A comparative UHPLC-Q/TOF–MS-based eco-metabolomics approach reveals temperature adaptation of four Nepenthes species

Nepenthes, as the largest family of carnivorous plants, is found with an extensive geographical distribution throughout the Malay Archipelago, specifically in Borneo, Philippines, and Sumatra. Highland species are able to tolerate cold stress and lowland species heat stress. Our current understanding on the adaptation or survival mechanisms acquired by the different Nepenthes species to their climatic conditions at the phytochemical level is, however, limited. In this study, we applied an eco-metabolomics approach to identify temperature stressed individual metabolic fingerprints of four Nepenthes species: the lowlanders N. ampullaria, N. rafflesiana and N. northiana, and the highlander N. minima. We hypothesized that distinct metabolite regulation patterns exist between the Nepenthes species due to their adaptation towards different geographical and altitudinal distribution. Our results revealed not only distinct temperature stress induced metabolite fingerprints for each Nepenthes species, but also shared metabolic response and adaptation strategies. The interspecific responses and adaptation of N. rafflesiana and N. northiana likely reflected their natural habitat niches. Moreover, our study also indicates the potential of lowlanders, especially N. ampullaria and N. rafflesiana, to produce metabolites needed to deal with increased temperatures, offering hope for the plant genus and future adaption in times of changing climate.


Individual metabolomic fingerprints of the four Nepenthes species. The metabolomes of the four
Nepenthes species, subjected to varying temperature regimes, displayed significant differences (p < 0.01; Supplementary Table S2). A dendrogram based on pearson distances and average clustering showed very distinct grouping of N. ampullaria and N. minima under all three temperature conditions compared to N. northiana and N. rafflesiana which were a little bit more mixed ( Fig. 2A). Principle components 1 and 2 derived from the PLS-DA, showed the total variance among the species at 39.7% (Fig. 2B). Despite grouping distinctly on its own, Figure 1. Pie chart depicting percentages of the 89 putatively identified metabolites from each of the metabolite groups, as well as the regulation of the metabolites of each group in response to the different temperature conditions. Each of the layer indicate a Nepenthes species, from inner layer to outer layer: N. minima, ampullaria, northiana, rafflesiana. NE no effect/particular pattern, LL metabolites expressed highest at lowland condition, HL metabolites expressed highest at highland condition.
Universal metabolite response to temperature stress. Our result showed that the metabolites were greatly affected by both lowland. Heat stress and highland cold stress conditions. While the metabolites expressed differed significantly among the four species (Fig. 3), high or lowland stress also led to a similar response in metabolite regulation across all our species ( Fig. 4; Supplementary Table S4). Adenine, berberastine and 1-naphthoic acid were, for example, all expressed the highest under highland cold stress, whereas l-tryptophan (except N. rafflesiana), 18-oxononadecanoic acid, olealdehyde and indole-3-acrylic acid were all expressed the highest under lowland heat stress ( Fig. 3; Supplementary Table S4). Interestingly, a flavone baicalein together with its isomers showed the highest accumulation at both highland and lowland conditions (Supplementary Table S4). Among the identified compounds, certain groups showed consistent expression among all the Nepenthes sp. such as purine bases (highest expression at highland condition), fatty acyls, amino acid, and indoles (highest expression level at lowland condition). Biosynthetic pathways and metabolic networks. We identified 14 metabolites that are involved in 32 metabolic pathways, including biosynthesis of phenylpropanoid and flavones derivatives, flavonols, flavonoids, amino acids, secondary metabolites and lignins, as well as coniferin metabolism (Table 1). A metabolic network was created to summarize the major heat and cold stresses adapting strategies found in the 4 Nepenthes species (Fig. 5).

Discussion
Till today, we have no clear understanding on the adaptation or survival mechanisms acquired by the different Nepenthes species to their climatic conditions. The distinctiveness of N. ampullaria and N. rafflesiana (capable to inhabit lowland and highland altitudes), and N. minima (unique highlander that is able to grow at lowland conditions), as well as N. northiana from an extreme habitat (limestone vegetation), made them the target species for the present study. We aimed to shed light on their unique adaptation mechanisms by studying their metabolomes in response to different environmental conditions. For the purpose of this study, we considered temperature to be the main environmental difference between highland (with temperatures down to 8 °C during the night) and lowland (up to 33 °C during the day) and exposed all four Nepenthes to temperatures representative of lowland, intermediate and highland conditions. Nepenthes are known to produce a wide range of metabolites but their metabolites have so far only been studied for their enzymatic properties and pharmacological potential [26][27][28] . The role of metabolites as regulatory/signaling agents, or in defense against biotic and abiotic stress (such as temperature changes), has been described in other plant species such as Arabidopsis, Triticum, and Vitis [29][30][31] . It is important to note that no additional experimental validation of the candidates has been carried out in this study. Nonetheless, our study provides the first insight into the response (at metabolome level) of Nepenthes plants originating from different climatic niches to temperature stress.
Universal adaptation strategy. Plants possess various acclimatization strategies to survive temperature stresses, which includes the accumulation of flavonoids, alterations in the membrane lipid composition and signaling, phytohormones regulation and signaling, increased in transpiration, ROS scavenger accumulation www.nature.com/scientificreports/ and many more [32][33][34][35][36][37][38][39][40] . Our results indicate that some of these strategies are important in Nepenthaceae as well (Figs. 2,3,5). Under lowland high temperature condition, Nepenthes seems to overcome the heat stress by increased phytohormone metabolism and related lateral root development. l-tryptophan is known to be involved in auxin indole-3-acetate synthesis 41,42 , while isoleucine is known to be the key amino acid that activates endogenous phytohormone jasmonic acid 43 . Phytohormone auxin was previously recorded to be positively affected by heat stress [44][45][46] and is responsible for plant root formation 47,48 . We hypothesize that the observed increased auxin metabolism might increase lateral root development 48 . This would potentially contribute to a reduction in water loss caused by increased transpiration, and provide a cooling effect for the plant. Similar strategies have been found in Arabidopsis thaliana, where high temperature increased water loss via transpiration and enhanced leaf cooling capacity 49 . Besides that, the importance of α-oxidation (olealdehyde) and ω-oxidation (18-oxononadecanoic acid) seem to play a role in the response to heat stress in Nepenthes plants as well. Both oxidation processes, which involve aldehyde dehydrogenase as the key enzyme, are known to yield unsaturated fatty acid (α-oxidation) and dicarboxylic fatty acid (ω-oxidation) in which the unsaturated fatty acid is responsible in maintaining the fluidity of plant cell membrane lipids and dicarboxylic acid is essential for the cuticle formation in plant [50][51][52][53] . A similar increased expression of aldehyde dehydrogenase genes was also observed in Arabidopsis plants under heat stress exposure 54 . Interestingly, the amino acid norvaline was previously recorded in cold and drought stress responses 29,55 and showed a universal contribution in our data ( Fig. 3; Supplementary Table S4).
Our results highlighted the potential utilization of energy metabolism (ATP) by Nepenthes plants to overcome the stress caused by exposure to low temperature. Both adenine and adenosine, which were expressed the highest level under cold condition, are known for their importance for proper functioning of plant cell, nucleic acid synthesis and energy metabolism 56,57 . Similar effects were also observed in other plants. For example, increase of energy in the form of ATP was recorded during the cold acclimation of winter rape plants when temperatures dropped down to 5 °C 58   Species-specific adaptation strategies. The metabolite regulation patterns showed that responses to temperature changes are linked to the plant habitats. Thermal specialization in tropical plant species indicate further inabilities for highland Nepenthes plants to adapt to changing environments [72][73][74] . The pyrophytic species Nepenthes minima, however, is unique as it adapted well in the greenhouse under lowland conditions. Its habitat is known to experience high temperatures up to 38 °C and prone to seasonal burning, with re-growing observed from the plant rootstock after the wildfires 75 . This could be the reason the species developed heat tolerance. Our temperature metabolome study revealed the ability of this species to manipulate production of amino acids and phytohormones in their heat stress adaptation. We discovered that this highland species possesses the same heat adaptation strategy like the lowlander N. rafflesiana, such as increased nitrogen storage, and polyamides synthesis via l-arginine accumulation 76 . Nitrogen storage is known to be crucial in heat shock protein production which plays a vital role in surviving heat stress 77 , while polyamines and l-arginine play a major role in activating ROS-scavenging enzymes under abiotic stress 76,78 .
Apart from the increased production of the phytohormone auxin observed in all 4 Nepenthes species studied, N. minima upregulated the production of two other phytohormones: jasmonic acid and trigonelline. The importance of the endogenous phytohormone jasmonic acid for heat and cold toleration in plants have been previously recorded 44,79,80 . Trigonelline was previously linked with various regulatory roles in relation to plant cell cycle regulation, nodulation, oxidative stress, as well as the growth of the plant 81-83 . Interestingly, a similar www.nature.com/scientificreports/ manipulation of the two hormones can also been observed in the lowlander N. ampullaria. However, instead of the lowland condition, N. ampullaria up-regulated their production under highland cold stress. As a lowlander, N. ampullaria was found to be capable of inhabiting a wide altitude range-including highland environments (up to 2100 masl 3 ). That is to say, the species even as a lowlander, is capable of tolerating low temperatures. Besides the two phytohormones mentioned above, we also observed an increased production of norbergenin, which possesses both antioxidant and ROS scavenging activities. It is likely that N. ampullaria uses it to protect themselves from cold stress induced oxidative damage 33,34 . A similar potential protection strategy was also observed in N. rafflesiana, which has been recorded growing at 1500 masl according to Adam et al. 84 .
While two metabolites involved in lignin biosynthesis pathways (coniferin and syringin) were detected in all four species, two of the lowland species (N. northiana and rafflesiana) displayed significantly higher accumulation of syringin under lowland conditions. N. northiana is commonly found on limestone hills, a harsh environment composed of calcium carbonate, alkaline pH and highly susceptible to drought 85 , while N. rafflesiana can be found in open habitats such as degraded, dry laterite and podsols 4,86 . In Matang, Kuching, Sarawak, Malaysia, N. rafflesiana has also been observed in open areas with direct exposure to sunlight and heat (anecdotal observations). Based on our data, it seems that in response/adaptation to the sun, heat, and drought, both N. northiana and rafflesiana developed a self-protection strategy by increasing lignification to inhibit water loss from plant tissue 87 . Similar adaptation mechanisms have been shown for Norway spruce, Ctenanthe setosa, and wheat 88-90 . Survival in a changing climate. Past studies revealed the importance of ecological adaptation of Nepenthes as the key determining factor driving, not only the diversification of pitcher morphology and their prey trapping mechanisms, but also the evolution of plant nutrient sequestration strategies 91,92 . In this study, we observed significant changes in the individual metabolomes of four Nepenthes species towards high and low temperature heat stress. Some pf the observed responses, such as the lignification, are verily likely linked to their habitat niches (Fig. 5).
Nepenthes are known to be susceptible towards climate change. Due to the narrow endemism geographical distribution of certain species, especially some highlanders that are confined to single mountain summits, they are at particularly high risk of species extinction 92,93 . Previously ecological niche modeling and maxent modeling have determined the climatically suitable area (habitat) for some species such as N. rafflesiana, N. tentaculate, N. macrophylla and N. lowii, via application of the climatic (such as rainfall and temperature) and edaphic (such as landform, soil association, soil parent material and soil suitability) variables 93,94 . The present eco-metabolomic study has highlighted the flexible responses (in terms of metabolite production) of the plant genus to adapt to environmental heat and cold stress. Our data does suggest that some lowlander species are indeed able to produce metabolites required to deal with increased high temperature stress. Hence, the future for selected species might not be as bleak as predicted.

Conclusion and final remark
Our eco-metabolomic study on the impact of lowland heat stress and highland cold stress revealed different metabolic fingerprints and potential adaptation strategies based on the species ecological niches. Our study demonstrated both universal (shared across all four species studied) and species-specific responses increase in selected metabolites under heat and cold stress. The metabolites found indicate the importance of several adaptation strategies ranging from increased ATP and ROS production, to the potential increased root development via auxin production. Lastly, we suggest more studies on plant metabolomes to achieve a better understanding of the adaption of Nepenthes (and other plant) species to their habitats.

Nepenthes. Four Nepenthes species namely Nepenthes minima, Nepenthes ampullaria, Nepenthes northiana
and Nepenthes rafflesiana were pre-adapted at lowland greenhouse for a period of 6 months, at least, before subjected into climatic chamber with control environment conditions. The Nepenthes in this study represented highland and lowland climate conditions (Table 2; for morphological details of the plant, please refer to Jebb and Martin 3 and Adam et al. 84 ). We hypothesized that distinct metabolite regulation patterns can be distinguished between the Nepenthes species due to their adaptation towards different geographical and altitudinal distribution. www.nature.com/scientificreports/ Plant materials and growth conditions. All plants were grown in a Pol-Eko-Aparatura climatic chamber with phytotron system (Model: KK 750 FIT P) in a mixture of cocopeat and perlite (at a ratio of 10:0.5; g/g Sample preparation. One (± 0.1) mg freeze-dried plant samples were weighted, ground, and exhaustedly extracted with 600 µL of solvent mixture of methanol:chloroform:ultrapure water (with resistance of 18.2 Ω cm −1 ) with 1% sodium chloride added (1:1:1 v/v/v). Mixtures were vortexed for 30 min at room temperature, followed by 30 min centrifugation at 3000 × g maintained at 4 °C. The lower layer was then transferred into a new borosilicate tube and vacuum dried using a speed concentrator. Dried extracts were then reconstituted using 400 µL of methanol and filtered using 0.2 μm PTFE membrane filter before subjected to liquid chromatography and mass spectrometry analysis.
Metabolome profiling. The extracted samples were profiled based on previously published method 95 .
Briefly, 10 µL of the samples were injected into Kinetex F5 (2.1 × 100 mm, 2.6 μm; Phenomenex, Torrance, CA, USA) for chromatographic separation via Vanquish™ Horizon UHPLC system (Thermo Fisher Scientific, USA). During analysis, the column was maintained at 40 °C with the flow rate of 600 µL/min. The mobile phase was composed of 2 solvents; solvent A (H 2 O-0.1% HCOOH-1% 10 mM NH 4 OAc) and solvent B (acetonitrile/ methanol [6:4 v/v]-1% of 0.1% HCOOH-1% 10 mM NH 4 OAc). The gradient elution program was initiated from 1 to 40% solvent B in 5 min, followed by 100% solvent B from 5.1 to 8 min and maintained for 2 min. Before injecting the next sample, the initial gradient was employed to condition the column for 3 min. UHPLC system was coupled with electrospray ionization Impact II QToF-mass spectrometry system (Bruker Daltonic, Germany). Mass-to-charge ratio (m/z) was set between 50 and 1500 for data acquisition. The heated electrospray ionization (ESI) was deployed at 4200 V for positive. Ion source gas temperature and flow rate was set at 300 °C and 12 L/min, respectively.
Mass calibration solution, 10 mM sodium formate was introduced post-column through a 6-port valve diverted between 0.1 and 0.3 min. Acquired m/z was calibrated against the introduced sodium formate, and then subsequently converted into a mzXML file format.
Metabolomics data processing. Raw data was exported in .mzXML format prior to MZmine 2 analysis 96 .
The software provides noise filtering, peak detection, alignment, normalization, alignment, and gap-filling and exported data in .csv format. Exported .csv files were used for multivariate analyses with MetaboAnalyst 4.0 97 . Metabolite features with missing values > 45% were removed, and missing values imputed using K-nearest neighbors 98 . The data was log transformed and pareto scaled. Metabolite features (ANOVA p < 0.01) between the 4 Nepenthes species under 3 environmental conditions further underwent compound matching and analysis. The .csv file with significantly changed metabolite features is provided in the supplementary file. The current analysis focused on polar the layer only as the non-polar layer demonstrated no significant difference (data excluded). All statistical analyses were performed on the positive ion data sets.
Metabolite annotation and identification. Metabolite features, including accurate m/z, possible chemical formula, and the fragmentation pattern, were queried against biological databases (highest priority was given to the database KEGG, followed by PubChem, and the others such as ChEBI and ChemSpider) using in silico fragmenter MetFrag 99 . The candidate was chosen based on the following criteria: (a) highest score with at least 80% match of the major fragment ions towards the databases (b) lowest relative mass deviation error when compared to the theoretical value (c) lowest relative mass deviation error from the fragment ions matched. To increase the accuracy of the identified metabolites, we cross checked the matched compound with earlier literature on similar compound especially in Nepenthes or in other plants. Pathway Tools Omics Viewer, developed by the Plant Metabolic Network (PMN), was used to identify highly correlated metabolites and to visualize the biosynthetic pathways 100 .

Statistical analyses.
Multiple comparison of mean tests, bar chart and pie chart were performed using Microsoft Excel. The data were pre-transformed using generalized logarithm transformation method via Meta-boAnalyst 4.0. A two-way ANOVA with Tukey's Post hoc analysis performed using PAST software 101 . Multivariate analyses including analysis of variance (ANOVA), partial least squares-discriminant analysis (PLS-DA), hierarchical cluster analysis and heat map were performed using MetaboAnalyst 4.0 97 . Venn diagram was created using Venny 2.1-developed at Bioinformatics for Genomics and Proteomics (BioinfoGP) 102 . Correlation values of the highly correlated metabolites with the biosynthetic pathways was performed using Omics Viewer 100 . Figure 5  www.nature.com/scientificreports/ Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creat iveco mmons .org/licen ses/by/4.0/.