Glycinebetaine accumulation is more effective in chloroplasts than in the cytosol for protecting transgenic tomato plants against abiotic stress

Authors

  • EUNG-JUN PARK,

    1. Department of Horticulture, ALS 4017, Oregon State University, Corvallis, OR 97331, USA and
    Search for more papers by this author
    • *

      Present addresses: Plant Metabolism Research Center, Kyung Hee University, 446-701, Republic of Korea.

  • ZORAN JEKNIĆ,

    1. Department of Horticulture, ALS 4017, Oregon State University, Corvallis, OR 97331, USA and
    Search for more papers by this author
  • MARÍA-TERESA PINO,

    1. Department of Horticulture, ALS 4017, Oregon State University, Corvallis, OR 97331, USA and
    Search for more papers by this author
    • Present address: Institute of Agricultural Research, INIA, Santiago 7083150, Chile.

  • NORIO MURATA,

    1. National Institute for Basic Biology, Okazaki 444-8585, Japan
    Search for more papers by this author
  • TONY HWEI-HWANG CHEN

    Corresponding author
    1. Department of Horticulture, ALS 4017, Oregon State University, Corvallis, OR 97331, USA and
      T. H.-H. Chen. Fax: +1 541 737 3479; e-mail: chent@hort.oregonstate.edu
    Search for more papers by this author

T. H.-H. Chen. Fax: +1 541 737 3479; e-mail: chent@hort.oregonstate.edu

ABSTRACT

Tomato (Lycopersicon esculentum Mill. cv. Moneymaker) plants were transformed with a gene for choline oxidase (codA) from Arthrobacter globiformis. The gene product (CODA) was targeted to the chloroplasts (Chl–codA), cytosol (Cyt–codA) or both compartments simultaneously (ChlCyt–codA). These three transgenic plant types accumulated different amounts and proportions of glycinebetaine (GB) in their chloroplasts and cytosol. Targeting CODA to either the cytosol or both compartments simultaneously increased total GB content by five- to sixfold over that measured from the chloroplast targeted lines. Accumulation of GB in codA transgenic plants was tissue dependent, with the highest levels being recorded in reproductive organs. Despite accumulating, the lowest amounts of GB, Chl–codA plants exhibited equal or higher degrees of enhanced tolerance to various abiotic stresses. This suggests that chloroplastic GB is more effective than cytosolic GB in protecting plant cells against chilling, high salt and oxidative stresses. Chloroplastic GB levels were positively correlated with the degree of oxidative stress tolerance conferred, whereas cytosolic GB showed no such a correlation. Thus, an increase in total GB content does not necessarily lead to enhanced stress tolerance, but additional accumulation of chloroplastic GB is likely to further raise the level of stress tolerance beyond what we have observed.

INTRODUCTION

In natural environments, productivity and the geographical distribution of plants are limited by adverse conditions, such as cold, salinity and drought. Among their various protective mechanisms for ensuring survival is the accumulation of compatible solutes, for example, organic metabolites that are of low molecular weight, highly soluble in water and non-toxic at high concentrations (Bohnert, Nelson & Jensen 1995). Compatible solutes, which differ among plant species, include betaines, polyols and sugars and amino acids (Rhodes & Hanson 1993).

Glycinebetaine (GB) is a quaternary ammonium compound that occurs naturally in a wide variety of plants, animals and microorganisms (Rhodes & Hanson 1993). GB is accumulated in many plant species at elevated levels in response to various kinds of abiotic stress, resulting in their improved tolerance (Rhodes & Hanson 1993). Possible roles for GB under such conditions include the stabilization of complex proteins and membranes, protection of the transcriptional and translational machineries and intervention as a molecular chaperone in the refolding of enzymes (for reviews, see Sakamoto & Murata 2000; Chen & Murata 2002). In addition, GB may indirectly induce H2O2-mediated signalling pathways, for example, through enhanced catalase expression and activity (Park, Jeknić & Chen 2006).

In both higher plants and Escherichia coli, two enzymes are required for the production of GB via a two-step oxidation of choline (Hanson et al. 1985; Landfald & Strøm 1986; Yuwansiri et al. 2002). By contrast, GB synthesis in two soil bacteria, Arthrobacter globiformis and Arthrobacter pascens, requires only a single enzyme – choline oxidase (COD) – to catalyse the direct conversion of choline to GB (Ikuta et al. 1977). Genetic engineering to introduce GB-biosynthetic pathways into non-accumulator species appears to be a promising approach in efforts to increase their tolerance against environmental stress (for reviews, see Sakamoto & Murata 2000; Chen & Murata 2002). However, levels of GB, on a fresh-weight (FW) basis, differ considerably in transgenic plants (0.05–5 µmol g−1 FW) and among natural accumulators under stress conditions (4–40 µmol g−1 FW) (Rhodes & Hanson (1993).

Tomato plants do not normally accumulate GB. Under drought or saline conditions, for example, foliar application of GB enhances their stomatal conductance without affecting abscisic acid (ABA) metabolism, while also increasing protein and chlorophyll contents, and decreasing photo-respiration (Mäkeläet al. 1998, 1999). The majority of that exogenously applied GB accumulates in the cytosol and leads to an increase in tolerance to chilling stress (Park et al. 2006). On the other hand, the levels of GB accumulated in the chloroplasts of transgenic tomatoes are positively correlated with the degree of chilling tolerance (Park et al. 2004). However, GB contents in those transgenic plants are relatively low, only ∼0.3 µmol g−1 FW (Park et al. 2004). Therefore, it is important to know whether it is the localization of GB synthesis in various subcellular compartments and/or the amount of GB accumulation that determines the extent of tolerance to various abiotic stresses in transgenic plants.

In the current study, we generated three transgenic tomato plant types that targeted the gene product CODA to chloroplasts, the cytosol or both compartments simultaneously. For each transgenic type, five independent lines with varying levels of GB accumulation were used to investigate whether it is the subcellular location of GB accumulation or the total amount of GB that is the critical determinant for enhancing stress tolerance.

MATERIALS AND METHODS

Transformation vectors

The binary vector pCG/codA for chloroplast-targeted expression of the codA gene was constructed as described by Park et al. (2004). For non-targeted, i.e. cytosolic, expression of the codA gene, the binary vector pGAH/codANT was used (Hayashi et al. 1997).

For simultaneous expression of codA in both chloroplasts and cytosol, a third binary vector was constructed by transferring the codANT expression cassette from pUC/codANT (Hayashi et al. 1997) into the HindIII site of pGAH/codA (Hayashi et al. 1997). For purposes of clarification, we included the specifically targeted compartment when naming these three binary vectors pChl/codA (chloroplasts), pCyt/codA (cytosol) and pChlCyt/codA (both compartments) (Fig. 1).

Figure 1.

Schematic representation of T-DNA regions for pChl–codA, pCyt–codA and pChlCyt–codA expression cassettes. codA, the codA gene for choline oxidase; P-35S, CaMV 35S promoter; TP, sequence for transit peptide of small subunit of ribulose 1·5-bisphosphate carboxylase/oxygenase (Rubisco); T-NOS, terminator sequence of nos gene for nopaline synthase; P-NOS, promoter sequence of nos gene; GUS, the gus gene for β-glucuronidase; Int, intron of cat gene for catalase; nptII, neomycin phosphotransferase II gene conferring resistance to kanamycin; hpt, hygromycin phosphotransferase gene.

Plant material and transformation

The general transformation procedure for tomato (Lycopersicon esculentum cv. ‘Moneymaker’) followed that described by Park et al. (2004). Plants were selected on hygromycin (2 mg L−1) or kanamycin (50 mg L−1) selection media, and stable integration of the hpt or npt gene was confirmed by PCR. Five transgenic homozygous (T3) lines for each construct (pChl-codA, pCyt-codA or pChlCyt-codA) showing various levels of GB accumulation were chosen for the evaluation of GB dose-dependent stress tolerance. For other experiments, we selected from each construct (pChl-codA, pCyt-codA or pChlCyt-codA) one line that had accumulated the greatest amount of leaf GB.

Biochemical analyses

General biochemical analyses, including GB and H2O2 quantifications and a catalase assay, were conducted as described by Park et al. (2004).

Chloroplast isolation

Six-week-old transgenic plants, at the preflowering stage, were placed in the dark for 2 d, then intact chloroplasts were extracted using an isolation kit (Sigma, St. Louis, MO, USA) according to the manufacturer's protocol. The percentage of intact chloroplasts was determined by measuring ferricyanide photo-reduction before and after osmotic shock. Total chlorophyll concentrations were determined in 80% (v/v) acetone, per the manufacturer's instructions.

Measurement of chlorophyll fluorescence and ion leakage

The induction of chlorophyll fluorescence was recorded at RT using a pulse-modulated fluorescence monitoring system (FMS1) (Hansatech, Norfolk, UK). After adaptation in the dark for 30 min, the ratio of variable to maximum fluorescence (Fv/Fm) was measured.

Five leaf discs (0.8 cm diameter) per sample were used to determine ion leakage, as described previously (Park et al. 2004). Both measurements were conducted with the third to fifth compound leaves.

General plant growth and abiotic stress treatments

Wild-type (WT) plants and those from five transgenic lines per construct were examined at the stages of seed germination, in-vitro seedling development and the non-flowering phase. General growth and treatment conditions included: greenhouse (25 ± 3 °C, 16-h photo-period, 400–500 µmol m−2 s−1); growth chamber (3, 7, or 25 ± 0.5 °C, 16-h or 24-h photo-period, 500 µmol m−2 s−1); and cold room (3 ± 1 °C, 16-h photo-period, 100 µmol m−2 s−1).

Tolerance by WT and transgenic plants (Chl–codA, Cyt–codA and ChlCyt–codA) to three major abiotic stresses was determined as follows.

Chilling stress

Uniform 10-day-old in vitro seedlings were excised just above their root systems (approximately 2.0 cm long) and transplanted into Magenta GA-7 vessels (Magenta, Chicago, IL, USA) containing 50 mL medium [a full-strength Murashige and Skoog (MS) basal medium with 30 g L−1 sucrose and 7 g L−1 agar; pH 5.7]. After 3 d, they were chilled at 3 °C for 7 d in a cold room under continuous light, then transferred to a warm growth chamber (25 °C; 16 h photo-period). Seedling responses were quantified by hypocotyl length (cm). This experiment was repeated three times with 15 replicates each (45 seedlings/genotype/treatment).

Salt stress

Uniform 10-day-old in vitro seedlings were excised as described earlier for chilling stress, and were transplanted into Magenta GA-7 vessels containing 50 mL of an MSP medium with 125 mm NaCl. They were incubated in a growth chamber (25 °C) under continuous light for 2 weeks before individual FWs were measured.

Oxidative stress

Leaf discs (0.8 cm diameter) were collected from the third or fourth compound leaves of 5-week-old greenhouse-grown WT and transgenic plants (Chl–codA, Cyt–codA and ChlCyt–codA), and were subjected to vacuum infiltration for 30 min in the dark with either water or 5 µm methyl viologen (MV). They were then transferred to 6.0 cm Petri dishes containing 5 mL of either water or 5 µm MV, and were placed under lights (100 µmol m−2 s−1) at 25 °C for 16 h.

Statistical analyses

Data were statistically examined using analysis of variance (anova) methodology, and differences in value means were compared according to Duncan's multiple range tests. All analyses were performed with the SAS statistical program (SAS 2000).

RESULTS

Targeting CODA to the cytosol or both cytosol and chloroplasts significantly increased GB content in leaves

When targeted to the chloroplasts, cytosol or both compartments at once (Fig. 1), constitutive over-expression of the codA gene for CODA caused transgenic tomato plants to accumulate various amounts of GB in their leaves. GB levels were analysed in intact chloroplasts isolated from leaves, as was the total content in leaves from five transgenic lines of each construct (Table 1). Compared with leaf contents in the Chl–codA plants (24 nmol GB mg−1 chlorophyll or 0.3 µmol GB g−1 FW), those measured from Cyt–codA (96 nmol GB mg−1 chlorophyll or 1.5 µmol GB g−1 FW) and ChlCyt–codA (120 nmol GB mg−1 chlorophyll or 2.0 µmol GB g−1 FW) plants showed 5.0- and 6.6-fold higher accumulations of GB, respectively.

Table 1.  Glycinebetaine (GB) levels in chloroplasts isolated from leaves of five transgenic lines per constructa
Transgenic lineGB in leaves (nmol mg−1 chlorophyll)GB in isolated chloroplasts (nmol mg−1 chlorophyll)Intact chloroplasts (%)Total GB in chloroplasts (nmol mg−1 chlorophyll)GB in chloroplasts (%)
  • a

    Mean values ± SD from three experiments. GB content in chloroplasts was corrected for the percentage of broken chloroplasts present. The percentage of GB found in the chloroplast was calculated by comparing leaf and chloroplast contents, expressed on a chlorophyll basis.

Chl–codA-123.5 ± 2.89.3 ± 1.857.9 ± 5.116.1 ± 2.768.5 ± 6.3
Chl–codA-215.3 ± 1.96.9 ± 1.258.7 ± 5.811.7 ± 2.176.7 ± 7.7
Chl–codA-314.7 ± 2.17.1 ± 1.155.9 ± 4.812.7 ± 2.086.3 ± 6.9
Chl–codA-411.3 ± 1.84.3 ± 0.663.2 ± 5.76.8 ± 1.259.9 ± 5.6
Chl–codA-510.8 ± 1.53.6 ± 0.555.2 ± 6.76.5 ± 1.260.1 ± 7.5
Cyt–codA-195.6 ± 10.11.9 ± 0.166.8 ± 7.32.9 ± 0.63.0 ± 0.3
Cyt–codA-261.4 ± 5.81.8 ± 0.251.5 ± 6.33.6 ± 0.75.7 ± 0.7
Cyt–codA-344.9 ± 4.91.8 ± 0.259.8 ± 6.13.0 ± 0.66.7 ± 0.8
Cyt–codA-439.2 ± 3.11.7 ± 0.262.3 ± 4.52.8 ± 0.37.0 ± 0.7
Cyt–codA-529.1 ± 3.81.3 ± 0.270.1 ± 5.31.9 ± 0.46.4 ± 0.7
ChlCyt–codA-1119.7 ± 12.26.7 ± 0.950.1 ± 6.813.4 ± 2.111.2 ± 0.2
ChlCyt–codA-272.8 ± 6.05.1 ± 0.657.9 ± 7.18.8 ± 1.812.1 ± 0.2
ChlCyt–codA-369.6 ± 8.54.5 ± 0.755.3 ± 7.38.1 ± 1.611.7 ± 0.2
ChlCyt–codA-460.1 ± 8.03.9 ± 0.864.3 ± 4.56.1 ± 0.910.1 ± 0.2
ChlCyt–codA-547.5 ± 6.33.7 ± 0.568.3 ± 8.25.4 ± 0.811.4 ± 0.2

Although leaf contents were greater in Cyt–codA and ChlCyt–codA plants, the levels of chloroplastic GB were much higher in the Chl–codA plants (6.5–16.1 nmol GB mg−1 chlorophyll) than in the Cyt–codA plants (1.9–2.9 nmol GB mg−1 chlorophyll) (Table 1). ChlCyt–codA plants also accumulated GB in their chloroplasts, but to a lesser extent (5.4–13.4 nmol GB mg−1 chlorophyll). In other words, 60–86% of the total GB in leaves was localized to the chloroplasts in Chl–codA plants, compared with only 3–7% and 10–12% in the Cyt–codA and ChlCyt–codA plants, respectively (Table 1).

GB contents were measured from several tissue types in the transgenic plants (Fig. 2), with the highest levels being found in the reproductive organs. For example, GB accumulations were two- to fourfold higher in the anthers and pistils than in the leaves (Fig. 2). The distribution of GB in various organs from Chl–codA, Cyt–codA and ChlCyt–codA plants was directly proportional to the total content of GB in the leaves, i.e. Chl–codA < Cyt–codA < ChlCyt–codA.

Figure 2.

Glycinebetaine (GB) levels in various organs from 7-week-old transgenic plants. Chl–codA, Cyt–codA and ChlCyt–codA represent lines in which product of codA gene was targeted to chloroplast, cytosol or both subcellular compartments, respectively. Means of three independent experiments; bars indicate standard errors. L, leaves; SD, seeds; Pit, pistils; A, anthers; P, petals; St, stems.

GB accumulation in chloroplasts was more effective in protecting leaves against chilling stress-induced electrolyte leakage

One visible symptom of tomato chilling injury is necrotic leaves (Park et al. 2003, 2004). Therefore, to quantify the degree of cellular damage, we measured electrolyte leakage in the leaves following chilling treatment.

Under normal greenhouse conditions, seedling growth did not differ among the WT and three transgenic plant types (data not shown). At day 0, the % ion leakage showed a small but significant difference (P ≤ 0.05) among the four genotypes. After treatment at low temperatures, electrolyte leakage from the leaves was significantly greater (≤ 0.05) from the WT plants than from the transgenics (Fig. 3). After chilling at 3 °C for 3 d, the order of leakage was WT > Cyt–codA > ChlCyt–codA > Chl–codA. During a further 3 d of incubation at 25 °C, leakage continued to increase from both WT and transgenic cells (Fig. 3), but with the Chl–codA lines exhibiting a significantly lower percentage of leakage than the other two transgenic types (P ≤ 0.05). Thus, Chl–codA plants that had the lowest leaf GB content were more tolerant to chilling than were Cyt–codA or ChlCyt–codA plants that contained five- to sixfold higher levels of GB in their leaves.

Figure 3.

Effects of chilling on ion leakage from leaves of three codA transgenic plant types. Chl, Cyt and ChlCyt represent lines in which product of codA gene was targeted to chloroplast, cytosol or both subcellular compartments, respectively. Five-week-old greenhouse-grown transgenic plants, one line per construct, were incubated at 3 °C for 3 d, then at 25 °C for 3 d. Percentages of ion leakage were calculated using 100% to represent values obtained after autoclaving. Each reading is average of nine leaf discs (three discs per plant, three independent plants per genotype). Bars not labelled with same letter are significantly different by Duncan's multiple range test at P < 0.05.

GB accumulation in chloroplasts was more effective in protecting photosystem II (PSII) against photo-inhibition under chilling and oxidative stress

Photo-inhibition, the interruption of PSII activity under intense lighting, is caused by an imbalance between the rate of photo-damage to PSII and the rate of its repair (Nishiyama, Allakhverdiev & Murata 2006). Normally, photosynthetic organisms can overcome photo-damage via rapid and efficient PSII repair. However, under stressful conditions, such as strong light, chilling or high salt, the rate of photo-damage exceeds the repair rate, resulting in photo-inhibition (Nishiyama et al. 2006). One by-product of this phenomenon is the generation of excess reactive oxygen species (ROS), leading to oxidative stress. Here, we compared the degree of photo-inhibition in our WT and transgenic plants following chilling and oxidative stress treatments.

After 3 d of incubation at 3 °C, PSII activity in the leaves of WT plants declined to 40% of the original value (i.e. at day 0) while transgenic plants maintained 49–52% of their initial activity (Fig. 4a). The latter plant type appeared to be significantly less harmed than the former (P ≤ 0.05). Moreover, there was no significant difference in activity among the three types of transgenics even though they were quite varied in their leaf GB contents.

Figure 4.

Effects of chilling or oxidative stress on photo-inhibition of photosystem II (PSII). (a) Leaf discs (0.8 cm2) collected from 5-week-old wild-type (WT) and transgenic plants were incubated at 3 °C for 3 d under lights at 100 µmol m−2 s−1. (b) Leaf discs (0.8 cm2) collected from 5-week-old WT and transgenic plants were vacuum infiltrated for 30 min in water or 5 µm methyl viologen (MV), then incubated for 16 h. WT and transgenic plants, one line per construct, were used. Extent of photo-inhibition was defined by calculating means of Fv/Fm values. Each reading is average of 27 leaf discs (three discs per plant, three independent plants per genotype, three replications of entire experiment). Bars not labelled with same letter are significantly different by Duncan's multiple range test at P < 0.05.

We also examined transgenic tomato plants to determine if they were more tolerant of oxidative stress induced by MV. The reduction in PSII activity was used here to illustrate the extent of photo-inhibition. Under non-stress conditions, no significant differences in Fv/Fm values were noted between WT and transgenic plants. Treatment of leaves with light at an intensity of 100 µmol m−2 s−1 in the presence of MV caused a severe decline in PSII activity in leaf discs from both WT and transgenic plants (Fig. 4b), although activity remained significantly higher in the transgenic lines (P ≤ 0.05). Most importantly, plants of the Chl–codA line, which had only 1/5 to 1/6 as much leaf GB accumulation as the other transgenic types, still exhibited PSII activity similar to the other two.

To evaluate GB-dependent tolerance against photo-inhibition, we selected, from each construct, five transgenic lines accumulating various amounts of GB in their cytosol and/or chloroplasts, and exposed them to the same stress conditions described in Fig. 3. A positive correlation was found between the level of GB in the chloroplasts and plant tolerance to photo-inhibition (Fig. 5a) (R2 = 0.50; P < 0.01), whereas no correlation was apparent between cytosolic GB (Fig. 5b) and stress tolerance (R2 = 0.07; P > 0.33).

Figure 5.

Correlation between glycinebetaine (GB) content and degree of photo-inhibition under oxidative stress in transgenic plants. Chl, Cyt and ChlCyt represent lines in which product of codA gene was targeted to chloroplast, cytosol or both subcellular compartments, respectively. Extent of photo-inhibition, as measured by means of Fv/Fm values, was correlated with level of GB in chloroplasts (a) and cytosol (b). Data from five lines per construct for all three transgenic types were combined for calculation of correlation coefficients.

GB accumulation in chloroplasts was more effective in reducing the level of H2O2 and enhancing catalase activity

Under non-stress conditions, levels of H2O2 were 17–21% higher in transgenic plants than in the WT (Fig. 6a). However, after incubation at 3 °C for 1 d, the amount of leaf H2O2 in the latter was 3.7-fold higher than on day 0, whereas the accumulation of H2O2 in transgenic plants increased 2.2- to 2.4-fold (Fig. 6a). Among the three types of transgenics, Chl–codA plants accumulated the lowest amount. After further incubation at 25 °C for 3 d, the levels of H2O2 continued to increase, at similar rates, in both WT and transgenic plants, but with the transgenic lines again showing significantly lower accumulations of H2O2 (P ≤ 0.05). Among the three transgenic types, the order of H2O2 content was Chl–codA = ChlCyt–codA < Cyt–codA.

Figure 6.

Effects of chilling on levels of hydrogen peroxide (a) and activities of catalase (b) in leaves of wild-type (WT) and codA-transgenics. Chl, Cyt and ChlCyt represent lines in which product of codA gene was targeted to chloroplast, cytosol or both subcellular compartments, respectively. Five-week-old greenhouse-grown transgenic plants, one line per construct, were incubated at 3 °C for 1 d, then at 25 °C for 3 d. Each reading is average of 27 measurements (three discs per plant, three independent plants per genotype and three replications of entire experiment). Bars not labelled with same letter are significantly different by Duncan's multiple range test at P < 0.05.

Under non-stress conditions, catalase activities in the transgenic plants were slightly, but not significantly, higher (P < 0.05) than in the WT (Fig. 6b). After chilling at 3 °C for 1 d, leaf catalase activity in the latter remained unchanged, while activity was significantly elevated in all three types of transgenic lines. After three more days of incubation at 25 °C, catalase activity was severely reduced in both WT and transgenic plants, compared with initial activity (day 0), but this decline was much smaller in the transgenics. On both days 1 and 4, there was no significant difference in catalase activity among the three types of transgenics.

GB accumulation in chloroplasts was more effective in protecting seedlings against chilling and salt stress

To examine whether our observations were specific to leaves or could be applicable to other tissues, we extended our tolerance comparison to evaluate seedling growth among the three transgenic plant types and their response to salt stress as well.

No difference in plant heights was observed among seedlings of WT and transgenic plants grown in vitro at 25 °C for 7 d (data not shown). However, after incubation at 3 °C for 7 d, the growth of transgenic seedlings was significantly faster (data not shown). After incubation at 25 °C for an additional 7 d, the WT hypocotyls (2.8 cm) were significantly shorter (P ≤ 0.05) than those of the transgenic seedlings (4.0–6.0 cm; Fig. 7a). Among the three transgenic types, Chl–codA seedling growth was significantly greater (P ≤ 0.05).

Figure 7.

Effects of chilling (a) and high salt (b) on growth of wild-type (WT) and transgenic seedlings. Chl, Cyt, and ChlCyt represent lines in which product of codA gene was targeted to chloroplast, cytosol or both subcellular compartments, respectively. (a) Length of hypocotyls after chilling. WT and transgenic plants, grown for 10 d from seeds, were then excised just above roots (approximately 2.0 cm long) and transplanted into growth media. After initial incubation for 3 d, seedlings were held at 3 °C for 7 d, then at 25 °C for 7 d. Effects of chilling were examined by calculating means of hypocotyl length. (b). WT and transgenic seedlings were transplanted to Murashige and Skoog (MS) media containing 125 mm NaCl, and incubated for 2 weeks. Effects of salt stress were examined by calculating means of fresh weight (FW) for individual seedlings. Each reading is average of 27 measurements (three discs per plant, three independent plants per genotype and three replications of entire experiment). Bars not labelled with same letter are significantly different by Duncan's multiple range test at P < 0.05.

Accumulation of GB in the transgenic lines also improved their tolerance to salt stress, as seen by their performance in 125 mm NaCl in vitro (Fig. 7b). At 14 d after treatment, the transgenic seedlings were 72% heavier than the WT. Although seedlings from all three transgenic lines grew better than the WT, the Chl–codA line, with the lowest GB level in its seeds, showed the best growth under salt stress (P ≤ 0.05).

GB accumulation in chloroplasts was more effective in protecting seed germination against chilling and salt stress

To evaluate whether higher levels of GB accumulation in the seeds could further increase chilling tolerance, those from the WT and three transgenic lines were exposed to 3 °C for 14 d. Afterwards, 23% of the seeds from each transgenic line germinated compared with only 1.4% of the WT seeds (Fig. 8a). One additional day of incubation at 25 °C resulted in a rate of 80–90% for all transgenic seeds, compared with a germination frequency of only 16% for the WT (Fig. 8a). Moreover, five more days at 25 °C were required for the WT seeds to reach a comparable (∼90%) rate. Even though the amount of GB in seeds from the Chl–codA line (0.10 µmol GB g−1 FW) was much lower than from those of the ChlCyt–codA (1.0 µmol GB g−1 FW) and Cyt–codA (0.3 µmol GB g−1 FW) lines (Fig. 8a), all their germination rates were similar, suggesting that the GB that accumulated in the plastids was most effective in protecting seeds against chilling stress.

Figure 8.

Effects of chilling (a) and high salt (b) on seed germination for wild-type (WT) and transgenic plants. Chl, Cyt and ChlCyt represent lines in which product of codA gene was targeted to chloroplast, cytosol or both subcellular compartments, respectively. (a) Seeds from WT and transgenic plants, one line per construct, were incubated for 14 d at 17/7 °C (light/dark) under 16 h photo-period, then at 25 °C for 7 d. (b) Seeds from WT and transgenic plants, one line per construct, were allowed to imbibe water for 1 d, then placed for 14 d on media containing 125 mm NaCl. All results shown are means ± SE from three independent experiments, each with 50 seeds per genotype per experiment.

To determine whether our GB-accumulating transgenic tomato plants were also more tolerant of induced salt stress, we allowed the seeds from WT and transgenic plants to imbibe in water for 24 h before being exposed to 125 mm NaCl. After 14 d, the frequency of germination was 25% in WT plants and >90% in all of the transgenics (Fig. 8b). These data indicate that GB was very effective in protecting transgenic seeds against salt stress.

DISCUSSION

GB contents in transgenic tomato plants can be increased by five- to sixfold

We have previously transformed tomato plants with the bacterial codA gene for COD. This resulted in the accumulation of GB and enhanced chilling tolerance during their entire life cycle, from seed germination to the reproductive stage (Park et al. 2004). Targeting of CODA into tomato chloroplasts, however, elicited only a very low level of GB accumulation in transgenic leaves (0.09–0.30 µmol g−1 FW) (Park et al. 2004). This amount is far less than that detected under stress conditions in natural GB accumulators, such as spinach (30–40 µmol g−1 FW) (Rhodes & Hanson 1993). A positive correlation has been reported between GB accumulation and the acquisition of tolerance to salt stress in maize (Zea mays) (Rhodes et al. 1989) and to cold stress in barley (Hordeum vulgare) (Kishitani et al. 1994). These studies have suggested that it might be possible to further improve stress tolerance via genetic manipulation that increases GB contents in natural non-accumulator plants (McCue & Hanson 1990).

In the current study, we generated transgenic tomato plants with CODA targeted to the chloroplasts (Chl–codA), cytosol (Cyt–codA) or both chloroplasts and cytosol simultaneously (ChlCyt–codA). The Cyt–codA and ChlCyt–codA lines accumulated up to 5.0- and 6.6-fold more, respectively, in their leaves compared with the Chl–codA lines (0.3 µmol g−1 FW) (Table 1). However, even the highest level of GB (2.0 µmol g−1 FW), in the ChlCyt–codA leaves, was still at least 20-fold lower than the amount measured in natural GB accumulators (cf. up to 40 µmol g−1 FW in spinach under stress conditions). Nuccio et al. (1998) have reported that the activity of phosphoethanolamine N-methyltransferase (P-EAMT) is 30- to 100-fold lower in tobacco compared with spinach, which suggests that this may be the main reason why the endogenous supply of choline is limited in non-accumulators, such as tobacco. Therefore, it is likely that the increased accumulation of GB in both Cyt–codA and ChlCyt–codA lines was caused by the greater availability of choline in the cytosol, whereas the lower activity of P-EAMT in tomato plants may have restricted the increase in GB accumulation in codA-transgenic plants compared with natural accumulators.

Two major factors limit the accumulation of GB in chloroplasts of transgenic plants: the availability of endogenous choline itself (Huang et al. 2000) and the transport of choline across the chloroplast envelope (McNeil et al. 2000). Huang et al. (2000) have introduced the metabolic steps for oxidation of choline to GB into three diverse species –Arabidopsis thaliana, Brassica napus and Nicotiana tabacum. In all three, the use of exogenous choline significantly increases GB accumulation, suggesting that this supplement is required for the enhancement of GB levels in transgenic plants. Furthermore, study of the labelling kinetics for choline metabolites has revealed that the import of choline into chloroplasts limits GB synthesis in those compartments (McNeil et al. 2000; Nuccio et al. 2000). In plants, the biosynthesis of choline occurs exclusively in the cytosol (McNeil et al. 2000). Therefore, McNeil et al. (2001) have attempted to increase the endogenous choline supply through a transgenic approach, using P-EAMT, a key enzyme in the plant choline biosynthetic pathway. Those P-EAMT transgenic plants contain up to 50-fold more free choline and accumulate 30-fold more GB than plants transformed with the vector alone. However, those previous experiments did not evaluate the effects on abiotic stress tolerance by transgenic lines with elevated GB contents.

Compartmentalization of GB accumulation in transgenic tomato plants

In analyzing the levels of GB in various organs from all three transgenic tomato types (Fig. 2), we found that the highest accumulation was in floral tissues. We had made a similar observation in codA-transgenic plants of A. thaliana, although COD levels were nearly the same among various organs (Sulpice et al. 2003). Therefore, we hypothesize that the increase in GB concentrations in the reproductive organs might be evidence of GB transport from leaves to sink organs. We have also reported that GB exogenously applied to the leaves of tomato plants is taken up and translocated to actively growing tissues, such as shoot apexes and flowers (Park et al. 2006). Thus, it is likely that GB accumulation in leaves, either through exogenous application (Park et al. 2006) or from endogenous synthesis by COD (this study), can be translocated to and accumulated in the flowers.

Among our transgenic lines, chloroplasts of the Cyt–codA lines accumulated 3.0–7.0% of the total GB in their leaves, compared with the Chl–codA line, in which such accumulation was 60–86% of the total (Table 1). Although GB levels in chloroplasts of the ChlCyt–codA lines accounted for only 9.6–13.2% of the total leaf GB, this concentration was much higher than in the Cyt–codA lines, but lower than in the Chl–codA lines. Targeting the GB-catalyzing enzyme to different subcellular compartments, including the chloroplasts, mitochondria and cytosol, has also been shown to enhance tolerance of the photosynthetic machinery to salinity and chilling (Hayashi et al. 1997; Takabe et al. 1998; Huang et al. 2000). Moreover, exogenous application of GB to tomato plants improves their tolerance to salt, drought and chilling (Mäkeläet al. 1999; Park et al. 2006). When GB is exogenously applied to leaves by spray, the absorbed GB remains mostly in the cytosol, but a very limited amount is also transported into the chloroplasts (Park et al. 2006). Although that amount of transported GB may be minimal, it appears to be sufficient to confer significant tolerance to chilling stress (Park et al. 2006).

We can also hypothesize that transport of GB from the cytosol to different subcellular compartments is restricted. In spinach, for example, the majority of salt-induced GB accumulation is in the chloroplasts; the GB concentration in chloroplasts isolated from control and salt-stressed plants is 0.67 and 6.65 µmol mg−1 chlorophyll, respectively (Robinson & Jones 1986). Such a large concentration gradient across the chloroplast envelope suggests the existence of a specific transport mechanism. Although little is known about the transport of compatible solutes in plants, Schwacke et al. (1999) have demonstrated that the product of the tomato gene LeProT1, a homologue to an Arabidopsis proline transporter, transports GB with high affinity, and proline and γ-amino butyric acid with low affinities, when expressed in yeast. Therefore, GB transport from the cytosol to chloroplasts in tomato plants is very inefficient, resulting in only a limited amount of GB that can be accumulated in the Cyt–codA lines.

When one considers the small volume within a chloroplast, the actual GB concentration in the Chl–codA lines may be much greater there than in the cytosol, when the same amounts of GB are found in both locations. Likewise, in chloroplast-targeted codA Arabidopsis plants, the leaves contain 1.0 µmol GB g−1 FW, and the GB concentration in chloroplasts is about 50 mm when one takes into account the intro-chloroplastic volume (Hayashi et al. 1997). Therefore, the chloroplastic GB concentration in our tomato Chl–codA lines may be much greater than that expressed by the absolute amount, suggesting that the concentration of GB in chloroplasts from those lines is likely to be high enough to confer substantial protective effects, as shown in this study.

GB accumulation in the chloroplasts is more effective in protecting tomato plants against abiotic stress

Genetic engineering of GB biosynthesis with the codA gene has been mostly targeted to the chloroplasts in a number of higher plants (reviewed by Sakamoto & Murata 2000). Such transgenic plants accumulate GB primarily in their chloroplasts, and exhibit tolerance to various abiotic stresses at a wide range of developmental stages (Park et al. 2004). Furthermore, Sakamoto, Alia & Murata (1998) have reported that, in rice plants transformed with chloroplast-targeted CODA, their photosynthetic machinery is protected against salt and cold stresses more efficiently than in non-targeted CODA plants, even though the latter accumulates five times more GB. Therefore, it is important to know whether it is the localization of GB synthesis in various subcellular compartments and/or the level of that accumulation that determines the extent of tolerance to various abiotic stresses in transgenic tomato. To investigate these possibilities, we compared the responses of WT and transgenic plant types at different developmental stages/organs.

In leaves of 5-week-old WT and transgenic seedlings exposed to chilling stress, all three transgenic lines showed improved tolerance over the WT, as defined by the extent of ion leakage (Fig. 3). Interestingly, the Cyt–codA plants with high GB contents in their cytosol appeared to be less efficient than the Chl–codA plants in protecting the cellular membrane. GB accumulations prevent membrane damage from various environmental stresses (Deshnium et al. 1997; Chen, Li & Chen 2000) through direct membrane stabilization (Rudolph, Crowe & Crowe 1986), maintenance of the water shell surrounding the surface-exposed membrane proteins (Coughlan & Heber 1982) or a reduction in lipid peroxidation that protects the membranous organelles, chloroplasts and mitochondria (Prasad, Anderson & Stewart 1994). When cold treatment is accompanied by illumination, a very strong correlation is found between the electrolyte efflux and lipid peroxidation in various chilling-sensitive plants, including maize, cucumber and millet (Parkin & Kuo 1989; Lukatkin 2003). Furthermore, Lukatkin (2003) has suggested that cold impairment of the cellular membrane reaches a peak after the main increase in ROS produced at a very early stage in the chilling period. As shown here in Fig. 6, our Chl–codA lines maintained the lowest level of H2O2 while the Cyt–codA line produced the greatest amount of H2O2 at day 4. This suggests that GB accumulation in chloroplasts more effectively reduces the H2O2 level, and subsequently, less damage to the integrity of the plasma membrane occurs in the Chl–codA line compared with the Cyt–codA line. Furthermore, the ion leakage rate in the ChlCyt–codA line was significantly lower than that in the Cyt–codA line, evidence of the additional protective effect gained via simultaneous chloroplastic and cytosolic GB accumulations in the former line (Fig. 3). However, chloroplastic GB contents in the ChlCyt–codA lines were still lower than those in the Chl–codA lines (Table 1), probably resulting in a lower level of protection than in the latter. This suggests that GB accumulation is more effective in the chloroplasts than in the cytosol for protecting membranes against damage from chilling stress.

We were surprised to find that, under non-stress conditions, the percentage of ion leakage was slightly higher in the transgenic lines than in the WT plants. The cause for this phenomenon is not yet known. As shown in Fig. 3, at day 0, the amount of leakage seemed to be proportional to the total amount of GB accumulation (Table 1). Therefore, transgenics that show a higher level of GB should also produce more H2O2 catalysed by COD. One possibility is that elevated levels of H2O2 in the transgenic plants may have increased this leakage.

We also compared the responses of WT and three transgenic types to chilling and salt stress at the seedling and the germination stage, respectively (Figs 7 & 8), as well as photo-inhibition under both chilling and oxidative stress (Fig. 4). In all parameters measured, the transgenics were generally more tolerant of these abiotic challenges. Among the three types of transgenic plants produced, those from the Chl–codA line exhibited equal (Fv/Fm, Fig. 4; and germination rate, Fig. 8) or higher (seedling growth, Fig. 7) levels of stress tolerance even though their GB contents were lower than in the other two transgenic types. Based on these observations, we can conclude that GB accumulation in the chloroplasts is more effective in protecting plants against such stresses.

Previously, we have shown that codA-transgenic tomato plants exhibit efficient protection of their PSII in a GB dose-dependent manner (Park et al. 2004). Here, we also found a significant correlation between levels of GB in the chloroplasts (R2 = 0.50, P ≤ 0.01) and degrees of tolerance to oxidative stress, whereas no correlation was evident between cytosolic GB and stress tolerance (Fig. 5). As we reported earlier (Park et al. 2004, 2006), the threshold level of GB to confer full protection against chilling stress is rather low (0.09 µmol g−1 FW). In Cyt–codA transgenic plants, cytosolic GB may have no positive effects on abiotic stress tolerance. However, a small amount of cytosolic GB can be transported into chloroplasts. Such small amount of chloroplastic GB may contribute to the observed increase in abiotic stress tolerance. This would explain at least partially the lack of correlation between cytosolic GB and stress tolerance. Our finding of a positive correlation between chloroplastic GB levels and degree of stress tolerance may suggest that increases in chloroplast GB levels may further enhance stress tolerance.

GB accumulation in the chloroplasts effectively reduces the level of H2O2 and enhances catalase activity under chilling stress

Under non-stress conditions, H2O2 levels were higher in the transgenic lines than in the WT plants (Fig. 6a), an observation similar to that noted for numerous plants transformed with the codA gene (Alia et al. 1999; Park et al. 2004). All of those have higher amounts of H2O2 under non-stress conditions, which suggest that maintaining H2O2 at a particular non-toxic threshold level in codA-transgenic plants might induce the expression of genes responsible for enzymatic detoxification of H2O2. Furthermore, exogenous application of GB can increase levels of H2O2 in tomato plants under non-stress conditions (Park et al. 2006). Demiral & Türkan (2004) have also reported that exogenously applied GB elevates catalase activity in a salt-sensitive rice cultivar under high-salt stress. They have suggested that such enhanced activity might result from the increased synthesis of catalase induced by greater H2O2 production during salt stress. Sulpice et al. (2002) have shown that application of GB to both canola and Arabidopsis leaf discs induces their accumulation of both glutamine and glycine. That of the latter in canola, however, is restricted when GB is supplied along with glycolate pathway inhibitors, suggesting a possible interaction between GB accumulation and photo-respiration in mitochondria. Glycolate from the chloroplasts diffuses to the peroxisome where it is oxidized to glyoxylate by a glycolate oxidase (GO)-mediated reaction that yields H2O2. Therefore, it is very likely that the higher accumulation of H2O2 in our transgenic lines was caused by both COD and GO, resulting in greater catalase activity compared with the WT plants.

GB accumulation in the chloroplasts protects PSII against photo-inhibition under chilling and oxidative stress

Although exposure to light can cause photo-damage to PSII in most photosynthetic organisms, such harm can be overcome by the rapid and efficient repair of PSII under normal lighting conditions (Nishiyama et al. 2006). When photosynthetic cells are subjected to environmental stresses, for example, high salt or temperature extremes (Alia et al. 1999; Allakhverdiev et al. 2002; Ohnishi & Murata 2006; Yang et al. 2007), the degree of photo-damage to PSII is unaffected, whereas the repair of damaged PSII is inhibited because of the production of ROS, resulting in severe photo-inhibition. Under non-stress conditions, a ROS-scavenging system consisting of antioxidant enzymes and metabolites can reduce ROS amounts to tolerable levels. However, under stress conditions, the absorption of photons may exceed the capacity of the photosynthetic machinery for photosynthesis, thereby accelerating ROS production and causing elevated levels of ROS that give rise to oxidative stress. Nishiyama et al. (2006) have presented evidence to suggest that: (1) the primary sites of photo-damage are the oxygen-evolving complex and the D1 protein; (2) ROS do not accelerate photo-damage to PSII but inhibit its repair; and (3) protein synthesis is a specific target of the actions of ROS, specifically the step of translational elongation. The protective effect conferred by GB has been described as one that guards the machinery required for degradation and synthesis of D1 under stresses from high salt (Ohnishi & Murata 2006), chilling (Alia et al. 1999) and high temperature (Yang et al. 2007). In the current study, we found that, compared with the WT, all three transgenic types exhibited enhanced tolerance of PSII to photo-inhibition under both chilling and oxidative stress (Fig. 4). The accumulation of GB in the chloroplasts might be the most efficient in protecting cells against chilling and oxidative stress-inducible photo-inhibition. Therefore, our working hypothesis is that the GB accumulated in the chloroplasts may reverse the detrimental effects of ROS on the repair of photo-damaged PSII. Further studies are needed to address this question.

Ancillary