Transcriptome analysis for Fraxinus mandshurica seedlings from different carbon xation and growth provenances in response to nitrogen deciency

Background The carbon xation characteristic of F. mandshurica seedlings from 20 provenances has been evaluated by our research group. In order to explore whether growth and carbon sequestration are related to nutritional adaptability, the foliage and roots of the annual Wuchang (WC) seedlings with high carbon content and Hailin (HL) seedlings with low carbon content which were grown in nitrogen-decient nutrition and total nutrition were used for RNA-seq determination. Results Eight transcriptome libraries by high-throughput sequence were analysis. 1,235,174,984 clean reads and 88,655 unigenes with N50 length of 1,259 bp were obtained. Under normal nitrogen condition, 783 differentially expressed genes (DEGs) between WC and HL were identied, the number in foliage (669) of DEGs between HL and WC was more than that in roots (149). The number of transcription factors (TFs), hormone, and Protein kinase (PK) genes was signicantly more in WC than that in HL. Compared to the normal nitrogen, 8173 DEGs related to nitrogen deciency were identied and the number of DEGs in roots (6999) was more than that in foliage (1616). Several nitrogen deciency-related metabolic pathways and many DEGs involved in nitrogen absorption and assimilation, carbon metabolism, hormones, transcription factors and kinases were identied. The numbers of DEGs encoding nitrate transporters, PK, TFs and hormone in WC were less than that in HL, which indicated the response of HL provenance seedlings to nitrogen deciency was stronger than that of WC provenance seedlings. However, under nitrogen deciency, the number of up-regulated DEGs in the WC provenance seedlings is more than that in the HL, which indicated that the WC provenance seedlings were more tolerant to nitrogen deciency than HL. Conclusions The data showed that high carbon content and high growth rate of WC provenance seedlings are mainly attributed to the high transcriptional expression of many metabolic genes in foliage. The response of HL provenance seedlings to nitrogen deciency is signicantly greater than that of WC provenances seedlings, but WC provenance seedlings were more

groups. Especially, the carbon content and carbon xation quality of the Wuchang (WC) provenance seedlings is signi cantly higher than that of the Hailin (HL). In this study, in order to explore the mechanism of high carbon xed provenance formation, especially the response mechanism under nitrogen de ciency stress, the annual WC seedlings with high carbon content and HL seedlings with low carbon content were selected for RNA-seq determination. What's more, to identify genes associated with nitrogen de ciency and explore the differences in response to nitrogen de ciency, the two provenances seedlings treated with nitrogen de ciency were also used for RNA-seq determination.

RNA-seq and de novo transcriptome assembly
To obtain an overall view of the nitrogen de ciency transcriptome in the two provenances, RNA samples were prepared from the roots and foliage of both provenances on 30 days after nitrogen de ciency treatment. RIN values indicate that RNA quality can be used for cDNA library construction (Additional le 2). Gene expression pro les of the high carbon xation Fraxinus mandshurica provenance (WC) and low carbon xation Fraxinus mandshurica provenance(HL) seedlings roots and foliage under both normal nitrogen and nitrogen de ciency conditions were analyzed. For the eight samples, three biological replicates were performed in sequencing. In total, 1,235,174,984 clean reads originated from the 24 Fraxinus mandshurica cDNA libraries were obtained following removal of adaptors and low-quality reads (Additional le 3). The percent G+ C and fraction of bases with quality scores of Q30 for the 24 libraries averaged 45.2% and 94.2%, respectively. These numbers showed that the data was of high quality. Trinity software was then used to assemble the clean reads into 88,655 unigenes with N50 length of 1259 bp. The relative length of the assembled sequences is one standard by which a transcriptome assembly can be assessed and the summary statistics are shown in Fig.1. In the nal assemblies there were 48,754 unigenes ranging from 200 to 500 bp in length, which accounted for 55% of the total. There were 19,220 (21.68%) and 20,681 (23.33%) unigenes from 500 to 1000 bp and > 1000 bp in length, respectively, and assembled sequences < 200 bp were not taken into account.

Annotation and functional classi cation of unigenes
The assembled sequences were used as queries in BLAST searches (E ≤ 10 −5 ) against the NR, KOG, Swissprot and KEGG databases, and a total of 55,537 were annotated (Fig. 2a). The majority of the unigenes (53,603; 60.5%) were annotated from the NR database. In contrast, only 24,959 (28.2%), 35,879 (40.5%) and 42,650 (48.1%) annotated unigenes were matched to the KEGG, KOG and Swissprot database, respectively. As shown in Fig.2a, a total of 21,084 unigenes were annotated with all four databases. To further utilize the transcriptome data, the KOG, GO, and KEGG databases were used to identify the molecular processes that occur during nitrogen de ciency of Fraxinus mandshurica seedlings roots and folium. The annotated unigenes were initially classi ed into 25 KOG categories with the largest group being "General function prediction only" (10,345; 28.83%) (Fig.2b). The Gene Ontology (GO), which is an internationally standardized classi cation system, was used to assign 15,827 unigenes to the three principal GO domains: "cellular component", "molecular function" and "biological process" (Fig.2c). In the "cellular component" category, "cell" and "cell part" were the most abundant terms. For the category of "molecular function", "catalytic activity" and "binding" were the most prominent. "Metabolic process" and "cellular process" were the most abundant terms in the "biological process" category.

Identi cation of differentially expressed genes (DEGs)
The transcriptional levels were normalized using the FPKM method. Meanwhile, FDR<0.05 was used as screening thresholds to test the signi cance of difference in transcript abundance. More down-regulated DEGs were found in CP1 and CP8, which indicated that the numbers of DEGs up-regulated in WC were 34.74% and 204.55% greater than that in HL under normal nitrogen for foliage and nitrogen de ciency for roots. In order to study the difference in DEGs between two provenances, the DEGs in CP1, CP2 and CP7, CP8 were further analyzed together. Consequently, 1057 DEGs were identi ed using pair-wise comparison of each accession between WC and HL under normal condition and nitrogen de ciency (Additional le 4). 783 DEGs were identi ed between WC and HL under normal condition, including both up-regulated (345) and down-regulated (438) genes ( g.3a and g.3b). Interestingly, DEGs in the CP1 (669) were nearly 4.5 times as much as those in the CP2 (149) (table1), which indicated that the number of DEGs in foliage was signi cantly higher than that in roots under normal nitrogen. Similarly, 300 DEGs were identi ed between WC and HL under nitrogen de ciency, including both up-regulated (102) and down-regulated (198) genes ( g.3a and g.3b). And downregulated DEGs in the CP8 (134) were nearly twice as much as those in the CP7 (74) (table1), which indicated that the number of DEGs in roots was signi cantly higher than that in foliage under nitrogen de ciency.
Moreover, analysis of DEGs in CP3, CP4 and CP5, CP6 can be used to identify genes that respond to nitrogen de ciency in WC and HL, and then the differences in response to nitrogen de ciency between the two provenances can be further analyzed. Consequently, 8173 DEGs under nitrogen de ciency were identi ed using pair-wise comparison of each accession between normal and nitrogen de ciency in WC and HL (Additional le 4: Figure S2). However, there were 1959 DEGs commonly found in both HL and WC. 4447 DEGs were identi ed between normal nitrogen and nitrogen de ciency in WC, including both up-regulated (2019) and down-regulated (2431) genes ( g.3c and g.3d). Interestingly, DEGs in the CP4 (4225) were nearly 14 times as much as those in the CP3 (305), which indicated that the number of DEGs in roots was signi cantly higher than that in foliage in WC. Similarly, 5685 DEGs were identi ed between normal and nitrogen de ciency in HL, including both up-regulated (3210) and down-regulated (2518) genes ( g.3c and g.3d). And DEGs in the CP5 (4531) were nearly 3.2 times as much as those in the CP6 (1414), which indicated that the number of DEGs in roots was signi cantly higher than that in foliage in HL. In addition, the transcriptional expression of DEGs in the two provenances foliage was signi cantly different in response to nitrogen de ciency. The two provenances displayed dissimilar expression patterns, in which the mount of total (1414), up-regulated (991) and down-regulated (423) DEGs in the CP6 were 4.6 times, 6.0 times and 3.0 times as much as those (305,165 and 140) in the CP3(table1), respectively, which indicated that the foliage of HL provenance responded to nitrogen de ciency more strongly than that of WC provenance.
To validate the gene expression results derived from the transcriptome data, nine DEGs related to transcription factors in CP1 and six DEGs involved in the phenylpropanoid pathway for the biosynthesis of lignin in CP4 and CP5 were selected for qRT-PCR. The results of experiment showed a strong correlation with the RNA-Seq data (Fig.4), which indicates that the results of the transcriptomic analysis are reliable.

GO annotation for DEGs
Gene annotation of DEGs was using GO database. Annotated genes were grouped into three major functional categories: cellular component, molecular function and biological process, and then were divided into subcategories ( Fig.5; Additional le 5). For ease of presentation, data was expressed as the average of genes in each subcategory in WCKF vs. the average in HCKF. Within the category of biological process in WCKF vs. HCKF (CP1), genes involved in single-organism process, cellular process, metabolic process, localization, biological regulation, response to stimulus and developmental process were 11.32%,11.80%,13.24%, 3.03%, 3.03%, 2.87% and 1.28% of the total, respectively. In the cellular component category, 5.10%, 4.47%, 3.83%, 3.83%, 1.28% and 1.12% genes were related to cell, membrane, organelle, membrane part, organelle part and cell junction, respectively. In the molecular function category, 14.99%, 9.73% and 1.75% genes were involved in catalytic activity, binding and transporter activity, respectively (Fig.5). Annotation for DEGs in CP2, CP3, CP4, CP5, CP6, CP7 and CP8 pairwise comparisons was also carried out. In all three functional categories, genes belonging to the above three groupings showed a similar distribution pattern to those of CP1. However, macromolecular complex showed higher levels in CP3, CP4, CP5 and CP6, which suggested that macromolecular complex might play an important role in responding to nitrogen de ciency ( Fig.5; Additional le 5).

KEGG pathway analysis for DEGs
Transcriptome sequencing can also be taken to identify a variety of metabolic processes. In this study, a total of 1057 DEGs were con rmed in CP1, CP2, CP7 and CP8 and 8173 DEGs between normal and nitrogen de ciency were identi ed in CP3, CP4, CP5 and CP6. After identifying DEGs, KEGG annotation was used for DEGs from the above comparisons. In CP1, CP2, CP3, CP4, CP5, CP6, CP7 and CP8,126, 19, 63, 1298, 1165, 232, 21 and 30 DEGs were assigned to 73, 24, 47, 111, 117, 90, 20 and 30 pathways of the KEGG database, respectively (Additional le 6). The most reliable signi cantly (p<0.05) enriched pathways of DEGs were represented as Additional le 3. Integrate CP3, CP4, CP5 and CP6 comparisons, the major pathways involved in the mechanism for responding to nitrogen de ciency were "Phenylpropanoid biosynthesis", "Nitrogen metabolism", "Diterpenoid biosynthesis", "Isoquinoline alkaloid biosynthesis". In addition, in the roots that responded most to nitrogen de ciency, "Ribosome" pathway had the highest number of DEGs in both provenances. "Oxidative phosphorylation" and "Starch and sucrose metabolism" pathways were speci cally enriched in the roots of WC and HL, respectively. Similarly because only 2 DEGs in CP2 are annotated by KEGG, we used the DEGs in CP1 to analyze the difference between two provenances. In CP1, there were 18 pathways which could be divided into eight types involving in the growth mechanism: "Carbohydrate metabolism", "Energy metabolism", "Lipid metabolism", "Amino acid metabolism", "Metabolism of other amino acids", "Metabolism of terpenoids and polyketides", "Biosynthesis of other secondary metabolites" and "Signal transduction" (Additional le 6).

DEGs involved in carbon and energy metabolism
The expression levels of many genes involved in carbohydrate metabolism, lipid metabolism and energy metabolism were differentially between HL and WC. For example, six genes involved in "starch and sucrose metabolism" pathway in CP1 showed increased transcript abundance in WC (Additional le 6, Additional le8). These genes encoded polygalacturonase (Unigene0056877), beta-glucosidase (Unigene0064953, Unigene0055864, Unigene0054701, Unigene0040728) and trehalose-phosphate phosphatase J(Unigene0037044) were identi ed and their expressions were increased by 2.3 to 42-fold.
Five DEGs involved in "Nitrogen metabolism" pathway in CP1 showed increased transcript abundance in WC (Additional le 6). These genes encoded carbonic anhydrase (Unigene0063117, Unigene0054409, Unigene0063118), glutamine synthetase (Unigene0053381), high a nity nitrate transporter 2.5 (Unigene0050395) were identi ed and their expressions were increased by 3.4 to 16.7-fold. Nine DEGs involved in Lipid metabolism in CP1 showed increased transcript abundance in WC (Additional le 6).

DEGs between normal nitrogen and nitrogen de ciency in Fraxinus mandshurica
In order to study the two Fraxinus mandshurica provenances transcriptional expression difference in response to nitrogen de ciency, DEGs between normal condition and nitrogen de ciency from roots and foliage of WC and HL were obtained, respectively. KEGG enrichment analysis showed that the transcriptional expression of foliage from two provenances was signi cantly different in response to nitrogen de ciency. For example, the number of KEGG pathways which were signi cantly enriched was 8 in WC foliage (CP3), less than that (14) in HL foliage (CP6). The pathways with a larger number of genes enriched in WC foliage (CP3) were "Sulfur metabolism" and "Nitrogen metabolism", however, that in HL foliage (CP6) were "Plant-pathogen interaction", "Plant hormone signal transduction", "Protein processing in endoplasmic reticulum", "Phenylpropanoid biosynthesis" and "Starch and sucrose metabolism" (Fig6d). This indicated that the response patterns of WC and HL to nitrogen de ciency were different in the foliage. Similarly, the DEGs of the roots were analyzed. For example, the number of KEGG pathways which were signi cantly enriched was 8 in WC roots (CP4), less than that (17) in HL roots (CP5). The pathways with a larger number of genes enriched in WC roots (CP4) were "Ribosome", "Oxidative phosphorylation", "Phenylpropanoid biosynthesis", "Cyanoamino acid metabolism" and "Diterpenoid biosynthesis", however, that in HL roots (CP5) were "Ribosome", "Starch and sucrose metabolism", "Oxidative phosphorylation" , "Plant hormone signal transduction", "Pentose and glucuronate interconversions", "Tyrosine metabolism" and "Phenylpropanoid biosynthesis" (Fig.6c). This indicated the response patterns of WC and HL to nitrogen de ciency were also different in the roots.
DEGs related to hormone signaling responding to nitrogen de ciency In addition to the basic roles in growth and development, phytohormones are also involved in various environmental responses, such as light, salt and drought. It has been proposed that some hormones coordinate demand and acquisition of nitrogen. In this study, the number of up-regulated DEGs related to hormone signaling was greater than that of down-regulated DEGs in respond to nitrogen de ciency. Totally 0 (0 up-and 0 down-regulated), 22 (17 and 5), 40 (34 and 6) and 15 (14 and 1) DEGs were identi ed in CP3, CP4, CP5 and CP6 (Additional le 7). Obviously, the number of up-regulated DEGs related to hormone signaling in CP5 (CP6) was greater than that of CP4 (CP3), which indicated the response to nitrogen de ciency, HL is stronger than WC. Among them, auxin, ABA, SA, JA and CK pathways were predominatly induced. In the auxin signaling pathway, there were 8, 19 and 2 genes upregulated in CP4, CP5 and CP6; in the SA pathway, the number of up-regulated genes was 5, 5 and 1; in the ABA pathway, the number of up-regulated genes was 0, 5 and 3; in the JA pathway, the number of upregulated genes was 0, 1 and 7; in the CK pathway, the number of up-regulated genes was 2, 5 and 0 (Additional le 7). In addition, the response patterns of DEGs related to hormones under nitrogen de ciency in roots and foliage were different. For example, the hormones that predominantly played a role in the foliage were JA and ABA but in the roots were auxin, SA, CK and ABA.
DEGs related to nitrogen assimilation responding to nitrogen de ciency Many genes involved in nitrogen absorption and assimilation were differentially expressed under nitrogen de ciency relative to the normal condition. In the current study, 10 DEGs encoding nitrate transporters were detected (Additional le 12, Additional le 8). Compared to normal nitrogen, 7 DEGs were upregulated and 3 DEGs were down-regulated under nitrogen de ciency. Whereas, the number of these DEGs in WC was less than that in HL, and there were 2 up-regulated DEGs (Unigene0034112 and Unigene0050395) which encoded NRT3.2 and NRT2.5 and 2 down-regulated DEGs (Unigene0046155 and Unigene0047301) which encode NRT2.7 and NRT in both WC and HL. In addition, four DEGs (Unigene0018686, Unigene0022313, Unigene0026134, Unigene0050965) which encoded NRT3.1 and NRT2.1 were up-regulated and one DEG (Unigene0039451) which encoded NRT3.1 was down-regulated only in HL, and one DEG (Unigene0071953) which encoded NRT was up-regulated in WC, respectively. Four nitrogen de ciency responsive genes encoding ammonium transporters (Unigene0034277, Unigene0045039, Unigene0045040, Unigene0054529) were identi ed (Additional le 12), in which, Unigene0054529 was responsive to nitrogen de ciency only in HL. Moreover, fourteen, six and four DEGs encoding amino acid, lysine histidine and oligopeptide transporters were found (Additional le 12), respectively. There were also three, six, two and two DEGs encoding glutamine synthetase (GS), glutamate dehydrogenase (GDH), nitrite reductases (NiR) and nitrate reductase (NR) which were the key enzymes in nitrate assimilation and their transcription expression were all down-regulated except two GS unigenes (Unigene0009471, Unigene0053381).
In this study, there were also one, one, three and one DEGs encoding Dof zinc nger protein DOF1.5-like (CDF4), Pyruvate kinase (PYK1), Ribulose-1,5-bisphosphate carboxylase (RuBP) and neutral invertase A (INVA) which were the key enzymes in carbon assimilation and their transcription expression were all down-regulated except one INVA unigene (Additional le 12). Meanwhile, expression of many genes associated with absorption or translocation of other nutrients changed under nitrogen de ciency, such as phosphate (9), potassium (3), sulfate (5), Zinc (4), iron (3) and molybdate transporter (2), indicating that uptake of these nutrients in Fraxinus mandshurica is affected by nitrogen metabolism under cross-talking regulation.

DEGs between WC and HL under normal nitrogen
Under normal conditions, compared with HL provenance, there were 669 DEGs in the foliage of WC provenance, of which 285 were down-regulated and 384 were up-regulated; 149 DEGs were in roots, of which 77 were down-regulated, and 72 were up-regulated. It indicated that the difference in growth rate of the two provenances under normal nitrogen is mainly caused by the DEGs in the foliage. Through comparative analysis, it was found that the number of up-regulated genes in WC provenance foliage was more than that in HL. Through KEGG annotation analysis, it was found that the DEGs were signi cantly enriched in 8 metabolic processes such as biosynthesis of other secondary metabolites, signal transduction, metabolism of terpenoids and polyketides, carbohydrate metabolism, energy metabolism, lipid metabolism, amino acid metabolism and metabolism of other amino acids, that is, 17 metabolic pathways (Additional le 6: CP1). Further analysis of DEGs on these signi cant enrichment pathways revealed that transcription factor, hormone, and kinase-related genes were signi cantly more in WC provenance than in HL, but among the DEGs of carbon metabolism, nitrogen metabolism and lipid metabolism, only the number of lipid-related genes in WC provenance was signi cantly higher than that in HL. Five of the nine lipid metabolism-related differentially expressed genes are in the alpha-Linolenic acid metabolism pathway, and the transcription level of four jasmonic acid synthesis-related genes (4CL5, lipoxygenase 2.1, 3-ketoacyl-CoA thiolase 2, SAMT) was signi cantly higher in WC than that in HL. Correspondingly, the JAZ protein transcription level in the jasmonic acid signaling pathway was higher in the WC than that in HL (TIFY10A, TIFY10B, TIFY11B, TIFY3B).
Studies have shown that in addition to MYC2 transcription factors [ 25,26], JAZ proteins can bind directly or indirectly to transcription factors such as BHLH [ 27], MYB [ 28], AP2/ERF [ 29] and WRKY [ 30]. In Fraxinus mandshurica, there are three BHLH, three ERF, three WRKY transcription factors whose expression were consistent with JAZ expression. In addition to the regulation of transcription factors and hormones on the plants, the kinase also has a cascade of ampli cation functions in the growth and signaling process of the plant. In the signi cantly enriched KEGG metabolic pathway, twelve kinases whose transcription level in the WC provenance is signi cantly higher than that in HL provenance. Six of the twelve kinases belong to the serine/threonine protein kinase, two are leucine-rich protein kinases, and two are receptor-associated protein kinases on the cell wall.
Since growth and carbon storage of WC provenance seedlings are higher than that of HL provenance seedlings, we wanted to nd genes with high expression of carbon metabolism-related genes in WC provenance seedlings. Three genes related to carbon metabolism, INVA, β-glucosidase and trehalose phosphate phosphatase J (TPPJ) were found. Invertase, was also known as sucrose invertase. The sucrose is broken down into fructose and glucose by invertase. Studies have shown that highly active invertases in plants are associated with rapid growth of tissues, such as in seedlings, young foliage, radicles and young fruits. The analysis showed that the transcription expression of sucrose invertase from WC provenance was signi cantly higher than that of HL provenance, indicating that the utilization rate of sucrose in WC was higher. Studies have shown that trehalose 6-phosphate (T6P) is a signal substance that can re ect the adequacy of carbon sources. In this study, it was found that the transcription expression level of T6P phosphorylase TPPJ in WC provenance was 5 times that of HL provenance, but there was no signi cant difference in the transcription expression of T6P metabolic enzyme TPS, indicating that trehalose content in WC provenance seedlings may be higher than HL provenance. It is speculated that the carbon use e ciency of WC provenance was higher than that of HL provenance.
In summary, plants produce a large amount of carbohydrates through photosynthesis. Sucrose, is the main form of carbon transport, whose content increase causes plants to produce higher levels of T6P. In the WC provenance, on the one hand, the high concentration of sucrose is decomposed by invertase, which is fully used for plant growth and development; on the other hand, the content of T6P is decreased to regulate growth by increasing the transcriptional expression of TPPJ. Due to the low utilization rate of sucrose, coupled with the inhibition of carbon source utilization by high concentration of T6P, the growth of HL provenance seedlings is slower than that of WC seedlings.
Identi cation of DEGs related to nitrogen de ciency DEGs related to nitrogen de ciency have been identi ed in many species, such as duckweed [ 31], Magnaporthe oryzae [ 32], rice (Oryza sativa L.) [ 33], Tibetan wild barley [ 34], Maize [ 35], soybean [ 24], Arabidopsis [ 23,36], etc. However, little is known about the DEGs related to nitrogen de ciency in Fraxinus mandshurica. In current study, analysis of DEGs in CP3, CP4 and CP5, CP6 can be used to identify genes that respond to nitrogen de ciency in Fraxinus mandshurica, and then the differences in response to nitrogen de ciency between two provenances can be further analyzed. Consequently, 8173 DEGs under nitrogen de ciency were identi ed using pair-wise comparison of each accession between normal and nitrogen de ciency in WC and HL. Different parts of plants respond differently to nutrient stress and the number of DEGs in roots was signi cantly higher than that in foliage under nitrogen de ciency. For example, DEGs in the CP4 (4225) were nearly 14 times as much as that in the CP3 (305) and DEGs in the CP5 (4531) were nearly 3.2 times as much as that in the CP6 (1414).
Many genes involved in nitrogen absorption and assimilation were differentially expressed under nitrogen de ciency relative to the normal condition. In plants, nitrogen is rst actively absorbed by nitrate transporters in roots. In the current study, 10 DEGs encoding nitrate transporters were detected (Additional le 12), such as Unigene0034112, Unigene0050395, Unigene0046155, Unigene0047301, Unigene0018686, Unigene0022313, Unigene0026134, Unigene0050965, Unigene0039451 and Unigene0071953. Four, three, six, two and two DEGs encoding AMT, GS, GDH, NiR and NR were identi ed (Additional le 12), which were key enzymes in nitrogen assimilation. The transcription expression of genes encoding GS, NR and GOGAT was also down-regulated under nitrogen starvation in duckweed [ 31]. What's more, macromolecular complex showed higher levels in CP3, CP4, CP5 and CP6, which suggested that macromolecular complex might play an important role in responding to nitrogen de ciency.
Nitrogen de ciency also has an important impact on carbon metabolism [ 31,37] [ 38]. In the current study, the DEGs in the phenylpropanoid pathway for the biosynthesis of lignin (Fig9) were also signi cantly up-regulated such as the 4CL gene and lignin content is increased but the increase is not signi cant in roots under nitrogen de ciency (Additional le 13: Figure S7). Suppression of 4CL gene expression exerted the biggest impact on lignin production of all of the genetic manipulations of phenylpropanoid related genes in conifers [ 39] and 4CL RNAi lines exhibited a reduced lignin content of approximately 50% [ 40]. The lignin content of T 0 and T 1 generation plants over-expressed antisense 4CL gene was also reduced by 45.77% and 31.97% in Alfalfa (Medicago truncatula), respectively [ 41].
Induction of lignin biosynthesis is an adaptive response of plants subjected to many abiotic stresses [ 42].
What's more, in maize shoots, N luxury signi cantly reduces lignin total content as well as the generation of S, H and G monomers of the lignin [ 43]. Transcriptional and physiological analyses identify a regulatory role for hydrogen peroxide in the lignin biosynthesis of copper-stressed rice roots [ 44].
Peroxidases are the main enzymes which are involved in the process of lignin biosynthesis [ 45]. What's more, gibberellin (GA) is a positive regulator of ligni cation [ 38]. The key genes for lignin synthesis, 4CL and CAD1, are signi cantly induced by GA, which in turn changes the composition of the cell wall [ 46]. In this study, the DEGs in the terpenoids and polyketides metabolism pathway for the biosynthesis of abscisic acid (ABA) and gibberellin A4 (GA4) (Fig8) were also signi cantly up-regulated in roots under nitrogen de ciency. Compared with the control, the endogenous ABA content and GA content were signi cantly reduced and increased in the Fraxinus mandshurica roots by the nitrogen de ciency, respectively (Fig10). Meanwhile, transcription expression of many genes associated with absorption or translocation of other nutrients changed under nitrogen de ciency, such as phosphate (9), potassium (3), sulfate (5), Zinc (4), iron (3) and molybdate transporter (2), indicating that uptake of these nutrients in Fraxinus mandshurica is affected by nitrogen metabolism under cross-talking regulation.
Hormones, transcription factors and kinases are the three major regulators in plants. Studies have shown that mitogen-activated protein kinase kinase 9 plays an important role in plant nitrogen tolerance [ 47].
Protein phosphorylation may be induced by nitrogen de ciency in Fraxinus mandshurica. For example, a total of 14, 92, 132 and 152 DEGs encoding protein kinases were found in CP3, CP4, CP5 and CP6, respectively (Additional le 10). Transcriptional expression of nitrogen metabolism genes may be activated by transcription factors under nitrogen de ciency [ 21]. The results indicated that in the response to nitrogen de ciency, bHLH, MYB and AP2/ERF play a major role in roots, and WRKY plays a major role in foliage. The hormones homeostasis in uences plant growth by regulating nutrients metabolism, such as proteins, nucleic acids and soluble carbohydrates [48][49][50]. Nitrogen de ciency caused a signi cant decrease in GA and IAA content but increase in ABA content in rice foliage [ 51]. In this study, Hormone (auxin, ABA, SA, JA and CK) pathways were predominatly induced by nitrogen de ciency, especially ABA and GA pathways (Fig.8). Compared to control, endogenous ABA content was reduced by nitrogen de ciency, but endogenous ZR, GA and IAA content was increased. Therefore, Fraxinus mandshurica is different from other species in response to nitrogen de ciency in endogenous hormones, and the reason for this difference needs further research. In addition, the response patterns of DEGs related to hormones under nitrogen de ciency in roots and foliage were different. For example, the hormones that predominantly played a role in the foliage were JA and ABA but in the roots were auxin, SA, CK and ABA.
Several nitrogen de ciency-related metabolic pathways were also identi ed by KEGG analysis [ 31]. Integrate CP3, CP4, CP5 and CP6 comparisons, the major pathways involved in the mechanism for responding to nitrogen de ciency were "Phenylpropanoid biosynthesis", "Nitrogen metabolism", "Diterpenoid biosynthesis", "Isoquinoline alkaloid biosynthesis". In addition, the pathway which had the most DEGs in the roots that responded most to nitrogen de ciency in both provenances was the "Ribosome" pathway. Studies on Magnaporthe oryzae proteomics have shown that nitrogen de ciency induces the synthesis of many extracellular proteins, and protein degradation and translation have also undergone extensive changes [ 32].
The transcriptional expression difference of two provenances in response to nitrogen de ciency The number of DEGs of HL provenance seedlings was less under nitrogen stress (CP8) but signi cantly more in response to nitrogen de ciency than that of WC provenance seedlings, respectively. For example, the foliage of HL provenance responded to nitrogen de ciency more strongly than that of WC provenance. The numbers of KEGG pathways which were signi cantly enriched both in WC foliage (CP3) and roots (CP4) were less than that both in HL foliage (CP6) and roots (CP5), respectively. KEGG enrichment analysis showed that the response patterns of WC and HL to nitrogen de ciency were different both in the foliage and roots. The pathways with a larger number of genes enriched in WC foliage (CP3) were "Sulfur metabolism" and "Nitrogen metabolism", however, that in HL foliage (CP6) were "Plant-pathogen interaction", "Plant hormone signal transduction", "Protein processing in endoplasmic reticulum", "Phenylpropanoid biosynthesis" and "Starch and sucrose metabolism" (Fig6d).
Similarly, the pathways with a larger number of genes enriched in WC roots (CP4) were "Ribosome", "Oxidative phosphorylation", "Phenylpropanoid biosynthesis", "Cyanoamino acid metabolism" and "Diterpenoid biosynthesis", however, that in HL roots (CP5) were "Ribosome", "Starch and sucrose metabolism", "Oxidative phosphorylation", "Plant hormone signal transduction", "Pentose and glucuronate interconversions", "Tyrosine metabolism" and "Phenylpropanoid biosynthesis" (Fig.6c), which indicated that "Oxidative phosphorylation" and "Starch and sucrose metabolism" pathways were speci cally enriched in the roots of WC and HL, respectively. The numbers of DEGs encoding NRTs, PK, TFs and hormone in HL were more than that in WC. The higher expression of NRTs could be attributed to its more nitrogen uptake and higher low nitrogen tolerance [ 34]. Studies have shown that trehalose can increase plant growth by increasing the expression of nitrogen metabolism genes under nitrogen de ciency conditions [ 52]. In plants, UDP-glucose and glucose-6-phosphate are catalyzed by trehalose 6phosphate synthases (TPSs) to synthesize T6P, which is catalyzed by trehalose 6-phosphate phosphatases (TPPs) to form trehalose [ 53,54]. In this study, one and four TPS genes transcription expression was up-regulated and down-regulated signi cantly under nitrogen de ciency in HL provenance seedlings roots, respectively. The endogenous ZR content in HL foliage was signi cantly higher than that in WC foliage under nitrogen de ciency (Fig10). The root lignin content in HL was higher than that in WC under nitrogen de ciency but not signi cant (Additional le 13: Figure S7). In summary, the response of HL provenance seedlings to nitrogen de ciency was stronger than that of WC provenance seedlings. However, under nitrogen de ciency, the number of up-regulated DEGs in the WC provenance seedlings is more than that in the HL (CP7 and CP8), especially in the roots (CP8), which indicated that the WC provenance seedlings were more tolerant to nitrogen de ciency than HL.

Conclusion
The data showed that high carbon content and high growth rate of WC provenance seedlings are mainly attributed to the high transcriptional expression of many metabolic genes in foliage under normal condition. The response of HL provenance seedlings to nitrogen de ciency is signi cantly greater than that of WC provenances seedlings, but the WC provenance seedlings were more tolerant to nitrogen de ciency than HL. Many genes related to nitrogen de ciency were identi ed, which will expand our current understanding of nitrogen responses. containing 0 (nitrogen de ciency) or 1000 (normal nitrogen) μM NH 4 NO 3 , respectively, and the nutrient solution was adjusted to pH 5.5. The LA solution was added every 2 days with 1000ml until August 20.

Methods
The roots and foliage samples used for cDNA library construction were collected on 30 days after treatment, with three biological replicates per group. All samples were frozen in liquid nitrogen immediately upon collection and stored at -80°C.
cDNA library construction and sequencing De novo cDNA assembly and functional annotation To obtain clean reads, sequencing adaptors and low quality reads were removed from each library. Transcriptome de novo assembly was carried out with short reads assembling program -Trinity. Trinity is a modular method and software package which combines three components: Inchworm, Chrysalis and Butter y. Firstly, Inchworm assembles reads by a greedy k-mer based approach, resulting in a collection of linear contigs. Next, Chrysalis clusters related contigs that correspond to portions of alternatively spliced transcripts or otherwise unique portions of paralogous genes, and then builds a de Bruijn graphs for each cluster of related contigs. Finally, Butter y analyzes the paths taken by reads and read pairings in the context of the corresponding de Bruijn graph, and outputs one linear sequence for each alternatively spliced isoform and transcripts derived from paralogous genes. And the transcriptome feference database was obtained. All raw read data were deposited in the Genome Sequence Archive with the project ID. The unigene expression was calculated and normalized to RPKM (Reads Per kb per Million reads).

Determination of phytohormone and lignin content
From each treatment, the fully expanded leaves were selected for the indoleacetic acid (IAA), gibberellin (GA), zeatinriboside (ZR) and abscisic acid (ABA) analyses. For the method of extracting and measuring phytohormone content, please refer to He [ 55]. The method of extracting and measuring lignin content is the same as phytohormone. Declarations Figure 1 Length distribution of Fraxinus mandschurica unigenes    The relative expression levels of (a)nine DEGs identi ed in the comparison CP1 and (b)six DEGs identi ed in the comparison CP4 and CP5 between RNA-Seq and qRT-PCR. The genes relative expression levels were determined by 2-ΔΔCT as expressed, and were normalized to the expression level of TU.    PDS, 15-cis-phytoene desaturase; CYP97A3, beta-ring hydroxylase; ABA2, zeaxanthin epoxidase; NCED, 9cis-epoxycarotenoid dioxygenase; AAO3, abscisic-aldehyde oxidase.  The content of endogenous ABA, IAA, ZR and GA. Each data point represents the average of three replicates, and nine seedlings were used for each experiment.