Information theory examines important predictions of plant defense theory for metabolism

Different plant defense theories provide important theoretical guidance for explaining the patterns of plant specialized metabolism, but their key predictions remain to be tested. Here, we used unbiased tandem mass spectrometry (MS/MS) analysis to systematically explore the metabolome of tobacco attenuated strains from individual plants to populations and closely related species, and processed a large number of mass spectrometric feature theories based on compound spectra in the information. Framework to test the key predictions of optimal defense (OD) and moving target (MT) theories. The information component of plant metabolomics is consistent with OD theory, but contradicts the main prediction of MT theory on metabolomics dynamics caused by herbivores. From the micro to the macro evolutionary scale, the jasmonate signal was identified as the main determinant of OD, while the ethylene signal provided fine-tuning of the herbivore-specific response annotated by the MS/MS molecular network.
Special metabolites with diverse structures are the main participants in plant adaptation to the environment, especially in defense of enemies (1). The amazing diversification of the special metabolism found in plants has stimulated decades of in-depth research on its many aspects of ecological functions, and has formed a long list of plant defense theories, which are the evolutionary and ecological development of plant-insect interactions. Empirical research provides important guidance (2). However, these plant defense theories did not follow the normative path of hypothetical deductive reasoning, in which key predictions were at the same level of analysis (3) and were tested experimentally to advance the next cycle of theoretical development (4). Technical limitations restrict data collection to specific metabolic categories and exclude comprehensive analysis of specialized metabolites, thus preventing inter-category comparisons that are essential for theoretical development (5). The lack of comprehensive metabolomics data and a common currency to compare the processing workflow of the metabolic space between different plant groups hinders the scientific maturity of the field.
The latest developments in the field of tandem mass spectrometry (MS/MS) metabolomics can comprehensively characterize the metabolic changes within and between species of a given system clade, and can be combined with computational methods to calculate the structural similarity between these complex mixtures. Prior knowledge of chemistry (5). The combination of advanced technologies in analysis and computing provides a necessary framework for long-term testing of many predictions made by the ecological and evolutionary theories of metabolic diversity. Shannon (6) introduced information theory for the first time in his seminal article in 1948, laying the foundation for the mathematical analysis of information, which has been used in many fields other than its original application. In genomics, information theory has been successfully applied to quantify sequence conservative information (7). In transcriptomics research, information theory analyzes the overall changes in the transcriptome (8). In previous research, we applied the statistical framework of information theory to metabolomics to describe the metabolic expertise of the tissue level in plants (9). Here, we combine the MS/MS-based workflow with the statistical framework of information theory, characterized by the metabolic diversity in the common currency, to compare the key predictions of the plant defense theory of the metabolome induced by herbivores.
The theoretical frameworks of plant defense are usually mutually inclusive and can be divided into two categories: those that try to explain the distribution of plant-specific metabolites based on defense functions, such as optimal defense (OD) (10), moving target (MT) (11) ) And appearance (12) theory, while others seek mechanical explanations of how changes in resource availability affect plant growth and the accumulation of specialized metabolites, such as carbon: nutrient balance hypothesis (13), growth rate hypothesis (14 ), and the growth and differentiation equilibrium hypothesis (15). The two sets of theories are at different levels of analysis (4). However, two theories involving defensive functions at the functional level dominate the conversation about plant constitutive and inducible defenses: OD theory, which assumes that plants invest in their expensive chemical defenses only when needed, for example, when they are ingested When a grass animal attacks, therefore, according to the possibility of future attack, the compound with a defensive function is assigned (10); the MT hypothesis proposes that there is no axis of directional metabolite change, but the metabolite changes randomly, thus creating the possibility Obstruct the metabolic “motion target” of attacking herbivores. In other words, these two theories make opposite predictions about the metabolic remodeling that occurs after the attack of herbivores: the relationship between the unidirectional accumulation of metabolites with defensive function (OD) and non-directed metabolic changes (MT) (11 ).
The OD and MT hypotheses involve not only the induced changes in the metabolome, but also the ecological and evolutionary consequences of the accumulation of these metabolites, such as the adaptive costs and benefits of these metabolic changes in a specific ecological environment (16). Although both hypotheses recognize the defensive function of specialized metabolites, which may or may not be expensive, the key prediction that distinguishes the OD and MT hypotheses lies in the directionality of the induced metabolic changes. The prediction of OD theory has received the most experimental attention so far. These tests include the study of the direct or indirect defense functions of different tissues of specific compounds in greenhouses and natural conditions, as well as changes in the stage of plant development (17-19). However, so far, due to the lack of a workflow and statistical framework for global comprehensive analysis of metabolic diversity of any organism, the main difference prediction between the two theories (that is, the direction of metabolic changes) remains to be tested. Here, we provide such an analysis.
One of the most significant characteristics of plant-specific metabolites is their extreme structural diversity at all levels from single plants, populations to similar species (20). Many quantitative changes in specialized metabolites can be observed at the population scale, while strong qualitative differences are usually maintained at the species level (20). Therefore, plant metabolic diversity is the main aspect of functional diversity, reflecting the adaptability to different niches, especially those niches with different invasion possibilities by special insects and common herbivores (21). Since Fraenkel’s (22) groundbreaking article on the reasons for the existence of plant-specific metabolites, interactions with various insects have been regarded as important selection pressures, and these interactions are believed to have shaped plants during evolution. Metabolic pathway (23). Interspecies differences in the diversity of specialized metabolites may also reflect the physiological balance associated with constitutive and inducible plant defense against herbivorous strategies, as the two species are often negatively correlated with each other (24). Although it may be beneficial to maintain a good defense at all times, timely metabolic changes associated with defense provide clear benefits in allowing plants to allocate valuable resources to other physiological investments (19, 24), and avoid the need for symbiosis. Collateral damage (25). In addition, these reorganizations of specialized metabolites caused by insect herbivores may lead to destructive distribution in the population (26), and may reflect direct readings of substantial natural changes in the jasmonic acid (JA) signal, which may be maintained in the population. High and low JA signals are trade-offs between defense against herbivores and competition with specific species (27). In addition, specialized metabolite biosynthetic pathways will undergo rapid loss and transformation during evolution, resulting in patchy metabolic distribution among closely related species (28). These polymorphisms can be rapidly established in response to changing herbivore patterns (29), which means that the fluctuation of herbivore communities is a key factor driving metabolic heterogeneity.
Here, we specifically solved the following problems. (I) How does the herbivorous insect reconfigure the plant metabolome? (Ii) What are the main information components of metabolic plasticity that can be quantified to test the predictions of long-term defense theory? (Iii) Whether to reprogram the plant metabolome in a way unique to the attacker, if so, what role does plant hormone play in tailoring a specific metabolic response, and which metabolites contribute to the species specificity of defense? (Iv) Since the predictions made by many defense theories can be extended across all levels of biological tissues, we asked how consistent the metabolic response caused is from internal comparison to interspecies comparison? To this end, we have systematically studied the leaf metabolome of tobacco nicotine, which is an ecological model plant with rich specialized metabolism, and is effective against the larvae of two native herbivores, Lepidoptera Datura (Ms) ( Very aggressive, mainly eaten) On Solanaceae and Spodoptera littoralis (Sl), cotton leaf worms are a kind of “genus”, with the host plants of Solanaceae and other hosts of other genera and families Plant food. We parsed the MS/MS metabolomics spectrum and extracted information theory statistical descriptors to compare OD and MT theories. Create specificity maps to reveal the identity of key metabolites. The analysis was extended to the native population of N. nasi and closely related tobacco species to further analyze the covariance between plant hormone signaling and OD induction.
In order to capture an overall map about the plasticity and structure of the leaf metabolome of herbivorous tobacco, we used a previously developed analysis and calculation workflow to comprehensively collect and deconvolute high-resolution data independent MS/MS spectra from plant extracts ( 9). This undifferentiated method (called MS/MS) can construct non-redundant compound spectra, which can then be used for all compound-level analyses described here. These deconvoluted plant metabolites are of various types, consisting of hundreds to thousands of metabolites (about 500-1000-s/MS/MS here). Here, we consider metabolic plasticity in the framework of information theory, and quantify the diversity and professionalism of the metabolome based on the Shannon entropy of the metabolic frequency distribution. Using the previously implemented formula (8), we calculated a set of indicators that can be used to quantify metabolome diversity (Hj indicator), metabolic profile specialization (δj indicator) and metabolic specificity of a single metabolite (Si indicator) . In addition, we applied the Relative Distance Plasticity Index (RDPI) to quantify the metabolome inducibility of herbivores (Figure 1A) (30). Within this statistical framework, we treat MS/MS spectrum as the basic information unit, and process the relative abundance of MS/MS into a frequency distribution map, and then use Shannon entropy to estimate metabolome diversity from it. The metabolome specialization is measured by the average specificity of a single MS/MS spectrum. Therefore, the increase in the abundance of some MS/MS classes after herbivore induction is transformed into spectral inducibility, RDPI and specialization, that is, the increase in the δj index, because more specialized metabolites are produced and a high Si index is produced. The lowering of the Hj diversity index reflects that either the number of generated MS/MS is reduced, or the profile frequency distribution changes in a less uniform direction, while reducing its overall uncertainty. Through the Si index calculation, it is possible to highlight which MS/MS are induced by certain herbivores, on the contrary, which MS/MS does not respond to the induction, which is a key indicator to distinguish MT and OD prediction.
(A) Statistical descriptors used for herbivorous (H1 to Hx) MS/MS data-inducibility (RDPI), diversity (Hj index), specialization (δj index) and metabolite specificity (Si index). An increase in the degree of specialization (δj) indicates that, on average, more herbivorous specific metabolites will be produced, while a decrease in diversity (Hj) indicates a decrease in the production of metabolites or uneven distribution of metabolites in the distribution map . The Si value assesses whether the metabolite is specific to a given condition (here, herbivorous) or conversely maintained at the same level. (B) Conceptual diagram of defense theory prediction using information theory axis. OD theory predicts that herbivore attack will increase defense metabolites, thereby increasing δj. At the same time, Hj decreases because the profile is reorganized toward the reduced uncertainty of metabolic information. The MT theory predicts that the attack of herbivores will cause non-directional changes in the metabolome, thereby increasing Hj as an indicator of increased metabolic information uncertainty and causing a random distribution of Si. We also proposed a mixed model, the best MT, in which some metabolites with higher defensive values ​​will be particularly increased (high Si value), while others exhibit random responses (lower Si value).
Using information theory descriptors, we interpret the OD theory to predict that herbivore-induced special metabolite changes in an uninduced constitutive state will lead to (i) an increase in metabolic specificity (Si index) driving metabonomic specificity (δj index) The increase of) certain special metabolite groups with higher defense value, and (ii) the decrease of metabolome diversity (Hj index) due to the change of metabolic frequency distribution to more leptin body distribution. At the level of a single metabolite, an ordered Si distribution is expected, where the metabolite will increase the Si value according to its defense value (Figure 1B). Along this line, we explain the MT theory to predict that excitation will lead to (i) non-directional changes in metabolites resulting in a decrease in the δj index, and (ii) an increase in the Hj index due to an increase in metabolic uncertainty. Or randomness, which can be quantified by Shannon entropy in the form of generalized diversity. As for the metabolic composition, the MT theory will predict the random distribution of Si. Taking into account that certain metabolites are under specific conditions under specific conditions, and other conditions are not under specific conditions, and their defense value depends on the environment, we also proposed a mixed defense model, in which δj and Hj are distributed in two along Si Increase in all directions, only certain metabolite groups, which have higher defense values, will particularly increase Si, while others will have a random distribution (Figure 1B).
In order to test the redefined defense theory prediction on the axis of the information theory descriptor, we raised expert (Ms) or generalist (Sl) herbivore larvae on the leaves of Nepenthes pallens (Figure 2A). Using MS/MS analysis, we retrieved 599 non-redundant MS/MS spectra (data file S1) from methanol extracts of leaf tissue collected after caterpillar feeding. Using RDPI, Hj, and δj indexes to visualize the reconfiguration of information content in MS/MS configuration files reveals interesting patterns (Figure 2B). The overall trend is that, as described by the information descriptor, as the caterpillars continue to eat leaves, the degree of all metabolic reorganization increases over time: 72 hours after the herbivore eats, the RDPI increases significantly. Compared with the undamaged control, Hj was significantly reduced, which was due to the increased degree of specialization of the metabolic profile, which was quantified by the δj index. This apparent trend is consistent with the predictions of the OD theory, but is inconsistent with the main predictions of the MT theory, which believes that random (non-directional) changes in metabolite levels are used as a defensive camouflage (Figure 1B). Although the oral secretion (OS) elicitor content and feeding behavior of these two herbivores are different, their direct feeding resulted in similar changes in the directions of Hj and δj during the 24-hour and 72-hour harvest periods. The only difference occurred at 72 hours of RDPI. Compared with that caused by Ms feeding, the overall metabolism induced by Sl feeding was higher.
(A) Experimental design: common pig (S1) or expert (Ms) herbivores are fed with desalinated leaves of pitcher plants, while for simulated herbivory, the OS of Ms (W + OSMs) is used to handle the puncture of standardized leaf positions wound. S1 (W + OSSl) larvae or water (W + W). The control (C) is an undamaged leaf. (B) Inducibility (RDPI compared with the control chart), diversity (Hj index) and specialization (δj index) index calculated for the special metabolite map (599 MS/MS; data file S1). Asterisks indicate significant differences between the direct herbivore feeding and the control group (Student’s t-test with paired t-test, *P<0.05 and ***P<0.001). ns, not important. (C) Time resolution index of main (blue box, amino acid, organic acid and sugar; data file S2) and special metabolite spectrum (red box 443 MS/MS; data file S1) after simulated herbivory treatment. The color band refers to the 95% confidence interval. The asterisk indicates the significant difference between the treatment and the control [quadratic analysis of variance (ANOVA), followed by Tukey's honestly significant difference (HSD) for post hoc multiple comparisons, *P<0.05, **P<0.01 And *** P <0.001]. (D) Specialization of scatter plots and special metabolite profiles (repeated samples with different treatments).
To explore whether the herbivore-induced remodeling at the metabolome level is reflected in the changes in the level of individual metabolites, we first focused on the metabolites previously studied in the leaves of Nepenthes pallens with proven herbivore resistance. Phenolic amides are hydroxycinnamamide-polyamine conjugates that accumulate during the herbivory process of insects and are known to reduce insect performance (32). We searched the precursors of the corresponding MS/MS and plotted their cumulative kinetic curves (Figure S1). Unsurprisingly, phenol derivatives that are not directly involved in the defense against herbivores, such as chlorogenic acid (CGA) and rutin, are down-regulated after herbivory. In contrast, herbivores can make phenol amides highly potent. The continuous feeding of the two herbivores resulted in almost the same excitation spectrum of phenolamides, and this pattern was especially obvious for de novo synthesis of phenolamides. The same phenomenon will be observed when exploring the 17-hydroxygeranyl nonanediol diterpene glycosides (17-HGL-DTGs) pathway, which produces a large number of acyclic diterpenes with effective anti-herbivore functions (33), of which Ms Feeding with Sl triggered a similar expression profile (Figure S1)).
The possible disadvantage of direct herbivore feeding experiments is the difference in leaf consumption rate and feeding time of herbivores, which makes it difficult to eliminate the herbivore-specific effects caused by wounds and herbivores. In order to better solve the herbivore species specificity of the induced leaf metabolic response, we simulated the feeding of Ms and Sl larvae by immediately applying the freshly collected OS (OSM and OSS1) to the standard puncture W of consistent leaf positions. This procedure is called W + OS treatment, and it standardizes the induction by precisely timing the onset of the response elicited by the herbivore without causing confounding effects of differences in the rate or quantity of tissue loss (Figure 2A) (34). Using the MS/MS analysis and calculation pipeline, we retrieved 443 MS/MS spectra (data file S1), which overlapped with the spectra previously assembled from direct feed experiments. The information theory analysis of this MS/MS data set showed that the reprogramming of leaf-specialized metabolomes by simulating herbivores showed OS-specific inducements (Figure 2C). In particular, compared with OSS1 treatment, OSM caused an enhancement of metabolome specialization at 4 hours. It is worth noting that compared with the direct herbivore feeding experimental data set, the metabolic kinetics visualized in two-dimensional space using Hj and δj as coordinates and the directionality of the metabolome specialization in response to simulated herbivore treatment over time Increase consistent (Figure 2D). At the same time, we quantified the content of amino acids, organic acids and sugars (data file S2) to investigate whether this targeted increase in metabolome expertise is due to the reconfiguration of central carbon metabolism in response to simulated herbivores ( Figure S2). To better explain this pattern, we further monitored the metabolic accumulation kinetics of the previously discussed phenolamide and 17-HGL-DTG pathways. The OS-specific induction of herbivores is transformed into a differential rearrangement pattern within phenolamide metabolism (Figure S3). Phenolic amides containing coumarin and caffeoyl moieties are preferentially induced by OSS1, while OSMs trigger a specific induction of ferulyl conjugates. For the 17-HGL-DTG pathway, differential OS induction by downstream malonylation and dimalonylation products was detected (Figure S3).
Next, we studied the OS-induced transcriptome plasticity using the time-course microarray data set, which simulates the use of OSMs to treat the leaves of the rosette plant leaves in herbivores. The sampling kinetics basically overlap with the kinetics used in this metabolomics study (35). Compared with the metabolome reconfiguration in which metabolic plasticity is particularly increased over time, we observe transient transcription bursts in leaves induced by Ms, where transcriptome inducibility (RDPI) and specialization (δj) are at 1 There was a significant increase in hours, and diversity (Hj) at this time point, the expression of BMP1 was significantly reduced, followed by the relaxation of transcriptome specialization (Figure S4). Metabolic gene families (such as P450, glycosyltransferase, and BAHD acyltransferase) participate in the process of assembling special metabolites from structural units derived from primary metabolism, following the aforementioned early high-specialization model. As a case study, the phenylalanine pathway was analyzed. The analysis confirmed that the core genes in phenolamide metabolism are highly OS-induced in herbivores compared with unattracted plants, and are closely aligned in their expression patterns. The transcription factor MYB8 and structural genes PAL1, PAL2, C4H and 4CL in the upstream of this pathway showed early initiation of transcription. Acyltransferases that play a role in the final assembly of phenolamide, such as AT1, DH29, and CV86, exhibit a prolonged upregulation pattern (Figure S4). The above observations indicate that the early initiation of transcriptome specialization and the later enhancement of metabolomics specialization are a coupled mode, which may be due to the synchronous regulatory system that initiates a powerful defense response.
The reconfiguration in plant hormone signaling acts as a regulatory layer that integrates herbivorous information to reprogram the physiology of plants. After the herbivore simulation, we measured the cumulative dynamics of key plant hormone categories and visualized the temporal co-expression between them [Pearson correlation coefficient (PCC)> 0.4] (Figure 3A). As expected, plant hormones related to biosynthesis are linked within the plant hormone co-expression network. In addition, metabolic specificity (Si index) is mapped to this network to highlight the plant hormones induced by different treatments. Two main areas of herbivorous specific response are drawn: one is in the JA cluster, where JA (its biologically active form JA-Ile) and other JA derivatives show the highest Si score; the other is ethylene (ET). Gibberellin showed only a moderate increase in herbivore specificity, while other plant hormones, such as cytokinin, auxin, and abscisic acid, had low induction specificity for herbivores. Compared with using W + W alone, the amplification of the peak value of JA derivatives through the OS application (W + OS) can basically be transformed into a strong specific indicator of JAs. Unexpectedly, OSM and OSS1 with different elicitor content are known to cause similar accumulation of JA and JA-Ile. In contrast to OSS1, OSM is specifically and strongly induced by OSMs, while OSS1 does not amplify the response of basal wounds (Figure 3B).
(A) Co-expression network analysis based on PCC calculation of herbivore-induced plant hormone accumulation kinetics simulation. The node represents a single plant hormone, and the size of the node represents the Si index specific to the plant hormone between treatments. (B) Accumulation of JA, JA-Ile and ET in leaves caused by different treatments indicated by different colors: apricot, W + OSM; blue, W + OSSl; black, W + W; gray, C (control) . Asterisks indicate significant differences between treatment and control (two-way ANOVA followed by Tukey HSD post hoc multiple comparison, *** P <0.001). Information theory analysis of (C)697 MS/MS (data file S1) in JA biosynthesis and impaired perception spectrum (irAOC and irCOI1) and (D)585 MS/MS (data file S1) in ETR1 with impaired ET signal Two simulated herbivore treatments triggered plant lines and empty vehicle (EV) control plants. Asterisks indicate significant differences between W+OS treatment and undamaged control (two-way ANOVA followed by Tukey HSD post hoc multiple comparison, *P<0.05, **P<0.01 and ***P<0.001). (E) Scattered graphs of scattered opposition to specialization. The colors represent different genetically modified strains; the symbols represent different treatment methods: triangle, W + OSS1; rectangle, W + OSM; circle C
Next, we use a genetically modified strain of attenuated Nepenthes (irCOI1 and sETR1) in the key steps of JA and ET biosynthesis (irAOC and irACO) and perception (irCOI1 and sETR1) to analyze the metabolism of these two plant hormones on herbivores The relative contribution of reprogramming. Consistent with previous experiments, we confirmed the induction of herbivore-OS in empty carrier (EV) plants (Figure 3, C to D) and the overall decrease in the Hj index caused by OSM, while the δj index increased. The response is more pronounced than the response triggered by OSS1. A two-line graph using Hj and δj as coordinates shows the specific deregulation (Figure 3E). The most obvious trend is that in strains lacking JA signal, the metabolome diversity and specialization changes caused by herbivores are almost completely eliminated (Figure 3C). In contrast, the silent ET perception in sETR1 plants, although the overall effect on changes in herbivorous metabolism is much lower than that of JA signaling, attenuates the difference in Hj and δj indices between OSM and OSS1 excitations (Figure 3D and Figure S5). . This indicates that in addition to the core function of JA signal transduction, ET signal transduction also serves as a fine-tuning of the species-specific metabolic response of herbivores. Consistent with this fine-tuning function, there was no change in the overall metabolome inducibility in sETR1 plants. On the other hand, compared with sETR1 plants, irACO plants induced similar overall amplitudes of metabolic changes caused by herbivores, but showed significantly different Hj and δj scores between OSM and OSS1 challenges (Figure S5).
In order to identify specialized metabolites that have important contributions to the species-specific response of herbivores and fine-tune their production through ET signals, we used the previously developed structural MS/MS method. This method relies on the bi-clustering method to re-infer the metabolic family from MS/MS fragments [normalized dot product (NDP)] and similarity score based on neutral loss (NL). The MS/MS data set constructed through the analysis of ET transgenic lines produced 585 MS/MS (data file S1), which was resolved by clustering them into seven main MS/MS modules (M) (Figure 4A) . Some of these modules are densely packed with previously characterized special metabolites: for example, M1, M2, M3, M4 and M7 are rich in various phenol derivatives (M1), flavonoid glycosides (M2), acyl sugars (M3 And M4), and 17-HGL-DTG (M7). In addition, the metabolic specific information (Si index) of a single metabolite in each module is calculated, and its Si distribution can be seen intuitively. In short, MS/MS spectra exhibiting high herbivory and genotype specificity are characterized by high Si values, and kurtosis statistics indicate the distribution of fur on the right tail corner. One such lean colloid distribution was detected in M1, in which phenol amide showed the highest Si fraction (Figure 4B). The previously mentioned herbivorous inducible 17-HGL-DTG in M7 showed a moderate Si score, indicating a moderate degree of differential regulation between the two OS types. In contrast, most constitutively produced specialized metabolites, such as rutin, CGA, and acyl sugars, are among the lowest Si scores. In order to better explore the structural complexity and Si distribution between special metabolites, a molecular network was constructed for each module (Figure 4B). An important prediction of the OD theory (summarized in Figure 1B) is that the reorganization of special metabolites after herbivory should lead to one-way changes in metabolites with high defense value, especially by increasing their specificity (as opposed to random distribution) Mode) Defensive metabolite predicted by MT theory. Most of the phenol derivatives accumulated in M1 are functionally related to the decline of insect performance (32). When comparing the Si values ​​in the M1 metabolites between the induced leaves and the constituent leaves of the EV control plants at 24 hours, we observed that the metabolic specificity of many metabolites after herbivory insects has a significant increasing trend (Figure 4C). The specific increase in Si value was detected only in defensive phenolamides, but no increase in Si value was detected in other phenols and unknown metabolites coexisting in this module. This is a specialized model, which is related to the OD theory. The main predictions of metabolic changes caused by herbivores are consistent. In order to test whether this particularity of the phenolamide spectrum was induced by OS-specific ET, we plotted the metabolite Si index and caused a differential expression value between OSM and OSS1 in the EV and sETR1 genotypes (Figure 4D ). In sETR1, the phenamide-induced difference between OSM and OSS1 was greatly reduced. The bi-clustering method was also applied to MS/MS data collected in strains with insufficient JA to infer the main MS/MS modules related to JA-regulated metabolic specialization (Figure S6).
(A) The clustering results of 585 MS/MS based on shared fragment (NDP similarity) and shared neutral loss (NL similarity) result in the module (M) being consistent with the known compound family, or by unknown or poorly metabolized Metabolite composition. Next to each module, the metabolite (MS/MS) specific (Si) distribution is shown. (B) Modular molecular network: Nodes represent MS/MS and edges, NDP (red) and NL (blue) MS/MS scores (cut-off,> 0.6). The graded metabolite specificity index (Si) colored based on the module (left) and mapped to the molecular network (right). (C) Module M1 of EV plant in constitutive (control) and induced state (simulated herbivore) at 24 hours: molecular network diagram (Si value is the node size, defensive phenolamide is highlighted in blue). (D) The M1 molecular network diagram of the spectrum line sETR1 with impaired EV and ET perception: the phenolic compound represented by the green circle node, and the significant difference (P value) between W + OSM and W + OSS1 treatments as Node size. CP, N-caffeoyl-tyrosine; CS, N-caffeoyl-spermidine; FP, N-ferulic acid ester-uric acid; FS, N-ferulyl-spermidine; CoP, N’, N “-Coumarolyl-tyrosine; DCS, N’, N”-dicaffeoyl-spermidine; CFS, N’, N”-caffeoyl, feruloyl-spermidine; Lycium barbarum in wolfberry Son; Nick. O-AS, O-acyl sugar.
We further extended the analysis from a single attenuated Nepenthes genotype to natural populations, where strong intraspecific changes in herbivorous JA levels and specific metabolite levels have been previously described in natural populations (26). Use this data set to cover 43 germplasms. These germplasms consist of 123 species of plants from N. pallens. These plants were taken from seeds collected in different native habitats in Utah, Nevada, Arizona, and California (Figure S7), we calculated the metabolome diversity (here called population level) β diversity) and the specialization caused by OSM. Consistent with previous studies, we observed a wide range of metabolic changes along the Hj and δj axes, indicating that germplasms have significant differences in the plasticity of their metabolic responses to herbivores (Figure S7). This organization is reminiscent of previous observations about the dynamic range of JAs changes caused by herbivores, and has maintained a very high value in a single population (26, 36). By using JA and JA-Ile to test the overall level correlation between Hj and δj, we found that there is a significant positive correlation between JA and the metabolome β diversity and specialization index (Figure S7). This suggests that the herbivore-induced heterogeneity in JAs induction detected at the population level may be due to key metabolic polymorphisms caused by selection from insect herbivores.
Previous studies have shown that tobacco types differ greatly in type and relative dependence on induced and constitutive metabolic defenses. It is believed that these changes in anti-herbivore signal transduction and defense capabilities are regulated by insect population pressure, plant life cycle, and defense production costs in the niche where a given species grows. We studied the consistency of leaf metabolome remodeling induced by herbivores of six Nicotiana species native to North America and South America. These species are closely related to Nepenthes North America, namely Nicolas Bociflo. La, N. nicotinis, Nicotiana n. attenuated grass, Nicotiana tabacum, linear tobacco, tobacco (Nicotiana spegazzinii) and tobacco leaf tobacco (Nicotiana obtusifolia) (Figure 5A) (37). Six of these species, including the well-characterized species N. please, are annual plants of the petunia clade, and obtusifolia N. are perennials of the sister clade Trigonophyllae (38). Subsequently, W + W, W + OSM and W + OSS1 induction were performed on these seven species to study the species-level metabolic rearrangement of insect feeding.
(A) A bootstrap phylogenetic tree based on maximum likelihood [for nuclear glutamine synthesis (38)] and the geographical distribution of seven closely related Nicotiana species (different colors) (37). (B) A scatter plot of specialized diversity for the metabolic profiles of seven Nicotiana species (939 MS/MS; data file S1). At the species level, metabolome diversity is negatively correlated with the degree of specialization. The analysis of the species-level correlation between metabolic diversity and specialization and JA accumulation is shown in Figure 2. S9. Color, different types; triangle, W + OSS1; rectangle, W + OSM; (C) Nicotiana JA and JA-Ile dynamics are ranked according to OS excitation amplitude (two-way ANOVA and Tukey HSD post-multiple comparison, * P <0.05, ** P <0.01 and * ** For comparison of W + OS and W + W, P <0.001). Box plot of (D) diversity and (E) specialization of each species after simulating herbivorous and methyl JA (MeJA). The asterisk indicates the significant difference between W + OS and W + W or lanolin plus W (Lan + W) or Lan plus MeJA (Lan + MeJa) and Lan control (two-way analysis of variance, followed by Tukey’s HSD post hoc multiple comparison , *P<0.05, **P<0.01 and ***P<0.001).
Using the dual cluster method, we identified 9 modules of 939 MS/MS (data file S1). The composition of MS/MS reconfigured by different treatments varies greatly among different modules between species (Figure S8). Visualizing Hj (referred to here as species-level γ-diversity) and δj reveals that different species aggregate into very different groups in the metabolic space, where species-level division is usually more prominent than excitation. With the exception of N. linear and N. obliquus, they exhibit a wide dynamic range of induction effects (Figure 5B). In contrast, species such as N. purpurea and N. obtusifolia have a less obvious metabolic response to treatment, but the metabolome is more diverse. The species-specific distribution of the induced metabolic response resulted in a significant negative correlation between specialization and gamma diversity (PCC = -0.46, P = 4.9×10-8). OS-induced changes in JAs levels are positively correlated with metabolome specialization, and negatively correlated with the metabolic gamma diversity exhibited by each species (Figure 5B and Figure S9). It is worth noting that the species colloquially referred to as “signal response” species in Figure 5C, such as Nepenthes nematodes, Nepenthes nepenthes, Nepenthes acute, and Nepenthes attenuated, caused significant signs at 30 minutes. The recent OS-specific JA and JA-Ile outbreaks, while other bacteria called “signal unresponsive”, such as Nepenthes mills, Nepenthes powdery and N. obtusifolia only show JA-Ile Edge induction without any OS specificity (Figure 5C). At the metabolic level, as mentioned above, for attenuated Nepenthes, the signal-responsive substances showed OS specificity and significantly increased δj, while reducing Hj. This OS-specific priming effect was not detected in species classified as signal non-reactive species (Figure 5, D and E). OS-specific metabolites are shared more frequently between signal-responsive species, and these signal clusters cluster with species with weaker signal responses, while species with weaker signal responses show less interdependence (Figure S8 ). This result indicates that the OS-specific induction of JAs and the OS-specific reconfiguration of the downstream metabolome are coupled at the species level.
Next, we used a lanolin paste containing methyl JA (MeJA) to treat plants to investigate whether these coupling modes are restricted by the availability of JA applied by exogenous JA, which will be in the cytoplasm of plants. Rapid deesterification is JA. We found the same trend of the gradual change from signal-responsive species to signal-non-responsive species caused by the continuous supply of JA (Figure 5, D and E). In short, MeJA treatment strongly reprogrammed the metabolomes of linear nematodes, N. obliquus, N. aquaticus, N. pallens, and N. mikimotoi, resulting in a significant increase in δj and a decrease in Hj. N. purpurea only showed an increase in δj, but not Hj. N. obtusifolia, which has previously been shown to accumulate extremely low levels of JAs, also responds poorly to MeJA treatment in terms of metabolome reconfiguration. These results indicate that JA production or signal transduction is physiologically restricted in signal-unresponsive species. To test this hypothesis, we studied the four species (N. pallens, N. mills, N. pink and N. microphylla) induced by W + W, W + OSMs and W + OSS1 Transcriptome (39). Consistent with the pattern of metabolome remodeling, the species are well separated in the transcriptome space, among which N. attenuated showed the highest OS-induced RDPI, while N. gracilis had the lowest (Figure 6A). However, it was found that the transcriptome diversity induced by N. oblonga was the lowest among the four species, contrary to the highest metabonomic diversity of N. oblonga previously shown in seven species. Previous studies have shown that a set of genes related to early defense signals, including JA signals, explains the specificity of early defense responses induced by herbivore-related elicitors in Nicotiana species (39). Comparing the JA signaling pathways between these four species revealed an interesting pattern (Figure 6B). Most genes in this pathway, such as AOC, OPR3, ACX and COI1, showed relatively high levels of induction in these four species. However, a key gene, JAR4, converts JA into its biologically active form of JA-Ile accumulated transcripts, and its transcription level is very low, especially in N. mills, Nepenthes pieris and N. microphylla. In addition, only the transcript of another gene AOS was not detected in N. bifidum. These changes in gene expression may be responsible for the extreme phenotypes induced by the low JA production in signal anergic species and the induction of N. gracilis.
(A) Information theory analysis of the reprogramming of early transcriptional responses of four closely related tobacco species sampled 30 minutes after herbivory induction. RDPI is calculated by comparing the leaves induced by the herbivore OS with the wound control. The colors indicate different species, and the symbols indicate different treatment methods. (B) Analysis of gene expression in JA signaling pathways among four species. The simplified JA path is shown next to the box plot. Different colors indicate different processing methods. The asterisk indicates that there is a significant difference between the W + OS treatment and the W + W control (for the Student’s t-test for pairwise differences, *P<0.05, **P<0.01 and ***P<0.001). OPDA, 12-oxophytodienoic acid; OPC-8: 0,3-oxo-2(2′(Z)-pentenyl)-cyclopentane-1-octanoic acid.
In the last part, we studied how insect species-specific remodeling of the metabolome of different plant species can be resistant to herbivores. Previous research emphasized the Nicotiana genus. Their resistance to Ms and larvae differ greatly (40). Here, we studied the connection between this model and their metabolic plasticity. Using the above four tobacco species, and testing the correlation between the diversity and specialization of the metabolome caused by herbivores and the resistance of plants to Ms and Sl, we found the resistance, diversity and specialization to the generalist Sl All are positively correlated, while the correlation between resistance to expert ladies and specialization is weak, and the correlation with diversity is not significant (Figure S10). With regard to S1 resistance, both the attenuated N. chinensis and N. gracilis, which were previously shown to exhibit both JA signal transduction levels and metabolome plasticity, had greatly different responses to herbivore induction, and they also showed similar high resistance. Sex.
In the past sixty years, plant defense theory has provided a theoretical framework, based on which researchers have predicted a considerable number of evolution and functions of plant specialized metabolites. Most of these theories do not follow the normal procedure of strong inferences (41). They propose key predictions (3) at the same level of analysis. When the testing of key predictions allows specific theories to be analyzed, this will make the The field is moving forward. Be supported, but reject others (42). Instead, the new theory makes predictions at different levels of analysis and adds a new layer of descriptive considerations (42). However, the two theories proposed at the functional level, MT and OD, can be easily explained as important predictions of specialized metabolic changes caused by herbivores: OD theory believes that changes in specialized metabolic “space” are highly directional . The MT theory believes that these changes will be non-directional and randomly located in the metabolic space, and tend to have high defense value metabolites. Previous examinations of OD and MT predictions have been tested using a narrow set of a priori “defense” compounds. These metabolite-centric tests preclude the ability to analyze the extent and trajectory of metabolome reconfiguration during herbivory, and do not allow testing within a consistent statistical framework to require key predictions that can be considered as a whole Quantify changes in plant metabolome. Here, we used the innovative technology in metabolomics based on computational MS and performed deconvolution MS analysis in the general currency of information theory descriptors to test the distinction between the two proposed at the global metabolomics level. The key prediction of this theory. Information theory has been applied in many fields, especially in the context of biodiversity and nutrient flow research (43). However, as far as we know, this is the first application used to describe the metabolic information space of plants and solve ecological problems related to temporary metabolic changes in response to environmental cues. In particular, the ability of this method lies in its ability to compare patterns within and between plant species to examine how herbivores have evolved from different species to inter-species macroevolutionary patterns at different levels of evolution. Metabolism.
Principal component analysis (PCA) converts a multivariate data set into a dimensionality reduction space so that the main trend of the data can be explained, so it is usually used as an exploratory technique to parse the data set, such as deconvolution metabolome. However, dimensionality reduction will lose part of the information content in the data set, and PCA cannot provide quantitative information about characteristics that are particularly relevant to ecological theory, such as: how herbivores reconfigure diversity in specialized fields (for example, richness, distribution) And abundance) metabolites? Which metabolites are predictors of the induced state of a given herbivore? From the perspective of specificity, diversity and inducibility, the information content of the leaf-specific metabolite profile is decomposed, and it is found that the eating of herbivores can activate specific metabolism. Unexpectedly, we observed that, as described in the information theory indicators implemented, the resulting metabolic situation has a large overlap after the attacks of the two herbivores (the night-fed generalist Sl) and the Solanaceae expert Ms. Although their feeding behavior and concentration are significantly different. Fatty acid-amino acid conjugate (FAC) initiator in OS (31). By using herbivore OS to treat standardized puncture wounds, simulated herbivore treatment also showed a similar trend. This standardized procedure for simulating the response of plants to herbivore attacks eliminates the confounding factors caused by changes in the eating behavior of herbivores, which lead to varying degrees of damage at different times (34). FAC, which is known to be the main cause of OSM, reduces JAS and other plant hormone responses in OSS1, while OSS1 reduces hundreds of times (31). However, OSS1 caused similar levels of JA accumulation compared to OSM. It has previously been demonstrated that the JA response in attenuated Nepenthes is very sensitive to OSM, where FAC can maintain its activity even if diluted 1:1000 with water (44). Therefore, compared with OSM, although the FAC in OSS1 is very low, it is sufficient to induce sufficient JA outbreak. Previous studies have shown that porin-like proteins (45) and oligosaccharides (46) can be used as molecular clues to trigger plant defense responses in OSS1. However, it is still unclear whether these elicitors in OSS1 are responsible for the accumulation of JA observed in the current study.
Although there are few studies describing the differential metabolic fingerprints caused by the application of different herbivores or exogenous JA or SA (salicylic acid) (47), no one has perturbed the herbivore species-specific perturbation in the plant grass network and its effects on the specific The overall impact of metabolism is systematically studied. personal information. This analysis further confirmed that the internal hormone network connection with other plant hormones other than JAs shapes the specificity of the metabolic reorganization caused by herbivores. In particular, we detected that ET caused by OSM was significantly greater than that caused by OSS1. This mode is consistent with more FAC content in OSM, which is a necessary and sufficient condition for triggering an ET burst (48). In the context of the interaction between plants and herbivores, the signaling function of ET on plant-specific metabolite dynamics is still sporadic and only targets a single compound group. In addition, most studies have used exogenous application of ET or its precursors or various inhibitors to study the regulation of ET, among which these exogenous chemical applications will produce many non-specific side effects. To our knowledge, this study represents the first large-scale systematic examination of the role of ET in the use of ET to produce and perceive impaired transgenic plants to coordinate plant metabolome dynamics. Herbivore-specific ET induction can ultimately modulate the metabolome response. The most significant is the transgenic manipulation of ET biosynthesis (ACO) and perception (ETR1) genes that revealed the herbivore-specific de novo accumulation of phenolamides. It has previously been shown that ET can fine-tune JA-induced nicotine accumulation by regulating putrescine N-methyltransferase (49). However, from a mechanical point of view, it is not clear how ET fine-tunes the induction of phenamide. In addition to the signal transduction function of ET, the metabolic flux can also be shunted to S-adenosyl-1-methionine to regulate the investment in polyaminophenol amides. S-adenosyl-1-methionine is ET and Common intermediate of polyamine biosynthetic pathway. The mechanism by which ET signal regulates the level of phenolamide needs further study.
For a long time, due to the large number of special metabolites of unknown structure, the intense attention to specific metabolic categories has been unable to strictly assess the temporal changes of metabolic diversity after biological interactions. At present, based on information theory analysis, the main result of MS/MS spectrum acquisition based on unbiased metabolites is that herbivores eating or simulating herbivores continue to reduce the overall metabolic diversity of the leaf metabolome while increasing its degree of specialization. This temporary increase in metabolome specificity caused by herbivores is associated with a synergistic increase in transcriptome specificity. The feature that contributes the most to this larger metabolome specialization (having a higher Si value) is the special metabolite with the previously characterized herbivorous function. This model is consistent with the prediction of OD theory, but the prediction of MT related to the randomness of metabolome reprogramming is not consistent. However, this data is also consistent with the prediction of the mixed model (best MT; Figure 1B), because other uncharacterized metabolites with unknown defense functions may still follow a random Si distribution.
A noteworthy pattern further captured by this research is that from the micro-evolution level (single plant and tobacco population) to a larger evolutionary scale (closely related tobacco species), different levels of evolutionary organization are in the “best defense” There are significant differences in the abilities of herbivores. Moore et al. (20) and Kessler and Kalske (1) independently proposed to convert the three functional levels of biodiversity originally distinguished by Whittaker (50) into the constitutive and induced temporal changes of chemical diversity; these authors neither summarized The procedures for large-scale metabolome data collection also do not outline how to calculate metabolic diversity from these data. In this study, minor adjustments to Whittaker’s functional classification will consider α-metabolic diversity as the diversity of MS/MS spectra in a given plant, and β-metabolic diversity as the basic intraspecific metabolism of a group of populations Space, and γ-metabolic diversity will be an extension of the analysis of similar species.
The JA signal is essential for a wide range of herbivore metabolic responses. However, there is a lack of rigorous quantitative testing of the contribution of intraspecific regulation of JA biosynthesis to metabolome diversity, and whether JA signal is a general site for stress-induced metabolic diversification on a higher macroevolutionary scale is still elusive. We observed that the herbivorous nature of Nepenthes herbivorous induces metabolome specialization and the variation of metabolome specialization within the population of Nicotiana species and among closely related Nicotiana species is systematically positively correlated with JA signaling. In addition, when the JA signal is impaired, the metabolic specificity induced by a single genotype herbivore will be cancelled (Figure 3, C and E). Since the metabolic spectrum changes of the naturally attenuated Nepenthes populations are mostly quantitative, the changes in the metabolic β diversity and specificity in this analysis may be largely caused by the strong excitation of metabolite-rich compound categories. These compound classes dominate part of the metabolome profile and lead to a positive correlation with JA signals.
Because the biochemical mechanisms of the tobacco species closely related to it are very different, the metabolites are specifically identified in the qualitative aspect, so it is more analytical. Information theory’s processing of the captured metabolic profile reveals that herbivorous induction exacerbates the trade-off between metabolic gamma diversity and specialization. The JA signal plays a central role in this trade-off. The increase in metabolome specialization is consistent with the main OD prediction and is positively correlated with JA signal, while JA signal is negatively correlated with metabolic gamma diversity. These models indicate that the OD capacity of plants is mainly determined by the plasticity of JA, whether on a microevolutionary scale or on a larger evolutionary scale. Exogenous JA application experiments that circumvent JA biosynthesis defects further reveal that closely related tobacco species can be distinguished into signal-responsive and signal-non-responsive species, just like their mode of JA and metabolome plasticity induced by herbivores. Signal non-responsive species cannot respond due to their inability to produce endogenous JA and are therefore subject to physiological limitations. This may be caused by mutations in some key genes in the JA signaling pathway (AOS and JAR4 in N. crescens) of. This result highlights that these interspecies macroevolutionary patterns may be mainly driven by changes in internal hormone perception and responsiveness.
In addition to the interaction between plants and herbivores, the exploration of metabolic diversity is related to all important theoretical advances in the research of biological adaptation to the environment and the evolution of complex phenotypic traits. With the increase in the amount of data acquired by modern MS instruments, hypothesis testing on metabolic diversity can now transcend individual/category metabolite differences and perform global analysis to reveal unexpected patterns. In the process of large-scale analysis, an important metaphor is the idea of ​​conceiving meaningful maps that can be used to explore data. Therefore, an important result of the current combination of unbiased MS/MS metabolomics and information theory is that it provides a simple metric that can be used to construct maps to browse the metabolic diversity on different taxonomic scales. It is the basic requirement of this method. The study of micro/macro evolution and community ecology.
At the macro-evolutionary level, the core of the plant-insect co-evolution theory of Ehrlich and Raven (51) is to predict that the variation of interspecies metabolic diversity is the cause of the diversification of plant lineages. However, in the fifty years since the publication of this seminal work, this hypothesis has been rarely tested (52). This is largely due to the phylogenetic characteristics of comparable metabolic characteristics across long-distance plant lineages. The rarity can be used to anchor target analysis methods. The current MS/MS workflow processed by information theory quantifies the MS/MS structural similarity of unknown metabolites (without prior metabolite selection) and converts these MS/MSs into a set of MS/MS, thus in professional metabolism These macro-evolutionary models are compared in classification scale. Simple statistical indicators. The process is similar to phylogenetic analysis, which can use sequence alignment to quantify the rate of diversification or character evolution without prior prediction.
At the biochemical level, the screening hypothesis of Firn and Jones (53) shows that metabolic diversity is maintained at different levels to provide raw materials to exert the biological activities of previously unrelated or substituted metabolites. Information theory methods provide a framework in which these metabolite-specific evolutionary transitions that occur during metabolite specialization can be quantified as part of the proposed evolutionary screening process: biologically active adaptation from low specificity to high specificity Inhibited metabolites of a given environment.
All in all, in the early days of molecular biology, important plant defense theories were developed, and deductive hypothesis-driven methods are widely considered to be the only means of scientific progress. This is largely due to the technical limitations of measuring the entire metabolome. Although hypothesis-driven methods are particularly useful in choosing other causal mechanisms, their ability to advance our understanding of biochemical networks is more limited than the computational methods currently available in contemporary data-intensive science. Therefore, theories that cannot be predicted are far beyond the scope of available data, so the hypothetical formula/test cycle of progress in the research field cannot be abolished (4). We foresee that the computational workflow of metabolomics introduced here can rekindle interest in the recent (how) and final (why) issues of metabolic diversity, and contribute to a new era of theoretically guided data science. The era re-examined the important theories that inspired previous generations.
Direct herbivore feeding is carried out by raising a second instar larva or Sl larva on a single Pale-colored pitcher plant leaf of a single rose blooming plant, with 10 plant replicates per plant. The insect larvae were clamped with clamps, and the remaining leaf tissue was collected 24 and 72 hours after infection and quick-frozen, and the metabolites were extracted.
Simulate herbivorous treatment in a highly synchronized manner. The method is to use fabric pattern wheels to puncture three rows of thorns on each side of the midrib of the three fully expanded leaves of the plant during the growth stage of the fabric garland, and immediately apply 1:5 Diluted Ms. Or use gloved fingers to insert S1 OS into the puncture wound. Harvest and process a leaf as described above. Use the previously described method to extract primary metabolites and plant hormones (54).
For exogenous JA applications, the three petiole leaves of each of the six rose blooming plants of each species are treated with 20μl of lanolin paste containing 150μg MeJA (Lan + MeJA), and 20μl of lanolin plus wound treatment (Lan + W), or use 20μl pure lanolin as a control. The leaves were harvested 72 hours after treatment, snap-frozen in liquid nitrogen, and stored at -80°C until use.
Four JA and ET transgenic lines, namely irAOC (36), irCOI1 (55), irACO and sETR1 (48), have been identified in our research group. irAOC strongly showed a decrease in JA and JA-Ile levels, while irCOI1 was not sensitive to JAs. Compared with EV, JA-Ile accumulation increased. Similarly, irACO will reduce the production of ET, and compared with EV, sETR1, which is insensitive to ET, will increase the production of ET.
A photoacoustic laser spectrometer (Sensor Sense ETD-300 real-time ET sensor) is used to perform ET measurement non-invasively. Immediately after treatment, half of the leaves were cut and transferred to a 4-ml sealed glass vial, and the headspace was allowed to accumulate within 5 hours. During the measurement, each vial was flushed with a stream of 2 liters/hour of pure air for 8 minutes, which had previously passed through a catalyst provided by Sensor Sense to remove CO2 and water.
The microarray data was originally published in (35) and saved in the National Center for Biotechnology Information (NCBI) Gene Expression Comprehensive Database (accession number GSE30287). The data corresponding to the leaves caused by the W + OSMs treatment and the undamaged control were extracted for this study. The raw intensity is log2. Before statistical analysis, the baseline was converted and normalized to its 75th percentile using the R software package.
The original RNA sequencing (RNA-seq) data of Nicotiana species was retrieved from the NCBI Short Reading Archives (SRA), project number is PRJNA301787, which was reported by Zhou et al. (39) and proceed as described in (56). The raw data processed by W + W, W + OSM and W + OSS1 corresponding to Nicotiana species were selected for analysis in this study, and processed in the following manner: First, the raw RNA-seq readings were converted into FASTQ format. HISAT2 converts FASTQ to SAM, and SAMtools converts SAM files into sorted BAM files. StringTie is used to calculate gene expression, and its expression method is that there are fragments per thousand base fragments per million sequenced transcription fragments.
The Acclaim chromatographic column (150 mm x 2.1 mm; particle size 2.2μm) used in the analysis and the 4 mm x 4 mm guard column consist of the same material. The following binary gradient is used in the Dionex UltiMate 3000 Ultra High Performance Liquid Chromatography (UHPLC) system: 0 to 0.5 minutes, isocratic 90% A [deionized water, 0.1% (v/v) acetonitrile and 0.05% formic acid], 10% B (Acetonitrile and 0.05% formic acid); 0.5 to 23.5 minutes, the gradient phase is 10% A and 90% B, respectively; 23.5 to 25 minutes, isocratic 10% A and 90% B. The flow rate is 400μl/min. For all MS analyses, inject the column eluent into a quadrupole and time-of-flight (qTOF) analyzer equipped with an electrospray source operating in positive ionization mode (capillary voltage, 4500 V; capillary outlet 130 V ; Drying temperature 200°C; drying airflow 10 liters/min).
Perform MS / MS fragment analysis (hereinafter referred to as MS / MS) that is irrelevant or indistinguishable from the data to obtain structural information about the overall detectable metabolic profile. The concept of the indiscriminate MS/MS method relies on the fact that the quadrupole has a very large mass isolation window [therefore, consider all mass-to-charge ratio (m/z) signals as fragments]. For this reason, because the Impact II instrument was unable to create a CE tilt, several independent analyses were performed using increased collision-induced dissociation collision energy (CE) values. In short, first analyze the sample by UHPLC-electrospray ionization/qTOF-MS using single mass spectrometry mode (low fragmentation conditions generated by in-source fragmentation), scanning from m/z 50 to 1500 at a repetition frequency of 5 Hz. Use nitrogen as the collision gas for MS/MS analysis, and perform independent measurements at the following four different collision-induced dissociation voltages: 20, 30, 40, and 50 eV. Throughout the measurement process, the quadrupole has the largest mass isolation window, from m/z 50 to 1500. When the front body m/z and isolation width experiment is set to 200, the mass range is automatically activated by the instrument’s operating software and 0 Da. Scan for mass fragments as in single mass mode. Use sodium formate (50 ml isopropanol, 200 μl formic acid and 1 ml 1M NaOH aqueous solution) for mass calibration. Using Bruker’s high-precision calibration algorithm, the data file is calibrated after running the average spectrum in a given time period. Use the export function of Data Analysis v4.0 software (Brook Dalton, Bremen, Germany) to convert the raw data files into NetCDF format. The MS/MS data set has been saved in the open metabolomics database MetaboLights (www.ebi.ac.uk) with the accession number. MTBLS1471.
MS/MS assembly can be realized through correlation analysis between MS1 ​​and MS/MS quality signals for low and high collision energy and newly implemented rules. The R script is used to realize the correlation analysis of the distribution of the precursor to the product, and the C# script (https://github.com/MPI-DL/indiscriminant-MS-MS-assembly-pipeline) is used to implement the rules.
In order to reduce false positive errors caused by background noise and false correlation caused by detecting certain m/z features in only a few samples, we use the “filled peak” function of the R package XCMS (for background noise correction) Should be used to replace the “NA” (undetected peak) intensity. When the fill peak function is used, there are still many “0″ intensity values ​​in the data set that will affect the correlation calculation. Then, we compare the data processing results obtained when the filled peak function is used and when the filled peak function is not used, and calculate the background noise value based on the average corrected estimated value, and then replace these 0 intensity values ​​with the calculated background value. We also only considered features whose intensity exceeded three times the background value and regarded them as “true peaks.” For PCC calculations, only the m/z signals of the sample precursor (MS1) and fragment data sets with at least eight true peaks are considered.
If the intensity of the precursor quality feature in the entire sample is significantly correlated with the reduced intensity of the same quality feature that is subjected to low or high collision energy, and the feature is not labeled as an isotope peak by CAMERA, it can be further defined. Then, by calculating all possible precursor-product pairs within 3 s (the estimated retention time window for peak retention), the correlation analysis is performed. Only when the m/z value is lower than the precursor value and MS/MS fragmentation occurs in the same sample location in the data set as the precursor from which it is derived, is it considered a fragment.
Based on these two simple rules, we exclude the specified fragments with m/z values ​​greater than the m/z of the identified precursor, and based on the sample position where the precursor appears and the specified fragment. It is also possible to select the quality features generated by many in-source fragments generated in MS1 ​​mode as candidate precursors, thereby generating redundant MS/MS compounds. In order to reduce this data redundancy, if the NDP similarity of the spectra exceeds 0.6, and they belong to the chromatogram “pcgroup” annotated by CAMERA, we will merge them. Finally, we merge all four CE results associated with the precursor and fragments into the final deconvoluted composite spectrum by selecting the highest intensity peak among all candidate peaks with the same m/z value at different collision energies. The subsequent processing steps are based on the concept of composite spectrum and take into account the different CE conditions required to maximize the probability of fragmentation, because some fragments can only be detected under a specific collision energy.
RDPI (30) was used to calculate the inducibility of the metabolic profile. The metabolic spectrum diversity (Hj index) is derived from the abundance of MS/MS precursors using the Shannon entropy of the MS/MS frequency distribution using the following equation described by Martínez et al. (8). Hj = −∑i = 1mPijlog2(Pij) where Pij corresponds to the relative frequency of the i-th MS/MS in the j-th sample (j = 1, 2,…, m) (i = 1, 2, …, m) t).
Metabolic specificity (Si index) is defined as the expression identity of a given MS/MS in relation to the frequency between the samples being considered. MS/MS specificity is calculated as Si = 1t (∑j = 1tPijPilog2PijPi)
Use the following formula to measure the metabolome-specific δj index of each j sample, and the average of MS/MS specificity δj = ∑i = 1mPijSi
MS/MS spectra are aligned in pairs, and the similarity is calculated based on the two scores. First, using standard NDP (also known as cosine correlation method), use the following equation to score the segment similarity between spectra NDP = (∑iS1 & S2WS1, iWS2, i) 2∑iWS1, i2∑iWS2, i2 where S1 and S2 Correspondingly, for spectrum 1 and spectrum 2, as well as WS1, i and WS2, i represents the weight based on the peak intensity that the difference of the i-th common peak between the two spectra is less than 0.01 Da. The weight is calculated as follows: W = [peak intensity] m [quality] n, m = 0.5, n = 2, as suggested by MassBank.
A second scoring method was implemented, which involved analyzing the shared NL between MS/MS. To this end, we used the 52 NL lists frequently encountered during the MS fragmentation process in tandem, and the more specific NL (data file S1) that has been previously annotated for the MS/MS spectrum of the secondary metabolites of the weakened Nepenthes species ( 9, 26). Create a binary vector of 1 and 0 for each MS/MS, corresponding to the current and non-existent of some NL respectively. Based on the Euclidean distance similarity, the NL similarity score is calculated for each pair of binary NL vectors.
To perform dual clustering, we used the R package DiffCoEx, which is based on an extension of Weighted Gene Co-expression Analysis (WGCNA). Using the NDP and NL scoring matrices of MS/MS spectra, we used DiffCoEx to calculate the comparative correlation matrix. Binary clustering is performed by setting the “cutreeDynamic” parameter to method = “hybrid”, cutHeight = 0.9999, deepSplit = T, and minClusterSize = 10. The R source code of DiffCoEx was downloaded from additional file 1 by Tesson et al. (57); The required R WGCNA software package can be found in https://horvath.genetics.ucla.edu/html/CoexpressionNetwork/Rpackages/WGCNA.
In order to perform MS/MS molecular network analysis, we calculated paired spectral connectivity based on NDP and NL similarity types, and used Cytoscape software to visualize network topology using organic layout in the CyFilescape yFiles layout algorithm extension application.
Use R version 3.0.1 to perform statistical analysis on the data. Statistical significance was assessed using two-way analysis of variance (ANOVA), followed by Tukey’s honestly significant difference (HSD) post-hoc test. In order to analyze the difference between the herbivorous treatment and the control, the two-tailed distribution of the two groups of samples with the same variance was analyzed using Student’s t test.
For supplementary materials for this article, please see http://advances.sciencemag.org/cgi/content/full/6/24/eaaz0381/DC1
This is an open access article distributed under the terms of the Creative Commons Attribution-Non-Commercial License, which allows the use, distribution and reproduction in any medium, as long as the final use is not for commercial gain and the premise is that the original work is correct. Reference.
Note: We only ask you to provide your email address so that the person you recommend to the page knows that you want them to see the email and that it is not spam. We will not capture any email addresses.
This question is used to test whether you are a visitor and prevent automatic spam submission.
Information theory provides a universal currency for the comparison of special metabolomes and the prediction of test defense theories.
Information theory provides a universal currency for the comparison of special metabolomes and the prediction of test defense theories.
©2021 American Association for the Advancement of Science. all rights reserved. AAAS is a partner of HINARI, AGORA, OARE, CHORUS, CLOCKSS, CrossRef and COUNTER. ScienceAdvances ISSN 2375-2548.


Post time: Feb-22-2021