The experiment was carried out at Gansu Agricultural University in 2012 and 2013. The potato tetraploid cultivar “Shepody” was used in all tests. The plants were propagated in vitro by subculturing single-node cuttings on MS liquid medium supplemented with 3% (w/v) sucrose and grown under light intensity of 60–80 μmol m⁻² s⁻¹ and a photoperiod of 8 h d⁻¹ at 22 ± 1°C. After culturing for 25 d, MS liquid medium was removed and the microtuber inducing medium was added with the supplement of 8% (w/v) sucrose and 5 mg L⁻¹ of 6-BA. Plantlets were grown under dark at 22 ± 1°C and microtubers were harvested from the 8-week-old plants.

The 867 bp cDNA of the StPIP1 gene (DQ999080) isolated from potato cultivar “Gannongshu No.2” was cloned into the BamH I –Sac I site of the plasmid pStPIP1 containing CaMV 35S promoter (Wu et al., 2009), such that the β-glucuronidase (GUS) gene was replaced by StPIP1 gene. The recombinant plasmid pStPIP1 was introduced into Agrobacterium tumefaciens strain LBA4404 using the freeze-thaw method (Höfgen and Willmitzer, 1988). The presence of the plasmid was verified by restriction enzyme digestion and PCR amplification.

Slices of microtubers of the potato cultivar “Shepody” were used as the receptor for Agrobacterium-mediated transformation, which was performed following the procedures described elsewhere (Si et al., 2003). The regenerated shoots of 10.0 mm in height were transferred to the selective rooting medium (MS+100 mg L⁻¹ Kanamycin+200 mg L⁻¹ Carbenicillin). Genomic DNA was isolated from 0.1 g of leaves from putative transgenic and non-transgenic control plants using the cetyl-trimethyl ammonium bromide (CTAB) method (Edwards et al., 1991). The presence of transferred neomycin phosphotransferase II (Npt II) gene was confirmed using standard PCR technique. The Npt II gene was amplified with the forward primer 5′-GCTATGACTGGGCACAACAG-3 ′ (1–18 bp) and the reverse primer 5′-ATACCGTAAAGCACGAGGAA-3′(657–676 bp), which generated an expected PCR product fragment of 676 bp. Amplification was performed in a thermal cycle (UNO II, Biometra) programmed for one cycle of 3 min at 94°C, followed by 30 cycles of 45 s at 94°C, 45 s at 59°C, and 1 min at 72°C. A final extension step was performed for 5 min at 72°C. The PCR products were electrophoresed on a 0.8% agarose gel, denatured, and transferred to a positively charged nylon membrane (Roche). Membranes were hybridized at high stringency in hybridization buffer at 42°C for 16 h. Membranes were washed twice in 2 × SSC, 0.1% (v/v) SDS at room temperature for 15 min each, and twice in 0.5 × SSC, 0.1% (v/v) SDS at 65°C for 15 min each. The gene-specific probe of the Npt II gene was labeled using a DIG High Prime DNA Labeling and Detection Starter Kit I (Roche). The hybridization signal generated using NBT/BCIP color substrate was detected after 20 h.

Genomic DNA was extracted using the CTAB method (Edwards et al., 1991). Forty micrograms of genomic DNA were digested with the restriction endonuclease HindIII and separated on 0.8% agarose gel, denatured, and transferred to a positively charged nylon membrane (Roche). Membranes hybridization and detection methods were the same as those used in the PCR-Southern analysis described above.

Total RNAs were isolated using the RNA simple Total RNA Kit (TIANGEN, lot#N2822) following the manufacturer's instructions. Reverse transcription was performed in 20 μL reaction mixtures using the PrimeScript™ RT reagent Kit with gDNA Eraser (TaKaRa, RR047A). The expressed Npt II gene was confirmed using standard PCR techniques. The Npt II gene primers and amplification methods were the same as those in the PCR analysis described above. The RT-PCR products were electrophoresed on a 0.8% agarose gel.

Seven transgenic potato lines (designated as T1, T2, …T7) and the control (a non-transgenic potato cultivar, NT) were included in the test. For plant mannitol osmotic abiotic stress treatments, the six-leaf-stage T7 and the non-transgenic (NT) plants were grown in liquid MS medium containing 3% (w/v) sucrose, and then the liquid medium was removed and replaced by 50 mL of liquid MS medium containing 3% (w/v) sucrose and 25 mM mannitol. Plant leaves were collected at 0, 6, 12, 24, 48, and 96 h after stress treatment, respectively. The entire experiment was run three times. The samples were cleaned with sterilized water, immediately frozen in liquid nitrogen, and stored at −80°C for subsequent analysis.

Total RNA was extracted from mannitol-treated samples of T7 and non-transgenic (the control) potato plantlets using the RNA simple Total RNA Kit (TIANGEN, lot#N2822). The RNA was quality checked and quantified using a Nanodrop ND-1000 (Nanodrop Technologies, USA). Reverse transcription was performed in 20 μL reaction mixtures using the PrimeScript™ RT reagent Kit with gDNA Eraser (TaKaRa, RR047A), and qRT-PCR amplification was conducted in 20 μL reaction mixtures with the SYBR® Premix Ex Taq™ II(Tli RnaseH Plus) (TaKaRa, DRR820A), and 10 μM of each primer (ef1a as internal control gene and forward and reverse primers: 5′-CAA GGATGA CCC AGC CAA G -3′ and 5′-TTCCTT ACC TGAACGCCT GT-3′; StPIP1 gene-specific forward and reverse primers: 5′-TGTGGG TAT TCA AGGAGTTGC T-3′and 5′ -CCA AGA ACA GAC CAA ATG TCA C-3′). Reactions were conducted on a Mxprosystem (Applied Stratagene Mx3005p real-time PCR) using the default cycling conditions (30 s at 95°C and 40 cycles of 5 s at 95°C, and 34 s at 60°C). The transcripts of stress-responsive genes are shown as relative transcripts compared with NT under non-stress growth conditions. Each Quantitative Real-Time PCR was repeated three times independently. A blank control was included where no cDNA was added in the reaction mixtures. After each reaction, melt curve analysis was used to verify the specificity of amplification and the relative expression levels calculated by 2^−ΔΔCt.

Mini-tubers of the transgenic potato lines carrying the StPIP1 gene were evaluated for their stress tolerance in controlled-environment greenhouses. Potato mini-tubers were produced from test-tube plantlets following the established procedures (Li et al., 2012). One mini-tuber recovered from dormancy was seeded in each pot (25 cm in height and 32 cm in diameter) filled with 15% vermiculite (v/v) and 85% loessal soil under room temperatures. A daily watering was provided to each pot to maintain soil moisture at 70–80% field capacity. Healthy plants with a similar physical size were selected for drought treatments and those plants that survived the drought challenge were measured further.

All seven transgenic lines and the non-transgenic check were tested under three water treatments: 80–90% field water holding capacity (FC) as non-stress treatment, 55–60% FC as moderate drought, and 30–35% FC as severe drought. The three water stress treatments were implemented by measuring soil moisture at 10:00 and 16:00 h each day using moisture meters TDR-300 (Spectrum®, USA) that were pre-installed in each pot, and an amount of water was, respectively added to each plot to bring the soil water content to the pre-designed level.

The treatments were arranged in a completely randomized design with three replicates. Each treatment in each replicate had 10 pots, one plant per pot, with a total of 720 pots (8 lines × 3 stress-treatments × 3 replicates × 10 pots each). The sufficient number of plants allowed all measurements and analyses to be taken below. The day/night temperatures in the greenhouse were set at 25/15°C and relative humidity at 70 to 75%. The day/night photoperiod was 14/10 h. Light intensity was 3500 lx. At Day 30 after drought treatments, plant heights were measured from the soil level to the tip of the main stem, the leaf size was measured with the weighted estimate of paper outline, and the stem thickness were measured using a vernier caliper. The relative expansion rate of leaf area was expressed as: leaf area expansion rate = (the average leaf area at Day 30 after drought treatment–the average leaf area before the treatment)/30. Daily gain of plant height = (the average plant height at Day 30 after drought treatment−the average plant height before the treatment)/30. Stem thickness expansion rate = (the average stem thickness at Day 30 after drought treatment−the average stem thickness before the treatment)/30. A portable photosynthetic system LI-6400XT (Li-COR, Lincoln, NE, USA) was used to estimate PN, Tr, Gs and Ci for the second terminal leaflet of fully grown second leaf at an irradiance of 1500 μmol m⁻² s⁻¹ and CO₂ concentration of 400 μmol mol⁻¹. The fifth or sixth leaves below the growing point were selected for the determination of soluble sugar, starch, nonstructural carbohydrates (NSC), total chlorophyll content, proline content, relative water content (RWC), malondialdehyde (MDA), electrolyte leakage, superoxide dismutase (SOD), catalase (CAT) and peroxidase (POD). The soluble sugar, starch and NSC were measured using the method described by Yu (Yu et al., 2011). The electrolyte leakage and RWC content was determined with the method of Yu (Yu et al., 2006). The total chlorophyll content, proline content, MDA, SOD activities, POD activities and CAT activities were determined according to the method described by Li (2000). The survival rate was calculated at Day 65 after drought treatments and expressed as the ratio of number of plants survived drought treatments to total number of plants before the drought treatments.

All plants were hand-harvested at the full maturity, about 90 days after seedling emergence. For each plant, shoot, roots, and tubers were weighed separately for fresh wrights, and then the samples were oven-dried at 70°C to a constant weight and weighed for the dry weights of shoot, roots, and tubers. The entire experiment was run three times, using three planting dates April through August in 2013.

In addition to the plant biomass and tuber yield, we determined “root activity” of the plants under the different levels of drought stress. Root activity is considered one of the important traits that reflect root respiration, oxidizing activity, or reducing activity of plants in response to stress. Measurement of root activity gives an indication of how plant roots perform under stress (Muler et al., 2014; Whalley et al., 2017; Wu et al., 2017). For instance, where the nutrient absorption by roots is inhibited due to drought stress, the oxidizing activity of the roots is low (Muler et al., 2014). A number of methods can be used to measure root activity. In the present study, we used the triphenyl tetrazolium chloride (TTC) method (Zhang et al., 2015a). In brief, plant roots were washed with deionized water and excised at 5 cm in length from the root tips on the 30th day after stress treatment. The reaction mixture consisted of 5 g samples of root tips, 5 ml of 0.4% TTC, and 5 ml of phosphate buffer solution (pH 7.0) in a beaker, with root tips fully immersed in the solution for 1 h at 37°C in the dark. When root tips turned red in color, 2 mL of 1 M sulfuric acid was added to stop the reaction. Root tips were dried with filter paper and homogenized with liquid nitrogen in an ice cold mortar and pestle. The red extraction was moved into a centrifuge tube to make the total volume of 10 ml using ethyl acetate. The extraction was vortexed for 30 s, and centrifuged at 1000 rpm for 5 min. The extraction was measured at 485 nm against the blank of ethyl acetate. The dehydrogenase activity is regarded as an indicator of the root activity. This analysis was repeated for three times.

Genomic DNA was isolated from 0.1 g of leaves from the putative transgenic and the NT control plants. The Npt II gene was demonstrated using standard PCR techniques. The genomic DNA isolation methods, the Npt II gene' primers and the PCR reaction programs were the same as those described in the above sections. The PCR products were electrophoresed on a 0.8% agarose gel.

To quantify the degree of drought tolerance, we employed an integrated evaluation index where a cluster analysis was combined with standard deviation coefficient allocation weighted method. Detailed methodology was described previously (Wang et al., 2013). In brief, an evaluation index was developed using the following three steps:

Data standardization, where a subordinator function was formed to standardize the measured raw data of each parameter as follows: (1)μ(Xij)=(Xij-Xmin)(Xmax-Xmin) (2)μ(Xij)=(Xmax-XijXmax-Xmin), where, Xij is the value of the jth variable (daily gain of plant height, leaf area expansion rate, stem thickness expansion rate, plant tissue water contents, tuber yield per plant, and rate of plant survival to drought treatment.) of the ith genotype, Xmin and Xmax represent the minimal or maximal value among the Xij values for the jth variable, respectively. The subordinator function with a positive correlation between a particular growth variable and drought tolerance was expressed by Equation (1), while a negative correlation was expressed by Equation (2).

Preliminary analysis showed that the treatment effects followed a similar trend in the first, the second, and the third runs of the experiment, and there were no significant treatment by run interactions for most of the variables measured. Therefore, the data from the three runs were pooled together in the analysis. All data were subjected to Tukey's HSD test using the SPSS package (SPSS Software, 19.0, SPSS Institute Inc., USA). Cluster analysis and the quantitative indices were used to categorize the seven genetic lines into the different levels of drought tolerance: very high, high, low, and very low. Significances between treatments in the ANOVA or among the categories of drought tolerance were considered at P < 0.05.

The plasmid pStPIP1 containing the StPIP1 gene was constructed and both StPIP1 and Npt II genes were verified with PCR amplification (Figure S1). Green shoots were produced directly from the transformed micro-tuber slices after 4 to 5 weeks of culture in the selective medium (Figure S2). Roots were formed about 5 days after green shoots were transferred to the selective rooting medium (Figure S3). These growth traits did not appear in the non-transgenic control lines. Potato plantlets with well-developed roots were propagated for further molecular analyses as follows.

PCR analysis using Npt II gene-specific primers showed that the putatively transformed plants had an amplification product of 676 bp and this product was missing in the control plant (Figure S4). The PCR amplification results were further confirmed by PCR-Southern blot analysis, which showed that the amplification product of 676 bp was the Npt II gene where 9 transformed plants had the Npt II gene and the control plants did not (Figure S4B). The Southern blot analysis revealed that Npt II gene was integrated into the potato genome, with the 7 transgenic plant lines having the hybridization signal, of which the line T4 having two copies of Npt II gene (Figure S5). The expression profiles for Npt II mRNA in the leaves of the transgenic plants showed that the Npt II gene was expressed and no transcript was found in the control plants (Figure S6). The qRT-PCR analysis showed that StPIP1 was expressed in all plant tissues with higher expressions in the stems and roots than in the leaves (Figure 1A). In each tissue, the relative expression of the StPIP1 gene was significantly higher in transgenic plants than in the control plants. Also, there were significant differences among the transgenic lines; T6 and T7 lines had the greatest StPIP1 expression, which were 2–15 times that of the four other transgenic lines, and 5–15 times that of the control plant (Figure 1A). In both transgenic and control plantlets, the expression of the StPIP1 gene was enhanced with the time of osmotic treatment until 96 h (Figure 1B). At a given time, the relative expression of StPIP1 gene was highest in the T7 lines.

The PCR amplifications of Npt II proved that six (T1, T2, T4, T5, T6, and T7) of the seven transgenic lines carried the complete coding sequence of Npt II gene in 676 bp, and no specific fragment was detected in the control or in the T3 line (Figure S7). These results confirmed the genomic integration and stabilized inheritance of the Npt II gene carried by the majority of transgenic potato lines. Thus, the six transgenic lines (excluding T3) were carried in the evaluation of drought tolerance for over-expression of StPIP1 gene.

The following three steps were taken for the assessment of plant tolerance to drought. First, the six key plant growth traits were individually included in the evaluation formula: (i) daily gain of plant height, (ii) leaf area expansion rate, (iii) stem thickness expansion rate, (iv) plant tissue water contents, (v) tuber yield per plant, and (vi) rate of plant survival to drought treatment. Evaluated using the six individual growth traits, the control plants had a significantly (P < 0.05) lower drought tolerance than all the transgenic lines, reflected by lower coefficients (Table 1). There were significant differences in the degree of drought tolerance among the transgenic lines. The line T7 had a significantly higher coefficient of drought tolerance with higher water content (95%) than the T6 line, while T7 had a lower coefficient to tuber yield than T6. Both lines had an equivalent value on leaf area expansion and survival rate. Thus, using the individual growth traits to rank the relative drought tolerance among plants was complicated because each growth trait had a different degree of influence on the magnitude of drought tolerance.

Then, the six growth traits were integrated together to form an evaluation index using Equations (1–5) and to provide quantitative values in assessing the magnitude of the drought tolerance. The integrated evaluation indices (i.e., D-values, Table 2) were calculated by determining the relative weights of each of the six growth traits in influencing the drought tolerance. Subordinator functions were used to generate correlation coefficients ranging from 0.01 to 1.00, and a higher coefficient represents a greater contribution of the growth trait to the drought tolerance. These indices were used to rank the drought tolerance of the lines quantitatively. All the six transgenic lines had significantly higher D-values than the control (Table 2). The lines T6 and T7 had the D-values 8–9 times that of the control and 1.6–2.0 times that of the four other transgenic lines. These quantitative evaluations suggest that T6 and T7 have the greatest drought tolerance, followed by T5, T1, T2, and T4, and the control had the lowest drought tolerance.

Finally, a cluster analysis (cluster diagram not shown) was used to categorize the drought tolerance of the genetic materials into four categories: drought tolerance being very high (T6 and T7), high (T4 and T5), low (T1 and T2), and very low (the control). The persistence of the drought tolerance in the four groups was further verified in a controlled-environment experiment where the different levels of drought stresses (no stress, moderate stress, or severe stress) were imposed. At Day 45 in the stress treatments, the transgenic line T7 plants grew vigorously with more roots, while the control plants started wilting with fewer roots (Figure S8). At harvest, all plants produced tubers and no significant differences were shown under non-stress treatments (with normal irrigation). With the stress level increased from normal to moderate, the control plants decreased tuber yield by 37.8%, while the lines T7 and T6 plants decreased tuber yield by 5.5% (Table 3); With the drought stress further increased to severe level, the control plant decreased tuber yield by 97.4% while the lines T7 and T6 decreased by 39.7%.

With the stress level increased from normal to moderate, the control plants decreased root fresh weight by 67.6% and root dry weight by 40.0%; and with the drought stress furthered to the severe level, the control plants decreased root fresh weight by 69.8% and root dry weight by 80.0% (Table 4); by comparison, the values of the decreases for the lines T7 and T6 plants were 27.8 and 13.8%, respectively. The latter lines had significantly smaller decreases in root weights than the control plant. Similar trends of treatment effect on shoot fresh and dry weights were observed as the effects on the root traits: with the drought stress increased from normal to severe level, the control plant decreased shoot fresh weight by 82.5% and shoot dry weight by 62.9%, and the percent decreases were significantly smaller in the lines T7 and T6 plants. There were no differences in either fresh or dry weight of shoot or those of roots among the genetic lines under non-stress conditions.

Drought stress affected a number of physiological traits in potato plants. Severe stress significantly decreased photosynthesis efficiency (Pn), stomatal conductance (Gs), and transpiration rate (Tr) in both transgenic and control plants. However, the magnitude of the decrease was significantly lower in the transgenic lines than in the control (Figure 2). Measured Day 30 in stress treatment, the transgenic lines in the “Very high” category decreased Pn slightly under the severe drought stress, while the control plant decreased Pn by 33% under moderate stress and 56% under severe stress (Figure 2A); a similar trend of response to drought stress was found in Gs and Tr, where the control plants decreased Gs by 39% under moderate stress and 65% under severe stress (Figure 2B); with the values of Tr decreases being 49 and 69%, respectively (Figure 2C). These defending responses of the transgenic lines to drought stress enhanced water use efficiency (WUE), as evidenced that under severe stress the transgenic lines increased WUE more prominently than the check (Figure 2D).

The soluble sugar, starch, total nonstructural carbohydrates (NSC) concentrations, superoxide dismutase (SOD), peroxidase (POD), catalase (CAT) in the leaves of young potato plants with overexpression of StPIP1 gene were measured at each of the three stress levels. Significant (P < 0.05) differences in soluble sugar concentration were detected among the different lines and among treatments (Figure 3A). The transgenic lines increased soluble sugar concentration remarkably compared to the control under moderate and severe drought stress treatments. The plants in the “Very high” and “High” categories had the similar responses to the stress. In contrast, the control plants had the lowest sugar and starch concentrations (Figure 3B). Also, the responses of total NSC to drought stress varied significantly among the lines (Figure 3C). Overall, the control plants had the lowest total NSC concentration under drought stress. With severe stress, total NSC concentration in the transgenic lines was 1.5 times higher than that in the control under (Figure 3C).

Furthermore, the biochemical analysis showed that the drought stress treatments affected the activity of many anti-oxidative enzymes, such as SOD, POD, and CAT. Moderate and severe drought stresses stimulated enzymatic activities in general, but the magnitude of the responses differed between the transgenic and control plants (Figure 4). In particular, under severe stress, the transgenic lines increased the enzymatic activities significantly (P < 0.05) whereas the control plants had a low enzymatic activity.

Total chlorophyll content decreased significantly (P < 0.05) with the increase of drought stress levels for all the genotypes evaluated in the study (Figure 5A). The magnitude of the decrease in chlorophyll content varied largely, with the transgenic plants in the “Very high” and “High” categories decreasing significantly less than the control. By day 30 of stress treatments, the control plants lowered chlorophyll content from 2.5 mg g⁻¹ of FW at no-stress to 1.72 mg g⁻¹ at moderate stress, and furthered to 1.56 mg g⁻¹ at severe stress, representing a decrease of 32 and 38%, respectively. On average, total chlorophyll contents of the transgenic lines in the “Very high” and “High” groups were 25% higher than in the control. The reverse trend of effect was found for proline content: with drought stress the control plants increased proline content significantly more than the transgenic lines (Figure 5B). By Day 30 of severe stress treatment, proline content in the control plant reached a level of 29.1 μg⁻¹ g⁻¹ of FW, which was about 2 times the average of the transgenic lines. The MDA content (Figure 5C) followed a similar trend of effect as the proline content. The magnitude of the increases in MDA varied among the genotypes. Treated for 30 days under drought stress, the control plants increased MDA content from 1.23 mg g⁻¹ of FW at no-stress to 6.50 at moderate stress and furthered to 11.09 at severe stress, representing an increase of 5.3 and 9.0 times, respectively. The transgenic lines increased a least amount of MDA. In contrast to the variables described above, the response of electrolyte leakage to drought stress varied substantially among genotypes (Figure 5D). A similar level of electrolyte leakage was found in the transgenic lines among the three levels of drought stress, but electrolyte leakage in the control plant increased significantly both in the moderate and severe stresses (Figure 5D).

The root activity assay showed that the continuous drought challenges decreased root activity for all genotypes evaluated, with the control plant decreasing the root activity more substantially (P < 0.05) than the transgenic lines in the moderate and severe drought stress. At the high stress levels, the roots activity of the control plants was about half of the transgenic lines (Figure 5E). The over-expression of StPIP1 gene effectively improved the root activity of potato plants.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.