Many efforts have been made to engineer stress tolerance by accumulating polyols. Transformants that accumulate polyols often show growth inhibition, because polyols are synthesized as a dead-end product in plants that do not naturally accumulate polyols. Here, we show a novel strategy in which a sorbitol cycle was engineered by introducing apple cDNA encoding NAD-dependent sorbitol dehydrogenase (SDH) in addition to sorbitol-6-phosphate dehydrogenase (S6PDH). Tobacco plants transformed only with S6PDH showed growth inhibition, and very few transformants were obtained. In contrast, many transgenic plants with both S6PDH and SDH were easily obtained, and their growth was normal despite their accumulation of sorbitol. Interestingly, the engineered sorbitol cycle enhanced the accumulation of sucrose instead of fructose that was expected to be increased. Sucrose, rather than fructose, was also increased in the immature fruit of tomato plants transformed with an antisense fructokinase gene in which the phosphorylation of fructose was inhibited. A common phenomenon was observed in the metabolic engineering of two different pathways, showing the presence of homeostatic regulation of fructose levels.
Sorbitol-6-phosphate dehydrogenase (S6PDH) catalyses the NADPH-dependent reduction of glucose-6-phosphate to sorbitol-6-phoshate (Kanayama 1998). S6PDH is a key enzyme in the biosynthesis of sorbitol, which is important in the translocation of photosynthate in fruit trees of the Rosaceae (Sakanishi et al. 1998). Apple cDNA encoding S6PDH has been cloned (Kanayama et al. 1992), and the abundance of S6PDH mRNA is high in source tissues (Kanayama et al. 1995). The enzymatic properties of S6PDH indicate that S6PDH is involved in the biosynthesis of sorbitol in vivo (Kanayama & Yamaki 1993). Sorbitol accumulates in species that do not synthesize it by transgenically introducing S6PDH cDNA (Tao, Uratsu & Dandekar 1995).
Inducible expression systems for determinants of stress tolerance have been proposed because engineered accumulation of sorbitol and mannitol causes abnormalities in growth (Sheveleva et al. 1998; Abebe et al. 2003). Alternatively, a polyol would be synthesized as an end product in an engineered pathway if the plant does not naturally possess the polyol. Here, we show a novel strategy in which a sorbitol cycle was engineered by introducing SDH in addition to S6PDH (Fig. 1). As a result, transgenic tobacco plants that accumulated sorbitol were easily obtained, and their growth was normal. In addition, an interesting change in sugar composition was observed in the transformants.
MATERIALS AND METHODS
Construction of binary vectors
The partial sequence that contained the open reading frame of apple S6PDH cDNA (Kanayama et al. 1992; accession number D11080) was inserted in a sense direction between the cauliflower mosaic virus 35S promoter (35S) and nopaline-synthase (NOS) termination site in the binary vector pBI121. The S6PDH cDNA was digested with SpeI (−68) and BstBI (+ 953), and was ligated into pBI121 after the β-glucuronidase (GUS) sequence was removed by digesting with XbaI and SacI. The amplified sequence of apple SDH cDNA (Yamada et al. 1998; accession number AB 016256) was also inserted in a sense direction between the 35S and NOS in pBI121. The primers for the amplification contained engineered XbaI and SacI sites. The partial sequence between −35 and + 1147 of the SDH cDNA was amplified by PCR using a high-fidelity enzyme, and was ligated into pBI121 after the GUS sequence was removed by digesting with XbaI and SacI. These binary vectors were used for transformation as pSPD and pSDH. Furthermore, the 35S-S6PDH-NOS cassette in pSPD was digested with EaeI, and was inserted into the EcoRI site in pSDH. This binary vector containing the coding regions of S6PDH and SDH was used as pSPSD. Blunt ends were made with T4 DNA polymerase if they were necessary for ligation. We used pBI121 as a control vector.
Transformation of tobacco and culture of T0 generations
Each binary vector was introduced into Agrobacterium tumefaciens LBA4404. Tobacco (Nicotiana tabacum cv. SR1) leaf segments were inoculated with the mixture of Agrobacterium tumefaciens cells containing pSPD and cells containing pSDH at the same concentrations. Leaf segments were also inoculated with A. tumefaciens cells containing pSPSD. Selection and regeneration were carried out in Murashige and Skoog medium (Murashige & Skoog 1962) supplemented with 0.1 mg L−1 naphthylacetic acid, 1.0 mg L−1 benzylaminopurine, 500 mg L−1 cefotaxime and 100 mg L−1 kanamycin. When distinct stems and leaves were visible in the regenerated shoots, the excised shoots were transferred to a phytohormone-free medium. The rooted shoots were grown in potting soil under a 16 h photoperiod in artificial light at 25 °C. Primary transformants (T0 generation) were identified by PCR analysis.
Growth of T1 generations
T0 generations with pSPSD (SPSD plants) were selected for further investigation by measuring S6PDH and SDH activity. The selected T0 plants with high activity were self-pollinated. T1 plants (SPSD lines) from independent T0 plants were grown in potting soil (one plant per pot) under a 16 h photoperiod in artificial light (approximately 100 µmol m−2 s−1) at 25 ± 5 °C. The leaves and roots of these plants were sampled 3 months after sowing and were stored at −80 °C until use.
Lines SPSD2, SPSD3, SPSD19 and SPSD35 were used for analyses. Each line was grown as a single independent experiment. Because the growth conditions for each line were somewhat different, segregated populations that were grown simultaneously in each line were used as control (azygous) and transformant (homozygous and heterozygous). The numbers of plants used for each analysis are described in the figure legends. The transgene copy number and segregation were investigated using Southern blot analysis and PCR according to the methods of Kanayama et al. (1997) and Odanaka, Bennett & Kanayama (2002). Each line had a single copy of the transgene.
Fructokinase antisense tomato plants
Lines AF1-72 and AF2-14 of fructokinase antisense tomato (Lycopersicon esculentum Mill.) plants were produced and grown as described by Odanaka et al. (2002). Immature fruits were sampled 20 d after flowering and were used for analysis.
Enzyme activity analyses
Crude extracts for S6PDH and SDH were prepared from fifth and sixth leaves from the tops of the transgenic tobacco plants according to the method of SDH extraction described by Suzuki, Odanaka & Kanayama (2001). S6PDH and SDH activity were measured by the methods of Tao et al. (1995) and Suzuki et al. (2001), respectively. Crude extracts for sucrose synthase were prepared from roots of the transgenic tobacco plants and from immature fruits of fructokinase antisense tomato plants according to the method of Schaffer & Petreikov (1997). Sucrose synthase activity was measured by the method of Miron & Schaffer (1991). Protein content in the enzyme extracts was determined by the method of Bradford (1976).
Determination of soluble carbohydrate content
Soluble carbohydrates were extracted from the third and fourth leaves from the top of the plants or from roots using the method of Suzuki et al. (2001). The samples were analysed by gas chromatograph G-300 (Hitachi, Tokyo, Japan) with a CP-Sil 5CB column (Varian, Palo Alto, USA) after trimethylsilylation. The gas chromatography conditions were injector temperature, 200 °C; flame ionization detector temperature, 300 °C; running temperature program, 50 °C for 1 min, then increasing at 25 °C min−1 to 280 °C followed by a 10 min hold at 200 °C. Helium was used as a carrier gas. Sorbitol was separated from mannitol in the condition. Xylitol was used as an internal standard.
Transformation of tobacco with pSPD, pSDH and pSPSD
The binary vector pSPD contained the coding region of S6PDH that functions in sorbitol biosynthesis. The binary vector pSDH contained the coding region of SDH that functions in the conversion of sorbitol to fructose. Tobacco leaf segments were inoculated with a mixture of the same concentrations of A. tumefaciens cells that contained each vector. The inoculation was performed using 40 segments in 10 plates. We observed many regenerated plants that did not grow well and had lesions as described by Sheveleva et al. (1998). These plants that could not be acclimated because of poor growth were identified by PCR as the SPD line. Table 1 shows the number of transformants that generated shoots and roots, that were acclimated and then grew for 7 weeks without lesions. SPD plants harbouring only 35S::SPD and SDH plants harbouring only 35S::SDH were obtained with SPD + SDH plants harbouring both constructs. There were fewer SPD and SPD + SDH plants than SDH plants.
Table 1. Number of independent acclimated plants, S6PDH activity and internode length in T0 generation
Forty leaf segments in 10 plates were inoculated with the mixture of Agrobacterium tumefaciens cells containing pSPD and cells containing pSDH at the same concentrations.
Forty leaf segments in 10 plates were inoculated with A. tumefaciens cells containing pSPSD.
c S6PDH activity and internode length were shown in each four transgenic plants with high S6PDH activity. Each value represents the mean (the associated SE). In internode length, differences between control plants and transgenic plants were tested using one-way analysis of variance (anova) (**, P < 0.01).
The binary vector pSPSD contained 35S-S6PDH-NOS and 35S-SDH-NOS in tandem. The inoculation of A. tumefaciens cells that contained pSPSD was also performed using 40 leaf segments in 10 plates. In contrast to SPD plants, 50 transformants grew without lesions for 7 weeks after acclimation (Table 1). Lesion formation was rare.
SPD, SPD + SDH and SPSD plants with high S6PDH activity were selected, and their growth was compared (Table 1). Growth was assessed by measuring internode length, and we found that even SPD plants that grew without observable lesions were stunted compared with control plants. In contrast, the internode lengths of SPSD and SPD + SDH plants were longer than those of SPD plants, despite the same range of S6PDH activity in the three lines. In particular, the internode length of SPSD plants was almost the same as that of control plants.
S6PDH and SDH activity in T1 generations
T0 generations (SPSD plants) that had high S6PDH and SDH activity were selected and self-pollinated. Four SPSD lines (T1 plants) from independent T0 plants were used for analyses because they had a single copy of the transgene. Segregated populations in each line that were simultaneously grown were used for the control (azygous) and the transformant (homozygous and heterozygous). The growth and carbohydrate content in control plants were not always the same among the four lines because the growth conditions for each line were somewhat different. S6PDH and SDH activity were shown in lines SPSD2, SPSD3, SPSD19 and SPSD35 (T1 generation) harbouring S6PDH and SDH (Fig. 2). Sorbitol accumulation in tobacco plants that do not naturally contain sorbitol has been achieved by the introduction of the S6PDH gene (Tao et al. 1995; Sheveleva et al. 1998). Sorbitol was also detected in leaves and roots in the SPSD lines (Figs 3 & 4). These results indicate that the introduced S6PDH gene functioned in the SPSD lines.
To our knowledge, this is the first report of heterologous expression of the SDH gene. SDH activity was similar to S6PDH activity in SPSD lines except SPSD2 that had low SDH activity (Fig. 2). SDH cDNA cloned from apples was used in this study (Yamada et al. 1998). SDH activity was assayed in crude enzyme extracted from apples using the same methods used in this study (Suzuki et al. 2001). According to Suzuki et al. (2001), SDH activity was approximately 2 nmol min−1 g−1 fresh weight (FW) when the activity was highest during fruit development. In our study, SDH activity was 9.6 and 4.1 nmol min−1 g−1 FW in SPSD3 and SPSD19, respectively. This result showed that the introduced SDH gene expressed high activity in non-Rosaceae plants.
Sorbitol and myo-inositol content and growth in SPSD lines
Sorbitol content was 3.0–4.5 µmol g−1 FW in the leaves of SPSD plants (Fig. 3). Myo-inositol content was determined, because the stunted growth of tobacco plants transformed with S6PDH gene is reportedly caused by a drastic decrease in myo-inositol (Sheveleva et al. 1998). SPSD2 and SPSD3 plants retained a myo-inositol content of 1.5–2.0 µmol g−1 FW, although the content was lower than that in control plants. The myo-inositol content in SPSD19 and SPSD35 plants were similar to that in control plants. The growth of SPSD plants was assessed by plant height, the number of leaves and FW (Fig. 5). SPSD plants, which showed significant activity of S6PDH and SDH as well as the accumulation of sorbitol, grew similarly to control plants. Soluble protein content was also determined to show the normal growth of SPSD plants. As shown in Fig. 6, it was also similar between SPSD plants and control plants.
Soluble carbohydrate composition in the roots of SPSD lines
Soluble carbohydrate composition was measured in the roots of SPSD lines to better understand the effect of the engineered sorbitol cycle on carbohydrate metabolism in sink tissues (Fig. 4). The sorbitol content was similar to, or slightly lower than, the glucose and fructose content in SPSD lines. A difference in hexose content was not observed between SPSD and control plants. However, the sucrose content was higher in SPSD3, SPSD19 and SPSD35 plants compared with the control. The results indicate that total soluble carbohydrate content was increased in these SPSD lines. In contrast, the sucrose content was not higher in SPSD2 plants that had low SDH activity (Fig. 2). We then assayed the activity of sucrose synthase, which plays a key role in sucrose metabolism in tobacco roots, in relation to the increase in sucrose content in lines SPSD3 and SPSD19 (Fig. 7). The result shows that sucrose synthase activity in SPSD lines was lower than that in the control. Acid and neutral invertase activity was scarcely detected.
Sucrose synthase activity in the fruit of fructokinase antisense tomato
The engineered sorbitol cycle that was expected to facilitate carbon flow from glucose-6-phosphate to fructose (Fig. 1) did not increase fructose content, but did increase sucrose content. An increase in fructose content was also expected in the suppression of fructose metabolism by the antisense transformation of fructokinase (Fig. 1). However, sucrose, rather than fructose, was increased in the fruit of fructokinase antisense tomato (Odanaka et al. 2002). Sucrose content and sucrose synthase activity in the antisense tomato fruit are shown in Fig. 8. Sucrose content was increased in the fruits of both antisense lines compared with the control line, while sucrose synthase activity was suppressed in the antisense lines.
Sorbitol synthesis is attributed to S6PDH, while SDH functions in fructose formation as described in the Introduction. Several strategies can be used to obtain transgenic plants having both S6PDH and SDH, for example, (1) crossing between a transformant with S6PDH and another transformant with SDH; (2) inoculation of the mixture of A. tumefaciens cells harbouring S6PDH and cells harbouring SDH; and (3) inoculation of A. tumefaciens cells harbouring both genes of S6PDH and SDH inserted into a single binary vector. Although Sheveleva et al. (1998) produced more than 100 transgenic tobacco plants that accumulated various concentrations of sorbitol, many of these did not show normal growth. In our study, it was also very difficult to obtain SPD plants harbouring only S6PDH, and abnormal growth was often observed as described by Sheveleva et al. (1998). Therefore, a problem with strategy (1) is the difficulty in obtaining a fertile SPD plant for crossing. With regard to strategy (2), the growth of SPD + SDH plants with both genes was more robust. However, only a few transgenic plants were obtained, probably because it is necessary for both genes, for example, S6PDH and SDH, to be inserted into genomic DNA in the same plant cell and to be expressed with high activity. In contrast, in strategy (3), many more transformants (SPSD plants) were obtained compared with strategy (2).
The growth of SPSD plants was better than that of acclimated SPD plants in the T0 generation, even though S6PDH activity in SPSD plants was in the same range as S6PDH activity in the SPD plants (Table 1). These results suggest that the introduction of SDH in addition to S6PDH results in normal growth of SPSD plants.
Tao et al. (1995) first reported that the introduction of S6PDH is sufficient for sorbitol biosynthesis in non-Rosaceae plants that do not naturally contain sorbitol. However, all transformants accumulated a very low content of sorbitol (0.2–0.45 µmol g−1 FW in leaves), and their growth was normal. Brown et al. (1999) and Bellaloui, Brown & Dandekar (1999) also reported the enhancement of boron transport in transgenic tobacco plants that grew normally with a very low content of sorbitol (less than 0.9 µmol g−1 FW in leaves). However, tobacco plants transformed with S6PDH show slower growth at 2–3 µmol g−1 FW of sorbitol content in leaves (Sheveleva et al. 1998). Furthermore, at 3–5 µmol g−1 FW of sorbitol content, transgenic tobacco plants show regions in which chlorophyll is partially lost and necrotic lesions are visible. Transgenic plants accumulating mannitol have been reported by the expression of the bacterial mltD gene for the biosynthesis of mannitol in tobacco (Karakas et al. 1997) and wheat (Abebe et al. 2003). The transgenic tobacco plants accumulating mannitol show slower growth and are 20–25% smaller than wild-type plants. The concentration of mannitol accumulated in transgenic tobacco leaves is 3.2 µmol g−1 FW, which is calculated from 3.8 mg g−1 dry weight (DW) using the value of relative water content of 85% reported in the same study (Karakas et al. 1997). The transgenic wheat plants accumulating more than 1.5 µmol g−1 FW in leaves are also short (Abebe et al. 2003). In our study, SPSD plants showed normal growth despite sorbitol accumulation in leaves at approximately 3.0–4.5 µmol g−1 FW. Therefore, SDH activity most likely contributed to the normal growth of the transgenic plants accumulating sorbitol.
The myo-inositol content is less than 1 µmol g−1 FW in almost all tobacco plants transformed with S6PDH (Sheveleva et al. 1998). Many of these transformants accumulate 4 µmol g−1 FW of sorbitol or less. In our study, such a low concentration of myo-inositol was not observed despite the accumulation of 3.0–4.5 µmol g−1 FW sorbitol. Myo-inositol and its phosphate ester are important in cell wall biosynthesis and cellular signal transduction (Sheveleva et al. 1998). Extreme interference with myo-inositol biosynthesis results in the inhibition of growth in tobacco plants transformed with S6PDH (Sheveleva et al. 1998). Therefore, the normal growth of SPSD plants may be caused by the maintenance of the myo-inositol pool.
Collectively, many transgenic plants accumulating sorbitol have been readily obtained by an engineered sorbitol cycle, rather than by transformation with only S6PDH. Moreover, the growth of transgenic plants with an engineered sorbitol cycle is normal. One explanation for this useful result is most likely that the myo-inositol pool is maintained.
The phosphoglucose isomerase reaction between glucose-6-phosphate and fructose-6-phosphate is generally the only pathway of isomerization of glucose and fructose in plants that produce sucrose for translocation. Because the sorbitol cycle was engineered as an additional pathway for isomerization of the hexoses in this study (Fig. 1), it was expected that soluble carbohydrate composition would be changed and/or that homeostatic regulation of the composition would occur. Sorbitol synthesized by S6PDH in leaves is converted to fructose by SDH in fruit, resulting in apparently more fructose than glucose accumulation in apple and pear fruits (Yamaki & Ishikawa 1986; Yamaki & Moriguchi 1989). However, in transgenic tobacco plants with an engineered sorbitol cycle, sucrose, rather than fructose, was increased in roots, which, like fruits, are sink organs. Sucrose content is not increased by transformation with only S6PDH (Sheveleva et al. 1998). Indeed, sucrose content did not differ between transgenic plants and control plants in line SPSD2, in which SDH activity was much lower than other SPSD lines, while S6PDH activity was similar to other SPSD lines (Figs 2 & 4). Therefore, the increase in sucrose content most likely requires SDH activity in addition to S6PDH activity.
Fructokinase plays a key role in the phosphorylation of fructose to fructose-6-phoshphate (Kanayama et al. 1997). Thus, an increase in fructose content is also expected in transgenic tomato with the antisense fructokinase gene (Fig. 1). However, sucrose rather than fructose was increased in immature fruits of antisense tomato plants. A common phenomenon was observed in the metabolic engineering of two different pathways, showing the presence of the homeostatic regulation of fructose levels.
Acid and neutral invertase activity was scarcely detected in tobacco roots in this study. Sucrose synthase is the major enzyme that breaks down sucrose in immature tomato fruit (Schaffer & Petreikov 1997). Therefore, sucrose synthase activity was determined as one candidate for homeostatic regulation. Our results show decreases in sucrose synthase activity in both the engineered sorbitol cycle and the down-regulation of fructokinase, suggesting that sucrose synthase activity is one of the regulation points for fructose levels. This regulation may be important in the suppression of sucrose unloading when fructose levels are increased by surplus sucrose supplies from source tissues.
Collectively, the engineered polyol cycle represents a strategy that avoids polyol-induced growth inhibition in transgenic plants. In this strategy, plants do not always accumulate excessive amounts of polyols. Nevertheless, many previous reports have indicated that marginal accumulation of polyols and other osmolytes increases stress tolerance (Tarczynski et al. 1993; Kavi Kishor et al. 1995; Thomas et al. 1995; Holmström et al. 1996; Karakas et al. 1997; Shen et al. 1997; Abebe et al. 2003). In these cases, the enhanced stress tolerance could not be caused by osmotic adjustment, but rather by other mechanisms such as scavenging of reactive oxygen and stabilization of macromolecular structures. Therefore, an engineered sorbitol cycle is meaningful as a method for easily obtaining plants that accumulate intermediate amounts of polyols and that exhibit normal growth. Alternatively, an engineered cycle is useful as an analytical method for the dynamic regulation of carbohydrate metabolism. Our study showed the homeostatic regulation of fructose levels in plant cells. Exogenous application of sugars to plant tissues has been conducted to investigate responses to sugars. However, this method only provides information about responses to extracellular signals. In contrast, the engineered cycle is a useful tool for precisely investigating responses to intracellular signals.
Practically, the engineered sorbitol cycle could be a technique for increasing sucrose levels. Even if antisense sucrose synthase RNA is expressed to inhibit sucrose breakdown, fruit setting is decreased as has been described in tomato (D’Aoust et al. 1999). Antisense inhibition of fructokinase also decreases fruit setting, although sucrose content is increased (Odanaka et al. 2002). Such negative effects will be a problem in the practical use of the technique for increasing sucrose levels. In contrast, the growth of roots as a sink organ was not inhibited in this study (Fig. 5), that is, engineered sorbitol cycle is potentially valuable for enhanced sucrose accumulation.
We thank A. Nakamura for transforming the tobacco plants. We also thank S. Odanaka for analysing the transgenic tomato plants.