1 Introduction

The plant kingdom synthesizes hundreds of thousands of diverse specialized metabolites that play key roles in chemical defenses against herbivores and resilience to environmental stress (Duke et al. 2000; Wagner 1991). Some find commercial value as pharmaceuticals, fragrances, food additives and natural pesticides (Balandrin et al. 1985; Cragg and Newman 2005). Additionally, since biosynthetic pathways of specialized metabolites have been modified by mutation and natural selection, plants function as efficient “chemical factories” to produce bioactive chemicals (DellaPenna and O’Connor 2012). In the family Solanaceae, which includes numerous food and ornamental crops including tomato, potato, petunia, tobacco, eggplant, and pepper, hair-like epidermal appendages known as glandular trichomes (GTs) (Fahn 2000) play critical roles in synthesis, storage and secretion of bioactive specialized metabolites (Schilmiller et al. 2008). GTs can be selected as targets for plant metabolic engineering to increase resistance to insects (Glas et al. 2012). Among trichome-enriched specialized metabolites, acylsugars have drawn broad attention owing to their importance in plant resistance against numerous insects (McKenzie and Puterka 2004; Rodriguez et al. 1993; Weinhold and Baldwin 2011). In Solanaceous plants, acylsugars have been documented as glucose or sucrose cores esterified by several C2–C12 aliphatic acids (Ghosh et al. 2014; King et al. 1990, 1993). Previous work showed that diversity in acylsugars, which derives from variation in acyl chain lengths, aliphatic chain branching, number of ester groups and positions of esterification, affects insecticidal properties (Chortyk et al. 1997; Maluf et al. 2010). An improved understanding of the genetic controls of acylsugar biosynthesis is essential to manipulate plant metabolic engineering (Kim et al. 2012; Leckie et al. 2014), and advances in this area would be accelerated by a more complete understanding of the structures of acylsugars produced in nature.

Methods for acylsugar characterization were described in a recent review(Ghosh and Jones 2017). Recent efforts to identify acylsugars produced by plants in the family Solanaceae have relied on mass spectrometry (MS) and nuclear magnetic resonance (NMR) spectroscopy of purified metabolites (Ghosh et al. 2014; Kim et al. 2012; Schilmiller et al. 2010). Ultra-high performance liquid chromatography (UHPLC) and time-of-flight mass spectrometry (TOF-MS) serve as powerful tools for metabolomic profiling (Dettmer et al. 2007). Individual specialized metabolites are resolved by LC separation, recognized as acylsugars from their mass spectra using relative mass defect (RMD) filtering (Ekanayaka et al. 2015) and annotated based on retention times and accurate molecular and fragment ion masses. In addition, MS-based metabolite profiling of multiple genotypes builds connections between phenotype and genotype that facilitate discovery of gene and enzyme functions involved in specialized metabolite biosynthesis (Kliebenstein and Osbourn 2012). In tomato, a BAHD acetyltransferase SlAT2 (recently renamed SlASAT4) specifically catalyzes acetylation on the 2-position of acylsucroses, and was identified based on acylsugar profiles of cultivated tomato and S. lycopersicum × S. pennellii introgression lines (Schilmiller et al. 2012). A recent and extensive investigation identified four BAHD acyltransferases that account for all acylation reactions in the biosynthesis of tomato acylsucroses (Fan et al. 2016). These efforts relied on a detailed understanding of acylsugar structures that describes both the diversity of acyl groups and their attachment positions on the carbohydrate core.

Current understanding of Petunia acylsugar biosynthesis is based on limited reports largely focused on metabolites extracted from hybrids generated from crosses of Petunia species (Moser et al. 1999; Ohya et al. 1996) including pooled tissues from 114 Petunia hybrid cultivars (Chortyk et al. 1997). One notable recent paper profiled acyl groups released from acylsugar-containing extracts from five Petunia hybrids, and found a predominance of malonate and straight-chain aliphatic acyl groups ranging from 2 to 8 carbon atoms (Kroumova et al. 2016). We are aware of one report of structures of two acylsucroses isolated from Petunia species other than hybrids (Singh et al. 2003). A small number (~9) of Petunia acylsucroses have been documented with exact structure information supported by NMR of purified or pooled metabolites (Moser et al. 1999; Singh et al. 2003) with acyl substitutions detected on 2, 3, 4, 6-positions on the pyranose ring of sucrose by C5–C8 aliphatic acyl groups and 1′, 4′, and 6′-positions on the furanose ring by acetyl and malonyl chains. Unique triacylsucroses with pyranose acylation on 2, 3, 4-positions with either iC6 or iC7 groups were derived from P. nyctaginiflora (synonymous with P. axillaris) (Singh et al. 2003). However, much characterization of Petunia acylsugar has been performed on incompletely purified mixtures. Acylglucoses from P. hybrida were also reported but positions of specific acyl groups were not established (Chortyk et al. 1997). Deeper and more detailed structural information, particularly with regard to selectivity of acylations at specific positions, is necessary to establish roles of position- and substrate-selectivity of acyltransferase enzymes and how these evolved to control acylsugar accumulation.

In this report, we profiled acylsugar metabolites derived from three Petunia species, P. integrifolia, P. axillaris, and P. exserta using UHPLC/MS and demonstrated acylsugar chemical diversity within species based on structures generated from NMR spectra of purified metabolites. The broader goal of this investigation is to support discoveries of biosynthetic genes involved in specialized metabolite accumulation.

2 Materials and methods

2.1 Plant material preparation and extraction

Plants (P. integrifolia, P. exserta and P. axillaris) were germinated from seeds and grown in a greenhouse. Additional plant growth metadata are provided in Supplemental Material, Sect. 4. For acylsugar profiling, leaflets of each were extracted with isopropanol:acetonitrile:water (3:3:2, v/v/v). Additional extraction metadata details are presented in Supplemental Material, Sect. 1.1.

2.2 UHPLC/MS analyses

Metabolite profiling was performed using a Waters LCT Premier™ time-of-flight mass spectrometer coupled with Shimadzu LC-20AD pumps. Experimental details are provided in Supplemental Material, Sect. 1.2.

2.3 Purification of acylsugars from Petunia species

Acidic malonate ester acylsugars in P. integrifolia, P. exserta and P. axillaris were separated from neutral acylsugars using Supelclean™ LC-SAX SPE tubes with Cl− counter ion (3 mL, bed wt. 500 mg). Subsequent purification of individual acylsugars was achieved using HPLC. Details regarding metabolite fractionation and purification are provided in Supplemental Material, Sect. 2.

2.4 Structure elucidation of purified acylsugars

NMR spectra were obtained using either Agilent DirectDrive2 500 MHz (Agilent) or Avance 900 MHz (Bruker) NMR spectrometers at the Max T. Rogers NMR Facility at Michigan State University. CDCl3 was used as NMR solvent and chemical shift reference (δ = 7.24 ppm for 1H, δ = 77.0 ppm for 13C) for individual acylsugars. Additional NMR metadata for acylsugar structure elucidation are presented in Supplemental Material, Sect. 6.

3 Results and discussion

Twenty-nine distinct acylsucroses were purified from P. integrifolia, P. axillaris and P. exserta, including malonate ester acylsucroses that have not, to our knowledge, been isolated from other Solanaceae species. All are described in Table 1 with UHPLC retention times, accurate molecular masses and structures proposed based on NMR spectroscopy. NMR 1D and 2D spectra, chemical shift assignments, PubChem CID numbers, and InChI keys are presented in Supplemental Material. Since these metabolites were identified based on NMR and mass spectrometric characterization without available authentic standards or synthetic confirmation, their structures should be considered as putative, meeting level 2 criteria of the Metabolomics Standards Initiative guidelines (Sumner et al. 2007).

Table 1 NMR elucidated structures of acylsugars purified from leaf extracts of P. integrifolia, P. exserta and P. axillaris
Full size table

We propose a modified nomenclature for Petunia acylsugars derived from that proposed for acylsugars from tomato and its wild relatives (Schilmiller et al. 2010; Ghosh et al. 2014). “S” refers to the sucrose core, and designation of the letter “m” in parentheses identifies compounds containing a malonate ester. The first number indicates the number of acyl groups attached to the sugar core, and the number following the colon represents the total number of carbons in acyl chains. Numbers in parentheses indicate the number of carbons in the individual acyl chains. Isomeric acylsucroses with the same molecular mass are ordered based on increasing chromatographic retention times, which are labeled as a number in brackets (e.g. S(m)5:25[2]). Branching of acyl groups is specified by using the abbreviations ‘i’ for iso and ‘ai’ for anteiso branching.

3.1 Acylsugar profiling using UHPLC/TOF-MS

UHPLC/TOF-MS analyses demonstrated numerous chromatographic peaks that were annotated based on retention times and accurate pseudomolecular and fragment ion masses. Most observed metabolites were assigned as acylsucroses.. Thirty-seven distinct acylsucrose pseudomolecular masses were resolved from P. integrifolia, P. exserta and P. axillaris (Fig. 1), consisting of 2 triesters, 16 tetraesters and 19 pentaesters of sucrose. More than 100 distinct acylsugar forms were resolved using an 80-min LC gradient. All neutral aliphatic acyl groups were substituted on the pyranose ring as judged from fragment ions generated using non-selective CID in positive-ion mode. To our knowledge, none of these Petunia acylsugars have been identified in tomato or its wild relatives.

Fig. 1
figure 1

LC/MS Base Peak Intensity (BPI) chromatograms of leaflet extracts from three Petunia species. Peaks for specialized metabolites are annotated. Data were generated using ESI in negative-ion mode with 80-min LC gradient. a P. integrifolia; b P. exserta; c P. axillaris

Full size image

Extracted ion chromatograms for carboxylate fragment ions (m/z 87, 101, 115…) generated at the highest collision potential provide information about the number of carbon atoms in the aliphatic acyl groups. All three genotypes yielded evidence for acyl groups ranging from C4 (m/z 87) to C9 (m/z 157), with relative amounts varying across genotypes. In contrast to a recent report of acyl groups in acylsugars of Petunia hybrids (Kroumova et al. 2016), our LC/MS data indicated low levels of C9 esters in acylsucroses from P. axillaris and P. exserta but not P. integrifolia. The isomeric diversity of acylsucroses was extensive based on extracted ion chromatograms for [M+formate]− ions (Fig. 2), and suggest about 18 isomers of S4:21 across all three Petunia species. This finding is in accord with variation in positions of esterification by iso- and anteiso- groups.

Fig. 2
figure 2

LC/MS extracted ion chromatograms for m/z 737.36 ± 0.05, corresponding to neutral acylsucrose S4:21, obtained in negative-ion mode for leaflet extracts from a P. integrifolia, b P. exserta, and c P. axillaris showing evidence for a total of 18 isomers, some incompletely resolved, across the three Petunia genotypes

Full size image

Acylsucroses that possess a malonyl group on the furanose ring were abundant in all three Petunia species and could be distinguished by ions with different nominal masses than the aliphatic ester series. For example, neutral acylsugars were detected in negative ion mode as [M+formate]− ions as a series of homologs differing in nominal mass by 14 Da (m/z 709, 723, 737, 751, 765…; e.g. S4:23 appeared at m/z 765), whereas malonate esters formed a series of [M–H]− ions 2 Da lighter (m/z 763, 777, 791, 805…; e.g. S(m)5:23 at m/z 763). In addition, multiplexed CID spectra distinguished malonate acylsugars, which undergo facile decarboxylation to [M–H–CO2]−at the low collision potential of 25 V (Fig. 3), while neutral acylsugars did not yield significant fragment ions under these relatively gentle conditions.

Fig. 3
figure 3

Multiplexed CID mass spectra of S(m)5:25 (RT ~ 36.16 min) from P. exserta. a positive-ion ESI spectrum (collision potential = 40 V), b negative-ion ESI spectra (collision potentials = 80 V, 25 V, 10 V; top to bottom)

Full size image

CID spectra generated at multiple collision voltages provided important information to distinguish isomeric acylsugars. In negative-ion mass spectra, compositions of acyl chains and esterification positions in acylsugars were suggested by fragment ions formed at intermediate collision potentials (25, 40 and 55 V) corresponding to neutral losses of acyl groups as ketenes (e.g. 84 Da for C5, 112 Da for C7). Ions corresponding to deprotonated fatty acids (e.g. m/z 101 Da for C5 carboxylates and 129 Da for C7 carboxylates) were detected at the highest collision potential (80 V). In contrast, positive-ion mass spectra yielded fragment ions corresponding to cleavage of the glycosidic bond as the collision voltage increased. The masses of abundant key fragment ions facilitated assignment of acyl groups to either the pyranose or furanose rings. For neutral acylsucroses from tomato and its wild relatives, the most abundant fragment ion was usually formed with retention of charge on the furanose ring and its mass reflected the sum of acyl substitutions. However, for Petunia acylsucroses containing a malonate ester, the presence of malonate ester shifted the most abundant positively-charged fragment to the pyranose ring ion (e.g. m/z 527 for S(m)5:25; Fig. 3A) and the malonylfructose fragment at m/z 249 was less abundant than typical when fructose groups are esterified by neutral acids. Figure 3 illustrates how negative and positive mode mass spectra were applied to establish carbon number of acyl chains and their distribution on each ring of sucrose in malonate acylsugar S(m)5:25. Observation of [M–H]− (m/z 791) at lowest CID potential and [M–H–CO2]− (m/z 747) formed at 25 V suggested malonyl ester decarboxylation. When CID potential was increased to 80 V, ions corresponding to additional neutral aliphatic ketene losses ([M–H–CO2–C5]−, m/z 663) and [M–H–CO2–C5–C7]−, m/z 551) were observed, as were carboxylate ions for C5 (m/z 101) and C7 (m/z 129) aliphatic acids. In positive-ion mass spectra, [malonylfuranosyl]+ (m/z 249) was observed at a collision potential of 40 V, consistent with substitution of only the malonyl ester on the furanose ring. More specific assignments of acyl group attachments at specific positions on sucrose were made based on one-dimensional and two-dimensional NMR (1D/2D NMR) spectra of purified compounds as described below.

3.2 Isolation of malonate acylsugars using LC-SAX SPE

Fractionation on a strong anion exchange (SAX) solid phase extraction column (SPE) separated acidic malonate ester acylsugars from neutral metabolites and facilitated subsequent purification (Fig. 4). Several malonate ester acylsugars that have similar UHPLC retention times (Table 1) as neutral acylsugars were challenging to isolate using only reversed phase HPLC. A solid phase anion exchange column retained malonate ester acylsugars which are negatively-charged at pH 6, and neutral acylsugars were eluted using organic solvent followed by elution of acidic acylsugars using acidified methanol/water (details in Supplemental Material, Sect. 2.2).

Fig. 4
figure 4

Workflow for malonate acylsugar fractionation from plant extracts using LC-SAX SPE column

Full size image

3.3 Acylsugar structure elucidation using NMR

To fully understand Petunia acylsugar chemical diversity with a broader goal to accelerate discoveries of biosynthetic genes involved in acylsugar metabolites accumulation, 14 neutral acylsucroses and 15 malonate acylsucroses were purified from three Petunia species. Their proposed structures were suggested based on 1D (1H and 13C) and 2D (COSY, HSQC and HMBC) NMR data. There are three distinct chemical shift regions for proton resonances in 1H spectra of all purified acylsugars: (1) 0.7–1.8 ppm is the region for aliphatic H from acyl groups; (2) 2.0–2.5 ppm is the region for α-H of carbonyl; (3) 4.0–5.8 ppm is the region for protons on the pyranose ring. H attached to substituted C in sucrose ring fell in the 4.3–5.8 ppm range, and 3.6–4.3 ppm is the region for protons on furanose ring carbons that are not acylated. Acylation causes a downfield shift in resonances for protons attached at the esterified carbon positions and this information was used to identify substituted positions in simple cases (Fig. 5a–c). 1H and COSY NMR spectra were applied to assign protons on sucrose carbons based on 1H–1H coupling, starting with the most downfield peaks (δ 5.6–5.8 ppm) that correspond to the 1-position proton on the pyranose ring. 13C and HSQC spectra were used to not only assign carbon atoms on sucrose core and for acyl chains but also to distinguish ring protons when 1H peaks overlapped in the region of 3.8–4.3 ppm since chemical shifts of their direct attached C differ between esterified and nonesterified positions (Supplemental Material, Table S2). Substitution positions for each acyl chain on the sucrose core were recognized using HMBC spectra, which showed correlation between H and C nuclei that are 2–4 bonds away from each other. Carbonyl carbons (δ 170–180 ppm) generated cross peaks with both pyranose ring protons and α-position protons on acyl chains to demonstrate specific esterification positions for individual acyl groups.

Fig. 5
figure 5

NMR spectra for P. exserta acylsugars. 1H NMR spectra indicate that there are 2 substitution positions for malonate groups in P. exserta acylsugars in a S4:23(5,5,5,8), b S(m)5:26(m,5,5,5,8)[1], and c S(m)5:26(m,5,5,5,8)[2]. d HMBC spectrum of S(m)5:26(m,5,5,5,8)[1] provided evidence for presence of the malonate group and its attachment on the 6′ position

Full size image

1H and HMBC spectra provided clear evidence for malonyl ester groups. Compared with neutral acylsugar 1H spectra, a unique singlet of the malonyl α-proton was present around δ 3.5 ppm. Furthermore, two malonate carbonyl carbon resonances at 166–171 ppm correlated with the α-proton resonance at 3.5 ppm, producing two corresponding cross-peaks in HMBC spectra. For example, S(m)5:26(m,5,5,5,8) showed malonate α-proton resonance at 3.5 ppm correlated with carbonyl carbon resonances at 167.1 and 170.2 ppm, and the former correlated with 6´-position protons (δ 4.33, 4.48 ppm) that demonstrated attachment of the malonyl chain at the 6′ position (Fig. 5d). Because of the facile decarboxylation of the malonyl group, the presence of a malonyl group on the furanose ring was confirmed by mass spectra. A new cross peak in HMBC spectra indicative of an acetyl group, instead of a malonyl group, substituted on furanose ring appeared for old extracts (stored >2 weeks at −20 °C after purification; Supplemental Material, Figure S2). This finding may explain why esterification by acetyl groups has been reported in previous petunia acylsugar analyses. Table 1 lists all structures of purified acylsugars from P. integrifolia, P. exserta and P. axillaris elucidated from 1D and 2D NMR spectra, with relative abundances evaluated using chromatographic peak areas normalized to dry weight of leaflets.

3.4 Comparison of acylsugar profiles among 3 Petunia species and wild tomato

Although the three species P. integrifolia, P. exserta and P. axillaris all belong to the genus Petunia, acylsugar profiles revealed substantial differences, not only in qualitative aspects (e.g. presence of acylsugar isomers differing in substitution pattern or acyl chain branching), but quantitative levels of both neutral and malonate acylsugars among the three species (see Table 1; Fig. 1). UHPLC/MS chromatograms of three Petunia plants exhibited significant differentiations in acylsugar profiles based on chromatographic retention times, fragment ion patterns and relative abundances of specific acylsugars based on integrated extracted ion chromatograms (Fig. 1). Malonate ester acylsugars were more abundant relative to neutral acylsugars in P. exserta and P. axillaris, while in P. integrifolia, the levels of the two types were similar. Such diversity demonstrated within Petunia species may reflect differences in selectivity of acyltransferases as well as different levels of acyl-CoA pools.

Additionally, other specialized metabolites were detected in extracts of the three Petunia species using UHPLC/MS. Several lipids were found in three Petunia species; flavonoids quercetin monoglucoside, diglucoside and 3-methylquercetin were detected in P. exserta and a dimethylquercetin in P. axillaris. Neither terpenes nor alkaloids were detected based on MarkerLynx screen.

3.5 Diversity and conservation in position selective acylation

NMR structures of Petunia acylsugars summarized in Table 1 define sites of acyl group attachment and provide evidence of chemical diversity among three Petunia species. All hydroxyl groups on the pyranose ring of Petunia acylsucroses are substituted by acyl groups, while malonyl chains were detected only on the furanose ring. Acyl group diversity was greatest on the 3-position of the pyranose ring, with all acyl groups except iC4 detected at this site. Moreover, in 26 of 29 purified Petunia metabolites, this 3-position was preferably substituted by the longest acyl group among a specific acylsucrose. In contrast, the 4-position in P. exserta, P. axillaris and 6-position in P. integrifolia is conserved by esterification by aiC5 groups, and aiC5 chains were also attached at the 2-position in all three species. Position-selective acylation by aiC5 and longer acyl groups (C6–C8) was similar to an earlier report of acylsucroses from P. hybrida (Kraus 2003). Iso-C4 acylation at position 6 was observed in both P. exserta and P. axillaris, whereas the same substitution happened at the 2- and 4-positions in P. integrifolia. Only 3 purified acylsugars in P. axillaris contained aiC6 ester groups, all of them at the 3-position. The similar position-selectivity in attachment of acylgroups of P. exserta and P. axillaris is consistent with reported Petunia genus evolutionary analyses that these two species are more closely related than P. integrifolia (Kulcheski et al. 2006).

To our knowledge, acylsugar related malonyltransferases have not been reported in Petunia glandular trichomes and we anticipate that identification of malonate acylsugars will aid efforts to discover these metabolic functions. Esterification by malonate esters at the 1′-position of the furanose ring was predominant in P. integrifolia and P. axillaris, while either 1′- or 6′-position in P. exserta was esterified by malonyl chain. Previous reports described similar findings for P. hybrida malonate acylsugars (Chortyk et al. 1997; Ohya et al. 1996), while we suggest that their observations of acetate groups at the 6´-position may come from malonyl decarboxylation during metabolite isolation.

Despite limited prior understanding of Petunia acylsugar biosynthetic mechanisms, some hypotheses can be generated by comparing with previous studies of tomato and closely related wild species (Fan et al. 2016; Kim et al. 2012; Schilmiller et al. 2012). The highly conserved acyl groups at specific sugar positions might be attributed to the order of acylation catalyzed by position-selective acyltransferases or selective acyl chain removal and remodeling.

3.6 Diversity in acyl groups and implications for elongation mechanisms

Acyl CoA precursors of acylsugars are often derived from branched-chain ketoacids produced by metabolism of branched chain amino acids, often involving chain elongation. Although wild tomato species use two-carbon elongation as is common for fatty acid biosynthesis, elongation in Petunia has been proposed to use a one-carbon elongation known as the α-KAE pathway (Kroumova and Wagner 2003). Structure elucidation of Petunia acylsugars also assists the investigation of aliphatic acid elongation mechanisms. Confirmed by NMR data in this study, no acetate esters were observed in Petunia acylsugars (except via decarboxylation of malonate esters). Excluding malonate, the lengths of acyl chains in Petunia acylsugars ranged from C4 to C8, which is consistent with reported acylsugar profiles for other plants including Nicotiana spp. that use one-carbon metabolic elongation (Forkner and Hare 2000; Kroumova and Wagner 2003). At least two C5 moieties were present in all isolated Petunia acylsucroses, most of which are aiC5. Aside from the observation of aiC6 esters in P. axillaris, all other acyl groups longer than C5 were iso-branched, which could result from elongation of either valine-derived iC4 or leucine-derived iC5 to yield products as long as iC8 via either one- or two-carbon elongation. Anteiso-branching was only observed for aiC5 and aiC6 derived from isoleucine, the latter consistent with one-carbon elongation. Further experiments need to be performed to verify the proposed aliphatic acid elongation pathways in Petunia plants. Compared with structures of neutral acylsugars, malonate acylsugars only have one more position on the fructose ring that is acylated by a malonyl group, while all hydroxyls on the pyranose ring were esterified by aliphatic chains. Therefore, purified neutral acylsucroses in this study could be used as substrates for potential screening of malonyltransferases in Petunia species.

4 Conclusions

A more comprehensive determination of diversity and complexity in Petunia acylsucrose metabolites has been reported. Structures have been assigned to 28 acylsucroses, including malonate esters that have never been reported in metabolite profiles of other Solanaceous plants. In the current study, these structures point to acylation position selectivity, particularly in the case of aiC5 on the pyranose ring within P. exserta and P. axillaris but less so in P. integrifolia. This result is consistent with phylogenetic analyses that P. exserta and P. axillaris are more closely related than P. integrifolia. The selectivity (and promiscuity) of acylation at specific positions is consistent with action of multiple acyltransferase enzymes and these findings should serve as the foundation for investigations into the functions of Petunia acyltransferase genes.