Standards of delphindin 3-rutinoside (Extrasynthese, Genay, France), chlorogenic acid, nicotine, and tryptophan (Sigma, St Louis, MO, USA) were used in concentrations ranging from 3.12 to 200 μg/ml. cDNAs from ROS1 and DEL were amplified by PCR using genomic DNA of E8:ROS/DEL tomato fruits obtained from (Butelli et al., 2008). PCR products were gel purified and TOPO cloned into the pCR8/GW/TOPO-TA vector (Invitrogen). After sequence verification the ROS1 and DEL fragments were transferred by GATEWAY recombination to pK7WG2 ¹ to create 35S-ROS1 and 35S-DEL. The plasmids obtained were then introduced into Agrobacterium tumefaciens AGL0 (Lazo et al., 1991). AGLO harboring pBINPLUS (pBIN) plasmid (van Engelen et al., 1995) was used as a negative control.

Nicotiana benthamiana infiltrations were carried out as described previously (van Herpen et al., 2010). Briefly, Agrobacterium strains were grown at 28°C for 24 h in LB medium with antibiotics. Cells were resuspended in MES buffer (10 mM MES; 10 mM MgCl2; 100 mM acetosyringone) to a final OD₆₀₀ of 0.5, followed by incubation for 150 min. Agrobacterium strains (pBIN, 35S-ROS1 and/or 35S-DEL) were mixed in equal volumes. N. benthamiana plants were grown in a greenhouse with 16 h light at 28°C. Strain mixtures were infiltrated into the abaxial side of leaves of four-week-old plants using a 1 mL syringe. The plants were grown under greenhouse conditions before further analysis.

For each reported metabolite, data derive from three independent infiltration experiments, performed on different dates. Within each infiltration experiment, the same Agrobacterium solutions were used for three independent leaves on different N. benthamiana plants. Each leaf was analyzed individually.

Metabolite analysis was essentially performed as described in De Vos et al. (2007). Leaves were harvested 5 days post infiltration, snapfrozen in liquid nitrogen and ground to a fine powder. Exactly 200 mg (+/–2 mg) of powder was extracted with 800 μL of methanol containing 1% formic acid. Extracts were sonicated for 15 min, centrifuged at 12500×g for 10 min and filtered through 0.45 μm filters (Minisart SRP4, Biotech GmbH, Germany). An Accela High Pressure Liquid Chromatography system with a photodiode array (HPLC-PDA; Thermo) coupled to an LTQ Ion Trap-Orbitrap Fourier Transformed Mass Spectrometer (FTMS; Thermo) hybrid system was used to detect, identify and quantify compounds (van der Hooft et al., 2012a). A LUNA 3 μ C18 (2) 150 × 2.00 mm column (Phenomenex, USA) was used to separate the extracted metabolites, with MQ water with 0.1% formic acid (A) and acetonitrile with 0.1% formic acid (B) as solvents. A linear gradient from 5 to 35% B at a flow rate of 0.19 ml/min was used. The FTMS was set at a mass resolution of 60,000 HWHM and a mass range of m/z 85-1200, using electrospray ionization in positive mode. Identification of detected compounds was based on retention time, accurate masses of both the parent and fragment ions, in combination with any PDA absorbance spectra (recorded at 240–600 nm).

Data analysis was performed in an untargeted manner, essentially as described (De Vos et al., 2007). Visualization of the HPLC-PDA-FTMS data was performed using Xcalibur 2.1 software (Thermo). The MetAlign software package ² was used for baseline correction, noise estimation, and mass peak alignment (Lommen, 2009). Threshold for peak detection was set at 10000 ions per scan. Mass peaks originating from the same metabolite, including adducts, fragments and isotopes, were subsequently clustered into so-called reconstructed metabolites, using MSClust software (Tikunov et al., 2012). Zero values for compound intensity, i.e., absent in sample, were replaced by random values between 0 and 1. The resulting relative intensity levels of compounds were then used for further statistical analysis, using Student’s t-test. Metabolites of interest were annotated by querying literature data for metabolites, previously detected in N. benthamiana or other Nicotiana species. In addition, we checked for the presence of compounds that have previously been identified from other plant sources and analyzed at the same LC-MS conditions (Moco et al., 2007; van der Hooft et al., 2012b). Quantification of selected compounds in the leaf extracts was performed using the same HPLC-PDA-Orbitrap FTMS system and conditions, using dilution series of authentic standards.

Manduca sexta bioassays were carried out as follows. Infiltrated N. benthamiana leaves were collected after 4 days of infiltration. Leaves were placed in Petri dishes with a wet filter paper and their petioles were placed in Eppendorf tubes that contained 1% agar in water. Per infiltration construct four replicates were used, and every replicate included five larvae. Leaves and larvae were incubated in the dark at 25°C. Leaves were replaced on day 3, and on day 5 the weights of individual larvae were recorded.

We set out to investigate the effect of ROS1 and DEL in N. bentamiana on metabolite profiles by transient expression using agroinfiltration. Leaves of four-week-old N. benthamiana plants were infiltrated with Agrobacterium suspensions and harvested, unless indicated, 5 days after infiltration. Anthocyanin formation could be observed by occurrence of a purple color in the leaves, but no other phenomena could be observed that would distinguish a ROS1 and DEL infiltrated leaf from a pBIN infiltrated leaf (supplemental Figure S1). Aqueous-methanol extracts of the leaves were subjected to HPLC-PDA-Orbitrap FTMS based global profiling, enabling the detection of a large variety of secondary metabolites, including anthocyanins (De Vos et al., 2007). Infiltration of ROS1 and DEL, induced a single peak with UV/Vis-absorbance at 520 nm, as detected by the photodiode array detector (PDA). This peak displayed a typical anthocyanin absorption spectrum and a monoisotopic mass of [M+H] = 611.158 m/z (Figure 1A, inserted panel). The observed m/z, retention time, absorption spectrum, and MSMS fragmentation pattern (Table 1) corresponded to that of the delphinidin 3-rutinoside (D3R) standard. Surprisingly, however, the LC-MS analysis revealed that, next to D3R, a range of other compounds not absorbing light at 520 nm, were also highly upregulated by ROS1 and DEL expression (Figure 1B). To investigate the reproducibility of induction of these compounds, we performed two new independent infiltration experiments, each in three replicates. Data were processed in an untargeted manner and analyzed for compounds that were significantly altered (students t-test, p < 0.05) in ROS1 and DEL infiltrated plants as compared to the control pBIN plants (Table 1). Most of the induced compounds were already present in the control pBIN plants, and could be annotated by comparison with authentic standards, or by MSMS analysis and comparison to literature data of Nicotiana compounds. Figure 2 illustrates their relative changes in content induced by ROS1 and DEL infiltration.

As expected, D3R was highly upregulated in the ROS1 and DEL infiltrated plants. In addition, a number of compounds which are derived from the phenylpropanoid pathway but which do not qualify as anthocyanins or flavonoids were detected. Among these was a group of phenolamides, consisting of hydroxycinnamic acids conjugated to polyamines such as p-coumaroyl-, caffeoyl-, and feruloyl-putrescine. On the other hand, down-regulation of two higher molecular weight caffeoyl-polyamine conjugates, i.e., N-(3-aminopropyl)-N-[4-[[(2E)-3-(3,4-dihydroxyphenyl)-1-oxo-2-propen-1-yl]amino]butyl]-3,4-dihydroxybenzenepropan amide, abbreviated as NB13, and N-[3-[[4-[[3-(3,4-dihydroxyphenyl)-1-oxo-2-propen-1-yl]amino]butyl]amino]propyl]-3,4-dihydroxybenzenepropanamide, abbreviated as NB16, were down-regulated (Table 1). Caffeoylquinic acids and tryptophan were found to be upregulated, indicating a general up-regulation of products of the shikimate pathway. Surprisingly, while nicotine levels were found to be down-regulated, a number of nornicotine conjugates carrying butanoyl, hexanoyl, and octanoyl hydrophobic moieties, which are not related to the phenylpropanoid pathway, were significantly (students t-test, p < 0.05) up-regulated.

Thus, ROS1 and DEL expression results in a range of both phenolic and non-phenolic metabolites in N. benthamiana (Figure 2). The mechanism of this induction (by direct gene activation of by indirect effects) remains elusive.

Earlier reports indicate that in N. tabacum, which is a close relative of N. benthamiana, ROS1 alone is sufficient for the induction of anthocyanins (Orzaez et al., 2009). We compared the effect of infiltration of combination of 35S-ROS1&35S-DEL against 35S-ROS1 alone and 35S-DEL alone on the global secondary metabolite profile (Figure 3). LC-MS analysis revealed that the anthocyanin D3R was still induced by ROS1, albeit two to four fold less than upon ROS&DEL co-infiltration. Other compounds, such as the putrescine conjugates and the nornicotine conjugates, were much less induced in the ROS1 only infiltration, while the DEL only infiltration hardly induced any changes in biochemical profile. In our interpretation, this indicated that ROS1 is essential for the changes in the metabolite profiles that were observed in N. benthamiana, while DEL is required for robust ROS1 activity. Levels of D3R, nicotine and 5-0-(E)-caffeoylquinic acid were quantified in the LC-MS using a dilution series of purified compounds (Figure 3B). D3R was undetectable in both the empty vector pBIN and DEL infiltrated plants. ROS1 infiltration alone yielded about 0.3 mg/g fresh weight of D3R. Combining ROS1 with DEL increased the D3R level about twofold. In control plants, the phenylpropanoid 5-0-(E)-caffeoylquinic acid was already present in well-detectable amounts. Infiltration of ROS1 alone increased its level significantly (students t-test, p < 0.05), while ROS&DEL co-infiltration showed doubled this effect, to 0.8 mg/g fresh weight. Thus, 5-0-(E)-caffeoyl quinic acid increased by a similar manner as D3R upon ROS1&DEL infiltration, relative to ROS1 alone. This result suggests that a general activation of the phenylpropanoid pathway has occurred, rather than a specific activation of the flavonoid/anthocyanin pathway. Nicotine levels were significantly (students t-test, p < 0.05) down-regulated by fivefold in the combination of ROS1&DEL, as compared to the control pBIN infiltrated leaves. Small nicotine down-regulation was also observed in ROS alone infiltration but that change was not significant.

Both nornicotine conjugates and phenolamines have been described as plant defensive molecules against insect herbivores (Zador and Jones, 1986; Severson et al., 1988; Laue et al., 2000; Gaquerel et al., 2014). In addition there is a growing evidence that anthocyanin by itself might have a role in the plant defense against herbivores (Lev-Yadun and Gould, 2009; Markwick et al., 2013). We investigated if plant leaves ectopically expressing ROS1 and DEL exert altered resistance against the tobacco hornworm, M. sexta. M. sexta is a solanaceous specialist adapted to nicotine in tobacco, but not to phenolamines and octanoyl nornicotine (Severson et al., 1988; Kaur et al., 2010). We have fed M. sexta first instar larvae for 5 days with leaves infiltrated with ROS1 and DEL, both alone and in combination, and have compared them to pBIN controls. In Figure 4, the average larval weights after 5 days of feeding are shown. M. sexta larvae fed on pBIN or DEL infiltrated leaves did not differ significantly in weight. In contrast, a significant weight reduction (t-test, p < 0.05) was observed when larvae were fed on leaves expressing ROS1 or ROS1 and DEL.

These results indicate that anthocyanin-regulating ROS1 and DEL TFs act as defensive anti-herbivore TFs in N. benthamiana. The available data do not allow to distinguish the roles of the different compound classes in the anti-herbivore effect.