in Methods. In an elo1 background, slender (sln1sln1) seedlings cannot be distinguished at the first leaf stage from Sln1− seedlings, but after transplanting and further growth, the early stem elongation characteristic of sln1sln1 plants, which still occurs in an elo1 genetic background, allowed the genotype to be determined.

For the wild type and the first two classes, the mean values of different parameters were estimated from the fitted curves: LER_max at zero and saturating concentrations of GA₃ (R_min and R_max, respectively); the Hill interaction coefficient (p), which provids a measure of the “steepness” of the response to increasing concentration of GA₃; and [H]₅₀, the concentration of GA₃ at which 50% of the maximal response to GA₃ is attained (Table I). The values of these parameters highlight the similarity between the wild type and the first class of mutant in their response to GA₃, with p values close to unity (reflecting near Michaelis-Menten behavior), and the estimated concentration of GA₃ required for a half-maximal response falling within a narrow range of 56 to 120 nm. The R_MIN values of the three dwarf mutants in this class were lower than those of the wild type, which is consistent with their dwarf nature, but their R_max values were not quite as high as those of the wild type, possibly because the grains of the dwarf mutants were smaller by 5% to 20% on a dry-weight basis (data not shown). As a consequence, in mature grains the L₁ primordium of the dwarf would probably be smaller than that of the wild type, and its capacity to increase the elongation rate in response to exogenous GA could potentially be compromised.

Table I.

Parameters estimated from GA₃ dose-response curves

Genotype	RMIN	RMAX	p	[H]50
	mm d−1			nm
Wild type	36.5  ± 3.1	67.5  ± 5.8	0.96  ± 0.08	120
grd Mutants
M117 (grd1)	12.0  ± 1.1	56.5  ± 5.3	0.98  ± 0.09	58
M359 (grd2)	15.0  ± 1.4	60.3  ± 5.8	0.98  ± 0.09	84
M411 (grd3)	11.6  ± 1.0	56.9  ± 5.1	0.91  ± 0.08	56
gse Mutants
M121 (gse1)	12.8  ± 1.0	50.0  ± 4.0	0.47  ± 0.04	9,500
M488 (gse1)	11.2  ± 0.9	52.5  ± 4.3	0.61  ± 0.05	43,000

Mean values (and their 95% confidence limits) of parameters (see text for explanation of symbols) were estimated for the fitted curves shown in Figure 2 using PEST software (Weyers et al., 1987).

The values shown in Table I contrast the behavior of the wild type and the first class of mutants (M117, M359, and M411) against that of the second class (M121 and M488). In particular, there was a much broader response in the latter mutants (p values considerably less than unity), and much higher (90- to 350-fold) concentrations of GA₃ were required for half-maximal response ([H]₅₀). R_max values for these mutants were again somewhat less than those for the wild type, and an argument similar to that above could be made based on grains that are approximately 25% smaller (dry weight) than those of the wild type. In addition, even at 10⁻² m GA₃ (the highest concentration tested), the response may not have been saturated. The estimated values of the parameters in Table I for this second class of mutant might be less reliable than those estimated for the wild type and the first mutant class because saturation was barely attained. Nevertheless, the main differences (a broader transition and a displacement of the response to higher [GA₃]) are clearly discernible in the curves of Figure 2.

Similar experiments were carried out with GA₁, which is also an important bioactive GA for leaf elongation. A less extensive range of concentrations and mutants was studied, but the results (data not shown) were in close agreement with those described for GA₃. The principal difference was that the wild type and the first class of mutant had an [H]₅₀ value for GA₁ (500–1700 nm) that was approximately 10-fold higher than that estimated for GA₃. If we assume that GA₁ and GA₃ have equal intrinsic activity and similar rates of uptake, the difference in [H]₅₀ values may reflect more rapid catabolism of GA₁. The behavior of the three mutant classes in their responses to GA₁, including the estimates for p and the relative differences in [H]₅₀, paralleled the behavior observed for GA₃.

Comparing the Effects of GA₃ on LER_max and on Final Blade Length

Seedlings of two mutants (M117 and M411) were maintained on a range of GA₃ concentrations until growth of L₁ stopped, allowing the final length of the L₁ blade to be determined. To allow a direct comparison between the effects of GA₃ on LER_max and on final blade length, each response was normalized to an R_MIN value of 1 (Fig. 3). Final blade length responded to increasing concentration of GA₃, but the response was lower in magnitude than that of LER_max and occurred over a wider range of GA₃ concentrations. These effects may have resulted from differences in the duration of elongation.

Figure 3.

Dose-response curves comparing the effects of GA₃ on either LER_max or final blade length. Grains of the indicated lines were surface-sterilized, placed in paper envelopes moistened with the appropriate [GA₃], stratified, and incubated in low light, and LER_max (mean ± se) or final blade length of seedling L₁ was determined as described in Methods. To allow direct comparison, response ratios are plotted in which the LER_max or L₁ blade length in the absence of GA₃ is assigned a value of unity. Curves are the same (minus data points) as shown in Figure 2. Individual data points (▾) are for final blade length.

Definition of Mutant Classes

Based on the dose-response curves (Fig. 2), three classes of dwarf mutants can be defined. The first, grd, responds to GA₃ over the same concentration range as the wild type. These mutants are proposed to have normal GA signaling and their dwarfism is associated with low levels of endogenous bioactive GAs (P.M. Chandler and J.R. Lenton, unpublished results). The three mutants in this class represent three different genetic loci: grd1 (M117), grd2 (M359), and grd3 (M411). Mutants in the second class are primarily characterized by an alteration in GA sensitivity (the gse mutants). M121 and M488 represent alleles at the gse1 locus, because no complementation is observed when they are intercrossed. The greatly reduced sensitivity to GA of these two mutants probably explains why they showed only poor growth responses to GA₃ in preliminary experiments (see Methods). Mutants in the third class show no elongation response to GA. On the basis of the results presented below, these mutants are proposed to be defective in the specific processes that are required for leaf elongation (the elo mutants), rather than in GA signaling. M21 and M626 represent different elo loci. For each of these three mutant classes, additional experiments aimed at a more detailed characterization were performed.

Response of grd Mutants to GA-Biosynthetic Intermediates

GA-biosynthetic intermediates may be active or inactive in promoting elongation in dwarf mutants, depending on the concentration at which they are applied, the severity of the dwarfing mutation, and the step in the GA biosynthetic pathway in which the mutant is blocked. The growth responses of L₁ of the three grd mutants to late intermediates of the early 13-hydroxylation pathway (Grosselindemann et al., 1992) were determined (Fig. 4). For each mutant, GA₁ treatment resulted in LER_max values that were greater than the wild-type values (approximately 37 mm d⁻¹). The [GA] used in these experiments (2 × 10⁻⁶ m) was only slightly greater than the [H]₅₀ value determined for GA₁ (approximately 1 × 10⁻⁶ m) so that LER_max would be highly responsive to the content of active GAs. GA₂₀ was very effective in stimulating LER_max of the grd1 and grd3 mutants, and GA₄₄ and GA₁₉ were slightly less effective. This pattern differed markedly from that seen for the grd2 mutant, in which each of the intermediates had very low activity in stimulating elongation. We inferred from this that the grd1 and grd3 mutants convert “inactive” GA precursors such as GA₂₀ to growth-active GAs, whereas grd2 mutants do not (or do so at a greatly reduced rate). This pattern would be consistent with grd2 mutants having reduced levels of 3β-hydroxylation (converting GA₂₀ to GA₁), whereas the other two loci are presumably blocked earlier in the pathway.

Figure 4.

LER_max (means ± se) of different grd mutants in the presence of GA-biosynthetic intermediates. Grains of M117 (grd1), M359 (grd2), and M411 (grd3) were surface-sterilized, placed in paper envelopes moistened with the indicated GA at 2 μm, stratified, and incubated in low light, and LER_max (means ± se) of seedling L₁ was determined as described in Methods. GA₄₄, GA₁₉, and GA₂₀ are successive biosynthetic intermediates in the early 13-hydroxylation pathway leading to the formation of the bioactive GA₁.

α-Amylase Production by gse1 Grains in Response to GA₃

Mutants in the gse1 locus were characterized by reduced sensitivity to GA₃ for leaf elongation; therefore, α-amylase production by aleurone tissue was also examined. α-Amylase activity in de-axised wild-type grains increased with time in the presence of 10⁻⁸ to 10⁻⁷ m GA₃ at approximately one-half the maximal rate observed with 10⁻⁶ and 10⁻⁵ m GA₃ (Fig. 5). In contrast, α-amylase activity of de-axised gse1 grains showed no increase with time at GA₃ concentrations ≤ 10⁻⁶ m, intermediate rates of accumulation with 10⁻⁵ m GA₃, and high rates of accumulation at 10⁻⁴ to 10⁻³ m GA₃. This pattern parallels that observed for leaf elongation, in which responses equivalent to those of the wild type required at least 100-fold-higher concentrations of GA₃. The maximal rate of α-amylase accumulation in the mutant was less than that of the wild type, possibly because the grains were 25% smaller (on a dry-weight basis) than those of the wild type, and perhaps there was an equivalent reduction in aleurone cell number. We concluded that the gse1 mutants are defective in a component of GA signaling that is required for two independent GA responses: leaf elongation and α-amylase production by aleurone.

Figure 5.

α-Amylase activity of de-axised grains of the wild type and a gse1 mutant incubated with different [GA₃]. The embryonic axes of wild type and M488 (gse1) grains were removed, and the resulting de-axised grains were surface-sterilized, placed in paper envelopes moistened with the indicated [GA₃], stratified, and incubated in low light as described in Methods. At the indicated times, duplicate samples (five grains each) were harvested and α-amylase was extracted and assayed. Each data point is the mean of duplicate samples.

Conditional Regulation of LER_max in elo Mutants by GA₃

The two elo mutants were characterized by low rates of leaf elongation even at very high concentrations of GA₃ (Fig. 2); however, the aleurone of both mutants showed near-normal responses to GA₃ for α-amylase production (data not shown). We considered the possibility that leaf-elongation rates were limited by defective components involved in leaf elongation rather than in GA signaling. Inhibitors of GA biosynthesis such as tetcyclacis induce dwarfing in barley, but this effect can be overcome by GA₃ (Zwar and Chandler, 1995). Grains were germinated of the two elo mutants and of the grd1 mutant (as a control), and the seedlings were grown in control conditions in the presence of tetcyclacis alone or tetcyclacis plus GA₃.

Significant additional dwarfing was induced in all of the lines by tetcyclacis treatment, as revealed by the LER_max values (Fig. 6). When GA₃ was also present, the LER_max values returned to control levels for both elo mutants but greatly exceeded control levels for the grd1 mutant (as expected). We inferred from this that the elo mutants are capable of responding to GA₃ provided leaf elongation occurred at a lower rate than in control conditions. In more detailed experiments with the elo1 mutant, the concentration dependence for restoration of LER_max by GA₃ was examined. The results (data not shown) indicated that in the presence of tetcyclacis, concentrations of GA₃ as low as 10⁻⁷ m were able to restore LER_max to control levels, indicating that the elo mutants were capable of responding to low concentrations of GA₃. The failure of such mutants to respond to GA₃ in the dose-response experiment (Fig. 2) was presumably because they were already elongating at their maximal rate.

Leaf Elongation of the sln1 Mutant in Different Genetic Backgrounds

The sln1 mutant of barley (Foster, 1977) exhibits rapid leaf elongation without added GA₃, yet has lower than normal levels of active GAs in its leaves (Croker et al., 1990). On this basis, and because it shows high rates of α-amylase production by aleurone incubated without added GA₃, sln1 is regarded as a constitutive GA-response mutant (Chandler, 1988; Lanahan and Ho, 1988). We previously showed that sln1 derivatives of M117 (grd1) elongated rapidly despite the dwarfing background (Smith et al., 1996). Similar results were found for sln1 in grd2 or grd3 backgrounds (P.M. Chandler, unpublished data). To determine whether double mutants of sln1 with either gse1 or elo would also elongate rapidly without added GA₃, LER_max values were determined for segregating Sln1− and sln1sln1 types in the different genetic backgrounds. The results (Fig. 7) show that sln1 homozygotes elongate equally rapidly in the wild type, grd1, and gse1 genetic backgrounds, indicating that sln1 is epistatic to gse1 and grd1. In contrast, the LER_max value of sln1 homozygotes in an elo1 (Fig. 7) or elo2 genetic background (data not shown) did not differ significantly from the Sln1− segregants, indicating that sln1 was hypostatic to these two elo loci, which is consistent with the proposal that these mutants were already elongating at their maximal rate. The effect of homozygosity at sln1 on LER_max values for the wild type, grd1, and gse1 was equivalent to that of saturating concentrations of GA₃ (compare Figs. 2 and 7).

DISCUSSION

Hormone dose-response studies on the effects of differing concentrations of GA on LER_max provided an effective basis for discriminating between different classes of dwarf mutants in barley. In earlier studies GA application to seedlings had readily distinguished fully responsive dwarfs (thought to be affected in GA biosynthesis) from mutants that showed no response to GA. Dwarf mutants that gave a partial response to GA were problematic, because they may have involved alterations either in the magnitude of response at saturating hormone concentrations or in the concentration range over which a response occurred. Dose-response experiments distinguished between these possibilities. The gse mutants fit the latter category, and thereby define a novel type of mutant that is involved in GA signaling. Growth responses to GA in a recently described pea mutant (lgr) have similar properties (Ross et al., 1997).

The current interest in hormone signal transduction requires that quantitative assays be used to characterize the mutants that are affected in such processes. The GA dose-response curves described here provide a framework for future characterization of the remaining barley dwarf mutants in our collection. Sensitivity parameters were estimated from the dose-response curves using PEST software (A'Brook, 1987; Weyers et al., 1987), which fits data to a modified Hill equation. It was significant that near Michaelis-Menten responses were observed even for growth rates of whole leaf blades (at least for the wild type and grd mutants), because there are presumably many steps between GA perception and the final leaf growth rate at which the initial magnitude of a response to GA could be modified.

In many previous studies there was a broader-than-expected GA concentration range over which responses occurred (Nissen, 1988). In contrast, the range of GA concentrations over which a response occurred in the wild type and the grd mutants was relatively narrow (p ≈ 1; see Table I). This difference was probably because we used LER_max as a measure of hormone response (Weyers et al., 1987) rather than the extent of response, which was used in the earlier studies. In some cases such broad transitions may be genuine, perhaps reflecting attenuation so that a wide range in hormone content can be accommodated. However, until they are analyzed in terms of rate rather than final extent, the wide concentration range might also be misleading. When we monitored final blade length rather than LER_max, a broader transition was observed and the magnitude of the response was smaller (Fig. 3). The probable explanation for this difference is a shorter duration of the response at high concentrations of GA₃, so that the effect of GA₃ on growth rate was never exactly matched by the effect on final length.

A [GA₃] of approximately 10⁻⁷ m stimulated LER_max in grd dwarfs to that of the wild type given only water. This concentration is close to the [H]₅₀ value estimated for GA₃ (Table I), a condition in which LER_max changes most rapidly as the concentration of applied GA₃ changes. In this range there was a 20% change in LER_max for a 2-fold change in [GA₃], illustrating the potential for relatively small changes in the content of endogenous bioactive GAs to have considerable effects on the leaf-elongation rate when, for example, plants respond to different environmental factors. It is difficult to compare the [GA₃] applied in a treatment (e.g. 10⁻⁷ m) with the endogenous contents of bioactive GAs, because we know neither the relative contributions of different GAs (GA₁, GA₃, and possibly other GAs) in determining leaf growth rate, nor the most appropriate part of the leaf (zones of cell division or elongation, or perhaps only the epidermis of such regions) in which to determine GA content.

Tonkinson et al. (1997) determined GA₁ and GA₃ contents in the elongation zones of the second leaves of wheat seedlings and, assuming uniform distribution, their values correspond to 2 to 7 nm. These estimates are considerably lower than the [H]₅₀ values above, but there are many factors that could account for such a discrepancy, including species and leaf differences and the assumption of uniform GA distribution. It is apparent that the maintenance of normal growth rates requires an adequate supply of and an ability to sense endogenous GAs, because mutants that affect either process are dwarfed. The relative importance of these two processes in explaining natural variation in growth rate is difficult to assess. Weyers et al. (1995), in discussing hormonal control in a general sense, argued that combined control should always be assumed unless there is evidence to the contrary. In this context it is interesting that the growth rate of the grd1gse1 double mutants was considerably lower than either of the single mutants (P.M. Chandler, unpublished observations).

Interpretation of the grd mutants is relatively straightforward, because equivalent mutants have been isolated in a range of other plant species, and have generally involved mutations in the GA-biosynthetic pathway (for review, see Ross et al., 1997). For example, the growth responses of the grd2 mutant (Fig. 4) are typical of 3β-hydroxylase mutants that have been isolated in several species. There are two GA-responsive dwarf mutants in barley that have been studied in some detail (Hentrich et al., 1985; Boother et al., 1991), and both of these are allelic with the grd1 locus described here (P.M. Chandler, unpublished data).

An important advantage of barley and some other cereals is the availability at the seedling stage of two well-defined GA responses (leaf elongation and α-amylase production) that involve different components (the meristem-leaf-elongation zone and aleurone, respectively). This has been important in interpreting the gse mutant category. In both assays the gse mutants were capable of responding to GA₃, probably to the same extent as the wild type and grd mutants, but the gse mutants required approximately 100-fold higher concentrations of GA₃. These recessive mutants are unique in showing reduced sensitivity for two different GA responses. One interpretation is that they define receptors that have a lower affinity for GA than the receptors in the wild type. Loeb and Strickland (1987) showed that dose-response curves can reflect the activity of components involved in signal transduction, rather than initial receptor-hormone interactions; thus the gse1 mutants may also involve changes in the downstream components of GA signaling. An alternative interpretation is that a “primary” GA receptor or signaling pathway is rendered nonfunctional in the gse mutants, and the activity of a redundant pathway(s) with different properties is revealed. An interesting feature of the dose-response curves of both gse mutants was the broader range of concentrations over which the response occurred (Fig. 2). The associated lower values of p (Table I) may result from negative cooperativity in the binding of interacting components (e.g. a ligand to its receptor), either as a result of mutational change or because a different signaling pathway was operating.

Two other interpretations of the gse phenotype are possible. The first, involving overproduction of an enzyme that inactivates GAs, is considered unlikely for several reasons: (a) we would expect the trait to show some degree of dominance if it resulted from increased levels of a catabolic enzyme; (b) the dose-response curves indicate that extremely high concentrations of GA₃ (by in vivo standards) are still subsaturating, yet during normal growth the gse mutants are not severely dwarfed, and an altered enzyme that was capable of inactivating such high concentrations of exogenous GA₃ might be expected to have an extreme effect on the endogenous GA content, resulting in a much more severe dwarf phenotype; and (c) determination of the endogenous GA content of developing grains of M488 exhibits a profile that is very similar to that of the wild type (P.M. Chandler, unpublished data), indicating that there are no major changes in GA metabolism. This includes stages when the gse phenotype of developing grains is being expressed, revealed by the failure of 10 μm GA₃ to induce germination of isolated immature grains.

The second alternative interpretation of the gse mutants is that their reduced sensitivity to GA may have resulted either from increased levels of endogenous ABA or from enhanced responses to ABA (Cutler et al., 1996), because in barley grains and seedlings, ABA antagonizes many of the effects of GA. This interpretation is difficult to exclude until more information is available, but two lines of investigation have failed to provide support: first, the quantitative hormone analysis of developing grains of M488 (see above) revealed ABA contents similar to those of the wild type, and second, there were similar relative reductions in the L₁ growth rate observed when gse1, grd1, and wild-type grains were germinated in the presence of 1 μm ABA (P.M. Chandler, unpublished data).

The two elo mutants showed no significant growth stimulation by GA₃, yet their ability to perceive and initially respond to GA₃ was probably not affected. For example, LER_max was responsive to GA₃ with approximately the normal concentration dependence when the mutants were further dwarfed either by chemical means (Fig. 6) or by making double mutants with a grd locus (P.M. Chandler, unpublished data). This observation and the epistasis of elo to sln1 (Fig. 7) suggest that the mutations affect specific components required for leaf elongation rather than those involved in GA signaling. There was no restoration of normal growth when these mutants were germinated on other growth-related hormones such as brassinolide, IAA, or kinetin (P.M. Chandler, unpublished observations).

The characterization of these dwarf mutants suggests that they are representative of three broad areas involving GA control of growth: GA biosynthesis (grd), GA signaling (gse), and the growth processes themselves (elo). In the simplest model, GA elicits a positive signaling pathway and growth is stimulated. According to this model, the low growth rates of the grd and gse mutants are due to the effects of reduced GA content and GA sensitivity, respectively, on GA signaling. The sln1 mutant is recessive, and slender plants show “constitutive” GA responses. Thus, the product of the wild-type Sln1 gene (SLN) presumably functions as a negative regulator of GA signaling (if sln1 involves a loss of function). Is SLN a negative regulator that plays a direct role in GA signaling, or does it play an indirect role? Other signaling pathways in the plant could modulate flux through a positive GA-signaling pathway via SLN acting as a negative regulator of this pathway.

Slender derivatives of the grd mutant showed a typical slender phenotype rather than a dwarf phenotype. The same result was observed with slender derivatives of the gse mutant. Thus, in an sln1 background, mutations such as grd and gse that result in reduced GA signaling had no effect on growth rate. If SLN is an indirect negative regulator of a positive GA-signaling pathway, we might still expect to see reduced growth rates in the double mutants because of reduced GA signaling. We favor the view that SLN is a negative regulator whose activity is directly involved in GA signaling. If the GA-signaling pathway is under negative control, the positive responses observed when, for example, GA is applied must involve reducing the extent of negative regulation mediated by SLN. In the same manner, ‘Himalaya’ barley grows at wild-type rates because with wild-type levels of GA signaling, it can substantially reduce the extent of negative regulation imposed by SLN. By contrast, the grd and gse mutants have lower levels of GA signaling and are less able to reduce the extent of negative regulation by SLN, and consequently their growth is slow.

This interpretation is similar to that reached for the product of the GAI gene in Arabidopsis, which, according to genetic evidence, also functions as a negative regulator involved in GA signaling and whose activity is proposed to be regulated by GA (Peng et al., 1997; Harberd et al., 1998). In Arabidopsis there are now three different proteins that, on the basis of mutant studies, are proposed to be negative regulators of GA signaling. GAI (Peng et al., 1997) and RGA (Silverstone et al., 1997) are closely related; based on sequence comparisons, they are putative VHIID transcription factors. SPY is a protein with a sequence closely related to O-linked GlcNAc transferases (Jacobsen et al., 1996). Only for SPY has the proposed role as a negative regulator of GA signaling been confirmed: transient expression of barley SPY (HvSPY) largely prevented GA-induced α-amylase promoter activity in aleurone (Robertson et al., 1998). Is barley SLN related to these other proteins? It is known that SLN does not correspond to HvSPY (Robertson et al., 1998), but there is no evidence yet concerning its relationship to GAI or RGA. Scott (1990) suggested that (semi)-dominant GA “insensitive” mutants (encoded by gai, Rht3, and D8) might involve the same gene that is affected in recessive constitutive GA-response mutants such as sln1: GAI would involve a gain of function (a negative regulator whose activity was no longer regulated by GA), whereas the mutant SLN would involve a loss of function. The cloning of GAI and RGA should allow their relationship to SLN to be investigated.

Abbreviations:

L₁

first leaf or leaf one

LER_max

maximal leaf-elongation rate