Molecular tailoring of farnesylation for plant drought tolerance and yield protection

Authors


(fax +1 613 545 3618; e-mail huangy@performanceplants.com).

Summary

Protecting crop yield under drought stress is a major challenge for modern agriculture. One biotechnological target for improving plant drought tolerance is the genetic manipulation of the stress response to the hormone abscisic acid (ABA). Previous genetic studies have implicated the involvement of the β-subunit of Arabidopsis farnesyltransferase (ERA1) in the regulation of ABA sensing and drought tolerance. Here we show that molecular manipulation of protein farnesylation in Arabidopsis, through downregulation of either the α- or β-subunit of farnesyltransferase enhances the plant's response to ABA and drought tolerance. To test the effectiveness of tailoring farnesylation in a crop plant, transgenic Brassica napus carrying an ERA1 antisense construct driven by a drought-inducible rd29A promoter was examined. In comparison with the non-transgenic control, transgenic canola showed enhanced ABA sensitivity, as well as significant reduction in stomatal conductance and water transpiration under drought stress conditions. The antisense downregulation of canola farnesyltransferase for drought tolerance is a conditional and reversible process, which depends on the amount of available water in the soil. Furthermore, transgenic plants were more resistant to water deficit-induced seed abortion during flowering. Results from three consecutive years of field trial studies suggest that with adequate water, transgenic canola plants produced the same amount of seed as the parental control. However, under moderate drought stress conditions at flowering, the seed yields of transgenic canola were significantly higher than the control. Using protein farnesyltransferase as an effective target, these results represent a successful demonstration of engineered drought tolerance and yield protection in a crop plant under laboratory and field conditions.

Introduction

Drought conditions occur ubiquitously during the growing season of many plants, and in the case of crops, it can have a profound negative effect on agricultural productivity. For example, in corn, a mild drought of 4 days at the flowering and silking stage of development can result in up to a 50% decrease in seed production (Claassen and Shaw, 1970). Plants have evolved to various levels of adaptation in response to water stress conditions, and of these, the best understood is the modulation of stomatal aperture by the phytohormone abscisic acid (ABA) to reduce transpirational water loss. Under water stress conditions, the endogenous level of ABA increases, which, through a complex signaling cascade, results in stomatal closure (Blatt, 2000). When water relations return to optimal conditions for growth, ABA concentration decreases to reverse the process. Thus, the ability to regulate ABA synthesis and response makes this hormone an excellent target for improving drought tolerance in crop species.

One approach to identify genes that regulate ABA synthesis or signaling in crops is to use mutational analysis in a model genetic system and then transfer this information to a crop species. Arabidopsis thaliana is the model of choice because mutational and molecular analysis of target genes is experimentally tractable. ABA normally inhibits seed germination and early seedling development in Arabidopsis. Thus, genetic screens for mutations that alter the sensitivity of the plant to ABA at the germination level are relatively straightforward. A large number of genes have now been identified that either increase or decrease the plant's response to the hormone (Finkelstein et al., 2002). Of these, ERA1 and ABH1 are of particular interest, as loss-of-function alleles of these genes increase sensitivity of the guard cells to ABA (Cutler et al., 1996; Hugouvieux et al., 2001; Pei et al., 1998). As a consequence, era1 or abh1 mutants show reduced wilting during drought stress. In the presence of exogenous ABA, the leaves of era1 display the tightest stomatal aperture closure. Patch clamping of the guard cells isolated from the complete loss-of-function era1-2 mutant indicated enhanced ABA sensitivity of plasma membrane calcium channel currents (Allen et al., 2002; Hugouvieux et al., 2002). ERA1 encodes the β-subunit of Arabidopsis farnesyltransferase (AtFTB), suggesting that a negative regulator of ABA sensitivity must be farnesylated to modulate ABA response in Arabidopsis (Cutler et al., 1996). All plant farnesyltransferases identified to date consist of a heterodimer of an α- and a β-subunit, each of which belongs to a single-gene family. In addition to farnesyltransferase, a related enzyme, geranylgeranyltransferase type I, is involved in protein prenylation (Galichet and Gruissem, 2003; Rodríguez-Concepción et al., 1999). Geranylgeranyltransferase type I is also a heterodimer that shares a common α-subunit with farnesyltransferase, but has a distinct β-subunit (Caldelari et al., 2001). A recent study indicates the involvement of the Arabidopsis α-subunit (AtFTA) in meristem growth and development in Arabidopsis (Running et al., 2004).

Although loss-of-function of ERA1 in Arabidopsis results in reduced water loss, this mutation also causes severe pleiotropic phenotypes in growth and development, including delayed senescence, altered structures in apical and floral meristem, increased number in floral organs and dramatic reduction in seed yield even under normal growth conditions (Donetta et al., 2000; Yalovsky et al., 2000; Ziegelhoffer et al., 2000). Thus, the usefulness of AtFTB as a biotechnological target for yield protection under drought conditions was previously unproved. In this study, we demonstrated that downregulation of either AtFTA or AtFTB resulted in both increased ABA sensitivity and improved drought stress tolerance in Arabidopsis. Furthermore, use of a drought-inducible Arabidopsis rd29A promoter to drive the antisense expression of AtFTB in canola conferred similar drought protection in this species. Transgenic canola plants showed enhanced ABA response during seedling development. In addition, under drought conditions, ABA-hypersensitive plants showed greater reduction in stomatal conductance and leaf transpiration than the parental non-transgenic plants. We demonstrated that the downregulation of targeted gene expression driven by this drought-inducible promoter is a conditional, reversible process that is controlled by the available soil water content. More importantly, results obtained from three consecutive years of field trials indicated that in the field under well-irrigated conditions, these transgenic plants showed similar or greater seed yield than the parental plants. When irrigation was reduced at the peak of flowering, transgenic canola showed significantly greater seed yield than the parental control. Together, these results demonstrate that a novel regulation system through genetic manipulation of plant farnesyltransferases for improving drought tolerance and seed yield protection in Arabidopsis and canola model systems and may be useful in a variety of crop plants.

Results

Increased ABA sensitivity and drought tolerance by downregulation of AtFTA

We obtained a total of 12 independent, single-locus insertion homozygous Arabidopsis lines containing the 35S-driven, anti-AtFTA antisense construct. ABA tests with the T3 homozygous seeds indicated that seven of the lines showed increased inhibition of seedling development by the hormone compared with the wild type (data not shown). Three ABA hypersensitive lines, FTA1, FTA2 and FTA3, were subjected to further molecular and physiology studies. Under optimal growth conditions, transgenic lines were phenotypically similar to the wild-type plants, except for a slight delay in the emergence of the first flowers (1–2 days). Northern blot analysis with total RNA isolated from 2-week-old seedlings showed the AtFTA antisense transcript was present in the three transgenic lines, but was absent in the wild-type control (Figure 1a). The level of AtFTA sense transcript was reduced in the transgenic lines compared with the wild-type (Figure 1a). Transgenic seedlings had arrested development at the mid-cotyledon stage after growing in 0.5x MS medium containing 0.5 μm ABA for 2 weeks, and failed to develop the first pair of true leaves (Figure 1b). The length of the lateral roots in the transgenic lines was also substantially shorter than that of the wild-type control, indicating that reduction in the AtFTA transcript results in enhanced ABA response. However, as shown in Figure 1(b), seedling development in the antisense lines was less inhibited than that of era1-2.

Figure 1.

35S-driven, antisense downregulation of AtFTA results in increased ABA sensitivity and enhanced drought tolerance in Arabidopsis.
(a) Northern analysis of AtFTA transcript levels in 2-week-old seedlings of three independent homozygous AtFTA-antisense lines (FTA1, FTA2 and FTA3) and in wild-type (WT) Columbia. The upper panel shows the AtFTA antisense transcripts detected by a single-stranded, sense RNA probe of AtFTA. The middle panel shows the sense AtFTA transcripts detected by a single-stranded, antisense RNA probe on the same, stripped blot.
(b) Wild-type Columbia, era1-2, FTA1, FTA2 and FTA3 plants were grown on 0.5x MS agar plates in the presence of 0.5 μm abscisic acid for 2 weeks.
(c) Comparison of soil water content of the three independent AtFTA-antisense lines (FTA1, FTA2 and FTA3 are in closed circle, diamond and triangle, respectively) and wild-type Columbia (in open circle) during a drought treatment. The drought treatment began when the plants started to flower.
(d) Water loss during the first 5 days of drought treatment divided by the final shoot dry weight (SDW) in wild-type Columbia and the three antisense lines.
(e) Comparison of wild-type Columbia and the three transgenic lines after 7 days of drought treatment. Error bars represent standard errors (n = 8).

To determine whether FTA1, FTA2 and FTA3 were more drought stress tolerant than wild-type Columbia, plants were grown to the start of flowering, then subjected to a drought stress treatment in 4 in pots. In this study, soil water content was calculated as a percentage of the initial level, after subtracting soil and pot weights from the total weight. As shown in Figure 1(c), all three transgenic lines exhibited higher soil water content than the wild-type control during the course of the stress, suggesting that antisense AtFTA lines lose less water than the control under limited water conditions. Furthermore, to account for any difference in soil water content due to any subtle difference in plant size, the ratio of total water loss in the first 5 days of drought treatment divided by final shoot dry weight was determined. The results indicated that the transgenic lines had significantly less (P < 0.05) water loss per unit of shoot weight than that of the control (Figure 1d). As there was no difference in stomatal density of the rosette leaves between the transgenic and the control plants (M. Kuzma and Y. Huang, unpublished data), the lower water loss in the FTA lines was likely a result of reduction in leaf transpiration through stomates. On day 7 of drought treatment, shoots of the AtFTA antisense lines appeared greener and more turgid than those of the wild-type plants (Figure 1e), indicating that they are more drought stress tolerant.

Increased ABA sensitivity and drought tolerance by downregulation of AtFTB

We obtained 12 independent, single-locus insertion homozygous Arabidopsis lines of the 35S-driven AtFTB hairpin construct. ABA test results indicated that eight lines showed enhanced inhibition of seedling development by exogenous application of the hormone (data not shown). Two ABA hypersensitive lines, FTB1 and FTB2, were subjected to further molecular and physiology studies. Northern blot analysis with total RNA isolated from 2-week-old seedlings detected no AtFTB transcript in mutant era1-2, and the transcript level was markedly reduced in the two transgenic lines compared with the wild type (Figure 2a). Similar to the ABA response observed in the 35S anti-AtFTA lines, these hairpin AtFTB lines also exhibited intermediate inhibition of seedling development between wild-type Columbia and era1-2 in the presence of 0.5 μm ABA (Figure 2b).

Figure 2.

Downregulation of AtFTB (ERA1) results in increased abscisic acid (ABA) sensitivity and enhanced drought tolerance in Arabidopsis.
(a) Northern analysis of AtFTB transcript levels in wild-type (WT) Columbia, era1-2, and two hairpin AtFTB lines (FTB1 and FTB2). The blot was hybridized with a double-stranded, AtFTB cDNA probe.
(b) Wild-type Columbia, era1-2, FTB1 and FTB2 plants were grown on 0.5x MS agar plates in the presence of 0.5 μm ABA for 2 weeks.
(c) Water loss during the first 2 days of drought treatment divided by the final shoot dry weight (SDW) in wild-type Columbia, FTB1 and FTB2. Error bars represent standard errors (n = 6).

A growth study under well-watered conditions indicated that the hairpin AtFTB plants had a slight delay in flowering time (1–2 days), an increase in final shoot dry weight, and a slight decrease in seed yield (data not shown). In this drought study, FTB1 and FTB2 plants were more efficient in water usage, as their total water losses in the first 2 days of drought, divided by their final shoot dry weights, were significantly lower (P < 0.05) than that of the wild type (Figure 2c). As there was no reduction in plant size and there was no statistically significant change in stomatal density in the rosette leaves of the transgenic plants (M. Kuzma and Y. Huang, unpublished data), the lower water loss in the hairpin AtFTB plants was likely a result of a reduction in leaf transpiration.

Conditional downregulation of FTB in canola results in enhanced ABA response, reduced stomatal conductance and leaf transpiration, and increased drought tolerance

The double-haploid Brassica napus cv. DH12075 was used as the parent for all transgenic experiments. Transgene effects were evaluated in this double-haploid canola variety because of its stable and highly homogeneous genetic background, as well as its high seed yield and disease tolerance in the field (G. Rakow and J. Ying, personal communication). We had previously cloned a near-full-length B. napus FTB (BnFTB) cDNA (data not shown), and results of pairwise alignment analysis using the Clustal W program (EMBL-EBI) indicated that the BnFTB gene is highly homologous to AtFTB, showing 88 and 84% identity in nucleotide and amino acid sequence levels, respectively. A construct containing the full-length antisense AtFTB gene driven by the drought-inducible rd29A promoter was transformed into DH12075, and we obtained 10 homozygous, single-locus insertion lines for evaluation (data not shown). These transgenic lines were designated as Yield Protection Technology (YPT) lines.

To determine whether introduction of this construct into canola would result in a change in ABA response upon seed germination, ABA tests were performed with the T4 homozygous seeds of the 10 transgenic lines. Six of the transgenic lines showed various levels of enhanced ABA sensitivity (data not shown) but YPT1 and YPT2 were most sensitive. In the absence of exogenous ABA, germination rates of DH12075, YPT1 and YPT2 seeds were close to 100%. In the presence of 10 μm ABA, most YPT1 and YPT2 seeds were able to germinate, however, they were severely inhibited in subsequent seedling development (Figure 3). After growing for 11 days in ABA containing medium, germination rates of YPT1, YPT2 and DH12075 seeds were 81, 86 and 95%, respectively. At this stage, only 22 and 21% of YPT1 and YPT2 seeds started to form cotyledon leaves, whereas 80% of the DH12075 seed had formed green cotyledon leaves. The cotyledon leaves of the DH12075 seedlings were fully expanded, as opposed to partial emergence and formation of cotyledon leaves in the YPT1 and YPT2 seedlings (Figure 3). The development of the root system in YPT1 and YPT2 seedlings was also inhibited compared with the roots of DH12075.

Figure 3.

Enhanced inhibition of early seedling development of the rd29A:anti-AtFTB canola by exogenous abscisic acid (ABA). Seeds of DH12075 canola (WT) and two independent, homozygous transgenic lines (YPT1 and YPT2) were surface-sterilized and plated on 0.5x MS medium containing 10 μm of ABA. Shown here is one of the three replicates of the experiment.

We performed Northern blot analysis to examine the levels of the antisense AtFTB and the endogenous sense BnFTB transcripts under alternating water-sufficient and water-deficient conditions to determine whether or not the regulation of the targeted gene expression was a conditional and reversible process. As expected, no AtFTB antisense transcript was detected in the DH12075 leaf tissues throughout the course of the water treatment (Figure 4a). The level of AtFTB antisense transcript was undetectable in YPT1 and YPT2 during the water-sufficient days prior to drought stress, but reached the highest level on day 3 and day 4, respectively, during the drought treatment. No antisense message was detected when the drought-treated plants were re-watered (days R1–R3). The level of endogenous sense BnFTB transcript in the YPT1 and YPT2 plants was similar to that observed in DH12075 prior to drought stress, however, under water stress conditions, they were markedly lower than in DH12075 (Figure 4b). These decreases resulted from a reduction of the BnFTB transcript in YPT1 and YPT2 plants, while the level of the BnFTB transcript in DH12075 remained relatively constant over the drought treatment (data not shown). After re-watering, the level of sense BnFTB in YPT1 and YPT2 recovered to the level of DH12075 (Figure 4b). Therefore, the increase in accumulation of the AtFTB antisense transcript in YPT1 and YPT2 coincided with the decrease in the level of endogeneous BnFTB transcript, suggesting effective downregulation of FTB in YPT lines by the rd29A promoter-driven antisense construct during drought stress.

Figure 4.

Molecular and physiological responses of the rd29A:anti-AtFTB canola during an alternating water-sufficient, water-deficient and water-sufficient treatment cycle. Results shown in (a) and (b) are from the same experiment in which leaf samples were taken daily, starting a day before drought stress (−1), continuing during a 5-day drought treatment (1–5), finishing with a 3-day re-watering (R1–R3). The quantified values of BnFTB transcript in transgenic YPT1 and YPT2 were divided by the corresponding value of DH12075 on the same day of treatment (b). Panels (c), (d) and (e) represent data from an independent gas-exchange experiment conducted in a growth chamber. (a) AtFTB antisense transcripts detected in leaves of DH12075 (WT) canola, YPT1 and YPT2 during the treatment cycle. (b) Relative level of the endogenous Brassica napus FTB transcript detected during the treatment. Stomatal conductance (c), leaf transpiration (d) and leaf photosynthesis (e) were measured daily on the fifth leaf of DH12075 canola (solid circle), YPT1 (diamond) and YPT2 (square) by a Li-6400 Open Flow gas-exchange system over the treatment cycle. Error bars represent standard errors (n = 8).

To correlate the downregulation of BnFTB in YPT canola with a reduction in stomatal conductance and transpiration, gas-exchange measurements were performed on plants at the sixth leaf stage over a course of alternating water-sufficient and water-deficient treatments. Under water-sufficient conditions, there was no significant difference in stomatal conductance, leaf transpiration, and photosynthesis between YPT1, YPT2 and DH12075 (Figure 4c–e). Over the course of water stress, stomatal conductance of YPT1 and YPT2 was lower than in DH12075 (Figure 4c). In particular, on day 2 of the treatment, stomatal conductance of YPT1 and YPT2 was reduced by 77.3 and 76.2% of the initial conductance measured before stress (day 0), respectively, while DH121075 was only reduced by 67.2%. These differences were statistically significant (P < 0.05). Stomatal conductance in the transgenic and wild-type lines returned to the pre-stress level after re-watering. Similarly, leaf transpiration in YPT1 and YPT2 was reduced more than that observed in DH12075 during the water stress treatment (Figure 4d). On day 2 of the treatment, YPT1 and YPT2 had a 58.6 and 60.6% reduction in transpiration rate, respectively. This was significantly lower than a 42.9% reduction observed in DH12075 (P < 0.05). Transgenic lines also showed a greater reduction in leaf photosynthesis than DH12075, especially on day 2 of the water stress treatment (Figure 4e). Over the 4 days of drought treatment, the ratios of water loss relative to final shoot dry weight in YPT1 and YPT2 were 61.4 ± 1.03 and 61.1 ± 1.82, respectively, and they were significantly lower (P < 0.05) than that of DH12075 (65.1 ± 1.44). It is interesting to note that YPT1 and YPT2 showed a consistently better, but statistically non-significant, rate of recovery in stomatal conductance, leaf transpiration and photosynthesis when the plants were re-watered after drought stress treatment.

To further determine whether transgenic canola was more tolerant to water stress during flowering, YPT2 and DH12075 plants were grown in a chamber under optimal conditions until 8 days into flowering before being subjected to a 4-day drought stress. After re-watering, YPT2 recovered from stress better than DH12075. As shown in Figure 5, the YPT2 plant produced many healthy seed-filled pods, whereas water stress caused severe seed abortion in DH12075. This suggests that transgenic plants were more resistant to water deficit-induced seed abortion during flowering.

Figure 5.

Increased drought tolerance of rd29A:anti-AtFTB canola during flowering. The DH12075 (a) and YPT2 (b) plants were subjected to a 4-day drought treatment starting on day 8 after flowering. The pictures were taken on day 8 of re-watering after the drought stress.

Field trials on canola under different water conditions at mid-flowering

To assess the performance of rd29A:anti-AtFTB antisense canola in the field, standard confined field trials were conducted with either the bulk-harvested T4 or T5 homozygous seeds. Field trials were performed in three consecutive summer seasons in different sites of southern Alberta, Canada. The field sites traditionally lack natural rainfall during the months of July and August when plants are at their flowering peak, and canola plants grown in the area require irrigation to maximize seed yield. This provided an opportunity for us to examine the effect of drought stress on seed yield by applying different amounts of irrigation during this time. The amounts of water received in the trial sites from natural precipitation and irrigation during three field test seasons are summarized in Table 1.

Table 1.  Summary of natural precipitation and the amounts of irrigation applied to the field sites over 3 years of field trials of rd29A:anti-AtFTB transgenic canola and DH12075 (Precipitation data were provided by the Department of Statistics of Environment Canada). The 2002 irrigation was applied on July 14. The 2003 irrigations were applied on July 9 and 15. The 2004 irrigation was applied on June 30
Field siteYearMonth30-Year average precipitation (mm)Actual precipitation (mm)One irrigation (mm)Two irrigation (mm)
Lethbridge2002May48.351.7  
June62.7222.5  
July34.135.576 
August43.249.5  
Taber2003May53.566.3  
June36.563.0  
July47.55.876152
August45.84.5  
Taber2004May53.562.5  
June36.587.276 
July47.570.1  
August45.855.5  

In the summer of 2002, a first field trial was carried out to evaluate yield performance of eight homozygous YPT canola lines (YPT1–YPT8). The field trial for each transgenic line, as well as the DH12075 control, was standardized as a four-replicated plot trial in a randomized complete block design (RCBD) of all lines, with each plot (6.7 × 1.5 m) containing approximately 1800 sibling plants. In this field test, one irrigation (76 mm water) was provided to all plants in mid-July. As shown in Table 2, the mean values of seed yield of six transgenic YPT lines are higher than those obtained for DH12075, whereas two other transgenic lines (YPT7 and YPT8) produced yield similar to the control. Results of statistical analysis indicated that only YPT1 produced a significantly higher yield than DH12075 (P < 0.10). Under the one-irrigation field conditions, this first test indicated a positive trend on the agronomic benefits of conditional downregulation of BnFTB.

Table 2.  Seed yields of DH12075 (WT) and eight rd29A:anti-AtFTB transgenic canola lines (YPT) in 2002 field trial. Values represent the mean of four replicates ± standard error for each line
LineSeed yield (kg ha−1)
YPT13086 ± 129
YPT32865 ± 274
YPT42785 ± 200
YPT52697 ± 410
YPT62645 ± 277
YPT22503 ± 162
WT2440 ± 274
YPT72383 ± 148
YPT82365 ± 218

Based on the results of the first field trial, the two most ABA-sensitive YPT lines, YPT1 and YPT2, were selected along with the DH12075 control as the subjects of the second year field trial with a more defined control of irrigation, to further examine the impact of available water at flowering time on eventual seed yield. To increase the reliability of yield data from the trial, each line contained six replicates in each irrigation condition. In July and August 2003, the rainfall at the field trial site was 5.8 and 4.5 mm, respectively (as reported by Environmental Canada). In the month of July, one field received two irrigations for a total of 152 mm of water, compared with a second field that received one irrigation with half that amount of water. As shown in Figure 6, seed yields in the two-irrigation site were 2655, 2722 and 2660 kg ha−1 for DH12075, YPT1 and YPT2, respectively. This suggested that there was no yield drag for the YPT events under adequate water conditions. Under reduced irrigation conditions, the seed yields were 2291, 2634 and 2664 kg ha−1 for DH12075, YPT1 and YPT2, respectively. Thus, there was a 14% reduction in seed yield for DH12075 under reduced water conditions, whereas the seed reduction for YPT1 was only 3%, and seed yield for YPT2 was unaffected. Under reduced irrigation conditions, YPT1 and YPT2 produced 15 and 16% higher seed yield than that of DH12075, respectively, and the difference between the transgenics and the control was significant (P < 0.10). These results suggest that stress had a significantly greater negative effect on the yield of the control than on that of the YPT lines.

Figure 6.

Seed yields of DH12075 (WT) canola, YPT1 and YPT2 in 2003 (a) and 2004 (b) confined field trials. In (a) the solid bars represent seed yields for the two-irrigation condition, and the dotted bars represent seed yields for the one-irrigation condition. Irrigation was conducted during the flowering period. Error bars represent standard errors (n = 6).

In order to further confirm the observation of increased seed yield of the transgenic YPT1 and YPT2 lines, we conducted an additional year of field trials over the summer of 2004. This field trial was designed as a standard confined yield trial, containing YPT1, YPT2 and DH12075. There were six replicated plots for each of the testing events, and all entries were arranged in a RCBD. In 2004, the field site in Taber received noticeably higher amounts of natural precipitation over the summer than the corresponding 30-year averages in precipitation (Table 1). In particular, natural precipitation in July and August was higher than the previous 2 years (Table 1). In addition, the field was supplemented with 76 mm of water at the end of June to promote optimal growth and flowering. Under these field conditions, the seed yields were 2686, 2964 and 2841 kg ha−1 for DH12075, YPT1 and YPT2, respectively. Thus YPT1 and YPT2 produced approximately 10 and 6% higher seed yield, respectively, than that of DH12075, and the difference in yield between YPT1 and DH12075 was statistically significant (P < 0.05).

Conditional downregulation of FTB in canola does not have a negative impact on plant growth and seed quality

We recorded the key agronomic parameters during the 3 years of field studies. Except for the final seed yield, there were no significant differences between the DH12075 control, YPT1 and YPT2 in germination, seedling vigor, flowering time, plant height, lodging or maturity (data not shown). In addition, as shown in Figure 7(a), no significant differences in total oil content or protein content were found, and the fatty acid composition in the YPT1 and YPT2 seeds was the same as in DH12075 (Figure 7b). These data indicate that introduction of the drought-inducible gene expression cassette does not affect canola seed quality.

Figure 7.

Oil and protein contents (a) and fatty acid composition (b) of field trial seeds of DH12075 canola (solid bar), YPT1 (shaded bar) and YPT2 (dotted bar). Each data bar represents results of three replicates per entry sample.

Discussion

Depending on the crop, the average yield obtained by farmers using modern agricultural practices is on average 30% of the yield produced under ideal growth conditions (Boyer, 1982). Approximately 10% of the yield reduction is thought to be due to biotic stresses such as disease, insects and weeds. Of the remaining yield reduction, over 70% has been attributed to environmental stresses such as drought, salinity, and high and low temperatures (Bray et al., 2000). Due to the severe negative impact on yield and the ubiquitous nature of drought stress, engineering drought-tolerant crops has become one of the top priorities for crop scientists worldwide. Over the last decade, tremendous breakthroughs have been achieved in the understanding of drought stress response and signaling (Thomashow, 1999; Xiong et al. 2002; Zhu, 2002). Ectopic expression of several genes in drought signaling pathways has resulted in drought tolerance in model plants under laboratory conditions (Garg et al., 2002; Haake et al., 2002; Kasuga et al., 1999). However, translating drought-tolerant properties observed under laboratory conditions into yield improvement of crops under stress conditions in the field has remained a challenge (Sinclair et al., 2004). One reason for the difficulty is that seed yield is a complex trait, and is affected by a large number of factors during plant growth and development. In this study, we first produced drought-tolerant Arabidopsis plants by constitutive downregulation of either the α- or the β-subunit of the protein farnesyltransferase. We provided evidence that molecular tailoring of genes involved in protein farnesylation can effectively alter the plant's response to ABA at seed germination, seedling development and leaf guard cell levels. The ability to manipulate guard cell responses offered a way to reduce transpirational water loss, prolonging a plant's ability to cope with drought stress. We further demonstrated that downregulation of the β-subunit of farnesyltransferase in canola using an Arabidopsis drought-inducible promoter resulted in drought-tolerant, yield-protected (YPT) canola plants. In this system, the antisense downregulation of farnesyltransferase is a conditional and reversible molecular process that is governed by the amount of available soil water. In the field under adequate water conditions, these YPT plants showed normal growth and development and had a yield that was comparable with the DH12075 control, suggesting that expression of the YPT gene cassette had no adverse effects on plant growth and development. Under water stress conditions, YPT plants produced significantly greater seed yields than the control, and the quality of the seeds was not compromised.

Recently, an analysis of plant traits for increasing grain yield on limited water supplies was carried out (Sinclair and Muchow, 2001). Simulation studies on corn yield over 20 years indicate that earlier-than-normal stomatal closure in a crop is a positive trait for yield stability and enhancement, and delayed stomatal closure in a crop under water-deficient conditions would most likely result in decreased yield. These findings reinforce the idea that drought stress tolerance and yield protection could be achieved by controlling stomatal transpiration. Under drought conditions, earlier and tighter stomatal closure would reduce leaf transpiration, thereby maintaining higher available soil water content, and subsequently, reducing the damage to the plant. In our study, under water-sufficient conditions, the levels of stomatal conductance and leaf transpiration in YPT1, YPT2 and DH12075 were similar. However, stomatal conductance and transpiration were reduced further in YPT1 and YPT2 than in the control during water stress (Figure 4c,d). The reduction in leaf transpiration observed in the YPT lines is probably the result of greater reduction in stomatal aperture during stress, as there were no statistical differences in stomatal density between the YPT lines and DH12075 (M. Kuzma and Y. Huang, unpublished data).

The impact of water deficit on seed yield of canola is greatly dependent on the timing and duration of stress as well as on the stage of growth cycle (Mingeau, 1974). Plants are most susceptible to drought stress during flowering when a lack of moisture leads to pollination failure and seed abortion. In our field tests, we reduced irrigation at the peak of flowering to compare the impact of water deficit between the YPT and control plants. We also chose not to completely eliminate irrigation during that time, as moderate drought stress on yield potential is a more relevant agricultural problem with crops (Bruce et al., 2002). Data from our first field trial in 2002 indicated that conditional downregulation of BnFTB enhanced seed yield under mild water stress conditions (Table 2). In our second year of field trials, we compared the impact of two different irrigation conditions on seed yield. Reduction in irrigation (by 50%) caused a 14% reduction in seed yield of DH12075, but the yield of YPT canola lines was unaffected. Low coefficient of variation values for the field experiments (15% for the two-irrigation site and 13.5% for the one-irrigation site) indicated that the variation between plots was relatively low and the results of this study were reliable.

One of the key criteria for a viable drought tolerance biotechnology is that the introduction of the transgenic construct should have little negative effect on seed yield under optimal growth conditions. More importantly, the yield advantage of the transgenic plants should be present over the subsequent growing seasons and under variable climatic conditions. Data obtained from our third year field trial in 2004 indicated a significant increase in yield even though there was a considerably higher amount of moisture (Figure 6b). The 3-year combined data set from different field conditions suggests a consistently positive correlation between conditional downregulation of BnFTB and yield protection in canola.

Recent genetic studies have pinpointed several other mutations that affect ABA-induced guard cell activities. Disruption of an Arabidopsis heterotrimeric GTP binding protein α-subunit GPA1, as well as two ABA-activated protein kinase orthologs, Arabidopsis OST1 and Vicia faba AAPK, resulted in impairment of stomatal closure (Li et al., 2000; Mustilli et al., 2002; Wang et al., 2001). Thus, these three genes appear to be the positive regulators of stomatal closing. If true, such genes would also be potentially useful for genetic engineering crop plants with reduced stomatal responsiveness and enhanced transpiration for maximum yield production under optimal irrigation conditions.

Our data suggest that downregulation of an ABA-negative regulator, such as ERA1, offers an effective way to confer drought tolerance in a crop plant. One advantage of using the two subunits of plant farnesyltransferase as gene targets for drought tolerance is that both FTA and FTB belong to single-gene families in all plant species examined (Y. Wang and Y. Huang, unpublished data). This apparent lack of genetic redundancy will greatly simplify the design of genetic manipulation approaches, and it is likely that drought-tolerant phenotypes similar to those described here for Arabidopsis and canola could be engineered for many crop species for benefits to growers worldwide.

Experimental procedures

Constructs and transgenic materials

For the AtFTA antisense construct, the coding region of the α-subunit of Arabidopsis farnesyltransferase (AtFTA) was cloned downstream of the 35S promoter of pBI121 (Clontech, Palo Alto, California) in the antisense orientation. For the hairpin AtFTB construct, a 500-bp DNA fragment encoding the N-terminus of AtFTB was placed in pBI121 in sense and antisense orientations spaced by a truncated GUS gene as previously described (Chuang and Meyerowitz, 2000). For the rd29A:anti-AtFTB construct, the rd29A promoter (Yamaguchi-Shinozaki and Shinozaki, 1993) was placed upstream of the antisense AtFTB fragment. The first two binary plasmids were introduced into Agrobacterium tumefaciens GV3101 and subsequently transformed into A. thaliana (Columbia ecotype) wild-type plants by the floral dip method (Clough and Bent, 1998). The rd29A:anti-AtFTB construct was transformed into B. napus (canola cv. DH12075) as described (Moloney et al., 1989).

ABA sensitivity test

Seeds (30–80 seeds/plate) were placed on 0.5x MS plates and chilled for 3 days with or without (+/−) ABA (Sigma, St. Louis, MO, USA). They were subsequently transferred to room temperature, with continuous light for up to 2 weeks. Germination was scored by full penetration of a radicle tip through the seed coat.

RNA gel blot analysis

Bleach-sterilized Arabidopsis seeds were plated on full-nutrient MS plates and left in the dark for 3 days at 4°C. The plates were then transferred to a growth chamber, and seedlings were grown for 2 weeks under constant light at 22°C. Total RNA was isolated from whole seedlings with the RNeasy Plant Mini Kit (Qiagen, Mississauga, Ontario, Canada), separated by agarose gel electrophoresis, and blotted to a nylon membrane (Roche, Indianapolis, IN, USA) as described (Ausubel et al., 1996). For the anti-AtFTA samples, the blot was first hybridized with a [32P]-labeled, single-stranded sense AtFTA RNA probe to detect the antisense transcripts. The blot was then stripped and re-hybridized with a single-stranded antisense AtFTA RNA probe to detect the sense AtFTA transcripts. For the hairpin AtFTB samples, the blot was hybridized with a random-hexamer-labeled (Promega, Madison, WI, USA), double-stranded AtFTB cDNA probe.

Canola plants were grown in 4 in pots under optimal conditions with 16 h light/8 h dark cycles to the fourth leaf stage before being subjected to an alternating well-watered, drought stress, and well-watered treatment. The drought treatment was started on day 0 by withholding water. Three replicates of the fourth leaf were collected from each line 1 day before drought treatment, and on days 0, 2, 3, 4 and 5 of the drought treatment for isolation of total RNA. Leaf samples were also collected in the following three re-watering days. The RNA blot was prepared and hybridized first with the single-stranded sense AtFTB RNA probe to detect the antisense message using a PhosphoImager (Molecular Dynamics, Piscataway, NJ, USA). The blot was subsequently stripped and re-hybridized with a single-stranded antisense BnFTB RNA probe to detect the endogenous BnFTB transcript. The level of BnFTB transcript was quantified against the membrane-bound 23S rRNA using the UN-Scan-It gel automated digitizing system (Silk Scientific, Orem, UT, USA).

Drought stress treatment

Bleach-sterilized Arabidopsis seeds were plated on MS plates and cold-treated at 4°C for 3 days to synchronize germination. Plates were placed in the growth chamber at 22°C, 70% relative humidity and 16 h light/8 h dark cycles. After 7 days of growth, five seedlings were transplanted into either a 4-in pot (the antisense AtFTA experiment) or a 3-in pot (the hairpin AtFTB experiment) with equal amounts of homogenized soil mix (Promix). Eight replicate pots were used for the antisense AtFTA experiment and six replicate pots were used for the hairpin AtFTB experiment, arranged in a RCBD. Plants were maintained in the growth chamber under optimal conditions (watered daily) until the first open flower was observed. Before drought treatment, pots were watered to a set weight and drought treatment was started on day 0 by withholding water. Water loss from pots was monitored daily by weighing the pots. In the 4 in pots, water was withheld for 9 days, and in the 3-in pots water was withheld for 4 days. At the end of the treatment shoot biomass was harvested and fresh weight was determined. Shoots were then dried at 60°C for 4 days and dry weight was determined.

Canola drought and recovery test

The canola drought and recovery experiment was performed in a controlled-environment growth chamber (Conviron, Winnipeg, Manitoba, Canada). The plants (five replicates of DH12075 and T4 homozygous YPT2, respectively) were grown in 6 in pots in a RCBD under optimal watering and growth conditions until 8 days into flowering. The plants were not watered for 4 days, followed by re-watering, and the plants were allowed to grow to full maturity.

Canola gas-exchange study

This experiment was conducted in a Conviron growth chamber. The plants (DH12075 canola and two transgenic lines YPT1 and YPT2) were arranged in a RCBD with eight replicates for each entry. The plants were grown in 6 in pots under optimal conditions with 16 h light/8 h dark cycles to the sixth leaf stage before being subjected to drought stress treatment. Prior to the drought stress treatment, the pots were watered to near-saturation and encased by aluminum foil. Leaf gas-exchange rate was measured daily beginning 1 day prior to the drought stress treatment, over a 4-day course of drought stress and 3 days of re-watering. The fifth fully expanded leaf was measured using a portable, open-flow gas exchange system LI-6400 (LI-COR, Lincoln, NE, USA). The gas-exchange rate measured on day 0 of the treatment was considered as the maximum rate under optimum growing conditions. Leaf gas exchange was calculated by the LI-6400's operating software as described (Von Caemmerer and Farquhar, 1981).

Field trial study and seed quality analysis

The canola field trials were conducted by Ag-Quest Inc. in the summers of 2002, 2003 and 2004 in southern Alberta, Canada. Transgenics and control plant entries were arranged in the replicated plots according to a RCBD. Each plot was 6.7 m long and 1.5 m wide, containing eight rows with 15 cm row spacing. Seeds (1800) were planted using a cone seeder in each plot. Local recommended management practices were employed to achieve maximum yield. The 2002 field trial was conducted in Lethbridge, Alberta, and there were four replicated plots for each field trial entry. The field was irrigated once with 76 mm of water on July 14. The 2003 field trials conducted in Taber contained six replicated plots for each trial entry per irrigation condition. The two-irrigation field was irrigated twice during the flowering period. 76 mm of water was irrigated on July 9 and 15, respectively, while the one irrigation field was only irrigated once on July 9 with 76 mm of water. The 2004 summer trial was also conducted in Taber, and it contained six replicated plots for each field entry. The field was irrigated with 76 mm of water on June 30. Upon maturation, seeds were harvested from the middle 7.5 m2 area of each plot to determine seed yield. Seed oil and protein contents, as well as seed fatty acid composition, were analyzed for approximately 300 seeds of each replicate, three replicates per entry sample from the 2003 field trial, according to the protocols of AOCS Official Methods and Recommended Practices (POS Pilot Plant, Co., Saskatoon, Saskatchewan, Canada).

Statistical analysis

For the Arabidopsis and canola gas exchange experiments, statistical analysis was completed using JMP software (Version 5.0.1.2; SAS Institute Inc., Cary, NC, USA). After initial testing of residuals normality (Shapiro–Wilk test) and equal variance (Levene test), all data were subjected to analysis of variance (anova). If significant differences were found (P < 0.05), comparisons with control were performed using Dunnett's test and all pair comparisons were made using the Tukey–Kramer test at P < 0.05.

Canola seed yield data from the field trials were also subjected to an anova using JMP. Transgenic line, year and the transgenic line × year interaction were considered fixed effects and replications nested within year were treated as random effects in the variance model. Yield data of each year were analyzed separately if the interaction of transgenic lines and year was significant. Treatment effects were considered significant at P < 0.05 unless otherwise indicated.

Acknowledgements

The authors would like to thank Dr Gerhard Rakow of Agriculture and AgriFood Canada, Saskatoon, for providing the elite canola seed of DH12075 for our transformation. We are also grateful to Dr Michael Thomashow of the Michigan State University for his critical reading and insightful comments on this manuscript.

Ancillary