Establishing glucosylglycerol synthesis in potato (Solanum tuberosum l. cv. Albatros) by expression of the ggpPS gene from Azotobacter vinelandii

Plants and bacteria synthesize compatible solutes to adapt towards various environmental stress conditions. Moderate halotolerant cyanobacteria and some heterotrophic bacteria accumulate the compatible solute glucosylglycerol (GG), a compound not commonly found in higher plants. We aimed to investigate whether GG can be produced in potato and if it has an improving effect regarding stress tolerance. Potato plants were transformed with the ggpPS gene of Azotobacter vinelandii, which codes the bifunctional enzyme GG-phosphate phosphate/synthase (GgpPS) catalyzing the two-step synthesis of GG. Potato plants were generated expressing ggpPS under control of the constitutive CaMV-35S promoter or the stress-inducible rd29A promoter. The integration of T-DNA in transgenic clones was verified by PCR and transgene expression was detected on mRNA level. Transgenic potatoes with constitutive ggpPS expression accumulated GG up to 19 μmol/g fresh mass (FM) in leaves but no GG in tubers, while plants with rd29A-controlled ggpPS expression contained 10 μmol GG/g FM in leaves and up to 2.6 μmol GG/g FM in tubers. In greenhouse experiments, an increased shoot growth of the GG-accumulating potato plants was observed under salt and drought stress conditions. These results demonstrate that GG synthesis can be achieved in potato plants and might have a protective role on plant metabolism.


Introduction
Drought and elevated soil salinity limit the distribution and productivity of crop plants worldwide. Therefore, improvement of drought and salt tolerance is an important challenge in modern plant breeding. To achieve this goal classical breeding techniques as well as genetic engineering are used [1,2,3]. Central for successful salt acclimation is the avoidance of high concentrations of toxic ions in the cytosol by active extrusion or compartmentation of ions and a parallel accumulation of socalled compatible solutes in metabolic active parts of the cell. The compatible solutes act by enhancing the internal osmotic potential and thereby improving the water status of the cell. In addition, they are able to protect proteins and membranes directly against denaturation [4,5].
An increased accumulation of compatible solutes in plants was obtained by two strategies: (i) by the enhancement of synthesis or the reduction of breakdown of naturally occurring compatible solutes, e.g. proline [6] or trehalose [7], or (ii) by the expression of transgenes, which encode pathways for the synthesis of compatible solutes from halotolerant (micro) organisms (for reviews see [8,9]). Different chemical classes of compatible solutes were successfully transferred, e.g. the carbohydrate trehalose [10,11], the polyols mannitol [12] and sorbitol [13], or the amino acid derivatives glycine betaine [14,15] and ectoine [16]. Although the transgenes were expressed in most of the cases, the detected levels of compatible solutes were either relatively low (0.01-4.6 µmol/g fresh mass (FM)) or the accumulation was accompanied by pleiotrophic effects, when the detectable levels of these osmolytes were rather high (e.g. for sorbitol accumulation in transgenic tobacco, [13]). However, an improved growth of the transgenic plants under drought, high salt, cold, or high light stress conditions has been reported in several cases [9,11].
Microorganisms such as the photoautotrophic cyanobacteria adapted to almost all habitats on Earth including extreme environments such as hypersaline ponds. The molecular biology of the cyanobacterial salt acclimation has been intensively studied using different model strains [17]. Moderate halotolerant cyanobacteria accumulate the heteroside 2-O-(a-D-glucopyranosyl)-glycerol (glucosylglycerol, GG) as main compatible solute [18]. The biosynthesis of GG resembles the twostep synthesis mechanism of trehalose or sucrose [19]. Thereby, glycerol 3-phosphate and ADP-glucose are converted by the activity of the GG-phosphate synthase to GG-phosphate, which is dephosphorylated in a second step by the specific GG-phosphate phosphatase resulting in GG [20,21,22]. Beside cyanobacteria, GG was also found in salt-treated heterotrophic bacteria [23,24]. For the salt-induced GG synthesis these bacteria express the

Plant Science & Molecular Breeding ISSN 2050-2389
bi-functional protein GG-phosphate phosphatase/synthase (GgpPS) that is composed of an N-terminal phosphatase domain fused to a C-terminal synthase domain, thus this enzyme can catalyze both reactions. Moreover, the GgpPS enzymes were shown to accept ADP-glucose as well as UDP-glucose as glucosyldonors [25]. The resurrection plant Myrothamnus flabellifolia represents the only known land plant species accumulating GG under drought stress [26], while some Liliaceae synthesize it constitutively [27]. However, so far nothing is known about the GG synthesis mechanism in these plants. Additional to the osmotic function in the natural GGaccumulating cells, the protective ability of GG was verified, because it protects soluble enzymes as well as membranes against denaturation in vitro [28,29].
Recently we reported that the transfer of the GG biosynthesis enzyme GgpPS from the rhizobacterium Azotobacter vinelandii into the model Arabidopsis thaliana led to GG accumulation and improved stress tolerance [30]. In the present study we aimed to investigate if GG synthesis can also be achieved in the crop plant Solanum tuberosum. Such transgenic crop plants may show sustainable growth under saline or drought conditions. Moreover, significant GG accumulation in potato tubers could be used as raw material for GG purification, which is used under the trade name glycoin in cosmetics or pharmaceuticals as moisture or stabilizer (http://www.bitop.de; [31]).

Material and Methods Plant material, bacteria and growth media
Potato plants (Solanum tuberosum L. cv. Albatros) were obtained from the "NORIKA GmbH" (Germany). In vitro cultured plants were maintained on Murashige Skoog (MS) medium [32] in plastic containers under a 16 h light (22°C)/8 h dark (18°C) regime at 90 µmol photons/m 2 /s. Approximately 3 cm long shoots were cut and transferred onto fresh medium under sterile conditions every 8 weeks. In vitro plants were then transferred directly onto soil (4:1 mixture of soil (Type VM Mini Tray; Einheitserdewerk, Uetersen) and vermiculite, soaked with 0.2% Wuxal liquid fertilizer (Aglukon)), in which they differentiated roots. Regenerated potato plants were grown in a greenhouse on soil in pots of 20 cm diameter containing 7.5 l soil. Greenhouse control conditions were set to 16

Construction of vectors used for plant transformation
For experiments with constitutive expression of the ggpPS from Azotobacter by the CaMV 35S promoter, we used the vector pLH-AF as described previously [30]. In order to generate a vector expressing ggpPS via the stress-inducible rd29A promoter [34], the ggpPS gene was amplified using the primers AFggpPS-HpaI-5' (5'-CGTTAACATGCTACTTGCCACCGA-3') and AFggpPS-HpaI-3' (5'-CGTTAACTCAGGCGATGGCGCGGA-3'), which generated an HpaI restriction site at each end of the ggpPS coding sequence. The PCR product was then inserted into the vector pGEM-T (Promega) and sequenced. Correct ggpPS gene fragments were obtained by HpaI restriction and then fused to the CaMV 35S promoter and terminator in the SmaI-cut pGreen 35S-2 cassette-vector. The rd29A upstream region from Arabidopsis thaliana was amplified using primers that were described by [34] and modified by adding a 5'-HindIII (rd29A-5'-HindIII; 5'-CAAGCTTCGACTCAAAACAAACTTACG-3') and 3'-BamHI (rd29A-3'-BamHI; 5'-CGGATCCAATCAAACCCTTTATTCCTG-3') restriction site, respectively. The PCR product was cloned in the pGEM-T vector and verified by sequence analysis. By using the restriction enzymes HindIII and BamHI, the rd29A promoter fragment was obtained and inserted upstream of the ggpPS gene. Subsequently, the entire expression cassette containing rd29A promoter, ggpPS coding sequence and CaMV 35S terminator was cleaved out via restriction by HindIII and EcoRV and cloned into the vector pGreenII0229. Finally, the expression cassette was obtained via restriction by KpnI and XbaI and transferred into pCambia1305.1, which harbors the hygromycin phosphotransferase gene (hptII) for transgenic plant selection. The resulting vector was named pCambia-rd29A-AF.

Extraction of total RNA for RT-PCR
For RNA gel blots, total RNA from leaves of the transgenic potato plants was extracted using the Invisorb Spin Plant-RNA Mini Kit (Invitek). For RT-PCR the Ready-To-Go RT Beads Kit (Amersham Pharmacia Biotech) was used. The first strand cDNA synthesis was carried out with the random primer pd(N)6 and amplification of transgene specific cDNA was carried out with gene specific primer ggpPS RT fwd (5'-AGCCTGCTGCAATGCGACTA-3') and ggpPS RT rev (5'-TGAAGGCGACGATGTCGAAC-3'). The cDNA amounts were equalized according to the PCR product of the constitutively expressed 60S rRNA which was amplified using the primers 60s RT fwd (5'-GACAAGGCTGGAATGACTCACATTGTC-3') and 60s RT rev (5'-CTCCTCAAGAGCAACACGAGAGGTCTG-3').

Low molecular mass compound analysis
GG and other low molecular mass compounds were determined by gas-liquid chromatography [25]. These substances were extracted from 100 mg plant tissue. After homogenization, the material was treated with 1 ml 80% ethanol (containing 50 µg sorbitol as internal standard) at 68°C for 2 h, and then with 500 µl 80% ethanol at 68°C over night. After centrifugation (20,000 g, 10 min, 4°C) the supernatants were combined, dried in a vacuum centrifuge and re-suspended in 500 µl A. dest.. After a further centrifugation step, the supernatant was again evaporated by vacuum centrifugation. The pellets were re-suspended in 65 µl pyridine containing 20 mg/ml methoxylamine at 30°C for 90 min. Then, 35 µl N-methyl-Ntrimethylsilyl-trifluoracetamide were added. Subsequently, samples were incubated at 60°C for 1 h and centrifuged (20,000 g). Aliquots of the supernatants were injected into the gas-chromatograph (GC Focus including FID detector and Auto Sampler AS3000, Thermo Scientific).

Green house experiments
For the growth experiments transgenic plants from in vitro culture were planted in soil (7.5 l pots with 20 cm diameter) and transferred into a greenhouse (long day, 16 h light (22 °C)/8 h dark (20°C) regime at approximately 180 µmol photons/m 2 /s). Wild-type and transgenic plants (three independent clones with each five individuals from each series were selected) were first cultivated for 4 weeks under control conditions (see above). After this growth phase, the pre-stress samples were collected (3 rd and 7 th leaf from the top), frozen in liquid nitrogen and stored at -80°C. Then, the stress treatments were started and continued for the next 4 weeks. Preliminary experiments with wild-type plants were performed to define stress conditions affecting the potato growth without toxic effects. Accordingly, five individuals from each clone or wild type were held at normal irrigation (control, watered with 150 ml every day), five other plants were irrigated with 150 ml water containing 100 mM NaCl (salt stress, the salt addition summed up to a final concentration of 60 mM per pot volume), or irrigated with one half of the amount of water used for control conditions (75 ml, drought stress). At defined time points (1, 2, 3, or 4 weeks after onset of stress, i.e., with 5, 6, 7, 8 weeks old plants) or at the end of the 4 weeks stress period (i.e., 8 weeks old plants, just before starting flowering) samples were taken for GC analyses. Shoot growth of every plant was determined once a week. Shoot branches were removed to ensure comparable measurements of the main shoot. Due to space limitations, this pot size did not allow a reliable estimation of tuber yield. Each greenhouse experiment was repeated 3times. Statistical significant differences were verified using t-test (P-values = 0.5).

Generation and selection of transgenic potatoes
To establish the GG biosynthesis in potato, the gene ggpPS for GG-phosphate phosphatase and GG-phosphate synthase from Azotobacter vinelandii [25] was first cloned under the control of the constitutive CaMV 35S promoter (series Alba-AF). None of the selected transgenic plants showed obvious morphological differences in comparison to the wild type. The integration of the T-DNA in the potato genome was confirmed by PCR. This analysis showed the expected DNA fragment pattern: (i) in wild-type plants no transgene could be detected; (ii) for transgenic plants of the Alba-AF series the occurrence of nptII and ggpPS; and, (iii) for transgenic plants of the Alba-rd29A-AF series the occurrence of hptII and ggpPS were verified (Figure 1).

Expression of ggpPS in transgenic potato plants
Selected clones of the series Alba-AF were analyzed for expression of ggpPS at mRNA levels using RT-PCR analyses (Figure 2). These semi-quantitative data showed expression of the ggpPS gene in various organs: young and old leaves, stolons as well as tubers. Furthermore, the expression data indicate that the expression level of transgenes under control of the CaMV 35S promoter is almost similar in the different plant organs and showed no direct correlation with the GG contents (discussed below).

Biosynthesis of glucosylglycerol (GG)
The presence of GG in the transgenic potato plants verified that the ggpPS gene was not only transcribed but translated into an enzymatic active protein. The independently obtained transgenic clones of the series Alba-AF showed great differences in the detected GG levels. Plants grown on MS medium showed lower GG accumulation (0.02 up to 0.6 µmol GG/g FM) than plants cultivated in soil (0.2 up to 26 µmol GG/g FM in 3 rd leaves of 4 weeks old plants). Another striking difference in the GG amount became visible when leaves of different ages were compared. Older leaves, for example we analyzed different leaves of 4 weeks old plants (i.e., leaves 3 (youngest) -9 (oldest) from the top were sampled), showed generally lower GG levels than younger leaves (Figure 3). Nevertheless, the quantitative differences among different transgenic plants of the series Alba-AF remained visible during all experimental series, i.e. plants having higher GG values in young leaves retained higher GG levels also in older or stressed leaves. Moreover, we observed significant GG accumulation in stolons (up to 20 µmol GG/g FM), but GG was never found in fully developed tubers in potato plants with CaMV-35S-promoted ggpPS expression.

Growth of plants controlling ggpPS expression by the CaMV 35S promoter
For stress experiments three independently obtained transgenic clones (Alba-AF1, Alba-AF8, and Alba-AF9) were chosen. Alba-AF8 and Alba-AF9 plants showed rather high levels of GG accumulation (in young leaves 17 µmol GG/g FM), while plants of the transgenic line Alba-AF1 showed significantly lower GG accumulation (30% less) (Figure 3, Table 1). Under salt and drought stress all potato plants showed significant growth retardation compared to the plants grown under control conditions (Figure 4). However, we found significantly improved shoot growth of all transgenic GG-accumulating plants under drought stress conditions compared to drought-stressed wild-type plants. The transgenic clone Alba-AF1 also showed increased shoot heights under control and salt stress conditions compared to the wild-type and the other transgenic potatoes (Figure 4).

Experiments with plants controlling ggpPS expression by the rd29A promoter
Additionally, we generated transgenic potato plants with ggpPS under control of the rd29A promoter [35] (Series Alba-rd29A-AF). Despite the control of ggpPS expression by the salt-stress-induced rd29A promoter, we also found GG accumulation in these transgenic clones under control conditions ( Figure 5, Table 1). Under control conditions, the GG levels in young leaves of plants of the series Alba-rd29A-AF were 2-3times lower than in plants of the series Alba-AF. However, application of a salt stress doubled the GG amount in transgenic clones of the series Alba-rd29A-AF compared to plants of the same genotype grown under control conditions ( Table 1). Moreover, GG remained detectable in leaves of the salt-stressed Alba-rd29A-AF plants till the end of the 8 weeks growth period, i.e. we found no sharp decrease in GG levels of older leaves (Figure 5). Interestingly, in this plant series GG also was found in tubers. While under control conditions only low GG levels with high variation were detected, salt stress significantly increased the amount of GG in tubers, it reached up to 5 µmol GG/g FM (Figure 5).
Three transgenic clones with relatively high levels of GG accumulation (Alba-rd29A-AF5, Alba-rd29A-AF9, and Alba-rd29A-AF16) were chosen for salt stress experiments in the greenhouse. Again, under salt stress conditions all potato plants showed significant growth retardation compared to control conditions (Figure 6). However, we found significant improved shoot growth of the transgenic plants of clone Alba-rd29A-AF16 under salt stress conditions, comparable to the 6 doi: 10.7243/2050-2389-2-1 phenomenon in clone Alba-AF1. Interestingly, all transgenic lines also showed significantly increased shoot heights under control conditions compared to wild type (Figure 6).

Effects of GG accumulation on contents of soluble sugars in potato leaves
Additional to GG, we estimated the amounts of sucrose, glucose and fructose in leaves. Analyses of extracts from the 3 rd leaf of 4 weeks old potato plants (

Table 1. Contents of low molecular mass compounds in young leaves (leaf number 3) of wild-type potato plants and GGaccumulating transgenic clones (Means ± confidence intervals, α=0.05). Statistical significant differences between transgenic and corresponding wild-type plants are shown in bold face.
Samples were taken before the stress and 4 week after treatments with normal watering (control), with salt-enriched watering (salt stress), or with reduced watering (drought stress). as of the rd29A promoter. The increased amount of total soluble compounds is mainly due to the accumulated GG, whereas the levels of sucrose, glucose and fructose were slightly lower in transgenic plants compared to wild-type plants. The difference in total soluble sugars between wildtype and transgenic plants of the series Alba-AF disappeared when 8 weeks old plants were analyzed grown under control conditions. In those plants the amount of GG decreased to trace amounts, while the contents of sucrose, fructose and glucose were comparable between transgenic and wild-type plants ( increased the amount of soluble compounds under salt stress ( Table 1). This enhancement of soluble sugars is mainly based on the salt-triggered increase in the GG levels, since the expression of the ggpPS gene is controlled by the salt-stressregulated rd29A promoter. The amounts of sucrose, glucose and fructose became slightly lowered in transgenic compared to wild-type plants under salt stress conditions.

Discussion
In the present study the gene ggpPS, encoding the enzymes for biosynthesis of GG from A. vinelandii, was successfully expressed in potato as before in Arabidopsis [30]. The amount of GG detected in the transgenic lines was rather high; it exceeded about 10times the sucrose level in young leaves. The GG levels were similar to that observed in our previous study using Arabidopsis as a host [30]. Also in potato plants, these high amounts were tolerated without any strong pleiotropic effects, which can be taken as a clear indication for the compatibility of GG with plant metabolism in general. This finding is in contrast to several other studies, where the synthesis of compatible compounds (e.g. sorbitol) after transgene expression was accompanied by pleiotropic effects such as infertility or necrosis, when the content of the compound exceeded 5 µmol/g FM (e.g. [13]). In fact, the high level GG-accumulating potato plants showed even better growth under standard conditions, whereas the GG-accumulating Arabidopsis plants rather showed a negative correlation between GG level and growth, i.e. transgenic lines with GG-levels of 20 µmol/g FM showed smaller rosette diameters whereas plants accumulating only 2 µmol/g FM were not affected [30].
In contrast to GG-accumulating Arabidopsis [30], we observed changes in GG accumulation depending on the age of the plant when the ggpPS gene was expressed by the constitutive CaMV 35S promoter. Generally, young developing leaves and also stolons contained high amounts of GG, whereas fully developed source leaves and tubers were virtually free of GG. This change was not expected because the ggpPS gene in plants of the series Alba-AF is expressed by the CaMV 35S promoter thought to be constitutively active in plants. However, there are reports on expression differences in plant organs of different ages using this promoter for transgene expression [36,37,38]. Our semi-quantitative RT-PCR showed almost similar ggpPS-mRNA levels in young and old leaves, stolons and tubers as well. Obviously, expression changes are not sufficient to explain the developmental change of GG accumulation. Interestingly, sucrose contents showed a behavior opposite to that of GG, i.e. young organs contained rather low sucrose amounts which increased during further age (see Table 1). Both, the biosynthesis of GG and sucrose use UDP-glucose as precursor. Moreover, older leaves develop to source organs exporting high amounts of sucrose to other plant parts. This differentiation is accompanied by an increased expression of enzyme sucrose-phosphate synthase (SPS) in older potato leaves [39]. Possibly, the competition for precursors and/or the developmental increased sucrose synthesis could explain the diminished GG levels. We also noticed that the GG amount in transgenic potato plants was much lower when shoots are grown on sucrose-containing MS medium, which indicates that sucrose itself might have a negative regulatory effect on GG synthesis. Sucrose and especially hexoses are known to act as direct regulators on numerous processes in plants [40]. Moreover, a GG degrading activity or its export could become active in older leaves, since sampling of the same leave at different time points revealed also a disappearance of GG in leaves initially characterized by high GG levels.
We also generated plants in which the expression of this transgene was controlled by the salt-stress-regulated promoter rd29A. This promoter was identified in Arabidopsis and shown to have a basal expression, which is significantly increased under high salt and other stress conditions [35,41]. There are many reports that the stress-regulation of rd29A is not only found in its native host Arabidopsis but also in crop plants such as potato (own results, [34,42]), rice [43], mulberry [44], soybean [45], and peanut [46]. The relatively stable GG accumulation in transgenic potatoes of the series Alba-rd29A-AF indicated that changes in the relative expression of ggpPS and genes for competing enzymes such as sps might be at least partly responsible for the developmental change in GG accumulation in plants expressing ggpPS by CaMV 35S promoter. The increase of rd29A-controlled ggpPS expression under salt stress conditions (data not shown) resulted in doubled GG amounts in older leaves ( Table 1) and elevated GG levels in tubers. Again, a reverse relation of GG and sucrose is visible, i.e. salt-stressed plants with enhanced GG levels decreased the sucrose levels, while leaves of older control plants with slightly decreased GG amounts contained more sucrose. Summarizing, the use of the stress-inducible rd29A promoter offered clear advantages for the stable expression of ggpPS in potatoes compared to the CaMV 35S promoter control.
The presence of high amounts of GG in the transgenic potatoes made it promising to search for improvements of tolerance towards salt stress in greenhouse experiments. In the case of GG accumulation in transgenic potato plants we found an increase in shoot growth under control and drought stress conditions of up to 30%. This growth improvement was found with plants obtained from independent explants. Similar growth promoting effects under control and stress conditions were reported before for the GG-accumulating model Arabidopsis [30]. Comparable to Arabidopsis, only plants of the clone Alba-AF1 and Alba-rd29A-AF16 with an intermediate GG amount showed an improved growth under salt stress conditions, while plants of clones with higher GG accumulation (e.g. Alba-AF8 or Alba-AF9) showed no salt-stress improvement. Whereas the plants of the series Alba-rd29A-AF contained high GG amounts during the whole experiment, plants of the series Alba-AF retained only traces of GG. These findings point for a more direct protective effect of GG for critical macromolecules rather than an osmotic effect. That