A pleiotropic drug resistance transporter in Nicotiana tabacum is involved in defense against the herbivore Manduca sexta


(e-mail marc.boutry@uclouvain.be).


Pleiotropic drug resistance (PDR) transporters are a group of membrane proteins belonging to the ABCG sub-family of ATP binding cassette (ABC) transporters. There is clear evidence for the involvement of plant ABC transporters in resistance to fungal and bacterial pathogens, but not in the biotic stress response to insect or herbivore attack. Here, we describe a PDR transporter, ABCG5/PDR5, from Nicotiana tabacum. GFP fusion and subcellular fractionation studies revealed that ABCG5/PDR5 is localized to the plasma membrane. Staining of transgenic plants expressing the GUS reporter gene under the control of the ABCG5/PDR5 transcription promoter and immunoblotting of wild-type plants showed that, under standard growth conditions, ABCG5/PDR5 is highly expressed in roots, stems and flowers, but is only expressed at marginal levels in leaves. Interestingly, ABCG5/PDR5 expression is induced in leaves by methyl jasmonate, wounding, pathogen infiltration, or herbivory by Manduca sexta. To address the physiological role of ABCG5/PDR5, N. tabacum plants silenced for the expression of ABCG5/PDR5 were obtained. No phenotypic modification was observed under standard conditions. However, a small increase in susceptibility to the fungus Fusarium oxysporum was observed. A stronger effect was observed in relation to herbivory: silenced plants allowed better growth and faster development of M. sexta larvae than wild-type plants, indicating an involvement of this PDR transporter in resistance to M. sexta herbivory.


ATP binding cassette (ABC) transporters form a large group of proteins in organisms from all kingdoms of life (Kos and Ford, 2009). Membrane-bound ABC transporters perform active transport of molecules across biological membranes energized by ATP hydrolysis. They consist of one or two transmembrane domains (TMDs) and one or two nucleotide-binding domains (NBDs), which contain the characteristic Walker A and Walker B motifs as well as the ABC signature motif. Depending upon the organization of these structural elements, ABC transporters are classified into different sub-families (Verrier et al., 2008). The ABCG sub-family contains half- and full-size transporters organized in the so-called reverse orientation (NBD–TMD and NBD–TMD–NBD–TMD, respectively). Full-size ABCG transporters, also called pleiotropic drug resistance (PDR) transporters, have only been identified in plants, fungi, oomycetes, brown algae and slime molds (Kang et al., 2011). In plants, PDR transporters have been shown to be involved in a variety of functions, including defense against pathogens, resistance to cadmium, lead and high salt concentrations, formation of the cuticular layer, and transport of antimicrobial terpenoids, auxin derivatives, abscisic acid and strigolactone (Jasinski et al., 2001; Lee et al., 2005; Ito and Gray, 2006; Stein et al., 2006; Kim et al., 2007, 2010; Bultreys et al., 2009; Krattinger et al., 2009; Strader and Bartel, 2009; Yazaki et al., 2009; Kang et al., 2010; Ruzicka et al., 2010; Bessire et al., 2011; Kretzschmar et al., 2012).

The involvement of PDRs in resistance to pathogens has been demonstrated comprehensively across plant species. In Nicotiana plumbaginifolia, NpPDR1 has been shown to function in the transport of diterpenes such as sclareol and in resistance to fungal pathogens (Jasinski et al., 2001; Stukkens et al., 2005; Bultreys et al., 2009). Recently it was shown that growth inhibition of Ralstonia solanacearum on Arabidopsis or tobacco plants by pre-treatment with sclareol was attenuated in AtPDR12 knockout or NtPDR1-silenced plants, respectively, suggesting that these PDR transporters function in resistance against bacterial wilt disease (Seo et al., 2012). In Arabidopsis, deletion of ABCG36/PDR8 leads to increased susceptibility to Pseudomonas syringae infections (Kobae et al., 2006), while the importance of the PDR transporter Lr34 in resistance to multiple fungal pathogens has been demonstrated in wheat (Krattinger et al., 2009). Despite several cases of PDR transporters being implicated in the resistance of various plant species to pathogens, no PDR, or other ABC transporter, has yet been shown to be involved in resistance to insects or herbivores.

Plants have developed mechanisms to differentiate biotic damage due to insects and herbivores from abiotic wounding and stresses. Detection is mediated by recognition of pathogen- or herbivore-specific mediators, such as pathogen-specific avirulence factors or fatty acid–amino acid conjugates in the oral secretions of chewing insects (Bisgrove et al., 1994; Wu et al., 2007). Plants respond in an attack-specific manner by reallocating resources, and by mounting both specific and general inducible defenses. Alterations in primary metabolites occur as in differential regulation of photosynthesis or photosynthetic genes in damaged tissue (Welter, 1989; Hui et al., 2003), and changes in source–sink relationships, which is reflected in up-regulation of sugar transporters in damaged tissues immediately following wounding, for example (Meyer et al., 2004; Schwachtje and Baldwin, 2008). Indirect defenses function to attract natural enemies of the perceived herbivore, for example production of green leaf volatile compounds to attract natural predators of herbivores (Baldwin et al., 2001; Allmann and Baldwin, 2010). Direct defenses target specialists or generalists or entire classes of herbivores, and include physical barriers, such as trichomes or waxes, and the production of inducible chemical defense compounds, such as secondary metabolites (Arimura et al., 2005).

One example of a herbivore-specific defense response in Nicotiana species is the induced response to herbivory by larvae of the nicotine-tolerant lepidopteran Manduca sexta, a Solanaceae specialist (Wu and Baldwin, 2010). These responses have been studied most extensively in the wild tobacco Nicotiana attenuata, especially initiation of changes at the transcriptomic, proteomic and metabolomic level (Wu et al., 2007). The induced defense compounds include trypsin protease inhibitors (van Dam et al., 2001; Zavala and Baldwin, 2004), anti-nutritive enzymes, such as polyphenol oxidases (Karban et al., 2000), nicotine (Steppuhn et al., 2004), diterpene glycosides (Jassbi et al., 2008; Heiling et al., 2010), and the phenylpropanoid–polyamine conjugates caffeoylputrescine and dicaffeoylspermidine (Kaur et al., 2010).

Importantly, in both laboratory and field studies, herbivore performance has been correlated with levels of inducible defense compounds. Transgenic N. attenuata plants silenced for the putrescine N-methyl transferase (pmt) gene demonstrated a whole-plant decrease in nicotine content and corresponding improved M. sexta performance in the laboratory. When planted in their native habitat, the plants showed greater susceptibility to herbivory in terms of the variety of herbivore species that attack them and the leaf area lost (Steppuhn et al., 2004).

While numerous studies have identified genes involved in the production of secondary metabolites or in the signaling cascade transmitting attack signals, little is known about the transporters involved in facilitating the translocation of defense or signaling compounds to and from the site of physical damage.

Here, we identified ABCG5/PDR5, a full-size PDR transporter from Nicotiana tabacum, expression of which is negligible in leaves under standard growth conditions, but that is induced by treatment with methyl jasmonate (MeJA), wounding, and herbivory by the specialist herbivore Msexta. Furthermore, silencing of ABCG5/PDR5 expression results in faster growth and development of M. sexta, indicating an important role for ABCG5/PDR5 in herbivore performance.


Cloning of ABCG5/PDR5

Using degenerate primers corresponding to highly conserved regions in PDR genes, and RNA from N. plumbaginifolia roots, we obtained a partial cDNA for an unknown PDR gene. As this gene, named NpABCG5/PDR5, belonged to a PDR sub-family previously uncharacterized in a Nicotiana species, we obtained a full cDNA from root RNA by RACE RT-PCR. The cDNA contains a 4497 bp open reading frame encoding a 1499 amino acid protein with a predicted size of 170 kDa. An in-frame stop codon was found upstream of the initiation codon, indicating that no N-terminal sequence was missing (Figure S1).

Homologs of NpABCG5/PDR5 from N. tabacum were also amplified, and two sequences, NtABCG5/PDR5a and NtABCG5/PDR5b, were obtained and found to share high nucleotide identity with NpABCG5/PDR5 (97.6 and 98.0%, respectively). As N. tabacum is an amphitetraploid, these sequences probably represent orthologs originating from the two ancestors, Nicotiana tomentosiformis and Nicotiana sylvestris. This was confirmed by PCR on genomic DNA extracted from N. tomentosiformis and N. sylvestris using ortholog-specific primers (Figure 1).

Figure 1.

ABCG5/PDR5 has two orthologs in N. tabacum. PCR on N. tabacum genomic DNA with specific primers for the two orthologs present in the genome (a, forward 5′-TAGACTTCGAATCAGCCAG-3′, reverse 5′-GAACTGGTACGACTGCTCA-3′; b, forward 5′-CCCAAGACTTAAAATCAGCCAGT-3′, reverse 5′-TGGATATGACTGCTCATTCGTC-3′). A non-specific region in ABCG5/PDR5 was chosen as a control (forward 5′-GGCACAACTAGATTGGGTCTG-3′; reverse 5′-CTTTGCTGCGACACCATTT-3′) (bottom panel).

Phylogenetic analysis revealed that the NtABCG5/PDR5 and NpABCG5/PDR5 sequences aligned with known PDR transporter genes in cluster III (Figure S2). The closest ABCG5/PDR5 homologs were found to be Arabidopsis AtABCG35/PDR7 and AtABCG36/PDR8, with approximately 74% amino acid identity. The deduced ABCG5/PDR5 amino acid sequence contained the typical PDR and ABC transporter motifs (Figure S3).

Expression of ABCG5/PDR5 in organs and tissues

In order to determine the expression profile of ABCG5/PDR5, an antibody was generated against a 61 amino acid sequence in a region unique to ABCG5/PDR5 (Figure S4). Immunoblotting analysis of various plant tissues showed that ABCG5/PDR5 expression was high in the petals, stem and roots, but weak in the whole leaf, although it was enriched in epidermal strips (Figure 2a).

Figure 2.

 ABCG5/PDR5 expression analysis in various plant tissues and organs.
(a) Immunoblotting of microsomal fractions of various tissues and organs of N. tabacum plants using an ABCG5/PDR5-specific antibody and a general anti-H+-ATPase antibody as the loading control.
(b–h) Analysis of GUS staining in ProPDR5:GV transgenic plants grown either hydroponically (root) or on soil (aerial tissue). (b) Petal, (c) stem, (d) leaf, (e) root with side roots, (f) cross-section of the mature part of the root, (g) root tip and elongation zone, (h) cross-section of the root tip. In (h), GUS staining was performed for a shorter time period compared to (g) to improve visualization of expression in the epidermis. Scale bars = 0.5 mm (b,c), 0.1 mm (d), 1 cm (e,g) and 50 μm (f,h).

To confirm the ABCG5/PDR5 expression profile, we cloned its transcription promoter region by inverted PCR and obtained a 1479 bp genomic fragment upstream of the translation initiation codon. The putative promoter sequence (ProPDR5) was fused to the hybrid reporter gene, GUSVENUS (GV) (Navarre et al., 2011) and introduced into N. tabacum plants by Agrobacterium tumefaciens-mediated transformation. Twenty kanamycin-resistant ProPDR5:GV plants that showed GUS expression were selected. GUS activity was consistently detected in the petals (Figure 2b), stem (Figure 2c) and roots (Figure 2e,g), and at a low level in leaves (Figure 2d). Microscopic analysis of cross-sections of roots showed staining in the outer cell layers (Figure 2f,h).

ABCG5/PDR5 is localized in the plasma membrane

In order to determine the subcellular localization of ABCG5/PDR5, ABCG5/PDR5–GFP and GFP–ABCG5/PDR5 fusion proteins were transiently expressed in N. tabacum and N. benthamiana leaves. Both fusion proteins were detected as a thin ring at the periphery of epidermal cells and co-localized with FM4-64, a marker labeling the plasma membrane, shortly after its infiltration (Figure S5). As the tonoplast is located in close proximity to the plasma membrane in epidermal cells, we used a YFP fusion to tonoplast intrinsic protein 2;1 (AtTIP2;1–YFP) located in the tonoplast to distinguish these two membranes and localize ABCG5/PDR5 more precisely. The GFP signal resulting from GFP–ABCG5/PDR5 (Figure 3a) co-localized with FM4-64 (Figure 3c) but was distinct from the YFP signal of TIP2;1–YFP (Figure 3e), demonstrating the plasma membrane localization of ABCG5/PDR5. As an additional control, we used a marker for the endoplasmic reticulum, CFP–HDEL. Although some GFP–ABCG5/PDR5 co-localized with CFP–HDEL, most likely protein that is passing through the secretory pathway to the plasma membrane, the majority of the GFP signal was distinct from CFP–HDEL (Figure 3g–i), which is in agreement with the plasma membrane localization of ABCG5/PDR5.

Figure 3.

 Subcellular localization of ABCG5/PDR5.
(a–c) Localization of (a) the GFP–PDR5 fusion protein and (b) the plasma membrane marker FM4-64 (b). (c) Merged image.
(d) Localization of AtTIP2;1–YFP as a marker for the vacuolar membrane.
(e,f) Merged images of (e) GFP–PDR5 and AtTIP2;1–YFP, and (f) FM4-64 and AtTIP2;1–YFP (f).
(g–i) Localization of (g) GFP–PDR5 and (h) CFP–HDEL as a marker of the endoplasmic reticulum. (i) Merged image.
(j) Immunoblotting of the microsomal fraction (MF) and plasma membrane-enriched fraction (PMF) from N. tabacum roots using an ABCG5/PDR5-specific antibody, a general anti-H+-ATPase antibody and an anti-Nt-TIPa antibody. Scale bars = 10 μm (a–f) and 6 μm (g–i).

To confirm the plasma membrane localization of ABCG5/PDR5, a plasma membrane-enriched fraction was isolated from N. tabacum roots and subjected to immunoblotting, using an antibody recognizing the H+-ATPase (Morsomme et al., 1998) as a positive control for the plasma membrane and an antibody recognizing the tonoplast intrinsic protein Nt-TIPa (Gerbeau et al., 1999) as a negative control. As shown in Figure 3(j), ABCG5/PDR5 was enriched in the plasma membrane fraction like the H+-ATPase and unlike Nt-TIPa, confirming its plasma membrane localization.

ABCG5/PDR5 expression is induced by methyl jasmonate

The sequence of the ABCG5/PDR5 promoter region revealed the presence of a T/G box motif (5′-AACGTG-3′) at position -456 bp (Figure S1), which has been shown to be involved in the MeJA response in Arabidopsis (Guerineau et al., 2003; Boter et al., 2004). In order to study the effect of hormone treatment on ABCG5/PDR5 expression, we submerged leaves of wild-type N. tabacum plants for 16 h in water or 500 μm MeJA. Non-submerged leaves from intact plants were used as a control. As shown in Figure 4(a), the abundance of ABCG5/PDR5 transcript resulting from MeJA-submerged leaves was increased compared to the water- and non-treated samples. The plant tissues used in Figure 4(a) were fractionated and used to extract RNA from which cDNA was synthesized and to prepare microsomal fractions to analyze protein expression. ABCG5/PDR5 expression was strongly induced by the presence of MeJA (Figure 4b). Interestingly, submersion in water increased ABCG5/PDR5 protein expression (Figure 4b), but this difference was not seen at the RNA level (Figure 4a). This may indicate that ABCG5/PDR5 expression is regulated at the post-transcriptional level. To confirm the expressional regulation of ABCG5/PDR5, GUS activity was measured on soluble protein extracts of leaves from ProPDR5:GV plants (Figure 4c). In non-treated leaves, GUS activity was low but increased significantly when the leaves were submerged in water, and even more so in the presence of MeJA, with GUS activity after MeJA treatment being significantly higher than in the non-treated and water-treated samples.

Figure 4.

 Induction of ABCG5/PDR5 expression by MeJA.
(a) Semi-quantitative PCR on cDNA of N. tabacum leaves taken directly from the plant (Fresh) or after submersion for 16 h in water + methanol (Water) or 500 μm methyl jasmonate (MeJA). Primers specific for ABCG5/PDR5 were used to follow ABCG5/PDR5 transcript abundance (forward 5′-GGCACAACTAGATTGGGTCTG-3′; reverse 5′-CTTTGCTGCGACACCATTT-3′). NtATP2, encoding the beta-subunit of mitochondrial ATP synthase (Boutry and Chua, 1985), was used as a loading control (forward primer 5′-TCTTTGCTGGTGTTGGTGAA-3′; reverse primer 5′-TGAGCTCATCCATACCCAAA-3′).
(b) Microsomal fractions of the same material as used in (a) were probed using an ABCG5/PDR5-specific antibody and a general anti-H+-ATPase antibody as a loading control.
(c) Leaves of ProPDR5:GV transgenic plants were either directly taken from the plant (Fresh) or submerged in water + methanol (Water) or 500 μm MeJA (MeJA) for 16 h. GUS activity was measured as pmol 4-MU (methyl umbellifoerone) per μg protein per minute using a fluorimeter. Standard errors are shown. Asterisks indicate statistically significant differences compared with wild-type (*< 0.05, **< 0.001, Student’s t-test).

ABCG5/PDR5 expression is induced upon wounding

MeJA is involved in a number of physiological processes, including responses to pathogen attack and wounding (Creelman and Mullet, 1997). As stated earlier, the promoter region of ABCG5/PDR5 contains a T/G box motif that has not only been shown to be responsive to MeJA, but is also found in promoters of genes whose expression is induced upon wounding (Delessert et al., 2004). This prompted us to examine the responsiveness of ABCG5/PDR5 expression to wounding. N. tabacum plants were wounded either by forceps pressure without rupturing the leaf or by perforating the leaf (Figure S6), and microsomal fractions were prepared 16 h later. Control samples were taken from a non-injured area of the wounded leaf and from systemic leaves. As shown in Figure 5, both methods of wounding resulted in a marked increase in ABCG5/PDR5 expression in leaves, whereas expression remained marginal in a non-injured area of the wounded leaf and in systemic leaves. In order to study induction of ABCG5/PDR5 expression over time and to determine whether it remained local or became systemic, we isolated microsomal fractions of wounded leaf areas at 1–72 h after wounding, and from non-wounded areas of the wounded leaf and of systemic leaves at the same time. As shown in Figure 5(b,c), induction of ABCG5/PDR5 expression was observed at 7 h after wounding in the wounded area, and remained high to at least 72 h after wounding, whereas ABCG5/PDR5 expression did not change over time in the non-injured areas of the wounded leaf or systemic leaf. Thus, induction of ABCG5/PDR5 expression is restricted to the wounded area, implying that the protein plays a major role close to the region of infection or physical damage.

Figure 5.

 Induction of ABCG5/PDR5 expression by wounding.
(a) Microsomal fractions of N. tabacum leaves wounded by forceps pressure (Wounded) or non-wounded tissue on the same leaf (Same leaf), and of a systemic leaf (Systemic) were probed using an ABCG5/PDR5-specific antibody and a general anti-H+-ATPase antibody as a loading control.
(b,c) Immunoblotting of leaf microsomal fractions taken from wounded tissue (Wounded) or tissue from a non-wounded area on the same leaf (Same leaf) (b), or from a region on the leaf that was perforated with a metal fabric pattern wheel (Perforated) and a systemic leaf (Systemic) (c) at various times after wounding using the same antibodies as in (a).

ABCG5/PDR5 expression is induced upon pathogen infiltration

Expression of a number of PDR transporters has been shown to be induced following pathogen infiltration or infection (Stukkens et al., 2005; Ruocco et al., 2011). We therefore investigated ABCG5/PDR5 expression following infiltration of leaves with Fusarium oxysporum, Rhizoctonia solani, Phytophthora nicotianae or Botrytis cinerea. Immunoblotting showed that ABCG5/PDR5 expression was strongly induced by P. nicotianae, while F. oxysporum and R. solani had a smaller effect, and B. cinerea even less (Figure 6).

Figure 6.

 ABCG5/PDR5 is induced upon pathogen infiltration. P. nicotianae (Pn), R. solani (Rs), F. oxysporum (Fo) or B. cinerea (Bc) were infiltrated into N. tabacum leaves using a syringe. Expression of ABCG5/PDR5 was followed by immunoblotting after 48 h (Pn, Rs and Fo) or 96 h (Bc) of infection using an ABCG5/PDR5-specific antibody and a general anti-H+-ATPase antibody as a loading control.

Silencing of ABCG5/PDR5

To determine the physiological role of ABCG5/PDR5, we restricted ABCG5/PDR5 expression by expressing an artificial micro RNA (amiRNA) (Schwab et al., 2006) directed against a specific region in ABCG5/PDR5. Two amiRNA constructs, driven by the constitutive En2pPMA4 promoter (De Muynck et al., 2009), were introduced into N. tabacum, and efficiently silenced lines were selected by immunoblotting of root tissues. Only one of the amiRNA constructs, amiRNA3, resulted in efficient silencing of ABCG5/PDR5 protein expression. Fourteen independent silenced plant lines were obtained. Figure 7(a) shows the results for the two lines, pdr5-1 and pdr5-2, that were chosen for further studies. The efficiency of silencing was checked regularly. In addition, we investigated whether silencing of ABCG5/PDR5 affected the protein expression of PDR1, another PDR gene expressed in roots. The PDR1 expression level was comparable in pdr5-1, pdr5-2 and wild-type plants (Figure S7).

Figure 7.

 Silencing of ABCG5/PDR5 slightly increases susceptibility of N. tabacum to F. oxysporum.
(a) ABCG5/PDR5 expression was silenced using an amiRNA construct directed against a specific region in ABCG5/PDR5 (see Experimental procedures). Microsomal fractions of hydroponically grown N. tabacum roots were isolated and subjected to immunoblotting using an ABCG5/PDR5-specific antibody and a general anti-H+-ATPase antibody as a loading control. Two independent lines, pdr5-1 and pdr5-2, showing efficient silencing, were selected.
(b) Mortality of ABCG5/PDR5- and NpPDR1-silenced plants in N. tabacum 26 days after F. oxysporum infection.
(c) Difference in leaf width of wild-type and ABCG5/PDR5-silenced plants 20 days after F. oxysporum infection. Standard deviations are shown. Asterisks indicate statistically significant differences compared with wild-type [*< 0.05, Tukey’s (HSD) test].

Silencing of ABCG5/PDR5 slightly increases the susceptibility of N. tabacum to F. oxysporum

As expression of ABCG5/PDR5 is induced upon pathogen infiltration, we investigated its role in plant defense and compared it to that of NpPDR1. NpPDR1-silenced lines of N. plumbaginifolia were more sensitive than wild-type to F. oxysporum, R. solani, P. nicotianae and B. cinerea in a root susceptibility test; total death of the inoculated plants was usually observed (Bultreys et al., 2009). We obtained an NpPDR1-silenced line of N. tabacum (Figure S8), and investigated whether the same effects were observed. In this case, total mortality was observed with F. oxysporum (Figure 7b).

Comparison of susceptibility of the ABCG5/PDR5- and NpPDR1-silenced plants against F. oxysporum indicated that silencing of ABCG5/PDR5 had a lower influence on plant resistance than silencing of NpPDR1. Whereas all the NpPDR1-silenced plants were killed, only one plant of 24 was killed in one ABCG5/PDR5-silenced line, and none in the other silenced line or the wild-type (Figure 7b).

As silencing of ABCG5/PDR5 does not show major effects on the susceptibility of N. tabacum plants towards F. oxysporum, we developed a test to detect weak effects in plant defense that consisted of measuring the leaf width 20 days after inoculation (Figure 7c). Under these conditions, ABCG5/PDR5-silenced plants displayed a low but significant increase in susceptibility, indicating a role of ABCG5/PDR5 in plant defense.

ABCG5/PDR5 expression is induced by the specialist herbivore Manduca sexta

As expression of ABCG5/PDR5 is induced after wounding and treatment with MeJA, we examined whether the presence of insects also has an effect on ABCG5/PDR5 expression in leaves. M. sexta is used as a model organism for studying plant–herbivore interactions (Steppuhn et al., 2004). Oral secretions from larvae are commonly used as an elicitor to mimic insect feeding on plants (Alborn et al., 1997, 2003). In order to be able to distinguish ABCG5/PDR5 expression induced by wounding as a result of chewing from the induction as a specific response to M. sexta larva oral secretions, we infiltrated leaves with water or oral secretions through the stomata, prepared a microsomal fraction 16 h after infiltration, and performed immunoblotting. As shown in Figure 8(a), the presence of oral secretions significantly increased ABCG5/PDR5 expression in leaves, suggesting its involvement in the response to herbivory.

Figure 8.

 Induction of ABCG5/PDR5 expression by Manduca sexta oral secretions and Manduca sexta feeding.
(a) Immunoblotting of microsomal fractions of leaves infiltrated with water (Water) or an equal volume of water and M. sexta oral secretions (OS) through the stomata using a syringe, or a non-treated area of the same leaf (Same leaf). The samples were probed using an ABCG5/PDR5-specific antibody or a general anti-H+-ATPase antibody as a loading control.
(b) Immunoblotting of leaf microsomal fractions taken from the area surrounding the chewed part of the leaf after browsing by M. sexta for 24 h (Herbivored) and a non-chewed area of the same leaf as a control (Non-herbivored) using the same antibodies as in (a).
(c,d) Leaf in situ GUS activity of plants carrying the ProPDR5:GV fusion construct that were not subjected to herbivory (c) and after M. sexta herbivory for 24 h (d). Arrows indicate the non-herbivored (c) and herbivored (d) leaf areas. Scale bars = 0.5 mm.

To examine whether ABCG5/PDR5 expression was induced after caterpillar feeding, we placed 4-day-old caterpillars on wild-type N. tabacum plants for 24 h, and performed immunoblotting on microsomal fractions from chewed areas and non-chewed areas of the same leaf. ABCG5/PDR5 expression was induced in the area around the feeding site but not at a distant site on the same leaf (Figure 8b).

Furthermore, ProPDR5:GV plants were used to visualize ABCG5/PDR5 promoter activity following herbivory. Four-day-old caterpillars were placed on leaves of transgenic N. tabacum plants for 24 h, then samples were taken from the feeding site and a non-injured area for GUS staining. GUS staining was hardly visible in the control samples (Figure 8c), but was intense at the site of herbivory (Figure 8d), confirming local induction of ABCG5/PDR5 expression upon caterpillar feeding.

Silencing of ABCG5/PDR5 favors herbivore performance

The performance of M. sexta larvae on ABCG5/PDR5-silenced plants was compared to that on wild-type plants in the glasshouse by weighing the larvae after 3, 6, 9 and 12 days of feeding. Larvae grown on the ABCG5/PDR5-silenced lines, pdr5-1 and pdr5-2, weighed significantly more than those grown on wild-type plants, with a weight difference of 31% between caterpillars grown on ABCG5/PDR5-silenced plants compared to wild-type after 12 days (Figure 9a). The increased mass gain correlated with more rapid development of the larvae grown on the silenced lines over the observation period (Figure 9b). For instance, on day 12, 19 and 27% of the caterpillars on the pdr5-1 and pdr5-2 transgenic plants, respectively, had reached stage 5, but none of the caterpillars on wild-type plants had reached this molting stage.

Figure 9.

 Silencing of ABCG5/PDR5 expression improves Manduca sexta growth and development.
(a) Mean larval mass of M. sexta caterpillars fed on ABCG5/PDR5-silenced lines and wild-type plants for 3, 6, 9 or 12 days after hatching. Three caterpillars were placed on each plant (n = 15), and the mean was determined (first the mean of 3 caterpillars per plant was taken and then the mean of 15 plants per line); the bars represent the 95% confidence interval. Asterisks indicate statistically significant differences compared with wild-type (***< 0.001, **P < 0.01, *< 0.05, Student’s t-test).
(b) Development of M. sexta on ABCG5/PDR5-silenced lines and wild-type plants during 12 days after hatching.

Metabolite analysis

We isolated secondary metabolites from perforated leaf tissue that had been treated with M. sexta oral secretions and from healthy plant tissue to determine whether the faster development and decreased feeding restriction of M. sexta larvae on ABCG5/PDR5-silenced lines were due to a difference in the abundance of major anti-herbivore compounds. We used M. sexta oral secretions to mimic the presence of the caterpillar, as they contain chemicals that trigger the plant defense mechanism, including production of secondary metabolites that restrict caterpillar feeding. However, although a lower content of chlorogenic acid was observed in the two silenced lines compared to wild-type, neither the content of this metabolite nor that of nicotine, dicaffeoyl spermidine or rutin was significantly different from wild-type (Figure S9).

Alternatively, ABCG5/PDR5 may be involved in the plant-wide reallocation of carbohydrates from damaged tissue to other tissues in the plant (Meyer et al., 2004; Schwachtje and Baldwin, 2008). Concentrations of glucose, sucrose and fructose moieties were measured but did not vary significantly between wild-type and ABCG5/PDR5-silenced lines (Figure S10).


In this study, we cloned a full-length ABC transporter sequence, ABCG5/PDR5, that clusters in sub-group III of the PDR transporter sub-family (Figure S1). Using ProPDR5:GUS analysis and immunoblotting, we showed that, under standard growth conditions, ABCG5/PDR5 expression was constitutively high in roots, stems and flowers, but marginal in leaves (Figure 2). In agreement with the presence of a T/G box motif in the promoter region of ABCG5/PDR5, which has been shown to be involved in MeJA and wounding responses (Guerineau et al., 2003; Boter et al., 2004; Delessert et al., 2004), ABCG5/PDR5 expression was induced in leaves upon treatment with MeJA and wounding (Figures 4 and 5). Exogenous application of MeJA causes induction of various plant defense-related proteins (Repka et al., 2001), which suggests an involvement of ABCG5/PDR5 in the response to biotic stress.

The plant biotic stress response involves several defense mechanisms that allow the plant to cope with the challenges of the living environment. Accumulation of plant secondary metabolites and inhibitory proteins restricts pathogen invasion and insect feeding. Increasing evidence suggests that full-size ABC transporters of the ABCG/PDR sub-family mediate the secretion of plant defense compounds into the apoplast, onto the plant surface, or into the rhizosphere (Kretzschmar et al., 2011). N. plumbaginifolia plants that are silenced for the expression of NpPDR1 show spontaneous infections and are highly susceptible to controlled fungal infections (Stukkens et al., 2005; Bultreys et al., 2009). Similarly, PDR transporters from Arabidopsis, AtABCG36/AtPDR8, and wheat, LR34, were found to function in pathogen resistance (Kobae et al., 2006; Stein et al., 2006; Krattinger et al., 2009). NpPDR1 is thought to transport the anti-microbial secondary metabolite sclareol onto the leaf surface to restrict pathogen proliferation. The substrates of AtABCG36/AtPDR8 and LR34 in the context of pathogen defense are not known.

On the basis of these reports, we investigated the expression of ABCG5/PDR5 in the presence of fungal pathogens, and showed that expression was induced after infiltration with fungal pathogens (Figure 6). However, when ABCG5/PDR5-silenced lines of N. tabacum were inoculated with various fungal pathogens, including F. oxysporum, there was only a marginal difference in sensitivity compared to infected wild-type plants (Figure 7b,c). In contrast, N. tabacum plants silenced for NpPDR1 were heavily infected in the presence of F. oxysporum, comparable to the infections observed on N. plumbaginifolia plants silenced for the same gene (Bultreys et al., 2009). This may indicate that ABCG5/PDR5 does not play a prominent role in defense against F. oxysporum, but it is also possible that its role is masked by the presence of other important players in the defense mechanism that compensate for the decrease in ABCG5/PDR5 protein levels.

Recently, it was suggested that ABCG/PDR transporters may play a role in defense against herbivores, as they are involved in the secondary metabolite-based pathogen response (Kretzschmar et al., 2011), but there is currently no experimental evidence for this. Here, we showed that ABCG5/PDR5 expression was induced locally in leaves by M. sexta oral secretions and herbivory, indicating a role in the defense response (Figure 8). Expression was seen in close proximity to the physical damage and not in untreated areas of the same leaf. Interestingly, silencing of ABCG5/PDR5 rendered N. tabacum plants more susceptible to feeding by M. sexta (Figure 9a), suggesting that this diet was more appealing than wild-type leaves. In addition, M. sexta larvae developed faster on ABCG5/PDR5-silenced lines compared to those fed on wild-type plants (Figure 9b). These data provide experimental evidence for the involvement of a PDR transporter in the response to herbivory.

Damage by herbivory induces the production of compounds that limit feeding and protect against further pathogen invasions (Delessert et al., 2004). We analyzed the major anti-herbivore compounds nicotine, chlorogenic acid, dicaffeoyl spermidine and rutin, but did not observe any significant difference between wild-type and ABCG5/PDR5-silenced plants (Figure S9). At this stage, we cannot exclude the possibility that another metabolite important in herbivore repellence is affected in the ABCG5/PDR5-silenced lines. A thorough study of a wide range of secondary metabolites may answer this question.

In addition, plants respond to herbivore attacks by reallocating carbohydrate resources from the damaged tissue (Meyer et al., 2004; Schwachtje and Baldwin, 2008). ABCG5/PDR5 may be involved in the transport of carbohydrates from damaged tissue to other tissues in the plant. As we did not observe any significant difference in the concentration of glucose, sucrose or fructose in the ABCG5/PDR5-silenced plants compared to wild-type (Figure S10), the improved caterpillar performance cannot be attributed to altered carbohydrate transport by ABCG5/PDR5 within damaged tissue.

ABCG5/PDR5 expression was induced in N. tabacum leaves following physical damage, and was maintained for at least 72 h after wounding (Figure 5b,c). This may indicate that the protein is involved in a defense response in tissues on which the caterpillar fed earlier and moved away from. In this case, the importance of ABCG5/PDR5 may not be in limiting actual feeding, but in limiting continuous feeding on injured tissue or in wound healing.

Attack by insects triggers the expression of genes involved in herbivore deterrence and also in wound healing and other defense-related processes (Karban and Baldwin, 1997). Whereas systemically induced genes are thought to play a role in driving the defense mechanism to prevent further herbivore or pathogen attacks, locally induced genes are predicted to function in wound healing and repair, and in protecting the wound from water loss and further attacks (Delessert et al., 2004). Abscisic acid (ABA), a hormone that is involved in the response to water stress, accumulates in close proximity to the wound site and is thought to play a role in maintaining the physiological state of the plant (Birkenmeier and Ryan, 1998). Several PDR transporters have been shown to either directly transport ABA or be part of the ABA signaling cascade (Kanai et al., 2010; Kang et al., 2010; Kuromori et al., 2010, 2011). The possible involvement of ABC transporters in ABA transport or accumulation after wounding has not been investigated, but represents an interesting field for further investigation.

Wound healing also involves the accumulation of waxes and repair of the cuticle at the wound site. The involvement of PDR transporters in formation of the cuticle layer has been reported recently. Disruption of ABCG32/PEC1 in Arabidopsis has a significant impact on the formation of the cuticle (Bessire et al., 2011). Furthermore, HvABCG31 has been reported to be important in leaf water retention due to a physiological function in cutin deposition (Chen et al., 2011). The cuticle acts as the first barrier against pathogens and herbivores, meaning that resistance starts here. Little is known about the role of the plant cuticle in the induced defense against herbivores. However, it is possible that induced defense mechanisms act by production of waxes or biosynthesis of secondary metabolites that are deposited on the plant cuticle after herbivore attack and serve as a deterrent to, or are toxic for, the attacker (Müller, 2008). The potential involvement of ABCG5/PDR5 in fatty acid transport or the deposition of secondary metabolites onto the wounded leaf area remains to be investigated.

In addition to its induction by herbivores in leaf tissues, ABCG5/PDR5 expression is constitutively found in other tissues, including roots and stems. In roots, the transporter may function in the extrusion of secondary metabolites into the rhizosphere to defeat soil-borne pathogens or communicate with the microbial community. In stems, the function may consist in defeating pathogens such as fungi that move along the vascular tissue from the roots to the aerial part of the plant. The substrate of ABCG5/PDR5 in these tissues may be similar or very different from that in leaves. Whereas some ABCB transporters have been shown to have high substrate specificity (Knöller et al., 2010; Bailly et al., 2011), other ABC transporters are able to transport various substrates with unrelated structures. For instance, AtABCG36/AtPDR8 is thought to extrude cadmium (Kim et al., 2007) and indole-3-butyric acid (Strader and Bartel, 2009) from roots. In which form these molecules are transported by AtABCG36/AtPDR8 is unknown, but this may indicate that this PDR transporter has a broad substrate spectrum. For the mouse P-glycoprotein, an ABC transporter for which a 3D structure has been elucidated, this pleiotropy is explained by multiple binding sites in a large internal cavity that may accommodate several substrates (Aller et al., 2009). Addressing the question as to whether ABCG5/PDR5 also has multiple substrates and functions will require further work.

In conclusion, we present evidence that silencing of ABCG5/PDR5 renders N. tabacum plants more susceptible to M. sexta feeding. ABCG5/PDR5 expression is highly responsive to wounding and herbivory, and is induced only in close proximity to the site of physical damage. The observation that PDR transporters are involved in the defense response to herbivory opens up a new dimension in terms of the importance of this protein family in plants.

Experimental procedures

Plant growth

Nicotiana tabacum cv. Petit Havana SR1 (Maliga et al., 1973), Nicotiana tomentosiformis and Nicotiana sylvestris (Institut du Tabac, Bergerac, France) were grown on soil in a climate chamber under 12 h light (200 μmol photons m−2 sec−1) at 25°C and 12 h dark at 23°C.

For in vitro growth, seeds were sterilized for 20 min in 50% v/v commercial bleach. Seeds were vernalized at 4°C for 48 h prior to germination.

Selection of transgenic plants was performed on Murashige and Skoog medium [4.4 g l−1 MS salts (MP Biomedicals, http://www.mpbio.com/), pH 5.8 (KOH), 3% sucrose, 0.9% agar] supplemented with 100 μm kanamycin. In vitro cultures were grown at 25°C under a photoperiod of 16 h light (50 μmol photons m−2 sec−1)/8 h dark. T2 generations were used in all experiments described.

For analyses of root material, plants were grown in in vitro culture on MS medium, then transferred to hydroponic solution (1.25 mm Ca(NO)3, 1.25 mm MgSO4, 1.25 mm KNO3, 0.25 mm KH2PO4, 25 μm Fe(III)Na2EDTA, 12.5 μm H3BO3, 11.25 μm MnCl2, 9.5 μm ZnSO4, 0.75 μm CuSO4, 0.25 μm (NH4)6Mo7O24).

Gene cloning

The full-length NpABCG5/PDR5 sequence (accession number JQ808000) was obtained from root cDNA of N. plumbaginifolia by RACE PCR using degenerate primers. RNA was prepared using the Trizol method (Sigma, http://sigmaaldrich.com/) followed by DNase treatment (Roche, http://www.roche.com/) prior to cDNA synthesis. Full-length NpABCG5/PDR5 was amplified using forward primer 5′-ATGGAGGCAAATGGGGGACCG-3′ and reverse primer 5′-CTATCTAGTTTGGAAGTTCAATGTCTTGATAGC-3′, and cloned into the pGEM-T easy cloning vector (Promega, http://www.promega.com/).

NtABCG5/PDR5 was amplified from root cDNA of N. tabacum using a forward primer designed according to the homology of NpABCG5/PDR5 with an EST clone on the Gene Index Project website (http://compbio.dfci.harvard.edu/tgi/plant.html) (5′-GTGAAAATGGAGGGAAATG-3′) and the NpABCG5/PDR5 reverse primer (5′-CTATCTAGTTTGGAAGTTCAATGTC-3′). Two sequences corresponding to NtABCG5/PDR5 [accession numbers JQ808002 (a) and JQ808003 (b)] were identified.

ABCG5/PDR5 promoter–GUSVENUS fusion

To clone the NpABCG5/PDR5 promoter region (accession number JQ808001), 10 μg of N. plumbaginifolia genomic DNA was digested with several restriction enzymes and purified, then the linearized DNA was self-ligated. Inverted PCR reactions were performed on circularized DNA using specific primers. A fragment of 1482 bp upstream of the start codon was obtained and amplified using the forward primer 5′-GTACAGAGGATAATGTTATTCAA-3′ and reverse primer 5′-AAGCTTTCACAATAATCTCTTAT-3′. The NpABCG5/PDR5 promoter was inserted in front of the GUSVENUS (GV) coding sequence in pAUX3131 (Navarre et al., 2011) using NotI and HindIII sites, then the fusion construct was excised using I-SceI and inserted into the pPZP-RCS2-nptII plant expression vector (Goderis et al., 2002). The constructs were introduced into Agrobacterium tumefaciens (LBA4404 virGN54D; van der Fits et al., 2000) for subsequent N. tabacum leaf disk transformation (Horsch et al., 1986).

GUS activity

Various N. tabacum plant tissues were fixed and stained for GUS activity as described by Moriau et al. (1999). For GUS quantification, microsomal membrane fractions were isolated as described below. The supernatant containing the soluble proteins was used for an activity assay performed as described by Navarre et al. (2011).

ABCG5/PDR5–GFP fusions and plasma membrane localization

NpABCG5/PDR5 was cloned into pAUX3131 between the tNOS terminator and the constitutive NpPMA4 promoter fused to two copies of the same enhancer (En2pPMA4) (De Muynck et al., 2009). GFP (mgfp4) (Haseloff et al., 1997) was amplified by PCR using forward primer 5′-AATCCGCGGAAAATGAGTAAAGGAGAA-3′ and reverse primer 5′-CTTCCGCGGCCACCTCCTCCTTTGTATAGTTCATCCAT-3′, and fused at the N-terminus of the NpABCG5/PDR5 coding sequence at the SacI site. For the C-terminal fusion of GFP, the stop codon of ABCG5/PDR5 was removed by PCR using forward primer 5′-TGACTCGAGTTCTTGAAACTTGT-3′ and reverse primer 5′-TTTCCTAGGTCTAGTTTGGAAGTTCA-3′, and the original sequence was replaced using XhoI and AvrII (NEB, http://www.neb-online.de/). The GFP coding sequence was ligated in-frame with the ABCG5/PDR5 coding sequence after amplification with forward primer 5′-AAACTCGAGACCTAGGGGAGGAGGTGGCGGAATGAGTAAAGGAGA-3′ and reverse primer 5′-TTGGATCCTTATTTGTATAGTTCATCC-3′ using XhoI and BamHI.

The fusion constructs were excised from pAUX3131 using I-SceI and inserted into pPZP-RCS2-nptII. Markers for the tonoplast (AtTIP2;1–YFP) or the ER (CFP–HDEL) were co-expressed with GFP–PDR5. After transformation of A. tumefaciens, the constructs were transiently expressed in N. tabacum or N. benthamiana leaves, and fluorescence was examined after 3 days of expression using a LSM 710 confocal microscope (Zeiss, http://microscopy.zeiss.com/microscopy/). FM4-64 (50 μm; Invitrogen, http://www.invitrogen.com/), a plasma membrane marker, was infiltrated into the transformed leaves shortly before examination. GFP, CFP and YFP were excited using the 488, 405 and 514 nm laser lines of an argon multi-laser, respectively. The emitted light was detected at 419–515 nm (GFP) and 519–566 nm (YFP) for GFP–PDR/TIP2;1–YFP co-localization, and 494–550 nm (GFP) and 434–495 nm (CFP) for GFP–PDR/CFP–HDEL co-localization. FM4-64 was excited using the 514 nm line of an argon multi-laser and the emitted light was detected at 640–758 nm.

Isolation of the microsomal fraction

All steps were performed at 4°C. Plant tissues (approximately 150 mg) were homogenized in 800 μl of 250 mm sorbitol, 60 mm Tris, 2 mm Na2EDTA, 10 mm dithiothreitol, 0.7% polyvinylpolypyrrolidone (Polyclar AT; Serva, http://www.serva.de/), 1 mm phenylmethanesulfonyl fluoride and a peptidase inhibitor cocktail (1 μg ml−1 each of leupeptin, pepstatin, aprotinin, antipain and chymostatin) at pH 8 (HCl), and the homogenate was centrifuged at 2700 g for 10 min. The supernatant was then centrifuged for 10 min at 3800 g and at 100 000 g for 15 min. The final pellet was suspended in 5 mm KH2PO4, 330 mm sucrose, 3 mm KCl, 10 mm phenylmethanesulfonyl fluoride and 1 μg ml−1 peptidase inhibitor cocktail, pH 7.8. Protein was quantified using the Bradford method (Bradford et al., 1976).


Microsomal fractions (20 μg) were separated by SDS–PAGE (8% or 12% polyacrylamide) and electroblotted onto poly(vinylidene difluoride) membranes (Millipore, http://www.millipore.com/). A rabbit ABCG5/PDR5-specific antibody was generated against the peptide QASDMEADQEESTGSPRLRISQSKRDDLPRSLSAADGNKTREMEIRRMSSRTSSS-GFYRNE. Antibodies were diluted as follows: anti-ABCG5/PDR5 antibody, 1:1000; anti-Nt-TIPa antibody (Gerbeau et al., 1999), 1:1000; anti-PDR1 antibody, 1:1000; anti-H+-ATPase antibody (Morsomme et al., 1998), 1:500 000; horseradish peroxidase-labeled goat anti-rabbit IgG antibodies (Biognost, http://www.biognost.be/), 1:10 000.

Plasma membrane purification

All steps were performed at 4°C. Hydroponically grown N. tabacum roots (25 g) were ground in 250 mm sorbitol, 60 mm Tris, 2 mm Na2EDTA, 0.6% polyvinylpolypyrrolidone, 20 mm dithiothreitol, 1 μg ml−1 of peptidase inhibitor cocktail and 1 mm phenylmethanesulfonyl fluoride in a blender. The homogenate was filtered through four layers of Miracloth (Calbiochem, http://www.calbiochem.com/), and centrifuged for 5 min at 10 000 g, then the supernatant was centrifuged for 30 min at 100 000 g. The resultant microsomal fraction was homogenized in 700 μl of 5 mm KH2PO4, 330 mm sucrose, 3 mm KCl at pH 7.8 (KOH). Plasma membranes were purified by partitioning in an aqueous two-phase system as described by Larsson et al. (1987).

Silencing of NtABCG5/PDR5

Silencing of NtABCG5/PDR5 in N. tabacum was performed using the artificial microRNA technique (Schwab et al., 2006). Two constructs were created as described at http://wmd3.weigelworld.org/. The hairpin construct was amplified using the primers described in Figure S11, and cloned downstream of the En2ProPMA4 promoter in pAUX3131 using KpnI and SacI. The expression cassette was then excised using I-SceI and inserted into pPZP-RCS2-nptII, and the constructs were transformed into A. tumefaciens (LBA4404) for subsequent N. tabacum leaf disk transformation (Horsch et al., 1986).

Pathogen susceptibility assay

An NpPDR1-silenced line of N. tabacum was obtained by introducing the small hairpin construct, previously used for N. plumbaginifolia, by A. tumefaciens-mediated transformation (Stukkens et al., 2005). Efficient silencing was confirmed by immunoblotting (Figure S5).

Wild-type and NpPDR1-silenced lines were compared in a root sensitivity test based on plant mortality in response to infection with the pathogen F. oxysporum (PB1). F. oxysporum PB1 was isolated in Belgium from a naturally infected N. plumbaginifolia NpPDR1-silenced plant; it was identified at the species level on the basis of morphological characteristics. Susceptibility toward F. oxysporum of NpPDR1- and ABCG5/PDR5-silenced lines was compared as described previously (Bultreys et al., 2009).

Differences in plant resistance between wild-type and ABCG5/PDR5-silenced lines were measured on homogeneous plantlets grown in glasshouses and maintained with sterilized water. Growth measurements (leaf width) were performed on the widest new leaf that appeared after 20 days of infection. The leaf width of each inoculated plant was subtracted from the mean leaf width of the non-inoculated plants.

Secondary metabolite analysis

Samples for secondary metabolite analysis were obtained from non-herbivorized stem leaves as a control, or from leaves that were wounded using a metal fabric pattern wheel followed by immediate application of M. sexta oral secretions. Samples were collected 72 h after elicitation. Leaf tissue samples were flash-frozen in liquid nitrogen and stored at −80°C until extraction. The method for secondary metabolite extraction was adapted from Keinänen et al. (2001), and the method for analysis by HPLC was adapted from Jassbi et al. (2008) and Kaur et al. (2010).

Sugar analysis

Frozen leaf extracts were homogenized, and 100 mg tissue was suspended in 500 μl of 80% v/v ethanol/water; the mixture was incubated for 10 min at 78°C. Samples were centrifuged at 20 800 g for 5 min at 4°C. The resulting supernatant was transferred and the pellet was re-dissolved in 500 μl of 50% v/v ethanol, and sugars were extracted as above. Both supernatants were pooled for final analysis. The concentrations of the soluble sugars sucrose, fructose and glucose were determined in this order by the coupled enzymatic assay described by Jones et al. (1997), using a 96-well microplate photometer (Infinite Pro 200; Tecan, http://www.tecan.com/).

M. sexta growth and development assay

Manduca sexta eggs from in-house reared populations were kept in a growth chamber (Snijders Scientific, http://www.snijders-scientific.nl/) at 26°C for 16 h light and at 24°C for 8 h of darkness until the larvae hatched. Three freshly hatched M. sexta neonates per plant were placed on separate leaves of 15 replicate elongating N. tabacum plants, and larval mass was measured after 3, 6, 9 and 12 days of herbivory.


This work was supported by the Interuniversity Poles of Attraction Program (Belgian State, Scientific, Technical, and Cultural Services), the Belgian National Fund for Scientific Research, the Walloon Agricultural Research Centre, and grants D31-1132/S2 and D31-1174/S2 from the Ministry of the Walloon Region, Head Office of Agriculture. The authors thank Joseph Nader, Sébastien Lievyns, Abdelmounaim Errachid and Nicolas Noël for technical help, Hagen Reinhardt and Adrien Chevalier for providing the TIP–YFP and CFP–HDEL constructs, respectively, Danny Kessler for supplying the Manduca sexta larvae, and Viviane Planchon (Department of Agriculture and Natural Environment, Centre Wallon de Recherches Agronomiques) for help with development and statistical analysis of fungal pathogenicity tests.