Potato virus X-delivered CRISPR activation programs lead to strong endogenous gene induction and transient metabolic reprogramming in Nicotiana benthamiana.

Programmable transcriptional regulators based on CRISPR architecture are promising tools for the induction of plant gene expression. In plants, CRISPR gene activation is effective with respect to modulating development processes, such as the flowering time or customizing biochemical composition. The most widely used method for delivering CRISPR components into the plant is Agrobacterium tumefaciens-mediated genetic transformation, either transient or stable. However, as a result of their versatility and their ability to move, virus-derived systems have emerged as an interesting alternative for supplying the CRISPR components to the plant, in particular guide RNA (gRNA), which represents the variable component in CRISPR strategies. In the present study, we describe a Potato virus X-derived vector that, upon agroinfection in Nicotiana benthamiana, serves as a vehicle for delivery of gRNAs, producing highly specific virus-induced gene activation. The system works in combination with a N. benthamiana transgenic line carrying the remaining complementary CRISPR gene activation components, specifically the dCasEV2.1 cassette, which has been shown previously to mediate strong programmable transcriptional activation in plants. Using an easily scalable, non-invasive spraying method, we show that gRNA-mediated activation programs move locally and systemically, generating a strong activation response in different target genes. Furthermore, by activating three different endogenous MYB transcription factors, we demonstrate that this Potato virus X-based virus-induced gene reprogramming strategy results in program-specific metabolic fingerprints in N. benthamiana leaves characterized by distinctive phenylpropanoid-enriched metabolite profiles.

INTRODUCTION

The emerging (CRISPR)/CRISPR‐associated protein (Cas) systems provide a battery of specific and efficient tools for gene editing and regulation (Mao et al., 2019; Moradpour & Abdulah, 2020). The recent achievements in targeted gene regulation in plants employing CRISPR/Cas‐based programmable transcription factors not only enhance our ability to accurately explore the links between gene expression and phenotype (Lowder et al., 2018; Pan, Wu, et al., 2021; Papikian et al., 2019), but also open a new paradigm for targeted crop improvement by, for example, tuning flowering time or customizing metabolic composition (Charfeddine et al., 2019; Lessard et al., 2002; Maeda & Nakamichi, 2022). CRISPR‐based transcriptional regulators typically comprise a catalytically inactive Cas enzyme with transcriptional regulatory domains anchored to its structure. This allows the specific binding of the chimeric transcriptional regulator to the genomic DNA, guided by the 20 nucleotides of the guide RNAs (gRNAs) (Lee et al., 2019). The most widely used Cas nuclease is the Cas9 endonuclease from Streptococcus pyogenes; however, other nucleases have been also optimized for achieving transcriptional regulation in plants (Zhang, Ren, et al., 2021). Currently, engineering efforts are focused on improving the efficiency and specificity of programmed gene regulation and also on the development of optimized delivery methods for the information encoded in the gRNA sequence. Thus, in an ideal scenario, crop plants could contain integrated CRISPR systems capable of executing gene activation instructions that would be delivered in the form of gRNA commands that move systemically throughout the plant (Molina‐Hidalgo et al., 2020).

Initial approaches for the delivery of CRISPR components to the plant were based on the widely used Agrobacterium‐mediated T‐DNA transformation, biolistic‐delivery or protoplast transformation methods (Mathur & Koncz, 1998; Wu et al., 2020). Agrobacterium‐mediated transformation relies on already established plant transformation protocols (Danilo et al., 2019; Tzfira & Citovsky, 2006). It is easy to handle and does not present strict limitations on the size of DNA constructs, but it requires the generation of stable transgenics through tissue culture and regeneration. Agrobacterium‐mediated transformation can operate transiently in some species, such as Nicotiana benthamiana (Wydro et al., 2006), without generating stable transgenic lines. In this case, however, the action of the CRISPR elements will be limited to the area of agroinfiltration. The biolistic delivery and protoplast transfection methods provide alternatives for plant species that cannot be transformed with Agrobacterium, and offer non‐transgenic options for genome editing (Hamada et al., 2017), such as the pre‐assembling of the Cas protein and the gRNA to form ribonucleoproteins (Liang et al., 2019; Zhang, Iaffaldano, & Qi, 2021). However, these delivery strategies present some disadvantages, such as the instability and the rapid degradation of the ribonucleoprotein complex, and do not circumvent the requirements of plant regeneration, which is not suitable for all plant species.

Classic work showed that viruses can be used to silence plant endogenous genes by simply harboring a sequence fragment homologous to the target gene in an approach known as virus induced‐gene silencing (VIGS) (Lu et al., 2003). In addition, recent studies highlight the potential use of viral vectors as transient delivery vehicles for CRISPR/Cas components in many biological systems including plants (Platt et al., 2014; Senís et al., 2014; Xu et al., 2019). This approach, commonly termed virus‐induced genome editing (VIGE), has mainly focused on the delivery of one or more gRNAs using RNA or DNA virus vectors in transgenic plants that stably express the Cas nuclease (Gentzel et al., 2022). The range of viral systems used for gRNA‐delivery in VIGE approaches is expanding, also enlarging the range of suitable plant hosts. Recently described VIGE vectors include the Tobacco rattle virus (TRV) (Ali et al., 2015; Ellison et al., 2020), the Cabbage Leaf Curl virus (CaLCuV) (Yin et al., 2015), the Tobacco mosaic virus (TMV) (Cody et al., 2017), the Pea early browning virus (Ali et al., 2018), the Foxtail mosaic virus (FoMV) (Mei et al., 2019), the Barley stripe mosaic virus (BSMV) (Hu et al., 2019; Li et al., 2021) and the Beet necrotic yellow vein virus (BNYVV) (Jiang et al., 2020). One of the VIGE systems reported previously that showed remarkable efficiency in terms of gene editing was based on the Potato virus X (PVX) (Ariga et al., 2020), a member of the genus Potexvirus that infects 62 plant species, several of them belonging to the Solanaceae family (Lico et al., 2015). In addition, PVX infection can be easily traced as a result of the phenotypic alterations produced in the host in a relatively short period. This recombinant virus was engineered to drive the expression of the gRNA by the sub‐genomic coat protein (CP) promoter. Also, this viral system entails a multiplexing strategy for gRNA expression with the particularity that the PVX‐based gRNA delivery does not need the presence of a tRNA‐processing system for properly expressing several gRNAs in tandem (Uranga, Aragonés, et al., 2021).

Viral vectors have been also proposed as shuttles for the transient delivery of exogenous, RNA‐encoded information to plant crops, leading, for example, to the ectopic and generalized expression of added‐value recombinant proteins, defense‐related genes or developmental regulators (Gleba et al., 2013; Marillonnet et al., 2005; Torti et al., 2021). This inspiring new breeding concept of transient genetic reprogramming can be expanded even further with the introduction of CRISPR programmable regulators. In this proposed scheme, gRNAs delivered to the plant using viral vectors would serve to reprogram endogenous gene expression. In the present study, we aimed to explore the potential of PVX‐based gRNA delivery for strong customized transcriptional activation in plants, thus expanding the toolbox of viral systems employed for gene regulation, which is currently based only on TRV (Ghoshal et al., 2020; Khakhar et al., 2021). To achieve this, the PVX‐based VIGE vector developed by Uranga, Aragonés, et al. (2021); Uranga, Vazquez‐Vilar, et al. (2021) was further engineered to express tandem repeats of modified gRNAs incorporating two RNA aptamers that bind the coat protein of the MS2 phage. This modification is an adaptation to the so‐called dCasEV2.1 system, a strong CRISPR programable activator described earlier by our group (Selma et al., 2019). In addition, the constant elements of the dCasEV2.1 system were stably transformed in N. benthamiana. The constant dCasEV2.1 module comprises two transcriptional units (TUs) driven by the CaMV 35S promoter: (i) a TU encoding an inactive version of the Cas9 (dCas9) fused to the plant activation domain EDLL (Tiwari et al., 2012) and (ii) a second TU encoding the coat protein of the MS2 phage fused to the VPR activator. VPR is a tandem fusion domain comprising the viral transactivation domains VP64, P65 and Rta (Chavez et al., 2015). Through a spray‐based agrodelivery method that is easy to handle, non‐invasive and adapted for large scale application, we demonstrated that the modified gRNAs were efficiently delivered to the plant, leading to potent PVX‐based virus‐induced gene reprogramming (VIGR). By targeting three different endogenous MYB factors (NbODO1, NbMYB24 and NbMYB21), allegedly involved in regulating the phenylpropanoid pathway, we generated three different phenylpropanoid‐enriched chemotypes in N. benthamiana leaves, obtaining information on the specific transcriptomic and metabolic fingerprint produced by each MYB factor.

RESULTS

Generation of dCasEV2.1 N. benthamiana chassis for CRISPRa

The reported ability of CRISPR‐based transcriptional activators to efficiently regulate gene expression in N. benthamiana prompted us to generate a transgenic line to serve as the chassis for CRISPR‐mediated targeted transcriptional activation (CRISPRa) experiments. To achieve this, a construct carrying the constant genetic elements in the dCasEV2.1 strategy, along with a yellow fluorescent protein (YFP) (Figure 1a), was employed to generate the transgenic lines, named YFP‐dCasEV. Upon transient delivery to the plant cells of one or more gRNAs targeting a specific promoter, a full dCasEV2.1 ribonucleoprotein complex will be assembled, which would expectedly result in a strong and specific gene activation (Figure 1b). The YFP‐positive transgenic plants obtained in Agrobacterium‐mediated transformation were then evaluated in a CRISPRa assay where a gRNA targeting the tomato Dihydroflavonol 4‐reductase (SlDFR) promoter was delivered via agroinfiltration. Targeted activation was scored using a luciferase reporter driven by the SlDFR promoter and co‐delivered together with the gRNA. Six transgenic lines were generated, although only two of them, namely YFP‐dCasEV_L3 and YFP‐dCasEV_L6, showed a strong gene activation upon specific gRNA expression and no activation in the absence of gRNA. As a positive control, a wild‐type N. benthamiana line was transiently co‐infiltrated with dCas9EV2.1 module, gRNA and the luciferase reporter. No statistically significant differences were found between the two positives, stably transformed lines and the transiently infiltrated control (Figure 1c). We decided to establish YFP‐dCasEV_L6 as our chassis line, from which we later obtained a homozygous T2 population.

Figure 1

Generation and selection of N. benthamiana chassis for CRISPRa.

(a) Representation of the DNA construct employed for N. benthamiana transformation that includes the constant module of the dCasEV2.1 (dCas9:EDLL and MS2:VPR), a YFP reporter (YFP) and a kanamycin resistance (NPTII) gene.

(b) Schematic representation of action mechanism of YFP‐dCasEV plants by the exogenous incorporation of gRNAs (gRNA2.1), forming the dCasEV2.1 complex as a transcriptional activation system.

(c) Normalized Fluc/Rluc ratios measured in transgenic lines containing the constant module of dCasEV2.1. The right side of the leaf transiently expresses a luciferase reporter driven by the SlDFR promoter and a gRNA that targets the position −150 relative to the TSS of the SlDFR promoter (gRNA:SlDFR). The left side of the leaf transiently expresses a luciferase reporter driven by the SlDFR promoter as a negative control of induction (no gRNA). WT represents a non‐transgenic plant that transiently expresses the dCasEV2.1 components. Asterisks indicate significant values by Student's t test (P < 0.05) compared to the control samples. L3, L4, L6, L7, L8 and L9 are independent YFP‐dCasEV T1 lines. Bars represent the mean ± SD RTAs, n = 3. Images generated with BioRender.com.

Engineering PVX‐based sgRNA delivery for systemic gene activation

Once the CRISPRa chassis was developed, it was necessary to optimize the gRNA delivery for gene activation. Taking the previous study by Uranga, Aragonés, et al. (2021) as a starting point, an alternative version of the pPVX‐based gRNA delivery vector was designed (PVX_VIGR vector), changing the native gRNA scaffold for the so‐called gRNA2.1 scaffold, which incorporates two RNA aptamers that recognize the MS2 phage coat protein at the 3′ end of the scaffold. Also, a multiplexing structure harboring two tandemly arrayed gRNAs under the control of the PVX CP promoter was incorporated into the recombinant pPVX viral vector (Figure 2a). Next, the efficiency of PVX_VIGR as an agroinfection‐mediated delivery agent for gRNAs in systemic gene activation was tested, using the CRISPRa‐ready dCasEV_L6 N. benthamiana line as recipient chassis. For this assay, we targeted the endogenous NbDFR gene, for which strong and highly specific responsiveness to dCasEV2.1 activation was reported previously (Selma et al., 2019). Two NbDFR‐specific gRNAs were designed for this purpose, targeting the positions −145 and −198 bp relative to the transcriptional start site (TSS) of the NbDFR gene, and cloned into the PVX_VIGR vector, generating the PVX::gDFR construct. In a first approach, the PVX_VIGR‐encoded gene activation program was agroinoculated in basal N. benthamiana leaves, employing an OD at 600 nm of 0.5 as described previously in Uranga, Aragonés, et al. (2021). The activation effects were recorded in the first and second symptomatic leaves, 9 and 14 days post‐inoculation (dpi), respectively. The symptoms are easily noticeable because of the characteristic phenotypic alterations generated in the leaves. As shown in Figure 2b, strong transcriptional activation of NbDFR was observed in symptomatic leaves in plants treated with PVX::gDFR construct, but not in those treated with a PVX_VIGR vector carrying control gRNA. No significant differences in the activation rates were found between the two analyzed symptomatic leaves, but a considerable increase in transcript levels was observed over time, reaching activation levels of 45–55‐fold at 9 dpi and 250‐ to 370‐fold at 14 dpi. The assay was repeated including a 20 dpi time point; however, no significant increase in the activation rates was observed at this time point (Figure S1).

Figure 2

Engineering and optimization of PVX_VIGR vector for gRNA delivery.

(a) Schematic representation of recombinant PVX_VIGR with the gRNA2.1. The gRNA2.1 is a modification of the native gRNA scaffold with the addition of two RNA aptamers at 3′ that recognize the MS2 phage coat protein.

(b) NbDFR mRNA fold change at 9 and 14 dpi measured in the first and second systemic symptomatic leaves (ST, systemic tissues), after targeting NbDFR at positions –145 and –198 (relative to the TSS) with PVX‐VIGA by syringe agroinoculation.

(c) As in (b) but using spray inoculation at an OD of 0.1.

(d) NbDFR mRNA fold changes observed when PVX‐VIGA was inoculated at OD values of 0.02, 0.004 and 0.0008 through spray application.

(e) NbDFR mRNA fold change in local tissues (LT) at 5 and 9 dpi using the same ODs as in (d). The C– represents a negative control where a non‐specific gRNA was delivered with the PVX_VIGR. Bars represent the mean ± SD RTAs, n = 4. Images generated with BioRender.com.

Our initial results prompted us to implement further optimizations, following the logic that the usability of VIGR in large scale applications, for example, would depend on the simplicity of the delivery method and loads of the delivery agent required. For that reason, a spray‐based agroinfection was employed as an alternative method to classical agroinfiltration with a syringe or vacuum. This method has some advantages such as a non‐invasive application and the possibility of infecting many plants at the same time with simple equipment. As a first approach, an Agrobacterium OD at 600 nm of 0.1 was employed, spraying two leaves per plant. After 14 dpi, NbDFR transcript levels in systemic tissues showed a successful activation of the gene, reaching up to 520‐fold in the first symptomatic leaf compared to control plants, where non‐specific gRNA was delivered with the PVX_VIGR (Figure 2c). The next analysis focused on obtaining a minimal effective OD for activation. In this case, also the youngest sprayed leaf was analyzed at 5 and 9 dpi to measure the effects in local tissues of the PVX_VGR recombinant virus. Interestingly, no significant differences in gene activation were found among the OD values of 0.02, 0.004 and 0.0008, both in systemic (Figure 2d) or local infections (Figure 2e). Surprisingly, the lowest OD values assayed in this second experiment resulted in activation rates even higher than those obtained in assays at an OD of 0.1, reaching up 11 000‐fold activation at 14 dpi in the first upper symptomatic leaf. As in the previous experiment, no further increases in the activation rates were observed at 20 dpi. By contrast, a slight decrease in the NbDFR transcript levels was observed at the later time point, perhaps as a result of leaf aging. Also, successful activation of NbDFR was achieved in local leaves by the spraying method, with maximum activation rates of 2500‐fold found at 9 dpi. These results confirm that the spray approach is an effective, easy‐to‐handle, non‐invasive method for delivering the activation information contained in the PVX‐VGR system to the plant.

PVX_VGR spray activates three different N. benthamiana MYB transcription factors generating distinctive transcriptional responses

To evaluate the ability of PVX_VGR to activate endogenous genes other than NbDFR, we designed specific activation gRNA programs for three endogenous MYB transcriptional factors (TFs), namely NbODO1 (NbD050495.1), NbMYB21 (NbD050797.1) and NbMYB24 (NbD027779.1). The criteria for the selection of these three factors were that they (i) show high levels of expression in at least one tissue (flowers) but low levels in the leaves and (ii) show homology with MYB factors from other species involved in the regulation of secondary metabolic profiles. With the first criterium, we wanted to ensure the inducibility of the selected genes, avoiding targeting pseudogenes or constitutively inactivated gene homeologues. With the second criterium, we aimed to show the possibility to create distinctive MYB‐specific metabolite fingerprints in leaves using PVX_VGR (Figure 3a). Highly homologous genes for all three TFs were earlier described as modifiers of the phenylpropanoid composition and/or of the volatile profiles in Arabidopsis, petunia and tomato, respectively (Dal Cin et al., 2011; Huang et al., 2020; Verdonk et al., 2005). Very little is known, however, about their role in N. benthamiana.

Figure 3

PVX_VIGR gRNA delivery for systemic activation of TFs.

(a) Schematic representation of the PVX_VIGR vector employed for altering the metabolite profiles.

(b) mRNA fold change at 14 dpi in the first symptomatic leaf (ST, systemic tissue) upon targeting the TFs NbODO1, NbMYB21 and NbMYB24 with dCasEV2.1 by spraying the PVX_VIGR vectors at an OD of 0.008.

(c) mRNA fold change of NbPAL, NbC4H, Nb4Cl, NbCHS and NbFLS genes at 14 dpi in the first symptomatic leaf in plants where the NbODO1, NbMYB21 or NbMYB24 were activated with dCasEV2.1 employing a spray inoculation of PVX_VIGR vectors at an OD of 0.008. The C– represents negative controls where a non‐specific gRNA was delivered with the PVX_VIGR. Asterisks indicate significant values by Student's t test (P < 0.05) compared to the control samples. Bars represent the mean ± SD RTAs , n = 5 in TF induction analysis and n = 3 in NbPAL, NbC4H, Nb4Cl, NbCHS and NbFLS transcriptomic analysis. Images generated with BioRender.com.

Two gRNAs were designed for activating each of the target genes. The criteria for protospacer selection were: (i) location between −100 and −300 bp upstream of the TSS (Pan, Sretenovic, & Qi, 2021); (ii) optimal on‐target score; and (iii) an absence of putative off‐targets in the N. benthamiana genome. Following the same multiplexing strategy described above, the gRNAs were incorporated in the PVX_VGR vector and delivered with the spray‐based method using an OD at 600 nm of 0.0008. This concentration was chosen because no differences were found after comparing the activation rates achieved with more concentrated PVX_VGR Agrobacterium suspensions. Next, the transcriptional activation of the targeted TFs was analyzed at 14 dpi in the first symptomatic leaves by a reverse trasncription quantitative real‐time polymerase chain reaction (RT‐qPCR). The results shown in Figure 3b indicate that all three PVX‐delivered programs resulted in significant activation of the target genes, which are otherwise repressed in leaves, reaching activation rates of > 600‐fold for NbODO1, > 110 fold for NbMYB21 and > 1200‐fold for NbMYB24 transcripts, respectively. As an indirect indication of the absolute mRNA levels obtained, the C t values of all three TFs exceeded those observed for the F‐box gene employed as a normalizing reference (between 25–26 Ct) (Table S1). Next, we analyzed the changes in the transcript levels of putative downstream‐regulated genes in the phenylpropanoid pathway: NbPAL, NbCHS, Nb4CL, NbC4H and NbFLS (Battat et al., 2019; Boersma et al., 2021). As shown in Figure 3c, only the NbODO1 activation program resulted in a strong up‐regulation of the NbPAL, Nb4CL and NbC4H genes, whereas the NbMYB24 activation program only generated a slight increase in NbFLS, and NbMYB21 did not result in significant activation in any of the analyzed genes.

PVX_VGR sprayed leaves show target‐specific metabolic profiles

Once the distinctive transcriptomic responses produced by each PVX_VIGR activation program were demonstrated, we decided to investigate whether this had a reflection on the leaf metabolite composition, yielding three distinctive metabolite profiles. For this, we first analyzed the volatile profiles of PVX_VGR‐treated leaves (Figure 4a), although no drastic differences were found in the volatile profiles when the 20 most significatively different volatiles were displayed in a hierarchical cluster. In general, a significant decrease of some monoterpenes such as alpha‐terpineol was observed in all the MYB‐activated samples compared with PVX_VGR‐treated control samples carrying gRNAs with no match in the genome. In up‐regulation terms, the greatest changes were found in the NbMYB21‐sprayed plants. Here, methoxycalamenene (a sesquiterpene) showed a 2.5‐fold increase, and other compounds with unknown biological functions, such as 3,4,4‐trimethyl‐2‐cyclopentene‐1‐one, 2,4‐diisopropylphenyl acetate and phenol 3,5‐dimethyl, showed NbMYB21‐specific overaccumulation (Figure 4b).

Figure 4

Analysis of the volatile profiles of N. benthamiana leaves activated in NbODO1, NbMYB21 and NbMYB24 expression.

(a) Hierarchical cluster analysis and heatmap representation of the volatile metabolite profiles obtained by GC‐MS from leaves treated with PVX_VIGR vectors targeting NbODO1, NbMYB21 and NbMYB24. Control samples (C–) were treated with a PVX_VIGR vector carrying a non‐specific gRNA. The cluster shows the 20 most significantly different metabolites by Student's t test (P < 0.05). The data were obtained using Euclidean distance and Ward's minimum variance method. Red indicates up‐regulated and blue indicates down‐regulated metabolites. Five biological replicates were employed per condition.

(b) Signal intensity analysis of the differential compounds found in at least one of the PVX_VIGR treatments (P < 0.05). Bars represent the mean ± SD fold change (n = 5). Asterisks indicate significant values by Student's t test (*P < 0.05, **P < 0.01 and ****P < 0.0001).

The modest differences observed in the non‐targeted analysis of the volatile compounds prompted us to evaluate the non‐volatile metabolic profiles. For that, an untargeted LC‐MS analysis was carried out with the same samples analyzed previously by RT‐qPCR and GC‐MS. Interestingly, LC‐MS revealed treatment‐specific metabolic fingerprints, as shown by the principal component analysis in Figure 5a. The class‐specific differences are depicted in Figure 5b, where the 350 most significantly different features are hierarchically clustered. The transcription factor that generates the most distinctive non‐volatile metabolic fingerprint is NbMYB24, followed by NbODO1. By contrast, NbMYB21 activation shows a lower ability to alter the metabolite content of N. benthamiana leaves. A tentative identification of differential compounds, performed by the differential m/z ions and their respective retention times, was carried out as detailed in Table 1 and summarized in Figure 5c. As expected, most perturbations corresponded to metabolites in the phenylpropanoid pathway. This is reflected by the increased levels of phenylalanine in NbODO1‐ (1.6‐fold) and NbMYB24‐ (1.9‐fold) induced samples, as well as by the increased levels of coumaroyl‐D‐quinic acid, the product of the PAL, C4H and CL4 gene activities, in the samples sprayed with NbODO1 (2.3‐fold) and NbMYB24 (2.7‐fold) activation programs. The accumulation of 5‐O‐caffeoyl shikimic acid and caffeic acid 3‐glucoside in the NbMYB24 (5.1‐ and 2.2‐fold, respectively) and NbODO1 samples (4.4‐ and 2.3‐fold, respectively) indicates that the activation of these two TFs leads to the up‐regulation of the caffeic acid branch of the phenylpropanoid pathway. By contrast, a clear down‐regulation of chlorogenic acid is observed in NbMYB21 samples (0.3‐fold). Furthermore, a general reduction in N‐caffeoylputrescine was observed in all three TF activated samples. Finally, in the NbODO1 samples, the metabolic flux of phenylpropanoid derivatives presents a strong shift toward lignans accumulation, as evidenced by the high accumulation of syringaresinol β‐d‐glucoside (> 22‐fold) and sinapoylglucose (> 4‐fold). Slight increases in these compounds can also be observed in NbMYB24 and NbMYB21 samples, although to a lower extent. Remarkably, flavonols over‐accumulate in all TF‐targeted samples, specifically kaempferol glucosides, with the highest accumulation observed in NbODO1 samples (11‐fold). Finally, other metabolic pathways also appear to be perturbed, although to a lower extent. A significant increase in phospholipids (linoleoylglycerol and lysoPC) and a decrease in hydroxy steroid derivatives (esthylestrenol) were found in NbMYB24‐induced samples.

Figure 5

Analysis of the metabolite profiles of N. benthamiana plants sprayed with PVX_VIGR for targeting NbODO1, NbMYB21 and NbMYB24.

(a) Principal component analysis and (b) hierarchical cluster analysis and heatmap representation resulting from the non‐targeted LC‐MS data obtained from leaves treated with PVX_VIGR targeting NbODO1, NbMYB21 or NbMYB24. Control samples (C–) were treated with a PVX_VIGR vector carrying a non‐specific gRNA. Five biological samples were employed per condition. The m/z features represented in the heatmap are the 350 most significantly different using anova (P < 0.05). The data were obtained using Euclidean distance and Ward's minimum variance method. Red indicates up‐regulated and blue indicates down‐regulated features. (c) Schematic representation of the metabolic pathways and metabolite subgroups that show differential abundance after inducting the N. benthamiana plants with a PVX‐VIGR for targeting NbODO1, NbMYB21 or NbMYB24 TFs compared to control samples (C–). The blue arrows represent an increment of the indicated metabolite group in the samples where NbODO1, NbMYB21 or NbMYB24 were up‐regulated with dCasEV2.1. The red arrows represent a down‐regulation of the indicated metabolite group in the samples where NbODO1, NbMYB21 or NbMYB24 were up‐regulated with dCasEV2.1. The orange asterisk represents that some metabolites that belong to the indicated chemical class are increased and other metabolites are down‐regulated in the samples where NbODO1, NbMYB21 or NbMYB24 were up‐regulated with dCasEV2.1. The equals (=) symbol represents no differential metabolite changes in the sample where one of the indicated MYB TF was up‐regulated with dCasEV2.1. The metabolites employed to generate this illustration were extracted from Table 1 where the differential metabolites tentatively identified are included.

Table 1

Differential tentatively identified compounds obtained from non‐targeted LC‐MS metabolic analysis of the leaves treated with PVX_VIGR targeting NbODO1, NbMYB21 or NbMYB24

Compounds	Mass	RT (min)	Fold change
			C‐	NbODO1	NbMYB21	NbMYB24
l‐Phenylalanine	165.08	4.81	1.00 ± 0.01	1.60 ± 0.45*	1.36 ± 0.17	1.90 ± 0.09***
N‐Caffeoylputrescine I	250.13	5.04	1.00 ± 0.62	0.33 ± 0.19**	0.12 ± 0.10***	0.31 ± 0.21*
N‐Caffeoylputrescine II	250.13	5.53	1.00 ± 0.64	0.30 ± 0.15**	0.12 ± 0.10***	0.30 ± 0.13*
N‐Caffeoylputrescine III	250.13	7.10	1.00 ± 0.53	0.33 ± 0.19	0.13 ± 0.09***	0.35 ± 0.22
Chlorogenic acid I	354.09	7.39	1.00 ± 0.28	0.43 ± 0.16	0.29 ± 0.22**	1.03 ± 0.32
Kaempferol‐3‐O‐glucuronide	462.10	7.59	1.00 ± 0.47	11.12 ± 3.13***	5.53 ± 2.23***	9.94 ± 9.07***
Chlorogenic acid II	354.09	7.63	1.00 ± 0.26	0.76 ± 0.21	0.27 ± 0.25**	1.49 ± 0.42
Chlorogenic acid III	354.09	7.80	1.00 ± 0.23	0.81 ± 0.22	0.27 ± 0.25**	1.59 ± 0.45
Benzenoid derivative	659.16	7.99	1.00 ± 0.52	0.06 ± 0.09***	0.36 ± 0.20	0.49 ± 0.10
Coumaroyl‐d‐quinic acid I	338.10	9.50	1.00 ± 0.12	1.30 ± 0.28	0.79 ± 0.16	2.28 ± 0.29***
Chlorogenic acid IV	354.09	10.17	1.00 ± 0.21	0.97 ± 0.38	0.42 ± 0.31	2.02 ± 0.26*
Caffeyl alcohol derivative	166.06	10.58	1.00 ± 0.09	2.66 ± 0.40***	1.08 ± 0.07	1.49 ± 0.28**
Caffeic acid 3‐glucoside	342.09	10.65	1.00 ± 0.17	2.29 ± 0.86***	1.52 ± 0.15	2.24 ± 0.17**
Chlorogenic acid V	354.10	10.87	1.00 ± 0.23	0.96 ± 0.34	0.40 ± 0.31*	1.95 ± 0.10**
Coumaroyl‐d‐quinic acid II	338.10	11.82	1.00 ± 0.13	1.98 ± 0.34***	1.24 ± 0.18	2.36 ± 0.13***
Coumaroyl‐d‐quinic acid III	338.10	12.31	1.00 ± 0.16	1.84 ± 0.38**	0.99 ± 0.25	2.38 ± 0.19***
Chlorogenic acid IV	354.09	12.38	1.00 ± 0.24	0.70 ± 0.27	0.41 ± 0.25*	1.47 ± 0.25
Coumaroyl‐d‐quinic acid IV	338.10	12.71	1.00 ± 0.08	2.34 ± 0.39***	1.04 ± 0.26	2.71 ± 0.15***
Kaempferol 3‐O‐gentiobioside‐7‐O‐rhamnoside	756.21	12.98	1.00 ± 0.13	2.86 ± 0.58***	1.52 ± 0.29*	2.03 ± 0.59***
5‐O‐Caffeoyl shikimic acid	336.08	13.00	1.00 ± 0.44	4.39 ± 2.22***	0.68 ± 0.49	5.10 ± 0.91***
Kaempferol 3‐O‐gentiobioside‐7‐O‐rhamnoside	756.21	13.09	1.00 ± 0.34	2.57 ± 0.72***	1.73 ± 0.25*	2.27 ± 0.48***
Sinapoylglucose (lignan)	386.12	13.27	1.00 ± 0.17	4.44 ± 0.94***	1.56 ± 0.25*	2.38 ± 0.12***
N‐Benzoylglutamic acid	251.08	13.67	1.00 ± 0.22	1.91 ± 0.4***	1.38 ± 0.13*	1.64 ± 0.09**
5‐O‐Coumaroyl‐d‐quinic acid	338.10	14.47	1.00 ± 0.14	1.60 ± 0.3**	0.95 ± 0.23	1.97 ± 0.09***
Lignan glucoside (Isolariciresinol 9′‐O‐beta‐d‐glucoside)	764.42	17.25	1.00 ± 0.14	3.01 ± 0.64**	1.40 ± 0.23	2.13 ± 0.33***
Kaempferol‐rutinoside II	594.16	18.25	1.00 ± 0.24	2.53 ± 0.76***	1.84 ± 0.19*	2.38 ± 0.61***
Coumaryl‐hexose malic acid I	442.18	19.05	1.00 ± 0.24	0.28 ± 0.34	0.33 ± 0.04	0.38 ± 0.15*
Syringaresinol β‐d‐glucoside	580.22	19.73	1.00 ± 0.14	22.96 ± 3.56***	1.84 ± 0.28***	3.49 ± 0.36***
Dihidroxy benzoic acid derivative I	420.11	23.03	1.00 ± 0.12	0.33 ± 0.09***	0.57 ± 0.05***	0.39 ± 0.03***
Coumarolyquinic acid glucoside derivative	1041.50	26.96	1.00 ± 0.24	0.52 ± 0.35	0.33 ± 0.20*	0.29 ± 0.18*
Coumarolyquinic acid glucoside derivative	1041.50	27.66	1.00 ± 0.29	0.46 ± 0.34	0.28 ± 0.18*	0.30 ± 0.17*
Coumarolyquinic acid glucoside derivative	1041.50	28.05	1.00 ± 0.36	0.30 ± 0.29	0.22 ± 0.16*	0.18 ± 0.18*
Coumarolyquinic acid glucoside derivative	1041.50	28.18	1.00 ± 0.31	0.48 ± 0.29	0.25 ± 0.16*	0.30 ± 0.15*
Coumarolyquinic acid glucoside derivative	1041.50	28.67	1.00 ± 0.30	0.54 ± 0.30	0.27 ± 0.16*	0.28 ± 0.17*
Coumarolyquinic acid glucoside derivative	1041.50	28.76	1.00 ± 0.31	0.43 ± 0.34	0.33 ± 0.17*	0.26 ± 0.16*
Coumarolyquinic acid glucoside derivative	1041.50	28.93	1.00 ± 0.23	0.42 ± 0.32	0.41 ± 0.20*	0.32 ± 0.14*
Hydroxy steroid derivative (ethylestrenol)	288.25	30.55	1.00 ± 0.41	0.31 ± 0.33	0.21 ± 0.20*	0.14 ± 0.14*
Hydroxy steroid derivative (ethylestrenol)	288.25	30.70	1.00 ± 0.44	0.27 ± 0.35	0.20 ± 0.17*	0.13 ± 0.12*
Linolenoylglycerol (monolinolenin)	352.26	33.84	1.00 ± 0.38	2.01 ± 0.36**	2.42 ± 0.36***	1.95 ± 0.63**
Lysophospholipid (lysoPC)	517.32	34.01	1.00 ± 0.11	3.56 ± 0.94***	1.95 ± 0.17**	1.47 ± 0.19**
Linoleoylglycerol (1‐linoleoyl‐2‐hydroxy‐sn‐glycero‐3‐PC)	519.33	34.89	1.00 ± 0.14	2.78 ± 0.97**	2.15 ± 0.13**	1.25 ± 0.24
Linolenoylglycerol (monolinolenin)	352.26	34.95	1.00 ± 0.18	1.59 ± 0.44**	2.01 ± 0.39**	2.18 ± 0.95**
Diterpenoid (Trigonosin C)	608.26	38.47	1.00 ± 1.01	6.71 ± 4.29**	8.39 ± 1.10**	8.81 ± 0.97**
Diterpenoid (Trigonosin C)	608.26	38.87	1.00 ± 1.01	7.87 ± 4.55**	8.63 ± 1.20**	8.07 ± 0.89**

The m/z signal intensity values employed for quantification are normalized with the negative control (C–) values. The colouration of table cells for each compound corresponds to the colors employed in the diagram of the metabolic routes in Figure 5c. Asterisks indicate Student's t test significant values (*P < 0.05, **P < 0.01 and ****P < 0.0001).

DISCUSSION

Viral systems have been widely used in plant sciences for a variety of applications, from recombinant protein production to reverse genetics. VIGS has been extensively used for interrogating gene function in diverse species and as a shortcut to gene knockout bypassing stable transformation. In VIGS strategies, viral clones incorporate sequences matching endogenous genes that, consequently, induce post‐transcriptional gene silencing in agroinfected plants. More recently, VIGE has emerged as an alternative to transgenics for efficient gene editing. In this latter case, the gene edition machinery is usually split into two parts: a constant module comprising the Cas enzyme and a variable module comprising the gRNA. The gRNA module is encoded in the viral vector, whereas the constant module, larger in size, can be either stably integrated into the plant genome (Uranga, Aragonés, et al., 2021) or delivered in a separate viral vector accepting larger cargo (Uranga, Vazquez‐Vilar, et al., 2021). Different viruses have been adapted for both VIGS and VIGE strategies. For VIGS, the TRV appears to be the ideal choice given its wide host range, mild silencing suppression and, consequently, strong gene silencing effect, whereas, in VIGE, other factors need to be taken into consideration, such as the gRNA accumulation levels, the ability to self‐process multiplexed gRNA arrays and the cargo capacity, all of which also play a role in editing efficiency. Therefore, other viruses such as PVX or TMV present advantageous characteristics for VIGE (Ariga et al., 2020; Cody et al., 2017).

Similar to VIGS and VIGE, VIGR represents a promising transient tool for plant biotechnology, with the ability not only to suppress, but also to strongly induce endogenous gene expression. Thus, VIGR offers a panoply of applications that range from gene discovery to transient genetic reprogramming of crops at a large scale, as recently proposed by Torti et al. (2021) for PVX and TMV‐based transgene expression. VipariNama, a VIGR strategy based on TRV, was pioneer in demonstrating the possibilities of virus delivery in CRISPR‐based transcriptional regulation in plants (Khakhar et al., 2021). Despite the elegant display of alternative VipariNama configurations, which includes both activation and repression strategies, the actual gene activation/repression levels observed with VipariNama tools were relatively modest (between 2‐ and 5‐fold), probably as a result of a limited intrinsic regulatory ability of the combination of TRV‐encoded gRNA and Cas9‐based synthetic transcription factors employed. These activation/repression levels were sufficient to induce measurable developmental and/or metabolic changes for some regulatory genes, although there is certainly room for improvement if robust responses are required in large‐scale applications. The results of the present study show that the PVX_VIGR system can induce a very strong gene activation in N. benthamiana, both local and systemic, opening the way to new and more robust applications. In this regard, a strong VIGA effect not only serves in research to elucidate gene function, but, in the case of N. benthamiana, also can be envisioned as a tool for shaping plant metabolic content. Recently, our group showed how CRISPR‐based programable transcriptional factors can be employed to channel metabolic fluxes toward the accumulation of selected metabolites in N. benthamiana leaves (Selma et al., 2021). In that work, gRNA‐encoded activation programs were delivered in the form of multiplex T‐DNA constructs via agroinfiltration, a strategy that limits the scalability of the approach. This limitation could be circumvented with the advantages offered by VIGR. In particular, the spraying strategy described here offers a solution for the large‐scale delivery of gene reprogramming instructions to N. benthamiana biofactories, with possibilities of extending the technology to other crops given the broad host range of PVX. In principle, in VIGR strategies, transcriptional activation/repression will not be sustained in the next generations without re‐inoculating the virus. However, the recent advances in CRISPR‐based epigenetic modifications open new perspectives on the ability to induce epigenetic heritable traits using VIGR (Ghoshal et al., 2020; Papikian et al., 2019; Selma & Orzáez, 2021).

The PVX recombinant virus presents an architecture that allows a multiplexing strategy for expressing the gRNAs without the necessity of pre‐tRNAs, as demonstrated previously by Uranga, Aragonés, et al., 2021. The addition of activation aptamers in the gRNA scaffold that are recognized by the MS2 protein in the recombinant virus structure has not been an obstacle to expressing active gRNAs. As a possible constraint of PVX_VIGR, cargo limitations might restrict the complexity of the gene activation/repression programs that can be delivered to the plant. In metabolic reprogramming using T‐DNA delivery via agroinfiltration, we previously employed up to 15 tandemly arrayed gRNAs to successfully redirect a metabolic pathway (Selma et al., 2021). However, it is unlikely that such large gRNA tandem repeats could work as efficient programs in PVX_VIGR. Future works would need to establish these limitations and circumvent them with other strategies such as the co‐delivery of non‐competing viruses.

The capacity of PVX_VIGR to activate genes in N. benthamiana was evaluated by targeting the transcriptional factors NbODO1, NbMYB21 or NbMYB24 and analyzing the specific changes in the metabolite profiles in the induced plants. In this case, although the selected targets are predominantly expressed in flowers, their role in N. benthamiana remains uncharacterized, giving added value to the employment of the PVX_VIGR to analyze the metabolite fingerprint of each TF. The MYB factors described previously as putative orthologues of our target genes were investigated in petunia, Arabidopsis and tomato. The over‐expression of these genes produced changes in the volatile and non‐volatile metabolic profiles in these species. In petunia, the overexpression of PhODO1 was linked to the production of compounds involved in the scent, such as benzenoids, phenylpropanoids, fatty acids and terpenoid derived volatiles (Verdonk et al., 2005). In addition, the petunia orthologs of MYB21 and MYB24, known as EOBI and EOBII (EMISSION OF BENZENOIDS), respectively, are also involved in the production of the components of the scent through tight interaction between them and PhODO1 (Spitzer‐Rimon et al., 2012). In tomato, ODO1 overexpression is also linked to the volatile production of benzenoids and phenylpropanoids; however, the biggest changes found in the tomato metabolome associated with ODO1 were the increase of non‐volatile phenylpropanoid derivatives, which includes some flavonoids, especially flavonols (Dal Cin et al., 2011; Xie et al., 2016). Concurrently, in Arabidopsis, the increase of flavonols in flowers was directly correlated with the expression of AtMYB99 (a homolog of ODO1), AtMYB21 and AtMYB24 (Battat et al., 2019; Zhang, He, et al., 2021). The interaction among all three paralogues appears to take place similarly in petunia and Arabidopsis. However, the metabolite profiles resulting from this interaction are different in the two species, suggesting that, in each species, these genes regulate the varied branches of the phenylpropanoid pathway differently. In addition, in Arabidopsis, AtMYB21 and AtMYB24 are known to interact with genes in the jasmonate pathway, leading to an increase in the production of terpenoids (Cheng et al., 2009). The up‐regulation of MYB homologs in N. benthamiana leaves using PVX_VIGR did not lead to dramatic changes in the volatile profiles, with the exemption of an increase in the levels of sesquiterpene methoxycalamenene observed in AtMYB21 samples. A probable explanation is a limited capacity of N. benthamiana leaves to produce volatile compounds, which may be related to a high activity of endogenous glycosyl transferases producing non‐volatile glycosylated derivatives (van Herpen et al., 2010; Louveau & Osbourn, 2019). By contrast, highly specific non‐volatile metabolic profiles were observed to be associated with each TF transcriptional activation. In all three treatments, most changes observed were associated with the phenylpropanoid pathway. This correlates with previous observations in Arabidopsis and tomato, where most changes associated with ODO1, MYB21 or MYB24 overexpression were in the non‐volatile fraction, especially phenylpropanoid derivatives (Battat et al., 2019; Xie et al., 2016). In particular, we found that all three treatments resulted in the overaccumulation of flavonols. By contrast, the accumulation of other phenylpropanoids appeared to be dependent on the activation program applied. For example, p‐coumaric acid and caffeic acid derivatives were enriched in NbMYB24 and NbODO1 activation programs but down‐regulated in NbMYB21‐induced samples. In the case of NbODO1, the observations correlate with the up‐regulation of NbPAL and NbC4H genes observed in the RT‐qPCR analysis. For NbMYB24, we did not observe a significant up‐regulation of the same genes. The observed metabolite accumulation could be explained by the activation of a set of homologous genes different from those evaluated by RT‐qPCR, or by a different timing in gene activation by NbMYB24. Other phenylpropanoid derivatives such as monolignans, although up‐regulated in all three treatments, resulted in a much higher accumulation in NbODO1 samples. These differential accumulations observed in certain groups of related metabolites, such as lignans or flavonols, suggest the distinctive specificities of each MYB factor in the activation of the different genes in the pathway and demonstrate the power of PVX_VIGR in reprogramming metabolic fluxes.

In conclusion, we show that PVX‐based gRNA delivery is an effective system for transient CRISPR activation. In the present study, we demonstrate that plant metabolic profiles can be customized through a spray agroinoculation of a PVX vector, expanding the current toolbox of viral systems employed for transcriptional activation. The precise and scalable spatiotemporal control of gene transcription offered by this tool opens the door to increasingly sophisticated designs in plant synthetic biology.

EXPERIMENTAL PROCEDURES

Generation and selection of N. benthamiana dCasEV lines

The N. benthamiana YFP‐dCasEV lines were generated following the transformation protocol described previously (Clemente, 2006). The construct (GB2618) was transferred to the LBA4404 Agrobacterium tumefaciens strain used for plant transformation. MS medium plates supplied with kanamycin at 100 mg L−1 were used to select the transgenic T0 lines. In addition, YFP analysis was carried out to double‐check the construct incorporation. Transgenic lines were sorted based on the T1 plants capable of generating efficient transcriptional activation. pDFR:Luciferase (GB1160) was employed as a reporter for testing the activation rates obtained incorporating the corresponding gRNA to target SlDFR promoter (gRNA:SlDFR, GB1121). A wild‐type N. benthamiana was infiltrated in parallel with the same reporter, the gRNA:SlDFR and dCasEV2.1 constant module construct for a fair comparison of the activation results obtained. In addition, the T1 plants were evaluated by segregation analysis for selecting a single‐copy T‐DNA insertion. The complete sequence of the GB constructs is available at: https://gbcloning.upv.es/.

The candidate single copy T2 of L3 and L6 lines were analyzed for homozygosity and, finally, YFP‐dCasEV‐L6.4.1 was selected as a homozygous population and kept for the following experiments. N. benthamiana transgenic lines were grown under light/darkness at 24°C/20°C with a 16:8 h photocycle in a growth chamber.

gRNA design and viral vector construction

Initially, the N. benthamiana endogenous gene NbDFR was selected for CRISPRa optimization. The protospacer was described previously by Selma et al. (2019). The design of gRNAs for NbODO1, NbMYB21 and NbMYB24 was performed employing the benchling CRISPR application tool (https://www.benchling.com). The protospacer sequences were selected within the activation window located between −100 and −300 bp upstream of the TSS for each gene (Pan, Sretenovic, & Qi, 2021). Protospacer sequences are listed in Table S2. Plasmid pPVX was described previously by Uranga, Aragonés, et al., 2021. The full‐length PVX cDNA (GenBank accession number MT799816) is flanked by the CaMV 35S promoter and the A. tumefaciens Nopaline synthase (Nos) terminator. The PVX CP promoter drives the expression of the gRNA, whereas the viral CP that harbors a deletion of 29 initial codons (Dickmeis et al., 2014) is expressed under the control of a heterologous promoter derived from the Bamboo mosaic virus (BaMV). The double gRNA2.1 for CRISPRa was included in recombinant virus plasmids through PCR amplification with high‐fidelity Phusion DNA polymerase (Thermo Fisher Scientific, Waltham, MA, USA), taking as template the GB construct GB2149 (https://gbcloning.upv.es), which contains a gRNA2.1 for activation. The two resulting amplicons containing the gRNAs were assembled into a pre‐digested pPVX/MluI plasmid through a Gibson DNA assembly reaction (NEBuilder HiFi DNA assembly master mix; New England Biolabs, Ipswich, MA, USA). The primers employed for gRNA2.1 construction and adaptation for PVX recombinant plasmids are listed in Table S3. All of the plasmids were confirmed by Sanger sequencing. As a negative control of infection, the PVX::gXT2 (un‐specific for activation) was employed without adding the 2.1 aptamers. The sequences of the pPVX plasmid and the PVX‐VIGR vectors are shown in Figure S2.

Plant inoculation

Transgenic N. benthamiana dCasEV2.1 plants were grown in growth chambers under light/darkness at 24°C/20°C with a 16:8 h photocycle. The inoculation was performed 4 weeks after sowing. The PVX_VIGR vectors were transferred to A. tumefaciens strain GV3101 by electroporation. Inoculation was carried out with overnight‐grown bacterial cultures. The cultures were pelleted and resuspended on agroinfiltration solution (10 mm MES, pH 5.6, 10 mm MgCl2 and 200 μm acetosyringone) and incubated for 2 h at room temperature with agitation. The OD of the bacterial cultures was adjusted to 0.5 at 600 nm for the syringe inoculation method and 0.1, 0.02, 0.004 and 0.0008 for the spray inoculation method. In the spray inoculation method, 0.04% of Fluvius (a non‐ionic adjuvant; CCL – Comptoir Commercial des Lubrifiants, Le Meux, France) was added. The final OD employed for the NbODO1, NbMYB21 and NbMYB24 spray induction was 0.0008. Two leaves per plant were treated in both inoculation methods, using a 1‐ml syringe on the abaxial side of the leaf or a complete spray of both sides of the leaves. Control plants were inoculated with PVX::gXT2 following the same procedure. The samples were collected at 9 and 14 dpi from the first and second symptomatic leaf employing the syringe approach. The symptoms are easily noticeable because of the characteristic phenotypic alterations generated in the leaves. Samples from the youngest sprayed leaf were collected at 5 and 9 dpi to measure the kinetics induction of NbDFR in local infection. Samples from the first symptomatic leaf were collected at 9, 14 and 20 dpi to measure the kinetics of the NbDFR gene in systemic infection. Finally, samples from the NbODO1, NbMYB21 and NbMYB24 induction assays were collected at 14 dpi from the first symptomatic leaf. All of the samples were immediately frozen in liquid nitrogen and stored at −80°C.

Luciferase/Renilla activity determination

The determination of the luciferase/renilla activity was carried out collecting one disc per leaf (d = 0.8 cm, approximately 18–19 mg) at 5 dpi. The samples were homogenized and extracted with 375 μl of ‘Passive Lysis Buffer’, followed by 10 min of centrifugation (14 000 g) at 4°C. Then, the supernatant was collected as working plant extract. Fluc and Rluc activities were determined using the Dual‐Glo® Luciferase Assay System (Promega, Madison, WI, USA) in accordance with the manufacturer's protocol with minor modifications: 10 μl of working plant extract, 40 μl of LARII and 40 μl of Stop&Glow Reagent were used. Measurements were made using a GloMax 96 Microplate Luminometer (Promega) with a 2‐s delay and a 10‐s measurement.

Fluc/Rluc ratios (RPUs) were determined as the mean value of three biological replicates coming from three independent agroinfiltrated leaves of the same plant. The RPUs were normalized to the Fluc/Rluc ratio obtained for a reference sample, which measures relative transcriptional activities (RTAs) of the evaluated promoter fused to the reporter.

RNA isolation and RT‐qPCR gene expression analysis

Total RNA was isolated from 100 mg of frozen leaf tissue employing an isolation kit (Gene Jet Plant Purification Mini Kit; Thermo Fisher Scientific) in accordance with the manufacturer's instructions. RNA was treated with an DNase‐I Invitrogen Kit and its concentration was adjusted for cDNA reaction. Then, 800 ng of total RNA was used for cDNA synthesis using a PrimeScript™ RT‐PCR Kit (Takara, Kusatsu, Japan) in a final volume of 20 μl in accordance with the manufacturer's instructions. Expression levels were measured in four biological replicates for NbDFR induction assays and five biological replicates for NbODO1, NbMYB21 and NbMYB24 induction assays. Each biological replicate was measured in three technical replicates reactions in the presence of a fluorescent dye (SYBR® Premix Ex Taq) using QuantStudio 3 equipment (Applied Biosystems, Waltham, MA, USA). The specific primers for detection of the target genes are listed in Table S4. F‐BOX protein gene was used as an internal reference gene (Liu et al., 2012). Basal expression levels were calculated from the control samples inoculated with PVX::gXT2 recombinant plasmid. Calculations of each sample were carried out according to the comparative ΔΔ C t method (Livak & Schmittgen, 2001).

Volatile organic compound and statistical analyses

Frozen and ground symptomatic leaf samples (70 mg) were incorporated in a 10‐ml headspace screw‐cap vial. The samples were prepared by adding 1 ml of 5 m CaCl2, 150 μl of 500 mm EDTA (pH = 7.5) and 3 μl 10 ppm lavadulol as an internal reference, followed by sonication for 5 min. Volatile compounds were captured by solid‐phase microextraction (SPME) with a 65‐μm polydimethylsiloxane/divinylbenzene SPME fiber (Supelco, Bellefonte, PA, USA). Volatile extraction was performed automatically using a CombiPAL autosampler (CTC Analytics, Zwingen, Switzerland). Vials were first incubated at 80°C for 3 min with agitation at 500 r.p.m. Then, the fiber was exposed to the vial through the headscape for 20 min under the same conditions of temperature and agitation. Desorption was performed at 250°C for 1 min (splitless mode) in the injection port of a 6890 N gas chromatograph (Agilent Technologies, Santa Clara, CA, USA). After desorption, the fiber was cleaned in an SPME fiber conditioning station (CTC Analytics) at 250°C for 5 min under a helium flow. Chromatography was performed on a DB5ms (60 m, 0.25 mm, 1 μm) capillary column (J&W; Agilent Technologies) with helium as the carrier gas at a constant flow of 1.2 ml min−1. The oven conditions were 40°C for 2 min, and a 5°C min−1 ramp was programmed until reaching 300°C with a final hold at 300°C for 5 min. Two blank controls with 1 ml of 5 m CaCl2 and 150 μl of 500 mm EDTA were included in the experiment.

Identification of the volatile compounds was performed by employing a customized library of N. benthamiana based on the NIST database (see Supplementary Material). A non‐targeted analysis for the differential emission of volatile compounds in the NbODO1, NbMYB21 and NbMYB24 samples was performed. Peak intensities were calculated by employing mass hunter Workstation Software for Quantitative Analysis (Agilent Technologies). Compounds with a peak intensity sample:blank ratio < 3 were removed for each condition. Peak areas were normalized using the total ion count.

The resulting compounds were evaluated via Student's t test (P < 0.05) analysis to identify differentially accumulated compounds between the control samples and the MYB‐targeted samples. The spectrum profile of the differential compounds is included in Table S5.

LC and untargeted analysis

The material used for the GC‐MS analysis was further analyzed by LC‐MS analysis. The first symptomatic leaf samples of five different plants infected with each PVX_VIGR vector were collected at 14 dpi. Frozen ground tissue (50 mg) was extracted in 75% acetonitrile in water (500 μl) with 1 ppm genistein as the internal standard. The homogenate was vortexed, sonicated for 10 min and centrifuged for 5 min at 24104 g. Supernatants were filtered with a 0.2‐μm filter. The analysis comprised five biological replicates per genotype.

LC‐MS analysis was performed with an Orbitrap Exploris 120 mass spectrometer coupled with a Vanquish UHPLC System (Thermo Fisher Scientific). Compounds were separated by reverse‐phase ultraperformance LC using an Acquity PREMIER BEH C18 UPLC column (1.7 μm particle size, dimensions 2.1 × 150 mm) (Waters Corp., Milford, MA, USA).

Mobile phases consisted of 0.1% formic acid in water (phase A) and 0.1% formic acid in acetonitrile (phase B). The solvent gradient program was conditioned as: 0.5% solvent B over the first 2 min, 0.5–30% solvent B over 25 min, 30–100% solvent B over 13 min, 2 min at 100% B, return to the initial 0.5% solvent B over 1 min, and conditioning at 0.5% B for 2 min. The flow rate was 0.4 ml min−1, and the injection volume was 1 μl. The column temperature was set at 40°C.

Ionization was performed with heated electrospray ionization in positive and negative modes. Samples were acquired in full scan mode (resolution set at 120 000 measured as the full width at half maximum) and mixes were acquired in both full scan and data‐dependent acquisition to help with compound identification. For data‐dependent acquisition, the resolution was set at 30 000 and the intensity threshold at 2e5. The mass range was set from 150 to 1500.

Data and statistical analyses were performed with compound discoverer, version 3.3 (Thermo Fisher Scientific) and metaboanalyst, version 5.0 (https://www.metaboanalyst.ca). Logarithmic transformation and autoscaling scaling were employed as normalization to perform the principal component analysis and the hierarchical clustering and also to produce a heatmap. Euclidean distance and Ward clustering algorithm were applied as parameters for making hierarchical clustering, and anova comprised the statistical method used for generating the list of 350 molecular features with significantly different accumulation. The molecular features determined as significantly different and tentatively identified through Kegg databases associated with compound discoverer, version 3.3, were manually added to Table 1. The quantification of the metabolites was performed employing the parental ion and normalized with the negative control samples (C–), where an unspecific non‐specific gRNA was delivered with the PVX_VIGR.

AUTHOR CONTRIBUTIONS

DO and SS designed the experiments. SS, GS, and UM conducted the experiments. GS, A‐RA, and V‐VM contributed to the data analysis. DO and SS drafted the manuscript. GS, DM, UM, FPD, and DJA contributed to writing and editing the manuscript. All of the authors discussed and revised the manuscript prior to submission.

CONFLICT OF INTEREST

The authors declare no conflict of interest.

Figure S1. Evaluation of PVX‐based sgRNA delivery for CRISPRa at 14 and 20 dpi.

Click here for additional data file.

Figure S2. Full sequence of Potato virus X, PVX::XT2 and PVX_VIGR vectors: PVX::gDFR, PVX::gODO1, PVX::gMYB21 and PVX::gMYB24.

Click here for additional data file.

Table S1. Ct values obtained from the qRT‐PCR assay for evaluating the transcriptional expression of NbODO1, NbMYB21 and NbMYB24 pPVX‐gRNA induced samples.

Table S2. Protospacer sequences designed for targeting the N. benthamiana genes with dCasEV2.1.

Table S3. Primers employed for sgRNA2.1 construction and adaptation for PVX recombinant plasmids through Gibson assembly.

Table S4. Primer pairs for RT‐qPCR analyses.

Table S5. m/z fragmentation spectrum F2 of the differential metabolites obtained in the non‐targeted analysis of the volatile content in negative control, NbODO1, NbMYB21 and NbMYB24 induced samples.

Click here for additional data file.