The transmission of insect-vectored diseases entails complex interactions among pathogens, hosts and vectors. Chemistry plays a key role in these interactions; yet, little work has addressed the chemical ecology of insect-vectored diseases, especially in plant pathosystems. Recently, we documented effects of Cucumber mosaic virus (CMV) on the phenotype of its host (Cucurbita pepo) that influence plant-aphid interactions and appear conducive to the non-persistent transmission of this virus. CMV reduces host-plant quality for aphids, causing rapid vector dispersal. Nevertheless, aphids are attracted to the elevated volatile emissions of CMV-infected plants. Here, we show that CMV infection (1) disrupts levels of carbohydrates and amino acids in leaf tissue (where aphids initially probe plants and acquire virions) and in the phloem (where long-term feeding occurs) in ways that reduce plant quality for aphids; (2) causes constitutive up-regulation of salicylic acid; (3) alters herbivore-induced jasmonic acid biosynthesis as well as the sensitivity of downstream defences to jasmonic acid; and (4) elevates ethylene emissions and free fatty acid precursors of volatiles. These findings are consistent with previously documented patterns of aphid performance and behaviour and provide a foundation for further exploration of the genetic mechanisms responsible for these effects and the evolutionary processes that shape them.
Increasing evidence suggests that vector-borne pathogens are capable of altering the phenotypes of their hosts in ways that influence the frequency and nature of interactions between hosts and vectors (Roy & Raguso 1997; Eigenbrode et al. 2002; Lacroix et al. 2005; Lefévre & Thomas 2008; Mauck, De Moraes & Mescher 2010; Bosque-Pérez & Eigenbrode 2011; Sugio et al. 2011; Mauck et al. 2012; Shapiro et al. 2012). In the case of plant viruses – which are transmitted primarily by insects such as aphids, whiteflies, thrips, leaf/plant-hoppers and beetles – a number of recent studies have documented virus effects on host-plant traits that influence patterns of vector attraction to, arrestment and performance on, and dispersal from infected plants (e.g. Eigenbrode et al. 2002; Maris et al. 2004; Belliure et al. 2005; Mauck et al. 2010, 2012; Bosque-Pérez & Eigenbrode 2011; Ingwell et al. 2012). Furthermore, there is ample reason to believe that such effects can have significant implications for virus transmission and epidemiology (Madden, Jeger & van den Bosch 2000; Zhang et al. 2000; Jeger et al. 2004; Sisterson 2008). To date, however, we know relatively little about the specific changes in host-plant physiology that mediate these effects in most systems, and still less about underlying biochemical and molecular mechanisms.
While few studies have directly explored the mechanisms responsible for plant virus effects on host-vector interactions, a great deal of research – focused primarily on elucidating the mechanisms of virus movement and replication within the host and on understanding impacts on crop plant yield – documents virus effects that are likely relevant to such interactions. For example, plant viruses can alter whole-plant and leaf morphology, metabolite profiles and plant responses to biotic and abiotic stress (Shalitin & Wolf 2000; Whitham & Wang 2004; Xu et al. 2008). Such effects occur as a consequence of exploitation of host resources by the pathogen and through direct interactions among viral components, host proteins and other cellular structures (Whitham & Wang 2004). Many plant defence pathways can also be activated or suppressed by virus infection, depending on the compatibility of the virus–host interactions (reviewed in Whitham & Wang 2004; Whitham et al. 2006; Lewsey et al. 2009, 2010). Virus infection has furthermore been shown to influence the production of the key plant hormone ethylene (Chaudhry et al. 1998) and may also have effects on precursors to defence pathways, such as membrane fatty acids, through modulation of organelle and plastid membranes to become virus replication scaffolds (Kim et al. 2002; Whitham & Wang 2004).
These and other virus effects on host-plant physiology may be expected to influence host-vector interactions through effects on a variety of ecologically relevant plant traits, including (1) visual and olfactory cues that mediate vector attraction to infected plants (e.g. leaf colour or the emission of volatile organic compounds) (Eigenbrode et al. 2002; Jiménez-Martínez et al. 2004b; Srinivasan et al. 2006; Medina-Ortega et al. 2009; Werner et al. 2009; Mauck et al. 2010); (2) constitutive or induced plant defences against herbivorous insect vectors (Belliure et al. 2005; Zhang et al. 2012); and (3) changes in plant nutritional quality for vectors or in the proximate (e.g. gustatory) cues by which vectors assess plant quality (Selman et al. 1961; Bozarth & Diener 1963; Lowe & Strong 1963; Blua et al. 1994; Herbers et al. 1997; Jiménez-Martínez et al. 2004a; Mauck et al. 2010). In turn, these altered traits may influence epidemiologically relevant patterns of vector behaviour and performance, including the frequency with which vectors visit infected, relative to healthy, plants; the likelihood that vectors will initiate feeding and the duration of feeding/residence on host plants; rates of growth and reproduction; and rates/timing of dispersal.
Recently, we explored the impacts of the widespread pathogen, Cucumber mosaic virus (CMV) on interactions between a common host-plant (cultivated squash, Cucurbita pepo) and aphid vectors (Mauck et al. 2010). CMV is a non-persistently transmitted virus that is acquired during brief probes of the outer cells of infected leaf tissue and forms only transient associations with the mouthparts of aphid vectors. For non-persistently transmitted viruses such as CMV, efficient transmission to a new host is typically aided by rapid vector dispersal from infected plants and can be hindered by the initiation of long-term feeding in the phloem (Martín et al. 1997; Wang & Ghabrial 2002; Ng & Falk 2006). Our field and laboratory experiments revealed a pattern of virus effects on vectors consistent with this transmission mechanism: the performance of the aphid vectors Myzus persicae (Sulzer) and Aphis gossypii (Glover) was dramatically reduced on infected plants, and aphids readily emigrated from infected to nearby healthy plants without initiating long-term feeding. Remarkably, however, both winged and wingless aphids were ‘deceptively’ attracted to the odours of infected plants, which exhibited elevated emissions of a volatile blend otherwise similar to that of healthy plants. In the current study, we build upon this research by exploring the physiological and biochemical changes induced by CMV in host plants and their relevance for vectors.
As aphid performance is largely regulated by available nutrients and ease of nutrient acquisition, as well as anti-herbivore defences of the host plant, we hypothesized that CMV infection significantly alters key nutrient levels and ratios for aphids (e.g. levels of free amino acids and simple carbohydrates, and carbohydrate to amino acid ratios, in leaf and phloem tissue) and/or constitutive and induced anti-herbivore defences (which we assayed via analyses of plant-produced phytohormone levels and the induction of defences by external application of synthetic phytohormones). Additionally, we expected CMV-induced changes in metabolites that function as precursors or hormone regulators of volatile organic compounds that might contribute to the previously observed enhancement of volatile emissions from CMV-infected plants.
Plant, insect and virus culture
Cultivated squash (C. pepo cv. ‘Dixie’, Willhite Seeds Inc., Poolville, Texas, USA) were grown in 12 cm3 square pots in autoclaved ProMix potting soil containing 5 g of slow-release fertilizer (Osmocote 14-14-14 N-P-K, Scotts, Marysville, Ohio, USA) and trace micronutrients (Micromax Micronutrients, Scotts, Marysville, Ohio, USA). All plants were grown in an insect-free, walk-in growth chamber with a 16:8 light:dark photoperiod maintained with banks of fluorescent and incandescent bulbs set to 23 °C during the day and 21 °C at night. When plants were at the cotyledon stage, they were inoculated with 5 cm2 of frozen stock tissue infected with CMV (stored at −80 °C). Frozen tissue was ground on a cold surface then combined with 15 mL of chilled 0.1 M potassium phosphate buffer. Carborundum powder was added and the solution was applied to cotyledon surfaces using cotton swabs. All inoculations employed the same stock collection of frozen tissue. Plants designated for the healthy treatment were mock-inoculated in the same manner, but with healthy squash tissue, to provide a control for the effects of the mechanical inoculation and the influence of plant-derived factors. Plants used for experiments were 3 weeks old (2.5 weeks since inoculation) or 5 weeks old (ethylene measurements). M. persicae were raised in colonies caged on Capsicum annuum (cv. California Wonder) and used as needed to induce plant defences or for population growth assays (details below).
Measurement of amino acids and carbohydrates
Analyses of amino acids and carbohydrates were performed on 3-week-old squash plants (10 infected and 10 healthy). Collection of tissue was performed between 1000 and 1200 h. From each plant, we used the two most recently expanded leaves from the apex, which we observed to be preferred feeding sites for aphids. Leaves used for phloem exudation (the most recently expanded leaf) were of a similar size across treatments. Leaves used for epidermal and mesophyll tissue analysis (the second most recently expanded leaf) tended to be larger for healthy plants, but tissue was collected from only a portion of the leaf and weighed to standardize. The leaves excised for phloem exudation were cut 8 cm down the petiole, immediately dabbed dry at the excision point using a Kimwipe (Kimberly-Clark Co., Neenah, Wisconsin, USA) to remove any wounding artifacts, and placed into 500 μL of 5 mM disodium ethylenediaminetetraacetic acid (EDTA), which was continuously chilled in an ice bath housed in a dark box (to prevent transpiration) during the 2-h collection period. (King & Zeevaart 1974). Samples were stored at −20 °C after collection. Although aphid stylectomy is frequently employed to collect phloem exudates, we used the well-established EDTA technique because (1) aphid species that feed on squash are too small for the stylectomy technique to be performed reliably and (2) Cucurbita phloem exudes sap copiously relative to other species (e.g. Girousse et al. 1991), allowing for collection of amounts adequate for analysis over a short period of time (2 h). The EDTA exudation technique is especially useful for determining amino acid composition and the ratio of sugars to amino acids (reviewed in Douglas 2003). From the second most recently expanded leaf, we collected leaf tissue using a 0.6 cm diameter cork borer that allowed for sampling between major veins. Tissue was collected quickly by punching 10 leaf discs per sample, weighing them and flash-freezing them in 2 mL Eppendorf tubes (Eppendorf North America, Hauppage, New York, USA). Tubes containing samples were frozen at −80 °C until processing.
To measure the target molecules in leaf tissue, the leaf-disc samples were ground in a Geno/Grinder (SPEX SamplePrep, Metuchen, New Jersey, USA) for 30 s at 1100 shakes per minute with two steel ball bearings as pestles. Samples were kept frozen during grinding using liquid nitrogen. Three samples (two healthy and one infected) were lost during grinding because of a failure of the Eppendorf tube caps; all other samples remained intact. All chemicals used were obtained from Sigma-Aldrich (St. Louis, Missouri, USA) (high-performance-liquid-chromatography grade, derivatization grade or higher), and all water used was distilled, deionized and filtered through a Millipore water sterilization system (EMD Millipore, Billerica, Massachusetts, USA). Extraction and analysis followed (Lisec et al. (2006), and we provide here a very brief summary of their detailed protocol. Each leaf sample received an internal standard (ribitol) in a known concentration, and metabolites of interest were extracted from ground tissue using methanol, chloroform and water. Phloem samples were combined with the internal standard and 300 μL of chloroform, vortexed, then centrifuged at 12 000 g for 1 min. Aqueous partitions containing amino acids and carbohydrates were removed to new Eppendorf tubes and dried in a Speedvac (Thermo Scientific, Franklin, Massachusetts, USA). Samples were stored overnight, dried again for 30 min and then derivatized using methoxyamine hydrochloride and N-methyl-N-(trimethylsilyl) trifluoroacetamide. Blank samples lacking leaf tissue or phloem exudates were also taken through the procedure. Details of the gas chromatography and mass spectrometry instrument configurations can be found in the Supporting Information.
Chromatograms were analysed using Chemstation software (Agilent Technologies, Santa Clara, California, USA). Peak areas for leaf samples were converted to nanogram amounts relative to the known amount of internal standard added to each sample, then corrected for the weight of leaf tissue collected. Peak areas for the phloem samples were similarly converted to nanogram amounts based on the internal standard. For phloem collections, amounts of each amino acid were converted to relative percentages to compare samples that varied slightly in exudate volumes. Simple carbohydrates were also summed and expressed as a proportion of the total amino acid concentration within that sample (ratio of carbohydrates to amino acids). If derivatization produced more than one peak for a compound (verified by standards), the amounts for the two products were summed. Comparisons for each compound based on the two treatments (infected and healthy) were done using non-parametric t-tests (Kruskal–Wallis test, Minitab v. 14 [Minitab Inc., State College, Pennsylvania, USA]). Total amino acids (leaf only) and carbohydrate to amino acid ratios were compared using one-way analysis of variance (anova) with log transformations where necessary to correct for normality (Minitab v. 14).
Measurement of in-plant phytohormones
Three-week-old squash plants were pre-sampled by excising one of the two most recently expanded leaves at the petiole to establish baseline levels of jasmonic acid (JA) and salicylic acid (SA), along with fatty acid precursors to many plant defences [linoleic acid (LA) and linolenic acid (LNA)]. From each excised leaf sample, a 2 cm diameter leaf disk was punched using a cork borer, weighed, placed in a screw-cap Fast-prep tube and then flash frozen in liquid nitrogen and stored at −80 °C. Previous tests determined that excising a leaf at the petiole causes no systemic change in plant chemistry, and this approach has previously been employed for sequential collections of tissue for metabolite analysis and herbivore bioassays (e.g. Thaler & Bostock 2004) and as a method of manipulating plant tissue removal without inducing plant defences (e.g. Agrawal et al. 1999). Ten third-instar M. periscae were applied to the remaining recently expanded leaf (confined in a supported, soft-clip cage lined with cotton at the contact point with the plant) and allowed to feed for 24 h. Prior tests determined that the presence of clip cages causes no change in plant chemistry. After 24 h, aphids were removed from the plant and the feeding area (2 cm diameter) was punched out, weighed and then flash frozen in liquid nitrogen and stored at −80 °C. Eight plants were used for each treatment.
Samples were processed and analysed using gas chromatography and mass spectrometry according to Schmelz et al. (2003), using gas-chromatography-grade chemicals and standards from Sigma-Aldrich and isotopic standards (JA and SA) from Tokyo Chemical Industry Co. (Tokyo, Japan) and Campro Scientific (Berlin, Germany), respectively. Briefly, internal isotopic or isomeric standards were added to each sample and samples were ground in a propanol/water/hydrochloric acid buffer using a Fast-prep machine (MP Biomedicals, Solon, Ohio, USA), then combined with dichloromethane and centrifuged to separate leaf material from the organic supernatant. The supernatant was transferred to glass vials, dried and the resultant residue was mixed with methanol and diethyl ether solvents and carboxylic acids and alcohols were derivatized to methyl esters using 2.0 M trimethylsilyldiazomethane. Methyl esters were recollected on Super-Q adsorbent using vapour-phase extraction. Collected compounds were eluted using dichloromethane and analysed using the instruments described above according to settings outlined in Schmelz et al. 2003. The amount of each target compound was calculated relative to the isotopic (SA and JA) or isomeric (LA and LNA) standard and then corrected for the weight of tissue used for each sample. Amounts of each compound were analysed using a general linear model with treatment, damage and treatment by damage as main effects and plant number as a random effect. Multiple comparisons using Tukey's test were performed for the treatment by damage interaction. All analyses were performed using Minitab v. 14 with log (SA, LA, LNA) or square root transformations (JA) to satisfy assumptions of normality.
Measurement of ethylene gas emissions
The most recently expanded leaf was excised from each plant and the petiole was immediately submerged in an Eppendorf tube with 1 mL of water. Each leaf + tube was enclosed in a 125 mL glass jar with a septum set in the lid and incubated for 2 h at 25 °C under fluorescent grow lights in a growth chamber. From each jar, two 500 μL aliquots of air were then sampled with a gas-tight syringe. Each aliquot was immediately injected manually into an Agilent 6890 gas chromatograph with a flame ionization detector (Agilent Technologies, Santa Clara, California, USA). The GC was fitted with a Restek RT-QS Bond PLOT column (30 m length, 0.53 mm internal diameter, 20 μm film) (Restek Co., Bellefonte, Pennsylvania, USA). The oven temperature and inlet were maintained isothermally at 70 °C with a helium carrier gas flow of 9 mL min−1. Chromatograms recorded during the run were saved and ethylene peaks were manually integrated. A standard curve of ethylene quantities was determined using incremental aliquots of 0.01 mg g−1 ethylene gas in nitrogen. Peak areas were converted to nanomoles based on the standard curve and are reported on a per jar volume, per gram of tissue basis. To test whether ethylene is up-regulated in response to infection during early and late phenological stages, we sampled 3- (n = 8) and 5-week-old (n = 12) plants. Comparison of infected and healthy plants within each age group was done using one-way anova (Minitab v. 14).
Measurement of induced plant defences and effects on aphid vectors
To determine if observed changes in JA because of virus infection and aphid feeding translate into the induction of plant defences, we assayed changes in leaf trichome production, a JA-inducible defence, in response to JA application to infected and healthy plants (Thaler et al. 1996; Traw & Bergelson 2003). Fourteen days after inoculations, we applied JA (1.5 mM or 158 nM in a 5% acetone in water solution) or a control (5% acetone in water) to infected and healthy plants – the acetone was needed to help disperse the JA in the water solution. JA was obtained by demethylation of methyl jasmonate (Sigma-Aldrich). Each plant received 600 μL of solution as a fine mist across the two most recently expanded leaves. Concentrations were chosen based on previous literature (1.5 mM) (Thaler et al. 1996, 2002, Traw & Bergelson 2003), and a concentration (158 nM) delivering a quantity of JA similar to that quantified in squash leaf tissues under damage by aphids (1–10 ng g−1 of plant tissue). Because of logistical constraints, we performed the 1.5 mM and 158 nM experiments separately, each having appropriate controls and four to five replicate plants per treatment. Trichomes were counted on a 50 mm2 leaf disc taken from the same location on the most recent fully expanded leaf of each replicate 10 days after application of the JA and control treatments. Leaf discs were photographed using an Olympus DP73 microscope camera fitted to an Olympus SZX10 stereo microscope (Olympus America Inc., Center Valley, Pennsylvania, USA), and trichomes were quantified with the aid of Adobe Photoshop CS6 (Adobe Systems Inc., San Jose, California, USA). Data were analysed for differences relative to controls using the Kruskal–Wallis test (Minitab v. 14). Experiments were also performed to determine the effects of single-event JA up-regulation and constitutive SA up-regulation on aphid population growth. Salicylic acid is constitutively up-regulated by CMV-FNY infection in squash, and JA up-regulation occurs in infected plants in response to aphid damage (see Results section). However, it is unknown whether such conditions result in the induction of defences against aphids in C. pepo. For the JA experiment, we used the same JA concentrations as for the trichome induction experiment and tested the effects of a single application of 600 μL of JA or water solution on subsequent aphid population growth on healthy C. pepo host plants. One day after JA application, 10 M. persicae of a standard age (second to third instar) were applied to each plant and confined using a large mesh cage around the entire plant. Aphid populations were counted after 10 days. To test the effects of constitutive SA up-regulation on aphid growth, 600 μL of an SA solution or a water control (1% ethanol in water) were sprayed on the first expanded true leaves of 2-week old squash plants, and a standard age population of M. persicae (10 second- or third-instar nymphs) was subsequently caged on the entire upper portion of the plant above the first true leaves. We used an SA concentration known to have biological activity (0.1 mM) (Senaratna et al. 2000, Traw & Bergelson 2003, Radwan et al. 2007) and a concentration (36 μM) that delivered a total quantity of SA similar to what we observed in CMV-infected plants (400–600 ng g−1 leaf tissue). The SA and water solutions were applied to the same leaves every other day for the following 10 days to mimic a constitutively up-regulated condition, after which aphid populations were assessed for each plant. For each experiment there were 10 replicates per treatment. Data for both experiments were rank transformed and analysed by anova (Minitab v. 14).
Amino acids and carbohydrates in leaf and phloem tissue
Mesophyll and epidermal cells of infected plants contained higher levels of free amino acids (Fig. 1a), with CMV-infection inducing the accumulation of all of the amino acids detected except alanine and proline (Fig. 1b). However, infected mesophyll and epidermal cells contained significantly lower levels of simple carbohydrates (Fig. 1c), and thus the ratio of simple carbohydrates (glucose, fructose and sucrose) to amino acids in infected plants was approximately 0.7:1, while the ratio of simple carbohydrates to amino acids in healthy plants was approximately 5.8:1 (Fig. 1d). Sucrose was abundant in leaf tissue, and again the ratio of sucrose to amino acids was higher in healthy plants relative to infected plants (Fig. 1d).
In contrast, the phloem of infected plants had significant differences in amino acid composition (percent of total amino acids) relative to healthy plants (Fig. 2a). While infected plants had higher relative percentages of a few non-essential amino acids (proline, alanine and asparagine), healthy plants had higher relative percentages of several essential amino acids (tryptophan, lysine, valine) as well as the non-essential amino acid tyrosine and the non-coding amino acid homoserine. Threonine was slightly higher in infected plants, although this amino acid was the lowest relative percent of the total amount of amino acids detected for both treatments. Ratios of simple carbohydrates to amino acids were approximately 3.3:1 for infected plants, and were significantly higher for healthy plants, at approximately 5.5:1 (Fig. 2b). We detected glucose, fructose and sucrose in our analysis, but sucrose was in relatively low amounts (roughly 2% of total amount of glucose and fructose detected). It is possible that very low levels of enzymatic hydrolysis of sucrose occurred during exudate collection [estimated to be less than 5% (van Bel & Hess 2008)]; however, our results are consistent with other studies of the genus Cucurbita, where researchers collected phloem sap (using different methods) and reported that fructose and glucose are major transport sugars in the phloem, with sucrose present in relatively low amounts (Fiehn 2003).
In-plant phytohormones and ethylene gas emissions
Levels of SA were higher in CMV-infected plants and were not influenced by aphid feeding (Fig. 3a). Undamaged CMV-infected and healthy plants exhibited similar baseline levels of JA, but aphid feeding induced significantly higher levels of JA in infected plants (Fig. 3b). LNA was higher in CMV-infected plants relative to healthy plants and aphid feeding did not cause a change in levels (Fig. 3c). LA was also higher in CMV-infected plants (Fig. 3d), and aphid feeding caused a decrease in LA levels, which was apparent in both treatments, but significant only for healthy plants (Fig. 3d). Ethylene emissions were uniformly higher, by about 30%, in CMV-infected plants at both 3 and 5 weeks post-infection (Fig. 4).
Induced plant defences and effects on aphid vectors
Trichome numbers in systemic leaves of plants receiving 1.5 mM JA were significantly higher than plants receiving the water control (Fig. 5a,b). Both CMV-infected and healthy plants showed the same induction in response to 1.5 mM JA, indicating that infected plants are capable of inducing JA-mediated responses despite constitutively high SA levels (Fig. 3a). In contrast, under the lower JA treatment of 158 nM, only the healthy plants showed induction of trichomes relative to the water control (Fig. 5c,d). Aphid growth on healthy plants receiving a single application of either 1.5 mM or 158 nM JA was not significantly different from the water control (Fig. 6a), and repeated application of 0.1 mM or 36 μM SA also did not result in reduced aphid population growth (Fig. 6b).
The results presented here document a range of CMV effects on host-plant physiology including (1) alteration of key nutritional traits that influence palatability and quality for aphids, (2) altered plant defence responses to aphid feeding, and (3) elevation of compounds implicated in the regulation of volatile emissions. Taken together, these effects are consistent with the pattern of interactions between CMV-infected plants and aphid vectors documented previously (Mauck et al. 2010), in which aphids are preferentially attracted to infected plants, but rapidly disperse from these plants after initial feeding. This pattern is conducive to the non-persistent transmission mechanism exhibited by CMV, consistent with our previous suggestion that the effects of vector-borne pathogens on host-vector interactions will typically have favourable (to neutral) effects on transmission, as selection will be expected to act against effects that negatively impact pathogen fitness (Mauck et al. 2012; Shapiro et al. 2012).
CMV-induced changes in host-plant nutrition
We previously reported a strong tendency of aphids to emigrate from CMV-infected to healthy plants without initiating long-term feeding. It is likely that this rapid dispersal is mediated by gustatory cues encountered during brief probes of leaf tissue, during which virions may be acquired. Aphids are stimulated to probe and taste surfaces with which they make tarsal contact (Powell et al. 1999), and, upon assessing cellular contents, are further stimulated to initiate vascular feeding or dispersal (Powell et al. 2006). Our data show that the nutritional cues available from non-vascular leaf cells of CMV-infected plants are significantly altered relative to those of healthy plants, which are strongly preferred by aphids in settling tests (Mauck et al. 2010). Phagostimulatory sugars, especially sucrose, are significantly lower in virus-infected plants, while free amino acids are higher (Fig. 1). This leads carbohydrate to amino acid ratios well below the 8:1 ratio that has been shown to be phagostimulatory and optimally nutritious for pea aphids (Acythosiphum pisum) – provided that free amino acids are above a minimum concentration threshold (Abisgold et al. 1994). Additionally, earlier work on M. persicae showed that performance and willingness to feed increase as the sucrose to amino acid ratio approaches 6.25:1 (Mittler 1967), and that fructose and glucose are both phagostimulatory and effectively utilized by M. persicae (Mittler et al. 1970). In our study, healthy plants exhibited higher levels of all three of these sugars (sucrose being most important), and more optimal ratios of these simple carbohydrates in relation to amino acids. The large body of research demonstrating a link between phagostimulatory carbohydrates, carbohydrate to amino acid ratios and the willingness of aphids to feed following acquisition of these taste cues (e.g. Auclair 1962; Mittler & Dadd 1965; Mittler 1967; Mittler et al. 1970; Abisgold et al. 1994; Douglas 2003; Pescod et al. 2007) strongly indicates that the condition found in non-vascular cells of CMV-infected C. pepo plants is likely to stimulate vector dispersal following probing and to discourage the initiation of phloem feeding, which would lead to a loss of virions acquired during probing.
While aphids infer information about plant quality from gustatory cues available during initial probes, aphid performance on plants is strongly influenced by the nutrition available in phloem. We previously reported reduced aphid population growth on CMV-infected relative to healthy squash plants (Mauck et al. 2010), and recent work in tobacco also suggests that CMV infection may lower the nutrient quality of phloem in this host, as aphids were shown to have longer average periods of phloem ingestion on infected plants, but to grow at a lower rate (Ziebell et al. 2011). Consistent with these observations, we found that infected plants had reduced relative percentages of several essential amino acids, as well as reduced percentages of homoserine and tyrosine, which are non-essentials. Homoserine, a precursor for some essential amino acids, and the aromatic amino acid, tyrosine, have both been shown to positively influence aphid performance when manipulated to constitute a larger percentage of the total amino acid composition (Maltais & Auclair 1962; Febvay et al. 1988a,b). Furthermore, the ratios of carbohydrates to amino acids were significantly reduced in CMV-infected plants (Fig. 2), suggesting that levels of key sugars utilized by M. persicae (Mittler et al. 1970) and other aphids (Febvay et al. 1999; Ashford et al. 2000) may be reduced in infected plant phloem. Additionally, as the ratio of simple carbohydrates to amino acids in ingested solution moves away from an optimal balance (roughly 6.25:1 for M. persicae and 8:1 for Ac. pisum), aphids are less stimulated to feed and exhibit reduced performance (Mittler 1967; Abisgold et al. 1994). As aphids often use gradients of sucrose or other carbohydrates to locate phloem (Hewer et al. 2010), reduced levels of carbohydrates might also make it difficult for aphids to efficiently feed on CMV-infected hosts. Furthermore, aphids carry symbiotic bacteria that synthesize essential amino acids from more abundant non-essentials, as well as from sucrose, glucose and fructose (Febvay et al. 1999; Ashford et al. 2000). Thus, a further consequence of the nutritional disruptions we documented is a reduction in intake of starting materials that are essential for subsidizing the inherently low free amino acid concentrations characteristic of a phloem sap diet (Febvay et al. 1999; Douglas 2003).
The reduction in phloem carbohydrates observed here may not occur uniformly in all hosts infected by CMV, of which there are many strains that infect a wide variety of hosts across multiple plant families (Roossinck 2002). For instance, Shalitin & Wolf (2000) reported high levels of sucrose in the phloem of melon plants (Cucumis melo) infected with CMV-FNY, as well as high levels of glucose and fructose in leaves (but did not report on amino acids in either tissue), later suggesting that these alterations might be because of the effects of the CMV movement protein on phloem loading (Shalitin et al. 2002). Tésci et al. (1994) reported reduced or unaltered concentrations of sucrose in the cotyledons of C. pepo plants recently inoculated with CMV, and increases in amino acid concentrations consistent with our results, a pattern also reported for older C. pepo plants infected by CMV (Fegla & Sheir 1975). Although none of these studies measured the same compounds in both phloem and leaf tissue, as was done here, this work indicates that CMV may not alter carbon and amino acid partitioning and/or metabolism in the same way across all hosts, raising the intriguing possibility that different host plants might be more or less efficient sources of virus inoculum. If this is the case, the virus-induced metabolite condition in a given host plant may influence the number of new infections arising from that source of inoculum through effects on aphid vector behaviour.
CMV effects on host-plant defences against aphids
Plant defences are another key factor influencing aphid performance and behaviour. We found that squash plants infected with CMV exhibited significant differences in levels of defence-related phytohormones compared with healthy plants (Fig. 3). Infected plants had significantly higher baseline levels of SA, although aphid damage had no effect on SA levels in either healthy or infected plants (Fig. 3a). In work with Arabidopsis thaliana, SA-regulated genes were expressed in response to feeding by aphids (Moran & Thompson 2001) and other phloem feeders such as whiteflies (Zarate et al. 2007), which may be a manipulation by these herbivores aimed at suppressing the up-regulation of the anti-herbivore signalling molecule JA, which is often negatively correlated with SA accumulation (Zarate et al. 2007). Research using tomato as a model organism has shown that SA accumulation is instead associated with reduced aphid performance and, significantly, with reduced aphid feeding preferences (Boughton et al. 2006; Avila et al. 2012). In our experiments with C. pepo, we did not find evidence of significant impacts of aphid feeding on SA induction in either infected or healthy plants (Fig. 3a), and we did not observe any effects of constitutive SA up-regulation via exogenous applications on aphid population growth (Fig. 6b). These results suggest that SA is not a key regulator of defences against aphids in this host, and that CMV-induced SA accumulation likely does not contribute to the reduced population growth and reduced feeding preference of aphids on CMV-infected C. pepo.
Active up-regulation of the SA pathway by CMV during the course of infection may confer other benefits to the pathogen that indirectly influence transmission and fitness. For instance, many pathogens are unable to successfully colonize and infect hosts that have experienced prior induction of the SA pathway (e.g. other viruses, bacteria and fungi), so by up-regulating SA over the course of infection, the CMV pathogen may avoid co-infections (Radwan et al. 2007; Sasu et al. 2010; Syller 2011). Arguing against this interpretation are previous reports that elevated SA levels can inhibit the replication and/or systemic movement of CMV and other viruses (Naylor et al. 1998; Mayers et al. 2005). However, up-regulation of SA in these studies occurred prior to inoculation. The current results, and those of other studies (e.g. Huang et al. 2005; Lewsey et al. 2010), indicate that a slow, CMV-induced accumulation of SA does not inhibit the CMV infection and may even be a necessary part of systemic CMV spread within a host.
Despite the relatively high levels of SA in infected plants, JA was induced (although not to particularly high levels) by aphid feeding in these plants while not being induced at all in healthy plants (Fig. 3b). The stronger induction of JA in infected plants might be partially explained by elevated baseline levels of the JA precursors LNA and LA in these plants (Fig. 3c,d). Interestingly, however, a pilot experiment examining JA induction in CMV-infected and healthy hosts by the chewing herbivore, Heliothis virescens, revealed the inverse pattern – healthy plants exhibited significant JA induction in response to this herbivore as would be expected, while this response was suppressed in CMV-infected plants (Supporting Information Fig. S1). Despite this very specific up-regulation of JA in response to aphid feeding, follow-up experiments examining the effects of a similar concentration of exogenously applied JA on aphid growth suggest that JA-based defences in C. pepo are not particularly effective at reducing aphid performance (Fig. 6a) and that while low concentrations of JA can induce plant defences (trichomes) in healthy C. pepo, the same concentration fails to induce defences in CMV-infected plants (Fig. 5c,d). In another study, reduced activation of JA-regulated genes in response to mechanical wounding of leaves or application of 0.25 mM methyl jasmonate was reported for CMV-infected tobacco (Lewsey et al. 2010). Our results suggest that a similar attenuation of JA-regulated defences occurs in CMV-infected C. pepo. However, higher levels of JA (1.5 mM) did induce greater production of trichomes in CMV-infected plants (Fig. 5a), suggesting that JA-regulated defences are still functional, but are less sensitive to the JA signal than in healthy plants. This effect of CMV infection could have important implications for interactions between the host plant and other organisms, such as non-vector herbivores or necrotrophic pathogens, against which JA-mediated defences play a greater role.
Effects of CMV infection on host-plant volatile chemistry
A key aspect of the previously reported effects of CMV on plant-vector interactions is the strongly elevated emission by infected plants of a volatile blend otherwise similar to that of healthy plants, which resulted in enhanced attraction of both winged and wingless aphids to the odours of infected plants (Mauck et al. 2010). We previously suggested that this effect might be partially mediated by CMV effects on the release of 2-hexenal (Mauck et al. 2010), a small molecular weight volatile belonging to a class of compounds (C6 alcohols and aldehydes) that act as signals to the plant and cause increases in volatile emissions (Kahl et al. 2000; Ruther & Kleier 2005; Gao et al. 2009). The current study also reveals a significant effect of CMV infection on the release of ethylene both early on in the host-plant life-cycle (3 weeks) and in older plants (5 weeks) (Fig. 4). Ethylene and C6 alcohols and aldehydes, including isomers of hexenal, have previously been found to exhibit synergistic effects on volatile emissions from leaves exposed to these cues (Ruther & Kleier 2005; Gao et al. 2009). We also observed elevated levels of key precursors for C6 alcohol and aldehyde volatiles (LNA and LA) in CMV-infected squash plants relative to healthy plants (Fig. 3c,d). Components of the CMV replicative machinery, namely the CMV 1a and 2a proteins, are known to interact with such molecules, which are membrane components, through manipulation of plastid membranes within the plant cell to create protected sites of replication (Kim et al. 2002). Thus, essential aspects of the replication process used by CMV, or other plant viruses, may have impacts on biochemical pathways that influence plant volatile emissions, creating the potential for subsequent selection to modify or fine tune such effects with respect to their influence on the olfactory cues presented to insect vectors.
The findings presented here confirm our previous report of a suite of CMV effects on interactions between C. pepo plants and aphid vectors that appear conducive to the non-persistent transmission mode of this pathogen (Mauck et al. 2010). More significantly, they provide key insights into the virus-induced changes in host-plant biochemistry and physiology that generate these effects. Our results indicate that reductions in host palatability and quality for aphid vectors are driven primarily by changes in key nutrients important to the aphid diet. CMV infection induces changes in the overall concentrations of phagostimulatory carbohydrates and free amino acids conducive to aphid rejection of plant tissues and rapid dispersal following virus acquisition (Fig. 1) (Abisgold et al. 1994). Meanwhile, CMV infection also reduces phloem sap quality, consistent with the reduced population growth that we observed for aphids growing on CMV-infected plants (Fig. 2) (Mauck et al. 2010). While we also found significant effects of CMV on baseline and aphid-induced levels of defence-related phytohormones (Fig. 3a,b), subsequent experiments (Figs 5 & 6) suggest that SA- and JA-mediated plant defences do not play a major role in reducing host-plant quality for aphids relative to the drastic changes in nutritional quality observed. However, we cannot rule out the possibility that in the presence of CMV-induced nutritional stress, SA- or JA-mediated defences may contribute to reductions in aphid performance and/or changes in feeding behaviour. In addition to changes in within-plant phytohormones, we also observed that CMV infection results in enhanced emission of the volatile hormone ethylene (Fig. 4) which can act synergistically with other volatiles emitted from infected plants (C6 alcohols and aldehydes) to up-regulate overall volatile emissions (Kim et al. 2002; Ruther & Kleier 2005; Gao et al. 2009), increasing the apparency of infected relative to healthy plants. CMV also caused plants to produce greater amounts of key fatty acid precursors of volatile alcohols and aldehydes, which may contribute to the enhanced emission of these volatile signalling molecules (Fig. 3C,D) (Mauck et al. 2010). Taken together, these findings provide much-needed insight into potential mechanisms influencing the vector-borne transmission of non-persistently transmitted viruses and support the hypothesis that mechanisms of vector transmission may play an important role in shaping virus effects on host-plant phenotypes (Mauck et al. 2010, 2012), particularly in commonly encountered hosts, such as C. pepo is for CMV, where selection may favour viral variants that cause specific, beneficial changes in host chemistry.
Thanks to Dr. John Tooker for comments on the manuscript. Thanks to Heike Betz, Erica Smyers and Janet Saunders for technical assistance and assistance with plant and insect maintenance. Thanks also to Dr. Rupesh Kariyat and Liz McCarthy for assistance with trichome photography and analysis methods. The CMV-FNY pathogen was provided by Dr. John Murphy (Auburn University). Funding for this project was provided by a NSF DDIG (DEB-1011122), a Penn State College of Agricultural Sciences Competitive Grant, and a USDA-AFRI NIFA Predoctoral Fellowship (2011-67011-30655) to K.E.M., a USDA-CSREES-NRI grant to M.C.M. and C.D.M. (2008-35302-04577), and by the David and Lucille Packard Foundation.