2.1 Analysis and comparison of LC–MS fingerprints

The present study employed the high-resolution RP-UHPLC-ESI-QqTOF-MS to determine the intra-organ variation and relative abundances of metabolites extracted from leaf, stem, root and fruit of P. sarmentosum. The aim was to achieve high chromatographic resolution to detect as many individual chemical species in a single batch analysis as possible. First, to determine the appropriate ion detection mode, the SWATH experiments were performed in both positive and negative ionization modes (See Figure S1 for TIC of each organ in negative ion mode). According to the results of these experiments, LC–MS analysis in positive ion mode revealed essentially more peaks, better sensitivity and more rich and explicit structural information. This can be explained by the presence of numerous metabolites in the test samples more likely to be ionized in the positive ion mode than in negative one, as e.g. alkaloids. Major peaks in the negative ion mode represent flavonoids (Figure S1), however, this compound class also could be detected and analyzed in the positive ion mode. Therefore, only the positive ionization mode was used for all further series of MS analyses.

To monitor the reproducibility and instrument performance, aliquots of a quality control (QC) sample (i.e. a pooled extract prepared from the samples of all organs) were systematically measured between sample sequences. Figure S2 shows the close clustering of QC samples in a principal component analysis score plot. The stability of retention times, UHPLC performance and especially the high-resolution mass accuracy was also examined by the presence of orcinol (C7H8O2) and kinetin (C10H9N5O) as internal standards (IS). Both cost-efficient and routinely used ISs eluted at specific retention times, (4.9 min for orcinol and 5.3 min for kinetin), showed good ionization properties and do not natively appear in any organ of P. sarmentosum. Ideally, internal standards should cover a wider polarity range of the chromatogram, however, the appearance of many compound peaks in the chromatogram region between 10 and 20 min did not allow the application of an internal standard eluting in this retention time range without overlap.

The representative total ion chromatograms (TICs) acquired from the extracts of the different organs are shown in Fig. 1. The efficient RP-UHPLC separation of P. sarmentosum metabolites was successfully achieved in 23 min. Most major peaks from different organs exhibit clear qualitative and quantitative (intensity) differences. For example, the TICs of fruit and root extracts showed a lower intensity of peaks eluting between 5 and 10 min compared to the extracts prepared from other organs (Fig. 1). On the other hand, the peaks eluting between 11 and 17 min demonstrated clearly higher intensity in these extracts. In contrast, the leaf extracts were featured with an opposite pattern of relative metabolite abundances, and stems were intermediate. Thus, our results clearly demonstrate that the quantitative patterns of characteristic P. sarmentosum metabolites depend on the plant organ. (For qualitative differences see below).

Fig. 1

Comparison of the total ion chromatograms (TICs) acquired from methanolic extracts of different P. sarmentosum organs (leaves, stems, roots and fruits) in the positive ion mode: ISA  internal standard orcinol, ISB  internal standard kinetin, and P peak number

In this study, these patterns were analyzed by electrospray (ESI) high-resolution-mass spectrometry (HR-MS) to enable the determination of the elemental composition. Based on these data, complemented with mass spectral fragmentation patterns obtained in MS/MS experiments, comparison with databases and co-elution experiments with authentic standard reference compounds (see chapter 2.4 for details) the most prominent peaks found in the tR range of 11–17 min (Fig. 1) could be assigned to the groups of piperamides and phenylpropanoids. For example, peaks P93, P95, P112, P114 and P136 were putatively identified as nigrinodine, trans-asarone, sarmentine, brachyamide B and guineensine, respectively (Fig. 2).

Fig. 2

The structures of selected compounds representing the most prominent peaks in the TICs of extracts from different organs of P. sarmentosum

2.2 Characterization of the inter-organ differences by untargeted metabolite profiling

Due to its high mass accuracy, sensitivity and resolution, the HR-QqTOF-MS is the most efficient analytical platform for online structural elucidation of multiple components in plants [35]. The inter-sample alignment of the individual TICs corresponding to all features in the dataset resulted in a consensus peak table (data matrix comprising individual features as defined pairs of tR-m/z values) containing 1272 entries. These matched features comprised protonated molecular ions and other adducts, as well as in-source fragmentation products associated with characteristic neutral losses (matrix data not shown). The data matrix was subjected to multivariate statistics analysis with MetaboAnalyst software.

Principal component analysis (PCA) is recognized to be an efficient tool for reducing complex data sets and providing important insight into the variation within and between experimental groups [24]. The results of the PCA analysis accomplished with our dataset (score and loadings plots) are shown in Fig. 3. As can be seen from the score plot, approximately 86% of the total variance was accounted for by the first two principal components. The first component (PC1) was responsible for 64.8% of the total variance, whereas the second component (PC2) explained 21.0% of the total variance.

Fig. 3

Principle component analysis (PCA) with scores (A) and loadings (B) plots accomplished with 1272 features detected in the RP-UHPLC-QqTOF-MS data of leaf, stem, root and fruit extracts from P. sarmentosum

The leaves were separated from the other organs in PC1, whereas PC2 indicates that stems appeared to be significantly different from other organs, here especially fruits, in terms of their metabolite patterns. The loadings plot gives access to the features contributing to the separation of organs, which is primarily due to differences in their abundance within the various test samples and specific p value (Fig. 3B). The presence of highly intense chromatographic signals corresponding to the neolignane andamanicin (P122), flavonoids (P30, P36) and the phenyl propanoid trans-asarone (P95) led to the positioning of leaves in a single group in PCA. Thereby, P122 was represented by two different features, namely its protonated molecular ion [M + H]+ (feature m765, tR 14.5 min, m/z 417.2256) and the corresponding ammonium adduct [M + NH4]+ (feature m793, tR 14.5 min, m/z 434.2530). Additionally, the existence of the unique aporphine-type compounds in stems, like magnoflorines (P12, P21), piperolactams A (P70) and B (P75), had a powerful discriminating effect for their separation from the other groups in the score plots. Finally, as can be seen from the TICs in Fig. 1, the metabolite profiles of fruits and roots are quite similar. Therefore, the corresponding individual samples are located in the same quadrant of the PCA score plot, though well separated. This fact can be explained by higher contents (manifested in the LC–MS dataset by higher relative intensities of corresponding molecular ionic species) of piperamides in these two organs, specifically pellitorine (P116), brachyamide B (P114) and guineensine (P136) (Fig. 3). Interestingly, peaks P93 (nigrinodine) and P112 (sarmentine), which are both derived from piperamide-type compounds, represent strong negative loadings located in the left side of PC1 compared to other common features (Fig. 3B). This indicates the relative high content of both P93 and P112 in all organs except the leaves. The relative abundances of individual metabolites, corresponding to the most significant loadings (i.e. the most significant ones for the separation between the plant organs) are illustrated by boxplots as displayed in Fig. 4. The observed results are strongly supported and further detailed by PCA and OPLS-DA pairwise comparisons of metabolite profiles from different organs (See Fig. S3). This was performed for all organs resulting in 6 analyses (leaves/stems, leaves/roots, leaves/fruits, stems/roots, stems/fruits and roots/fruits). Beside the already described metabolites, variation within the groups is partly caused by unidentified features and artifacts (e.g. feature m226, tR 10.4 min; m/z 230.2467 calculated for C14H32NO+).

Fig. 4

Boxplots illustrating the relative abundances of the individual secondary metabolites contributing to PCA separation demonstrate significantly different intensities of the corresponding chromatographic peaks, in fruits (red, A), leaves (green, B), roots (blue, C) and stems (cyan, D) of P. sarmentosum. Corresponding tR and m/z values of metabolites are summarized in Table 1 and Table S1. *The flavonoids P30 and P36 represent Vitexin 2ʺ-O-galactoside and Vitexin 4ʺ-(3-hydroxy-3-methylglutaroyl)-2ʺ-O-β-D-glucopyranoside, respectively

Our results clearly indicate that production and accumulation of secondary metabolites is organ-specific [25, 36]. Additionally, as demonstrated by the differential responses of plant organs to environmental stressors, the stress adaptation responses are organ-specific [37]. This results in the existence of well-defined organ-specific patterns of metabolite accumulation, which facilitates interaction with the environment and helps to establish defense mechanisms with minimum expenditure of energy and valuable nitrogen containing compounds [38, 39]. Obviously, this factor affects the nutritional and medicinal value of the plant organs.

2.3 Hierarchical clustering analysis

Hierarchical clustering with heat map representation proved to be an excellent technique to address the distribution and relative abundances of the identified individual phytochemicals in different organs of P. sarmentosum (Fig. 5). The heatmap was constructed using a hierarchical clustering algorithm based on the inter-sample similarities in the relative abundances of all annotated metabolites (Fig. 5). Altogether, the hierarchical clustering was performed based on 154 annotated compounds with assigned structure from different organs belonging to the two major groups, alkaloids (A) and flavonoids including other metabolites (B). The individual replicates obtained from the same organ cluster nicely together indicating good reproducibility of results. The heatmap clearly shows the organ-specific accumulation of individual metabolites in leaves, stems, roots and fruits, indicating essential metabolic or storage differences between the P. sarmentosum organs which will influence their medicinal properties. Based on this dataset, seven clusters could be assigned including Cluster I-IV for alkaloids and Cluster V-VII for flavonoids and other metabolites (Fig. 5).

Fig. 5

Hierarchical clustering with heatmap representation illustrating the relative inter-organ distribution of alkaloids (A) and flavonoids along with other metabolites (B), identified in different organs. The heatmap was established using a hierarchical clustering algorithm based on the similarities in abundances of aligned annotated metabolites. The red and blue color in the heatmap represent an increase and a decrease of metabolite levels compared to the average, respectively

Cluster I is primarily composed of lignanamides (P58, P60, P68, P74) and specific piperamides such as piperettylines (P92, P98, P102) and piperettines (P109, P111) containing a benzodioxole group (type A and E). The accumulation of these compounds was found in the roots where they may serve as defense against feeding and pathogens. Significant levels of similar piperamides were recently also detected in cells distributed in the cortex of black pepper roots [8]. Several plants, including Piper species such as P. puberulum and P. flaviflorum, have been reported to contain lignanamides [40, 41]. However, no comparative analysis of lignanamides among plant organs has been reported. The general biosynthesis pathway of lignanamides in another plant species, Cannabis sativa L., has been discussed in the past. However, detailed mechanisms and molecular events throughout the biosynthesis of lignanamides remain inconclusive [42]. Lignanamides are natural products which are formed via an oxidative coupling mechanism involving hydroxycinnamic acid amides (HCAAs) as intermediates [43]. The formation of HCAAs is catalyzed by hydroxycinnamoyl-CoA:tyramine N-(hydroxycinnamoyl) transferease (THT), an enzyme characteristic for several plant species, including potato and tobacco [44]. The function of HCAAs in strengthening plant defense, as a response to fungal and pathogen challenge, has been described in potato tubers, onion roots and tomatoes [45].

Aporphines and benzylisoquinolines belong to the large group of isoquinoline alkaloids mainly present in Cluster II. Isoquinoline alkaloids such as magnoflorine (P12, P21), cepharadione A (P81), aristolactam BIII (P97), coclaurines (P9, P26), and reticuline (P16) are among the metabolites responsible for the formation of this cluster. These compounds had elevated levels in stems. Several studies found that most of these phytochemicals are abundant in specialized cell sites [46, 47]. For example, the analysis of isoquinoline alkaloids in the stems of Sinomenium acutum showed a differential distribution of alkaloids, especially within the outer cortical regions, phloem and xylem [46]. In opium poppy (Papaver somniferum), sieve elements and specialized laticifers of the phloem produce and accumulate isoquinolines alkaloids, respectively [48]. The co-localization to vascular tissue might explain the predominant occurrence in stems. Basically, these compounds are synthesized based on phenylpropanoids and amino acids, and the primary specific part of their biosynthesis is common to all these plants i.e., from norcoclaurine to reticuline. However, occasionally some mixed pathways may also occur to provide structural divergence [49]. In general, isoquinoline alkaloids play a role in defense against pathogens, and their ecological functions also appear to include plant-animal interactions [50,51,52].

Cluster III is featured with a high number of phenolic amides. This distinct class of metabolites accumulated predominantly in leaves and fruits. Some specific components, such as P50 (paprazine) and P72 were most abundant in leaves. Paprazine represents a tyramine derivative that contributes to active plant defense responses in infected leaf material. Biosynthesis of paprazine was induced in tomato leaves in response to wounding [53], and it was found to be associated with the resistance reactions of pepper leaves infected with Xanthomonas campestris [54]. According to Newman et al. [54], the up-regulation of phenolic amides (e.g. paprazine) in plant tissues was preceded by an increase in the extractable activity of tyrosine decarboxylase in parallel to the enhancement in the transcription of genes encoding phenylalanine ammonia-lyase and tyramine hydroxycinnamoyl transferase. This gene–gene interaction is accompanied by cell death, which is connected to a necrotizing reaction or hypersensitive response of the plant. Zacares et al. reported that large amounts of hydroxycinnamic acid amides such as paprazine and trans-N-feruloyltyramine were detected in tomato leaves after bacterial infection. This suggested that phenolic amides could play a key role in plant resistance vs. pathogens [55].

The most promising group of phytochemicals for medicinal and food applications in Piper species are piperamides, including piperine (P91), pellitorine (P116), retrofractamide B (P128) and guineensine (P136), which are classified into Cluster IV. Multiple metabolites representing this compound class (which actually contributes to the pungent taste of pepper) were detected in fruits, but not in the other organs. Similar occurrences of these constituents were also reported in fruits of other Piper species such as P. nigrum L. and P. longum L. [22, 56]. The observations of these constituents in Piper fruits were supported by the study of Schnabel and co-workers dealing with piperine synthase from P. nigrum. The authors propose that the accumulation of piperine and piperamides is related to the activity of piperine synthase (piperoyl-CoA: piperidine piperoyl transferase) and other BAHD-type enzymes that are preferentially expressed in fruits [57]. These secondary metabolites have a significant ecological impact on fruit-frugivore interactions and plant reproductive success [58].

Not less importantly, piperamides possess numerous advantages for human health, food applications, and feature potential agricultural prospects. For example, pellitorine, the major metabolite in this organ, was reported to possess insecticidal, anticancer, and antiplatelet properties [59]. In addition, piperine, a well-known compound derived from this species, has a wide range of biological activities and a distinct sharp flavor [60, 61]. Therefore, the fruits from Piper are predominantly used as food complements. Thus, the results obtained here provide insights in the perspectives for the use of P. sarmentosum fruits as an alternative to other Piper species.

Cluster VI is formed mostly by glycosides of volatiles such as P4, P28, P45, and P53, which are primarily found in fruits. The Piper fruits are particularly appreciated for their characteristics flavor and aroma that develops as fruits ripen [62]. A significant proportion of potential flavor contributors have been reported to exist as volatile compounds. Volatile aroma compounds are typically discovered in plants in two forms: "free" and "bound" to a sugar unit. These compounds are not odoriferous when bound; however, upon glycoside hydrolysis, these metabolites are liberated and then can be volatilized [63]. Moreover, the hydrolysis of glycosides may occur during the fruit ripening (ageing, senescence), or chewing/mechanical disruption of storage organs (e.g. vacuoles or lipid bodies, then combined with glycosidases from other cell compartments or cells), followed by the subsequent release of the volatile compounds that can olfactorily enhance the fruit flavor. Previous studies, e.g. such performed with the model plant Arabidopsis thaliana, show that glycosides are the result of glycosyltransferase activity. These enzymes add an activated sugar unit to the aglycone in the cytosol of the plant cell [64,65,66]. Recently, the analysis of the 7-deoxyloganetin glucosyltransferase-like (GGT) gene during fruit development of various Piper nigrum L. varieties revealed that increased GGT expression in Piper fruits correlated with up-regulation of volatile and nonvolatile metabolites [67]. Interestingly, some of the glycosides detected in this study contain 3-hydroxy-3-methylglutaryl residues as also found in flavonoids in this species (see chapter 2.4.2.1.)

Finally, Cluster VII is mostly represented by flavonoids (apigenin derivatives), including flavone-C-glycosides, and other phenolic metabolites such as phenylpropanoids and specific lignans. The compounds present in this cluster were found in relatively high amounts in leaves and are congruent with the constituents isolated from leaves of P. sarmentosum in our previous study [17].

The nature of flavonoid accumulation in plants indicates their important biological role. For instance, flavones are involved in a broad range of physiological functions in plant development and defense, such as protection from pathogen attack and in plant resistance to unfavorable environmental conditions like oxidative stress, high-level UV radiation and water/drought stress [68]. In general, the precursors for flavonoid biosynthesis are delivered by the phenylpropanoid pathway. The last steps of flavonoid biosynthesis involve various chemical modifications on the flavonoid aglycons, such as methylation, hydroxylation, glycosylation, and acylation, which increase their diversity and contribute to the final chemical properties [69]. Most of the enzymes involved in flavonoid metabolism are found in leaf mesophyll [70, 71]. This explains the greater content of flavonoids detected in the leaves in comparison to other organs. Leaves are the plant parts from P. sarmentosum which are preferentially consumed in Malaysia and other Asian countries, and which were shown to possess vascular protective, neuroprotective, anti-obesity, and anti-hyperlipidemia activity [18,19,20]. These activities might be connected to the content of flavonoids and lignans which possess a high antioxidative potential [72]. Previously, Miean et al. reported that the total flavonoid content including apigenin (TFC) of P. sarmentosum leaves can reach 120.5 mg/kg of dry weight [73]. According to Ugusman et al. [74], there is a positive correlation between the presence of flavonoids in P. sarmentosum leaves and its protective effects against oxidative stress. Other, earlier published studies reported that plant leaf extracts exhibited anti-inflammatory, antibacterial and insecticidal properties [10, 75, 76]. Simple phenylpropanoid compounds as asarone and its derivatives have an inhibitory effect on mosquitos and other insects, particularly larvae by inhibiting the hatching rates of eggs [77, 78].

Based on these data, the conclusion about the predominant occurrence of specific metabolite classes in different organs of P. sarmentosum is summarized in Fig. 6.

Fig. 6

Predominant occurrence of compound classes in the organs of P. sarmentosum

2.4 Tentative identification of bioactives in P. sarmentosum

The untargeted metabolomic analysis which was used to characterize the metabolite composition of P. sarmentosum MeOH extracts revealed the differences between the plant organs. For annotation or identification of contributing constituents, their MS1 accurate monoisotopic ions were used to determine the likely elemental composition. As the mass accuracy of 5 ppm corresponded to the specification of the mass spectrometer used, this value was considered in the MS1 analysis. Moreover, a comparison of the fragmentation patterns with those reported in the literature and databases such as Reaxys and MassBank were performed to assign the compounds. The mass accuracy of fragment ions of 15 ppm was considered during the annotation process. This annotation strategy is well-established and is widely used [79,80,81,82]. In addition, isolated and fully elucidated compounds obtained from our previous work [17] were applied as reference compounds for the identification of individual metabolites by comparison of their m/z values, retention times and fragmentation patterns. Since structurally different compounds require different suitable collision voltages to obtain fragment information [83,84,85], a range of collision voltages between 15 and 60 V was used in the MS/MS analysis to identify metabolites.

As a result, a total of 154 compounds were annotated, either unambiguously identified by co-elution with authentic standards verified by NMR spectroscopy data or tentatively assigned by m/z values and MS/MS fragmentation patterns. The identified species were numbered according to the order of their elution time (Fig. 1 and Table 1 and S1). Alkaloids appeared to be the predominant metabolites, with 111 compounds annotated. In addition, 15 flavonoids and 28 other compounds were identified. Table 1 depicts the major organ-specific metabolites of P. sarmentosum characterized by both MS and MS/MS. The detailed list of all individual compounds annotated is summarized in Table S1.

Table 1 RP-UHPLC-QqTOF-MS/MS data of major and discriminating metabolites annotated in four organs of P. sarmentosum in positive ionization mode2.4.1 Alkaloids

Piper is a rich source of alkaloids, for which numerous biological activities have been reported [86, 87]. Furthermore, the quality of Piper species is usually assessed by their metabolite composition and the contents of some biologically active components [