Photosynthetic performance and photosynthesis-related gene expression coordinated in a shade-tolerant species Panax notoginseng under nitrogen regimes

Background Nitrogen (N) is an essential component of photosynthetic apparatus. However, the mechanism that photosynthetic capacity is suppressed by N is not completely understood. Photosynthetic capacity and photosynthesis-related genes were comparatively analyzed in a shade-tolerant species Panax notoginseng grown under the levels of low N (LN), moderate N (MN) and high N (HN). Results Photosynthetic assimilation was significantly suppressed in the LN- and HN-grown plants. Compared with the MN-grown plants, the HN-grown plants showed thicker anatomic structure and larger chloroplast accompanied with decreased ratio of mesophyll conductance (gm) to Rubisco content (gm/Rubisco) and lower Rubisco activity. Meanwhile, LN-grown plants displayed smaller chloroplast and accordingly lower internal conductance (gi). LN- and HN-grown individuals allocated less N to light-harvesting system (NL) and carboxylation system (NC), respectively. N surplus negatively affected the expression of genes in Car biosynthesis (GGPS, DXR, PSY, IPI and DXS). The LN individuals outperformed others with respect to non-photochemical quenching. The expression of genes (FBA, PGK, RAF2, GAPC, CAB, PsbA and PsbH) encoding enzymes of Calvin cycle and structural protein of light reaction were obviously repressed in the LN individuals, accompanying with a reduction in Rubisco content and activity. Correspondingly, the expression of genes encoding RAF2, RPI4, CAB and PetE were repressed in the HN-grown plants. Conclusions LN-induced depression of photosynthetic capacity might be caused by the deceleration on Calvin cycle and light reaction of photosynthesis, and HN-induced depression of ones might derive from an increase in the form of inactivated Rubisco.


Background
Nitrogen (N) is a major limiting factor in natural ecosystems and in most agricultural systems [1,2]. N is regarded as a necessary component of numerous biomolecules, such as DNA, RNA, proteins, chlorophyll (Chl) and cell envelope [3,4]. N shortage results in enormous changes in plant morphology and even destroys the balance of biological process, including N metabolism and photosynthesis [5,6]. N-deficient crops show the premature of leaves, and reduce leaf area expansion, plant height and ultimately yield of their own [5][6][7][8][9]. On the other hand, excessive N supply makes leaves dark green and stems frail and immature, and consequently cause an imbalance between the vegetative and reproductive growth [10][11][12]; For example, excessive N supply considerably reduces the biomass of cucumber (Cucumis sativus) [13] and of tomato (Lycopersicon esculentum) [14] . However, N surplus in plants receives relatively little attentions in comparison with N deficiency over the past decades.
It has been commonly accepted that photosynthesis is highly influenced by leaf anatomy and chloroplast ultrastructure. HN-grown Arabidopsis. thaliana displays thicker upper epidermises, lower epidermises, spongy tissue and palisade tissue, and increased thickness of anatomic structure would not facilitate CO 2 diffusion in the liquid phase of mesophyll cells [15]. N deficiency exhibits small chloroplast with lower internal conductance (g i, ) [16], and a large chloroplast with well-developed grana under high-N application has been reported in summer maize [17]. Indeed, Photosynthesis-related components are strongly regulated by leaf N and photosynthetic capacity is closely related to N content since more than 50% of total leaf N is allocated to photosynthetic machinery and proteins of Calvin cycle represent the majority of leaf N [18][19][20][21]. In leaves of developing maize (Zea mays), N deficiency results in an obvious decrease in photosynthesis with an reduction in activities of phosphor enolpyruvate carboxylase (PEPC), pyruvate orthophosphate di-kinase (PPDK) and ribulose 1, 5bisphosphate carboxylase (Rubisco) [17,[22][23][24]. A reduction in content of Rubisco and in effective and maximum quantum yield of photosystem II (ΔF v /F m & F v / F m ) has been recorded in A. thaliana and Oryza sative grown under N deficiency condition [25,26]. Likewise, Rubisco carboxylase activity considerably declines in spinach (Spinacia oleracea) and cassava (Manihot esculenta) due to N deficiency [27,28]. On the other hand, negative responses of photosynthetic capacity to excess N, including decreased Rubisco activity, lower N allocation to light-harvesting system (N L ) and lower photosynthetic efficiency, have been observed in field-grown wheat [29], rice [30] and cotton [31]. In general, the relationship between N levels and photosynthesis is nonlinear in a sufficiently broad range of leaf N content [32,33]. However, relatively little is known about its molecular mechanism on the nonlinear relationship between leaf N and photosynthesis.
N-suboptimal plants would suffer from greater excess of light energy, and this could produce excessive reactive oxygen species (ROS) [34]. Plants has employed a series of photoprotective mechanism to survive long periods of no-optimal N regimes. Non-photochemical quenching (NPQ) of excess light energy within the light-harvesting antennae of PSII (LHCII) are believed as an effective photoprotective mechanisms, as observed in A. thaliana plants subjected to low nitrogen (LN) [26,35,36] and in Coffea Arabica [37] and Lavandula angustifolia [38]. Excess N supply in benthic diatom (Entomoneis paludosa) [39] have revealed an up-regulation of xanthophyll cycle, which reduce the efficiency of PSII photochemistry and enhance NPQ. In addition, several studies have highlighted a positive effect of glycolytic pathway and pentose phosphate pathway (PPP) on energy and carbon balance in Nstressed plants [40,41]. Unexpectedly, molecular mechanisms of photoprotection are not completely clear in the N-stressed non-model species, especially in a shadetolerant plant.
Comparative transcriptomes have revealed that unigenes expression of Chl biosynthesis, Calvin cycle and ribosomal proteins were decreased in Scenedesmus acuminatus under high N (HN) supply [42]. Up-regulated gene transcripts are predominantly matched in kinds of amino acid metabolism, transport and stress, whereas repressed transcripts are overrepresented in categories of hormone metabolism and redox control in roots of A. thaliana under N deficiency [43]. During periods of Nlimitation, gata transcription factor (GNC), a gene regulating carbon (C) and N metabolism, operates to support A. thaliana survival by elevating Chl biosynthesis [44]. Genes encoding enzymes for C skeleton production are down-regulated in spinach plants under N-starvation, and plants also significantly show low contents of amino acid and high levels of glucose and consequently decelerate growth [45]. Light reaction center of photosynthesis by extrinsic proteins labled as PsbO, PsbP, PsbQ, PsbR, PsbU and PsbV are suppressed in Synechocystis under N stress [46]. PsbS protein is activated by the acidity of thylakoid lumen in A. thaliana plants under N-stressed condition [47]. In addition, the expression of NR and GOGAT was dramatically up-regulated in the cucumber exposed to HN [48]. Surprisingly, relatively less investigation has been conducted to elucidate the correlation of photosynthesis-related genes expression with photosynthetic performance in the context of N.
notoginseng is believed to be highly sensitive to high light, and 10% of full sunlight is suitable for its growth [50,52]. Besides, the development and growth of P. notoginseng is highly sensitive to high N [53][54][55]. HN application considerably enhance rust cracking, root decay and mortality rate of P. notoginseng [56]. Indeed, significant decreases in root, stem and leaf biomass have been observed in P. notoginseng grown under LN, along with narrow and yellow leaves [57,58]. However, these previous studies have mainly focused on effects of N input on agronomic traits, yield, and plant growth. Nowadays, the molecular mechanism of the sensitivity of P. notoginseng to N is still unclear.
Different N levels were applied to P. notoginseng, and photosynthetic capacity, photoprotection and photosynthetic pigments were comparatively analyzed in the plants grown under low N (LN), moderate N (MN) and high N (HN). Meanwhile, a comprehensive transcriptome was conducted to elucidate the expression of photosynthesis-related gene. The objective of our study was to elucidate the photosynthetic performance and the expression of photosynthesis-related genes in the typically shade-tolerant and N-sensitive plant P. notoginseng under different levels of N, and it was anticipated that photosynthetic performance might be coordinated with the expression of photosynthesis-related genes.

Results
Effect of N regimes on plant growth and leaf gas change HN-grown leaves were pretty dark-green, and LNgrown leaves were significantly smaller and yellowish (Additional file 1: Figure S1a). HN-grown plants possessed a low survival rate (Additional file 1: Figure  S1b). LN-grown leaves were dramatically reduced in the thickness of upper epidermis, lower epidermis, spongy tissue and palisade tissue, and biomass of leaf was significantly reduced in LN and HN treatments (Table 1). On the other hand, LN significantly decreased the size of chloroplasts accompanied with a reduction in chloroplast exposed to intercellular air space per unit leaf area (S c ), and correspondlingly an increase in the size of chloroplasts was observed in P. notoginseng under excessive N supply (Fig. 1 c; Table 2). The LN plants and HN plants showed 52.7 and 96.8% lower liquid phase (g lip ) than the MN plants, respectively. g lip can be expressed as g lip = C lip × S C , therefore, conductance per unit of exposed chloroplast surface area (C lip ) is one of a determinant of g lip . C lip were reduced in HN-grown plants as compared to the MN-grown individuals. Internal CO 2 condutance (g i ) is mainly determined by g lip , and LN-and HN-grown plants was decreased in g i ( Table 2) .

N-induced changes in photosynthetic capacity
The leaf exhibited a significant difference in a response of net photosynthetic assimilation (A net ) to incident photosynthetic photon flux density (PPFD) and to internal leaf CO 2 concentrations (C i ) within N regimes (Fig. 2). The maximum net photosynthetic assimilation (A max ), CO 2 response curves and carboxylation efficiency (CE), maximum electron transfer rate (J max )and maximum carboxylation efficiency (V cmax) were highest in MN-grown plants; however, these variables did not show apparent differences between LN and HN individuals except for A max (Table 3). N allocation to the photosynthetic system (N photo ) is the sum of N allocation to the carboxylation system (N C ), the bioenergetics component (N B ) and the light-harvesting system (N L ). N content per unit leaf area (SLN) was increased significantly with the increase in N application (Table 3). HN treatment caused a significant increase in N L , whereas there is a significant reduction in N C in HN-grown plants (Fig. 3 a). Most importantly, photosynthetic N use efficiency (PNUE) was significantly decreased from 45.2 to 20.3% with an increase in N supply ( Fig. 3 b). These results support that high N photo did not trigger an increase in PNUE. SLN and Rubisco content were greater when plants were exposed to high N as compared with ones to moderate and low N (Additional file 2: Figure S2; Tables 2,  3). MN treatment exhibited 44.4-98.4% more Rubisco activity than two other treatments ( Table 2). There were no significant differences in stomatal conductance (g s ) within treatments, but g m were higher in HN-grown individuals (Table 2). HN-grown plants had a decreased ratio of g m /Rubisco content and a lower Rubisco specific activity than two other treatments ( Table 2).

Photosynthetic electronic transport
The responses of photosynthetic electronic transport to continuous steady-state light were markedly different among N regimes (Additional file 3: Figure S3; Fig. 4). In the light response curves, the minimum values of PSII maximum quantum efficiency(F v`/ F m`) , PSII photochemical quantum yield (Φ PSII ), photochemical quenching(qP) as well as PSII total electron transport rate (J T ), rate of electron transport for oxidation reaction (J O ), carboxylation reaction (J C ) and the maximum values of non-photochemical quenching (NPQ) were generally recorded in the LN individuals, the maximum value of F v`/ F m`, Φ PSII , qP as well as J T , J O , J C were obtained in the MN ones (Additional file 3: Figure S3; Fig. 4).

Changes in photosynthetic-related pigments
The amounts of β-carotene (β-Cars) and the ratio of (V + A + Z)/Chl were enhanced in LN individuals, whereas total Chl decreased ( Table 4). LN resulted in a decrease in neoxanthin (N) and lutein (L), and an increase  Values are means ± SD. (n = 7). Different letters among nitrogen regimes indicate significant difference (P ≤ 0.05). g s : stomatal conductance; g m : mesophyll conductance; R d : dark respiration rate; g lip : liquid phase; C c : chloroplastic CO 2 concentration; S: the specificity factor of Rubisco for O 2 and CO 2 ; S*: apparent Rubisco specificity; S c : chloroplast exposed to intercellular air space per unit leaf area; C lip : conductance per unit of exposed chloroplast surface area; g i : internal CO2 condutance in violaxanthin(V), antheraxanthin (A), and zeaxanthin(Z). Violaxanthin de-epoxide activity ((A + Z)/(V + A + Z)) was greatest in the LN ones (Table 4).

Gene expression identification
Compared to the MN individuals, 1391 and 895 genes were classified as differentially expressed genes (DEGs) in the LN and HN groups. Whereas, there were 428 DEGs in both LN-and HN-treatments (Fig. 5). In the LN group, 467 DEGs were up-regulated, and 924 DEGs were down-regulated. Two hundred ninety-four genes were up-regulated and 601 genes were suppressed in HN individuals (Additional file 4: Figure S4). Moreover, 963 and 467 DEGs were typically detected in LN, HN groups. Two DEG sets were subjected to 34 Gene ontology (GO) classes (Fig. 6). Under the classification of molecular function, "catalytic activity" were largely represented, followed by "binding" (Fig. 6 a). The GO enrichment was further analyzed to identify specific GO enrichment terms among DEG sets. Based on Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analysis, the DEGs in LN ones were categorized into photosynthesis, carbon fixation, N metabolic, plant hormone signal transduction, starch and sucrose metabolism and galactose metabolism, the DEGs in HN ones were significantly overrepresented in pathway of citrate cycle (TCA cycle), alpha-Linolenic acid metabolism, carbon fixation in photosynthetic organism, N metabolism and galactose metabolism ( Fig. 6 b). In addition, the first 13 pathways widely related to the mechanism about photosynthesis and photo-protection were explored among KEGG enrich   analysis of all annotated unigenes (Additional file 5: Table  S1).

Transcriptional changes
GO enrichment analysis was presented and elucidated in Fig. 7. Enriched GO terms of further induced genes between two pairwise comparisons (MN vs LN., MN vs HN.) embraced photosynthesis, pigment metabolic process, carbohydrate catabolic process, thylakoid and so on. Common DEGs with suppressed expression were significantly enriched in cellular amino acid catabolic process, alpha-amino acid catabolic process, proline metabolic process and glutamine family amino acid metabolic (Fig. 7). KEGG pathway analysis was further certified distinct functional enrichments in biological process among common DEGs (Fig. 7 b, c), revealing that these induced expression of common DEGs were richen in TCA cycle, photosynthesis, carbon fixation in photosynthetic organism, glycolysis/ gluconeogenesis and carbon metabolism, and downregulated DEGs were primarily related to N metabolism, glutathione metabolism, biosynthesis of amino acids and starch and sucrose metabolism. In addition, a large number of specific DEGs involved in diverse biological processes were detected in the MN vs. HN ones, and response patterns in the LN and HN level also exhibited differences.

Genes expression related to Calvin cycle and light reaction
Interestingly, the expressions of the majority of genes encoding enzymes in Calvin cycle were down-regulated between LN and HN individuals, furthermore, the transcript levels of a substantial number of genes were reduced in LN ones (Additional file 6: Figure S5; Fig. 8).

Photoprotection-related genes
Both zeaxanthin epoxidase (ZEP) and violaxanthindeepoxidase (VDE) genes are positively induced by LN level (Additional file 8: Figure S7a). Glucose-6-phosphate dehydrogenase (G6PDH) and glutathione S-transferase (GSTs) involved in glucose metabolism was up-regulated in the LN individuals (Additional file 8: Figure S7b). Genes were found to be enriched in Chl degradation, and genes encoding chlorophyll b reductase (NYC) and red chlorophyll catabolite reductase (RCCR) were up-regulated in the LN groups (Additional file 8: Figure S7c). Under LN levels, transcript levels of genes involved in nitrate reduction were up-regulated (Additional file 8: Figure S7d). Expressions of genes encoding phytoene synthase (PSY), geranyl pyrophosphate synthase (GGPS), 1-deoxy-D-xylulose-5-phosphate reductoisomerase (DXR), isopentenyl pyrophosphate isomerase (IPI) and 1-deoxy-D-xylulose-5phosphate synthase (DXS) were up-regulated in LN groups (Additional file 8: Figure S7e). In addition, a substantial number of pathway regulating photosynthesis and photoprotection were activated by N-induction, including glutathione metabolism, N metabolism, carotenoid biosynthesis, photosynthesis-antenna proteins (Fig. 8).  In RT-qPCR analyses, melting curves of actin and 19 photosynthetic-related genes were clear and every curve all held a single and sharp peak (Additional file 9: Figure S8), indicating that the primer pairs could positively amplify specific products of 19 genes (Additional file 10: Table S2). The expressions of 19 genes were approximately similar to the results from RNA-Seq data in leaves (Additional file 11: Figure S9).

N-driven changes in photosynthesis is in part explained by leaf anatomy and N allocation
Photosynthetic capacity is at least in part determined by leaf anatomy and chloroplast ultrastructure [59] and A net is limited by the rate of CO 2 diffusion from the atmosphere to the chloroplast [60]. C lip has a close relationship with mesophyll cell thickness, and a thick tissue is always accompanied with a low C lip [30]. Nevertheless, C lip is positively related to the rate of CO 2 diffusion from the atmosphere to the chloroplast [61]. N deficiency obviously reduce the size of chloroplasts and consequently lead to a low chloroplast surface area exposed to S c [16]. Correspondingly, a large chloroplast has been documented in HN-grown rice, and the large chloroplasts would enhance chloroplastic CO 2 concentration (C c ) and g m [62]. Thicker upper epidermis, lower epidermis, spongy tissue and palisade tissue (Table 1) would reduce liquid phase diffusion of CO 2 in mesophyll cells, evidencing by lower C lip in the HN individuals (Table 2), and  Table 2), as has also been observed in Tosens & Laanisto [63]. C c positively reinforce photosynthetic capacity [64][65][66]. Nevertheless, HN supply resulted in a decline in A net (Fig. 2), it has been thought that the increase in g m is not enough to provide sufficient CO 2 to activate the increased Rubisco, the imbalance between g m and Rubisco content contributes to the reduction in A net in the HN individuals ( Table 2) as has also been observed in Yin & Struik [67]. LN-grown plants displayed small chloroplast and low S c ( Fig. 1; Table 2) and these characteristics suppress g i and A net , as has been proposed by Onoda et al [68] . 30-40% of leaf N is allocated to photosynthetic carboxylation and photosynthetic efficiency is determined by the proportion of N allocated to N C [69]. SLN is significantly increased, but V cmax , CE, Rubisco activity and N C is obviously decreased, and consequently photosynthetic efficiency is suppressed in wheat and rice grown in excessive N environment [29,30]. The LN plants exhibited lower SLN, Chl and N L (Tables 3, 4; Fig. 3 a), this would only limit the synthesis of light-trapping chlorophyll a/ b-protein complexes and effectively prevent absorption of excessive light energy [70]. Correspondingly, lower N C and activity of Rubisco were observed in the HN individuals (Fig. 3 a), this would suppress photosynthetic carboxylation and reduce photosynthetic efficiency.

The dark and light reaction of photosynthesis under nonoptimal N regimes
Rubisco is the key CO 2 -fixing enzyme in Calvin cycle. The depressed photosynthesis in the LN plants might be the consequence of low Rubisco catalytic capacity and CE [30,71]. Rubisco activity and CE were decreased in the LN individuals with down-regulated expression of genes encoding Rubisco (Tables 2, 3; Additional file 6: Figure S5). N deficiency in maize led to the decline in photosynthetic capacity and Rubisco catalytic capacity [22]. According to classic theories [72], V cmax positively reflects potential carboxylation capacity of Rubisco and regeneration rate of ribulose-bisphosphate (RuBP). V cmax and Rubisco activity were considerably reduced in the HN individuals, whereas Rubisco content were increased (Tables 2, 3). On the other hand, the expressions of Rubisco-catalyzing genes (CPN60A1, Os02g079470025 and RAF2) were down-regulated in the HN individuals (Tables 2, 3; Additional file 6: Figure S5). Overall, these results support the view that under HN supply, the majority of Rubisco function as N storage rather than as catalyzing enzyme [30], and the proportion of inactive Rubisco is greater in the HN leaves. However, high Rubisco activity is recorded in the HN-grown maize with high A max [73]. The difference might be explained by the fact that P. notoginseng is a shade-tolerant C3 species and highly sensitive to excess N [54,55], and maize is a high N-and sun-demanding C4 plant [74,75]. In the atmospheric environment, Rubisco has a low affinity with CO 2 , resulting in inferior catalytic capacity [76]. The potential catalytic capacity of Rubisco in C3 plants is much lower than in C4 plants [77], suggesting that relatively less Rubisco operate on photosynthesis, and that a high proportion of Rubisco serves as a storage of N in HNgrown C3 plants.
The expression of genes encoding enzymes involved in regeneration of RUBP was reduced in the LN individuals, including phosphor ribulokinase (PRK), phosphor glycerate kinase (PGK), glyceraldehyde-3phosphate dehydrogenase (GAPDH), triosephosphate isomerase (TPI), fructose-bisphosphate aldolase (FBA), transketolase (TKT) and sedoheptulose 1,7-bisphosphatase (SBP, Additional file 6: Figure S5), and correspondingly A net , Φ PSII , F v`/ F m`a nd J T were reduced. Likewise, N deficiency suppress photosynthetic capacity and inhibit the expression of genes involved in dark reaction of photosynthesis in barley (Hordeum vulgare) [78] and rice [79]. Our findings are in general agreement with several previous researches that down-regulated expressions of genes encoding enzymes involved in the regeneration of Rubisco is associated with abiotic stress [80,81].
Photosynthetic assimilation and photosynthetic electronic transport was considerably decreased in the LN and HN individuals (Figs. 2, 4; Additional file 3: Figure S3), and it is believed to derive from an interruptive synthesis of photosynthetic proteins [82]. The expression of genes encoding structural proteins of photosystems, including PsbA, PsbE, PsbF, PsbH and PsaN, were decreased in the LN-and HN-grown individuals, while the expression of genes encoding subunits of the PsbS and PetE was increased as compared with the MN individuals (Additional file 7: Figure S6a). The expression of genes encoding structural proteins of photosystems are positively correlated with photosynthetic capacity [83,84]. In addition, the down-regulation of genes involved in porphyrin and Chl metabolism and photosynthesis-antenna proteins might also in part explain the depressed photosynthetic capacity in the LN individuals (Fig. 8).

Photoprotection in N deprivation
Lower A max , V cmax and J max were recorded in the LN and HN individuals when compared with the MN individuals (Table 3). No-optimal N application induces suppressions of photosynthesis as has been recorded in the sun-demanding species Viciafaba [85], Lemna minor [86], Z. mays [87] and C. sativus [48], and in the shadetolerant species Abies fabri [88], Brassica juncea [89] and polypodiopsida [90]. On the other hand, Φ PSII was lower in the LN individuals than in the HN individuals (Fig. 4 b), indicating that a greater proportion of light energy absorbed by PSII would have to be expended by non-photochemical process in the LN individuals. The consumption of electrons by non-photochemical process would facilitate the formation of trans-thylakoid ΔpH [86]. ΔpH is also a precondition for the activation of VAZ cycle and NPQ [91]. This indicates that NPQ and V cycle pool might be reinforced in the LN individuals as confirmed by the present study (Fig. 4 d; Table 4).
The degradation of Chl has been believed to be a photoprotective mechanism for plants to cope with stress and to prevent photodamage [92,93]. Chl was considerably reduced in LN-grown plants (Table 4). Genes involved in Chl degradation were found to be enriched in LN-grown plants, and genes encoding NYC and RCCR were up-regulated (Additional file 8: Figure  S7c). The results obtained in our study indicate that low Chl in LN-grown plants might be caused by up-regulation of genes involved in Chl degradation. Similarly, the expression of genes involved in Chl degradation are also elevated in a shade-tolerant plant Neoregelia cruenta when grown under N-limited condition, together with the obvious decrease in leaf Chl [40]. LN-grown Spinach showed low Chl content and the corresponding elevation in expression of genes involved Chl degradation [94,95].
Car is not comprised of N atoms and their accumulation is beneficial for protecting photosystem from photodamage [42,96]. The genes involved in Car biosynthesis (GGPS, DXR, PSY, IPI and DXS) were down-regulated in the HN individuals (Additional file 8: Figure S7e), as has also been observed by Vidhyavathi et al [97]. Correspondingly, the expression of genes involved in Car biosynthesis were up-regulated in Haematococcus pluvialis by a combination of light and N-deprivation [98]. Further, a greater xanthophyll pool size (V + A + Z) and higher deepoxidation state ((A + Z)/(V + A + Z)) were observed in the LN individuals (Table 4), this suggests that LN cloud effectively improve the capacity of heat dissipation. The PsbS protein and violaxanthin de-epoxidase is believed to involve in regulating energy dissipation [93]. The latter catalyses the de-epoxidation of V to Z [99]. ΔpH-dependent quenching (qE) is activated by PsbS and the xanthophyll cycle [100]. The expression of genes encoding ZEP and VDE (key enzymes in xanthophyll cycle) were up-regulated in the LN individuals with the enhanced expression of PsbS genes (Additional file 8: Figure S7a), as has been observed in LN-induced maize where genes involved de-epoxidation state were up-regulated [34].
Nitrate assimilation is a process highly sensitive to N stresses [101]. Nitrite assimilation consumes six electrons from reduced ferredoxin [102,103]. This reaction is a strong alternative sink for photosynthetic electron transport chain. The expressions of genes encoding nitrite reductase (NIR) and nitrate reductase (NIA) were increased in the LN-grown individuals (Additional file 8: Figure S7d). It has been reported that maize [74], A. thaliana [104] and apple [105] show high activity of NIR and NIA in leaves in presence of suboptimal N application. The previous investigations and the present study both strongly evidence that LN-induced enhancement in nitrate assimilation might mitigate the accumulation of excess energy.

Conclusion
The non-optimal N supply significantly suppresses photosynthetic capacity in a typically shade-tolerant and N-sensitive species such as P. notoginseng. Thick leaf limits liquid phase diffusion of CO 2 in mesophyll cells and accordingly reduces internal conductance. Moreover, large chloroplast with low N c results in an imbalance between the increases in gm and in Rubisco content, consequently causing the decreased A net in the HN individuals. In addition, the expression of genes involved in Calvin cycle, Chl biosynthesis and antenna proteins are obviously repressed in the LN individuals; correspondingly, the expression of genes (e.g. RAF2, CAB and PetE) involved in Calvin cycle and light reaction is also inhibited, however, photosynthetic capacity might be primarily inhibited by the inactivated Rubisco in the HN individuals. Overall, our results indicate that photosynthetic performance and photosynthesis-related genes expression is coordinated in a shade-tolerant and N-sensitive plant grown along an N gradient.

Plant materials and growth conditions
Experimental plots were conducted at the teaching and experimental farm of Yunnan Agricultural University in Kunming, Yunnan, China. P. notoginseng is a perennial herb; farmers have cultivated this medicinal crop for more than 400 years. One-year-old P. notoginseng seedlings were collected from the Wenshan Miao Xiang P. notoginseng Industrial Co., Ltd., China (Longitude 104°32 ', latitude 23°53 '). 1-year-old healthy rootstalks of P. notoginseng were selected in our experiments and transplanted to a white plastic pot (30 cm in diameter and 40 cm in depth) on January 2017, and 3 individuals per pot, 120 pots per treatment were arranged.
The soil had the following chemical characteristics: organic mater 0.573%, total N 0.201%, pH (H 2 O) 6.42, total phosphorus (P) 0.727 g/kg, ammonium N 39.93 mg/kg, available potassium (K) 0.019 mg/g, available P 4.88 mg/ kg, soil water regime 12%. Pots were placed in environmentally controlled growth permeable black plastic net with growth irradiance of 10% full sunlight. Three Nfertilizer levels were applied in our experiments: (1)  P. notoginseng grown for 8 months were used to determine plant mortality, leaf morphology and photosynthetic performance, and to collect leaves for comparative transcritome, chlorophyll and elemental N analyses. Five biological replicates were quickly frozen in liquid N and stored at − 80°C for RNA extraction.

Leaf anatomy and chloroplast ultrastructure
The juvenile leaves achieved for morphological and anatomical traits were used after 8 months of the N regimes. Leaf anatomical properties were performed in the method of paraffin section, and then the leaves were dehydrated in an alcohol series. Leaf tissues were embedded in paraffin (Thermo Scientific Histostar™) and transversely sectioned at 10 mm thickness by means of microtome (Microm HM325, Walldorf, Germany). Finally, sections were stained with hematoxylin observed under a bright field Microscope (Zeiss Axio Cam HRC, Oberkochen, Germany).
A small piece of 1~2 mm 2 was cut between the middle leaf vein and leaf edge and fixed with 2.5% glutaraldehyde and 1% osmic acid. According to the conventional series of ethanol dehydration, epoxy resin embedding, ultra-thin slicer sectioning, sectioning was stained with uranyl acetate and then stained with lead citrate, the chloroplast ultrastructure was observed under JEM100CX-II transmission electron microscope. For the estimation of Sc, 700 nm-thick sections were used by the method of Hanba et al [59].

Steady-state gas exchange rate
Steady-state gas exchange measurements were carried out using the photosynthesis system (Li-6400-40, Li-Cor, USA) with the 2 cm 2 fluorescence leaf chamber. The leaf temperature and CO 2 in the chamber were maintained at 25°C and 400 μmol mol − 1 during measurements, respectively. Subsequently, Photosynthetic light response curves and photosynthetic CO 2 response curves were performed. Based on photosynthetic light response curves, full induction was complete, an automatic program of light response curves was run to measure the change in gas exchange rate with a set of PPFD. The level of PPFD was listed in the following order: 800, 500, 400, 300, 200, 100,80,60,40,20 and 0 μmol m − 2 s − 1 , each light intensity stabilized for 5 min. The relationship between A net and PPFD was fitted, A net = A max -A max-C 0 e -αPPFD/ Amax , where A max is the maximum net photosynthetic assimilation under saturating light, α is the apparent quantum efficiency (AQY), where AQY was estimated by the slope of the linear region of the light response curve. C 0 is the index to measure the net photosynthetic rate approaching 0 in low light. According to the parameters in the formula, dark respiration rate (R d ) = A max -A max C 0 .
A net and C i were evaluated at a range of reference CO 2 concentrations (400, 300, 200, 150, 100, 50, 400, 600, 800, 1000 and 1200 1500 μmol mol − 1 ). CO 2 response curves and CE were achieved by fitting the data to a nonrectangular hyperbola and the slope the linear region of the CO 2 response curve, respectively. V cmax and J max was gained according to the idea offered by Buckley and Diazespejo [106], this calibration requires measurements under low O 2 .

Chlorophyll fluorescence of PSII
At predawn, minimum and maximum Chl fluorescence yield (F O and F m ) was measured in the fully darkadapted leaves. Minimum, maximum and steady-state fluorescence intensity (F O`, F V`, F m`a nd F s ) were made in the process of light response curves. F v`/ F m`w as estimated as (F m`-F O`) / F m`; Φ PSII as (F m`-F s )/F m`; J T = PPFD×Φ PSII × α leaf × β, commercial fluorometers usually provide an estimate of PSII total electron transport rate (J T ) by assuming that 400-700 nm (PAR) leaf absorptance (α leaf ) equals 0.84 [107] and that absorbed photons (β) are equally distributed between the two photosystems (β = 0.5) [108]. This approximation is reasonable for comparison of J T between optically similar samples such as leaves of cultivars of a single plant species [109]. Moreover, there was a curvilinear relationship between α leaf and chlorophyll content, whereas the curvature was extremely low when the chlorophyll content was > 0.4 mmol m − 2 [30,110]. According to Evans and Poorter [110], the calculation of α leaf demonstrated that α leaf (0.84, 0.85, and 0.85, in leaves with low, moderate, and high N content, respectively) was similar to the value of 0.84 [111][112][113]. Therefore, in this study, α leaf also assumed to be 0.84, and β was assumed to be 0.5 [108,114]. NPQ as (F m -F m`) /F m`, and qP as (F m`-F s )/(F m-F 0`) . Jc and Jo was calculated according to the method of Valentini et al [115], J O = 2/3 × (J T − 4 × (A net + R d )), J C = 1/3 × (J T + 8 × (A net + R d )). According to the methods of Manter and Kerrigan [116], g m and C lip were calculated as where Γ* is the CO 2 compensation point. C c was calculated as C c = C i × S*/S. The initial slope of the regression of Jc/Jo to C i /O was used to S*(Additional file 12: Figure S10), O 2 concentration (210 mmol CO 2 mol − 1 ). S was calculated as follow: S=O/2Γ*. g lip can be showed that g lip = C lip × S c . g i was calculated by g i = A max /(C i -C c ) .

Calculation of N allocation in photosynthetic components
Leaf N was determined with Kjeldahl. SLN was calculated. Photosynthetic-related pigments were determined by the method of Xu et al. [91] and Thayer & Björkman [117]. N C , N B and N L were determined according to the method of Niinemets et al [118]. N photo is the sum of N C , N B and N L . PNUE is the ratio of leaf N used for C fixation per unit leaf area. The formula is as follows: V cr is the Rubisco specific activity with a value of 20.8 μmol CO 2 ·g − 1 Rubisco·s − 1 . J mc is the maximum electron transfer rate per unit cytochrome f (Cyt f) with a value of 155.6 μmol electrons·μmol-1 Cyt f·s − 1 . Cc is the leaf chlorophyll content (mmol·m-2 ), C B is the combined light system I (PSI), photosystem II (the chlorophyll in PSII) and PSII light-harvesting pigment complex (LHCII) with a value of 2.15 mmol·g − 1 N.

Leaf Rubisco content and activity
The Rubisco content was determined according to Makino et al [119]. Briefly, newly expanded leaves were stored at − 80°C. 0.5 g frozen leaves were ground in a solution containing 50 mM Tris-HCl (pH =8.0), 5 mM β-mercaptoethanol, and 12.5% glycerol (v/v), and then centrifuged at 1500 g for 15 min at 4°C. The supernatants were mixed with a solution containing 2% (w/v) SDS, 4% (v/v) β-mercaptoethanol and 10% (v/v) glycerol, boiled in a water bath for 5 min before SDS-PAGE using a 4% (w/v) stacking gel, and a 12.5% (w/v) separating gel. After electrophoresis, the gels were stained with 0.25% Commassie Blue for 12 h, and destained. Gel slices containing the large subunits and small subunits of Rubisco were transferred to a 10 mL cuvette containing 2 ml of formamide and incubated at 50°C in a water bath for 6 h. The absorbance of the wash solution was measured at 595 nm. Protein concentrations were determined using bovine serum albumin as a standard.
Bovine serum albumin (BSA) was measured at 595 nm as standard protein.
RNA extraction and library construction, sequencing RNA samples were extracted using RNA pre-pure Plant Kit (Tiangen, Beijing, China). After total RNA was extracted, mRNA was enriched by Oligo (dT) bads, and then the enriched mRNA was fragmented into short fragments using fragmentation buffer and reverse transcripted into cDNA with random primers. Secondstrand cDNA were synthesized by DNA polymerase I, RNase H, dNTP and buffer. The cDNA fragments were purified with QiaQuick PCR extraction kit, end repaired, poly(A) added, and ligated to Illumina sequencing adapters. The ligation products were selected by agarose gel electrophoresis, PCR amplified, and sequenced using Illumina HiSeqTM 4000 by Gene Denovo Biotechnology Co. (Guangzhou, China).

Raw reads filtering and de novo assembly
Low quality reads containing adapters, more than 10% of unknown nucleotides (N), were eliminated. Transcriptome de novo assembly was carried out with short reads assembling program-Trinity. The redundancy was eliminated by the TGICL software and further assembled into a set of non-redundant unigenes.
105G sequencing data were obtained and de novo assembled into 93,162 unigenes (Additional file 13: Table  S3) with an average length of 790 bp (Additional file 14: Table S4). Collectively, 41,569 (44.62%) unigenes were functionally annotated in accordance with their parallels with known genes/proteins in the databases. The particular statistics of the functional annotation are emerged as in Additional file 15: Figure S11. After eliminating adaptors, unknown nucleotides and low quality reads, the data generated 43,588,606, 46,978,940, 43,177, 242 paired-end 125-bP reads in the LN, MN and HN treatments, respectively, coinciding with approximately 6.48 Gb data (Additional file 16: Table S5). Q20 percentages exceeded 98%, uncalled base ("N") percentage was equal to 0% per sample (Additional file 16: Table S5). The GC contents were almost identical for all 15 leaves tissues, ranging from 43.08 to 44.20% (Additional file 16: Table S5). In general, between 83.23 and 84.79% of clean reads could be mapped on full gene set (Additional file 17: Table S6). A Pearson's correlation analysis revealed high correlations between biological replicates (R 2 = 0.8671 to 0.9769, Additional file 18: Figure S12).

Analysis of DEGs
To identify DEGs within N regimes, the normalized read counts from five replicates of each sample were analyzed and the edge R package (http://www.r-project.org) was used. We identified genes with a fold change ≥2 and a false discovery rate (FDR) < 0.05 in a comparison as significant DEGs. DEGs were then subjected to enrichment analysis of GO functions and KEGG pathways.

GO enrichment analysis and pathway enrichment analysis
All DEGs were mapped to GO terms in the Gene Ontology database (http://www.geneontology.org), gene numbers were calculated for each term, significantly enriched GO terms in DEGs comparing to genome background were defined by hyper geometric test. KEGG enrichment analysis was carried out through Genomes database (g" http://www.genome.jp/kegg). P-value of GO terms and KEGG pathway was gone through FDR Correction, taking FDR ≤ 0.05 as a threshold.

RT-qPCR assay
To validate the expression of 19 significant DEGs observed in RNA-Seq data, reaction was carried out using EvaGreen 2X qPCR MasterMix Kit (abm, Vancouver, Canada) in a Quanstudio™ 5 Real-Time PCR Intruments (Thermo Fisher Scientific, Inc.). First-strand cDNA was synthesized using the RevertAid™First strand cDNA Synthesis Kit (TransGen Biotech, Beijing, China). DEGs primers were designed using the Primer-Blast (/" https:// www.ncbi.nlm.nih.gov/tools/primer-blast/) and synthesized commercially (Shuoqing, Kunming, China). Actin were selected as reference genes [121]. The primers used in qRT-PCR analyses are listed in Table S1. Amplification reaction mixtures were made of 10 μL of Eva Green 2X qPCR Master Mix, 0.5 μL of each forward and reverse primer (10 mM), and 1 μL of cDNA template, and ddH 2 O was added to a final volume of 20 μL. The amplification cycling program was as follows: enzyme activation was operated at 95°C for 10 mins, moreover, 40 cycles of 95°C for 15 s, 58°C for 30 s and 72°C for 30 s. The results were analyzed using the software accompanying the Quanstudio™ 5 Real-Time PCR instruments. The relative expression values were obtained by using the 2 -ΔΔCt method [122].