The experiment was carried out at Northwest A&F University, Yangling (34° 15′ N, 108° 4′ E) Shaanxi, China. Seeds of M. hupehensis were collected in Pingyi (35° 70′ N, 117° 25′ E), Shandong Province. The seeds were disinfected and stored in sand for 2 months at 4°C to break dormancy. The germinated seeds were seeded in a nursery (12 cm × 12 cm). Plants of similar size were used for hydroponics and grown in 8–10 nurseries. Hoagland nutrient solution was the source of nutrients during hydroponics. The two N concentrations applied were 5 mmol L^–1 and 0.2 mmol L^–1. N was provided exclusively in the form of NO₃^–. Arginine was added 3 days before N-deficiency treatment to facilitate the adaptation of plants to the experimental environment. The temperature was maintained at 23–25°C in the daytime and 15–18°C at night. The sodium lamp provided light during the 14/10 h light/dark photoperiod (photon flux density was 160 mmol m^–2 s^–1). Oxygen was added to the nutrient solution to maintain the oxygen concentration at 8.0–8.5 mg L^–1 through the dissolved oxygen controller (FC-680; Corporation of Super, Shanghai, China). The nutrient solution was renewed every 5 days.

The plants were pre-cultivated for 15 days to permit adaptation to hydroponics conditions. To determine the optimal concentration of exogenous arginine, the plants were divided into seven treatments at 20 days: half-strength nutrient solution (5 mmol L^–1) and N-deficiency treatment with 20, 100, 500, 2500, and 10000 μmol L^–1 arginine. Arginine was supplemented throughout the entire experiment. The net photosynthetic rate (Pn) was measured at Day 0, 5, 10, 15, 20, and 25 after treatment; the dry weight, fresh weight, and plant height of all groups were measured at 25 days after treatment; and the phenotypic characteristics of the plants were recorded.

According to the optimal arginine concentration, the plants were divided into four treatments: control (1/2 Hoagland solution containing 5 mmol L^–1 N, CK), control + arginine (1/2 Hoagland solution containing 5 mmol L^–1 N with 100 μmol L^–1 arginine, CKA), N-deficiency solution (1/2 Hoagland solution containing 0.2 mmol L^–1 N, LN), and LN + arginine (1/2 Hoagland solution containing 0.2 mmol L^–1 N with 100 μmol L^–1 arginine, LNA). The physiological indexes of each treatment were measured.

The shoot height (SH) was measured from the top of the stem to the stem base; the leaf number was counted per plant at 20 days after treatment. On the first and last day of the treatments, the plants were divided into leaves, stems, and roots for weighing. The samples were then washed with tap water, distilled water, and double water, followed by fixation at 105°C for 30 min and drying at 72°C until a constant weight was achieved (at least 72 h). The fresh weight and dry weight were calculated from the weight of the leaves, stems, and roots before and after drying, respectively.

The malondialdehyde (MDA) content was measured per the instructions provided in an MDA reagent kit (Suzhou Comin Biotechnology Co., Ltd., China). Root activity was measured using triphenyl tetrazolium chloride (TTC) following the method of Joseph et al. (2010).

The tissues in the four treatments were collected at 20 days after treatment to estimate the N, P, and K concentrations. After quick-freezing, tissue (0.1 g) was ground and digested with concentrated sulfuric acid (H₂SO₄, AR, 98%) and hydrogen peroxide (H₂O₂, GR, ≥30%). The N and P content was measured by an AutoAnalyzer 3 continuous-flow analyzer (AA3; SEAL Analytical, Norderstedt, Germany), and the K content was measured by a flame photometer (M410; Sherwood Scientific, Cambridge, United Kingdom) according to the method of Li et al. (2013).

The formulas for nutrient uptake fluxes (TN) are as follows (Kruse et al., 2007):

where, RGR is the relative growth rate, DW is the dry weight, and C is the concentration of elements in each tissue. The uptake flux was expressed in units of milligrams per plant per day or micrograms per plant per day.

The transportation of N, P, and K was defined as the total amount of element transported to leaf and stem tissue per gram root DW per day. The accumulation of N, P, and K in the roots was defined as the total amount of nutrient absorbed into root per gram root DW per day. The formula is as follows (Kruse et al., 2002):

where, M represents the total content of elements, W represents the dry weight of roots, and T is the treatment time.

The element content of each part (root, stem, and leaf) is the product of the dry weight of each part and the element concentration.

The tissues of four treatments were collected at 20 days after treatment. The samples were ground into powder with liquid N and stored at –80°C. Samples were weighed (0.1 g) for each enzyme activity measurement. The activity of nitrate reductase (NR), nitrite reductase (NiR), glutamine synthetase (GS), and glutamate synthase (GOGAT) was measured per the instructions provided by corresponding kits (Suzhou Comin Biotechnology Co., Ltd., China). Absorbance of NR and GOGAT was recorded at 340 nm. Absorbance of NIR and GS was recorded at 540 nm. The standard was based on the instructions provided by the kits. NR activity was expressed as catalytic reduction of 1 nmol NADH per min per g of fresh weight. NiR activity was expressed as the reduction of 1 μmol NO₂^– per g tissue per hour. GS activity was expressed as 1 μmol of γ-glutamyl hydroxamic acid produced per gram of tissue per ml of reaction system per hour. GOGAT activity was expressed as 1 nmol of NADH consumed per gram of tissue per minute.

Chl was extracted with 80% acetone at 0 and 20 days. The total chl, chl a, chl b, and carotenoid (car) content was measured spectrophotometrically following the procedure of Arnon (1949) with minor modifications.

The Pn, stomatal conductance (gs), intercellular carbon dioxide concentration (Ci), and transpiration rate (Tr) were measured with a portable photosynthesis system (CIRAS3) from 9:00 am to 10:00 am. All measurements were performed at 1000 μmol photons m^–2 s^–1 and a constant airflow rate of 500 μmol s^–1. The cuvette CO₂ concentration was set to 400 μmol CO₂ mol^–1 air, with a vapor pressure deficit of 2.0–3.4 kPa. Measurements were taken from the fully extended leaves located at the same position from five randomly selected plants in each group.

Chl fluorescence was measured by a Dual-PAM-100 Chlorophyll Fluorometer (Walz, Germany). After 30 min of dark adaptation, the minimum fluorescence (F₀) and maximum Chl fluorescence yield (F_m) were measured. F_v/F_m was calculated as (F_m−F₀)/F_m.

Total RNA was extracted from leaves using a Wolact^® Plant RNA Isolation Kit (Vicband, HongKong, China) per the manufacturer’s instructions. qRT-PCR was performed on a LightCycler 480 (Roche, Indianapolis, IN, United States) Real-Time System using SYBR Premix Ex Taq II (Takara, Kyoto, Japan) following a previously described method (Gao et al., 2020). Sequences of the primers for MdAMT1;2, MdAMT2;1, MdFd-GOGAT, and MdNADH-GOGAT were designed with Primer Premier 5 software (Biosoft International, Palo Alto, CA, United States) (Supplementary Table 1). MdMDH was used as an internal reference gene.

Biological samples were crushed using a mixer mill (MM 400, Retsch). Extraction and detection methods were adjusted following the method of Chen et al. (2013). Briefly, 100 mg of powder was dissolved with 1.2 mL of 70% methanol solution. The solution was vortexed for 30 s every 30 min a total of 6 times. The sample was extracted at 4°C overnight, followed by centrifugation at 12,000 rpm for 10 min. The supernatant was filtrated through a 0.22-μm organic strainer before UPLC-MS/MS (UPLC, SHIMADZU Nexera X2; MS, Applied Biosystems 4500 Q TRAP) analysis. The UPLC was equipped with an Agilent SB-C18 column (1.8 μm, 2.1 mm × 100 mm). The mobile phase consisted of solvent A, pure water with 0.1% formic acid, and solvent B, acetonitrile with 0.1% formic acid. The gradient program was as follows: 0–9 min, 95% A to 5% A and 5% B to 95% B with a linear gradient and hold for 1 min; 5% A to 95% A and 95% B to 5% B in 1.10 min and hold for 2.9 min. The flow velocity was 0.35 mL min^–1, the column temperature was 40°C, and the injection volume was 4 μL. The effluent was alternatively connected to an ESI-triple quadrupole-linear ion trap (QTRAP)-MS.

LIT and triple quadrupole (QQQ) scans were acquired on a triple quadrupole-linear ion trap mass spectrometer (Q TRAP), AB4500 Q TRAP UPLC/MS/MS System, equipped with an ESI Turbo Ion-Spray interface, operating in positive and negative ion mode and controlled by Analyst 1.6.3 software (AB Sciex). The ESI source operation parameters were as follows: ion source, turbo spray; source temperature, 550°C; ion spray voltage (IS), 5500 V (positive ion mode)/-4500 V (negative ion mode); ion source gas I (GSI), gas II (GSII), and curtain gas (CUR) were set at 50, 60, and 25.0 psi, respectively; and the collision-activated dissociation (CAD) was high. Instrument tuning and mass calibration were performed with 10 and 100 μmol L^–1 polypropylene glycol solutions in QQQ and LIT modes, respectively. QQQ scans were acquired as MRM experiments with collision gas (N) set to medium. DP and CE for individual MRM transitions were done with further DP and CE optimization. A specific set of MRM transitions were monitored for each period according to the metabolites eluted within this period.

The amino acids were extracted and detected by LC-MS following the method of Jin et al. (2019). Fresh tissue was ground in liquid N, 500 mg of sample powder was weighed in tubes, and 1 mL of 50% ethanol (containing 0.1 mol L^–1 HCl) was added to tubes to extract amino acids. The supernatant was transferred to a new tube after centrifugation at 13,000 g and 4°C for 10 min. The liquid was filtered through a 0.22-μm organic strainer. The filtrate was diluted 20 times with methanol (AR, 98%) to a 1-mL volume. The LC-MS analysis was carried out on an Inertsil OSD-4C18 column (150 mm × 3.0 mm; packing size, 3.5 μm) at 25°C; 0.5% formic acid water (solvent A) and methanol (solvent B) were used as the mobile phase at a flow rate of 0.3 mL min^–1. The content was calculated according to the standard curve.

Data on the content of metabolites detected were used for principal component analysis (PCA), hierarchical cluster analysis, and orthogonal partial least squares-discriminant analysis in R¹ to analyze the accumulation of metabolites in response to arginine and N deficiency (Wang et al., 2015). The Pearson correlation coefficients were calculated in R to verify the repeatability between classes. Significantly regulated metabolites were filtered using the following criteria: VIP ≥ 1 and Log2FC (fold change) ≥ 2 or Log2FC ≤ 0.5. To avoid overfitting, a permutation test (200 permutations) was performed. To clarify the role of arginine in response to N-deficiency stress, the identified metabolites were annotated using Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis. MSEA (metabolite sets enrichment analysis) was then conducted on the pathways with significantly regulated metabolites, and the significance of these pathways was determined by hypergeometric tests. Annotated metabolites were mapped to the KEGG Pathway database.

The data were statistically analyzed by one-way ANOVA. Tukey’s multiple range test was used to determine significant differences between means (P < 0.05) using SPSS 25.0 software (IBM Corp., Armonk, NY, United States).

The growth rate of apple was significantly inhibited under N-deficiency conditions. The growth of apple treated with 2500 and 10000 μmol L^–1 arginine was severely inhibited, as indicated by the lower plant length, wilted leaves, absence of root growth, and even death on the 3rd d with 10,000 μmol L^–1 arginine and 6th d with 2,500 μmol L^–1 arginine (Figure 1A) under N-deficient solutions. Apple treated with 100 and 500 μmol L^–1 arginine had better growth performance, indicating that this concentration of arginine alleviated the inhibition induced by N-deficiency stress (Figure 1).

Under N deficiency, the dry weight and Pn decreased sharply. However, the dry weight was 59.26% higher in apple plants treated with 100 μmol L^–1 arginine than under N deficiency (Figure 1D). Exogenous application of 100 or 500 μmol L^–1 arginine resulted in higher Pn values during treatments compared with the control (Figure 1E). Based on these results, 100 μmol L^–1 arginine was the optimal concentration used in subsequent experiments.

The growth of M. hupehensis plants was inhibited significantly under N deficiency, and growth was restored via exogenous arginine application (Figure 2A). The dry weight and fresh weight of M. hupehensis plants were higher under exogenous arginine application compared with LN (Figures 2B,C); the growth of plants was also enhanced by arginine application under a normal nutrient supply. The SH and number of leaves were lower under N deficiency compared with when a normal level of N was supplied (Figures 2D,E). There were no significant differences observed between CK and CKA.

Root vitality of M. hupehensis was significantly affected by N deficiency. Root vitality was higher for plants under LNA compared with plants under LN (Figure 2G).

The MDA content was higher in LN than in LNA, and no clear differences were observed among CK, CKA, and LNA (Figure 2F). This indicated that arginine alleviates oxidative damage to the cell membrane.

These results indicated that exogenous arginine enhanced aboveground growth and root activity under N deficiency and alleviated the damage to cells caused by N-deficiency.

Nutrient concentrations in different tissues were altered at 20 days after N-deficiency treatment. Arginine increased the total N concentration of the leaves, stems, and roots under stress (Supplementary Figures 1A–C). In the leaves and roots (Supplementary Figures 1D,F), the total K concentration was higher under LNA than under LN, but in the stems (Supplementary Figure 1E), the total K concentration was higher under LN compared with CK. The total P concentration in the stems and roots increased after the application of arginine under N stress (Supplementary Figures 1H,I). The leaf P content was lower under LNA than under LN (Supplementary Figure 1G).

The uptake, accumulation, and partitioning of N, P, and K were affected under N deficiency. The uptake of N, P, and K was decreased under N deficiency in M. hupehensis. The use of arginine significantly alleviated this inhibition (Figures 3A–C).

The content of N, P, and K was highest in leaves and lowest in stems. The application of arginine significantly increased the content of N, P, and K in the roots, stems, and leaves of M. hupehensis, especially the stems. Compared with LN, arginine also increased the N, P, and K content in tissues of M. hupehensis under low N stress (Figures 3D–F).

Under N deficiency, arginine increased the accumulation of N, P, and K in the roots, as well as the rate of transport of these nutrients in the stems compared with LN. The N and P transport rate to leaves was decreased, and the N and K transport rate was increased (Figures 3G–I). These results indicated that exogenous arginine treatment had a significant effect on nutrient transport, accumulation, and partitioning under N-deficiency stress.

Given the observed effects of arginine on the absorption and distribution of nutrients, we measured the activity of these four enzymes. The activity of these four enzymes was significantly inhibited under N-deficiency stress, and exogenous arginine alleviated this inhibition (Figure 4). The ammonium transporter genes MdAMT1;2 and MdAMT2;1 and the ammonium assimilation genes MdFd-GOGAT and MdNADH-GOGAT were significantly up-regulated at 12 h after low N stress with arginine application (Figure 5); the NO₃^– transporter genes MdNRT1;1 and MdNRT2;7 were up-regulated 3.3 and 2.3 times, respectively, at 12 h, and no significant response was observed after 12 h.

The chl a, chl b, car, and total chl content under N deficiency decreased significantly at 20 days after treatment; chl b was most affected by N deficiency (Figure 6B). The chl a, chl b, car, and total chl content was higher under LNA than under LN (Figure 6).

The Pn, gs, and Tr of plants decreased at 20 days after low N treatment (Figures 7A–C). However, Pn was higher under LNA than under LN. Under normal nutrient conditions, Pn was higher under CK than under CKA at 20 days (Figure 7A). Starting on the fifth day, the Tr of LNA was higher than that of LN, and gs was higher under LNA than under LN on the 10 days (Figures 7B,C). Lower Ci aided the ability of plants to cope with stress, and Ci was high under LN (Figure 7D), indicating that plants were under greater nutrient stress; Ci was low under CK, CKA, and LNA.

F_v/F_m measurements indicated that F_v/F_m was significantly lower under LN than under CK and LNA, but no significant differences between CKA and CK were observed (Figure 7E). The application of exogenous arginine alleviated damage to the photosynthetic system under low N stress but had no effect under normal N conditions.

We analyzed the differential metabolites of M. hupehensis plants. PCA was used to conduct an unsupervised clustering analysis of the detected metabolites. The replications of the different treatments were clustered, and there was clear separation among the four treatments, indicating that the results were reliable. LN and CK were significantly separated along the first principal component (PC1), accounting for 23.29% of the total variance. LN and LNA were significantly separated along the second principal component (PC2), accounting for 18.45% of the total variance (Figure 8B). Differences between CKA and LNA along PC1 and PC2 were small (Figure 8B).

A total of 59, 58, 22, and 37 differential metabolites were identified in the four comparison groups (CK vs. CKA, CK vs. LN, CKA vs. LNA, and LN vs. LNA, respectively) (Figure 8A). In the CK vs. LN and LN vs. LNA comparisons, 18 and 23 metabolites were up-regulated, and 40 and 14 metabolites were down-regulated, respectively (Supplementary Tables 2, 3). A total of 43 down-regulated and 16 up-regulated metabolites were identified between CK and CKA, and 10 down-regulated and 12 up-regulated metabolites were identified between CKA and LNA (Figure 8A). Venn diagram analysis showed that only one differential metabolite was present in all four comparison groups (Figure 8C).

The differential metabolites detected under N deficiency were annotated to several pathways through KEGG analysis. Most of the metabolites were annotated to the metabolism and biosynthesis of specialized metabolites, which accounted for 82.14% and 60.71% of total metabolites. Approximately, 1/3 of the metabolites were related to amino acid synthesis (Figure 9A). N deficiency also affected lysine synthesis and degradation, as well as histidine, arginine, serine, threonine, glycine, alanine, proline, and ornithine synthesis (Figure 9A). Some metabolites were involved in carbon and sulfur metabolism, the TCA (Tricarboxylic Acid Cycle) cycle, and carbon fixation in photosynthetic organisms, which are important for plant growth. Nicotinate and nicotinamide, oxidative, phosphorylation, anthocyanin, and 2-oxocarboxylic acid metabolism and synthesis were also related to metabolites under N deficiency.

Thirty-seven differential metabolites were detected in the LN and LNA comparison: a total of 23 up-regulated metabolites, including 14 amino acids and derivatives, 2 nucleotides and derivatives, 3 flavonoids, 2 alkaloids, 1 phenolic acid, and 1 organic acid, and a total of 14 down-regulated metabolites, including 2 amino acids, 3 flavonoids, 1 phenolic acid, 2 lipids, 3 terpenoids, and 3 alkaloids (Supplementary Tables 3, 4). Most were annotated to the metabolism and biosynthesis of specialized metabolites, which accounted for 70.83% and 50% of total metabolites, respectively. Metabolites related to indole alkaloid, linoleic acid, and caffeine were detected in LNA (Figure 9B).

Next, we compared the differential metabolites between CK vs. LN and LN vs. LNA. A total of 17 metabolites detected in both showed opposite expression patterns, including 12 amino acids (L-serine, 5-aminovaleric acid, pipecolic acid, L-asparagine, L-ornithine, L-citrulline, L-cysteinyl-L-glycine, L-arginine, N-monomethyl-L-arginine, homoarginine, glutathione-reduced form, and L-homomethionine), 3 flavonoids (delphinidin-3-O-arabinoside, malvidin-3-O-arabinoside, and kaempferol-3-O-rhamnosyl (1→2) glucoside), 1 organic acid (3-ureidopropionic acid), 1 phenolic acid (4-O-methylgallic acid), 1 alkaloid (diethanolamine), 2 terpenoids (1-oxo-siaresinolic acid, myrianthic acid), and 1 lipid (LysoPC 20:4). These metabolites were up-regulated in LNA and down-regulated in LN, with the exception of terpenoids (Supplementary Table 3). Other metabolites were only related to synthesis under N deficiency (Supplementary Table 2), suggesting that exogenous arginine did not affect these pathways.

The content of arginine, glutamic acid, proline, serine, threonine, leucine, and valine was decreased (Figures 10A–C,E–H). However, the application of arginine restored the synthesis of these amino acids. The content of amino acids was higher in M. hupehensis growing under normal N conditions after arginine application. The content of histidine showed the opposite pattern: histidine content was high under LN but low when 100 μmol L^–1 arginine was added at 20 days after N deficiency (Figure 10D). Under normal N conditions, the content of histidine was similar to that of other amino acids. The content of isoleucine significantly increased only in CKA (Figure 10I).