To generate AtPLDε-OE soybean plants, the full-length coding region of Arabidopsis PLDε (At1g55180) was fused with a HA tag at the 3′ end, and then cloned into a binary vector pZY101 under the control of the CaMV 35S promoter. The resulting vectors were transformed into the Agrobacterium strain EHA101 by a freeze–thaw method. Soybean (cv. Williams 82) was subjected to Agrobacterium-mediated cotyledonary node transformation (Zeng et al., 2004). Transgenic soybean lines were screened by the resistance to the herbicide Basta and verified by PCR using Arabidopsis PLDε gene-specific primers (Supplementary Table 1). Soybeans were grown in a growth chamber at 16 h light 25°C/8 h dark 22°C.

For N treatments, soybean seeds were germinated on vermiculite watered with Hoagland solution containing 6 mM N (KNO₃) for 7 days in a growth chamber at 16 h light 25°C/8 h dark 22°C, and then transferred to Hoagland solution supplied with 6 mM N. After 7 days, the soybean plants were transferred to Hoagland solution supplied with 6 or 0 mM N for 7 days. For nodulation, soybean seeds were germinated on vermiculite watered with Hoagland solution containing 0.6 mM N (KNO₃) in a growth chamber at 16 h light 25°C/8 h dark 22°C. After 21 days, the soybean plants were infected with Bradyrhizobium diazoefficiens USDA110 (OD₆₀₀ = 0.2) grown in Yeast Mannitol (YM) medium. After infection, plants were maintained under the same condition, and watered with Hoagland solution containing 0.6 mM N until harvest stage.

Total RNA was isolated from soybean leaves using the RNeasy Plant Mini Kit (Qiagen), and the cDNAs were synthesized using the qScript cDNA Synthesis Kit (Quanta BioSciences) according to the manufacturer’s manual. Semi-quantitative PCR and real-time qPCR were performed to analysis gene expression. Actin (Glyma18g52780) was used as internal control to normalize the gene expression. The gene-specific primers are listed in Supplementary Table 1.

Total proteins were extracted from the leaves of soybean plants using an extraction buffer (50 mM Tris–HCl, pH 8.0, 150 mM NaCl, 10 mM EDTA, 1 mM PMSF, and 10 mM iodoacetamide). After centrifugation at 2,000 g for 5 min at 4°C, the supernatant was subjected to 8% SDS-PAGE and transferred onto a polyvinylidene difluoride membrane. PLDε-HA on the membrane was immunodetected using an anti-HA antibody (Invitrogen 26183; 1:4,000 dilution) as the primary antibody and the HRP-conjugated anti-rabbit antibody (Sigma) as the secondary antibody.

Soybean leaves were ground in liquid nitrogen and added to an extraction buffer (50 mM Tris–HCl, pH 8.0, 150 mM NaCl, 10 mM EDTA, 10 mM EGTA, 1 mM PMSF, 10 mM iodoacetamide, 0.5% Triton X-100). The extract was centrifuged at 5,000 g for 10 min at 4°C and the supernatant was used for enzyme assays. The total protein concentration was measured using Braford assay (Bio-Rad). All the enzyme activities were measured spectrophotometrically. The nitrate reductase activity was measured according to Hageman and Hucklesby (1971) with modifications. Briefly, 20 μl of extracts was added to 380 μl of a reaction buffer (50 mM KH₂PO₄, pH 7.5, 20 mM KNO₃, 0.35 mM NADH) and incubated at 30°C for 30 min. The reaction was stopped by adding 300 μl of 1% (w/v) sulfanilamide in 3 M HCl and 300 μl of 0.02% (w/v) N-1-naphthylethylenediamine dihydrochloride. The absorbance was measured at 540 nm, and NaNO₂ was used to prepare a standard curve. The nitrite reductase activity was measured according to Joy and Hageman (1966). Briefly, 25 μl of extracts were added to a 500 μl reaction buffer (50 mM KH₂PO₄, pH 7.5, 1 mM NaNO₂, 1 mM methyl viologen), and the reaction was started by adding 150 μl of 57.4 mM Na₂S₂O₄ in 290 mM NaHCO₃. After incubated at 30°C for 60 min, 40 μl of the reaction mixture was added to 360 μl H₂O in a new tube, followed by adding 300 μl of 1% (w/v) sulfanilamide in 3 M HCl and 300 μl of 0.02% (w/v) N-1-naphthylethylenediamine dihydrochloride. A₅₄₀ of the reaction mixture was measured, and NaNO₂ was used to generate standard curve. The glutamine synthetase activity was measured according to Liu et al. (2010). Briefly, 50 μl of extracts was added to 100 μl of a reaction buffer (70 mM MOPS, pH 6.8, 100 mM L-glutamic acid, 50 mM MgSO₄, 15 mM NH₂OH, 15 mM ATP) at 37°C for 30 min. The reaction was stopped by addition of 450 μl of 88 mM FeCl₃, 670 mM HCl, and 200 mM trichloroacetic acid. The absorbance was measured at 498 nm, and blank controls were obtained by incubating protein extracts in the reaction buffer without addition of ATP. GHA (γ-glutamyl hydroxymate) was used as standard.

Total lipids were extracted and analyzed by electrospray ionization–tandem mass spectrometry (ESI–MS/MS) according to Welti et al. (2002). Briefly, the samples were immediately immersed into 3 ml of preheated isopropanol containing 0.01% butylated hydroxytoluene (BHT) at 75°C for 15 min and then added with 1.5 ml of chloroform and 0.6 ml of water. After incubated in a shaking incubator at room temperature for 1 h, the extracts were transferred to glass tubes. The samples were then extracted with chloroform/methanol (2:1) containing 0.01% BHT four times. The combined extracts were washed once with 1 M KCl and then water. The lower phase was dried under nitrogen gas and resuspended with chloroform (250 μl chloroform/mg dry weight). Lipids were introduced by direct infusion to mass spectrometer (API4000, Sciex), and scanned in the positive ion mode as described previously (Welti et al., 2002). The seed oil contents were analyzed using gas chromatography (GC) according to Guo et al. (2014).

For artificial aging, soybean seeds were placed in a sealed container with a beaker filled with water and then stored in an incubator at 45°C for 5 days. The fresh harvested, stored under high humidity condition, and artificially aged seeds were surface sterilized with chloride gas and then washed with sterilized water five times. After placed on sterilized wet filter paper in petri dish, seeds were germinated at 25°C.

Approximately 20 mg of leaves were incubated in 2 ml of 95% ethanol and then placed in a rotating incubator overnight or until the leaf is bleached at 4°C under the dark. The absorbance was measured at 664 nm and 649 nm. The contents of chlorophyll a and b were calculated as C_a + b = 5.24A₆₆₄ + 22.24A₆₄₉ and normalized by fresh weight.

The soybean genome contains two PLDεs, GmPLDε1 (Glyma07g01310) and GmPLDε2 (Glyma15g02710), which are homologous to Arabidopsis PLDε (Zhao et al., 2012). The amino acid sequence identity between AtPLDε and GmPLDε1/GmPLDε2 is 62% and 62.2%, respectively (Supplementary Figure 1A). To examine how soybean PLDε responds to N deficiency, we measured the transcript level of two PLDεs in soybean under N-sufficient and -deficient conditions. When 2-week-old soybean seedlings were transferred to nitrogen-rich or -free medium for 7 days, the transcript level of GmPLDε1 and GmPLDε2 in soybean leaves was increased under the N-deficient condition by 132% and 254%, respectively (Figure 1A).

To study the effect of PLDε on soybean growth, we overexpressed PLDε from Arabidopsis in soybean (cultivar William 82) under the CaMV-35S promoter (Figure 1B), based on the previous reports that Arabidopsis PLDε enhanced plant growth in Arabidopsis and canola under N deprivation (Hong et al., 2009; Lu et al., 2016). The expression of AtPLDε in three independent transgenic lines was verified by semi-quantitative PCR and the production of hemagglutinin (HA) tagged-PLDε was confirmed by immunoblotting with an anti-HA antibody (Figures 1C,D). AtPLDε-OE1 showed the highest expression level of PLDε among three lines. T3 generation of homozygous transgenic plants was used for the future analysis. In contrast with WT soybean, the transcript level of GmPLDε1 and GmPLDε2 in AtPLDε-OE1was not induced under N-deficient condition (Supplementary Figure 1B).

To test the effect of PLDε-OE on soybean response to N deprivation, 2-week-old soybean seedlings grown in Hoagland solution containing 6 mM N were transferred to Hoagland solution supplied with 6 or 0 mM N for 7 days. Compared with wild-type (WT), AtPLDε-OE soybean plants displayed larger leaf area, and higher chlorophyll content under both N-sufficient and -deficient conditions (Figures 2A,B). However, the difference between AtPLDε-OE and WT plants in leaf area and chlorophyll content was greater under the N-deficient than N-sufficient condition. The chlorophyll content of three AtPLDε-OE lines was, on average, 41% higher than that of WT under N deficiency whereas that difference was 12% under sufficient N. The leaf area of three AtPLDε-OE lines was, on average, 16% higher than that of WT under N deficiency whereas that difference was 10% under sufficient N. Under N deficiency, leaf growth was inhibited, and chlorophyll content was decreased, but the primary root elongation was promoted. The primary root length of AtPLDε-OE1 and -OE2 with higher expression of AtPLDε was, on average, 14% longer under N-sufficient but 10% longer than that of WT under N-deficient condition (Figure 2B). These results indicate that AtPLDε-OE plants display less N-deficiency-induced growth alterations than WT, and PLDε-OE enhances leaf growth and chlorophyl content more under N-deficient than N-sufficient condition (Supplementary Figure 2).

Nitrate is the major form of N in soil taken up by plants, where it is reduced sequentially by nitrate reductase (NR) and nitrite reductase (NiR) to ammonium that is incorporated into amino acids by glutamine synthetase (GS; Masclaux-Daubresse et al., 2010). The activities of NiR, and GS in AtPLDε-OE leaves were higher than those of WT under sufficient N, and the effect was greater under N deficiency (Figure 2C). Nitrate deficiency led to a decrease in NR, NiR and GS in both WT and AtPLDε-OE leaves. However, compared to those grown under sufficient N, AtPLDε-OE soybean still maintained relatively higher activities of NR, NiR and GS when grown under N deficiency (Supplementary Figure 3). Three AtPLDε-OE lines had an average of 13% higher relative NR activity and 18% higher relative NiR activity than those of WT plant. The GS activity in AtPLDε-OE1 and -OE3 leaves was 7% higher than that of WT (Supplementary Figure 3). These results suggest that PLDε-OE enhances N assimilation.

To test the effect of PLDε-OE on soybean growth under N deficiency with and without nodulation, 3-week-old soybean seedlings grown on vermiculite watered with Hoagland solution containing 0.6 mM N were infected with B. diazoefficiens USDA110. The inoculation of soybean plants with rhizobia enhanced plant growth under N deprivation, but the enhancement was greater in AtPLDε-OE than WT plants (Figures 3A,B). Four weeks after infection, the leaf chlorophyll content in three AtPLDε-OE lines with inoculation was, on average, 33% higher than that of WT, whereas that in three AtPLDε-OE lines at the same growth stage without inoculation was 19% higher than that of WT (Figure 3B). However, there was no significant difference among the total weight of nodules collected from WT and AtPLDε-overexpressing soybean roots 28 days after inoculation (Figure 3C).

Compared with WT, the seed yield in PLDε-OE1 and -OE2 without inoculation was increased by an average of 12%, and the seed yield in PLDε-OE1 and -OE2 with inoculation was increased by an average of 18% (Figure 3D). The seed number was significantly increased in PLDε-OE1 and -OE2 plants with inoculation, which had higher expression of AtPLDε, and the average seed weight of WT and AtPLDε-OE soybean lines remained similar (Figure 3D).

PLDε hydrolyzes phospholipids to produce PA. To analyze the effect of PLDε-OE on lipids composition, we used an ESI-MS/MS approach to quantitatively profile lipid species in leaves from Hoagland solution-grown plants with and without N. After 7 days of N starvation, the amounts of galactolipids monogalactosyldiacylglycerol (MGDG) and digalactosyldiacylglycerol (DGDG), which are the major lipids of chloroplast membranes, and PC were greatly decreased in WT and AtPLDε-OE soybean plants (Figure 4A). The total PA level tended to be higher in PLDε-OE lines, but only PLDε-OE1 that had the highest expression of AtPLDε displayed a significantly higher level of total PA than WT under sufficient N (Figure 4A). However, when PA molecular species were compared, AtPLDε-OE1 leaves exhibited increases over WT in some PA species, including 36:6 PA under N-sufficient and -deficient conditions, and 34:3 PA and 36:5 PA under N deficiency (Figure 4B). PLDε-OE3 leaves showed an increase in 36:5 PA, but PLDε-OE2 leaves had no significant PA increase (Figure 4B). On the other hand, most PE and PC species, including 34:2, 34:1, 36:4, 36:2 and 36:1 were decreased in AtPLDε-OE plants under N deficiency (Figures 4C,D).

We then analyzed lipid species in roots and nodules of AtPLDε-OE and WT plants grown in vermiculites with and without nodulation, Total lipids were extracted from nodules and roots with and without inoculation 28 days after rhizobium inoculation and analyzed by ESI-MS/MS. There were no significant differences on lipid contents (mole percentage) of MGDG, DGDG, phosphatidylglycerol (PG), PC, and PE in roots or nodules between WT and PLDε-OE soybean (Figure 5A; Supplementary Figure 4). Compared to roots, PC contents were higher in nodules and PE contents were lower in nodules. Without inoculation, 36:5 PA was higher in roots of AtPLDε-OE1 and OE3 lines, but 36:4 PA in OE3 was lower than WT roots, whereas 34:3 PA and 36:6 PA in roots was higher in all three AtPLDε-OE lines than WT after rhizobium nodulation (Figure 5B).

Soybean is a major protein and oil crop, so we assessed the effect of PLDε-OE on seed and oil production of soybean plants grown in soil. AtPLDε-OE and WT soybean plants were grown side by side to maturity in greenhouse. The seed number per plant was higher in all three AtPLDε-OE lines than WT. The seed number per plant in three AtPLDε-OE lines was, on average, 13% higher than that of WT (Figure 6B). There were no significant differences among the total seed yield and seed weight for AtPLDε-OE and WT (Figures 6A,C). The plants in soil grew much better than those grown under the vermiculite condition, and the seed yield per plant in soil was about 3-fold and 9-fold higher than that of the vermiculite-grown plants with and without nodulation, respectively (Figures 3D, 6A).

The total seed oil content of AtPLDε-OE soybean lines was comparable with that of WT soybean (Supplementary Figure 5), whereas the fatty acid composition exhibited a change in levels of oleic acid (C18:1) and linoleic acid (C18:2) in AtPLDε-OE soybean seeds (Figure 6D). Two of the PLDε-OE lines which had higher expression of AtPLDε displayed a significant decrease in the oleic acid content with an increased linolenic acid content; significant changes in the two fatty acids from WT seeds (Figure 6D). In particular, AtPLDε-OE1 with the highest expression of AtPLDε had 41% decrease in oleic acid and 9% increase in linoleic acid than WT (Figure 6D).

We further tested the effect of PLDε-OE on seed viability because earlier studies indicated that alterations of PLDα expression affected seed viability in Arabidopsis (Devaiah et al., 2007) and soybean during storage (Lee et al., 2012; Zhang et al., 2019a). The germination rate of freshly harvested AtPLDε-OE and WT soybean seeds was similar (Figure 7A). After stored under a low temperature (4°C) and high humidity condition for 12 months, the germination rate of soybean seed was dramatically decreased in AtPLDε-OE seeds (Figure 7A). Compared with WT seeds, the germination rate of AtPLDε-OE1, OE2, and OE3 seeds was decreased by 53%, 41%, and 21%, respectively. High temperature and high humidity can decrease the seed viability by accelerating deterioration process (Zhang et al., 2019b). To further verify the role of PLDε in seed storability, we exposed fresh seeds to a high temperature (45°C) and high humidity condition (100%) for 5 days to accelerate seed aging. AtPLDε-OE displayed greater sensitivity to this artificial aging than WT seeds. The germination rate of AtPLDε-OE1, OE2, and OE3 seeds was on average 50% lower than WT seeds (Figure 7B). These results suggest that PLDε-OE reduces seed viability after storage and artificial aging.