- © 2015 American Society of Plant Biologists. All rights reserved.
Abstract
Lignin is an aromatic polymer derived from the combinatorial coupling of monolignol radicals in the cell wall. Recently, various glycosylated lignin oligomers have been revealed in Arabidopsis thaliana. Given that monolignol oxidation and monolignol radical coupling are known to occur in the apoplast, and glycosylation in the cytoplasm, it raises questions about the subcellular localization of glycosylated lignin oligomer biosynthesis and their storage. By metabolite profiling of Arabidopsis leaf vacuoles, we show that the leaf vacuole stores a large number of these small glycosylated lignin oligomers. Their structural variety and the incorporation of alternative monomers, as observed in Arabidopsis mutants with altered monolignol biosynthesis, indicate that they are all formed by combinatorial radical coupling. In contrast to the common believe that combinatorial coupling is restricted to the apoplast, we hypothesized that the aglycones of these compounds are made within the cell. To investigate this, leaf protoplast cultures were cofed with 13C6-labeled coniferyl alcohol and a 13C4-labeled dimer of coniferyl alcohol. Metabolite profiling of the cofed protoplasts provided strong support for the occurrence of intracellular monolignol coupling. We therefore propose a metabolic pathway involving intracellular combinatorial coupling of monolignol radicals, followed by oligomer glycosylation and vacuolar import, which shares characteristics with both lignin and lignan biosynthesis.
INTRODUCTION
Lignin is an aromatic polymer mainly deposited in secondary-thickened plant cell walls (e.g., woody tissues) where it provides mechanical strength as well as hydrophobicity to the cell wall and allows the transport of water and nutrients. Angiosperm lignins are mainly composed of guaiacyl (G), syringyl (S), and minor amounts of p-hydroxyphenyl (H) units that are derived from the monolignols coniferyl, sinapyl, and p-coumaryl alcohol, respectively (Figure 1). Monolignols are synthesized from phenylalanine via the general phenylpropanoid and monolignol biosynthetic pathways (Figure 1) (Boerjan et al., 2003; Vanholme et al., 2008; Bonawitz and Chapple, 2010; Vanholme et al., 2010, 2012a, 2012b). These pathways are most probably active in the cytoplasm, as the three P450 enzymes involved in monolignol biosynthesis have been localized to the outer surface of the endoplasmic reticulum, facing toward the cytosol (Ro et al., 2002; Li et al., 2008; Bassard et al., 2012; Sundin et al., 2014). The other enzymes of the biosynthetic pathway have no particular targeting signal, but immunolocalization has confirmed the cytoplasmic localization for some of them (Takabe et al., 1985; Chen et al., 2000). Once synthesized, monolignols are transported to the cell wall (Alejandro et al., 2012; Liu, 2012) where they are oxidized by peroxidases (Fagerstedt et al., 2010; Lee et al., 2013) and/or laccases (Berthet et al., 2011; Lu et al., 2013; Zhao et al., 2013). The resulting electron-delocalized monolignol radical can couple at several positions with another monolignol radical or, more often, with a radical at the free-phenolic end of the growing polymer. This results in the various linkage types present in the lignin polymer (mainly 8–8, 8–O–4, and 8–5 leading to resinol, β-aryl ether, and phenylcoumaran bonding units, respectively; Figure 1). Coupling reactions in lignin formation are combinatorial and under chemical control, leading to nonstereospecific, racemic products (Ralph et al., 2004).
Monolignol Biosynthesis and Oligolignol Formation.
Monolignol and oligolignol radicals dimerize or cross-couple with the formation of quinone methide intermediates. Rearomatization involves the attack of an external nucleophile, e.g., water (8–O–4-linkage), or trapping by an intramolecular hydroxyl function (8–8- and 8–5-linkages). For convenience, only the coupling of coniferyl alcohol is shown. In the trimer, only the product from 8–O–4-coupling is shown. Shorthand naming of the oligolignols is based on the nomenclature described by Morreel et al. (2014). In brief, the linkage type is noted between parentheses, whereas the units are denoted in bold: guaiacyl (G) and syringyl (S) units are derived from coniferyl and sinapyl alcohol. 4CL, p-coumarate:CoA ligase; C3H, p-coumarate 3-hydroxylase; CA, caffeic acid; CAD, cinnamyl alcohol dehydrogenase; CCR, cinnamoyl-CoA reductase; CSE, caffeoyl shikimate esterase; F5H, ferulate 5-hydroxylase; Fer-CoA, feruloyl-CoA; HCT, p-hydroxycinnamoyl-CoA:quinate/shikimate p-hydroxycinnamoyl transferase; PAL, phenylalanine ammonia-lyase; pCA, p-coumaric acid; POX, peroxidase; LAC, laccase.
Lignin oligomers (dimers, trimers, etc., together called oligolignols) can often be observed in the phenolic profiles of plant tissue extracts. Houghton (1985) initially described the identification of a few oligolignols in the stems of Buddleia davidii. Later, many oligolignols have been extracted from poplar (Populus spp) xylem and their structures characterized by mass spectrometry and supported by NMR and chemical syntheses (Morreel et al., 2004). This study revealed that (1) all major units and linkage types found in lignin are present in the oligolignols, (2) both diastereomers (threo or syn and erythro or anti) from 8–O–4-coupling exist in the oligolignols, and (3) all oligolignol structures reflect the known monolignol cross-coupling propensities, since their relative abundances matched with those generated by synthetic dehydrogenation reactions. Thus, all oligolignol structures are consistent with combinatorial radical cross-coupling reactions in the cell wall; hence, they are considered to be small lignin polymers present in the apoplast (Morreel et al., 2004).
In addition to being used for lignification, monolignols are precursors of lignans and neolignans (Figure 2; Wang et al., 2013). Lignans are secondary metabolites derived from the coupling of two coniferyl alcohol monomers via an 8-8-bond, whereas neolignans arise from 8–5- or 8–O–4-coupling (Moss, 2000). The exact function of these coupling products is not well understood, but they have been postulated to serve as antioxidants (Kitts et al., 1999) or to play a role in plant defense (Harmatha and Nawrot, 2002; Hano et al., 2006; Schroeder et al., 2006). In contrast to the free-radical lignin polymerization, the monolignol cross-coupling toward (neo)lignans is guided by dirigent proteins (Davin et al., 1997) that have a “dirigent domain” that acts as a scaffold that orients two monolignol radicals during dimerization, leading to the formation of a specific stereoisomer. Dirigent proteins have been localized to the cell wall (Burlat et al., 2001), suggesting that (neo)lignans are formed in the apoplast. To date, there is no evidence that lignans are incorporated into lignins (Ralph et al., 1999; Akiyama et al., 2014), although the dirigent domain-containing protein ENHANCED SUBERIN1 is required for the correct patterning of lignin deposition in Casparian strips (Hosmani et al., 2013).
Monolignol-Biosynthesis-Dependent Pathways in Arabidopsis.
Hydroxycinnamic acid derivatives, such as sinapoyl glucose, malate esters, and monolignol hexosides, are likely stored in the vacuole. Monolignols also serve for the production of (neo)lignans and lignin. The stereospecific coupling between two monolignol radicals is guided by dirigent proteins that are presumably located in the apoplast. Finally, monolignols are also precursors of oligolignol hexosides. The pathway shown in gray, in which the oligolignol aglycones arise from intracellular coupling and are stored as their glycosides in the vacuole, is the focus of this article.
Monolignols and other phenylpropanoids can also be glycosylated. It is generally hypothesized that free phenylpropanoids are unstable and/or toxic to the plant (Whetten and Sederoff, 1995; Meyermans et al., 2000). To increase their stability and reduce their toxicity, coniferyl and sinapyl alcohol are phenol-glucosylated, yielding coniferin and syringin (Whetten and Sederoff, 1995; Bowles et al., 2006). In addition, phenylpropanoic acids in Arabidopsis thaliana are often derivatized to glucose esters (e.g., sinapoyl glucose). Since sinapoylglucose:malate sinapoyltransferase, catalyzing the conversion of sinapoyl glucose to sinapoyl malate, is present in the vacuole (Sharma and Strack, 1985; Hause et al., 2002), and because monolignol glucosides can be transported through the tonoplast via ATP binding cassette-like transporters (Miao and Liu, 2010), sinapoyl glucose and the monolignol glucosides are likely stored in the vacuole. A vacuolar localization of coniferin and other monolignol glucosides has been speculated (Leinhos and Savidge, 1993; Dharmawardhana et al., 1995) but has not yet been unambiguously proven (Kaneda et al., 2008).
In addition to the glycosylated monolignols, a series of hexosylated oligolignols, such as G(8–O–4)G hexoside, G(8–O–4)FA hexoside (FA, unit derived from ferulic acid), and G(8–5)FA hexoside, were found in Arabidopsis stems and accumulated in several mutants with an altered monolignol biosynthesis (Vanholme et al., 2012b). This was most striking in stems of Arabidopsis mutants with a concomitant upregulation of FERULATE 5 HYDROXYLASE and downregulation of CAFFEIC ACID O-METHYLTRANSFERASE (COMT) expression (Vanholme et al., 2010; Weng et al., 2010). These mutants have an increased flux toward 5-hydroxyconiferyl alcohol, an alternative monolignol that results in lignins enriched in 5-hydroxyguaiacyl units. The stems of these mutants also accumulated a series of 5-hydroxyguaiacyl-containing dilignol and trilignol hexosides (Vanholme et al., 2010). Recently, a wide range of oligolignol hexosides has been detected in wild-type Arabidopsis leaves (Morreel et al., 2014), and some were also found in flax stems (Huis et al., 2012; Chantreau et al., 2014).
The detection of oligolignol hexosides is intriguing because the aglycones are expected to be formed by oxidative radical coupling within the cell wall, whereas glycosylation occurs in the cytosol. This raises speculation as to whether oxidative coupling between monolignols also takes place intracellularly. Such intracellular oligolignols could then be transported to the vacuole after glycosylation (Figure 2, gray background). In order to study this, we first cataloged the oligolignols and their derivatives from vacuoles of Arabidopsis using candidate substrate product pair (CSPP) networks (Morreel et al., 2014). This revealed that a plethora of oligolignol hexosides is stored in the vacuole and that their aglycone structures were consistent with combinatorial coupling of phenylpropanoid radicals. Next, feeding studies of protoplasts with 13C-labeled mono- and dilignols were performed, and the results provided strong evidence that combinatorial cross-coupling of monolignol radicals is not restricted to the apoplast as currently thought, but also occurs in the cytoplasm. Together, these results unveil a biochemical sink for monolignols that is produced via an intracellular pathway sharing characteristics of lignin polymerization (combinatorial cross-coupling) and (neo)lignan biosynthesis (reduction, hexosylation, and malate esterification).
RESULTS
Leaf Vacuoles Contain a Plethora of Hexosides and Malate Esters of Oligolignols
Earlier metabolite profiling studies on Arabidopsis stems and leaves revealed the presence of a large series of (neo)lignan/oligolignol hexosides (Morreel et al., 2004; Vanholme et al., 2012b). To investigate (1) whether the leaf (neo)lignan/oligolignols are localized in the vacuole and (2) whether their structures are in agreement with those expected from combinatorial monolignol radical coupling, Arabidopsis leaf vacuoles were prepared for subsequent phenolic profiling.
Crucial for the demonstration of the subcellular localization of specific phenolic compounds was the purity of the isolated vacuoles, and this was verified by light microscopy and protein gel blotting as described previously (Ranocha et al., 2013) (Supplemental Figures 1A and 1B). To further confirm the purity of the isolated vacuoles, we performed silver-stained protein gel analyses (Supplemental Figure 1C). Rubisco, the most abundant protein in leaf parenchyma cells (accounting for ∼50% of the total soluble leaf protein) (Fan et al., 2009), was almost invisible in the vacuolar extracts. Based on these results, we concluded that the isolated vacuoles could be used for subcellular metabolomics.
The purified vacuoles were then subjected to phenolic profiling using ultra-high-performance liquid chromatography (UHPLC)-electrospray ionization-Fourier transform-ion cyclotron resonance-mass spectrometry in negative ionization mode (Figure 3A) to identify as many peaks as possible associated with (neo)lignans/oligolignols. For this purpose, all detected peaks were used to construct a CSPP network. Within the obtained network, we focused on peaks that showed specific mass differences corresponding to well-known reactions occurring in lignin polymerization (i.e., the radical-radical cross-coupling reactions) and in (neo)lignan metabolism (e.g., the hexosylation and glucose-malate transesterification) (Table 1). By connecting all peak pairs in the network, structurally related compounds could easily be traced. A section of the network containing most of the (neo)lignans/oligolignols is shown in Figure 3B (Supplemental Data Set 1). In total, among the various profiled phenolics, coniferin, coniferin hexoside, and 34 glycosylated (neo)lignans/oligolignols were detected in the vacuoles (Table 2). The vacuolar (neo)lignan/oligolignols were mainly composed of G and FA units, but S units were also present. The FA units were always coupled at their 4–O position and appeared as end groups. A few compounds had an end group derived from sinapic acid. The units were 8–8-, 8–5-, and 8–O–4-coupled, and in the case of the 8–8-linkage, both resinol and lariciresinol, comprising a reduced tetrahydrofuran ring, units were present. In addition to the dimers, trimers and a tetramer were also detected. For the 8–O–4-linkage, both erythro and threo configurations were observed based on the mass spectrometry fragmentation spectrum (Morreel et al., 2010a) (Table 1). In general, the observed structural composition of the vacuolar (neo)lignans/oligolignols (i.e., the units and linkages and the detection of higher-order oligomers and both stereomers for the 8–O–4-linked compounds) was largely in agreement with their synthesis via the same radical-radical combinatorial coupling reactions that occur during lignin polymerization.
Vacuolar Phenolic Profiling.
(A) LC-MS chromatogram of Arabidopsis leaf vacuoles. See Methods for experimental details. Inset: light microscopy image of isolated vacuoles. The y axis represents the relative abundance of the peak height.
(B) Illustrative CSPP subnetwork for the vacuolar (neo)lignans/oligolignols. CSPP conversions and shorthand naming are mentioned in Table 1 and the footnote of Table 2, respectively. The subnetwork was obtained from the full network via network propagation (Morreel et al., 2014). Some edges with benzenoids and phenylpropanoic acids (see Supplemental Data Set 1 for their LC-MS-based structural characterization data) are present in the network.
A particularly remarkable feature of all (neo)lignans/oligolignols detected in the vacuoles is that they contain hexose moieties. For some oligolignols having a hydroxycinnamic acid-derived end group, this end group was esterified by malate. These secondary enzymatic modifications, as well as the presence of Gred(8–8)G [lariciresinol, i.e., a reduced form of G(8–8)G or pinoresinol], and their vacuolar localization do not allow us to assign them unambiguously as oligolignols, which are, strictly defined, lignin oligomers (Morreel et al., 2004). Taken together, our profiling data demonstrate that (neo)lignan/oligolignol hexose derivatives are present in the leaf vacuoles and that their syntheses share characteristics of combinatorial radical coupling as in lignin biosynthesis and postcoupling modifications of (neo)lignan biosyntheses.
Leaf Protoplast Feeding
One explanation for the presence of glycosylated oligolignols in leaf vacuoles is that the core structures of the vacuolar (neo)lignan/oligolignol derivatives arise from combinatorial coupling inside the cells. To examine this possibility, we performed monolignol feeding studies with Arabidopsis leaf protoplast cultures, followed by the generation of separate phenolic profiles from the medium and the protoplast fractions. Our first feeding experiments of Arabidopsis leaf protoplast cultures with coniferyl alcohol resulted in the accumulation of dilignols both in the medium and in the protoplasts separated from the medium (Supplemental Figure 2). Hence, it was not immediately clear whether the dilignols in the protoplast cells were derived from intracellular coupling or from (1) import of dilignols from the medium through the plasma membrane, (2) adherence to the plasma membrane of dilignols that were made in the medium, or (3) contamination of the protoplast cells with remnants of the medium (containing dilignols) after separation of the protoplasts from the medium. To circumvent these biases, collectively termed here as “import/adherence/contamination” for convenience, a more diagnostic experiment had to be designed. This involved the simultaneous feeding of the protoplast cultures with both [13C6]-coniferyl alcohol and [13C4]-dilignols. Whereas the protoplast concentrations of the [13C4]-dilignols solely arise from “import/adherence/contamination” of these labeled dilignols, those of the [13C12]-dilignols will additionally result from intracellular coupling of [13C6]-coniferyl alcohol (Figure 4). Thus, the fed [13C6]-coniferyl alcohol might be coupled in the medium and then actively imported into the protoplasts or via diffusion (Figure 5, upper model), or it might in addition be immediately imported and then coupled intracellularly (Figure 5, middle model). Because the [13C4]-dilignols and the [13C12]-dilignols represent the same compounds, they undergo the same reactions, i.e., “import/adherence/contamination” and any intra- and extracellular metabolic or chemical conversions, with the same kinetics for both 13C-labeled compounds. However, because the [13C4]-dilignols and their corresponding [13C12]-dilignols are different isotopes, they can be clearly distinguished in a liquid chromatography-mass spectrometry (LC-MS) chromatogram because of their different m/z value, allowing their relative abundances in the protoplasts versus the medium to be calculated. [13C4]-G(8–O–4)G was chosen for this cofeeding experiment rather than [13C4]-G(8–8)G or [13C4]-G(8–5)G because (1) G(8–8)G and G(8–5)G are precursors for two well-known secondary metabolic pathways, i.e., lignan and neolignan biosynthesis, which might complicate interpretation of the results, and (2) G(8–O–4)G synthesis yields both the threo- and erythro-isomers of [13C4]-G(8–O–4)G. Assuming nonenzymatic intracellular oxidative coupling, the cofeeding results are expected to be similar for both stereomers.
Cofeeding Principle of Protoplasts.
To the medium of protoplast cultures, 1 mM [13C6]-coniferyl alcohol and 1 mM [13C4]-G(8–O–4)G (containing both the threo- and erythro-diastereomers) were fed. [13C6]-coniferyl alcohol will dimerize in the medium to [13C12]-G(8–O–4)G (reaction rate v3) but might also diffuse through the plasma membrane (reaction rate v1a) and be dimerized inside the protoplast (reaction rate v4). However, protoplastic synthesized pools of [13C12]-G(8–O–4)G will additionally be formed from transport through the plasma membrane of dimers formed in the medium (reaction rate v2a) and/or adhering of the latter to the plasma membrane (reaction rate v2b). To determine intracellular dimerization, the ratios of the [13C12]-G(8–O–4)G concentrations in the protoplast and the medium were compared with those of the [13C4]-G(8–O–4)G concentrations (see formula “biological interpretation”). Protoplast pools of the latter [13C4]-dilignol only arise from transport through or adhering to the plasma membrane (reaction rates v2a and v2b). Both [13C4]-G(8–O–4)G and [13C12]-G(8–O–4)G represent the same compound and participate noncompetitively in the same (bio)chemical reactions, but they can be distinguished in a mass spectrometer as they represent different isotopes of the same compound. Therefore, a statistical advantage of the cofeeding experiment becomes clear after rearranging the “biological interpretation” formula to that mentioned under “statistical interpretation”: by taking the ratios of the [13C4]-G(8–O–4)G and [13C12]-G(8–O–4)G concentrations in one particular compartment (protoplast or medium) and, hence, between the two corresponding m/z peaks in one particular chromatogram, biological and experimental variations are reduced. v1b, reaction rate for plasma membrane adherence of [13C6]-coniferyl alcohol; v2, overall reaction rate for contribution of [13C12]-G(8–O–4)G from the medium to the [13C12]-G(8–O–4)G protoplast pool.
Cofeeding Models.
The upper model pictures the odds ratio when protoplast dilignol pools are solely the result of contribution from the medium due to, e.g., transport through the plasma membrane (odds ratio = 1). The middle and lower models provide explanations for the observation of an odds ratio higher than one due to intracellular coupling (intracellular coupling model) and via membrane-localized peroxidase-mediated coupling followed by channeling the reaction products to the cytoplasm via a transporter (dimer uptake model), respectively. In both the middle and lower models, additional diffusion of dimers from the medium through the plasma membrane and vice versa is illustrated. Molecules labeled in red are either [13C6]-coniferyl alcohol or derived from [13C6]-coniferyl alcohol; molecules labeled in blue are [13C4]-dilignols.
Samples from the medium and protoplasts were taken following incubation times of 3, 6, and 15 h (chosen based on a preliminary time-course feeding study with [13C2]-coniferyl alcohol; Supplemental Figures 3 and 4, Supplemental Table 1, and Supplemental Methods). For labeled G(8–O–4)G, it is expected that the [13C12]/[13C4] ratio in the protoplasts will be significantly larger than that in the medium whenever intracellular coupling takes place in addition to dilignol “import/adherence/contamination” (see “statistical interpretation” in Figure 4; combination of upper and middle model in Figure 5). It should be stressed that, by considering dimensionless ratios of peaks instead of their absolute abundances, variation from chromatogram to chromatogram is eliminated. Therefore, cofeeding studies are less biased by the technical variation due to the extensive protoplast isolation procedure than time-course-based flux studies would be. Noteworthy, for the threo- as well as the erythro-isomer of 13C-labeled G(8–O–4)G, the average odds ratios (an odds ratio is the ratio between the [13C12]/[13C4] proportion in the protoplasts to that in the medium) obtained after 3, 6, and 15 h of incubation were similar in magnitude, ranging between 3 and 4 (Table 3). Consequently, all data across incubation times were combined and subjected to one-sided t tests (see Methods). Significantly higher odds ratios than unity were evident for both the threo- and erythro-isomers (column “averaged” in Table 3; threo-isomer, P = 1.6E-04; erythro-isomer, P = 2.7E-04). Furthermore, at 15 h of incubation, no protoplast trilignol pools were observed. Based on one-sided t tests (three biological replicates), these results were confirmed in a second experiment in which cofeeding occurred for 20 h (Table 3; threo-isomer, P = 3.8E-03; erythro-isomer, P = 4.4E-02). These results indicate that the [13C12]-G(8–O–4)G pools in the protoplasts were larger than would be expected if they had been established solely by “import/adherence/contamination” from the medium.
Perturbation of Monolignol Biosynthesis Affects the Vacuolar Oligolignol Pool
It is generally known that perturbation of monolignol biosynthesis results in a modified amount and composition of lignin in the cell wall (Vanholme et al., 2008, 2010). If the vacuolar oligolignols result from combinatorial monolignol radical cross-coupling reactions similar to those in the apoplast, perturbation of monolignol biosynthesis is expected to result in modifications of the vacuolar oligolignol pool as well. To verify this, we profiled Arabidopsis comt mutants, defective in COMT, which catalyzes the methylation of 5-hydroxyferulic acid and 5-hydroxyconiferaldehyde into sinapic acid and sinapyl alcohol, respectively. Lignin in comt mutants is characterized by benzodioxane units formed by the incorporation of 5-hydroxyconiferyl alcohol into the lignin polymer (Vanholme et al., 2012b). Comparative vacuolar profiling of wild-type Arabidopsis and comt mutants (Supplemental Figure 5) revealed the appearance of multiple isomers of the benzodioxane oligolignol hexoside G(8–O–4)5HFA hex, the aglycone that is made through the combinatorial coupling of the COMT substrate 5-hydroxyferulic acid with coniferyl alcohol. In addition, phenolic profiling of ccoaomt1 mutants (Supplemental Figure 5), defective in CAFFEOYL-CoA O-METHYLTRANSFERASE1 (CCoAOMT1), revealed the presence of multiple isomers of the benzodioxane G(8–O–4)CA hex in the vacuolar oligolignol pool, as a consequence of the combinatorial coupling of coniferyl alcohol with the substrate of the CCoAOMT1 enzyme, caffeic acid. In addition, we aimed to investigate whether the overall vacuolar oligolignol pool was less abundant in the Arabidopsis c4h mutant, which is defective in CINNAMIC ACID 4-HYDROXYLASE (C4H). Mutations in C4H reduce the overall flux through the phenylpropanoid and monolignol biosynthetic pathways; hence, c4h mutants deposit less lignin (Schilmiller et al., 2009). However, making conclusions about a reduced abundance based on the quantification of compounds in vacuoles is technically very tedious. Therefore, in contrast to the glycosylated benzodioxane-containing compounds described above, for which the observation in vacuoles was obvious because they were undetectable in wild-type plants, we verified the reduced flux through the pathway in c4h mutants based on metabolite profiles of leaves. The latter analyses were based on the abundance of the most prominent oligolignol hexosides, i.e., G(8–5)G hex, G(8–8)G hex, and G(8–5)FA hex and the threo- and erythro-isomers of G(8–O–4)FA hex. The abundances of all of these oligolignols were significantly lower in Arabidopsis c4h mutants compared with their levels in wild-type profiles (Supplemental Figure 6). In conclusion, perturbation of the flux through the monolignol biosynthetic pathway leads to changes in the abundance and the composition of the vacuolar oligolignols. The latter observation further supports our argument that the vacuolar oligolignols arise from combinatorial radical coupling, as is the case for apoplastic lignin.
DISCUSSION
Coniferyl, sinapyl, and p-coumaryl alcohol are the end products of the monolignol biosynthetic pathway but, once produced in the cytoplasm, they can be further processed in several ways. In recent years, it has been suggested that monolignols can be glucosylated and stored as such in the vacuoles (Miao and Liu, 2010), a feature they share with some phenylpropanoid precursors (Figure 2). For example, sinapoyl glucose and sinapoyl malate are present in the vacuoles of Arabidopsis cells (Hause et al., 2002). However, the bulk of the monolignols will be exported to the cell wall for oxidation by peroxidase and/or laccase whereupon the resulting radicals cross-couple nonstereospecifically to form the lignin polymer (Ralph et al., 2004; Vanholme et al., 2012a). In addition, the monolignols serve as building blocks for the production of (neo)lignans. The latter biosynthetic routes start with a dirigent protein-guided stereospecific dimerization of two monolignol radicals. Circumstantial evidence suggests that this oxidative coupling also occurs in the cell wall because dirigent proteins contain a signal peptide for secretion, and immunolocalization has revealed the apoplastic localization of these proteins (Burlat et al., 2001).
These different apoplastic coupling products can serve as substrates for enzymatic conversion (mainly glycosylation) within the cell, on the condition that they are transported across the plasma membrane into the cytoplasm. Here, we present substantial evidence that oxidative cross-coupling reactions between monolignol and oligolignol radicals also occur in the cytoplasm whereupon the products are glycosylated and stored in the vacuole as oligolignol derivatives.
The Core Structures of the Vacuolar Oligolignols Are Consistent with a Combinatorial Coupling Process
Recently, oligolignol derivatives have been found in the stems of flax (Huis et al., 2012) and in the stems and rosette leaves of Arabidopsis (Vanholme et al., 2012b; Morreel et al., 2014). To verify whether these compounds are present in the vacuole, leaf vacuoles were purified and as many oligolignols as possible traced using UHPLC-mass spectrometry followed by CSPP network generation. The core structures (aglycones) of the various dilignols observed in the vacuolar phenolic profiles represented combinations of all linkage types (8–8-, 8–5-, and 8–O–4-linkages) and units (mainly G, but also S) that are present in lignin as well. All trimers and tetramers observed in the vacuolar oligolignol pool had 8–O–4-linkages in both the threo- and erythro-configurations. This type of linkage is prominent in lignin, since it is formed by an endwise polymerization process (Ralph et al., 2004), i.e., when monolignol-oligolignol cross-coupling is favored. This is in contrast with a bulk polymerization process that would favor more 4–O–5- and 5–5-linkages resulting from oligolignol-oligolignol cross-coupling along with more dimerization from coupling of monolignols.
The outcome of the combinatorial cross-coupling of phenylpropanoids leading to lignin is known to be highly dependent on the flux through the phenylpropanoid and monolignol biosynthetic pathways (Vanholme et al., 2012b). Here, we show that perturbation of this pathway also affects the vacuolar oligolignol pool. For example, when the methylation steps are perturbed, as in the comt and ccoaomt1 mutants, o-dihydroxybenzene-type phenylpropanoids (e.g., 5-hydroxyferulic acid and caffeic acid in comt and ccoaomt1 mutants) are synthesized that, upon combinatorial cross-coupling, lead to a benzodioxane-type unit following 4–O–8-coupling, for which the formation is purely governed by chemical reaction propensities. Although cross-coupling of monolignol radicals with the radicals of these o-dihydroxybenzene-type phenylpropanoids yield 8–O–4-linkages reminiscent of the radical-radical cross-coupling of, e.g., coniferyl and sinapyl alcohols, the subsequent rearomatization of the quinone methide intermediate occurs via internal trapping by the 3- or 5-hydroxyl group rather than by an external nucleophile (e.g., water) as happens during 8–O–4-coupling of the traditional monolignols (Morreel et al., 2004; Vanholme et al., 2010, 2012b). The fact that benzodioxane-containing molecules were observed in the vacuolar oligolignol pools of the comt and ccoaomt1 mutants indicates that perturbation of the monolignol biosynthetic pathway affects both the intracellular and extracellular oligolignol pools.
Evidence for Intracellular Combinatorial Coupling in Oligolignol Derivative Synthesis
When protoplast cultures were fed with monolignols, dilignols were abundantly present in the protoplast medium (Supplemental Figure 2). These extracellular oligolignol pools might thus be expected to contribute to the protoplast oligolignol pools due to import through or adherence to the plasma membrane or due to contamination of the purified protoplasts by the medium. However, if this was the major or only route leading to the intracellular oligolignol pools and if both oligolignol pools (intracellular and medium) would be equilibrated, the concentration profiles of the 13C-labeled oligolignols in the protoplasts would reflect those in the medium. In contrast, different flux dynamics in protoplasts and medium were observed. Different flux dynamics can be expected if (1) diffusion does occur, but too slowly for equilibration to occur during the time course of the feeding experiment; (2) an irreversible active import of dilignols from the medium into the protoplasts occurs that would prevent the pools in the medium and protoplasts from equilibrating; and/or (3) the protoplast 13C-labeled oligolignol pools are derived from intracellular combinatorial coupling. The latter possibility, i.e., the occurrence of intracellular coupling, was further verified via cofeeding of the protoplast cultures with both [13C6]-coniferyl alcohol and [13C4]-labeled dilignols. Whereas the protoplast pools of the [13C4]-labeled dilignols only arise from “import/adherence/contamination,” those of the [13C12]-labeled dilignols additionally result from the intracellular coupling of the imported [13C6]-coniferyl alcohol (Figure 4). If the protoplast pools were only derived via “import/adherence/contamination” from the medium, then the ratio of the protoplast versus the medium pools of the [13C12]-dilignols should be the same as that obtained for the [13C4]-dilignols because neither the import mechanism nor the intracellular downstream (bio)chemical conversions distinguish the 13C-labeled dilignol isotopes (see “biological interpretation” in Figure 4). Likewise, the ratio between the [13C12]- and [13C4]-labeled dilignol abundances in the protoplast would be equal to the corresponding abundance ratio in the medium if dilignol formation would only occur in the medium. However, and in agreement with the occurrence of intracellular coupling, the ratio between the [13C12]- and [13C4]-labeled dilignol abundances in the protoplast was much larger than that in the medium. The observed odds ratios range between ∼3 and 4 (Table 3), meaning that for each molecule of intracellular G(8–O–4)G arising from “import/adherence/contamination” from the medium, two to three molecules are produced via intracellular coupling. In conclusion, our feeding experiments of Arabidopsis leaf protoplasts indicate that protoplast oligolignol pools were mainly derived from intracellular coupling.
Can Oxidative Coupling at the Plasma Membrane Followed by Import into the Cytoplasm Explain the Cofeeding Results?
An odds ratio significantly larger than one could also arise when the site of monolignol radical formation, e.g., by peroxidase, is localized at the plasma membrane and the newly generated [13C12]-labeled dilignols are transported to the cytoplasm (Figure 5, lower model). The plasma membrane-localized peroxidase would enrich the concentrations of [13C12]-labeled dilignols compared with those of [13C4]-labeled dilignols at these particular spots on the outer side of the plasma membrane and thus could lead to a preferential uptake of the [13C12]-labeled dilignols by an adjacent transporter. The channeling via a transporter is a prerequisite for observing high protoplast [13C12]-dilignol concentrations. Assuming a nonchanneling mechanism in which the transporter was randomly distributed within the plasma membrane and not necessarily in contact with, e.g., the peroxidase, lower protoplast concentrations of the [13C12]-labeled than those of the [13C4]-labeled dilignols would be expected. Equal concentrations of [13C6]-coniferyl alcohol and [13C4]-G(8–O–4)G were fed to the protoplast cultures. Thus, assuming that encounters between the fed molecules and the protoplast plasma membrane were diffusion-limited, this would yield equal concentrations of [13C6]-coniferyl alcohol and [13C4]-G(8–O–4)G at the plasma membrane. Even if all [13C6]-coniferyl alcohol were to be converted to [13C12]-G(8–O–4)G, this would lead to plasma membrane-localized concentrations of the latter [13C12]-labeled dilignol being only half of those of their [13C4]-labeled counterparts, since two monolignol molecules are consumed for the production of each dilignol molecule. Thus, if a preferential uptake mechanism would explain our results, it should involve a channeling mechanism between membrane-localized oxidation and a transporter. Such a model is possible, since it has recently been demonstrated that lignification in the Casparian strip employs a protein-complex-bound peroxidase (Lee et al., 2013). Therefore, our results can equally well be explained by either intracellular coupling (Figure 5, middle model) or by a preferential uptake model (Figure 5, lower model). However, additional biochemical considerations favor the intracellular coupling model rather than the preferential uptake model. First, in addition to dilignol hexosides, trilignol hexosides and a tetralignol hexoside also were detected in the phenolic profiles of the vacuoles. Given the multitude of (hexosylated) dimers and higher order oligolignols detected in the vacuole, the proposed metabolic channel would need to be able to accommodate a large diversity of structures. Second, if intracellular combinatorial coupling occurs, coupling products between additional classes of phenolics are expected to be present. Indeed, flavonolignans, coumarinolignans, and stilbenolignans are well known in plants, being consistent with oxidative cross-coupling of monolignols with flavonoids, coumarins, or stilbenes (Begum et al., 2010). For example, several coupling products between the dihydroflavonol taxifolin and coniferyl alcohol have been observed in the thistle Silybum marianum: both the 8–O–4- (silybin A and B, isosylibin A and B) and the 8–5-coupled compounds (silychristin), as well as all possible diastereomers and regioisomers of compounds involving the former linkage (Dewick, 2009). Finally, further support for intracellular coupling has recently been obtained from the comparative profiling of wild-type poplars and poplars deficient in phenylcoumaran benzylic ether reductase. In these plants, different cysteine adduct analogs of 8–O–4-dilignols have been identified (Niculaes et al., 2014). In contrast to the formation of the 8–5- and the 8–8-units following the radical coupling reactions, that of the 8–O–4-units needs an external nucleophile to rearomatize the intermediate quinone methide. Commonly, this role is fulfilled by a water molecule. However, in the phenylcoumaran benzylic ether reductase-deficient poplars, the thiol function of cysteine competes with water, a reaction that is assumed to occur in the cytoplasm. Together, our data indicate that oxidative coupling of monolignols also occurs intracellularly.
Enzymatic Modifications Following Combinatorial Coupling
At least two types of enzymatic modifications following combinatorial coupling were evident from the vacuolar oligolignol structures, i.e., reduction of resinol units and glycosylation of phenolic and aliphatic end groups. Upon 8–8-coupling of two monolignol radicals, a quinone methide is formed that rearomatizes via internal trapping by the 9-OH functions, leading to the formation of a resinol unit consisting of two tetrahydrofuran rings (furofuran type; Figure 1). However, among the vacuolar oligolignols, 8–8-linked moieties were present as resinol units and as their reduced forms, i.e., units in which one of the tetrahydrofuran rings was reduced (furan type). Such a reduction can only occur postcoupling and is likely mediated by an NADPH-dependent pinoresinol-lariciresinol reductase, an enzyme known to convert pinoresinol to lariciresinol in Arabidopsis in vivo (Nakatsubo et al., 2008). The lack of a signal peptide suggests that pinoresinol-lariciresinol reductase is a cytoplasmic enzyme.
A second type of enzymatic modification is the glycosylation of phenolic end groups of phenylpropanoids/monolignols. Upon glycosylation, these compounds become more hydrophilic, improving their transport to the vacuole (Wink, 1997). The conversion to glycosides stabilizes and might detoxify the monolignol/phenylpropanoid (Whetten and Sederoff, 1995). Glucosyltransferases (GTs) are known to occur in the cytosol (Bowles et al., 2006), and Arabidopsis GTs that convert monolignols/phenylpropanoids to their glucosides belong to the UGT72E1-E3 subfamily (Lanot et al., 2008). It remains to be demonstrated whether the same GT subfamily is able to glucosylate the phenolic function of dimers and higher order oligolignols.
The structures of some of the vacuolar oligolignols were adorned with malate or hexose esterified to units derived from phenylpropanoic acids, more specifically ferulic or sinapic acid. Both feruloyl and sinapoyl glucose are known to be synthesized in the cytosol from their corresponding phenylpropanoic acid aglycones by Arabidopsis GTs belonging to the UGT84A1-A4 subfamily (Lim et al., 2002). In Arabidopsis, both feruloyl and sinapoyl glucose can be transesterified to feruloyl and sinapoyl malate, the latter being a well-known UV protectant (Fraser et al., 2007). In agreement, a vacuole-localized 1-O-sinapoylglucose:malate sinapoyltransferase (Hause et al., 2002) is present in Arabidopsis, indicating that the phenylpropanoyl glucose first has to be transported to the vacuole to allow transesterification. A combination of glycosylation followed by transport to the vacuole and transesterification to the malate ester might also occur for oligolignols containing units derived from phenylpropanoic acids. Alternatively, the latter type of oligolignol esters could arise from oxidative cross-coupling of monolignols with feruloyl or sinapoyl esters that contain a free phenolic function available for oxidation. Because feruloyl and sinapoyl malate are formed in the vacuole, their oxidative cross-couplings with monolignols would involve vacuolar peroxidases. However, because monolignols likely enter the vacuole as their glucosides, they first need to be deglycosylated prior to oxidation of the monolignols by peroxidases. This scenario is possible because vacuolar glucosidases have been described in Arabidopsis (Carter et al., 2004; Xu et al., 2012). However, vacuolar oxidative cross-coupling currently cannot explain the presence of oligolignols containing phenolic glycosides. For the biosynthesis of the latter oligolignols, a postcoupling glycosyltransferase reaction is necessary, and the presence of such enzyme activity in vacuoles has not yet been described. Therefore, the oligolignol glucosides are most likely formed via oxidative cross-coupling of the monolignol and oligolignol radicals in the cytosol, followed by glycosylation and transport to the vacuoles. Indeed, if monolignols are oxidized and combinatorially coupled in the cytoplasm, there is no reason to argue that trimers would not be generated in this compartment as well.
Is There a Biological Role for These Oligolignol Derivatives?
In view of intracellular combinatorial coupling, it is evident that glycosylation of the phenolic function will stabilize and/or detoxify phenolic compounds and prevent the coupling of monolignol radicals to proteins or amino acids (Cong et al., 2013; Diehl et al., 2014; Niculaes et al., 2014). However, after damage of the cell, e.g., during pathogen attack, the oligolignol glycosides stored in the vacuoles will come into contact with the apoplast, after which they may be hydrolyzed by extracellular β-glucosidases and polymerized due to the presence of reactive oxygen species and peroxidases. The cross-coupling of oligolignols would then yield highly heterogeneous and branched lignin polymers. Such a mechanism for leaf parenchyma cells might quickly seal off any apertures after wounding, hence preventing water loss, and forming a physical barrier against pathogen invasion (Lange et al., 1995). Furthermore, this preexisting battery of combinatorially coupled phenylpropanoids in the vacuole can be viewed as a “chemical library” helping to fight off pathogens, i.e., producing such a high number of phenolic compounds making it likely that at least one of them is toxic to the invading pathogen. In addition to this preexisting battery of defense molecules, Arabidopsis might also produce some of them as (neo)lignans in response to pathogen attack (Floerl et al., 2012). It is noteworthy that to produce such a “chemical library,” evolution has apparently made beneficial use of the oligolignol coupling products that are formed as a consequence of the oxidative environment in the cell.
Using vacuolar metabolomics and feeding studies, we have provided strong evidence that monolignols are not only oxidized in the cell wall, as commonly accepted, but also inside the cell and that these monolignol radicals combinatorially couple into a plethora of dimers and small lignin oligomers that are subsequently glycosylated and stored in the vacuole. Together, these results unveil a biochemical sink for monolignols that is produced via a pathway sharing characteristics of lignin polymerization (combinatorial cross-coupling) and (neo)lignan biosynthesis (reduction, hexosylation, and malate esterification).
METHODS
Synthesis
[8,9-13C2]-labeled coniferyl alcohol was prepared from vanillin as follows. Acetylated ethyl [8,9-13C2]-ferulate was prepared as described previously (Ralph et al., 1992; Kim et al., 2003) using acetylated vanillin as the starting material for the Wittig-Horner reaction with triethyl phosphonoacetate-13C2 (99 atom% 13C). The acetylated ethyl [8,9-13C2]-ferulate was obtained as a pale-yellow solid in 93% yield. 1H-NMR (d6-acetone, 500 MHz): δ 1.28 (3H, t, J = 7.2 Hz, CH3CH2), 2.25 (3H, s, Ac Me), 3.87 (3H, s, OMe), 4.19 (2H, qd, J = 7.2, 3.1 Hz, CH3CH2), 6.54 (1H, ddd, J = 162.3, 16.0, 2.4 Hz, H-8), 7.10 (1H, d, J = 8.2 Hz, H-5), 7.24 (1H, dd, J = 8.2, 1.9 Hz, H-6), 7.44 (1H, d, J = 1.9 Hz, H-2), 7.64 (1H, ddd, J = 16.0, 6.9, 3.1 Hz, H-7). Note: protons (CH3CH2, H-7, and H-8) are long-range-coupled to the labeled 8- and 9-13C nuclei with coupling constants of 2.4, 3.1, and 6.9 Hz. A proton (H-8) shows one-bond C–H coupling to the labeled 8-13C with a coupling constant of 162.3 Hz. 13C-NMR (d6-acetone, 125 MHz): δ 119.20 (d, J = 75.7 Hz, C-8), 166.94 (d, J = 75.7 Hz, C-9). [8,9-13C2]-coniferyl alcohol was finally obtained using diisobutylaluminium hydride (10 eq.) at 0°C, as described previously (Quideau and Ralph, 1992), as pale-yellow crystals in quantitative yield; simultaneous phenolic deacetylation also occurs during the reduction (Terashima et al., 1995). 1H-NMR (d6-acetone, 500 MHz): δ 3.76 (1H, m, H-9-OH), 3.85 (3H, s, OMe), 4.18 (2H, dddd, J = 140.2, 10.2, 5.5, 1.6 Hz, H-9), 6.21 (1H, ddtd, J = 150.7, 15.9, 5.4, 4.2 Hz, H-8), 6.48 (1H, ddt, J = 15.9, 7.0, 1.7 Hz, H-7), 6.76 (1H, d, J = 8.1 Hz, H-5), 6.85 (1H, dd, J = 8.1, 1.9 Hz, H-6), 7.05 (1H, d, J = 1.9 Hz, H-2), 7.62 (1H, s, phenol-OH). Note: protons (H-7, H-8, and H-9) are long-range-coupled to the labeled 8- and 9-13C nuclei with coupling constants of 10.2, 7.0, and 4.2 Hz and are also one-bond C–H coupled with coupling constants of 140.2 and 150.7 Hz. 13C-NMR (d6-acetone, 125 MHz): δ 63.36 (d, J = 47.4 Hz, C-9), 128.09 (d, J = 47.4 Hz, C-8).
The [13C4]-coniferyl alcohol dehydrodimers were prepared by oxidative radical coupling using FeCl3, as described previously (Tanahashi et al., 1976). The products were separated by preparative thin-layer chromatography (CHCl3-MeOH, 10:1).
Guaiacylglycerol-β-coniferyl ethers (8–O–4-dehydrodimers): 1H-NMR (d6-acetone, 500 MHz, 60:40 erythro:threo mixture): δ 3.48 (1H, dm, J = 142.6 Hz, Aγ1t), 3.68 (2H, dm, J = 133.5 Hz, Aγ2t and Aγ1e), 3.79 (1H, dm, J = 125.8 Hz, Aγ2e), 3.79 (3H, s, A3t-OMe), 3.80 (3H, s, A3e-OMe), 3.83 (3H, s, B3e-OMe), 3.88 (3H, s, B3t-OMe), 4.19 (5H, dm, J = 141.1 Hz, Aβt and Bγ), 4.30 (1H, dm, J = 144.0 Hz, Aβe), 4.48 (1H, brs, Aαt-OH), 4.59 (1H, brs, Aαe-OH), 4.88 (2H, brs, Aα), 6.28 (2H, dm, J = 153.2 Hz, Bβ), 6.51 (2H, m, Bα), 6.76 (2H, m, A5), 6.89 (4H, m, A6 and B6), 7.08 (6H, m, A2, B2, and B5), 7.53 (2H, brs, phenol-OH). Note: Protons (Aγ1t, Aγ2t, Aγ1e, Aγ2e, Aβt, Aβe, Bγ, and Bβ) have one-bond C–H coupling constants of 142.6, 133.5, 125.8, 141.1, 144.0, and 153.2 Hz. 13C-NMR (d6-acetone, 125 MHz): δ 61.69 (d, J = 42.0 Hz, Aγe), 61.77 (d, J = 42.2 Hz, Aγt), 63.19 (d, J = 47.2 Hz, Bγt and Bγe), 86.50 (d, J = 42.0 Hz, Aβe), 88.31 (d, J = 42.2 Hz, Aβt), 129.39 (d, J = 47.2 Hz, Bβe), 129.50 (d, J = 47.2 Hz, Bβt).
The aromatic-ring-labeled [13C6]-coniferyl alcohol was produced using the same preparation method as for the [8,9-13C2]-coniferyl alcohol, using labeled vanillin (ring-13C6, minimally 99 atom% 13C; formerly Isotec, now Sigma-Aldrich) and triethylphosphonoacetate via ethyl ferulate. 1H-NMR (d6-acetone, 500 MHz): δ3.84 (3H, d, J = 4.0 Hz, OMe), 4.20 (2H, br d, J = 5.8 Hz, H-9), 6.22 (1H, dm, J = 16.0 Hz, H-8), 6.49 (1H, dm, J = 16.0 Hz, H-7), 6.76 (1H, dm, J = 157.3 Hz, H-5), 6.84 (1H, dm, J = 158.0 Hz, H-6), 7.03 (1H, dm, J = 156.0 Hz, H-2), 7.76 (1H, brs, phenol-OH). 13C-NMR (d6-acetone, 125 MHz): δ 109.78 (tm, J = 63.4 Hz, C-2), 115.66 (tm, J = 63.4 Hz, C-5), 120.25 (td, J = 58.6, 5.8 Hz, C-6), 130.05 (td, J = 59.2, 7.1 Hz, C-1), 146.95 (tm, J = 69.5 Hz, C-4), 148.37 (tm, J = 69.5 Hz, C-3).
Growth and Harvest Conditions
Arabidopsis thaliana Columbia-0 seeds were stratified at 4°C in the dark for 48 h and sown on presoaked Jiffy7 pellets (one seed/pellet). Plants were grown in a growth chamber under short-day growth conditions (8 h/16 h light/dark, 21°C day/18°C night, 55% relative humidity, 120 PAR light intensity) for 2 months to obtain large rosettes. Leaves (2 g) were harvested in the morning (10 am) for protoplast and vacuole profiling. Two protoplast and two vacuole samples could be prepared per day. All samples were immediately frozen in liquid nitrogen and stored at −80°C. Growth and harvest were performed in a completely randomized set up. Using the same growth conditions, new Arabidopsis plants were grown for protein analyses and protoplast feeding experiments.
Protoplast and Vacuole Isolation
Protoplast and vacuole isolation from each leaf pool sample was performed according to Robert et al. (2007) with minor modifications. Instead of 0.1% neutral red solution, a 0.3% solution was used and rosettes were grown for 60 d under short-day conditions instead of for 35 d. Following protoplast isolation, a 1-mL fraction was retained for further metabolite analyses, whereas the remaining 29 mL of the protoplast solution was used for vacuole isolation. Protoplasts and vacuoles from each sample were each counted in a Fuchs-Rosenthal chamber and stored at −80°C. The purity of the vacuole preparations and the intactness of the neutral red-stained vacuoles were checked by light microscopy (Leitz Diavert inverted microscopy; Wetzlar).
Protein Extraction
Fully expanded rosette leaves of 60-d-old Arabidopsis plants were snap-frozen in liquid nitrogen and ground in a prechilled mortar. Approximately 1.7 mL powder was transferred to a 2-mL Eppendorf tube and 500 μL lysis buffer was added (1% Nonidet P-40, 200 mM NaCl, 10 mM Tris-HCl, pH 7.5, 5 mM EDTA, 10% glycerol, and complete protease inhibitor cocktail [Roche]). Samples were vortexed and incubated on ice for 5 min, a procedure that was repeated four times. Afterward, samples were centrifuged for 20 min at 4°C and 14,000 rpm in a tabletop centrifuge (5417R; Eppendorf) to clarify the lysate. The supernatant was transferred to a 1.5-mL Eppendorf tube, and the centrifugation step was repeated.
SDS-PAGE Separation
Laemmli SDS-sample buffer (Laemmli, 1970) was added to the protein extracts in a 1/1 (v/v) ratio, and samples were incubated for 8 min at 96°C and for 5 min at 37°C in the case of leaf and vacuolar extracts, respectively. After a short centrifugation step (1 min; 14,000 rpm), the samples were loaded onto an SDS-PAGE gel (20 μL leaf protein extract and 30 μL vacuole sample). The electrophoresis was performed in TGS (Tris/glycine/SDS; Bio-Rad) buffer using 80 and 180 V for proteins migrating in the 4% stacking and 12% separation gel, respectively. Visualization of the separated proteins was performed by Coomassie Brilliant Blue R 250 or silver staining (Shevchenko et al., 1996). The protein gel blot was used previously to visualize specific proteins (Ranocha et al., 2013).
Protoplast Feeding
Protoplasts for feeding experiments were isolated using a Tape-Arabidopsis Sandwich method (Wu et al., 2009). The peeled leaves were placed in protoplast enzyme solution (Robert et al., 2007) and incubated on a shaker at 70 rpm, at 27°C in the dark, for 2 h. Washing steps were performed according to Robert et al. (2007). Because the protoplasts used for metabolite profiling of vacuoles were isolated using the razor blade method (Robert et al., 2007), the metabolite profiles of protoplasts using the two different methods were compared and shown to be highly similar. Similar results were also obtained for profiles of purified vacuoles isolated following these two methods. After washing, protoplasts were placed in MSMO medium (minimal organic Murashige and Skoog medium [Sigma-Aldrich], supplemented with 0.8 mM sucrose, 2.6 μM 1-naphthaleneacetic acid, 0.23 μM kinetin, and 0.8 M mannitol, pH 5.7) and divided into 1-mL samples. The nonlabeled or 13C-labeled standard(s) was (were) added to 1 mL of this protoplast culture to a final concentration of 1 μM for feeding experiments. After incubation, the protoplasts were spun down (80g, 20 min, 20°C). The supernatant (medium) was transferred to a new tube and the remaining pellet was washed twice with 2 mL fresh wash buffer. Following centrifugation (80g, 20 min, 20°C) and removal of the wash buffer, the pellet containing the protoplasts was redissolved in 1 mL Milli-Q water, giving the final 1:1000 dilution factor. Samples were quenched in liquid nitrogen and both purified protoplasts and the medium were stored at −70°C. Phenolics from the protoplasts and from the media were extracted as described below. Several feeding experiments (see also Supplemental Methods) were performed and, unless otherwise mentioned, three replicate cultures were taken for each feeding experiment: (1) Protoplast cultures were fed with 1 mM each of [13C6]-coniferyl alcohol and [13C4]-G(8–O–4)G (comprising both the threo- and erythro-isomers). Medium and protoplasts were analyzed after 3, 6, and 15 h. (2) Protoplast cultures were fed with 1 mM each of [13C6]-coniferyl alcohol and [13C4]-G(8–O–4)G (comprising both the threo- and erythro-isomers). Medium and protoplasts were analyzed after 20 h.
Metabolite Extraction of MSMO Medium, Protoplasts, and Vacuoles
MSMO medium, protoplast, and vacuole samples were thawed and sonicated for 10 min. To remove the lipids, metabolite extraction was performed via solid-phase extraction rather than liquid-liquid extraction as the former method also removes the contaminating Ficoll, a polysaccharide used in the vacuole isolation procedure that causes ion suppression upon LC-MS analysis. Solid-phase extraction was performed using Extract Clean C18 cartridges (4 mL volume, 200 mg sorbent; Grace). The cartridges were conditioned with 2 mL methanol and equilibrated with 2 mL 0.1% formic acid, after which the MSMO medium, protoplast, or vacuole samples were loaded. Following a washing step with 1.5 mL 0.1% formic acid, the metabolites were eluted with 1 mL methanol. Methanol extracts were freeze-dried and dissolved in 200 μL water. The samples were centrifuged (14,000 rpm) for 15 min and 20 μL of the supernatants was metabolically profiled.
Metabolite Profiling
Metabolite profiling was performed by reversed phase UHPLC (Accela; Thermo Electron) coupled to electrospray ionization-Fourier transform-ion cyclotron resonance-mass spectrometry (ESI-FT-ICR-MS; LTQ FT Ultra; Thermo Electron) operated in the negative ionization mode. This FT-ICR-MS instrument is preceded by a linear ion trap (LIT). LC-MS analysis was performed on an Acquity UPLC BEH C18 column (2.1 × 150 mm, 1.7 μm; Waters) using a column temperature and flow of 80°C and 300 μL/min and using water/acetonitrile (99/1, v/v, 0.1% acetic acid; solvent A) and acetonitrile/water (99/1, v/v, 0.1% acetic acid; solvent B). Besides a longer run (gradient: 0 min 95% A, 30 min 55% A, 35 min 0% A), a shorter run (gradient: 0 min 95% A, 20 min 55% A, 22 min 0% A) was applied for some of the feeding experiments because of the limited time available on the LTQ FT Ultra. Ionization source parameter values for the spray voltage, sheath gas, aux gas, and capillary temperature were 5 kV, 20 (arb), 40 (arb), and 300°C, respectively. Full Fourier transform-mass spectrometry (FT-MS) spectra between m/z 120 and 1400 using a resolution of 100,000 and, in parallel, data-dependent LIT-MSn scans using 35% collision energy were recorded. MSn scans consisted of a MS2 scan, two MS3 scans of the two most abundant first product ions, and a MS4 scan of the most abundant second product ion derived from the most abundant first product ion. Except for the MS2 scans where a minimum signal threshold of 5000 was required, all other MSn scans were performed with a minimum signal threshold of 500.
In the case of feeding experiments, chromatographic peak integration was performed using Xcalibur 2.0 SR2. Before integration, the selected “mass range” chromatogram containing solely the full FT-MS scans was Gaussian smoothed (11 points). Peak integration was performed using the parameter default values. In the case of non-targeted profiling experiments, integration was preceded by slicing the full FT-MS spectra from all chromatograms and converting the resulting chromatogram raw files to netCDF files using RecalOffline and Xcalibur 2.0 SR2, respectively. The netCDF files were subsequently imported into the XCMS package (Smith et al., 2006) in R vs. 2.6.1 for the integration and alignment of the full FT-MS chromatograms as previously described (Morreel et al., 2014).
Structural annotation of the peaks was based on retention time alignment and MS2 spectral matching with data from a previous profiling study of Arabidopsis leaves (Morreel et al., 2014). Structural characterization of unknown peaks was based on MSn spectral interpretation taking into account known gas phase fragmentation behavior of particular metabolite classes, such as the flavonoids (Fabre et al., 2001; Morreel et al., 2006), the oligolignols/(neo)lignans (Morreel et al., 2010a, 2010b), and the glucosinolates (Fabre et al., 2007; Rochfort et al., 2008). A rigorous search for all profiled (neo)lignans/oligolignols was applied via CSPP networks.
Comparative Profiling of Vacuole and Leaf Oligolignol Hexosides in Mutants
Wild-type Arabidopsis Columbia-0 and Arabidopsis mutant lines that were downregulated for C4H (ref3-2 mutant) (Schilmiller et al., 2009; Vanholme et al., 2012b), COMT (comt-1 mutant) (Vanholme et al., 2012b), or CCoAOMT (ccoaomt1-3 mutant) (Vanholme et al., 2012b) were grown and harvested as previously described (Morreel et al., 2014). Following phenolic extraction of the leaves of 20 biological replicates of each mutant line and of the wild type (Morreel et al., 2014), LC-MS analysis was performed as described above. Vacuolar isolation and LC-MS profiling conditions were as described above.
Statistics
For the first cofeeding experiment (1), the ratios of the [13C12]-dilignol concentration versus the [13C4]-dilignol concentration in the protoplast and in the medium were computed for the threo- and erythro-isomers of [13C]-G(8–O–4)G. Dividing the protoplast-based ratio by the medium-based ratio yielded the odds ratio. The odds ratios obtained across the various time points were compared using a one-way ANOVA in R version 6.2.0 using the lm function. As no differences were observed after 3, 6, or 15 h of feeding, data of the various time points were combined and a one-sided, one-sample Student’s t test with Welch correction was performed in R to verify whether the odds ratio was significantly different from unity. Such a t test was also performed on the odds ratio obtained for the 20 h cofeeding experimental data (2). Comparative metabolite profiling of leaf extracts was based on one-way ANOVA followed by post-hoc tests using the lm and the Tukey HSD function in R version 6.2.0, respectively.
Accession Numbers
Sequence data from this article can be found in the GenBank/EMBL libraries under the following accession numbers: At2g30490 (ref3-2), At5g54160 (comt-1), and At4g34050 (ccoaomt1-3).
Supplemental Data
Supplemental Figure 1. Purity of Vacuole Isolation.
Supplemental Figure 2. Coniferyl Alcohol Feeding.
Supplemental Figure 3. Time-Course Feeding Study of Protoplast Cultures (Medium Fraction).
Supplemental Figure 4. Time-Course Feeding Study of Protoplasts Cultures (Protoplast Fractions).
Supplemental Figure 5. Vacuolar profiling of Arabidopsis Wild-Type and Mutant Lines.
Supplemental Figure 6. Comparative Profiling of Leaf Oligolignol Hexosides.
Supplemental Data Set 1. Structurally Characterized (Neo)Lignans/Oligolignols in Arabidopsis Vacuoles.
Supplemental Table 1. Statistical Time-Course Models for the [13C]-Labeled Mono- and Oligolignols.
Acknowledgments
We thank Frank Van Breusegem and Pavel Kerchev for critical reading of the article. We thank Annick Bleys for help preparing the article. J.R. and H.K. were funded by the DOE Great Lakes Bioenergy Research Center (DOE BER Office of Science DE-FC02-07ER64494). We thank Stanford University’s Global Climate and Energy Projects “Towards New Degradable Lignin Types” and “Efficient Biomass Conversion: Delineating the Best Lignin Monomer-Substitutes.” We also acknowledge the Hercules Program of Ghent University for the Synapt Q-Tof (AUGE/014); the Bijzonder Onderzoeksfonds-Zware Apparatuur of Ghent University for the Fourier transform ion cyclotron resonance mass spectrometer (174PZA05); and the Multidisciplinary Research Partnership “Biotechnology for a Sustainable Economy” (01MRB510W) of Ghent University.
AUTHOR CONTRIBUTIONS
O.D., K.M., and W.B. designed research. O.D. performed research. H.K. and J.R. contributed new analytical tools. O.D., B.V., and K.M. analyzed data. O.D., K.M., and W.B. wrote the article.
Footnotes
The author responsible for distribution of materials integral to the findings presented in this article in accordance with the policy described in the Instructions for Authors (www.plantcell.org) is: Wout Boerjan (wout.boerjan{at}psb.vib-ugent.be).
Glossary
- CSPP
- candidate substrate product pair
- UHPLC
- ultra-high-performance liquid chromatography
- LC-MS
- liquid chromatography-mass spectrometry
- GT
- glucosyltransferase
- FT-MS
- Fourier transform-mass spectrometry
- Received December 2, 2014.
- Revised December 2, 2014.
- Accepted February 7, 2015.
- Published February 19, 2015.
References
