Specific decorations of 17-hydroxygeranyllinalool diterpene glycosides solve the 1
autotoxicity problem of chemical defense in Nicotiana attenuata 2
3
Sven Heiling1, Lucas Cortes Llorca1, Jiancai Li1, Klaus Gase1, Axel Schmidt2, Martin 4
Schäfer1, Bernd Schneider3, Rayko Halitschke1, Emmanuel Gaquerel4,5 and Ian Thomas 5
Baldwin1 6
7 1Max Planck Institute for Chemical Ecology, Department of Molecular Ecology, Hans 8
Knöll Straße 8, 07745 Jena, Germany 9
2Max Planck Institute for Chemical Ecology, Department of Biochemistry, Hans Knöll 10
Straße 8, 07745 Jena, Germany 11
3Max Planck Institute for Chemical Ecology, Research Group Biosynthesis/NMR, Hans 12
Knöll Straße 8, 07745 Jena, Germany 13
4Centre for Organismal Studies Heidelberg, Im Neuenheimer Feld 360, 69120 14
Heidelberg, Germany 15
5Institut de Biologie Moléculaire des Plantes, CNRS UPR 2357 Université de 16
Strasbourg, 12 rue du Général Zimmer, 67084 Strasbourg, France 17
18
CORRESPONDING AUTHORS: 19
Ian T. Baldwin1 and Emmanuel Gaquerel2 20
1Max Planck Institute for Chemical Ecology, Molecular Ecology Department, Hans Knöll 21
Straße 8, 07745 Jena, Germany, E-mail: [email protected]; Fax: +49 (0)3641 22
571102; 2Institut de Biologie Moléculaire des Plantes, CNRS UPR 2357 Université de 23
Strasbourg, 12 rue du Général Zimmer, 67084 Strasbourg, France, E-mail: 24
26
Short Title: Glycosylation and Phytotoxicity of HGL-DTGs 27
28
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted August 27, 2020. ; https://doi.org/10.1101/2020.08.26.267690doi: bioRxiv preprint
The author responsible for distribution of materials integral to the findings presented in 29
this article in accordance with the policy described in the Instructions for Authors 30
(www.plantcell.org) is Ian T. Baldwin ([email protected]). 31
32
Abstract 33
34
17-hydroxygeranyllinalool diterpene glycosides (HGL-DTGs) are abundant and potent 35
anti-herbivore defense metabolites in Nicotiana attenuata whose glycosylation and 36
malonylation biosynthetic steps are regulated by jasmonate signaling. To characterize 37
the biosynthetic pathway of HGL-DTGs, we conducted a genome-wide analysis of 38
uridine diphosphate glycosyltransferases (UGTs) and identified 107 members of family-1 39
UGTs. Tissue-specific time-course transcriptional profiling revealed that the transcripts 40
of three UGTs were highly correlated with two HGL-DTG key biosynthetic genes: 41
geranylgeranyl diphosphate synthase (NaGGPPS) and geranyllinalool synthase 42
(NaGLS). NaGLS’s role in HGL-DTG biosynthesis was confirmed by virus-induced gene-43
silencing. Silencing the UDP-rhamnosyltransferase, UGT91T1, indicated its role in the 44
rhamnosylation of HGL-DTGs. In vitro enzyme assays revealed that UGT74P3 and 45
UGT74P4 use UDP-glucose for the glucosylation of 17-hydroxygeranyllinalool (17-HGL) 46
to lyciumoside I. UGT74P3 and UGT74P5 stably silenced plants were severely 47
developmentally deformed, suggesting a phytotoxic effect of 17-HGL. Applications of 48
synthetic 17-HGL and silencing of these UGTs in HGL-DTG-free plants confirmed the 49
phytotoxic effect of 17-HGL. Feeding assays with Manduca sexta larvae revealed the 50
defensive functions of the glucosylation and rhamnosylation steps in HGL-DTG 51
biosynthesis. Glucosylation is a critical step that contributes to the metabolites’ 52
defensive function and solves the autotoxicity problem of this potent chemical defense. 53
54
55
56
57
58
59
Introduction 60
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted August 27, 2020. ; https://doi.org/10.1101/2020.08.26.267690doi: bioRxiv preprint
61
In the course of evolution, plants developed versatile defense strategies against 62
biotic stresses, such as herbivores and pathogens. Among these defensive strategies 63
are physical and chemical barriers such as trichomes (Fordyce and Agrawal, 2001), 64
volatiles that attract predators (Kessler and Baldwin, 2001; Halitschke et al., 2008) and a 65
wide array of defensive specialized metabolites. Many small molecules produced as part 66
of a plant's specialized metabolism function as direct defenses by being toxic, repellent, 67
or anti-nutritive for herbivores of different feeding guilds and degrees of specialization; 68
compounds with a broad spectrum of toxicity are likely to be of greater defensive value 69
against a greater diversity of attackers. Noteworthy examples are glucosinolates, 70
alkaloids, and terpenoids. However, these broad-spectrum toxins force the producers to 71
solve the "toxic waste dump problem" of chemical defense, as many direct defense 72
compounds are generally cytotoxic and can damage the tissues of non-adapted 73
producers. 74
Plants have evolved numerous ways of solving this “toxic waste dump problem” 75
for broad-spectrum chemical defenses. One frequently used solution is glycosylation, 76
which is one of the most prevalent and widespread biochemical modifications 77
contributing to the structural and functional diversity of specialized metabolites in plants. 78
Incorporation of sugar molecules into small lipophilic metabolites can regulate 79
storage/localization of defensive metabolites (Gachon et al., 2005; Yadav et al., 2014) 80
and change their bioactivity by detoxifying phytotoxic intermediates. For example, for the 81
defensive deployment of hydrogen cyanide (Gleadow and Moller, 2014) or steroidal 82
saponins (Mylona et al., 2008), plants store these toxins as glycosides, sometimes in 83
particular compartments, away from lytic enzymes that liberate the active toxins in 84
response to the tissue damage that frequently accompanies herbivore and pathogen 85
attack. Similarly, glucosinolates are compartmentalized away from the myrosinases that 86
rapidly hydrolyze them to toxic isothiocyanates and other biologically active products 87
(Matile, 1980; Halkier and Gershenzon, 2006). Steroidal alkaloids are also safely stored 88
as glycosides and silencing GAME-1, a UDP-galactosyltransferase responsible for the 89
glycosylation of these defense compounds in tomato, results in the accumulation of the 90
aglycone, α-tomatidine with severe developmental consequences (Itkin et al., 2011). 91
Glycosylation of the saponin, hederagenin in Medicago truncatula (Naoumkina et al., 92
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted August 27, 2020. ; https://doi.org/10.1101/2020.08.26.267690doi: bioRxiv preprint
2010), and of avenacin A-1 in Avena sativa (Mylona et al., 2008) are other examples 93
which point to a similar chemical sequestration role exerted by glycosylation. 94
Diterpene glycosides (DTGs) are a diverse compound class whose members are 95
often associated with phytotoxic activities (Macias et al., 2008) and have potent anti-96
herbivore resistance/deterrence effects. For example, the abundance of monomers and 97
dimers of capsianosides is correlated with thrips resistance in pepper (Macel et al., 98
2019) and 17-hydroxygeranyllinalool diterpene glycosides (HGL-DTGs) in Nicotiana 99
species have been shown to function in resistance against larvae of the specialist 100
herbivore Manduca sexta (Lou and Baldwin, 2003; Jassbi et al., 2008; Heiling et al., 101
2010) and the generalist herbivore, Heliothis virescens (Snook et al., 1997). 102
HGL-DTGs constitute an abundant and structurally diverse group of specialized 103
metabolites whose function in plants is poorly understood (Heiling et al., 2010). HGL-104
DTGs occur in the aboveground tissues of the native diploid tobacco, Nicotiana 105
attenuata (Heiling et al., 2010) as well as other Nicotiana species (Shinozaki et al., 1996; 106
Snook et al., 1997; Lou and Baldwin, 2003; Jassbi et al., 2006; Heiling et al., 2010; 107
Jassbi et al., 2010; Poreddy et al., 2015; Heiling et al., 2016), several other solanaceous 108
genera including Capsicum (Izumitani et al., 1990; Hashimoto et al., 1997; Iorizzi et al., 109
2001; Lee et al., 2006, 2007, 2008; Lee et al., 2009) and Lycium (Terauchi et al., 1995; 110
Terauchi et al., 1997b, a; Terauchi et al., 1998b; Terauchi et al., 1998a; Roda et al., 111
2003), and the Asteraceae, Blumera lacera (Akter et al., 2016). HGL-DTGs consist of an 112
acyclic 17-hydroxygeranyllinalool aglycone, which is conjugated at its C-3 and C-17 113
hydroxyl groups to different sugar moieties, such as glucose and rhamnose. These 114
sugars can be further glycosylated at the C'-2, C'-4 or C'-6 hydroxyl groups and 115
acetylated or malonylated at the C'-6 hydroxyl group of the glucose(s) (Taguchi et al., 116
2005; Yu et al., 2008, Heiling et al., 2010; Jassbi et al., 2010; Heiling et al., 2016). To 117
date, 45 HGL-DTGs, which differ in their sugar or malonyl decorations, have been 118
putatively annotated or identified in N. attenuata (Figure 1) (Heiling et al., 2016). 119
However, it is unclear which of the many different HGL-DTGs or which structural 120
components of HGL-DTGs are responsible for the observed deterrent (Jassbi et al., 121
2006) and resistance effects against different herbivores (Snook et al., 1997). For 122
example, the geranyllinalool precursor is known to be insecticidal in pine wood and the 123
same compound can be found in the defensive secretions of the termite Reticulitermes 124
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted August 27, 2020. ; https://doi.org/10.1101/2020.08.26.267690doi: bioRxiv preprint
lucifugus (Baker et al., 1982; Lemaire et al., 1990). Jassbi and colleagues (Jassbi et al., 125
2010) suggested that the 17-HGL aglycone is responsible for the feeding-deterrent 126
characteristics of HGL-DTGs, but this hypothesis has not been tested. 127
While the chemical identity and anti-herbivore effects of HGL-DTGs have been 128
the focus of major investigations, less is known about the possible phytotoxic effects of 129
HGL-DTGs and how producing plants cope with these phytotoxic effects. This is, in part, 130
because we know very little about the enzymes required for the biosynthesis of these 131
metabolites. The 17-HGL aglycone is derived from the condensation of three five-carbon 132
units of isopentenyl-pyrophosphate (IPP) and dimethylallyl-pyrophosphate (DMAPP) to 133
produce the diterpenoid precursor, geranylgeranyl-pyrophosphate (GGPP; (Ohnuma et 134
al., 1998; Dewick, 2002)). This reaction is catalyzed by a plastidial geranylgeranyl 135
pyrophosphate synthase (GGPPS; (Jassbi et al., 2008; Heiling et al., 2010)). Formation 136
of GGPP is followed by its allylic rearrangement by a geranyllinalool synthase (GLS; 137
(Falara et al., 2014)) that produces the tertiary alcohol geranyllinalool. However, the 138
enzymes necessary for further hydroxylation and glycosylation steps, which determine 139
most of the structural diversity of HGL-DTGs, remain to be characterized. Li and 140
colleagues (Li et al., 2018) identified and silenced a malonyltransferase (NaMAT1), 141
which is responsible for the malonylation of HGL-DTGs. However, all malonyl moieties 142
of HGL-DTGs are rapidly lost when leaves are ingested by M. sexta larvae, suggesting 143
that the malonylation of HGL-DTGs does not play a central role in anti-herbivore defense 144
(Poreddy et al., 2015). Interestingly, disruption of the uniform malonylation patterns of 145
HGL-DTGs leads to a specific reduction in the floral style lengths of N. attenuata flowers 146
(Li et al., 2018). This shows that specific decorations of a plant's specialized metabolites 147
can play a crucial, but poorly understood, role in plant development. Malonylation of 148
specialized metabolites, such as flavonoids or phenolic glucosides, is a common 149
phenomenon and it has been shown that malonyl moieties alter the molecular properties 150
of these compounds can enhance their solubility in water (Heller and Forkman, 1994) or 151
sequester compounds to molecular compartments, such as vacuoles (Taguchi et al., 152
2010). Whether other intermediates in the HGL-DTG pathway are similarly influential for 153
plant development remains to be explored. 154
To evaluate the defensive value of glycosylation and the potential (auto)toxicity of the 155
17-HGL aglycone for both plant cells and insect herbivores, we investigated the UDP-156
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted August 27, 2020. ; https://doi.org/10.1101/2020.08.26.267690doi: bioRxiv preprint
glycosyltransferases (UDP-GTs) responsible for the glycosylation of HGL-DTGs. To this 157
end, we analyzed the tissue- and herbivory-specific co-regulation of the 107 predicted 158
UDP-GTs in N. attenuata UDP-GTs with two bait genes previously characterized for 159
their involvement in the biosynthesis of 17-HGL, which pinpointed on three novel UGT 160
candidates. RNAi silencing of these UGTs by RNAi approach in N. attenuata and the 161
related HGL-DTG-producing sympatric diploid species N. obtusifolia (syn. 162
N. trigonophylla Dunal) (Heiling et al., 2016) as well as enzyme assays with recombinant 163
proteins identified that UGT74P3 and UGT74P4 are responsible glucosylation of the C-3 164
and C-17 hydroxyl-groups to form lyciumoside I, while UGT91T1 in N. attenuata and 165
NoUGT91T1-like in N. obtusifolia are rhamnosyltransferases in the HGL-DTG pathway. 166
As summarized in one of our previous studies (Heiling et al., 2017), all rhamnosylated 167
HGL-DTG identified so far possess rhamosyl moieties attach to glucose ones, pointing 168
to the conclusion that rhamnosylation requires prio glucosylation. Finally, genetic 169
manipulations revealed that glycosylation of the 17-HGL aglycone by UGT74P3 is 170
crucial to prevent toxic effects to the plant as well as contributes to the metabolites` 171
defensive function during the attack by M. sexta larvae 172
173
174
Results 175
176
Putative identification of UGTs responsible for HGL-DTG biosynthesis 177
To identify UGTs, a genome-wide survey of N. attenuata was performed and a 178
total of 107 putative UGT sequences containing the PSPG motif at the C-terminus were 179
detected (Supplemental Figure 1). These were phylogenetically characterized 180
(Supplemental Figure 2, Supplemental Table 2), their amino acid sequence examined 181
(Supplemental Figure 3, Supplemental Table 1, 3, 4) and their expression analyzed in a 182
full-transcriptome microarray dataset obtained from leaf and root tissues collected at 183
several time-points following simulated leaf herbivory (wounding and application with 184
oral secretions [OS]) as described in supplemental method file 1. 185
HGL-DTGs are secondary metabolites of N. attenuata, whose biosynthetic steps 186
of glycosylation are regulated by the defense phytohormone, jasmonic acid (Heiling et 187
al., 2010). Genes involved in a shared biological process tend to be co-expressed in 188
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted August 27, 2020. ; https://doi.org/10.1101/2020.08.26.267690doi: bioRxiv preprint
large-scale expression datasets due to the fact that they are frequently under the control 189
of a common regulatory network (Saito et al., 2008). Following this rationale, we 190
identified UGT candidates putatively responsible for the biosynthesis of HGL-DTGs by 191
exploring gene co-expression in tissue-specific (local and systemic leaves as well as 192
root tissue) transcriptomes analyzed at 1, 5, 9, 13, 17 and 21 h after the simulated 193
herbivory treatment. For the resulting compendium consisting of 150 microarray 194
expression profiles, we calculated the Pearson Correlation Coefficients (PCC) for 150 195
microarrays of the expressed UGTs with previously identified genes of the HGL-DTG 196
pathway (NaGGPPS and NaGLS, Supplemental Table 5a/b). Transcript accumulation of 197
three UGTs was increased in response to wounding and simulated herbivory and 198
significantly correlated with GGPPS (UGT91T1 – PCC = 0.823, UGT74P3 – PCC = 199
0.608 and UGT74P5 – PCC = 0.684) and GLS (UGT91T1 – PCC = 0.899, UGT74P3 – 200
PCC = 0.872 and UGT74P5 – PCC = 0.868). The PCC value between GLS and GGPPS 201
transcript levels was 0.799 (Supplemental Figure 4). Furthermore, we compared the 202
expression levels between shoot and root tissues of the above genes. Consistent with 203
the absence of HGL-DTG accumulation in N. attenuata roots, transcript accumulation 204
was 50-fold and 3225-fold lower in roots compared to leaves for GGPPS and GLS, 205
respectively. The three candidate UGTs showed a similar profile with 2190-, 127-, and 206
20-fold lower root transcript levels of UGT91T1, UGT74P3, and UGT74P5, respectively. 207
Focusing on the 5 h time point collected from systemic leaves, we detected that GGPPS 208
(increased 6-fold), GLS (increased 8-fold), UGT91T1 (increased 7-fold), UGT74P3 209
(increased 8-fold) and UGT74P5 (increased 8-fold) were all highly up-regulated, relative 210
to untreated control and mechanical wounding conditions, in response to the simulated 211
herbivory treatment (Supplemental Figure 4 and 5, Supplemental Data 1). 212
A phylogenetic alignment with functionally characterized UGTs (Supplemental 213
Figure 6) showed a close relationship of UGT74P3 and UGT74P5 protein sequences 214
with two previously characterized diterpene glucosyltransferases (SrUGT74G1 – steviol 215
glycoside glucosyltransferase from Stevia rebaudiana and CsGIT2 - crocetin 216
glucosyltransferase from Crocus sativus). UGT91T1 showed a close phylogenetic 217
relationship to three of the very few functionally characterized flavonoid 218
rhamnosyltransferases (CmF7G12RT – flavonoid-1, 2-rhamnosyltransferase from Citrus 219
maxima, GmF3G6R – flavonol-3-O-glucoside-α-1, 6-L-rhamnosyltransferase from 220
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted August 27, 2020. ; https://doi.org/10.1101/2020.08.26.267690doi: bioRxiv preprint
Glycine max and PhA3ART – anthocyanidin-3-O-glucoside-α-1, 6-L-221
rhamnosyltransferase from Petunia x hybrida). 222
Based on the co-expression with known genes of the HGL-DTG pathway, the 223
localization of high expression levels in leaf tissues and their phylogenetic relationships, 224
we selected these three candidate UGTs for further characterization. 225
226
Virus-induced gene-silencing reveals a role of three UGTs in HGL-DTG production 227
Using a well-established transient virus-induced gene-silencing (VIGS) approach 228
(Saedler and Baldwin, 2004), we first examined the consequences of independently 229
silencing the three candidate UGTs for HGL-DTG production. Seventeen days after 230
inoculation with Agrobacterium tumefaciens harboring the appropriate constructs, leaf 231
transcript abundance was reduced by 98.5% for UGT91T1, 85% for UGT74P3 and 232
94.3% for UGT74P5 relative to the empty vector controls (EV, pTV00) (Supplemental 233
Figure 7). Co-silencing resulted in reduced transcript abundance of UGT74P3 and 234
UGT74P5 in both, pTVUGT74P3 and pTVUGT74P5 plants (Supplemental Figure 8). 235
To test the hypothesis that UGT91T1, UGT74P3 and UGT74P5 control the 236
glycosylation steps in the HGL-DTG pathway, we analyzed the leaf metabolome by 237
UPLC/TOF-MS of the different VIGS plants under control conditions (lanolin, Lan) or 238
after treatment with a lanolin paste containing methyl jasmonate (Lan + MeJA), known to 239
strongly induce the de novo production of HGL-DTGs (Heiling et al., 2010). We 240
putatively identified and annotated HGL-DTGs using a previously established rapid de-241
replication and identification workflow, which is based on high resolution MS data 242
analysis and a library of putative and identified HGL-DTGs (Heiling et al., 2016). 243
Levels of rhamnosylated HGL-DTGs, most particularly of lyciumoside IV and 244
attenoside, were reduced compared to pTV00 controls, while the non-rhamnosylated 245
HGL-DTGs lyciumoside I and lyciumoside II increased in the HGL-DTG chemotype of 246
pTVUGT91T1 VIGS plants (Figure 2A, Supplemental Data 2). This shift was more 247
pronounced after the Lan + MeJA treatment: in transiently transformed pTVUGT91T1 248
plants, most rhamnosylated HGL-DTGs were barely detectable and non-rhamnosylated 249
HGL-DTGs were dramatically increased compared to pTV00 controls (Figure 2A and B). 250
Furthermore, we detected the 17-HGL aglycone as well as several novel compounds in 251
pTVUGT74P3 and pTVUGT74P5 VIGS plant profiles (Figure 2A). We defined this class 252
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted August 27, 2020. ; https://doi.org/10.1101/2020.08.26.267690doi: bioRxiv preprint
of compounds as intermediate HGL-DTGs and our MS annotation workflow suggested 253
compounds with only one or two sugar moieties, malonylated or not at either the C-3 or 254
C-17 hydroxy-group of the aglycone. 255
Furthermore, we searched for orthologues to these three UGTs in N. obtusifolia, a 256
sympatric tobacco species to N. attenuata. We identified NoUGT74P4 (95% homology 257
to UGT74P3), NoUGT74P6 (93% homology to UGT74P5) and NoUGT91T1-like (92% 258
homology to UGT91T1). Using VIGS, we inoculated A. tumefaciens harboring the 259
appropriate constructs into 25-day-old N. obtusifolia plants and detected transcript 260
reductions of 91.2% in NoUGT91T1-like, 98.5% in NoUGT74P3 and 93.1% in 261
pTVNoUGT74P6, 14 days after inoculation (Supplemental Figure 9). 262
pTVNoUGT91T1-like VIGS plants showed an overall decrease in rhamnosylated 263
HGL-DTGs compared to pTV00 controls and an increase in non-rhamnosylated 264
compounds (Figure 3A and 3B, Supplemental Data 3). In pTVNoUGT74P4 and 265
pTVNoUGT74P6 VIGS plants compared to pTV00 controls, the abundance of 266
intermediate HGL-DTGs as well as the 17-HGL aglycone was increased and non-267
rhamnosylated HGL-DTGs were decreased. Silencing both glucosyltransferases in N. 268
obtusifolia by means of a double construct resulted in the same phenotypic alterations, 269
namely the appearance of the intermediate HGL-DTGs and of the 17-HGL aglycone. 270
Ten novel intermediate HGL-DTGs were putatively identified based on the de-replication 271
workflow for HGL-DTGs. A detailed MS analysis and putative structure description of N. 272
attenuata and N. obtusifolia HGL-DTGs can be found in the Supplemental Table 6 and 7 273
and Supplemental Figure 10a and b and 11a-c. 274
275
17-HGL glucosylation activities of UGT74P3/P4 and UGT74P5 proteins 276
The co-silencing in both UGT74P3 and UGT74P5 resulted in strong overlap in the 277
metabolic alterations of the HGL-DTG profiles produced by pTVUGT74P3 and 278
pTVUGT74P5 plants and suggested that at least one of these enzymes might play a 279
concerted role in the formation of lyciumoside I and lyciumoside II through glucosylation 280
of the C-3 or C-17 hydroxyl-group of 17-HGL. HGL-DTGs typically contain one D-Glc 281
moiety at the C-3 and one at the C-17 hydroxyl group. Lyciumoside II derives from 282
lyciumoside I by the attachment of a second D-Glc at the C’-2 position of the C-17 D-283
Glc. To test this hypothesis, we analyzed the in vitro activity of the UDP-GT enzymes 284
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted August 27, 2020. ; https://doi.org/10.1101/2020.08.26.267690doi: bioRxiv preprint
when recombinantly produced in E. coli. UDP-glucose was used as sugar donor with a 285
mixture of all six isomers of synthesized 17-HGL (Supplemental Figure 12) as 286
substrates. NaUGT74P3 and its N. obtusifolia homologue, NoUGT74P4, readily used 287
UDP-glucose as a donor producing several novel products with a mass-to-charge ratio 288
of m/z 491.3003 [M+Na]+ corresponding to a mono-glucosylated 17-HGL and two 289
additional products with m/z 653.3509 [M+Na]+ corresponding to lyciumoside I and 290
another di-glucosylated 17-HGL (Figure 4). Lyciumoside I was identified by retention 291
time and MS/MS fragmentation to a purified authentic standard and the novel 292
compounds were annotated based comparisons on their MS/MS fragmentation to an in-293
house database of HGL-DTG spectra. Thus, NaUGT74P3 and NoUGT74P4 function as 294
UDP-glucosyltransferases for HGL-DTGs synthesis via their ability to catalyze 295
glucosylation at C-3 and C-17 of the 17-HGL aglycone. When using UDP-glucose as 296
donor and 17-HGL as substrate, NaUGT74P5 and NoUGT74P6 did not show any 297
detectable product in the enzymatic assays. Furthermore, the combination of 298
NaUGT74P5 with NaUGT74P3 or NoUGT74P4 did not result in any further detectable 299
products. 300
301
302
303
Virus-induced gene-silencing of UGTs causes morphological defects and necrosis 304
In addition to the mild TRV infection symptoms such as curly leaves and local 305
chlorosis that were also seen in EV (pTV00) and pTVUGT91T1 plants (Supplemental 306
Figure 13a), we noticed that VIGS plants expressing the pTVUGT74P3 and 307
pTVUGT74P5 constructs displayed severe morphological alterations that ranged from 308
the presence of necrotic spots to necrotic apical meristems and a high percentage of 309
stalled flower buds (Figure 2C, Supplemental Figures 13b and c). Furthermore, we 310
isolated organs of mature flowers collected at anthesis (Figure 2D) and observed 311
significant reductions, compared to observations for pTV00 controls, in the lengths of 312
styles and corolla tubes, and in the width of the corolla limb of pTVUGT74P3 and 313
pTVUGT74P5 VIGS plants, as well as an increased combined lengths of ovary + 314
nectary (Figure 2E). 315
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted August 27, 2020. ; https://doi.org/10.1101/2020.08.26.267690doi: bioRxiv preprint
N. obtusifolia plants showed the same symptoms. Plants transiently silenced with 316
pTVNoUGT91T1-like did not display any additional phenotypic alterations compared 317
with the pTV00 control plants. However, N. obtusifolia plants silenced in UGT74P4 and 318
UGT74P6 expression showed similar morphological modifications as observed in N. 319
attenuata, ranging from necrotic spots, misshaped and deformed leaves to stalled flower 320
buds (Figure 3C, Supplemental Figure 14) with the exception of the apical meristem 321
necrosis. Floral morphology was not examined in N. obtusifolia. 322
Finally, we generated a VIGS construct (pTVGLS) targeting the geranyllinalool 323
synthase NaGLS which was also identified from the co-expression analysis. The 324
encoded protein is predicted to be part of the HGL-DTG biosynthetic pathway by 325
hydroxylating geranyllinalool to produce 17-HGL (Falara et al., 2014). 14 days after 326
inoculation of A. tumefaciens with pTVGLS, we detected a 97.5% reduction of the GLS 327
transcript abundance compared to pTV00 controls, which was associated with a strong 328
reduction in almost all HGL-DTG types (Supplemental Figure 15 and Supplemental Data 329
4). No novel intermediate HGL-DTGs were detected, consistent with the expected 330
absence of the precursor molecule, geranyllinalool. Beyond the typical TRV infection 331
symptoms, no morphological alterations were detected, suggesting that the phenotype 332
of the glycosyltransferase-impaired lines resulted from accumulations of intermediate 333
HGL-DTGs or the aglycone 17-HGL. 334
335
Plants stably-silenced for UGT74P3 and UGT74P5 display severe developmental 336
defects 337
To further elucidate the role of UGT74P3 and UGT74P5 in controlling the flux of 338
HGL-DTG synthesis in N. attenuata as well as the mechanisms responsible for the 339
developmental defects detected during transient gene silencing, we generated 36 340
independent stably-transformed transgenic N. attenuata plants harboring UGT74P3 and 341
UGT74P5 inverted-repeat (IR) silencing constructs in their genomes. Similar to the VIGS 342
experiments, IRugt74p3 and IRugt74p5-silenced plants displayed strong effects on 343
growth and development. Almost all stable T0-transformants developed either a dwarfish 344
growth or a “broom-like” appearance (Figure 5D). Compared to wild-type (WT) plants, 345
many of the transformed plants were strongly retarded in their growth, with thicker 346
woody, or possibly suberized side branches and deformed or thick, succulent-like 347
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted August 27, 2020. ; https://doi.org/10.1101/2020.08.26.267690doi: bioRxiv preprint
leaves. Most flower buds of these transformed plants were small and aborted before 348
fertilization. Few T0-transformants of both constructs exhibited a milder phenotype and 349
produced mature flowers and seed capsules. The phenotypic characteristics of 350
IRugt74p3 transformants could not be transferred to the T2 generation, as T1 plants 351
aborted most flower buds early during development and did not produce fertile flowers. 352
IRugt74p5 transformants could be propagated and two lines from the same T1 parental 353
plant were established (Supplemental Figure 16). Similar to the T0 transformants, both 354
IRugt74p5 lines displayed a milder phenotype compared to IRugt74p3, but still with 355
severe morphological alterations ranging from necrotic spots and necrotic apical 356
meristem to small deformed or thicker succulent leaves (Supplemental Figure 17a-c) 357
and numerous stalled flower buds (Figure 5C). IRugt74p5 plants were also smaller 358
compared to WT (Supplemental Figure 18). Additionally, we established a heterozygous 359
double construct UGT74P3/UGT74P5 for which ~75% of the silenced plants exhibited 360
similar morphological alterations. Similar to the IRugt74p3 plants, the homozygous 361
double construct was lethal and did not produce seeds. 362
In contrast to the IRugt74p3 and IRugt74p5-silenced plants, two independent 363
stably-silenced UGT91T1 lines did not show any morphological alteration compared to 364
WT (Supplemental Figure 19). 365
In addition to the careful examination of shoot morphological alterations (roots 366
were not examined), we analyzed transcript abundance of UGT74P3, UGT74P5 and 367
UGT91T1 in leaves of 52-day-old plants of the different transgenic lines (Figure 6). The 368
silencing efficiency for UGT91T1 transcript abundance was 97.4% for IRugt91t1a and 369
97.6% for IRugt91t1b. UGT74P5 transcript abundance was strongly repressed in all 370
IRugt74p5 lines tested (83.9% in IRugt74p5b, 96.8% in IRugt74p5a and 92.3% in 371
IRugt74p3/ugt74p5). The silencing efficiency for UGT74P3 in IRugt74p3/ugt74p5 was 372
99.3%, but we also detected a strong co-silencing of UGT74P3 expression in 373
IRugt74p5b (96.7%) and IRugt74p5a (99.4%). 374
375
Metabolic profiling confirms the unique HGL-DTG profiles of stably-silenced UGT 376
lines 377
From UPLC/qTOF-MS measurements and the application of our de-replication 378
workflow, we identified 60 HGL-DTGs in the leaf extracts of the different stably-silenced 379
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted August 27, 2020. ; https://doi.org/10.1101/2020.08.26.267690doi: bioRxiv preprint
lines. HGL-DTGs were annotated and classified, based on their chemical composition, 380
into three categories as rhamnosylated, non-rhamnosylated and biosynthetic 381
intermediates only detectable upon disruption of glucosylation steps (Figure 5). Both 382
stable lines impaired in UGT74P5 expression showed the highest overall HGL-DTG 383
levels (Supplemental Figure 20). Non-rhamnosylated lyciumoside I and its malonylated 384
forms, nicotianoside IXa-d and Xa-e, were barely detectable and levels of 385
rhamnosylated lyciumoside IV was strongly reduced. While rhamnosylated nicotianoside 386
I and II remained constant compared to WT plants, both transgenic IRugt74p5 lines 387
exhibited elevated levels of lyciumoside II and attenoside and their malonylated forms, 388
nicotianoside XIa-d, XIIa-c, XIIIa-f and nicotianoside VIa-c, VIIa-c and VIIIa-b, 389
respectively. More complex reconfiguration patterns were also detected such as for 390
nicotianoside III which remained at unchanged levels in transgenic IRugt74p5 lines but 391
whose malonylated forms, nicotianoside IV and V, increased in concentrations. 392
Furthermore, we putatively identified 10 novel highly abundant intermediate HGL-DTGs 393
and the 17-HGL aglycone which were not detected in WT plants. Annotation of the 394
MS/MS spectra of these novel HGL-DTGs indicated lower molecular weight HGL-DTG 395
biosynthetic intermediates with either one (G-3-HGL, G-17-HGL) or two (RGHGL and 396
DTG 648) sugar moieties attached to the 17-HGL (Figure 5, Supplemental Figures 11a-397
c, Supplemental Table 7). Additionally, malonylated forms of these compounds were 398
also detected (DTG 572, DTG 718, DTG 734 and DTG 820) (Figure 5A and B, 399
Supplemental Figures 11a-c, Supplemental Table 7). The transgenic line IRugt74p5b 400
was used for all further experiments. The heterozygous double construct 401
IRugt74p3/ugt74p5 exhibited an almost identical pattern of accumulation of known and 402
novel intermediate HGL-DTGs. The most prominent difference compared to the 403
IRugt74p5 lines was the specific increase detected for lyciumoside II and lyciumoside IV. 404
In both stable lines impaired in UGT91T1 expression, all rhamnosylated HGL-405
DTGs were strongly decreased. Reciprocally, all non-rhamnosylated HGL-DTGs were 406
highly increased leading to a complete shift between rhamnosylated and non-407
rhamnosylated compounds within the HGL-DTG profile (Figure 5A and B). Neither 408
intermediate HGL-DTGs nor the 17-HGL aglycone could be detected in either of the 409
transgenic IRugt91t1 lines. IRugt91t1 line A was selected for all further experiments. 410
Additionally we analyzed the HGL-DTG profiles in IRggpps plants (Figure 5). 411
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted August 27, 2020. ; https://doi.org/10.1101/2020.08.26.267690doi: bioRxiv preprint
Geranylgeranyl-pyrophosphate synthase (GGPPS) is responsible for the synthesis of 412
GGPP, the precursor for all HGL-DTGs (Jassbi et al., 2008) and almost all HGL-DTGs 413
were decreased and no intermediates could be detected. Only levels of nicotianoside 414
VIIIa and VIIIb were increased. Surprisingly, traces of the 17-HGL aglycone were 415
detected. A more detailed analysis of the HGL-DTG profiles in all stable silenced lines 416
can be found in Supplemental Data 5. 417
In addition, we performed a detailed metabolite profiling of central carbon 418
metabolism intermediates and specialized metabolites, which is further described in 419
Supplemental Method File 2. This included quantitative profiles of 23 amino acids and 420
biogenic amines, small organic acids, phenylpropanoids and derivatives, sugars and 421
phytohormones, such as gibberellins, cytokinins and jasmonates (Supplemental Figures 422
21 and 22, Supplemental Data 6 and 7). 423
424
Abolishing 17-HGL aglycone synthesis in IRggpps prevents the strong 425
morphological alterations that result from silencing UGT74P3 and UGT74P5 426
To determine if the morphological alterations of plants transiently- and stably-427
silenced in UGT74P3 and UGT74P5, are mediated by altered HGL-DTG profiles, we 428
performed a VIGS experiment silencing UGT74P3 and UGT74P5 in N. attenuata plants 429
impaired in GGPPS expression (IRggpps) and WT plants. Quantification of 17-HGL 430
concentrations in leaf tissues using a quantitative U(H)PLC-triple quadrupole-MS 431
method (Supplemental Figure 23) showed accumulation of 17-HGL only in WT N. 432
attenuata plants transformed with the pTVUGT74P3 or pTVUGT74P5 VIGS construct 433
(Supplemental Figure 24). 17 days after inoculation, we analyzed the morphological 434
alterations triggered by UGT74P3 and UGT74P5 silencing (Figure 7). WT plants 435
transiently transformed with pTVUGT74P3 and pTVUGT74P5 exhibited the phenotype 436
described above, ranging from necrotic spots and necrotic apical meristems to stalled 437
and aborted flower buds (Figure 7A, Supplemental Figure 25, 26 and 27). The number 438
of side branches increased and the rosette diameter decreased. IRggpps plants 439
transformed with pTVUGT74P3 and pTVUGT74P5 did not show any severe 440
morphological alterations. Specifically, the number of side branches and rosette 441
diameters of these plants were not different from those of WT plants (Figure 7B). This 442
combination of genetic manipulations indicates that the ectopic accumulation of several 443
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted August 27, 2020. ; https://doi.org/10.1101/2020.08.26.267690doi: bioRxiv preprint
intermediates and of 17-HGL might be directly responsible for the observed 444
morphological alterations. 445
446
Excess of 17-HGL triggers cell necrosis in N. attenuata leaves 447
We determined the absolute concentration of 17-HGL in leaves of N. attenuata 448
plants (N=5) impaired in UGT74P5 or UGT74P3/UGT74P5 expression which exhibited 449
the severe phenotype reported above. The concentration (mean ± SD) of 17-HGL was 450
94 ± 8 nmol/g FW in IRugt74p5 and 85.4 ± 17.6 nmol/g FW in IRugt74p4/ugt74p5 451
(Figure 8A). For the determination of the phytotoxic effect of 17-HGL, we used 32-day-452
old early-elongated N. attenuata (N=3) WT plants and 48-day-old flowering N. attenuata 453
plants impaired in GGPPS expression (N=5). The average leaf mass was estimated 454
between 1.3 – 1.6 g for the rosette leaves of WT and the stem leaves of transgenic 455
IRggpps plants. Three leaves of each plant were treated with either 20 µL of DMSO or 456
DMSO with 140, 280, or 9800 nmol 17-HGL. Independent of the concentration of 17-457
HGL, DMSO application resulted in a mild dissolution of the epidermal surface. Analysis 458
of the damaged leaf area, 1 day after application revealed strong necrotic regions at all 459
three 17-HGL concentrations (Figure 8B). Significant increases in leaf damage were 460
detected starting at 140 nmol/g FW for IRggpps plants and 280 nmol/g FW in WT plants 461
(Figure 8C). 462
463
Manduca sexta larvae perform poorly on transformed lines impaired in HGL-DTG 464
glycosylation 465
HGL-DTGs are abundant and potent anti-herbivore defense metabolites in the 466
aboveground tissues of N. attenuata (Heiling et al., 2010). Although 17-HGL has been 467
suggested as a feeding deterrent (Jassbi et al., 2010), which of the many different HGL-468
DTGs or which structural component of HGL-DTGs accounts for the observed deterrent 469
(Jassbi et al., 2006) and resistance effects against herbivores (Snook et al., 1997) 470
remains unknown. To elucidate the defensive value of glucosylated and rhamnosylated 471
HGL-DTGs, we conducted performance assays with M. sexta larvae reared on leaf disks 472
of transgenic N. attenuata plants impaired in rhamnosylation (IRugt91t1) and 473
glucosylation (IRugt74p5, IRugt74p3/ugt74p5) of HGL-DTGs. Additionally, we included 474
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted August 27, 2020. ; https://doi.org/10.1101/2020.08.26.267690doi: bioRxiv preprint
IRggpps, which has been shown to dramatically decrease HGL-DTG levels and enhance 475
growth of M. sexta (Heiling et al. 2010). 476
Caterpillars feeding on IRggpps plants showed enhanced growth after 6 days 477
and gained 1.6-fold higher larval mass after 12 days (P=0.000) compared to larvae 478
feeding on WT leaf disks (Figure 9A). In contrast, M. sexta feeding on leaf disks from 479
IRugt74p5 (~0.6 fold, P=0.028) and IRugt74p3/ugt74p5 (~0.6-fold, P=0.001) plants 480
showed significantly reduced growth compared to larvae feeding on WT leaf disks. 481
Caterpillars fed leaf disks of IRugt91t1 displayed only slight reductions in growth (~0.8 482
fold relative to WT). 483
Additionally, we measured the mass of leaf tissue consumed by M. sexta larvae 484
fed IRggpps, IRugt74p5, IRugt74p3/ugt74p5 and IRugt91t1 leaf disks between days 8 to 485
10 and observed a strong decrease in consumption of IRugt74p5, IRugt74p3/ugt74p5 486
and IRugt91t1 leaf disks relative to WT (Figure 9B). A detailed statistical analysis can be 487
found in Supplemental Data 8. 488
489
490
Discussion 491
492
Plants have evolved highly diversified specialized metabolism pathways to resist 493
both abiotic and biotic stresses as well as a series of mechanisms to mitigate 494
cost/benefit trade-offs of specialized metabolite production and thereby maintain 495
competitive ability. Innovations in specialized metabolism frequently result from the 496
modification or direct recruitment of pre-existing scaffolds produced by core metabolic 497
pathways. These scaffolds serve as substrates for modifying enzymes that add many 498
different types of decorations (Gachon et al., 2005). In this respect, the large numbers of 499
UGTs that populate plant genomes define a versatile glycosylation toolbox that has likely 500
facilitated the functional diversification of secondary metabolism across plant lineages. 501
Here we identified and phylogenetically characterized 107 UDP-502
glycosyltransferases of the superfamily 1 in N. attenuata. We identified three novel 503
UGTs, based on co-expression analysis, which are responsible for the synthesis of 17-504
HGL-DTGs in N. attenuata. UGT74P3 and UGT74P4 are GT-type enzymes that use 505
UDP-glucose to attach the first glucose moieties to the C-3 and C-17 hydroxyl groups of 506
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted August 27, 2020. ; https://doi.org/10.1101/2020.08.26.267690doi: bioRxiv preprint
17-HGL. Additionally, we show that the glucosylation of the 17-HGL aglycone is crucial 507
to prevent an autotoxic effect of 17-HGL accumulation, which causes severe necrosis of 508
the leaves. Collectively, this functional study provides novel insights into the 509
biosynthesis of broad-spectrum anti-herbivore HGL-DTGs and provides evidence 510
consistent with the hypothesis that UGTs play a central role in a plant’s ability to manage 511
the “toxic waste dump” problem of chemical defense. 512
513
A geranyllinalool synthase (GLS) is required for HGL-DTG production 514
Geranyllinalool (GL) is an acyclic diterpene alcohol which is widely distributed 515
across the plant kingdom, occurring in several essential oils (Sandeep and Paarakh, 516
2009) and is a precursor for both HGL-DTGs in solanaceous species and for the volatile 517
C16-homoterpene 4,8,12-trimethyltrideca-1,2,7,11-tetraene (TMTT) which is emitted from 518
the foliage of a wide range of plant species including Solanum lycopersicum (Ament et 519
al., 2004), Zea mays (Hopke et al., 1994), Medicago truncatula (Leitner et al., 2010) and 520
A. thaliana (Van Poecke et al., 2001). Although geranyllinalool is present in many plant 521
species, the enzymes responsible for its biosynthesis have been discovered only 522
recently in A. thaliana (Herde et al., 2008), S. lycopersicum, and N. attenuata in which 523
GLS is constitutively expressed in leaf and flower tissues and is induced by methyl 524
jasmonate treatment (Falara et al., 2014). Based on the tissue-specific correlation of 525
GLS transcript abundance with the levels of HGL-DTGs in reproductive organs, the 526
authors predicted that NaGLS is involved in the biosynthesis of HGL-DTGs (Falara et 527
al., 2014). Here we provide additional evidence demonstrating its coordinated 528
expression with other HGL-DTG biosynthetic genes (GGPPS, UGT91T1 and UGT74P3) 529
and demonstrate that HGL-DTG levels are highly reduced after the transient silencing of 530
NaGLS. 531
532
Gene-to-gene co-expression analysis identifies three novel UGTs responsible for 533
the glycosylation of HGL-DTGs 534
To identify the UGTs responsible for the glycosylation of HGL-DTGs, we mined 535
the expression profiles of 76 UGTs in a full-transcriptome datasets created for leaf and 536
roots collected from plants after simulated herbivory. More specifically, we explored co-537
linearity patterns in the regulation of this set of UGTs and GGPPS and GLS which are 538
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted August 27, 2020. ; https://doi.org/10.1101/2020.08.26.267690doi: bioRxiv preprint
involved in upstream steps of the biosynthesis of 17-HGL. A previous study by our group 539
(Gulati et al., 2013) detected, using the same transcriptome datasets, a tight co-540
regulation between genes of the non-mevalonate pathway (NaDXS, NaDXR) and 541
NaGGPPS. Our analysis pinpointed three novel glycosyltransferases strongly correlated 542
with GGPPS and GLS, which were not identified in the previous study. Gulati and 543
colleagues analyzed tissue-specific gene-to-gene and gene-to-metabolite associations 544
using fold-change data in response to herbivory and used very stringent statistical 545
parameters. Here, we used a simpler co-expression approach relying on overall gene-to-546
gene correlations across all tissues to identify significant correlations in expression 547
dynamics among UGTs and previously characterized upstream genes. Furthermore, a 548
detailed analysis of the expression pattern of each of the 76 UGTs (Supplemental Figure 549
4) revealed that herbivory has dramatic effects on the local induction of UGT expression, 550
an observation consistent with the importance of glycosylation reactions in maintaining 551
cellular homeostasis after damage. 552
Together with the three candidate UGTs identified in N. attenuata, we also 553
identified and biochemically characterized orthologues in N. obtusifolia whose genome 554
was recently published along with that of N. attenuata (Xu et al., 2017). 555
556
UDP-Rha: 17-HGL-DTG rhamnosyltransferase activity of UGT91T1 557
Very little is known about the physiological function of rhamnosylation in 558
secondary metabolism. For example, Hsu and colleagues (Hsu et al., 2017) suggested 559
that rhamnosylation is an essential step in the synthesis of lobelinin and therefore 560
responsible for the color of Lobelia flowers. Rhamnosylation of flavonols is thought to 561
modulate auxin homeostasis in rol1-2 mutants of A. thaliana (Kuhn et al., 2016). 562
However, the enzymatic characterization of UDP-rhamnosyltransferases is still thwarted 563
by the high costs of UDP-rhamnose (UDP-Rha). Despite important efforts in the 564
establishment of efficient UDP-Rha production systems (Irmisch et al., 2018), only a few 565
UDP-rhamnosyltransferases have been functionally characterized and enzymatic assays 566
remain challenging (Mo et al., 2016; Irmisch et al., 2018). 567
Here we show that UGT91T1 shares a high degree of sequence similarity with 568
functionally characterized rhamnosyltransferases for flavonoids and anthocyanins 569
(CmF7G12RT, GmF3G6R and PhA3ART – Supplemental Figure 6), is tightly and tissue-570
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted August 27, 2020. ; https://doi.org/10.1101/2020.08.26.267690doi: bioRxiv preprint
specifically co-regulated with the accumulation patterns of rhamnosylated HGL-DTGs 571
and exhibits strong co-expression with GLS, GGPPS and UGT74P3. Silencing of 572
UGT91T1 expression in N. attenuata results in the almost complete loss (88.5% - 98.8% 573
in different tissues compared to WT, Supplemental Table 8) of rhamnosylated HGL-574
DTGs. These changes in HGL-DTG rhamnosylation pattern suggest that UGT91T1 is 575
responsible for the rhamnosylation at the C’-4 hydroxyl-group of glucose on both the C-3 576
and the C-17 hydroxyl-group of the aglycone. Due to the high biological variability 577
among transiently-silenced plants and the complexity of the HGL-DTG profile in N. 578
obtusifolia, which produces lower levels of rhamnosylated HGL-DTGs, the biochemical 579
function of NoUGT91T1-like, the orthologue of UGT91T1 in N. obtusifolia, could not be 580
conclusively evaluated. 581
582
UGT74P3 and UGT74P4 function as UDP-Glu: 17-HGL-DTG glucosyltransferases 583
Both UDP-glycosyltransferases clustered together in the UGT74 family, which 584
belongs to phylogenetic group L and includes glycosyltransferases that are responsible 585
for the glucosylation of indole-3-acetic acid in Zea mays (Szerszen et al., 1994) and A. 586
thaliana (Jin et al., 2013). Importantly, additional members of this family show catalytic 587
activity towards diterpene and triterpene glycosides. For example, UGT74M1 588
glucosylates the carboxylic acid moiety of the triterpene gypsogenic acid 589
(Meesapyodsuk et al., 2007), UGT74-345-2 is involved in the glucosylation of mogroside 590
(Itkin et al., 2016, 2018), and UGT74G1 catalyzes the glucosylation of cyclic diterpene 591
glycosides in Stevia rebaudiana (Richman et al., 2005). Both UGT74P3 and UGT74P4 592
are closely related to UGT74G1 and CsGLT2, a UGT responsible for crocetin 593
glucosylation in Crocus sativus (Moraga et al., 2004). The expression profile of 594
UGT74P3 in N. attenuata revealed a strong tissue-specific correlation with HGL-DTG 595
accumulation and GLS expression. 596
Recombinant enzyme activity assays demonstrated that UGT74P3 and UGT74P4 597
encode enzymes that catalyze the transfer of glucose to the C-3 and C-17 hydroxyl 598
groups of 17-HGL in order to form the putative intermediate products 3-O-599
glucopyranosyl-17-HGL (G-3-HGL) and 17-O-glucopyranosyl-17-HGL (G-17-GHL) 600
(Figure 4), as well as lyciumoside I in which both hydroxyl groups are glucosylated. 601
Neither UGT74P3 nor UGT74P4 showed activity producing lyciumoside II, which 602
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted August 27, 2020. ; https://doi.org/10.1101/2020.08.26.267690doi: bioRxiv preprint
includes an additional glucose moiety linked to the C’-2 hydroxyl group of the glucose at 603
the C-17 hydroxyl group of 17-HGL, or other larger HGL-DTGs with additional sugar 604
moieties. However, the functional characterization of enzymes in the UGT family can be 605
very challenging and in vitro enzyme assays do not necessarily reflect native metabolic 606
activities in planta. 607
Metabolic profiling of transiently- and stably-silenced lines provided further 608
evidence that UGT74P3 and UGT74P4 are essential for the formation of lyciumoside I 609
through the attachment of two glucose moieties. Silencing of UGT74P3 and UGT74P4 610
expression significantly reduces the amount of lyciumoside I and its malonylated forms 611
in leaf tissue material. Interestingly, compounds further downstream in the HGL-DTG 612
biosynthetic pathway (lyciumoside II, lyciumoside IV, attenoside) are more affected in 613
the transiently than in the stably–silenced lines, which might be due to the infection with 614
the virus vector and a possible involvement of HGL-DTGs in the immune response of 615
plants against viruses (Ramegowda et al., 2014). In the stable lines impaired in 616
UGT74P3 expression, we observed a slight reduction of lyciumoside IV, but an increase 617
of attenoside and lyciumoside II, indicating that the downstream model of the HGL-DTG 618
biosynthetic pathway is incomplete and that other UGTs are involved in the glycosylation 619
of larger molecular weight HGL-DTGs. 620
In addition to the changes in known HGL-DTG patterns, we also observed an 621
accumulation of novel intermediate HGL-DTGs that lacked either glucosylation at the C-622
3 or C-17 hydroxyl group of the aglycone as well as the 17-HGL aglycone itself in both 623
stably- and transiently-silenced lines. This suggests that N. attenuata is unable to 624
reroute the excess of 17-HGL and provides us with the opportunity to study the effects of 625
glucosylation of HGL-DTGs in vivo. 626
In contrast to UGT74P3 and UGT74P4, we were not able to decipher the function 627
of UGt74P5 in N. attenuata and of its orthologue in N. obtusifolia. The high sequence 628
similarity between UGT74P3 and UGT747P5 of ~83%, indicated a related function as a 629
UGT. Furthermore, UGT74P5 expression highly correlates with that of all other genes 630
related to HGL-DTG biosynthesis (UGT74P3, UGT91T1, GLS and GGPPS), exhibits a 631
similar temporal dynamic after simulated insect herbivory and shows very low transcript 632
abundance in roots. UGT74P5 showed no activity towards the 17-HGL aglycone, 633
lyciumoside I or the putative intermediates G-3-HGL and G-17-HGL, indicating that 634
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted August 27, 2020. ; https://doi.org/10.1101/2020.08.26.267690doi: bioRxiv preprint
UGT74P5 might not directly contribute to the biosynthesis of HGL-DTGs or be 635
responsible for the glucosylation of HGL-DTGs that were not tested in these enzyme 636
assays. 637
The co-silencing of UGT74P5 and UGT74P3 expression in the stable and 638
transiently silenced lines (Figure 6, Supplemental Figure 8), likely due to the high 639
sequence similarity of both UGTs, did not allow us to further characterize the specific 640
function of UGT74P5. The metabolic alterations of these lines likely reflect the silencing 641
of both genes. Based on the enzymatic assays, we inferred that the reduction of 642
lyciumoside I and the accumulation of G-3-HGL, G-17-HGL and 17-HGL directly result 643
from the silencing of UGT74P3, leaving it unclear whether UGT74P5 is involved in the 644
synthesis of HGL-DTGs, is non-functional or has a still-unknown enzymatic function 645
beyond the diterpene metabolism. 646
647
Phytotoxicity observed in UGT74P3/UGT74P5 silenced lines 648
In addition to the striking shifts in HGL-DTG metabolism detected in the stably- and 649
transiently-silenced lines impaired in UGT74P3 and UGT74P5 expression, we also 650
observed strong morphological alterations that ranged from small deformed or thicker 651
succulent leaves to numerous stalled flower buds as well as necrotic spots, stunted 652
growth and apical meristem necrosis. 653
The loss of glycosylation results in the ectopic accumulation of hydrophobic 654
aglycones, which are known to be responsible for a wide range of morphological 655
alterations and growth retardation effects. Some of the best examples, come from the 656
biosynthesis of steroidal alkaloids. Silencing GAME-1, a UDP-galactosyltransferase 657
responsible for the glycosylation of steroidal alkaloids, results in severe developmental 658
defects due to the altered sterol composition in membranes triggered by the 659
accumulation of tomatidine (Itkin et al., 2011). Moreover, alterations of the glycosylation 660
of the saponin, hederagenin, severely affected growth in M. truncatula (Naoumkina et 661
al., 2010). Loss of function mutants, sad3 and sad4, in Avena sativa accumulate the 662
intermediate monodeglucosylated diterpeneglycoside, avenacin A-1, which disrupts 663
membrane trafficking and results in epidermal degeneration and reduced root hair 664
formation (Mylona et al., 2008). Knocking out UGT74B1, a UGT responsible for 665
glucosinolate biosynthesis in A. thaliana, leads to the accumulation of toxic levels of 666
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted August 27, 2020. ; https://doi.org/10.1101/2020.08.26.267690doi: bioRxiv preprint
thiohydroximates, increased auxin levels in seedlings and results in a chlorotic 667
phenotype (Grubb et al., 2004). 668
The severe growth defects, stalled flower buds and necrosis we observed in 669
UGT74P3/UGT74P5-silenced plants may have several explanations. First blocking 670
HGL-DTG glycosylation might increase the endogenous pool of UDP-glucose, resulting 671
in a preferential synthesis of other glycosides or the accumulation of other compound 672
classes (Naoumkina et al., 2010). This might influence hormone homeostasis and 673
therefore contribute to the severe alterations (Grubb et al., 2004; Kuhn et al., 2016). 674
Auxin levels, which have been suggested to mediate some of these observed 675
developmental alterations, are only changed in GGPPS-silenced plants but not in the 676
UGT-silenced lines showing the morphological defects. Alterations of the UDP-glucose 677
pool can also influence UDP-glucuronic acid (UDP-GlcA) biosynthesis. Reduction of 678
UDP-GlcA in ugd2 and ugd3 mutants of A. thaliana leads to swollen cell walls and 679
developmental defects associated with changes in the pectic network (Reboul et al., 680
2011). IRugt74p5 and IRugt74p3/ugt74p5 plants exhibit lower glucuronic acid levels 681
(Supplemental Data 6), suggesting that UDP-GlcA production is downregulated, which 682
could mediate the observed smaller leaves and the dwarfish growth phenotype. 683
However, we observed a similar reduction in glucuronic acid levels in GGPPS-silenced 684
plants, which do not show the altered growth effects. Alternatively, UGT74P3/UGT74P5-685
silenced plants also exhibited a dramatic increase in levels of intermediate HGL-DTGs 686
and the aglycone 17-HGL compared to other transgenic lines that did not show any 687
morphological phenotypes (Figure 8A). From these results we infer that the 688
accumulations of these intermediates may be toxic for the plant. 689
When UGT74P3 and UGT74P5 were silenced in the background of stably-690
silenced IRggpps lines, which do not produce the 17-HGL precursor, no developmental 691
abnormalities were detected, indicating that the developmental abnormalities are 692
associated with the altered HGL-DTG metabolism and not due to unexpected off-target 693
effects (Senthil-Kumar and Mysore, 2011). Applications of a synthetic mixture of 17-HGL 694
isomers to leaves of N. attenuata resulted in necrotic lesions that pheno-copied those 695
observed in IRugt74p5 and IRugt74p3/ugt74p5 plants. IRggpps plants, which are 696
strongly down-regulated in the expression of HGL-DTG biosynthetic genes and HGL-697
DTG accumulation, were more susceptible to the toxic effects of exogenous applications 698
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted August 27, 2020. ; https://doi.org/10.1101/2020.08.26.267690doi: bioRxiv preprint
of 17-HGL than WT plants and showed necrotic lesions in response to 17-HGL 699
concentrations lower than those observed in IRugt74p3/ugt74p5 plants. The lower toxic 700
effects observed in WT leaves may be due to the chemical sequestration of 17-HGL and 701
intermediate HGL-DTGs by active glycosylation with the HGL-DTG pathway. 702
Interestingly, N. sylvestris accumulates 17-HGL without producing the glycosylated 703
forms (Heiling et al., 2016), suggesting that other species might be tolerant to or store 704
17-HGL in special compartments. In short, silencing UGT74P3 and UGT74P5 results in 705
pleiotropic morphological affects that are associated with the accumulation of the 706
phytotoxic aglycone 17-HGL. 707
708
Defensive function of HGL-DTGs 709
HGL-DTGs have been studied for more than two decades and are described to be 710
potent anti-herbivore defense compounds (Heiling et al., 2010) with deterrent (Jassbi et 711
al., 2006) and resistance (Snook et al., 1997) effects against herbivores. However, so far 712
it remains unclear which of the many HGL-DTGs, their structural components or post-713
ingestive modifications account for their mode of action. The impressive structural 714
diversity of HGL-DTGs in N. attenuata plants results largely from glycosylation and 715
malonylation reactions. While all malonyl decorations are instantaneously lost in the 716
alkaline pH environment of the midgut of M. sexta larvae and therefor may play more 717
important roles in planta (Poreddy et al., 2015; Li et al., 2018), glycosylation leads to a 718
constant and stable pool of potential defensive compounds. Larvae feeding on leaf disks 719
of transgenic IRggpps plants grow larger than larvae feeding on WT leaves, indicating 720
that the overall abundance of HGL-DTGs is a major factor of the plant’s resistance 721
against M. sexta (Figure 9), consistent with earlier studies (Jassbi et al., 2008; Heiling et 722
al., 2010; Jassbi et al., 2010). The reduced growth of larvae feeding on leaf disks of 723
transgenic IRugt74p5 and IRugt74p3/ugt74p5 plants, which have higher total levels of 724
HGL-DTGs, is also consistent with the inference that the total abundance of HGL-DTGs 725
is defensively relevant. For example, N. obtusifolia is a perennial plant, which co-occurs 726
with the annual N. attenuata in the Great Basin Desert and is much less attacked by the 727
tobacco hornworm M. sexta. This species lacks N. attenuata’s strong OS-elicited 728
signaling system and hence produces attenuated jasmonate bursts and trypsin 729
proteinase inhibitors (TPI) accumulations (Pearse et al., 2006; Anssour and Baldwin, 730
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted August 27, 2020. ; https://doi.org/10.1101/2020.08.26.267690doi: bioRxiv preprint
2010; Xu et al., 2015). Furthermore, N. obtusifolia has constitutive levels of HGL-DTGs 731
that are five times higher than those of N. attenuata and M. sexta larval performance is 732
five times lower after 22 days feeding on N. obtusifolia (Jassbi et al., 2010). However, 733
the HGL-DTG profile in N. obtusifolia is structurally different (Heiling et al., 2016), 734
suggesting that the composition or degree of glycosylation is central for the plant’s 735
defense. 736
Importantly, the specific structures responsible for the mode of action of HGL-737
DTGs remain unclear. Feeding of purified lyciumoside IV, the most abundant HGL-DTG 738
in the leaves of N. attenuata, causes mortality in M. sexta larvae silenced in β-739
glucosidase 1 (BG1) expression (Poreddy et al., 2015). However, the single-compound 740
feeding experiment does not reflect the chemical diversity of HGL-DTGs that is normally 741
consumed by larvae when they feed on plants. In contrast to the results of Poreddy and 742
colleagues, we show that, while consuming less leaf tissue, M. sexta larvae grow 743
normally when feeding on leaf disks of transgenic IRugt91t1 plants, which have reduced 744
levels of lyciumoside IV, but increased levels of the non-rhamnosylated HGL-DTGs 745
lyciumoside I and II. Furthermore, we observed increased levels of novel intermediate 746
HGL-DTGs and the aglycone, 17-HGL, in IRugt74p5 and IRugt74p3/ugt74p5 plants, 747
from which we observed a strong antifeedant/deterrent effect, which showed a strong 748
antifeedant/deterrent effect (Figure 9B) and reduced growth performance of M. sexta 749
larvae (Figure 9A). The differential consumption and growth performance responses on 750
the transgenic plants with different HGL-DTG profiles suggest that both pre- and post-751
ingestive resistance mechanisms may be at play. Given the severe metabolic and 752
pleiotropic morphological phenotypes of IRugt74p5, IRugt74p3/ugt74p5 and IRggpps 753
plants, it is challenging to draw strong inferences about the defensive function of specific 754
intermediates in the HGL-DTG biosynthetic pathway. 755
We are still in the early stages of understanding how plants solve the “toxic waste 756
dump” problem. Two common solutions are well documented: 1) producing metabolites 757
(e.g., nicotine) that are specifically toxic to tissues or organs that plants lack (e.g., 758
nervous systems and neuro-muscular junctions populated with nicotinic acetylcholine 759
receptors); and 2) sequestering pro-toxins apart from their toxin-releasing lytic enzymes 760
(e.g., compartmentalization of cyanogenic glycosides and glucosinolates and their active 761
enzymes). However, these examples are likely to be special cases and the solutions that 762
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted August 27, 2020. ; https://doi.org/10.1101/2020.08.26.267690doi: bioRxiv preprint
plants have evolved to solve the “toxic waste dump” problem for a majority of their toxic 763
defensive metabolites lie in the details of their biosynthetic pathways. Here we advance 764
our understanding of the toxicity of HGL-DTGs by showing that glucosylation plays a 765
central role in N. attenuata’s solution of maintaining an HGL-DTG-based defense. 766
Extending the metabolomics analyses to the “digestive duet” that occurs between plant 767
and insect (“frassomics”) will allow for a better understanding of the post-ingestive fate of 768
DTGs in larval guts and will help to answer a central and festering question that remains 769
unanswered from this work: are the specific molecules that are responsible for the 770
defensive function of HGL-DTGs the same as those responsible for the clear 771
autotoxicity? 772
773
Methods 774
775
Plant material and growth conditions 776
777
Seed germination and growth condition have been described previously (Krugel et al., 778
2002). Seeds of the 31th generation of an inbred line of N. attenuata Torr. Ex. Watts 779
were used as wild type plants in all experiments. N. obtusifolia plants were cultivated 780
under the same growth conditions with the exception of not applying liquid smoke to the 781
seeds. Plants for virus-induced gene-silencing were transferred after 20 days to a York 782
Chamber with 22°C for a 16h light/ 8h dark cycle. 783
784
M. sexta growth conditions 785
786
M. sexta eggs were obtained from an in-house colony in which insects are reared 787
in a growth chamber (Snijders Scientific, Tilburg, Netherlands, 788
http://www.snijderslabs.com) at 26°C:16-h light and 24°C:8-h dark, 65% relative 789
humidity, until hatching. 790
791
Performance Assays 792
793
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted August 27, 2020. ; https://doi.org/10.1101/2020.08.26.267690doi: bioRxiv preprint
Freshly hatched neonates were fed leaf disk material, taken from the between-794
vein laminal tissue of the four lowest stem leaves of 43-day-old flowering transgenic N. 795
attenuata plants (WT, IRugt91t1, IRugt74p5, IRugt74p2/ugt74p5, IRggpps), in round 796
plastic PE-packing cups. Leaf disk material was exchanged every 2 days until day 6 and 797
then exchanged every day until day 12. 798
799
Plant transformation and screening of stably-silenced N. attenuata plants 800
801
Transformation of N. attenuata was performed as described in Krugel et al. 802
(2002) using the pRESC8 vector (Gase and Baldwin, 2012) containing the hygromycin 803
phosphotransferase II gene (hptII) from pCAMBIA-1301 (GenBank AF234297) and a 804
306 bp long fragment for UGT91T1, a 310 bp long fragment for UGT74P3 or a 295 bp 805
long fragment for UGT74P5. All primers are shown in supplemental table 9. Diploid 806
plants were selected by flow cytometry of leaf material of elongated N. attenuata 807
transformants performed on a CCA-II flow cytometer (Partec, http://www.partec.com) as 808
described by Bubner et al. (Bubner et al., 2006). Afterwards, seeds were collected and 809
individuals with the T-DNA insertion were selected for hygromycin resistance by addition 810
of 35 mg/L hygromycin B to the germination medium. After 10 days, the ratio of 811
seedlings surviving the antibiotic treatment was determined. Seedlings were chosen with 812
a survival rate of 50 – 90% and 12 T1 plants per line were checked for T-DNA insertion. 813
To confirm the integrity of the T-DNA insertion, we performed a diagnostic PCR using 814
the primer pairs PROM FOR/INT REV and INT FOR/TER REV (Gase et al., 2011). 815
Genomic DNA (gDNA) was isolated from leaves of N. attenuata using a modified 816
cetyltrimethylammonium bromide method (Bubner et al., 2004). PCR was performed 817
using DreamTaqTM DNA-Polymerase (Fermentas, http://www.fermentas.com) according 818
to the instructions of the manufacturer with 1 µg of gDNA. Homozygosity of T2 plants 819
was determined by screening for resistance to hygromycin B. To confirm single 820
insertions, we performed a Southern blot as described by Jassbi et al. (Jassbi et al., 821
2008), with the exception that a 287 bp hptII probe obtained by PCR with primer pair 822
(HYG1-18/HYG2-18) was used (Gase et al., 2011). Labelling was performed with the 823
GE Healthcare (http://www.gehealthcare.com) Readyprime DNA labelling system and 824
ProbeQuant g-50 microcolumn according to the manufacturer’s protocol. 10.5 µg of 825
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted August 27, 2020. ; https://doi.org/10.1101/2020.08.26.267690doi: bioRxiv preprint
gDNA was digested with the restriction enzymes EcoRV and XbaI from New England 826
Biolabs (http://www.neb.com) and blotted onto a nylon membrane (GeneScreenPlus; 827
Perkin Elmer, http://www.perkinelmer.com) according to the manufacturer’s instructions. 828
829
Virus-induced gene-silencing (VIGS) 830
831
Vector construction, plant growth, and inoculation conditions were as described 832
by Saedler and Baldwin (Saedler and Baldwin, 2004). Briefly, 200- to 300-bp fragments 833
of N. attenuata and N. obtusifolia target genes were amplified by PCR using primer pairs 834
as listed in supplemental table 9. Amplified fragments were cloned in the vector pTV00 835
(Ratcliff et al., 2001). Agrobacterium tumefaciens strain GV3101 was transformed by 836
electroporation with the resulting plasmids. We used the empty pTV00 vector as a 837
negative control in all experiments. Four leaves of 24-27 day old N. attenuata and 25 838
day old N. obtusifolia plants were infiltrated with a 1:1 mixture of A. tumefaciens 839
transformed with pBINTRA (Ratcliff et al., 2001) and one pTV00 derivative carrying a 840
fragment of a gene of interest. pTVPDS, targeting for a Phytoene desaturase, was used 841
as a positive control to monitor silencing progress. Due to the depletion of carotenoids, 842
silencing the phytoene desaturase gene causes bleaching of tobacco leaves. VIGS-843
silenced plants were treated 14 days after inoculation, when the bleaching phenotype 844
was fully established in the pTVPDS plants. 845
846
Plant treatment 847
848
In order to analyze the regulatory function of jasmonate signaling on HGL-DTG 849
biosynthesis, petioles of five elongated plants (38-days-old) were treated with either 20 850
µL lanolin paste containing 150 µg methyl jasmonate (Lan + MeJA) or with 20 µL pure 851
lanolin (Lan). Treated leaves were harvested from elicited and unelicited plants at 72 h 852
after treatment, flash-frozen in liquid nitrogen, and stored at -80°C until use. 853
854
Determination of 17-HGL phytotoxicity 855
856
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted August 27, 2020. ; https://doi.org/10.1101/2020.08.26.267690doi: bioRxiv preprint
For the determination of the phytotoxic effect of 17-HGL, we used 32-day-old 857
early-elongated N. attenuata WT plants (N=3) and 48-day-old flowering N. attenuata 858
plants impaired in GGPPS expression (N=5). Three leaves of each plant were 859
inoculated with either 20 µL DMSO, DMSO with 140 nmol HGL, DMSO with 280 nmol 860
HGL or DMSO with 9800 nmol HGL. The damaged leaf tissue was analyzed after 24 h 861
using ImageJ (Fiji, https://fiji.sc). 862
863
RT-qPCR analysis of transcript levels 864
865
Total RNA was extracted from an aliquot of approximately 200 mg of powdered 866
leaf material of N. attenuata and N. obtusifolia ground in liquid nitrogen following the 867
protocol of Kistner and Matamoros (Kistner and Matamoros, 2005). DNase treatment 868
was performed using the TURBO DNA-free™ kit (Invitrogen). RNA quality was checked 869
on a 1% agarose gel and the concentration was determined spectrophotometrically at 870
260 nm. A total of 1 µg of DNA-free RNA was reverse transcribed using oligo(dT)18 871
primers and the SuperScript II enzyme (Invitrogen) following the manufacturer’s 872
recommendations. All RT-qPCR assays were performed using TaykonTM No ROX 873
SYBR® Master Mix dttp Blue (Eurogenetics, http://www.eurogentec.com) on a 874
Stratagene MX3005P instrument (http://www.stratagene.com) as recommended by the 875
manufacturer. To normalize transcript levels, primers specific for the Nicotiana attenuata 876
elongation factor-1α gene (EF1-α; accession no. GBGF01000210.1) were used. Specific 877
primers in the 5′ to 3′ direction used for SYBR Green-based analyses are listed in 878
supplemental table 9. 879
880
Heterologous expression of UDP-glucosyltransferases and enzymatic activity 881
assays 882
883
The four UGT cDNAs coding for NaUGT74P3, NoUGT74P4, NaUGT74P5 and 884
NoUGT74P6 were cloned into a Gateway® pDEST17 expression vector 885
(ThermoFisherScientific, http://www.thermofisher.com) using pET28a empty vector as 886
control. Integrity of the sequence was checked by Sanger-sequencing using an ABI 887
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted August 27, 2020. ; https://doi.org/10.1101/2020.08.26.267690doi: bioRxiv preprint
PRISM 3130 Genetic analyzer (AppliedBiosystems, http://www.thermofisher.com) and 888
the appropriate gene-specific primers (Supplemental Table 9). 889
The resulting plasmid was transformed into BL21 (DE3) E. coli which is optimized 890
for the expression of eukaryotic genes. 50 mL LB medium containing 50 µg/mL 891
carbenicillin were inoculated with 500 µL of an overnight culture corresponding to each 892
candidate gene. Cultures were grown at 37°C until the OD600 reached 0.6. Protein 893
expression was induced by adding 1 mM IPTG and incubation at 18 °C overnight. The 894
cells were harvested by centrifugation for 10 min at 4,500 × g. The pellet was 895
suspended in 10 mL ice cold lysis buffer (50 mM Tris HCl pH 7.5, 1% Triton 100, 200 896
mM NaCl, 1 mg/ml lysozyme, and 1 tablet of Protease inhibitor Cocktail (Roche)) and 897
sonicated 6 times for 10 s with 10 s pauses at 200-300 W, followed by centrifugation at 898
10,000 × g for 60 min. The supernatant was purified using Ni-NTA agarose (QIAGEN) in 899
accordance with the manufacturer’s instructions. The purified protein was desalted using 900
the Amicon Ultra-15 Centrifugal Filter Unit (Merck) and the desalted protein was used for 901
activity assays. Reactions were performed in 100 μL reaction volumes containing 100 902
mM Tris HCl pH 7.5, 2.5 µg protein, 5 mM UDP-α-D-glucose (Calbiochem, 903
http://www.merckmillipore.com) and 250 µg/mL 17-hydroxygeranyllinalool, for 3 h at 904
30°C. The reaction was stopped by adding 400 µL methanol, and the mixture was used 905
for HGL-DTG analysis using a high resolution time-of-flight mass spectrometer. 906
907
Quantification of primary metabolites and phytohormones in plant tissues 908
909
For the quantitative analysis of primary metabolites and phytohormones, we used 910
leaf material of 42-day-old elongated N. attenuata transgenic lines silenced in GGPPS, 911
UGT91T1, UGT74P5 and UGT74P3/UGT74P5 expression as well as WT plants. 912
Sample preparation and analysis of primary metabolites and phytohormones were 913
performed based on Schaefer et al. (Schafer et al., 2016). Peak integration was 914
performed using the operating Software MS Workstation (Bruker Daltonics). For the 915
quantitative analysis of GPP, FPP and GGPP, we followed the protocol developed by 916
Nagel et al. (Nagel et al., 2014). 917
918
Quantification of 17-hydroxygeranyllinalool in plant tissues 919
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted August 27, 2020. ; https://doi.org/10.1101/2020.08.26.267690doi: bioRxiv preprint
920
Approximately 50 mg of leaf material was flash frozen and ground in liquid 921
nitrogen and then aliquoted. Each aliquot was extracted with 500 µL 80% methanol 922
containing 0.2 ng/µL testosterone, and shaken twice at 1000 strokes for 45 sec using a 923
GenoGrinder 2000 (SPEX SamplePrep, http://www.spexsampleprep.com/). 924
Homogenized samples were then centrifuged at 16 000 × g for 20 min at 4°C. The 925
supernatant was centrifuged again at 16 000 × g for 20 min at 4°C and then diluted 1:10 926
with 80% methanol. We established a chromatographic method using a mixture of 927
solvent A: water (Milli-Q, Merck, http://www.emdmillipore.com) with 0.1% acetonitrile and 928
0.05% formic acid and solvent B: methanol. U(H)PLC for the quantification of 17-HGL 929
was performed using a Zorbax Eclipse XDB-C18 column (particle size 1.8 µm, column 930
length 3.0 × 50 mm) from Agilent Technologies (http://www.agilent.com). The 931
chromatographic separation was achieved using a U(HPLC) Advance (Bruker Daltonics) 932
with the following gradient: 0-0.5 min at 10% of B, 0.5-1 min up to 90% of B, 1-4 min up 933
to 100% of B, 4-5 min at 100% of B, 5-5.05 min down to 10% of B and from 5.05-6 min 934
at 10% of B. The injection volume was 5 µL and the flow rate 0.5mL min-1. 935
MS detection was performed on an EvoQ Elite QqQ-MS equipped with a HESI 936
(heated electrospray ionization) ion source (Bruker Daltonics) and the following HESI 937
conditions: spray voltage 4500 V, cone temperature 350°C, cone gas flow 35a (arbitrary 938
units), heated probe temperature 500°C, probe gas flow 60a and nebulizer gas 60a. 939
Compounds were detected in multiple reaction monitoring (MRM) mode using specific 940
precursor ion/product ion after positive ionization: the [M-H2O+H]+ ion was used as 941
precursor for 17-HGL, 289/81 (quantifier), 289/107 (qualifier), the [M+H]+ ion was used 942
as precursor for testosterone, 289/97 (quantifier), 289/109 (qualifier). Further details are 943
given in supplemental figure 23. The areas were analyzed using the MS Workstation 944
operating software from Bruker Daltonics (http://www.bruker.com). 945
946
Structural determination by nuclear magnetic resonance spectroscopy (NMR) 947
948
17-Hydroxygeranyllinalool was provided by HPC24 Standards (www.hpc-949
standards.com) and the general structure was verified by 1D and 2D NMR spectroscopy 950
(for 1H NMR, see Supplemental Figure 12). A Bruker AVANCE 400 NMR spectrometer 951
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted August 27, 2020. ; https://doi.org/10.1101/2020.08.26.267690doi: bioRxiv preprint
(Bruker, Rheinstetten, Germany), equipped with a 5 mm BBFO probe, was used to 952
record 1H NMR, DEPT 135, 1H-1H COSY, HSQC and HMBC spectra in MeOH-d4 at 300 953
K. Spectra were processed using TOPSPIN 3.1. (BrukerBiospin). 954
955
Rapid screening of HGL-DTGs via ultrahigh-pressure liquid chromatography/time 956
of flight mass spectrometry 957
958
All materials were ground in liquid nitrogen and split into aliquots of 10-100 mg 959
fresh weight (FW), dependent on the tissue. Each aliquot was extracted in 100 µL - 1 mL 960
extraction solution (80% methanol; ratio 1/10 FW/extraction solution) containing two 961
steel balls by shaking twice at 1200 strokes/min for 60 sec using a Geno/Grinder 2000. 962
Homogenized samples were then centrifuged at 16 000 × g for 20 min at 4°C. The 963
supernatant was centrifuged again at 16 000 × g for 20 min at 4°C. Two independent 964
chromatographic methods were used to resolve HGL-DTGs. Both methods used a 965
mixture of solvent A: water with 0.1% acetonitrile and 0.05% formic acid and solvent B: 966
acetonitrile and 0.05% formic acid. U(H)PLC for method A was performed using a 967
Dionex UltiMate 3000 rapid separation LC system (Thermo Fisher, 968
http://www.thermofisher.com), combined with a Thermo Acclaim RSLC 120 C18 column 969
(particle size 2.2 µm, average pore diameter 120Å, column dimension 2.1 × 150 mm). 970
Gradient elution steps were as follows: 0-0.5 min at 10% of B, 0.5–6.5 min up to 80% of 971
B and 6.5–8 min at 80% of B followed by returning to the starting conditions and column 972
equilibration. For method B sample gradient steps were as follows: 0–3 min at 10% B, 973
3–12 min up to 20% B, 12–17 min up to 35% B, 17–23 min up to 40% B, 23–25 min up 974
to 45% B, 25–30 min up to 50% B, 30–40 min up to 90% B and 40–45 min at 90% B, 975
followed by returning to the starting conditions and column equilibration. The injection 976
volume was 2 µL and the flow rate 0.4 mL/min for method A and B. 977
MS detection was performed using a micrOTOF-Q II, an Impact II and a maXis 978
UHR-Q-TOF-MS system (Bruker Daltonics) equipped with an electrospray ionization 979
(ESI) source operating in positive ion mode. ESI conditions for the micrOTOF-Q II 980
system were: end plate offset 500 V, capillary voltage 4500 V, capillary exit 130 V, dry 981
temperature 180°C and a dry gas flow of 10 L min-1. ESI conditions for the Impact II 982
UHR-Q-TOF-MS system were capillary voltage 4500 V, end plate offset 500 V, nebulizer 983
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted August 27, 2020. ; https://doi.org/10.1101/2020.08.26.267690doi: bioRxiv preprint
2 bar, dry temperature 200°C and a dry gas flow of 8 L min-1. ESI conditions for the 984
maXis UHR-Q-TOF-MS system were capillary voltage 4500 V, end plate offset 500 V, 985
nebulizer 1.8 bar, dry temperature 200°C and a dry gas flow of 8 L min-1. MS data were 986
collected over a range of m/z from 100 to 1600. Mass calibration was performed using 987
sodium formate (50 mL isopropanol, 200 µL formic acid, 1 mL 1 M NaOH in water). Data 988
files were calibrated using the Bruker high-precision calibration algorithm. Lock mass 989
calibration was performed for the profiling of the stable lines using signal m/z 622.0289 990
(molecular formula C12H19F12N3O6P3) from the ESI Tuning Mix (Agilent Technologies, 991
http://www.agilent.com). MS/MS experiments were performed using AutoMS/MS runs at 992
various CID voltages from 12.5 to 22.5 eV for ammonium adducts. Instrument control, 993
data acquisition and reprocessing were performed using HyStar 3.1 (Bruker Daltonics). 994
Molecular formulae were determined using SmartFormula 3D. SmartFormula calculates 995
the elemental compositions from accurate mass as well as the isotopic pattern 996
information using MS (SmartFormula) and MS + MS/MS information (SmartFormula 3D) 997
(Krebs and Yates, 2008; Kind and Fiehn, 2010). The mass tolerance was set to 4 mDa, 998
and the filter H/C element ratio was set between 1 and 3. Isotope peaks were assigned 999
using the Simulate Pattern Tool of the DataAnalysis software version 4.2 (Bruker 1000
Daltonics). We used QuantAnalysis (Bruker Daltonics) to integrate the peak areas. 1001
1002
Dereplication of HGL-DTGs 1003
1004
The dereplication workflow relies on a comprehensive MS and MS/MS database 1005
constructed by our group for HGL-DTGs of several solanaceous species (Heiling et al., 1006
2016) and a detailed rule-set for the annotation of fragmentation patterns of the different 1007
moieties decorating the 17-hydroxygeranyllinalool (17-HGL) aglycone. The MS and 1008
MS/MS database is based on the retention time and mass spectrometric data of purified 1009
HGL-DTGs which are used as authentic standards. Novel HGL-DTGs are annotated 1010
based on their spectral similarity to the MS and MS/MS in-house database. For the 1011
visualization and identification of HGL-DTG profiles, we computed the extracted ion 1012
chromatogram EIC m/z 271.2420. This m/z fragment corresponds to the 17-HGL 1013
aglycone lacking both hydroxyl-groups and is produced by in-source fragmentation 1014
during ionization for all HGL-DTGs independently of the type and degree of metabolic 1015
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted August 27, 2020. ; https://doi.org/10.1101/2020.08.26.267690doi: bioRxiv preprint
decorations. The trace returned for m/z 271.2420 allows for the visualization of the 1016
complete HGL-DTG chemotype and to rapidly assess variations within this chemotype 1017
that result from the single gene manipulations. Following application of the de-replication 1018
workflow, we subdivided, based on the presence of diagnostic m/z signals, the 1019
chemotype between rhamnosylated and non-rhamnosylated HGL-DTGs. Lyciumoside I 1020
and lyciumoside II and their malonylated derivatives corresponded to non-1021
rhamnosylated HGL-DTGS, while lyciumoside IV, attenoside and nicotianoside III as 1022
well as their malonylated forms, represented the major rhamnosylated compounds. We 1023
provide a detailed description of the identification of known and novel HGL-DTGs in 1024
Supplemental Tables 1a/b. The identification levels are based on community standards 1025
reported in (Sumner et al., 2007). Raw MS metabolomics data have been deposited in 1026
the open metabolomics database Metabolights, www.ebi.ac.uk/metabolights (accession 1027
no. MTBLS1819). 1028
1029
1030
Statistical analysis 1031
1032
Data were analyzed using Excel (Microsoft, http://www.microsoft.com), SPSS 1033
20.0 (SPSS Inc, http://www-01.ibm.com/software/analytics/spss/) and RStudio (RStudio 1034
Inc, https://www.r-project.org) using the package xlsx. Unless otherwise stated, 1035
parametric data were compared using ANOVA followed by Fisher LSD/Holm-Bonferroni 1036
post hoc tests or Mann-Whitney-Wilcox Pairwise Test (for heteroskedastic data). The 1037
phylogenetic tree was constructed using the maximum-likelihood method in MEGA5.0 1038
(http://www.megasoftware.net/). 1039
1040
ACCESSION NUMBERS 1041
1042
For the analysis of the HGL-DTG pathway, we used the following GenBank 1043
accessions for N. attenuata: NaGLS, KJ755868; NaGGPPS, EF382626; NaUGT74P3, 1044
KX752207; NaUGT91T1, KX752209; NaUGT74P5, KX752208), N. obtusifolia 1045
(NoUGT74P4, KX752210; NoUGT74P6, KX752211; NoUGT91T1-like, MG051326). The 1046
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted August 27, 2020. ; https://doi.org/10.1101/2020.08.26.267690doi: bioRxiv preprint
MS metabolomics data set has been deposited in the open metabolomics database 1047
Metabolights, www.ebi.ac.uk/metabolights (accession no. MTBLS1819). 1048
1049
1050
Supplemental Data: 1051
1052
Supplemental Method File 1: Phylogenetic characterization of the UDP-1053
glycosyltransferases in N. attenuata and N. obtusifolia 1054
Supplemental Method File 2: Altered general and specialized metabolite levels 1055
1056
Supplemental Figure 1: Alignment of the UGT C-terminal consensus sequence of 112 1057
family 1 glycosyltransferases from N. attenuata and N. obtusifolia. 1058
Supplemental Figure 2: Phylogenetic analysis of the N. attenuata UGT superfamily 1059
shows 16 major groups 1060
Supplemental Figure 3: Amino acid composition of all identified UGTs of the 1061
superfamily 1 in N. attenuata 1062
Supplemental Figure 4: Phylogenetic relationships and herbivory-induced tissue-1063
specific expression of 110 predicted UDP-glycosyltransferases (UGT) 1064
Supplemental Figure 5: Transcriptomic variation of UGTs after treatment with OS in N. 1065
attenuata 1066
Supplemental Figure 6: Phylogenetic tree analysis for the UDP-glycosyltransferases 1067
used for stable and transient silencing. 1068
Supplemental Figure 7: Silencing efficiency for the three transiently-silenced 17-HGL-1069
DTG biosynthetic UGTs in pTVUGT91T1, pTVUGT74P3 and pTVUGT74P5. 1070
Supplemental Figure 8: Co-silencing efficiency of UGT74P3 and UGT74P5 in 1071
pTVUGT74P3 and pTVUGT74P5. 1072
Supplemental Figure 9: Silencing efficiency for the three transiently-silenced 17-HGL-1073
DTG biosynthetic UGTs in pTVUGT91T1-like, pTVUGT74P4 and pTVUGT74P6 in N. 1074
obtusifolia. 1075
Supplemental Figure 10a/b: Mass spectrometric characterization and annotation of 1076
novel HGL-DTGs in transiently-silenced N. obtusifolia plants impaired in NoUGT74P4 1077
and NoUGT74P6 expression. 1078
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted August 27, 2020. ; https://doi.org/10.1101/2020.08.26.267690doi: bioRxiv preprint
Supplemental Figure 11a/b/c: Characterization and annotation of novel HGL-DTGs via 1079
MS/MS in stably-silenced N. attenuata plants impaired in UGT74P3 and UGT74P5 1080
expression. 1081
Supplemental Figure 12: 1H NMR spectrum of synthetic 17-hydroxygeranyllinalool (17-1082
HGL, HPC24 Standards) 1083
Supplemental Figure 13a/b/c: Morphological characterization of N. attenuata plants 1084
transiently-silenced in UGT91T1, UGT74P3 and UGT74P5 expression 1085
Supplemental Figure 14: Morphological characterization of N. obtusifolia plants 1086
transiently-silenced in NoUGT91T1-like, NoUGT74P4, NoUGT74P6 and 1087
NoUGT74P4/UGT74P6 expression 1088
Supplemental Figure 15: Metabolite profiling and morphological characterization of N. 1089
attenuata plants transiently-silenced via virus-induced gene silencing (VIGS) of 1090
geranyllinalool synthase (GLS) 1091
Supplemental Figure 16: Southern Blot 1092
Supplemental Figure 17a/b/c: Morphological characterization of the stable transformed 1093
IRugt74p5 Line A, Line B and IRugt74p3/ugt74p5. 1094
Supplemental Figure 18: Characterization of growth parameters in IRugt91t1, 1095
IRugt74p5, IRugt74p3/ugt74p5 and IRggpps 1096
Supplemental Figure 19: Morphological characterization of IRugt91t1 1097
Supplemental Figure 20: Overall abundance of HGL-DTGs 1098
Supplemental Figure 21: Disrupting HGL-DTG glycosylation reorganizes general, 1099
specialized and hormonal metabolic pathways 1100
Supplemental Figure 22: Characterization of free prenyldiphosphates in IRggpps and 1101
WT 1102
Supplemental Figure 23: Quantitative 17-HGL method 1103
Supplemental Figure 24: 17-HGL concentration in IRggpps and WT plants transiently 1104
transformed with pTV00, pTVUGT74P3 and pTVUGT74P5 1105
Supplemental Figure 25: Characterization of the morphological phenotype of WT and 1106
IRggpps plants transiently transformed with pTV00. 1107
Supplemental Figure 26: Characterization of the morphological phenotype of WT and 1108
IRggpps plants transiently transformed with pTVUGT74P3. 1109
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted August 27, 2020. ; https://doi.org/10.1101/2020.08.26.267690doi: bioRxiv preprint
Supplemental Figure 27: Characterization of the morphological phenotype of WT and 1110
IRggpps plants transiently transformed with pTVUGt74P5. 1111
1112
Supplemental Table 1: Molecular weight of UGTs in N. attenuata 1113
Supplemental Table 2: Phylogenetic grouping of 107 UGTs in N. attenuata 1114
Supplemental Table 3: UGT Amino acid composition in N. attenuata 1115
Supplemental Table 4: SignalIP4.1 – signal peptide cleavage sites 1116
Supplemental Table 5a/b: Pearson Correlation of all UGTs to NaGLS and NaGGPPS 1117
Supplemental Table 6: MS/MS Measurements for HGL-DTGs in N. obtusifolia 1118
Supplemental Table 7: MS/MS Measurements for HGL-DTGs in N. attenuata 1119
Supplemental Table 8a-b: Tissue-specific HGL-DTG modulation 1120
Supplemental Table 9: Primers 1121
1122
Supplemental Data 1: Relative UGT expression vs. time in N. attenuata 1123
Supplemental Data 2: HGL-DTG profiles of transiently-silenced N. attenuata plants 1124
Supplemental Data 3: HGL-DTG profiles of transiently-silenced NaGLS N. attenuata 1125
plants 1126
Supplemental Data 4: HGL-DTG profiles of transiently-silenced N. obtusifolia plants 1127
Supplemental Data 5: HGL-DTG profiles of stably-silenced N. attenuata plants 1128
Supplemental Data 6: General and specialized metabolites in stably-silenced N. 1129
attenuata plants 1130
Supplemental Data 7: Phytohormone profiles in stably-silenced N. attenuata plants 1131
Supplemental Data 8: Statistical analysis of the performance assay and consumed leaf 1132
disk mass 1133
1134
ACKNOWLEDGEMENTS 1135
1136
We thank the gardening staff at the Max Planck Institute for Chemical Ecology. 1137
We thank Thomas Hahn, Nicolas Heinzel, Alexander Weinhold, Mario Kallenbach and 1138
Matthias Schoettner for the analytical and technical support as well as Dapeng Li and 1139
Felipe Yon for the fruitful discussions. We thank Raimund Nagel for his help with the 1140
terpene quantification. Additionally, we thank Eric McGale and Jyotasana Gulati for their 1141
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted August 27, 2020. ; https://doi.org/10.1101/2020.08.26.267690doi: bioRxiv preprint
statistical wisdom and Michael Court for assistance naming UGTs. We thank the Max 1142
Planck Society and the International Max Planck Research School on the Exploration of 1143
Ecological Interactions with Chemical and Molecular Techniques for financial support. 1144
We acknowledge the European Research Council advanced grant ClockworkGreen to 1145
I.T.B. (number 293926; I.T.B.) for funding. 1146
1147
References: 1148
1149
Akter, R., Uddin, S.J., Tiralongo, J., Grice, I.D., and Tiralongo, E. (2016). A new 1150
cytotoxic diterpenoid glycoside from the leaves of Blumea lacera and its effects 1151
on apoptosis and cell cycle. Nat Prod Res 30, 2688-2693. 1152
Ament, K., Kant, M.R., Sabelis, M.W., Haring, M.A., and Schuurink, R.C. (2004). 1153
Jasmonic acid is a key regulator of spider mite-induced volatile terpenoid and 1154
methyl salicylate emission in tomato. Plant Physiol 135, 2025-2037. 1155
Anssour, S., and Baldwin, I.T. (2010). Variation in antiherbivore defense responses in 1156
synthetic Nicotiana allopolyploids correlates with changes in uniparental patterns 1157
of gene expression. Plant Physiol 153, 1907-1918. 1158
Baker, R., Parton, A.H., and Howse, P.E. (1982). Identification of an acyclic diterpene 1159
alcohol in the defense secretion of soldiers of Reticulitermes-Lucifugus. 1160
Experientia 38, 297-298. 1161
Bubner, B., Gase, K., and Baldwin, I.T. (2004). Two-fold differences are the detection 1162
limit for determining transgene copy numbers in plants by real-time PCR. Bmc 1163
Biotechnol 4. 1164
Bubner, B., Gase, K., Berger, B., Link, D., and Baldwin, I.T. (2006). Occurrence of 1165
tetraploidy in Nicotiana attenuata plants after Agrobacterium-mediated 1166
transformation is genotype specific but independent of polysomaty of explant 1167
tissue. Plant Cell Rep 25, 668-675. 1168
Dewick, P.M. (2002). The biosynthesis of C-5-C-25 terpenoid compounds. Natural 1169
Product Reports 19, 181-222. 1170
Falara, V., Alba, J.M., Kant, M.R., Schuurink, R.C., and Pichersky, E. (2014a). 1171
Geranyllinalool synthases in solanaceae and other angiosperms constitute an 1172
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted August 27, 2020. ; https://doi.org/10.1101/2020.08.26.267690doi: bioRxiv preprint
ancient branch of diterpene synthases involved in the synthesis of defensive 1173
compounds. Plant Physiol 166, 428-441. 1174
Fordyce, J.A., and Agrawal, A.A. (2001). The role of plant trichomes and caterpillar 1175
group size on growth and defence of the pipevine swallowtail Battus philenor. J 1176
Anim Ecol 70, 997-1005. 1177
Gachon, C.M.M., Langlois-Meurinne, M., and Saindrenan, P. (2005). Plant secondary 1178
metabolism glycosyltransferases: the emerging functional analysis. Trends Plant 1179
Sci 10, 542-549. 1180
Gase, K., and Baldwin, I.T. (2012). Transformational tools for next-generation plant 1181
ecology: manipulation of gene expression for the functional analysis of genes. 1182
Plant Ecol Divers 5, 485-490. 1183
Gase, K., Weinhold, A., Bozorov, T., Schuck, S., and Baldwin, I.T. (2011). Efficient 1184
screening of transgenic plant lines for ecological research. Mol Ecol Resour 11, 1185
890-902. 1186
Gleadow, R.M., and Moller, B.L. (2014). Cyanogenic glycosides: synthesis, physiology, 1187
and phenotypic plasticity. Annual Review of Plant Biology, Vol 65 65, 155-185. 1188
Grubb, C.D., Zipp, B.J., Ludwig-Muller, J., Masuno, M.N., Molinski, T.F., and Abel, 1189
S. (2004). Arabidopsis glucosyltransferase UGT74B1 functions in glucosinolate 1190
biosynthesis and auxin homeostasis. Plant J 40, 893-908. 1191
Gulati, J., Kim, S.G., Baldwin, I.T., and Gaquerel, E. (2013). Deciphering herbivory-1192
induced gene-to-metabolite dynamics in Nicotiana attenuata tissues using a 1193
multifactorial approach. Plant Physiol 162, 1042-1059. 1194
Halitschke, R., Stenberg, J.A., Kessler, D., Kessler, A., and Baldwin, I.T. (2008). 1195
Shared signals - 'alarm calls' from plants increase apparency to herbivores and 1196
their enemies in nature. Ecol. Lett. 11, 24-34. 1197
Halkier, B.A., and Gershenzon, J. (2006). Biology and biochemistry of glucosinolates. 1198
Annu Rev Plant Biol 57, 303-333. 1199
Hashimoto, K., Kawagishi, H., Nakayama, T., and Shimizu, M. (1997). Effect of 1200
capsianoside, a diterpene glycoside, on tight-junctional permeability. Bba-1201
Biomembranes 1323, 281-290. 1202
Heiling, S., Khanal, S., Barsch, A., Zurek, G., Baldwin, I.T., and Gaquerel, E. (2016). 1203
Using the knowns to discover the unknowns: MS-based dereplication uncovers 1204
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted August 27, 2020. ; https://doi.org/10.1101/2020.08.26.267690doi: bioRxiv preprint
structural diversity in 17-hydroxygeranyllinalool diterpene glycoside production in 1205
the Solanaceae. Plant J 85, 561-577. 1206
Heiling, S., Schuman, M.C., Schoettner, M., Mukerjee, P., Berger, B., Schneider, B., 1207
Jassbi, A.R., and Baldwin, I.T. (2010). Jasmonate and ppHsystemin regulate 1208
key malonylation steps in the biosynthesis of 17-hydroxygeranyllinalool diterpene 1209
glycosides, an abundant and effective direct defense against herbivores in 1210
Nicotiana attenuata. Plant Cell 22, 273-292. 1211
Heller, W. and Forkman, G. (1994). In the Flavonoids: Advances in Research Since 1212
1986 (J.B. Harborne, ed. London: Chapman & Hall, pp. 499-535. 1213
Herde, M., Gartner, K., Kollner, T.G., Fode, B., Boland, W., Gershenzon, J., Gatz, 1214
C., and Tholl, D. (2008). Identification and regulation of TPS04/GES, an 1215
Arabidopsis geranyllinalool synthase catalyzing the first step in the formation of 1216
the insect-induced volatile C(16)-homoterpene TMTT. Plant Cell 20, 1152-1168. 1217
Hopke, J., Donath, J., Blechert, S., and Boland, W. (1994). Herbivore-induced 1218
volatiles - the emission of acyclic homoterpenes from leaves of Phaseolus-1219
Lunatus and Zea-Mays can be triggered by a beta-glucosidase and jasmonic 1220
acid. Febs Lett 352, 146-150. 1221
Hsu, Y.H., Tagami, T., Matsunaga, K., Okuyama, M., Suzuki, T., Noda, N., Suzuki, 1222
M., and Shimura, H. (2017). Functional characterization of UDP-rhamnose-1223
dependent rhamnosyltransferase involved in anthocyanin modification, a key 1224
enzyme determining blue coloration in Lobelia erinus. Plant J 89, 325-337. 1225
Iorizzi, M., Lanzotti, V., De Marino, S., Zollo, F., Blanco-Molia, M., Macho, A., and 1226
Munoz, E. (2001). New glycosides from Capsicum annuum L. var. acuminatum. 1227
Isolation, structure determination, and biological activity. J Agr Food Chem 49, 1228
2022-2029. 1229
Irmisch, S., Jo, S., Roach, C.R., Jancsik, S., Saint Yuen, M.M., Madilao, L.L., O'Neil-1230
Johnson, M., Williams, R., Withers, S.G. and Bohlmann, J. (2018). Discovery 1231
of UDP-Glycosyltransferases and BAHD-Acyltransferases involved in the 1232
biosynthesis of the antidiabetic plant metabolie Montbretin A. Plant Cell 30, 1864-1233
1886. 1234
Itkin, M., Rogachev, I., Alkan, N., Rosenberg, T., Malitsky, S., Masini, L., Meir, S., 1235
Iijima, Y., Aoki, K., de Vos, R., Prusky, D., Burdman, S., Beekwilder, J., and 1236
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted August 27, 2020. ; https://doi.org/10.1101/2020.08.26.267690doi: bioRxiv preprint
Aharoni, A. (2011). GLYCOALKALOID METABOLISM1 is required for steroidal 1237
alkaloid glycosylation and prevention of phytotoxicity in tomato. Plant Cell 23, 1238
4507-4525. 1239
Itkin, M., Davidovich-Rikanati, R., Cohen, S., Portnoy, V., Doron-Faigenboim, A., 1240
Oren, E., Freilich, S., Tzuri, G., Baranes, N., Shen, S., Petreikov, M., 1241
Sertchook, R., Ben-Dor, S., Gottlieb, H., Hernandez, A., Nelson, D.R., Paris, 1242
H.S., Tadmor, Y., Burger, Y., Lewinsohn, E., Katzir, N., and Schaffer, A. 1243
(2016). The biosynthetic pathway of the nonsugar, high-intensity sweetener 1244
mogroside V from Siraitia grosvenorii. P Natl Acad Sci USA 113, E7619-E7628. 1245
Itkin, M., Davidovich-Rikanati, R., Cohen, S., Portnoy, V., Doron-Faigenboim, A., 1246
Oren, E., Freilich, S., Tzuri, G., Baranes, N., Shen, S., Petreikov, M., 1247
Sertchook, R., Ben-Dor, S., Gottlieb, H., Hernandez, A., Nelson, D.R., Paris, 1248
H.S., Tadmor, Y., Burger, Y., Lewinsohn, E., Katzir, N., and Schaffer, A. 1249
(2018). The biosynthetic pathway of the nonsugar, high-intensity sweetener 1250
mogroside V from Siraitia grosvenorii. P Natl Acad Sci USA 115, E3862-E3862. 1251
Izumitani, Y., Yahara, S., and Nohara, T. (1990). Novel acyclic diterpene glycosides, 1252
capsianosides-A-F and capsianosides-I-V from Capsicum plants - Solanaceous 1253
studies .16. Chem Pharm Bull 38, 1299-1307. 1254
Jassbi, A.R., Zamanizadehnajari, S., and Baldwin, I.T. (2010). 17-1255
hydroxygeranyllinalool glycosides are major resistance traits of Nicotiana 1256
obtusifolia against attack from tobacco hornworm larvae. Phytochemistry 71, 1257
1115-1121. 1258
Jassbi, A.R., Zamanizadehnajari, S., Kessler, D., and Baldwin, I.T. (2006). A new 1259
acyclic diterpene glycoside from Nicotiana attenuata with a mild deterrent effect 1260
on feeding Manduca sexta larvae. Z Naturforsch B 61, 1138-1142. 1261
Jassbi, A.R., Gase, K., Hettenhausen, C., Schmidt, A., and Baldwin, I.T. (2008). 1262
Silencing geranylgeranyl diphosphate synthase in Nicotiana attenuata 1263
dramatically impairs resistance to tobacco hornworm. Plant Physiol 146, 974-986. 1264
Kessler, A., and Baldwin, I.T. (2001). Defensive function of herbivore-induced plant 1265
volatile emissions in nature. Science 291, 2141-2144. 1266
Kind, T. and Fiehn, O. (2010). Advances in structure elucidation of small molecules 1267
using mass spectrometry. Bioanal. Rev. 2, 23-60. 1268
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted August 27, 2020. ; https://doi.org/10.1101/2020.08.26.267690doi: bioRxiv preprint
Kistner, C., and Matamaros, M. (2005). RNA Isolation using phase extraction and LiCl 1269
precipitation. Lotus japonicas Handbook, pp 123-124. 1270
Krebs, I., and Yates, S. (2008). Fully unsupervised automatic assignment and 1271
annotation of sum formulae for product ion peaks, neutral losses in MS and 1272
product ion spectra. Proceedings of the ASMS 2008, MPG 152. 1273
(https://www.bruker.com/applications/life-1274
sciences/metabolomics/literature/latest-literature.html, Brochure: Posterhall 1275
Metabolomics, 09-2011, Page 23). 1276
Krugel, T., Lim, M., Gase, K., Halitschke, R., and Baldwin, I.T. (2002). 1277
Agrobacterium-mediated transformation of Nicotiana attenuata, a model 1278
ecological expression system. Chemoecology 12, 177-183. 1279
Kuhn, B.M., Errafi, S., Bucher, R., Dobrev, P., Geisler, M., Bigler, L., Zazimalova, E., 1280
and Ringli, C. (2016). 7-rhamnosylated flavonols modulate homeostasis of the 1281
plant hormone auxin and affect plant development. J Biol Chem 291, 5385-5395. 1282
Lee, J.H., Kiyota, N., Ikeda, T., and Nohara, T. (2006). Acyclic diterpene glycosides, 1283
capsianosides VIII, IX, X, XIII, XV and XVI from the fruits of paprika Capsicum 1284
annuum L. var. grossum BAILEY and Jalapeno Capsicum annuum L. var. 1285
annuum. Chem Pharm Bull 54, 1365-1369. 1286
Lee, J.H., Kiyota, N., Ikeda, T., and Nohara, T. (2007). Acyclic diterpene glycosides, 1287
capsianosides C, D, E, F and III, from the fruits of hot red pepper Capsicum 1288
annuum L. Used in Kimchi and their revised structures. Chem Pharm Bull 55, 1289
1151-1156. 1290
Lee, J.H., Kiyota, N., Ikeda, T., and Nohara, T. (2008). Three new acyclic diterpene 1291
glycosides from the aerial parts of paprika and pimiento. Chem Pharm Bull 56, 1292
582-584. 1293
Lee, J.H., El-Aasr, M., Ikeda, T., Oda, K., Miyashita, H., Yoshimitsu, H., Okawa, M., 1294
Kinjo, J., and Nohara, T. (2009). Two new acyclic diterpene glycosides from 1295
fruits of Habanero, Capsicum chinense. Chem Pharm Bull 57, 730-733. 1296
Leitner, M., Kaiser, R., Hause, B., Boland, W., and Mithofer, A. (2010). Does 1297
mycorrhization influence herbivore-induced volatile emission in Medicago 1298
truncatula? Mycorrhiza 20, 89-101. 1299
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted August 27, 2020. ; https://doi.org/10.1101/2020.08.26.267690doi: bioRxiv preprint
Lemaire, M., Nagnan, P., Clement, J.L., Lange, C., Peru, L., and Basselier, J.J. 1300
(1990). Geranyllinalool (diterpene alcohol) - an insecticidal component of pine 1301
wood and termites (Isoptera, Rhinotermitidae) in 4 european ecosystems. Journal 1302
of Chemical Ecology 16, 2067-2079. 1303
Li, J.C., Schuman, M.C., Halitschke, R., Li, X., Guo, H., Grabe, V., Hammer, A., and 1304
Baldwin, I.T. (2018). The decoration of specialized metabolites influences stylar 1305
development. Elife 7:e38611. 1306
Lou, Y., and Baldwin, I.T. (2003). Manduca sexta recognition and resistance among 1307
allopolyploid Nicotiana host plants. Proc Natl Acad Sci USA 100, 14581-14586. 1308
Macel, M., Visschers, I.G.S., Peters, J.L., Kappers, I.F., de Vos, R.C.H., and van 1309
Dam, N.M. (2019). Metabolomics of thrips resistance in pepper (Capsicum spp.) 1310
reveals monomer and dimer acyclic diterpene glycosides as potential chemical 1311
defenses. Journal of Chemical Ecology 45, 490-501. 1312
Macias, F.A., Lopez, A., Varela, R.M., Torres, A., and Molinillo, J.M.G. (2008). 1313
Helikauranoside A, a new bioactive diterpene. Journal of Chemical Ecology 34, 1314
65-69. 1315
Matile, P. (1980). The mustard oil bomb. Compartmentation of the myrosinase system. 1316
Biochem Physiol Pfl 175, 722-731. 1317
Meesapyodsuk, D., Balsevich, J., Reed, D.W., and Covello, P.S. (2007). Saponin 1318
biosynthesis in Saponaria vaccaria. cDNAs encoding beta-amyrin synthase and a 1319
triterpene carboxylic acid glucosyltransferase. Plant Physiol 143, 959-969. 1320
Mo, T., Liu, X., Liu, Y.Y., Wang, X.H., Zhang, L., Wang, J., Zhang, Z.X., Shi, S.P., 1321
and Tu, P.F. (2016). Expanded investigations of the aglycon promiscuity and 1322
catalysis characteristic of flavonol 3-O-rhamnosyltransferase AtUGT78D1 from 1323
Arabidopsis thaliana. Rsc Adv 6, 84616-84626. 1324
Moraga, A., Nohales, P., Perez, J., and Gomez-Gomez, L. (2004). Glucosylation of 1325
the saffron apocarotenoid crocetin by a glucosyltransferase isolated from Crocus 1326
sativus stigmas. Planta 219, 955-966. 1327
Mylona, P., Owatworakit, A., Papadopoulou, K., Jenner, H., Qin, B., Findlay, K., 1328
Hill, L., Qi, X., Bakht, S., Melton, R., and Osbourn, A. (2008). Sad3 and Sad4 1329
are required for saponin biosynthesis and root development in oat. Plant Cell 20, 1330
201-212. 1331
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted August 27, 2020. ; https://doi.org/10.1101/2020.08.26.267690doi: bioRxiv preprint
Nagel, R., Berasategui, A., Paetz, C., Gershenzon, J., and Schmidt, A. (2014). 1332
Overexpression of an isoprenyl diphosphate synthase in spruce leads to 1333
unexpected terpene diversion products that function in plant defense. Plant 1334
Physiol 164, 555-569. 1335
Naoumkina, M.A., Modolo, L.V., Huhman, D.V., Urbanczyk-Wochniak, E., Tang, 1336
Y.H., Sumner, L.W., and Dixon, R.A. (2010). Genomic and coexpression 1337
analyses predict multiple genes involved in triterpene saponin biosynthesis in 1338
Medicago truncatula. Plant Cell 22, 850-866. 1339
Nitao, J.K., Berhow, M., Duval, S.M., Weisleder, D., Vaughn, S.F., Zangerl, A., and 1340
Berenbaum, M.R. (2003). Characterization of furanocoumarin metabolites in 1341
parsnip webworm, Depressaria pastinacella. Journal of Chemical Ecology 29, 1342
671-682. 1343
Ohnuma, S., Hirooka, K., Tsuruoka, N., Yano, M., Ohto, C., Nakane, H., and 1344
Nishino, T. (1998). A pathway where polyprenyl diphosphate elongates in 1345
prenyltransferase - Insight into a common mechanism of chain length 1346
determination of prenyltransferases. J Biol Chem 273, 26705-26713. 1347
Pearse, I.S., Krugel, T., and Baldwin, I.T. (2006). Innovation in anti-herbivore defense 1348
systems during neopolypoloidy - the functional consequences of instantaneous 1349
speciation. Plant J 47, 196-210. 1350
Poreddy, S., Mitra, S., Schottner, M., Chandran, J., Schneider, B., Baldwin, I.T., 1351
Kumar, P., and Pandit, S.S. (2015). Detoxification of hostplant's chemical 1352
defence rather than its anti-predator co-option drives beta-glucosidase-mediated 1353
lepidopteran counteradaptation. Nat Commun 6. 1354
Ramegowda, V., Mysore, K.S., and Senthil-Kumar, M. (2014). Virus-induced gene 1355
silencing is a versatile tool for unraveling the functional relevance of multiple 1356
abiotic-stress-responsive genes in crop plants. Front Plant Sci 5. 1357
Ratcliff, F., Martin-Hernandez, A.M., and Baulcombe, D.C. (2001). Tobacco rattle 1358
virus as a vector for analysis of gene function by silencing. Plant J 25, 237-245. 1359
Reboul, R., Geserick, C., Pabst, M., Frey, B., Wittmann, D., Lutz-Meindl, U., 1360
Leonard, R., and Tenhaken, R. (2011). Down-regulation of UDP-glucuronic acid 1361
biosynthesis leads to swollen plant cell walls and severe developmental defects 1362
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted August 27, 2020. ; https://doi.org/10.1101/2020.08.26.267690doi: bioRxiv preprint
associated with changes in pectic polysaccharides. J Biol Chem 286, 39982-1363
39992. 1364
Richman, A., Swanson, A., Humphrey, T., Chapman, R., McGarvey, B., Pocs, R., 1365
and Brandle, J. (2005). Functional genomics uncovers three 1366
glucosyltransferases involved in the synthesis of the major sweet glucosides of 1367
Stevia rebaudiana. Plant J 41, 56-67. 1368
Roda, A.L., Oldham, N.J., Svatos, A., and Baldwin, I.T. (2003). Allometric analysis of 1369
the induced flavonols on the leaf surface of wild tobacco (Nicotiana attenuata). 1370
Phytochemistry 62, 527-536. 1371
Saedler, R., and Baldwin, I.T. (2004). Virus-induced gene silencing of jasmonate-1372
induced direct defences, nicotine and trypsin proteinase-inhibitors in Nicotiana 1373
attenuata. J Exp Bot 55, 151-157. 1374
Saito, K., Hirai, M.Y., and Yonekura-Sakakibara, K. (2008). Decoding genes with 1375
coexpression networks and metabolomics - 'majority report by precogs'. Trends 1376
Plant Sci 13, 36-43. 1377
Sandeep, S., and Paarakh, P. M. (2009). Jasminum grandiflorum Linn (Chameli): 1378
ethnobotany, phytochemistry and pharmacology: a review. Pharmacology Online 1379
Newsletter 2, 586-595. 1380
Schafer, M., Brutting, C., Baldwin, I.T., and Kallenbach, M. (2016). High-throughput 1381
quantification of more than 100 primary- and secondary-metabolites, and 1382
phytohormones by a single solid-phase extraction based sample preparation with 1383
analysis by UHPLC-HESI-MS/MS. Plant Methods 12. 1384
Senthil-Kumar, M., and Mysore, K.S. (2011). Caveat of RNAi in plants: the off-target 1385
effect. Methods Mol Biol 744, 13-25. 1386
Shinozaki, Y., Tobita, T., Mizutani, M., and Matsuzaki, T. (1996). Isolation and 1387
identification of two new diterpene glycosides from Nicotiana tabacum. Biosci 1388
Biotech Bioch 60, 903-905. 1389
Snook, M.E., Johnson, A.W., Severson, R.F., Teng, Q., White, R.A., Sisson, V.A., 1390
and Jackson, D.M. (1997). Hydroxygeranyllinalool glycosides from tobacco 1391
exhibit antibiosis activity in the tobacco budworm [Heliothis virescens (F)]. J Agr 1392
Food Chem 45, 2299-2308. 1393
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted August 27, 2020. ; https://doi.org/10.1101/2020.08.26.267690doi: bioRxiv preprint
Sumner, L.W., Amberg, A., Barrett, D., Beale, M.H., Beger, R., Daykin, C.A., Fan, 1394
T.W.M., Fiehn, O., Goodacre, R., Griffin, J.L., Hankemeier, T., Hardy, N., 1395
Harnly, J., Higashi, R., Kopka, J., Lane, A.N., Lindon, J.C., Marriott, P., 1396
Nicholls, A.W., Reily, M.D., Thaden, J.J., and Viant, M.R. (2007). Proposed 1397
minimum reporting standards for chemical analysis. Metabolomics 3, 211-221. 1398
Szerszen, J.B., Szczyglowski, K., and Bandurski, R.S. (1994). Iaglu, a Gene from 1399
Zea mays involved in conjugation of growth hormone indole-3-acetic acid. 1400
Science 265, 1699-1701. 1401
Taguchi, G., Shitchi, Y., Shirasawa, S., Yamamoto, H., and Hayashida, N. (2005). 1402
Molecular cloning, characterization, and downregulation of an acyltransferase that 1403
catalyzes the malonylation of flavonoid and naphthol glucosides in tobacco cells. 1404
Plant J 42, 481-491. 1405
Taguchi, G., Ubukata, T., Nozue, H., Kobayashi. Y., Takahi, M., Yamamoto, H. and 1406
Hayashida, N. (2010). Malonylation is the jey reaction in the metabolism of 1407
xenobiotic phenolic glucosides in Arabidopsis and tobacco. Plant J. 63, 1031-1408
1041 1409
Terauchi, M., Kanamori, H., Nobuso, M., Yahara, S., and Nohara, T. (1997a). 1410
Detection and determination of antioxidative components in Lycium chinense. 1411
Natural Medicines 51, 389-391. 1412
Terauchi, M., Kanamori, H., Nobuso, M., Yahara, S., and Nohara, T. (1997b). 1413
Seasonal variation of amounts of water-soluble components in Lycium chinense. 1414
Natural Medicines 51, 458-460. 1415
Terauchi, M., Kanamori, H., Nobuso, M., Yahara, S., and Yamasaki, K. (1998a). New 1416
acyclic diterpene glycosides, lyciumoside IV-IX from Lycium chinense Mill. 1417
Natural Medicines 52, 167-171. 1418
Terauchi, M., Kanamori, H., Nobuso, M., Fukuda, S., Yahara, S., and Yamasaki, K. 1419
(1998b). Antimicrobial components in leaves of Lycium chinense Mill. J Food Hyg 1420
Soc Jpn 39, 399-405. 1421
Terauchi, M., Kanamori, H., Nobuso, M., Sakamoto, I., Yahara, S., Nohara, T., and 1422
Kohda, H. (1995). Analysis of acyclic diterpene glycosides in Lycii Folium. 1423
Natural Medicines 49, 133-136. 1424
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted August 27, 2020. ; https://doi.org/10.1101/2020.08.26.267690doi: bioRxiv preprint
Van Poecke, R.M.P., Posthumus, M.A., and Dicke, M. (2001). Herbivore-induced 1425
volatile production by Arabidopsis thaliana leads to attraction of the parasitoid 1426
Cotesia rubecula: Chemical, behavioral, and gene-expression analysis. Journal of 1427
Chemical Ecology 27, 1911-1928. 1428
Xu, S.Q., Zhou, W.W., Pottinger, S., and Baldwin, I.T. (2015). Herbivore associated 1429
elicitor-induced defences are highly specific among closely related Nicotiana 1430
species. Bmc Plant Biol 15. 1431
Xu, S.Q., Brockmoller, T., Navarro-Quezada, A., Kuhl, H., Gase, K., Ling, Z.H., 1432
Zhou, W.W., Kreitzer, C., Stanke, M., Tang, H.B., Lyons, E., Pandey, P., 1433
Pandey, S.P., Timmermann, B., Gaquerel, E., and Baldwin, I.T. (2017). Wild 1434
tobacco genomes reveal the evolution of nicotine biosynthesis. P Natl Acad Sci 1435
USA 114, 6133-6138. 1436
Yadav, R.K., Sangwan, R.S., Sabir, F., Srivastava, A.K., and Sangwan, N.S. (2014). 1437
Effect of prolonged water stress on specialized secondary metabolites, peltate 1438
glandular trichomes, and pathway gene expression in Artemisia annua L. Plant 1439
Physiol Bioch 74, 70-83. 1440
Yu, X.H., Chen, M.H., and Liu, C.J. (2008). Nucleocytoplasmic-localized 1441
acyltransferases catalyze the malonylation of 7-O-glycosidic (iso)flavones in 1442
Medicago truncatula. Plant J 55, 382-396. 1443
Zangerl, A.R., and Berenbaum, M.R. (1998). Damage-inducibility of primary and 1444
secondary metabolites in the wild parsnip (Pastinaca sativa). Chemoecology 8, 1445
187-193. 1446
1447
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted August 27, 2020. ; https://doi.org/10.1101/2020.08.26.267690doi: bioRxiv preprint
Figure 1: HGL-DTG biosynthetic pathway
Components of the diverse HGL-DTGs structures previously identified and annotated in
the leaves of N. attenuata that differ with respect to their sugar and malonyl group
composition.
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted August 27, 2020. ; https://doi.org/10.1101/2020.08.26.267690doi: bioRxiv preprint
Figure 2: Metabolite profiling and morphologies of N. attenuata plants transiently-
silenced in the expression of HGL-DTG-predicted UGTs by virus-induced gene
silencing (VIGS)
A) Extracted ion chromatograms (EIC) for identified HGL-DTGs in leaves of 37 day-old
elongated N. attenuata plants silenced in UGT91T1, UGT74P3 or UGT74P5 transcript
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted August 27, 2020. ; https://doi.org/10.1101/2020.08.26.267690doi: bioRxiv preprint
accumulation as well as in the empty vector controls (pTV00). HGL-DTGs were
categorized into rhamnosylated, non-rhamnosylated and intermediates with one or two
glucose moieties to facilitate visualization. The 17-hydroxygeranyllinallool (17-HGL)
aglycone was only detected in transiently-silenced pTVUGT74P3 and pTVUGT74P5
lines. B) Heatmap visualization of the patterns of deregulation in control plants or in
plants treated with 150 µg methyl jasmonate (MeJA) in 20 µL lanolin paste (N=5). The
color gradient visualizes fold changes in individual HGL-DTGs for each of the VIGs
constructs compared to the average in the pTV00 empty vector VIGS plants. C)
Morphological alterations observed in pTV00, pTVUGT74P3 and pTVUGT74P5
transiently-transformed plants ranged from necrotic spots and tissues to necrotic apical
meristem and flower buds frequently stalled in the opening process. Additional
phenotypic details are provided in supplemental Figures S7a-c. D) and E) Morphological
alterations of the corolla tube, corolla limb, style, ovary and nectary (N=20). Asterisks
indicate significant differences between the empty vector control and pTVUGT74P3 or
pTVUGT74P5 VIGS plants (*P ≤ 0.05, ** P < 0.01, *** P < 0.001).
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted August 27, 2020. ; https://doi.org/10.1101/2020.08.26.267690doi: bioRxiv preprint
Figure 3: Metabolite profiling and morphologies of N. obtusifolia plants
transiently-silenced in the expression of HGL-DTG-predicted UGTs by VIGS
A) Extracted ion chromatograms (EIC) for the identified HGL-DTGs of 37 day-old
elongated N. obtusifolia plants silenced in NoUGT91T1-like, NoUGT74P4 and
NoUGT74P6 transcript accumulations as well as in the empty vector (pTV00) controls.
Rhamnosylated, non-rhamnosylated and intermediate HGL-DTGs with one or two
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted August 27, 2020. ; https://doi.org/10.1101/2020.08.26.267690doi: bioRxiv preprint
glucose moieties were color-categorized as in Fig. 3. The 17-HGL aglycone was only
detected in pTVNoUGT74P4, pTVNoUGT74P6 and the double construct
pTVNoUGT74P4/UGT74P6 VIGS plants. B) Heatmap visualization of deregulations in
the leaf HGL-DTG profiles of transiently-transformed pTVNoUGt91T1-like,
pTVNoUGT74P4, pTVNoUGT74P6 or pTVNoUGT74P4/UGT74P6 plants (N=5). The
color gradient visualizes fold changes in individual HGL-DTGs for each of the VIGS
constructs compared to the average in the pTV00 empty vector VIGS plants. C)
Morphological alterations in pTV00, pTVNoUGT91T1-like, pTVNoUGT74P4,
pTVNoUGT74P6 or pTVNoUGT74P4/UGT74P6 ranged from necrotic spots to a high
percentage of stalled flower buds. Additional phenotypic details are reported in
supplemental Figure 9.
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted August 27, 2020. ; https://doi.org/10.1101/2020.08.26.267690doi: bioRxiv preprint
Figure 4: Recombinant UGT74P3, UGT74P4, and UGT74P5 proteins glucosylate
the 17-HGL aglycone
Enzyme activity assays of the recombinant UGT74P3, UGT74P4 and UGT74P5 proteins
expressed in E. coli BL21 DE3 cells. Chromatograms (EIC traces for aglycone m/z
329.2475, HGL-DTGs with 1 glucose moiety m/z 491.3003 and HGL-DTGs with 2
glucose moieties m/z 653.3507) of analyses of 4 UGTs recombinant protein incubated
for 3 h in 50 mM Tris HCL pH 7.0 with 5 mM 17-HGL in the presence of 5 mM UDP-
glucose. Additionally, activity assays combining UGT74P5 with UGT74P3 or UGT74P4
were performed, but did not differ from the enzyme assays using only UGT74P3 or
UGT74P4.
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted August 27, 2020. ; https://doi.org/10.1101/2020.08.26.267690doi: bioRxiv preprint
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted August 27, 2020. ; https://doi.org/10.1101/2020.08.26.267690doi: bioRxiv preprint
Figure 5: Metabolite profiling and morphological characterization of stably
silenced N. attenuata plants
A) EICs for identified HGL-DTGs in leaves of 42 day-old elongated N. attenuata plants
silenced in GGPPS, UGT91T1, UGT74P3 and UGT74P5 transcript accumulation as well
as WT control plants. HGL-DTGs were categorized into rhamnosylated, non-
rhamnosylated HGL-DTG and intermediates with one or two glucose moieties to
facilitate visualization. The 17-HGL aglycone was only detected in IRugt74p5 and
IRugt74p3/ugt74p5 plants. B) Heatmap visualization of deregulations in the leaf HGL-
DTG profile of IRugt91t1, IRugt74p5, IRugt74p3/ugt74p5 and IRggpps (N=5). The color
gradient visualizes fold changes in individual HGL-DTGs for each of the stably-
transformed lines compared to the average in the WT plants. C) Morphological
alterations in IRugt74p5, IRugt74p3/ugt74p5 with milder phenotypes ranged from
necrotic spots and tissues, altered leaf shape and thickness, to apical meristem necrosis
and a high percentage of stalled flower buds and overall highly stunted growth.
Additional details of these phenotypes are shown in supplemental Figure S14a-c. D) 1-
year-old independent T0-transformants silenced in the expression of UGT74P3,
UGT74P5 and UGT74P3/UGT74P5. Strong morphological alterations ranging from
stunted growth, succulent leaves, stalled flower buds to a ‘broom’ like appearance were
consistently detected among T0-transformants. Viable seeds were produced by a few
transformants.
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted August 27, 2020. ; https://doi.org/10.1101/2020.08.26.267690doi: bioRxiv preprint
Figure 6: Silencing efficiency for the three 17-HGL-DTG biosynthetic UGTs in
IRugt91t1, IRugt74p5, IRugt74p3/ugt74p5 plants
Relative transcript abundance of UGT91T1, UGT74P3 and UGT74P5 in leaves of stably
transformed N. attenuata plants (N=4). Asterisks indicate significant differences between
WT control and stable transformants (*P ≤ 0.05, ** P < 0.01, *** P < 0.001).
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted August 27, 2020. ; https://doi.org/10.1101/2020.08.26.267690doi: bioRxiv preprint
Figure 7: Abolishing 17-HGL aglycone synthesis by silencing NaGGPPS
abrogates morphological alterations resulting from the silencing of UGT74P3 and
UGT74P5.
A) Morphological alterations of 42-day-old N. attenuata plants stably transformed to
silence NaGGPPS expression (IRggpps) and WT after inoculation with A. tumefaciens
harboring pTVUGT74P5, pTVUGT74P3 and the empty vector control (pTV00) VIGS
constructs. B) Stem height, number of side branches and rosette diameter in WT and
IRggpps. Asterisks indicate significant differences between empty vector control and
transient silenced lines (*P ≤ 0.05, ** P < 0.01, *** P < 0.001).
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted August 27, 2020. ; https://doi.org/10.1101/2020.08.26.267690doi: bioRxiv preprint
Figure 8: Application of 17-HGL aglycone results in necrotic lesions that
phenocopy those observed in IRugt74p5 and IRugt74p3/ugt74p5 plants
A) Concentrations of HGL in leaf material of WT, IRggpps, IRugt74p5,
IRugt74p3/ugt74p5 and IRugt91t1. B) Necrotic leaf tissue of 32 day-old elongated WT
and 48 day-old flowering IRggpps plants treated with DMSO, DMSO + 140nmol HGL,
DMSO + 280 nmol HGL and DMSO + 9800nmol HGL after 1 day. C) Percentage of
damaged leaf area in WT (N=3) and IRggpps (N=5). Asterisks indicate significant
differences between Control (DMSO) and treated (+HGL) leaves (*P ≤ 0.05, ** P < 0.01,
*** P < 0.001).
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted August 27, 2020. ; https://doi.org/10.1101/2020.08.26.267690doi: bioRxiv preprint
Figure 9: Performance assays show reduced growth of M. sexta fed on
transformed lines impaired in HGL-DTG glycosylation
A) Mass of M. sexta larvae feeding on leaf disk material of four stably transformed plants
impaired in glucosylation (IRugt74p5, IRugt74p3/ugt74p5) and rhamnosylation
(IRugt91t1) of HGL-DTGs as well as the formation of their precursor geranylgeranyl
diphosphate (IRggpps) (average ± SE; n=24 to 30). Larvae grow significantly larger on
IRggpps (P=0.005) and are significantly smaller on IRugt74p3/ugt74p5 (P=0.005) and
IRugt74p5 (P=0.006) by day 6 as determined by Mann-Whitney-Wilcox Pairwise Tests.
For clarity, significance is only shown for days 12: *P<0.05, **P<0.01, ***P<0.001. B)
Mass of consumed leaf disks after caterpillars fed leaf disks of transgenic plants
(average ± SE; n=26 to 30 leaf disks with one larva). Larvae fed IRugt74p5,
IRugt74p3/ugt74p5 and IRugt91t1 consumed significantly less leaf disk material as
determined by Mann-Whitney-Wilcox Pairwise Tests. *P<0.05, **P<0.01, ***P<0.001.
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted August 27, 2020. ; https://doi.org/10.1101/2020.08.26.267690doi: bioRxiv preprint