The present experiment was conducted at the experimental station of Southwest University (30°26′ N, 106°26′ E) in Beibei, Chongqing, China, from March 2022 to May 2023. Two-year-old citrus trees of Citrus sinensis (L.) Osbeck grafted on Poncirus trifoliata (L.) Raf. and Citrus junos (Sieb.) Tanaka rootstocks were planted in February 2022 in plastic pots with a top and bottom size of 38 and 30 cm diameter, respectively, and 40 cm height. The scion was Citrus reticulata Ehime No. 38. For potted plant growth, an alkaline purple soil with pH 7.25, organic matter concentration of 9.85 g kg^-1, plant available nitrogen of 44.2 mg kg^-1, plant available phosphorus of 11.2 mg kg^-1, plant available potassium of 172 mg kg^-1, exchangeable calcium of 2921 mg kg^-1, exchangeable magnesium of 125 mg kg^-1, plant available copper of 0.83 mg kg^-1, plant available iron of 17.3 mg kg^-1, plant available manganese of 24.6 mg kg^-1, plant available boron of 0.87 mg kg^-1 and plant available Zn of 0.42 mg kg^-1 was used.

All citrus trees were top grafted in March 2022 with scions of Citrus reticulata Ehime No. 38 and sampled in May 2023 ( Figure 1 ). Fertilizer applications were 200 mg N kg^-1, 90 mg P₂O₅ kg^-1, 150 mg K₂O kg^-1, and 25 mg Zn kg^-1 in 2022 and 50 mg N kg^-1, 90 mg P₂O₅ kg^-1, 150 mg K₂O kg^-1, and 10 mg Zn kg^-1 in 2023. The fertilizers used for these amendments were urea, superphosphate, potassium sulfate, and EDTA-Zn.

The sample collection for biomass measurements was conducted in May 2023, when the leaves of the spring shoots were fully mature. For this purpose, citrus trees were harvested, cleaned from soil, divided into young leaves, mature leaves, stems, and roots. All samples were dried at 105°C for 30 min, and then at 65°C to constant weight for dry weight determination. The shoot biomass was calculated as the sum of the dry masses of young leaves, mature leaves and stems. The total biomass was calculated as the sum of shoot and root biomass. Samples collected from individual plants were treated as one biological replicate, and three biological replicates were collected for each treatment. For physiological and metabolomic analyses, 5 young leaves and 5 mature leaves were collected from each plant.

The Zn concentrations of young leaves, mature leaves, stems and roots were determined by ICP-OES (ICP-OES5110010499, Agilent Technologies, USA) analysis after microwave digestion of tissue samples with HNO₃-H₂O₂ (Wang et al., 2023).

At clear weather conditions, the Li-6400 portable photosynthesis measurement system (Li-Cor, Inc., Lincoln, Nebraska, USA) was used for the determination of H₂O and CO₂ gas exchange of the leaves. The parameters inside the leaf cuvette were set as follows: photosynthetically active radiation (PAR) at 1200 μmol m² s^-1, CO₂ concentration at 400 μmol mol^-1, temperature at 30°C, and relative humidity between 45% and 65%. Measurements included net photosynthetic rate (P_n), intercellular CO₂ concentration (C _i ), stomatal conductance (g _s), and transpiration rate (T_r).

Young leaves (from spring branches) and mature leaves (from last year’s autumn branches) were collected, snap-frozen in liquid N₂ and stored at -80°C until further analyses. Before analyses, leaf samples were ground by mortar and pistil under liquid N₂. Chlorophyll a, chlorophyll b, and carotenoids were extracted from 0.1g fresh leaves with 80% acetone, and the absorbance of the extracts was measured at 665, 645, and 470nm with a spectrophotometer (Xiong et al., 2022). Soluble sugar, starch, soluble protein and free amino acid concentrations were determined with commercial test kits (Nanjing Boyan Biotechnology Co., Ltd., Nanjing City, Jiangsu Province, China), according to the manufacturer’s instructions. The anthrone method was applied to measure soluble sugar and starch at 620 nm absorbance (Feng et al., 2023), using 0.1g fresh leaves. For the determination of soluble protein and free amino acid concentrations (Song et al., 2015; Hameed et al., 2023), 0.1g leaf samples were ground at 4 °C and the supernatant was collected after centrifugation at 8000 g and 4 °C for 15 min. The absorbance of the extract was determined at 562nm and 570nm for soluble protein and free amino acid quantification, respectively. All standard substances used for detection were provided by the purchased assay kits.

The Shapiro-Wilk test indicated that data for all plant parameters were normally distributed (P > 0.05). Levene’s test confirmed homogeneity of variances between the groups (P > 0.05). Therefore, independent sample t–test was used to analyse the effect of the two rootstocks on plant parameters (P < 0.05). All statistical analyses were performed using SPSS version 16.0 (IBM Corp., Armonk, NY, USA). Graphical analysis was performed using the origin 2018 software (OriginLab Corp., Northampton, MA, USA).

The growth of citrus trees top-grafted on C. junos (Sieb.) Tanaka rootstock was better than that on Poncirus trifoliata (L.) Raf. rootstock ( Figure 2A ). Compared with Poncirus trifoliata rootstock, the shoot, root and whole tree biomass of citrus trees top-grafted on C. junos were significantly higher by 22.93%, 38.48% and 28.01%, respectively ( Figure 2B ). These results show that citrus top grafted on C. junos (Sieb.) Tanaka rootstocks had the better growth performance of all plant parts.

Compared with Poncirus trifoliata (L.) Raf. rootstocks, C. junos (Sieb.) Tanaka rootstocks significantly increased the Zn concentrations of young leaves, mature leaves, stems, and roots by 81.69%, 66.18%, 97.52%, and 45.94%, respectively ( Figure 3A ). In addition, the aboveground, root and whole tree Zn accumulation of citrus trees top-grafted on C. junos rootstocks were significantly higher than for citrus trees top-grafted on Poncirus trifoliata rootstock by 128.99%, 102.13% and 114.21%, respectively ( Figure 3B ). In citrus trees top grafted on C. junos rootstocks, total Zn partitioning was increased in favor of aboveground tissues by 6.89% compared to Poncirus trifoliata rootstocks ( Figure 3C ). Thus, C. junos (Sieb.) Tanaka rootstocks were more favorable for the absorption and utilization of Zn.

Compared with Poncirus trifoliata (L.) Raf. rootstocks, C. junos (Sieb.) Tanaka rootstocks significantly increased net photosynthetic rate (P_n), stomatal conductance (g _s), and transpiration rate (Tr) of young and mature leaves by 140.18%, 136.39%, and 73.23%, and by 169.74%, 99.94% and 55.04%, respectively ( Figures 4A, C, D ). In general, C. junos (Sieb.) Tanaka rootstock has a positive effect on photosynthetic efficiency of top-grafted citrus.

Young leaves of citrus trees top-grafted on Poncirus trifoliata rootstocks exhibited severe zinc deficiency mediated chlorisis ( Figure 5A ), whereas mature leaves showed only mild zinc deficiency symptoms. Compared with Poncirus trifoliata (L.) Raf. rootstocks, C. junos (Sieb.) Tanaka rootstocks significantly increased chlorophyll a, chlorophyll b, total chlorophyll, and carotenoids concentrations of young leaves by 22.96%, 25.82%, 23.93%, and 42.64%, respectively ( Figure 5B ). C. junos rootstocks significantly increased chlorophyll a, chlorophyll b, total chlorophyll, and carotenoids concentrations of mature leaf by 55%, 56.16%, 55.38%, and 60.09%, respectively ( Figure 5C ).

Compared with Poncirus trifoliata (L.) Raf. rootstocks, the C. junos (Sieb.) Tanaka rootstocks significantly increased starch, soluble sugar, soluble protein and free amino acid concentrations of young and mature leaves by 53.53%, 87.55%, 68.08% and 27.16%, and by 29.63%, 31.71%, 43.93%, and 30.65%, respectively ( Figure 6 ). Thus, C. junos (Sieb.) Tanaka Rootstocks better supported C and N nutrition of the citrus trees.

To identify key metabolites and pathways involved in carbon and nitrogen partitioning of young and mature leaves on citrus trees, top grafted on different rootstocks, we conducted non-targeted metabolome analysis ( Figure 7 ). The score of the first principal component (PC1) was 41.65%, the score of the second principal component (PC2) 20.98% ( Figure 7A ). This results indicates separation of metabolite concentrations between the different rootstocks. The heat-map showed that the replicate samples for different rootstocks are tightly clustered, indicating good data reproducibility. By analyzing the correlations between samples, we found good replicability within the groups ( Figure 7B ).

We identified a total of 4952 differentially accumulated metabolites. A total of 2425 differentially accumulated metabolites were identified in the comparison between young leaves on plants top-grafted with C. junos (Sieb.) Tanaka rootstocks and Poncirus trifoliata (L.) Raf. Rootstocks, with 1400 being upregulated and 1025 being downregulated ( Figure 7C ). In the comparison of mature leaves, a total of 2372 differentially accumulated metabolites were identified between plants with different rootstocks, with 1254 up-regulated and 1118 down-regulated ( Figure 7D ).

The differentially accumulated metabolites selected from each comparison group were matched to the KEGG database to obtain the pathways involved in these enrichments. We selected the top 20 pathways with P-values in ascending order to draw bubble charts. In the comparison group of young leaves ( Figure 8A ), 16 differentially accumulated metabolites were enriched in gyoxylate and dicarboxylate metabolism (ko00630), 17 in ascorbate and aldarate metabolism (ko00053), 20 in D-Amino acid metabolism (ko00470), 9 in valine, leucine and isoleucine biosynthesis (ko00290), and 10 in pyruvate metabolism (ko00620). In the comparison group of mature leaves ( Figure 8B ), 11 differentially accumulated metabolites were enriched in C5-Branched dibasic acid metabolism (ko00660), 15 in ascorbate and aldarate metabolism (ko00053), 12 in betalain biosynthesis (ko00965), 13 in glyoxylate and dicarboxylate metabolism (ko00630), and 10 in the pentose phosphate pathway (ko00030). Analysis of the top 5 pathways of significant enrichment in the comparison groups of young and mature leaves revealed that differentially accumulated metabolites in both comparisons were enriched in glyoxylate and dicarboxylate metabolism (ko00630) and ascorbate and aldarate metabolism (ko00053) ( Figure 8C ).

Compared to plants with Poncirus trifoliata (L.) Raf. rootstocks, most of the differentially accumulated metabolites in the leaves of citrus plants top-grafted on C. junos (Sieb.) Tanaka rootstocks were significantly up-regulated in the glyoxylate and dicarboxylate metabolism (ko00630) pathways ( Figure 8D ). Among these metabolites, pyruvate, D-glycerate, 3-phospho-D-glycerate, 3-propylmalate and L-glutamate were significantly up-regulated in both young and mature leaves. In addition, (S)-malate, glycolate, glycine and L-serine were significantly up-regulated in young leaves, while 2-hydroxy-3-oxopropanoate was significantly up-regulated in mature leaves. Compared with citrus plants on Poncirus trifoliata (L.) Raf. rootstocks, most of the down-regulated differentially accumulated metabolites in the leaves of citrus plants top-grafted on C. junos (Sieb.) Tanaka rootstocks were enriched in the ascorbate and aldarate metabolism (ko00053) pathways ( Figure 8D ). Among these metabolites L-dehydrosorbate, 3-dehydro-L-threonate, myo-inositol, D-glucarate, D-galactarate, D-galactaro-1,5-lactone, D-arabinono-1,4-lactone and L-arabinono-1,4-lactone were significantly down-regulated in both young and mature leaves In addition, L-gulono-1,4-lactone and 2,5-dioxopentanoate were significantly down-regulated in young leaves, while L-ascorbate was significantly down-regulated in mature leaves.

As shown in Supplementary Figure S1 , Zn concentration positively correlated with the physiological parameters analyzed in citrus leaves except for intercellular CO₂ concentration (C _i ). In both, young and mature leaves, Zn concentrations were positively correlated with pyruvate, D-glycerate, 3-phospho-D-glycerate, 3-propyImalate and L-glutamate, while negatively correlated with L-dehydrosorbate, 3-dehydro-L-threonate, myo-inositol, galactaric acid, D-galactaric acid, D-galactaro-1,5-lactone, D-arabinono-1,4-lactone and L-arabinono-1,4-lactone. In addition, in young leaves, (S)-malate (P<0.01), glycolate (P<0.001), glycine (P<0.001), and L-serine (P<0.01) were significantly positively correlated with Zn concentrations, while D-glucarate (P<0.001), L-gulono-1,4-lactone (P<0.001), and 2,5-dioxopentanoate (P<0.01) were significantly negatively correlated with Zn concentrations ( Supplementary Figure S1A ). In mature leaves, 2-hydroxy-3-oxopropanoate (P<0.001) was significantly positively correlated with the Zn concentration ( Supplementary Figure S1B ).

Consistent with our hypothesis (i), different rootstocks lead to significant differences in growth and leaf phenotypes of citrus plants. Citrus plants grafted on Poncirus trifoliata rootstock exhibit poor growth ( Figure 2A ), and display Zn deficiency symptoms such as chlorosis in the leaves ( Figure 4A ). This indicates that selecting appropriate rootstocks can alleviate weakened growth caused by top grafting. Also previous studies showed that different rootstocks have a significant impact on the growth and development of grafted plants (Zhu et al., 2020; Castle, 2010). Citrus plants grafted on Poncirus trifoliata rootstocks exhibited inhibited growth in zinc-deficient soils (Chen et al., 2014) and nutrient deficiency and stunted growth in alkaline soils (Wu et al., 2019). Poncirus trifoliata (L.) Raf. has short or only few root hairs under field conditions, which limits its nutrient absorption (Wu et al., 2011). C. junos rootstocks can regulate root hormone signaling pathways to help the roots adapt to alkaline environments, thus maintaining normal plant growth (Wu et al., 2019). As a consequence, Citrus plants grafted on C. junos rootstocks exhibit significantly higher shoot biomass, root biomass, and total biomass compared to those on Poncirus trifoliata rootstocks ( Figure 2B ).

Rootstock types have been reported to affect foliar mineral element concentrations of grafted trees (Mattos et al., 2003; Toplu et al., 2008). This may be related to differences in the root architecture, such as deep and medium branching, and their capacity for nutrient uptake (Ghimire et al., 2023). In the present study, different rootstocks induced variations in Zn nutrition in top-grafted citrus trees ( Figure 3 ), consistent with our hypothesis (i),. This indicates that replacing Poncirus trifoliata rootstocks with C. junos rootstocks can counteract Zn deficiency in citrus plants top-grafted under alkaline conditions. Differences in Zn concentrations of citrus leaves induced by rootstocks have already been reported (Yilmaz et al., 2018). In alkaline soils, when Poncirus trifoliata is used as rootstock, Zn absorption is very low, whereas C. junos rootstocks can absorb Zn at higher levels (Balal et al., 2011; Rodríguez-Gamir et al., 2010). This difference in Zn acquisition can be ascribed to the upregulation of Zn transport-related genes, such as ZIP1, ZIP5, and ZIP10, in C. junos under alkaline conditions (Wu et al., 2019). The present study also show that C. junos rootstocks significantly enhance Zn allocation to the shoots ( Figures 3B, C ). Previous study found that a Zn/Fe-regulated transporter (ZRT/IRT)-related protein (ZIP) gene was significantly induced in leaves and roots of Zn-deficient trifoliate orange plants (Fu et al., 2017). Therefore, rootstock-specific regulation of Zn transporters, such as ZIP genes, may contribute to the enhanced Zn uptake and transport observed in C. junos, though this remains to be experimentally confirmed. Notably, Zn plays a crucial role in enhancing fruit yield and quality (Arshad et al., 2024). This study also provides a theoretical basis for improving fruit quality.

Consistent with our hypothesis (ii), different rootstocks induced variations in photosynthetic efficiency and carbon-nitrogen accumulation in grafted citrus leaves ( Figures 4 - 6 ). C. junos rootstocks enhanced the net photosynthetic rate ( Figure 4A ), stomatal conductance ( Figure 4C ), and transpiration rate ( Figure 4D ) of the leves. Apparently, the rootstock affects the photosynthetic efficiency, energy metabolism, and protein synthesis of leaves of top-grafted citrus trees. Previous studies indicated that zinc enhances K⁺ influx into guard cells, leading to improved stomatal conductance and transpiration efficiency (Hassan et al., 2020). Moreover, Zn acts as a cofactor for photosynthetic enzymes like carbonic anhydrase and RuBisCO, thereby enhancing CO₂ fixation efficiency and increasing net photosynthetic rate (Dang et al., 2024).

Compared to Poncirus trifoliata rootstock, C. junos rootstock significantly increased the photosynthetic pigment concentration in the leaves of top-grafted citrus trees ( Figures 5B, C ). Also in previous studies, the chlorophyll concentration of leaves in citrus on C. junos rootstocks was higher than on Poncirus trifoliata rootstock (Yin et al., 2021; Zhu et al., 2021). This finding may be related to the differences in Zn deficiency-induced chlorosis, as Zn is an essential cofactor for enzymes involved in chlorophyll synthesis (Hussain et al., 2021). The C. junos (Sieb.) Tanaka rootstock increased the sugar levels in top-grafted citrus leaves ( Figures 6A, B ). In previous studies of Tamarix chinensis and tobacco, this effect was attributed to increased chlorophyll concentrations (Sun et al., 2021; Chen et al., 2016) that improved photosynthesis and positively affecting sugar production in leaves (Pacholczak and Nowakowska, 2020). Carbohydrates not only serve as energy reserves but also provide carbon skeletons for nitrogen assimilation into amino acids, and further on into proteins (Baslam et al., 2021). Consequently, our study found higher levels of free amino acids and protein in the leaves of top-graft citrus trees using C. junos (Sieb.) Tanaka as the rootstock ( Figures 6C, D ). Although fruit quality was not evaluated in this study, previous research has shown that C. junos rootstock can promote amino acid biosynthesis and increase organic acid content in citrus fruits (Xiong et al., 2023; Wang et al., 2024), both of which are key contributors to improved fruit flavor and nutritional quality. Based on this characteristic of the C. junos rootstock, our research group has also been continuously conducting studies on fruit quality and amino acid metabolism in citrus trees grafted onto C. junos (Zhao et al., 2025).

In this study, metabolomics analysis revealed that most of the metabolites in the glyoxylate and dicarboxylate metabolism pathway (ko00630) in citrus leaves top-grafted on Poncirus trifoliata rootstock were significantly downregulated ( Figure 8D ), with both young and mature leaf showing signs of Zn deficiency. Due to their higher metabolic activity, young leaves require more Zn, leading to a greater reduction in metabolites, whereas mature leaves, having lower metabolic demands, are less affected by Zn deficiency. Zn acts as an essential cofactor for many enzymes in carbon and nitrogen metabolism, participating in key reactions in glycolysis, the tricarboxylic acid cycle (TCA cycle), and the glyoxylate pathway (Yang et al., 2023). Additionally, Zn is crucial for amino acid synthesis (Barrameda-Medina et al., 2017). Therefore, metabolomic analysis revealed low abundances of glycine and L-serine in zinc-deficient citrus leaves induced by Poncirus trifoliata rootstock. Glycine is a critical intermediate in the photorespiratory pathway (Jiang et al., 2023), and its reduced levels can restrict photorespiration, thereby suppressing photosynthesis and limiting carbohydrate synthesis. Moreover, studies have shown that zinc deficiency reduces the activity of enzymes such as pyruvate kinase and malate dehydrogenase in plants (Wang et al., 2019; Navarro-León et al., 2016), leading to decreased accumulation of metabolites including pyruvate, malate, and 3-propylmalate. In summary, Zn deficiency limited carbon dioxide fixation and decreased the accumulation of non-structural carbohydrates in leaves. It also reduced the production of 3-phospho-D-glycerate, which in turn suppressed glycolysis and led to decreased pyruvate levels. This also limited the entry of pyruvate into the tricarboxylic acid (TCA) cycle, thereby reducing the production of intermediates such as malate and oxaloacetate. These metabolites are not only key nodes in energy metabolism, but also provide carbon skeletons for amino acid synthesis (Morley et al., 2023; Zhu et al., 2021). Their reduction further inhibited amino acid and protein biosynthesis, which supported our physiological findings.

Metabolome analysis also revealed that numerous metabolites in the ascorbate and aldarate metabolism pathway were significantly upregulated in citrus leaves top-grafted onto Poncirus trifoliata rootstock, with L-dehydroascorbate, 3-dehydro-L-threonate, myo-Inositol, D-glucarate, and D-galactarate notably increased in both young and mature leaves ( Figure 8D ). These metabolites play crucial roles in antioxidant pathways, essential to mitigate oxidative stress (Xu et al., 2018; Gu et al., 2024). Meanwhile, since Zn is a cofactor for various antioxidant enzymes, its deficiency results in ROS accumulation and oxidative stress (Ren et al., 2024). In addition, the divergent upregulation of some metabolites reveals the varying metabolic strategies between new and old leaves in their response to Zn deficiency. For instance, the significant upregulation of L-gulono-1,4-lactone and 2,5-dioxopentanoate in young leaves suggests that they respond more rapidly to oxidative stress by enhancing the synthesis of ascorbate precursors or intermediates (Wagschal et al., 2020). In contrast, the increase of L-ascorbate levels in mature leaves may indicate a greater reliance on existing antioxidant substances rather than a rapid response through enhanced synthesis of precursor metabolites. In general, the pronounced increase in L-dehydroascorbate and L-ascorbate indicates that the plant is strengthening its antioxidant defenses to eliminate excessive ROS and sustain cellular redox equilibrium (Wang et al., 2025).

Consistent with our hypothesis (iii), rootstocks significantly effected leaf metabolism in top-grafted citrus trees. Overall, Poncirus trifoliata rootstock induced a zinc deficiency response in both young and mature leaves of top-grafted citrus trees, leading to disruptions in carbon and nitrogen metabolism and activation of oxidative stress. In contrast, C. junos (Sieb.) Tanaka rootstock maintained better zinc nutrition and metabolic homeostasis.