Combining Targeted Metabolite Analyses and Transcriptomics to Reveal Specic Chemical Composition and Associated Genes in Resistant Soybean Infected with Soybean Cyst Nematode

Background: Soybean cyst nematode Heterodera glycines is one of the most devastating pathogens on soybean and causes severe annual yield loss worldwide. Different soybean varieties exhibit different responses to soybean cyst nematode infection at various levels, such as genomic, transcriptional, proteomic, and metabolomic levels. However, there were not yet any reports on the differential responses of resistant and susceptible soybeans infected with soybean cyst nematode by combining the metabolomic analyses and transcriptomics. Results: In this study, a highly-resistant variety PI 437654 and three susceptible varieties Williams 82, Zhonghuang 13 and Hefeng 47 were used as the test materials to clarify the differences in metabolites and transcriptomics between resistant and susceptible soybeans before and after SCN infection. A local metabolite-calibrated database was used to identify potential differential metabolites, and the differences of metabolites and metabolic pathways were compared between the resistant and susceptible soybean varieties after inoculation with SCN. Totally, 37 differential metabolites and 20 KEGG metabolic pathways were identied, which were divided into three categories including the metabolites/pathways overlapped among resistant and susceptible soybeans, specic in susceptible or resistant soybeans, respectively. Twelve differential metabolites were found to be involved in predicted KEGG metabolite pathways. Moreover, 14 specically differential metabolites such as signicantly up-regulated nicotine and down-regulated D-aspartic acid, and their involved KEGG pathways such as tropane, piperidine and pyridine alkaloid biosynthesis, alanine, aspartate and glutamate metabolisms, sphingolipid metabolism and arginine biosynthesis were signicantly changed and abundantly enriched in the resistant soybean, and likely played pivotal roles in defensing against SCN infection. Three key metabolites including N-acetyltranexamic acid, nicotine, and D, L-typotophan, which were found to be signicantly up-regulated in the resistant soybean PI437654 infected by SCN, classied into two types and used for combination analyses with the transcriptomic expression proling. Associated genes were predicted, and suggested their likely biological processes, cellular component, molecular function and involved pathways. Conclusions: Our results not only found out the potential novel metabolites and associated genes in the resistant response of soybean to soybean cyst nematode, but also provided new insights into the interactions between soybean and soybean cyst nematode. differential metabolites such as signicantly up-regulated nicotine and down-regulated D-aspartic acid, and their involved KEGG pathways such as tropane, piperidine and pyridine alkaloid biosynthesis, alanine, aspartate and glutamate metabolism, sphingolipid metabolism and arginine biosynthesis were signicantly changed and abundantly enriched specically in the resistant soybean PI 437654, and likely played pivotal roles in defensing SCN. Moreover, combination of the metabolomic analyses with the transcriptomic expression proling aided to predict potential associated genes involved in the most signicantly up-regulated metabolites of the resistant soybean variety PI437654, and suggest their likely biological processes, cellular component, molecular function and involved pathways. To our best our study was the rst to investigate the different responses of resistant and susceptible soybeans infected with SCN by combining metabolomic analyses and transcriptomics. Our results not only found out the potential novel metabolites and associated genes in the resistant response of soybean to soybean cyst nematode, but also provided new insights into the interactions between soybean and soybean cyst nematode.

system, and cause plant damages. Meanwhile, soybean roots produce many defense-related compounds such as reactive oxygen species and activate hormone signaling pathways in response to nematode infection [27,31], and the metabolites undergo tremendous changes in the soybean roots upon infection of SCN.
Recently, the use of "omics" technology to obtain biological data has been surged, the most commonly used omics are genomics, transcriptomics, proteomics, and metabolomics [53]. There are numerous research reports about soybean-SCN interactions by genomic and transcriptomic analyses. The genomes of SCN and many soybean varieties were uncovered as the solid foundations to further investigate their biology, characteristics and their potential interactions [38,43,61]. Many soybean resistant candidate loci and/or genes were identi ed based on the genome-wide association study (GWAS) [24,59,64]. Moreover, many studies on the expression pro ling of different plant tissues, such as the whole root samples and the syncytium structure were carried out in susceptible and resistant soybean varieties infected with SCN, and a lot of differential expression genes and related KEGG pathways were predicted [34,35,45].
However, there are few studies on soybean-SCN interactions by proteomic and metabolomic analyses.
Metabolomics is used to characterize the metabolic patterns and further study the functions and signi cance of the key differential metabolites, which play important roles in biological and medical elds [42]. Plant metabolomics has gained cumulative attentions, and became a mature method for studying plant responses to both biotic and abiotic stresses. Therefore, plant metabolomics is considered as an indispensable tool in the study of plant-pathogen interaction [12]. Plants respond to pathogen infection by producing the defense metabolites, especially, the plant secondary metabolites play important roles in plant host's resistance [33]. Metabolomics will provide technical support for the identi cation of plant metabolites which are potential sources of new compounds for nematode control [9]. Research on plant nematodes has become more and more abundant, especially on root-knot nematodes. Eloh et al. (2016) performed metabolomic analysis of root-knot nematodes treated with maleimide, and proved that maleimide could be used as a new potential nematicide [22]. In addition, they studied the levels of metabolites in tomato plants after root-knot nematode infection, and found that tomato roots altered the biochemical pathways in response to root-knot nematode infection [22]. Recently, Borges et al. (2018) investigated the metabolic pro les of both resistant and susceptible watermelon varieties against root-knot nematode infection by using nuclear magnetic resonance (NMR) and principal component analysis (PCA) [9]. However, there was few reports on SCN-soybean interactions. Differential proteins and metabolites between the resistant and susceptible soybean roots were identi ed with the traditional approaches more than ten years ago [1]. Recently, Kang et al. (2018) studied the effect of Bacillus simplex strain Sneb545 on soybean secondary metabolites under SCN infection by combining the advanced and high-throughput approaches of transcriptomic and metabolomic analyses, and found that Sneb545-treated soybeans showed a higher concentration of various nematicidal metabolites [32].
To date, there were no reports on the difference of metabolic pathways between resistant and susceptible soybeans with infection of SCN. In this study, a SCN-resistant soybean variety PI 437654 [3,4], and three SCN-susceptible soybean varieties Williams 82 (WM82), Zhonghuang 13 (ZH13) and Hefeng 47 (HF47) were selected as the research materials, and LC/MS full-scan detection technology was used to investigate the differences of metabolites and metabolic pathways between the resistant and susceptible soybean varieties after inoculation with SCN. The metabolites overlapped among resistant and susceptible soybeans, speci cally differential metabolites in resistant-or susceptible-soybeans and their potential KEGG metabolic pathways were identi ed. Moreover, the genes associated with the signi cantly up-regulated metabolites in the resistant soybean, which mainly included the resistant soybean PI437654-speci c up-regulated and the overlapping metabolite with signi cantly up-regulated but simultaneously down-regulated metabolites, were predicted. The results not only found out potential novel metabolites and associated genes in the resistant response of soybean to soybean cyst nematode, but also provided new insights into the interaction between soybean and soybean cyst nematode.

Results
Assessment of resistance of soybean to SCN At 8 dpi of SCN race 4, the average numbers of the second juveniles (J2s) within PI 437654 roots were signi cantly less than that of ZH13, while it did not show signi cant difference among PI 437654, WM82 and HF47 (Fig. 1a). However, the average numbers of the third juveniles (J3s) in PI 437654 were signi cantly decreased than those in the three other varieties ( Fig. 1a; Fig. 1b). At 60 dpi, the average numbers of cysts formed in the soybean PI 437654 were signi cantly decreased than that in the three other soybeans (WM82, ZH13 and HF47) (Fig. 1a). In addition, the J3s were much slimmer in the PI 437654 roots than in the three other soybeans (Fig. 1b). These results indicated that PI 437654 was a highly SCN-resistant soybean variety, while the other three varieties were the highly SCN-susceptible soybeans.
Differentiation of the metabolites in resistant and susceptible soybeans by the infection of SCN To investigate the differential root metabolites between the resistant and susceptible soybeans, the root samples of the resistant soybean PI 437654 and the three susceptible soybeans (WM82, ZH13 and HF47) inoculated with SCN race 4 ('_SCN') at 8 dpi, as well as their corresponding controls ('_0'), were collected and subjected to metabolomic analyses (PCA and PLS-DA analyses). In the PCA score chart of resistant soybean, all the PI 437654_SCN replicated samples were clustered together, and similarly, all the PI 437654_0 replicates were also clustered together (Fig. 2a), suggesting that the samples had sound repeatability. However, the PI 437654_SCN samples were clearly separated from the PI437654_0 samples (Fig. 2a), illustrating the signi cant induction of the changes of metabolites in the resistant soybean PI 437654 by the inoculation of SCN race 4 and the solid stability in the whole measurement process. The PLS-DA analyses results were similar to those of the PCA analyses, the samples of PI 437654_SCN were also clearly separated from those of PI437654_0 (Fig. 2b). The PCA and PLS-DA analyses results showed that all the three susceptible soybean lines WM82_SCN, ZH13_SCN and HF47_SCN were dramatically separated from each WM82_0, ZH13_0, and HF47_0 pairwise ( Fig. 2a; Fig. 2b), respectively, which indicated that the inoculation of SCN also caused obvious changes of metabolites of the susceptible soybean roots. Furthermore, according to the cumulative interpretation rate of model in the resistant and susceptible soybeans ( Fig. 2c; Table S1), both R 2 and Q 2 values were closer to 1, clearly explaining that the predictive power and model quality of the two groups of models were good and suitable for the subsequent experiments. All the results indicated that the SCN inoculation caused signi cant metabolic changes in both resistant and susceptible soybean lines.
Identi cation of differential metabolites in the resistant and susceptible soybeans At 8 dpi of SCN, 19 signi cantly differential metabolites were screened out in the resistant soybean PI 437654_SCN vs PI 437654_0 pairwise, among which seven metabolites were up-regulated, and 12 metabolites were down-regulated (Table 1). In the susceptible soybean WM82_SCN vs WM82_0 pairwise, 17 obviously differential metabolites were identi ed, including three up-regulated and 14 down-regulated metabolites (Table 1). Similarly, there were 12 dramatically differential metabolites in ZH13_SCN vs ZH13_0 pairwise, all of them were down-regulated (Table 1). There were 17 dramatically differential metabolites in HF47_SCN vs HF47_0 pairwise, including three up-regulated metabolites and 14 downregulated metabolites ( Table 1). The differential metabolites among the four soybean lines had different pro les, which were mainly divided into following three categories. Note: '*' represents the Log 2 (FC) value which was the ratio of the average expression of metabolites in the sample batch. Positive value represents up-regulated, while negative value represents downregulated. ' /' represents that soybean varieties did not contain such metabolites.
(1) Differential metabolites overlapped among resistant and susceptible soybeans. There were four differential metabolites overlapped (Table 1; Fig. S1), including D-leucine, D, L-typotophan, 16hydroxyhexadecanoic acid and linolenic acid, which belong to amino acids and fatty acids. Overall, linolenic acid, D-leucine and 16-Hydroxyhexadecanoic acid had an overall signi cantly down-regulated in resistant and three susceptible soybeans. However, D-leucine and 16-hydroxyhexadecanoic acid were obviously up-regulated in WM82_SCN vs WM82_0, and HF47_SCN vs HF47_0, respectively. Differently, D, L-typotophan was simultaneously dramatically down-regulated in the three susceptible soybean pairwise, while signi cantly up-regulated in the resistant soybean PI 437654_SCN vs PI437654_0 pairwise ( Table 1). The results indicated that D, L-typotophan potentially played an important role to defense SCN in the resistant soybean PI 437654.
(2) Differential metabolites speci c in the susceptible soybeans. There were 18 signi cantly differential metabolites speci cally presented in the susceptible soybeans but absented in the resistant PI437654_SCN vs PI437654_0. Both 2-oxo-4-methylthiobutanoic acid and 4-hydroxycoumarin were simultaneously overlapped in all the three susceptible soybeans (Table 1; Fig S1), while the other 16 metabolites only existed in one or two of the three susceptible soybeans (Table 1). Except four metabolites including PC (O-14:0/2:0), 3-hydroxy-7-methoxy avone, prunetin, and trans-ferulic acid were dramatically up-regulated in either WM82 or HF47 pairwise, the other 14 metabolites were obviously down-regulated among the three susceptible soybeans ( Table 1). The suppressed differential metabolites in the susceptible soybeans were probably linked to the soybean basal defense hijacked by the SCN infection for their establishment and development of the feed sites.
(3) Differential metabolites speci c in the resistant soybean. There were 10 uniquely differential metabolites in the resistant soybean PI 437654_SCN vs PI 437654_0 pairwise (Table 1; Fig. S1), including 7 down-regulated and 3 up-regulated metabolites (Table 1). Both D-aspartic acid and linoleic acid were the top two signi cantly down-regulated metabolites with about 70-fold more abundance in PI437654_SCN than that in PI 437654_0 (Table 1). N-palmitoyl alanine, cycloleucine and D, L-2,4diaminobutyric acid were all the sequentially differential down-regulated metabolites at about 8-, 3-and 2.5-fold more in PI 437654_SCN than that in PI 437654_0, respectively (Table 1). Among the 3 signi cantly up-regulated metabolites, N-acetyltranexamic acid and nicotine were the top two upregulated metabolites at about 12.9-and 5.8-fold more in PI 437654_SCN than in PI 437654_0, respectively (Table 1). In addition, 5 obviously differential metabolites including two down-regulated and three up-regulated metabolites in the PI437654_SCN vs PI437654_0 pairwise were identi ed. However, these ve metabolites were all reduced but not existed in all the three susceptible soybeans (Table 1). These results indicated that N-acetyltranexamic acid and nicotine might play potential roles to defense SCN infection in PI 437654, while D-aspartic acid and linoleic acid and N-palmitoyl alanine were likely involved into malfunction in feeding site establishment and maintenance for SCN development in PI 437654.
Metabolic pathways involved by differential metabolites in the resistant and susceptible soybeans The metabolic pathways involved by the differential metabolites were analyzed by the KEGG database searching. Totally, 14 metabolic pathways were changed in the resistant soybean PI 437654_SCN vs PI 437654_0, while 9, 9 and 8 metabolic pathways were changed in the susceptible soybean WM82_SCN vs WM82_0, ZH13_SCN vs ZH13_0, HF47_SCN vs HF47_0, respectively (Fig. 3, Table 2), which could also be divided into following three categories. (1) KEGG metabolic pathways overlapped among the resistant and susceptible soybeans. Three KEGG metabolic pathways including cutin, suberine and wax biosynthesis (ath00073), alpha-Linolenic acid metabolism (ath00592), and biosynthesis of unsaturated fatty acids (ath01040) were associated with the differential metabolites 16-hydroxyhexadecanoic acid and linolenic acid ( Table 2) and overlapped among all the resistant and susceptible soybeans. Moreover, the cutin, suberine and wax biosynthesis KEGG metabolic pathway (ath00073) was obviously activated in all the resistant soybean PI 437654 and the three susceptible soybeans infected with SCN (Fig. 3, Table 2), which indicated that this KEGG metabolic pathway was likely involved in the basal defense response of soybean varieties against SCN infection.

Associated genes involved in the up-regulated metabolites of the resistant soybean variety
In order to predict the associated genes involved in the up-regulated metabolites of the resistant soybean variety PI437654, the transcriptome of the resistant soybean PI437654 and the three susceptible soybean WM82, ZH13 and HF47, as well as their corresponding controls, were sequenced, and these transcriptome data were combined with the metabolomic data. The comparative transcriptome analyses identi ed 15,835 differential expressed genes (DEGs) ( respectively, infected by SCN ( Fig S2). As mentioned above, three key metabolites including Nacetyltranexamic acid, nicotine, and D, L-typotophan, which were signi cantly up-regulated in the resistant soybean PI437654 infected by SCN, were classi ed into two types.
One type is the resistant soybean PI437654-speci c signi cantly up-regulated metabolites containing Nacetyltranexamic acid and nicotine. The results of combination analyses showed that 14 potential associated genes ( Table S2). These associated genes were subject to GO analyses. The most abundant GO terms in the biological processes category were "translation", followed by "cytoplasmic translation" and "rRNA modi cation" (Fig. 5; Table S3). Regarding the cellular component category, the most abundant GO terms were "ribosome", "cytosolic ribosome", "cytosolic small ribosomal subunit", cytosolic large ribosomal subunit" and "nucleolus" (Fig. 5; Table S3). Concerning the molecular function category, the most abundant GO terms were "structural constituent of ribosome", "mRNA binding", "rRNA binding", "RNA binding" and "inorganic cation transmembrane transporter activity" (Fig. 5; Table S3). Correspondingly, these associated genes were enriched in KEGG pathways, which involving the largest number of unigenes were "Ribosome", followed by "RNA degradation", "Ribosome biogenesis in eukaryotes", "Peroxisome" and "Autophagy" (Fig S3, Table S4).
Another type is signi cantly up-regulated metabolite (D, L-typotophan) in the resistant soybean PI437654 but simultaneously dramatically down-regulated in the three susceptible soybean varieties. During the top 150 most associated genes, 80 genes had positive correlations, while 70 genes had negative correlations ( Fig. 4; Table S5). As for these associated genes, the most abundant GO terms in the biological processes category were "response to chitin", followed by "ethylene-activated signaling pathway" and "defense response" (Fig. 5; Table S6). Regarding the cellular component category, the most abundant GO terms were "extracellular region", "apoplast" and "photosystem I" (Fig. 5; Table S6). Concerning the molecular function category, the most abundant GO terms were "transcription factor activity, sequence-speci c DNA binding", "heme binding" and "hydroquinone:oxygen oxidoreductase activity" (Fig. 5; Table S6). Correspondingly, these associated genes were enriched in KEGG pathways, which involving the largest number of unigenes were "Plant hormone signal transduction", followed by "Phenylpropanoid biosynthesis", "MAPK signaling pathway -plant", "Starch and sucrose metabolism" and "Photosynthesis" (Fig S4, Table S7).
These results of the combination of metabolomic analyses and transcriptomics revealed potential associated genes involved in the most signi cantly up-regulated metabolites of the resistant soybean variety PI437654, and suggested their likely biological processes, cellular component, molecular function and involved pathways.

Discussion
A resistant soybean PI 437654 and three susceptible soybeans WM82, ZH13 and HF47 were used as the test materials to explore the changes of metabolites in soybean roots with and without SCN inoculation by LC/MS analyses in this study. Totally, 37 differential metabolites and 20 KEGG metabolic pathways were identi ed, which were divided into three categories including the metabolites/metabolic pathways overlapped among all the tested resistant and susceptible soybeans, speci c in the susceptible soybeans and speci c in the resistant soybeans (Table 1; Table 2). It was found that the same differential metabolites such as linolenic acid and L-arginine were probably involved in more than one KEGG metabolite pathways, while the differently differential metabolites such as nicotine and pipecolinic acid were likely involved in the identical tropane, piperidine and pyridine alkaloid biosynthesis pathway (Table 2). However, only 12 differential metabolites (32%) were found to be involved in the predicted KEGG metabolite pathways.
In this study, some basal defense-related potential metabolites, such as linolenic acid and 16hydroxyhexadecanoic acid, were found in the resistant soybean (PI 437654) and the three susceptible soybeans (WM82, ZH13 and HF47), and our metabolomic analyses results were consistent with some previous reports. Linolenic acid is involved in the metabolic pathways including alpha-linolenic acid metabolism and biosynthesis of unsaturated fatty acids, and it was thought to be related to plant tolerance to low temperature, and the content of linolenic acid was proportional to the cold tolerance of wheat [29,56,62]. In addition, linolenic acid plays an important role in the defense response to pathogen infection and traumatic stress in Arabidopsis thaliana [7]. Unsaturated fatty acids are considered as a new generation of plant disease inducers. Amruthesh et al. (2005) investigated the role of unsaturated fatty acids in the resistance of pearl millet to downy mildew, and found that its stimulation induced strong plant resistance [2]. Clay et al. (2006) found that the host resistance was related to the higher content of unsaturated fatty acids in the nuclear membrane of resistant species on Gossypium barbadense [16].
The metabolites and metabolic pathways speci c in the resistant soybean PI 437654 infected with SCN, mainly including L-arginine, phytosphingosine, D-aspartic acid and nicotine, probably played pivotal roles on plant stress, tolerance or resistance based on some related research literature. L-arginine is the precursor of nitric oxide (NO), while NO is synthesized by L-arginine through NO synthase in the Larginine-NO pathway, which is widely considered as an important regulator of cell function and communication in physiological and pathophysiological states [23,46,47]. NO is a second messenger in the stress response, and is involved in plant responses to biotic and abiotic stresses [18,66]. L-arginine is linked and involved in many KEGG metabolic pathways such as arginine and proline metabolism, aminoacyl-tRNA biosynthesis and ABC transporters. Many studies indicated that there was a positive correlation between proline accumulation and plant stress tolerance [8]. Aminoacyl-tRNA synthase is an important component of protein synthesis and catalyzes the connection of amino acids with their homologous tRNA. They form a different set of enzymes, ensuring the delity of the transfer of genetic information from DNA to proteins [50,55]. ABC transporter is a member of the protein superfamily, it uses the energy from the hydrolysis of ATP to transport material over membranes [60]. Recently, ABC transporters attracted more attentions because the ABC gene was associated with pesticide resistance [25]. Phytosphingosine (PHS), a member of the sphingosine family, shows fungicidal activity [57]. Castro et al. (2008) studied the effect of PHS on cells of the Neurospora crassa and found that PHS reduced the viability of fungal cells and induced fungal apoptosis [13]. D-aspartic acid participated in alanine, aspartate and glutamate metabolism, while glutamate dehydrogenase (GDH) is a central enzyme in glutamate metabolism [5]. The enhancement of drought resistance in wheat was related to ammonia assimilation and enhanced NADH-GDH activity under low osmotic stress and high osmotic stress [65].
Nicotine was widely used in killing many kinds of insects [28]. Vänninen at el. (2011) discovered that nicotine has a strong killing effect on the adults of miridae [58]. Imidacloprid is a new generation of nicotine pesticides with low toxicity, and nicotine and imidacloprid reduced aphid fecundity by inhibiting the reproduction of nymphs with certain concentrations [19]. Nicotine is involved in the biosynthesis of tropane, piperidine and pyridine alkaloid in metabolic pathways, and alkaloid was found to have obvious inhibitory effect on sun ower downy mildew fungus Plasmopara halstedii [51]. Thus, L-arginine, phytosphingosine, D-aspartic acid and nicotine may be involved in the resistance of soybean against SCN.
Differently, the differential metabolite 2-oxo-4-methylthiobutanoic acid, which was overlapped in all the three susceptible soybeans but absent in resistant soybean, was predicted to participate into three KEGG metabolic pathways, including cysteine and methionine metabolism, glucosinolate biosynthesis and 2oxocarboxylic acid metabolism. Moreover, 2-oxo-4-methylthiobutanoic acid was simultaneously downregulated in the three susceptible soybeans (ZH13, WM82 and HF47) infected with SCN, which likely resulted in decreased biosynthesis of glucosinolate. It was reported that glucosinolate was a plant secondary metabolite and had a preventive effect against plant pathogens and soil-borne plant pests [10,37]. Additionally, the metabolic pathways of avonoid biosynthesis, diterpenoid biosynthesis, which were reported to be involved in plant resistance to pathogens, were found to be enriched in one or two susceptible soybean varieties [6,48]. This type of differential metabolites and their related KEGG metabolite pathways might play potential roles in the manipulation of host innate immunity to establish and maintain the feeding sites for SCN development [26,54].
A local metabolite-calibrated database was used to identify potential differential metabolites in this study, which safeguarded the accuracy and precision [38,39], but also likely had some disadvantages to miss many important compounds due to its limited capacities. Nevertheless, in this study, the different metabolomic responses of the resistant and susceptible soybean varieties to SCN infection were thoroughly characterized, and some important and pivotal different metabolites and related KEGG metabolite pathways were identi ed, and their likely functions were predicted based on previous studies.
In this study, three key metabolites including N-acetyltranexamic acid, nicotine, and D, L-typotophan, which were signi cantly up-regulated in the resistant soybean PI437654 infected by SCN, were identi ed and classi ed into two types. Combination of the metabolomic analyses with the transcriptomic expression pro ling aided to predict potential associated genes involved in the most signi cantly upregulated metabolites of the resistant soybean variety PI437654, and suggest their likely biological processes, cellular component, molecular function and involved pathways.

Conclusion
In this study, the main metabolomic differences of the resistant and susceptible soybeans infected with SCN were characterized by using a broad-spectrum SCN-high resistant soybean PI 437654 versus three SCN-susceptible soybean varieties. Totally, 37 differential metabolites and 20 KEGG metabolic pathways were identi ed, which were divided into three categories including the metabolites/metabolic pathways overlapped among all the tested resistant and susceptible soybean, speci c in susceptible or resistant soybeans. Twelve differential metabolites were found to be involved in the predicted KEGG metabolite pathways. Moreover, 14 differential metabolites such as signi cantly up-regulated nicotine and downregulated D-aspartic acid, and their involved KEGG pathways such as tropane, piperidine and pyridine alkaloid biosynthesis, alanine, aspartate and glutamate metabolism, sphingolipid metabolism and arginine biosynthesis were signi cantly changed and abundantly enriched speci cally in the resistant soybean PI 437654, and likely played pivotal roles in defensing SCN. Moreover, combination of the metabolomic analyses with the transcriptomic expression pro ling aided to predict potential associated genes involved in the most signi cantly up-regulated metabolites of the resistant soybean variety PI437654, and suggest their likely biological processes, cellular component, molecular function and involved pathways. To our best knowledge, our study was the rst to investigate the different responses of resistant and susceptible soybeans infected with SCN by combining metabolomic analyses and transcriptomics. Our results not only found out the potential novel metabolites and associated genes in the resistant response of soybean to soybean cyst nematode, but also provided new insights into the interactions between soybean and soybean cyst nematode.

Soybean seedling preparation and SCN inoculation
All the soybean seeds including PI 437654, Williams 82 (WM82), Zhonghuang 13 (ZH13) and Hefeng 47 (HF47), which were stored in the Institute of Plant Protection, Chinese Academy of Agricultural Sciences, were sterilized with 1.0% (w/v) NaClO for 5 min [20], and the residual NaClO was washed with sterilized water. Then the seeds were placed between moist sterilized lter papers at 26 ℃. About 48 h, the germinated seeds were transplanted into the sandy soil (soil: sand = 3:1 (v/v)). The day and night temperature was set at 28 ℃ / 26 ℃ with a photoperiod of 12 h [15].
The cysts of SCN race 4 were soaked in the 3 mM ZnCl 2 solution and incubated in the dark at 25 ℃ [52].
The hatched SCN J2s were collected with a glass tube, and the concentration of SCN J2s suspension was adjusted to 500 J2s / tube [41]. The 14-day-old soybean seedlings were inoculated with 1000 SCN J2s / seedling. At 8 days post inoculation (dpi), the root samples of the resistant soybean PI437654 and the three susceptible soybean WM82, ZH13 and HF47, as well as their corresponding controls, were harvested. Each type of root sample was divided into two parts for the following metabolomic and transcriptomic analyses, respectively.

Juveniles Staining And Cysts Counting
Soybean root samples were collected at 8 dpi, and were stained by acid magenta staining as previously described [11], and the morphology were observed and the numbers of SCN juveniles within the root tissues were counted under a microscope (OLYMPUS-SZ2-ILST). The cysts were harvested from both the root tissues and root soil at 60 dpi, and their numbers were counted with a stereoscope (OLYMPUS-BX53F).

Lc/ms Metabolic Sequencing
At 8 dpi, the root samples of the resistant soybean PI437654 and the three susceptible soybean WM82, ZH13 and HF47, as well as their corresponding controls, were harvested, which were named PI437654_SCN, PI437654_0, WM82_SCN, WM82_0, ZH13_SCN, ZH13_0, HF47_SCN and HF47_0, respectively. Each soybean root sample was ground into powder in liquid nitrogen. Equivalent 80 mg root powder was dropped into a 1.5 mL centrifuge tube. Add 20 µL internal standard (0.3 mg/mL L-2chlorophenylalanine ) and 1 mL 70% methanol (methanol :water = 7:3 (volume ratio)); Place at -20 ℃ for 2 min to pre-cool, and grind with a grinding machine for 2 min (60 Hz); Then, ultrasonic extraction was placed at 4 °C for 30 min ; after standing at -20 °C for 20 min, the samples were centrifuged for 15 min (13,000 rpm, 4℃), and 200 µL supernatant was drawn with a syringe, each sample was repeated 3 times, and the same source supernatants were combined together. After ltering with a 0.22 µm organic phase pinhole lter, it was transferred to an LC injection vial and stored at -80 °C for LC-MS analysis [63]. Each treatment had six repetitions.

Metabolomic Analyses
The metabolites were identi ed using a local metabolite-calibrated database [38,39]. Multivariate statistical analysis, including principal component analysis (PCA) and partial least squares discriminant analysis (PLS-DA), was performed using SIMCA-P + 14.0 version software (Umetrics, Umeå, Sweden) to distinguish differences in metabolic pro les between pairwise groups. The Hotelling's T2 region, shown as an ellipse in score plots of the models, de nes the 95% con dence interval of the modeled variation. In the t-test, a variable with p-value < 0.05 was considered as a difference variable. Metabolites with the VIP (variable importance in the projection) > 1.0 and p-value < 0.05 were selected as differential metabolites.
KEGG pathways were predicted by the differential metabolites with the p-value < 0.05 [63].

Correlation Analysis Between Metabolome And Transcriptome Data
Correlation Analyses were preformed as previously described [14]. Pearson correlation coe cients were calculated for metabolome and transcriptome data integration. The mean content of metabolites of different soybean varieties (the resistant soybean PI437654 and the three susceptible soybean WM82, ZH13 and HF47) and the mean value of transcript abundance of DEGs together were calculated. The coe cients were calculated from log2 (fold change) of each metabolite and log2 (fold change) of each transcript. Correlations corresponding to a coe cient with R2 > 0.9 were selected. The relationships between metabolome and transcriptome data were demonstrated by using Cytoscape (The Cytoscape Consortium, San Diego, CA, USA, version 2.8.2). Top 150 associated genes were focused on in this study.