A non-persistent aphid-transmitted Potyvirus differentially alters the vector and non-vector biology through host plant quality manipulation

The association of plant viruses with their vectors has significant implications for virus transmission and spread. Only a few studies, with even fewer pathosystems, have explored non-persistent (NP) virus-vector interactions that are presumed to be transient. We studied how a NP virus, Papaya ringspot virus (PRSV) influenced the behavior and biology of its vector, the melon aphid (Aphis gossypii Glover) and the non-vector, silverleaf whitefly (Bemisia tabaci Gennadius). We also assessed whether the fitness effects on aphids are modulated through changes in the host plant, squash (Cucurbita pepo L.) nutrient profile. The overall performance of A. gossypii was substantially higher on PRSV-infected plants, along with increased arrestment on PRSV-infected than non-infected plants. No such PRSV-modulated fitness effects were observed with B. tabaci. PRSV-infected plants had increased concentrations of free essential amino acids: threonine, arginine and lysine; non-essential amino acids: glycine and homocysteine; and soluble carbohydrates: galactose, raffinose and cellobiose. In general, PRSV encouraged long-term feeding and enhanced fitness of A. gossypii through host plant nutrient enrichment. These findings provide evidence for a NP virus mediated positive fitness effects on its vector, with no spillover fitness benefits to the non-vector within the same feeding guild.


Results
Aphid and whitefly response to PRSV-infected and non-infected leaves. Dual-choice settling bioassays were conducted to determine the aphid and whitefly settling preferences to PRSV-infected versus non-infected squash plants. The aphid A. gossypii adults did not show any robust preference to either PRSV-infected or non-infected leaves (F (1,38) = 1.43, P = 0.23). Only ~25 of the 100 aphids in each trial settled on either of the leaves. The remaining aphids did not show any settling preference. The whitefly B. tabaci settling also followed a similar pattern. There was no clear preference to either of the leaves (F (1,38) = 1.45, P = 0.23). Unlike aphids, the majority of whiteflies (approx. 67%) settled on one of the two leaves.

Aphid emigration from and immigration to PRSV-infected and non-infected leaves.
The movement of A. gossypii away from and towards PRSV-infected versus non-infected plants was studied through emigration and immigration assays, respectively. Continuous observations of aphids at every 10 min for 1 h showed higher numbers of individuals emigrating from non-infected plants, when compared with the PRSV-infected plants at the first two time points (Fig. 1). As a result, a significant treatment (Z (1,38) = −3.84, P = 0.0001), time (Z (1,38) = −4.76, P < 0.0001) and treatment: time interactions (Z (1,38) = 3.23, P = 0.0012) were observed. The aphid numbers beyond the third time point; however, were nearly similar.
On the contrary, numbers of aphids immigrating towards both infected and non-infected leaves were nearly similar at every time point (Fig. 1). Furthermore, these numbers gradually declined at every time point, with highest numbers at the first time point (Non-infected = 12; Infected = 10) and the lowest at the last one (Non-infected = 2; Infected = 1). Thus, no significant effects of treatment (Z (1,38) = −0.093, P = 0.92) and treatment: time interactions (Z (1,38) = −0.11, P = 0.90) were observed. The time term (Z (1,38) = −4.76, P < 0.0001); however, was significantly different suggesting significant variation in mean number of aphids immigrating at different time points.

Discussion
We studied whether PRSV impacts the preference and fitness of a vector and a non-vector of the same feeding guild in a cucurbit pathosystem. Existing literature on select NP viruses highlights their inconsistent interactions with vectors. By exploring an economically important yet sparsely explored pathosystem, the present study demonstrated NP virus-mediated arrestment and long-term fitness benefits to its insect vector. These benefits can be evidenced by increased A. gossypii performance on PRSV-infected plants, in which the concentrations of five amino acids and three carbohydrates that are necessary to the nutrition and development of A. gossypii were elevated. Meanwhile, B. tabaci preference and development were not significantly altered on PRSV-infected squash. This was presumably due to very specific effects of PRSV-mediated nutritional manipulation on vectors, with no spill over effects to take advantage by B. tabaci although it belonged to the same phloem feeding guild as A. gossypii.
For both A. gossypii and B. tabaci, we did not find any significant settling differences between PRSV-infected and non-infected plants. Similar results have been presented previously in three studies with a NP virus, PVY, in which no settling responses of M. persicae and Macrosiphum euphorbiae (Thomas) were observed 1,33,34 . Results from the present study differed from those of similar studies in which CMV infection in squash 15 and in tobacco 35 , ZYMV infection in squash 11 , and TuMV infection in Nicotiana benthamiana 12,13 increased the attraction and/or settling of M. persicae and/or A. gossypii to their respective infected host plants. Phloem feeding insects are known to gauge plant quality information from gustatory cues available during initial probes 8 . Although the probing behavior may vary from one aphid species to the other for the same virus 10 and from one pathosystem to the other 36 . Since phloem-feeders principally use sucrose to locate phloem 37 , its natural low concentrations in squash may have made it difficult for A. gossypii and B. tabaci to prefer either of the plants in a robust way and may have contributed to poor overall settling of aphids (~25%).  www.nature.com/scientificreports www.nature.com/scientificreports/ Number of A. gossypii emigrating from non-infected plants was significantly higher at the first 10 min time interval and decreased subsequently at second-and third-time intervals. However, the actual numbers of emigrating aphids were substantially low in both treatments. On the contrary, these numbers were much higher for aphids immigrating towards both plants; however, no significant differences were observed. The minimal movement of aphids away from PRSV-infected plants show similarity with a few earlier studies [11][12][13] , while disparity with others 15,22 . In the latter, the aphid vectors M. persicae and A. gossypii readily emigrated from CMV-infected squash 15 , similarly Rhopalosiphum maidis (Fitch) readily emigrated from Soybean mosaic virus (SMV)-infected soybean 22 . M. persicae emigration; however, was not significantly different from PVY-infected potato over untreated control 1 . Soluble carbohydrates could partly explain the movement of A. gossypii towards and away from PRSV-infected plants. Earlier study by Hewer et al. 37 showed that in the absence of sucrose in leaves, raffinose strongly stimulates aphid probing. Since the concentrations of raffinose were significantly higher in PRSV-infected plants in the absence of sucrose, it is possible that A. gossypii responded to raffinose stimulated probing in the first 10-20 minutes 38 . As a result, no aphids emigrated from PRSV-infected plants, while relatively more emigrated from non-infected plants. Based on the first probing, aphids are either stimulated to initiate vascular feeding or to disperse 38 . The A. gossypii feeding appeared to be stimulated on PRSV-infected plants, but not on non-infected plants.
PRSV mediated changes in host plant quality influenced the performance of A. gossypii. Significant increase in the reproductive potential of A. gossypii, as evidenced by longer reproductive and adult periods and longevity, and increased fecundity and intrinsic rate appear to be due to profound increase in essential amino acids: threonine, arginine and lysine, and soluble sugars: galactose, raffinose and cellobiose in PRSV-infected plants. An array of previous studies has highlighted the roles of essential amino acids and soluble carbohydrates in elevating the performance of aphids [39][40][41][42][43][44][45][46] . Threonine, arginine and lysine, being essential amino acids, are critical constituents of aphid haemolymph, saliva and honeydew 39,46,47 . Furthermore, all three amino acids have been shown to positively influence aphid performance 39,46 . Glycine and homocysteine, although classified as non-essentials, when supplemented along with essential amino acids, increased aphid performance 48 . Furthermore, these and other non-essential amino acids are utilized by aphid endosymbionts as precursors to synthesize essential amino acids 45,46,49 .
Aphids principally use glucose, fructose, sucrose, galactose and raffinose to meet their dietary requirements 41,[49][50][51] . Consistent with multiple previous studies in Cucurbita genus 8,52 , we found sucrose in traces in both infected and non-infected squash plants and therefore its concentrations have not been reported. In the absence of sucrose, fructose and glucose were most abundant phloem sugars in both non-infected and PRSV-infected plants. In the present study, raffinose has been the most abundant sugar following fructose and glucose and its substantial increase in PRSV-infected plants, along with essential amino acids and sugars, may likely have encouraged A. gossypii growth. In a diet deficient in sucrose, raffinose has been reported to be critically important sugar that supports aphid growth 41 . A carbohydrate to amino acid ratio below 8:1 has been shown to be phagostimulatory for the pea aphid Acyrthosiphon pisum Harris 53 . Melon aphid, however, didn't encounter phagostimulatory response as it did not prefer either PRSV-infected or non-infected plants, despite these ratios being lower in non-infected plants and even more so in PRSV-infected plants. Significantly lower carbohydrate to amino acid ratios were also reported in CMV-infected plants 8 .
Showing both consistency and disparity with previous studies on NP viruses, PRSV modulated robust long-term effects on A. gossypii through enhanced nutritional quality of squash 5,15,35,54,55 . A thorough comparison of the present study with Blua et al. 9 and Mauck et al. 8 explains similarities and differences in aphid performance between these studies. The Blua et al. study 9 , as the current PRSV one, demonstrated that the longevity and intrinsic rate of increase of A. gossypii were significantly higher on ZYMV-infected vs. non-infected squash, mostly due to increased concentrations of individual amino acids. On the contrary, in Mauck et al. study 8 , the concentrations of essential amino acids, when compared with the respective non-infected squash plants, were significantly lower in CMV-infected plants. For instance, lysine and tryptophan were comparatively increased, while proline was significantly reduced in PRSV-infected plants. These two studies 8,9 reported similar results to the present study on PRSV with regards to soluble carbohydrate concentrations. Blua et al. reported relatively higher total soluble carbohydrate concentrations whereas, Mauck et al. 8 reported relatively higher and glucose and fructose concentrations in non-infected plants. Following the uptake of selectively manipulated amino acid and carbohydrate concentrations in PRSV-infected plants, the long-term effects of PRSV on A. gossypii appeared to be building up gradually during nymphal development and thrived later in the adulthood as evidenced by superior life history traits on PRSV-infected plants.
B. tabaci fecundity and developmental time showed no association with PRSV-induced changes in squash amino acids and sugars. This study showed disparity with recently published studies on NP 56 and persistently-transmitted viruses 57,58 . In the NP virus study, B. tabaci was reported to preferably settle on non-infected chilli plants, but interestingly develop faster on CMV-infected chilli plants 56 . In the persistent virus studies, Tomato spotted wilt virus (TSWV) and Pepper golden mosaic virus (PEPGMV) offered fitness benefits to non-vector insects 57,58 . No effects in the present PRSV study could be due to intrinsic differences in nutrient uptake and metabolism and requirements between vector and non-vector insects. For instance, histidine and ornithine were reported to be the predominant amino acids determining B. tabaci development time on Cucumis melo L. 59 . No changes in B. tabaci development time in the present study could be due to low levels and lack of significant differences in these amino acids between infected and non-infected squash. Furthermore, B. tabaci has been shown to convert a large portion of the soluble phloem carbohydrates, predominantly sucrose, to the disaccharide, trehalulose -an insect haemolymph sugar composed of glucose and fructose 60 . Since sucrose concentrations have been equally minimal in both PRSV-infected and non-infected squash, the other predominant sugars in squash may have been nutritionally less relevant to whitefly physiology. www.nature.com/scientificreports www.nature.com/scientificreports/ Deceptive signaling or attraction leading to short-term probing by vector on a NP virus infected plant and rapid dispersal thereafter is considered ideal for NP virus spread 2 . The probing behavior may or may not always be consistent with nutritional quality of plants, which appear to be increased, unaltered or reduced in NP virus infected plants 5,8 . Although these interactions appear to be context-specific, majority of NP viruses remain to be studied to draw any robust generic trends. Mauck and co-authors in their recent book chapter 7 highlighted differences between Potyviridae (which have mostly neutral and positive effects on plant quality) and Bromoviridae (which have mostly neutral and negative effects on plant quality) in the NP viruses. The present study supports this observation, in that a strong positive relationship between a NP virus and vector in our PRSV pathosystem can be evidenced by increased vector performance through enriched amino acid and carbohydrate profiles of squash. Such association may appear epidemically less significant in a conventional sense, but may serve as an alternate virus dispersal strategy in nature as discussed recently by Carr et al. 6 . For instance, increased reproductive potential of A. gossypii, in the present study, could facilitate its population buildup on PRSV-infected plants in a short time. This population surge can increase the chances of the early development of alate A. gossypii due to crowding and nutrient exhaustion 61 . Therefore, colonizing insects such as A. gossypii, due to its virus-modulated fitness benefits, may be able to effectively promote virus spread to greater distances and form large inoculum foci 6 . This alternate strategy may be the reason why A. gossypii has been reported to be one of the most efficient PRSV vectors 32 . Furthermore, no PRSV-mediated effects on B. tabaci development and fecundity suggest the targeted manipulation of host plant quality by PRSV-more relevant to vector and less relevant to non-vector development.
Concluding remarks. Our study demonstrated that PRSV selectively favors A. gossypii, but not B. tabaci fitness through host plant quality manipulation. The reduced emigration and increased fitness of A. gossypii on PRSV-infected plants can be attributed to increases in important amino acids: threonine, arginine, lysine, glycine, homocysteine and sugars: galactose, raffinose and cellobiose. Our findings shed light on a previously unexplored yet important pathosystem: PRSV-squash-A. gossypii and present a conclusive evidence of a strong positive correlation between a non-persistent virus and its vector. This relatively long-term favorable association between PRSV and A. gossypii could selectively favor virus spread. PRSV remains a chronic and severe threat to cucurbits in the southeastern US and elsewhere. The rapid A. gossypii population increase due to plant enrichment could lead to quicker dispersal and increased spatial spread of the virus. Increased alatae production could favor spread of virus farther and quicker. Perhaps, such spread is more conducive for a virus such as PRSV with a relatively restricted host range as opposed to viruses such as CMV with a broad host range. It is indeed obvious that additional NP virus pathosystems need to be investigated before any robust conclusions on virus-induced effects can be drawn, but this study has documented a previously unexplored pathosystem demonstrating virus-mediated selective manipulation of vector over non-vector insect via alteration in host plant quality.

Material and Methods
Host plants and PRSV inoculation. Two squash (cv. Goldstar) seeds per pot were planted using MiracleGro compost and grown in the greenhouse in whitefly-proof cages [47.5(l) × 47.5(w) × 93(h) cm] at 25 °C, 60% RH, and 16 L:8D photoperiod. Plants were thinned out one week after planting to retain one healthy seedling per pot. Two-week-old seedlings were mechanically inoculated with PRSV using 0.01 M phosphate inoculation buffer (pH 7). One-gram tissue of PRSV infected squash leaf was flash frozen with liquid nitrogen and mashed with 10 ml of 0.05 M phosphate buffer in a sterile mortar. The resulting inoculum was mixed with 0.1 g of each carborundum and Celite ® (Sigma Aldrich Inc., Darmstadt, Germany). Additional 0.1 g carborundum was dusted on a newly emerged leaf to induce abrasion. The sterile cheesecloth containing PRSV inoculum was rubbed lightly and spread over the entire leaf surface. Ten min following the inoculation, the leaves were lightly sprinkled with water to avoid excessive mechanical injury. Newly emerged leaves of inoculated plants showed intense mosaic pattern on the leaf lamina at two weeks post inoculation.
The infection status of PRSV-inoculated plants (n = 10) was confirmed using reverse transcription PCR. The total RNA from 100 mg symptomatic leaf tissue was extracted using RNeasy Plant Mini Kit (Qiagen Inc., USA).
Initially, cDNA was synthesized from each RNA sample using GoScript ® reverse transcription system (Promega www.nature.com/scientificreports www.nature.com/scientificreports/ Just like A. gossypii, B. tabaci settling assay was conducted using PRSV-infected and non-infected squash. The arena to test B. tabaci settling was more simple and consisted of a plastic cylinder (150 mm diameter × 310 mm height) closed with a 15-cm diameter petriplate at the top 62 . The arena was sectioned at 9 cm from the top with two narrow slits (each 5-mm wide × 70 mm long) exactly opposite to each other. The same-sized leaves intact to the each PRSV-infected and non-infected plants were positioned inside the arena using a rubber foam strip (1.9 mm wide × 11.1 mm thick). One hundred non-viruliferous whiteflies maintained on cotton were aspirated into a 10-ml glass vial (VWR, Radnor, PA) and were released at the bottom of the arena. The number of whiteflies settling on each plant was recorded after 24 h. The experiment was replicated twice with 10 arenas set up each time (n = 20) with previously unused infected and non-infected replicate plants under laboratory conditions (25 °C; 12 h L:12 h D).
A. gossypii emigration and immigration assays. No-choice emigration and immigration assays were conducted following the procedure of Eigenbrode et al. 1 , with slight modifications. The emigration arena consisted of a 25-cm flat wooden floor, on which fifty adult apterae, starved for 1 h, were released on a leaf from either a PRSV-infected or non-infected intact squash plant at one end and allowed to disperse to the other end. For immigration assays, 50 adult apterae starved for 1 h were released at the opposite end and allowed to immigrate towards a leaf from either PRSV-infected or non-infected intact squash plant. For both assays, numbers of emigrating and immigrating aphids were recorded and removed at every 10 min for 1 h. For each bioassay, experiment was repeated twice with 10 replicate plants (n = 20).

A. gossypii and B. tabaci life history traits.
A single randomly selected apterous A. gossypii adult was placed using a camelhair paintbrush on the underside of a leaf of the 6-week-old PSRV-infected and non-infected squash plants and confined with a clip cage. Each plant was fixed with two cages, and thirteen plants per treatment were used for a total of 26 replications per treatment. After 24 h of confinement, the females were removed, and only one neonate nymph was retained inside each cage. The life history of this single aphid was monitored throughout the entire life cycle. The nymphs laid by the subsequent generation adult were counted and removed daily. Fitness parameters such as nymphal period, adult period, pre-reproductive and reproductive periods, longevity, total fecundity and intrinsic rate of increase were estimated. The intrinsic rate of increase (r m ) for each aphid was calculated using the equation of Wyatt & White 63 : where N d is the number of nymphs produced during reproductive period, and "d" is the pre-reproductive period in days. B. tabaci development time (egg-adult) and fecundity was studied on 6-week-old PRSV-infected and non-infected plants. Two clip cages each containing two adult whiteflies post-48h mating period were set up on each of ten PRSV-infected and non-infected plants. Whiteflies were removed after an oviposition period of 48 h and six randomly selected eggs per clip cage were retained. At least one surviving nymph per clip cage was monitored until adult emergence (egg-adult development). One randomly selected emerged adult per clip cage was moved to cotton plant soon after adult emergence and its fecundity was recorded. Soluble carbohydrate analysis. Between 8 and 15 mg of lyophilized leaf tissues from PRSV-infected and non-infected plants were used for the soluble carbohydrate analysis (n = 5). The analysis was carried out at the Complex Carbohydrate Research Center of University of Georgia. The dried samples were ground and suspended in 1 ml of water. The suspensions were vortexed briefly and filtered through glass wool. 500 μl of the sample suspension was used for the glycosyl composition analysis and the remaining was used for the High-Performance Anion-Exchange Chromatography coupled with Pulsed Electrochemical Detection (HPAEC-PAD) analysis. Glycosyl composition analysis was performed by combined gas chromatography/mass spectrometry (GC/MS) of the per-O-trimethylsilyl (TMS) derivatives of the monosaccharide methyl glycosides. These glycosides were produced from the sample by acidic methanolysis as described previously by Santander et al. 64 . This technique quantified the monosaccharides viz. ribose, arabinose, rhamnose, fucose, xylose and mannose in squash samples. The samples were analyzed by HPAEC-PAD using a Dionex ICS3000 system equipped with a gradient pump, an electrochemical detector, and an autosampler. The samples were separated using a Dionex CarboPac PA20 (3 × 150 mm) analytical column with an amino trap. The HPAEC-PAD analyzed mono-and di-saccharides such as galactose, glucose, fructose, raffinose and cellobiose.
Free amino acid analysis. Free amino acids from PRSV-infected and non-infected plants were analyzed at the Molecular Structure Facility at the University of California, Davis (n = 5). Leaf tissue (150 mg) was flash frozen with liquid nitrogen and ground with sterilized mortar and pestle. The ground tissue was suspended in 600 μl of water: chloroform: methanol (3:5:12 v/v) extraction buffer. The resulting suspension was transferred to a 2 ml micro-centrifuge tube (Eppendorf North America, Hauppauge, NY) and was centrifuged at 14,000 rpm for 2 min. The supernatant was transferred to a new 2 ml micro-centrifuge tube. The pellet from centrifugation was re-suspended with 600 μl of extraction buffer and re-centrifuged as specified above. The 300 μl of chloroform and 450 μl of water was added to the combined supernatants. The suspension was centrifuged at 14,000 rpm for 2 min and the upper water: methanol phase was carefully transferred to a new 2 ml micro-centrifuge tube. The content was dried for about 3 h using a SpeedVac (Thermo Fisher Scientific, USA) and the resulting pellet was stored at −20 °C until further processing.