5 DISCUSSION
Assessing interactions between biotic stressors and plants in food webs is critical to understand the dynamics of these interactions. Plant responses to stressors are often specific to the attacker, and both phytochemical responses and plant nutritional status can affect susceptibility to specific stressors (van Geem, Gols, Raaijmakers, & Harvey, 2016; Shikano, 2017). We show plants responded in complex ways to unique biotic stressors, including a piercing-sucking herbivore (A. pisum ), a chewing herbivore (S. lineatus ), and a virus (PEMV). Our study is among the first to assess how the order of attack, and diversity of stressors, mediate the defensive responses of plants and plant nutritional status (see also Vos, Moritz, Pieterse, & Van Wees, 2015). Our results show that plants traits varied in response to the type of attacker, the number of stressors, and their order of arrival. Moreover, we show that assessment of multiple gene transcripts, phytohormones, and plant nutrients provides a more comprehensive perspective on mechanisms driving plant-insect-pathogen interactions than any isolated response.
We found PEMV caused broad defensive responses in P. sativum by inducing specific gene transcripts and phytohormones (Figs. 2, 3 & 4). Biotropic pathogens such as PEMV are known to activate salicylic acid signaling (Singh, Swain, Singh, & Nandi, 2018; Chisholm, Sertsuvalkul, Casteel, & Crowder, 2018). This was reflected by increased expression of the ICS1 biosynthesis gene, increased salicylic acid hormone levels, and increased expression of the downstream defensive transcriptPR1 when PEMV was present. However, effects of PEMV were not limited to salicylic acid, as PEMV induced gene transcripts often associated with biosynthesis of abscisic acid (AO3 ), and giberrellic acid (GA2ox ), while also affecting defense response genes that occur downstream from induction of these hormones (PR1, DRR230 , PsLectin ). However, PEMV alone can not induce gene transcript associated with biosynthesis of JA (LOX2 ), but PEMV infection after weevil herbivory induced LOX2 accumulation. Similar results in P. sativum have been observed in response to fungal infection by Mycosphaerella pinodes and Phoma koolunga , where infections induce defense related genes across multiple signaling pathways (Fondevilla et al., 2011; Tran, You, & Barbetti, 2018). However, increased expression of LOX2 and AO3 gene transcripts (Fig. 2) were not reflected by increased levels of jasmonic acid or abscisic acid (Fig. 4). This suggests that measuring phytohormones, or gene transcripts, in isolation may fail to reveal more complex pathways by which plants respond to stress (Kazan & Lyons, 2014).
While PEMV had broad effects on plants, S. lineatus attenuated these responses. When S. lineatus was present, before or after PEMV, the expression of three out of four biosynthesis gene transcripts (with the exception of LOX2 ) in response to infectious aphids were comparatively weaker (Fig. 2). Increased expression of LOX2may be due to S. lineatus inducing expression of defensive transcripts associated with JA-mediated chewing herbivore attack and the effect was further enhanced by PEMV infection after S. lineatusfeeding (Fig. 2). Effects of PEMV on plant defense genes (PR1 ,DDR230 , PsLectin ) were also affected by S. lineatus , but varied with attack order. Overall, in this study the order of attack seemed to have stronger effects on downstream plant defense gene transcripts than on hormone biosynthesis gene transcripts.
While S. lineatus increased expression of two (LOX2 andGA2ox ) of the four biosynthesis gene transcripts studied when PEMV was not present, expression of LOX2 was enhanced when PEMV was also present (Fig. 2). In contrast, PEMV caused decreased expression of two genes (PR1 , DRR230 ) that were induced by S. lineatus was present alone (Fig. 2). For plants attacked first by either PEMV or S. lineatus , we observed the strongest evidence for mutual antagonism at the gene transcript level rather than for phytohormones (Figs. 2-4). Our study shows interactions among a set of stressors can vary based on attack order. Here, we found that S. lineatus feeding following PEMV infection inhibited plant defense (mutual antagonism), while LOX2 expression was enhanced when PEMV infection followed S. lineatus feeding (synergistic effect). While the first effect is in line with studies showing “mutual antagonism” between chewing herbivores and biotropic pathogens (Thaler, Agrawal, & Halitschke, 2010; Vos et al., 2015), our study suggests the order of attack can lead to variation along a spectrum from antagonism to enhancement.
Our results provide evidence that the order of arrival of biotic stressors on plants can play a crucial role in determining plants’ response to these attackers. While mutual antagonism between S. lineatus and PEMV was common, for some genes these effects only occurred when S. lineatus attacked first, and for others whenS. lineatus attacked second (Figs. 2-4). Mutual antagonism has most often been studied as effects of a prior attacker affecting a subsequent attacker, such as when a herbivore alters gene activation or phytohormones in ways that attenuate performance of subsequent attackers (Kessler & Halitschke, 2007; Erb, Robert, Hibbard, & Turlings, 2011; Stam, Mantelin, McLellan, & Thilliez, 2014; Huang et al., 2017). However, our results suggest that a second attacker may also mitigate defensive responses against the first attacker in ways that might affect plant defense and propagation of pathogens. For example, we show that plants infected by PEMV had decreased defenses when subsequently attacked by S. lineatus (Fig. 3), which should promote PEMV replication. Moreover, our results suggest that, PEMV infection induces pathogen defense and S. lineatus inhibits that if they appear on plants after the infection has been established. This may be more strongly expressed as variation in gene transcripts rather than hormone levels, a result that has similarly been seen in Arabidopsis in response to pathogen infection (Anderson et al., 2004).
Mutual antagonism in plant signaling pathways has most commonly been examined in regard to tradeoffs between jasmonic acid and salicylic acid. Our results show these tradeoffs extend to other signaling pathways. For example, jasmonic acid exhibits antagonism with abscisic acid in Arabidopsis following attack from Fusarium oxysporum (Anderson et al., 2004). Mutual antagonism between jasmonic acid and gibberellic acid, and jasmonic acid and abscisic acid, have also been reported (Yang, Yang, & He, 2013; Okada et al., 2015; Liu & Hou, 2018). For example, jasmonic acid facilitates defense over growth by repressing degradation of DELLA protein in rice andArabidopsis , but elevated DELLA proteins interfere with the gibberellic acid pathway by binding to growth promoting transcription factors associated with gibberellic acid signaling (Yang et al., 2012, 2013; Okada et al., 2015). Antagonistic relationships between giberellic acid and abscisic acid have also been reported in both mono and dicot plants and regulated by various transcription factor regulators in response to diverse environmental cues (Liu & Hou, 2018). However, antagonisms between salicylic acid and abscisic acid may actually lead to synergism between jasmonic acid and abscisic acid, where elevated abscisic acid levels following infection with Pseudomonas syringae induce jasmonic acid in Arabidopsis , which in turn limits the levels of salicylic acid (Fan, Hill, Crooks, Doerner, & Lamb, 2009). Overall, these results suggest that a broad examination of genes and hormones are needed to elucidate pathways underlying plant-insect-pathogen interactions in P. sativum and other plants.
Our results suggest mutual antagonism may also occur among defense gene transcripts that are associated with a single signaling pathway. For example, the induction of PR1, a salicylic acid-responsive gene, was mitigated by S. lineatus attack after PEMV infection, as may be expected with antagonism between jasmonic acid and salicylic acid. However, the expression of ICS1, another gene associated with the biosynthesis of salicylic acid, was not responsive to S. lineatus . This has been seen in other studies where ICS1 was not induced by caterpillar feeding although other genes associated with salicylic acid were (Onkokesung, Reichelt, van Doorn, Schuurink, & Dicke 2016). These results suggest that a plant’s response to multiple stressors is unlikely to result from simple crosstalk but rather from interactions among multiple signaling pathways that may exhibit complex responses.
In addition to affecting plant gene expression and phytohormones, plant pathogens such as viruses can also alter nutritional quality of their host plants in ways that affect vectors (Mauck, Bosque‐Pérez, Eigenbrode, Moraes, & Mescher, 2012; Wang, Senthil-Kumar, Ryu, Kang, & Mysore, 2012; Patton et al., 2019). Similarly, non-vector herbivores may strongly affect the quantity and quality of plant nutrients (Ángeles-López, Rivera-Bustamante, & Heil, 2016). For example,pepper golden mosaic virus (PGMV) infection in Capsicum annuum increased levels of the amino acids proline, tyrosine, valine but decreased levels of histidine and alanine. In the same system, the greenhouse whitefly, Trialeurodes vaporarioum , reversed the levels of these amino acids (Ángeles-López et al., 2016). Arrival of S. lineatus before PEMV infection suppressed the amount of total amino acids in peas, while enhanced amino acid level was detected if S. lineatus damaged peas after PEMV infected was established. This suggests the intriguing possibility that antagonism between a pathogen and non-vector herbivore can occur at the level of amino acid production in plants.
Overall, our study provides example of complex interactions between a vector-borne plant pathogen and a non-vector herbivore that varies from antagonism to enhancement and manifest as changes in plant gene transcripts, phytohormones levels, and plant nutrients. However, we show that assessing the order of attack is necessary to best understand the complexity and mechanisms of plant-insect-pathogen interactions. Moreover, our study suggests complete pathways must be characterized as differences are evident even when a few transcripts and metabolites are analyzed., often measured with associated gene transcripts (Bedini, Mercy, Schneider, Franken, & Lucic-Mercy, 2018; Ángeles-López el al., 2016; Shi et al., 2019), may fail to capture mechanisms by which plants interact with multiple stressors. Our results demonstrate both the order of arrival, and the diversity of interactions, determine plant responses to stress through the combined action of defense gene activation, phytohormone accumulation, and modification of plant nutrients. Characterizing the pathways by which plants respond to single and multiple stressors, with varying attack order, can shed light on the mechanisms that shape food web interactions among plants, herbivores, and pathogens.