Overexpression of DWARF4 in the brassinosteroid biosynthetic pathway results in increased vegetative growth and seed yield in Arabidopsis


For correspondence (fax +1 310 317 8998; e-mail kfeldmann@ceres-inc.com).


Plants unable to synthesize or perceive brassinosteroids (BRs) are dwarfs. Arabidopsis dwf4 was shown to be defective in a steroid 22α hydroxylase (CYP90B1) step that is the putative rate-limiting step in the BR biosynthetic pathway. To better understand the role of DWF4 in BR biosynthesis, transgenic Arabidopsis plants ectopically overexpressing DWF4 (AOD4) were generated, using the cauliflower mosaic virus 35S promoter, and their phenotypes were characterized. The hypocotyl length of both light- and dark-grown AOD4 seedlings was increased dramatically as compared to wild type. At maturity, inflorescence height increased >35% in AOD4 lines and >14% in tobacco DWF4 overexpressing lines (TOD4), relative to controls. The total number of branches and siliques increased more than twofold in AOD4 plants, leading to a 59% increase in the number of seeds produced. Analysis of endogenous BR levels in dwf4, Ws-2 and AOD4 revealed that dwf4 accumulated the precursors of the 22α-hydroxylation steps, whereas overexpression of DWF4 resulted in increased levels of downstream compounds relative to Ws-2, indicative of facilitated metabolic flow through the step. Both the levels of DWF4 transcripts and BR phenotypic effects were progressively increased in dwf4, wild-type and AOD4 plants, respectively. This suggests that it will be possible to control plant growth by engineering DWF4 transcription in plants.


Brassinosteroids (BRs), a class of plant steroid hormones, mediate many important processes in plant growth and development including stimulation of cell division, cell elongation, vascular differentiation, and stress responses (Clouse and Sasse, 1998). The biological importance of BRs has been reinforced by the recent characterization of a variety of mutants perturbed in signaling or biosynthesis of this hormone (Altmann, 1998). BR-signaling mutants have been isolated from four different plant species: brassinosteroid insensitive1 (bri1) from Arabidopsis (Clouse et al., 1996; Li and Chory, 1997; Noguchi et al., 1999b); lka from pea (Nomura et al., 1999); curl-3 from tomato (Koka et al., 2000); and d61 from rice (Yamamuro et al., 2000). In contrast to the identification of only one locus involved in signaling in each of these species, a larger number of genes for BR biosynthetic enzymes have been cloned or defined by mutants. These include Arabidopsis DWARF1 (DWF1) (Choe et al., 1999a; Feldmann et al., 1989; Klahre et al., 1998; Takahashi et al., 1995); DE-ETIOLATED2 (DET2) (Fujioka et al., 1997; Li et al., 1996; Noguchi et al., 1999a); CONSTITUTIVE PHOTO MORPHOGENESIS and DWARFISM (CPD) (Szekeres et al., 1996); DWF4 (Azpiroz et al., 1998; Choe et al., 1998); DWF5 (Choe et al., 2000), DWF7 (Choe et al., 1999b); HYPER SENSITIVE TO ABSCISIC ACID and AUXIN1 (SAX1) (Ephritikhine et al., 1999); pea LKB (Nomura et al., 1997, 1999); and tomato DWARF and DUMPY (Bishop et al., 1996, 1999; Koka et al., 2000). In addition, Neff et al. (1999) isolated a bas1-D mutant (phyB activation-tagged suppressor1-dominant) that suppressed the long hypocotyl phenotype of the phytochrome B mutant hy5. BAS1 encodes a cytochrome P450 enzyme (CYP72B1) proposed to be involved in BR degradation.

The classic phenotype of Arabidopsis BR mutants includes short robust stems, dark-green and rounded leaves, and reduced fertility in light-grown plants, as well as abnormal de-etiolation in dark-grown seedlings exhibited by short hypocotyls and expanded cotyledons. The phenotype of BR biosynthetic mutants can be restored to wild type by exogenous application of bioactive BRs, whereas other growth-promoting hormones, such as gibberellins and auxins, do not rescue the phenotype (Azpiroz et al., 1998; Choe et al., 1998).

Brassinolide (BL), the most bioactive BR found to date, is a poly hydroxylated steroidal lactone that is synthesized using campesterol (CR) as a pathway precursor (Figure 1). Two consecutive pathways are necessary for the biosynthesis of BRs: the sterol-specific pathway and the BR-specific pathway (Choe et al., 1999b). The BR-specific pathway diverges into two parallel branches at campestanol (CN). Depending on the initiation at one of the two hydroxylation reactions, 6α- or 22α-hydroxylation, CN proceeds through the early or the late C-6 oxidation pathway, respectively (Figure 1). As the activity of the two enzymes, 6α- and 22α-hydroxylase, may determine which pathway dominates, these enzymes are hypothesized to play key regulatory roles in the BR biosynthetic pathway. Recently, a C-6 oxidase (CYP85) from tomato has been characterized (Bishop et al., 1999). The enzyme was shown to catalyze only C-6 oxidation of 6-deoxocastasterone (6-DeoxoCS), and not the C-6 oxidation of CN. dwf4 mutants were proposed to be defective in the 22α-hydroxylation step (Choe et al., 1998). The DWF4 gene has been shown to encode a cytochrome P450 enzyme (CYP90B1) that shares 43% amino acid sequence identity with CPD, a putative 23α-hydroxylase (CYP90A1) that mediates the biosynthetic step immediately downstream of DWF4.

Figure 1.

The brassinosteroid biosynthetic pathway and endogenous levels of BRs in light-grown plants of dwf4-1, wild type and AOD4-4, and in 7-day-old dark-grown seedlings of wild type and AOD4-4.

Enzymes named after corresponding mutants of Arabidopsis are labeled at the steps in the pathway. The endogenous BR levels are shown under the names of each compound (units ng g−1 FW). na = not analyzed, nd = not detected. Data with asterisks (*) were from different sets of analyses, with Ws-2 wild-type and AOD4-73 plants.

Previously, we have shown that the steady-state level of DWF4 mRNA was extremely low (Choe et al., 1998). In addition, Fujioka et al. (1995) showed that Catharanthus roseus accumulates DWF4 precursors that possess very low biological activity, whereas the downstream compounds of DWF4 are scarce but show relatively high bioactivities. Accordingly, the 22α hydroxylation step has been considered a rate-limiting step in both the early and late C-6 pathways (Choe et al., 1998; Fujioka et al., 1995). Thus we hypothesize that DWF4 transcription is tightly controlled, and this is an important mechanism for regulating enzyme activity and thus endogenous BR concentration. To test this, Arabidopsis and tobacco plants overexpressing DWF4 were generated and characterized.


Increased growth of DWF4 overexpression lines

To ectopically overexpress DWF4, a pOD4 construct, placing the DWF4 genomic DNA under the control of the CaMV 35S promoter, was made. This construct was introduced into Wassilewskija-2 (Ws-2) wild-type plants using a modified in planta transformation method, ‘spray transformation’. Spray transformation yielded a comparable or even a greater number of transformants relative to the traditional ‘floral dip’ (Clough and Bent, 1998) or ‘vacuum infiltration’ methods (Bechtold and Pelletier, 1998). Seeds collected from transformed plants were plated on agar-solidified media to isolate 80 transgenic plants using kanamycin as a selectable marker.

To compare the phenotypic effects resulting from DWF4 overexpression and exogenous addition of BRs, we measured the root and hypocotyl lengths of dwf4, wild-type control, wild-type plants supplemented with 10−7 m 24-epi-brassinolide (epi-BL), and two independent Arabidopsis overexpressors of DWF4 (AOD4) lines, grown in the light or dark (Figure 2a,b). Similar to a previous report (Azpiroz et al., 1998), dwf4-1 displayed reduced hypocotyl length both in the light and dark as compared to the wild type. The reduction in hypocotyl length of dwf4 was more pronounced in the dark, while the root length reduction was more dramatic in the light. The hypocotyl and root length of dwf4 decreased to 12 and 49%, respectively, of those of wild type in the dark; whereas the lengths were 20 and 23% of those of wild type in the light. In light- and dark-grown seedlings of the wild type, the ratios of root to hypocotyl length were 9 and 0.9, respectively, whereas in dwf4 the ratios were 11 and 4, respectively, suggesting that a co-ordinated reduction of root growth failed to occur in dark-grown dwf4.

Figure 2.

Figure 2.

Seedling phenotypes of 7-day-old dwf4-1, wild type, epi-BL-treated wild type and AOD4 lines grown in the light and dark and 14-day-old tobacco seedlings grown in the light.

(a) Quantification of hypocotyl and root length. The mean length of 16 seedlings is displayed with the standard deviation.

(b) Representative seedlings grown in the light (L) and dark (D). Inset, 16-day-old tobacco seedlings with control at left and TOD4 at right. Bars = 1 cm.

Figure 2.

Figure 2.

Seedling phenotypes of 7-day-old dwf4-1, wild type, epi-BL-treated wild type and AOD4 lines grown in the light and dark and 14-day-old tobacco seedlings grown in the light.

(a) Quantification of hypocotyl and root length. The mean length of 16 seedlings is displayed with the standard deviation.

(b) Representative seedlings grown in the light (L) and dark (D). Inset, 16-day-old tobacco seedlings with control at left and TOD4 at right. Bars = 1 cm.

The responses of light-grown hypocotyls and roots to 10−7 mepi-BL were different in the wild type. As compared to untreated controls, the hypocotyl length of epi-BL treated seedlings was 2.2-fold greater, while root length was reduced by 3.5-fold that of the control (Figure 2). In the dark, application of 10−7 mepi-BL resulted in negative growth effects on hypocotyls and roots of wild type, 90 and 40% those of controls. This suggests that 10−7 mepi-BL is beyond the physiologically effective concentration for seedlings grown in the dark.

The growth effects resulting from the overexpression of DWF4 were dramatic, and comparable to the effects observed in seedlings exogenously supplied with epi-BL. Similarly to the epi-BL-treated light-grown seedlings, AOD4-4 exhibited a 2.6-fold increase in hypocotyl length. In addition, both AOD4-4 and AOD4-51 lines also showed retarded root growth in both light and dark. Surprisingly, the hypocotyl length of dark-grown AOD4 seedlings was 1.4-fold greater than that of wild-type hypocotyls without epi-BL treatment, suggesting that the degree of DWF4 transcription could also be correlated to hypocotyl elongation in the dark.

To ascertain whether the Arabidopsis DWF4 enzyme could catalyze BR biosynthesis in tobacco, we generated 15 transgenic tobacco plants (TOD4) harboring the pOD4 construct. As shown in the inset in Figure 2(b), increased hypocotyl length was also observed in TOD4-11 plants. The hypocotyl length of TOD4-11 lines was 1.4-fold greater than that of their control (Figure 2a). However, root growth was not significantly inhibited.

The effect of DWF4 overexpression on plant growth was monitored during the course of development. The number of rosette leaves at bolting (20 days after germination) was not significantly different between wild-type and AOD4 plants, indicating that flowering time was not affected by the activity of the transgene (Table 1). The height of the wild type and the three independent AOD4 lines was comparable at 20 days after germination (DAG) (Figure 3). Later, the AOD4 lines outgrew the wild type (Figure 3, Figure 4a). Surprisingly, AOD4 plants continued to grow beyond 35 DAG at the time wild-type plants ceased elongation. At the termination of flowering, the height of AOD4 lines was approximately 40% greater than that of wild type (Figure 3, Figure 4f). Similarly, TOD4 plants displayed a 14% increase in plant height as compared to the control (Figure 4g). The increased inflorescence length in AOD4 plants appeared to be at the cost of stem rigidity. During development, AOD4 plants tend to fall over earlier than the Ws-2 wild type (data not shown). In addition to plant height, a comparison of rosette leaf size between wild type and AOD4 indicated that petioles and leaf blade length are longer, especially in adult AOD4 plants (Figure 4b,c). TOD4 plants also possessed leaves that were larger, and had longer petioles relative to the control (Figure 4d,e). In addition, both in Arabidopsis (arrows in Figure 4f) and tobacco overexpression lines (arrows in the inset of Figure 4g), additional secondary branches were found. In AOD4-65 and AOD4-73 plants, this additional branching was associated with a more than twofold increase in the number of siliques per plant, leading to an increase in seed weight per plant of 33 and 59%, respectively (Table 1). The increased seed production was mainly due to a greater number of seeds per plant, because seed size increase was within the range of standard deviation (Table 1).

Table 1.  Morphological comparison among control and AOD4 lines
ParameterWs-2AOD4-65AOD4-73Percentage of Ws-2
  • a

    Between rosette leaves and the first silique (N > 12).

  • b

    Length of the second silique on a main inflorescence, 35 days after germination.

  • c

    Number of rosette leaves at bolting, 20 days after germination.

  • d

    Number of branches on all inflorescences bearing siliques at maturity, 35 days after germination (N > 12).

  • e

    Weight of seeds per individual plant (N > 12).

 Height (cm)a11.4 ± 1.815.4 ± 3.316.8 ± 3.8135147
 Siliqueb length (mm)13.7 ± 2.018.4 ± 2.520.0 ± 3.2134145
 Seed length (mm)0.47 ± 0.040.51 ± 0.050.49 ± 0.03108105
 Seed width (mm)0.29 ± 0.030.32 ± 0.030.3 ± 0.03110102
 Rosette leavesc5.1 ± 0.55.2 ± 0.45.7 ± 0.5100112
 Branchesd9 ± 323.4 ± 5.721.6 ± 6.8260240
 Siliques106.7 ± 33.6373.5 ± 107.5344.2 ± 47.1350325
 Seed weighte (mg per plant)164 ± 48218 ± 33261 ± 37133159
Figure 3.

Increased inflorescence height for wild type and three independent AOD4 lines during the course of Arabidopsis development. Mean values of >12 plants with standard deviations are shown.

Figure 4.

Phenotypes of AOD4 and TOD4 plants in later developmental stages.

(a) Comparison of 25-day-old wild-type and AOD4-59 plants. Bar = 2 cm.

(b,c) A set of rosette leaves from 25-day-old wild type (b) and AOD4-59(c) , respectively. Bar = 2 cm.

(d) Comparison of leaves (sixth and seventh) from a 3.5-month-old control and TOD4-11 plants. Bar = 10 cm.

(e) Comparison of 45-day-old control and TOD4-11 plants. Bar = 10 cm.

(f) Phenotype of 5-week-old wild-type and AOD4-65 plants. Increased height and number of branches (arrows) are obvious in the AOD4 plant. Bar = 5 cm.

(g) Comparison of 3-month-old control and TOD4 plants. Bar = 20 cm.

Inset, thick and thin arrows indicate a main stem and secondary branch, respectively, coming out of the base of the main stem. Bar = 10 cm.

Elevated flow of BR biosynthetic intermediates

To understand the relationship between the level of DWF4 gene expression and the rate of BR biosynthesis, we examined the BR levels in dwf4, wild-type and AOD4 plants grown in the light and dark. In light-grown dwf4-1 plants, the levels of the four compounds upstream of the 22α-hydroxylation step, 24-methylenecholesterol (24-MC), CR, CN and 6-oxocampestanol (6-OxoCN), were elevated 1.9-, 1.7-, 1.4- and 1.8-fold, respectively, compared to the levels in the wild type (Figure 1). However, the levels of the downstream intermediates were just barely detectable: 6-deoxocathasterone (6-DeoxoCT), 6-deoxotyphasterol (6-DeoxoTY), typhasterol (TY) and 6-DeoxoCS; or were not detected in comparison to the wild type: 6-deoxoteasterone (6-DeoxoTE), castasterone (CS) and BL. The BR levels in dwf4-1 suggest that a biochemical block exists at the 22α-hydroxylation step, and the blocking is nearly complete. In light-grown AOD4 plants, the levels of DWF4 substrates, CN and 6-OxoCN decreased to 89 and 39% of those of the wild type, respectively, suggesting that they are being used by elevated DWF4 activity. The levels of compounds upstream of CN, including 24-MC and CR, were increased 1.7- and 1.3-fold, respectively. Indicative of facilitated metabolic flow at the 22α-hydroxylation step, the levels of downstream intermediates in the late C-6 oxidation pathway in light-grown AOD4 plants increased: 3.8-fold for 6-DeoxoCT; 4.8-fold for 6-DeoxoTE; 3.5-fold for 6-DeoxoTY; and 1.4-fold for 6-DeoxoCS. The levels of intermediates in the early C-6 oxidation pathway were either decreased; 40% (TY) and 82% (CS) those of wild type; or not detected: cathasterone (CT), teasterone (TE) and BL.

To learn how light affects the metabolic flow in the early and late C-6 oxidation pathways, we analyzed the endogenous levels of BRs in dark-grown seedlings of AOD4-4. The levels of CR and the two immediate precursors of the 22α-hydroxylation reaction, CN and 6-OxoCN, dropped to 72, 80 and 53% that of the wild type, respectively. Conversely, the levels of the downstream compounds were dramatically increased 57.8-fold (6-DeoxoCT), 4.7-fold (6-DeoxoTE), 8-fold (6-DeoxoTY) and 12-fold (6-DeoxoCS) in the late C-6 oxidation pathway, and 1.5-fold (typhasterol, TY) and 1.6-fold (CS) in the early C-6 oxidation pathway. Again, the BL level was lower than the detection limit in dark-grown seedlings.

Surprisingly, we were also able to detect novel 22α-hydroxylated steroids such as 22α-hydroxycampesterol (22-OHCR) (22S,24R)-22-hydroxyergost-4-en-3-one, and (22S,24R)-22-hydroxy-5α-ergostan-3-one (Figure 1 and data not shown). 22-OHCR was detected from both wild type and AOD4, with a dramatic increase in dark-grown seedlings; the level of 22-OHCR in dark-grown AOD4-4 seedlings was 488-fold greater compared to the wild type. Detection of 22 α-hydroxylated steroids implies that DWF4 has a broader range of substrate specificity than just CN and 6-OxoCN. Thus two enzymes, DWF4 and a Δ54 isomerase, can act on CR, leading to one of two pathways: either the early C-22 hydroxylation reaction by DWF4, or the late C-22 hydroxylation reaction initiated by a Δ54 isomerization reaction (Figure 1). Furthermore, we were able to detect another novel compound, 3-epi-6-DeoxoCT. The levels in dark-grown wild-type and AOD4-4 seedlings were 0.75 and 24 ng g−1 FW, respectively, and the levels in light-grown mature plants of wild type and AOD4-73 were 1.5 and 11 ng g−1 FW, respectively. Detection of 3-epi-6-DeoxoCT in both wild-type and AOD4 plants suggests that the 3-epimerization could occur before a conventional C-23 hydroxylation reaction (Figure 1 and Fujioka et al., 2000).

Altered transcript levels of BR-related genes in different DWF4 genetic backgrounds

As modification of DWF4 transcription levels led to changes in phenotype as well as endogenous BR levels, we tested whether the altered BR levels affect the transcript levels of BR-related genes using reverse transcriptase PCR (Figure 5). Five genes were examined: DWF4, DET2 and CPD, all of which code for BR biosynthetic enzymes; TCH4, the transcript level of which is upregulated by exogenous BR application (Xu et al., 1996); and BAS1, thought to be involved in BR degradation (Neff et al., 1999). First, the steady-state level of the DWF4 transcript was strikingly increased in AOD4 as compared to that of the wild type (Figure 5a). This suggests that the introduced CaMV 35S promoter successfully overexpresses the DWF4 gene as intended. The DWF4 transcript was also detected in the dwf4-1 knock-out allele that carries a T-DNA insert near the 3′ end of the gene (Choe et al., 1998). The presence of this transcript is due to the position of RT–PCR primers that were designed to amplify the 5′ end of the gene. The level of TCH4 was significantly increased in AOD4 compared to both wild-type and dwarf mutants. The levels of BAS1 transcripts were not detectable in dwf4, decreased in bri1, and slightly increased in AOD4 relative to the wild type. Both DET2 and CPD transcript levels were increased in the dwf4 and bri1 mutant plants, suggesting that decreased BR biosynthesis (dwf4) and deficiency of BR signaling (bri1) failed to feedback downregulate the transcript levels of DET2 and CPD. However, the levels of DET2 and CPD transcripts were not significantly affected in the AOD4 lines.

Figure 5.

RT–PCR detection of BR-related genes.

(a) RT–PCR detection of DWF4, TCH4, BAS1 and an ACT2 control in wild-type, dwf4-1, AOD4-4 and bri1-5 plants.

(b) RT–PCR detection of two BR biosynthetic genes, DET2 and CPD, which are upstream and downstream, respectively, of DWF4 in the BR biosynthetic pathways, and an ACT2 control in wild-type, dwf4-1, AOD4-4 and bri1-5 plants.


Studies on the abnormal development of dwarf mutants defective in BR biosynthesis and signaling led to the conclusion that the growth effects contributed by BRs are essential for proper growth and development of plants (Clouse and Feldmann, 1999). In this paper we show that ectopic overexpression of DWF4 resulted in (i) an increase in both hypocotyl and inflorescence height; (ii) prolonged flowering; (iii) an increased number of silique-bearing branches, resulting in an increase in seed yield; and (iv) increased BR levels (Table 1; Figures 3 and 4). These phenotypes are probably due to the elevated activity of the rate-limiting step enzyme, and thus increased BR biosynthesis.

The transcriptional activation of DWF4 using the 35S promoter resulted in larger plants in both light and dark (Figures 2 and 5). This phenotype is probably due to increased endogenous BR levels, as it mimics the morphology of plants exogenously supplied with epi-BL. In the dark, AOD4 hypocotyls showed elongation beyond the wild type, suggesting that plants retain the capacity for further cell elongation on top of that caused by normal etiolation. By contrast, wild-type seedlings exogenously supplied with epi-BL displayed a reduction in hypocotyl length, possibly due to toxic effects of excess BL. This negative growth effect suggests that BL was beyond the concentration that plants can efficiently handle, or that the presence of the end product BL alone is not enough to activate a degradation pathway.

The root/shoot ratios of dark-grown seedlings of wild type, dwf4-1, epi-BL-treated wild type and AOD4-51 are 0.9, 3.6, 0.5 and 0.1, respectively. The increased root/shoot ratio in dwf4-1 and decreased ratio in AOD4-51 suggest that limited availability of endogenous BRs de-repressed root growth in dwf4-1, whereas the increased BRs repressed growth in AOD4-51. Thus the level of endogenous BRs is tightly controlled to meet the growth and developmental needs of shoots and roots. Understanding the mechanisms that control the temporal and spatial endogenous BR levels, especially as contributed by tissue-specific synthesis and degradation, is a prioritized area for future study.

The increased seed production in AOD4 plants was positively correlated with an increased number of silique-bearing branches. The number of branches and siliques was greater in both AOD4-65 and AOD4-73 as compared to the wild type: 2.6- and 2.4-fold, respectively. Currently, it is not understood how the increased BR levels in AOD4 lead to prolonged growth and an overproduction of silique-bearing branches.

Analyses of the endogenous BRs in dwf4-1 revealed that the levels of BR intermediates after 22α-hydroxylation are either significantly lower or are undetectable in both late and early C-6 oxidation pathways (Figure 1). In both wild-type and AOD4 lines, the levels of biosynthetic intermediates in the late C-6 oxidation pathway are dramatically higher compared to the mostly undetectable levels in the early C-6 oxidation pathway. This suggests that the late C-6 oxidation pathway is the primary route leading to the end-product BL. Alternatively, it is possible that the lower levels of intermediates in the early C-6 oxidation pathway could be due to a greater turnover rate compared to the intermediates in the late pathway. However, because these data give only a snapshot of the levels, we are unable to determine if the two pathways employ different turnover rates. It will be important to investigate the activity of enzymes participating in the turnover pathway in AOD4 lines. Further, our data show that the metabolic flow of the intermediates in the late C-6 oxidation pathway is active in dark-grown seedlings. This is interesting given that the intermediates in the early C-6 oxidation pathway exhibit greater biological activity on etiolated tissues than those in the late C-6 oxidation pathway (Choe et al., 1998; Fujioka et al., 1997). To confirm the effects of the early C-6 oxidation pathway on etiolated seedlings, it is desirable to engineer plants for an increase in the activity of both C-6 oxidase and C-22α-hydroxylase. Such a transgenic plant may be able to direct the metabolic flow specifically to the early C-6 oxidation pathway, and possibly result in a phenotype other than AOD4.

Considering the enhanced BR effects in AOD4 plants attributable to elevated levels of endogenous BRs, undetectable levels of BL in both dark- and light-grown AOD4 plants were surprising. Possible explanations for this may include: BL production from CS did not occur in AOD4 plants; and/or the elevated BR effects seen in AOD4 are due to the bioactivity of BR biosynthetic intermediates preceding BL. This is based on the assumption that the precursors of BL also retain biological activity per se. To date, because there are no mutants in the steps downstream of CPD, it cannot be confirmed that the compounds preceding BL have nascent biological activity, or whether they start to possess activity after they are completely converted to BL. Studies by Bishop et al. (1999) suggest that the intermediates in the late C-6 oxidation pathway are not active BRs, because the tomato dx mutant accumulates significant amounts of 6-DeoxoCS in the late pathway, due to a block in 6-oxidation. Furthermore, we found that in the BR signaling mutant bri1 the endogenous levels of BL and its precursors were dramatically increased, indicating an active conversion of CS to BL in Arabidopsis (Noguchi et al., 1999b). Thus a more convincing explanation for the lack of BL accumulation is that the excess BL in AOD4 plants was rapidly metabolized in subsequent degradation pathways. The GA20-oxidase overexpression lines did not show a significant difference in the levels of endogenously active GA4, despite increased GA effects in the transgenic plants (Huang et al., 1998). Considering the dramatic increase in BL level in bri1, and an undetectable BL level in AOD4 plants, it appears that the phenotypic effects induced by BL signaling and the endogenous levels of BL are inversely correlated. Once BL is used by the signaling pathway it is rapidly degraded; however, when BL is not perceived by BRI1 the level remains high. We found that the transcript level of the gene coding for a candidate enzyme of BL degradation, BAS1 : CYP72B1, is down-regulated in BL-accumulating bri1, whereas its level is slightly increased in AOD4 plants (Figure 5).

Our analyses of sterol levels in dwf4-1, wild-type and AOD4 plants suggest that modulation of BR biosynthesis leads to a concomitant modification of phytosterol biosynthesis. Phytosterols are of great interest to human health. They have been shown to be effective in lowering serum cholesterol levels, resulting in reduced risk of arteriosclerosis (Jones et al., 1997). Future genetic engineering of crop plants using the BR and sterol biosynthetic genes should pursue modulation of the content and quantity of phytosterols, as well as increasing seed yield.

Because of a critical role of phytohormones in the plant life cycle, the processes of phytohormone biosynthesis and signaling have been a target for genetic engineering toward the development of desired crop traits. For instance, a substantial increase in wheat yield during the 1960s and 1970s ‘green revolution’ was possible because of the selection of a genetic variant shown to be less sensitive to gibberellins (Peng et al., 1999). It was proposed that increased seed yield was at the expense of vegetative biomass. We have found that bri1-5 displayed only 14% of wild-type height, but bears 45% the number of seeds of wild type (Noguchi et al., 1999b). This suggests that, like a reduced sensitivity to gibberellins, a reduced sensitivity to BRs could be utilized to increase seed yield at the expense of vegetative biomass. On the other hand, increased GA signaling in the spy mutant resulted in longer hypocotyls and increased stem elongation similar to AOD4 (Jacobsen and Olszewski, 1993). However, unlike AOD4, spy exhibits light green leaves, earlier flowering, parthenocarpy and partial male sterility. More agriculturally useful traits attributable to increased GA signaling were found in transgenic plants overexpressing the GA20-oxidase gene encoding a rate-limiting step enzyme in GA biosynthesis. The transgenic plants showed both rapid stem elongation and increased seed setting (Huang et al., 1998). Similarly, we found that overproduction of BRs could lead to an increase in both organ size and seed yield simultaneously.

Many of the traits enhanced in AOD4 plants open up the possibility of utilizing DWF4 for improving crop yield (Figures 3 and 4). Engineering of the BR metabolic pathways in conjunction with the genes involved in GA biosynthesis or signaling pathways may result in synergistic effects on plant performance.

Experimental procedures

Plant growth and BR analysis

To examine the growth of seedlings, Arabidopsis and tobacco seeds were surface-sterilized in a microcentrifuge tube by treating with 70% ethanol for 2 min, then for 15 min with a bleach solution consisting of 5% Clorox and 1% SDS, and finally by rinsing three times with sterile water. The seeds were resuspended with 0.15% top agar before pouring onto plates containing agar-solidified nutrient media [1 × Murashige and Skoog (MS) salts, 0.5% sucrose, 0.8% agar pH 6]. For growth in the dark, plates were wrapped with three layers of aluminum foil after 3 h light treatment, then grown under the same growth conditions as light-grown plants [16 : 8 light (200 µmol m−2 sec−1) : dark, 22 and 21°C, respectively, and 70–90% humidity]. For treatment with epi-BL, a 10−4 m stock solution in 95% ethanol was added before the media solidified. Epibrassinolide (product number E1641) was purchased from Sigma (St Louis, MO, USA). Hypocotyl and root lengths were measured on day 7 for Arabidopsis and on day 14 for tobacco.

Endogenous BRs were analyzed in 1-week-old, dark-grown seedlings of Ws-2 (17 g FW) and AOD4-4 (14.8 g FW); and 100 g FW each of 4-week-old light-grown Ws-2, AOD4-4 and 5-week-old light-grown dwf4-1 plants. Methods for chemical treatment and GC–MS have been described previously (Choe et al., 1999b). The endogenous BR levels of the wild type were taken from a previous publication (Choe et al., 1999b).

RT–PCR detection of BR-related gene transcripts

Total RNA was isolated from 3-week-old whole plants of Ws-2 and AOD4-4 and 4-week-old plants of dwf4-1 and bri1-5 using an RNeasy plant kit (Qiagen, Santa Clarita, CA, USA). After removing genomic DNA by digesting the total RNA with DNAase I (Boehringer Manheim-Roche, Indianapolis, IN, USA), the RNA was re-extracted and optically quantified. Total RNA (2 µg) was run on a formaldehyde-denaturing 1.1% agarose gel to confirm visually the quality and quantity of RNA. For each sample, 6 µg of total RNA was subject to reverse transcription (RT) using Retroscript RT kit (Ambion, Austin, TX, USA). The 50 µl PCR mixtures consisted of 1× Platinum Taq polymerase buffer (Life technologies, Rockville, MD, USA), 3 µl RT product, 0.2 mm of each dNTP, 2 mm MgCl2, 0.4 µm each of the forward and reverse primers, and 5 U of the Platinum Taq polymerase. The PCR program was composed of the initial denaturation for 4 min at 94°C, amplification by 28 cycles of 30 sec at 94°C, 30 sec at 55°C, and 1 min at 72°C, and followed by final elongation for 10 min at 72°C. Oligonucleotide sequences used for RT–PCR analysis (from 5′ to 3′) are D4RTF, TTCTTGGTGAAACCATCGGTTATCTTAAA; D4RTR, TATGATAAGCAGTTCCTGGTAGATTT; TCH4-F0, AAT GCCTCGAAACAGGGGACTAC; TCH4-R, GTAACAAAGAGAATA TTATAACTTTAACGAACTA; BAS1-F, TAATCTGCTGACGTGGAC GACCATCTT; BAS1-R, TGATAAGTAGGAGCCAAGTGAAAG GTGAA; DET2-F1, CATCGCCGCCTTGGCTTTCACCTT; DET2-R1, CCGGTTACTGGAAATTTTGACAAAGGTAGT; CPD-F1, GGCTAG GGTTGCACTCTCTGT; CPD-R1, CACAGAAGCTAAGGCTTTATA. To enhance the resolution between different genes, the template amount was adjusted following examination of the pilot PCR analysis. 10 µl each of the PCR products was run on a 1.8% agarose gel and transferred to nylon membrane for further visualization using the Southern method (Figure 5a), or pictures of ethidium bromide-stained gels were taken (Figure 5b). The membrane was hybridized with 32P-labeled probes of the cDNA: DWF4 (GenBank accession no. AF044216); TCH4 (AF051338); BAS1 (AC003105); ACT2 (U37281) at 65°C for 4 h, and washed for 1 h at 67°C with washing solution (0.1 × SSC and 0.1% SDS). The GenBank accession numbers of DET2 and CPD are U53860and X87367, respectively.

Overexpression constructs

For a DWF4 overexpression construct, PCR products were made using D4OVERF (5′-GaattctagaATGTTCGAAACAGAGCATCATA-3′) and D4R2 (5′-CCGAACATCTTTGAGTGCTT-3′) primers and Ws-2 genomic DNA as a template. The PCR products were restricted with XbaI and HindIII. The fragment was inserted into the same restriction sites as the genomic clone SCH25 containing a 2.5 kb HindIII fragment of the DWF4 genomic DNA corresponding to the 3′ half of the gene (Choe et al., 1998). The resulting recombinant DNA clone pD4CDS, containing the whole coding sequence from the translation initiation site to 694 bp downstream of the stop codon, was restricted with XbaI and transferred to bridge vector pART7, then to a binary vector pART27 (Gleave, 1992). The resulting binary construct was named pOD4. This construct was introduced into Agrobacterium by electroporation.

Spray transformation

As it has been shown that Agrobacterium-mediated transformation can work by seed infection (Feldmann and Marks, 1987) or by simply dipping the host plants into an Agrobacterium culture (Clough and Bent, 1998), we decided to try spraying the Agrobacterium directly onto the plants. In addition to ‘spraying’, the ‘floral dip’ method was used as described (Clough and Bent, 1998). About 20 Ws-2 wild-type seeds were sprinkled on 10 cm pots with pre-wet Metromix 350 (Grace Sierra, Milpitas, CA, USA) without prior covering with cheesecloth. The plants were thinned to five or six plants per pot 10 days after germination. When the primary inflorescences of the wild type reached 3–4 cm high they were decapitated to induce axillary bolts. For the preparation of Agrobacterium, a single colony selected on 60 µg ml−1 kanamycin in Luria–Bertani (LB) medium (10 g bacto-tryptone, 5 g bacto-yeast extract, 10 g l−1 NaCl, pH 7) was inoculated into 100 ml liquid LB media and grown for 3 days. 100 µl of the 3-day-grown cells (OD600 = 1) were used to inoculate fresh LB media (100 ml). The overnight-grown cells were collected by centrifugation, and resuspended with transformation media (5% sucrose and 0.05% Silwet L-77) as described by Clough and Bent (1998), and the concentration was adjusted to OD600 = 1. On the third day after decapitation, 50 ml of the Agrobacterium suspension (OD600 = 1) was sprayed onto six to eight pots of plants. To avoid physical contact with possibly hazardous Silwet vapor, protective glasses were used and the spraying was done in a fume hood. Plants were sprayed every third day for one week. Sprayed plants were grown to maturity and seeds harvested.

For seed sterilization, 60 mg seeds (approximately 3000 seeds) were surface-sterilized in 50 ml Falcon tubes. To plate the seeds, 25 ml of sterile top agar (0.15% agar in water) was added to the sterilized seeds and the seed mixture was poured equally onto three agar-solidified MS plates (100 × 15 mm) supplemented with kanamycin at 60 µg ml−1. Twelve days after germination, kanamycin-resistant seedlings were transferred to small pots and grown to maturity. T2 seeds were collected from individual transformants (T1) and plated again on the selection media to determine segregation ratios for kanamycin-resistant versus kanamycin-sensitive plants. Arabidopsis transformants harboring an overexpression construct pOD4 were named AOD4. The transgenes were determined to be homozygous when no sensitive T4 seedlings segregated from T3 individual plants. Morphometric analysis of AOD4 lines was performed using plants homozygous for the transgene.

Methods for Arabidopsis growth and morphometric analysis were as previously described (Choe et al., 1999b). Briefly, seeds of wild type and the two AOD4 lines were germinated on MS agar. At 10 DAG, 20 seedlings confirmed to be resistant to kanamycin were transferred to a single pot. Various morphological traits (Table 1; Figures 24) were measured. The two lines with distinguishably longer hypocotyls used for Figure 2 were chosen from among 40 randomly selected lines, most of which had long hypocotyls. The quantitative data in Figure 3 represent one line (AOD4-65) that had a long hypocotyl as in Figure 2, but the other two lines (AOD4-71 and AOD4-73) were chosen because they had hypocotyl lengths indistinguishable from the wild type. To determine seed yield, plants were dried for 2 weeks at room temperature. Seeds were harvested from individual plants and weighed. To measure seed size, harvested seeds were magnified 3× under a dissecting microscope, and the width and the length of five seeds from each plant (total >60 seeds per each line) were measured to the nearest 0.1 mm.

Transgenic tobacco plants (TOD4) harboring the pOD4 construct were produced in the Plant Tissue Culture Laboratory at the University of Arizona. Fifteen independent transformants for both control and pOD4 constructs were grown for seeds. One of the TOD4 lines showing the long hypocotyl phenotype was selected for root and shoot length measurement in Figure 2(a). Pictures of the TOD4 lines shown in Figure 4 were taken from T2 plants grown in the greenhouse.


We thank Frans Tax and Roger Pennell for critical reading of this manuscript, Byeong-ha Lee for technical assistance with cDNA cloning, and Randy Ryan and staff at the Plant Tissue Culture Laboratory, University of Arizona for generating transgenic tobacco plants. S.F. acknowledges support from the Ministry of Education, Science, Sports, and Culture of Japan (a Grant-in-Aid for Scientific Research (B), No. 10460050). K.A.F. acknowledges NSF Grant No. 9604439 and the Human Frontiers in Science Program (KAF).