Effects of Root-Colonizing Fluorescent Pseudomonas Strains on Arabidopsis Resistance to a Pathogen and an Herbivore

ABSTRACT

Rhizobacteria in the genus Pseudomonas can enhance plant resistance to a range of pathogens and herbivores. However, resistance to these different classes of plant antagonists is mediated by different molecular mechanisms, and the extent to which induced systemic resistance by Pseudomonas can simultaneously protect plants against both pathogens and herbivores remains unclear. We screened 12 root-colonizing Pseudomonas strains to assess their ability to induce resistance in Arabidopsis thaliana against a foliar pathogen (Pseudomonas syringae DC3000) and a chewing herbivore (Spodoptera littoralis). None of our 12 strains increased plant resistance against herbivory; however, four strains enhanced pathogen resistance, and one of these (Pseudomonas strain P97-38) also made plants more susceptible to herbivory. Phytohormone analyses revealed stronger salicylic acid induction in plants colonized by P97-38 (versus controls) following subsequent pathogen infection but weaker induction of jasmonic acid (JA)-mediated defenses following herbivory. We found no effects of P97-38 inoculation on herbivore-relevant nutrients such as sugars and protein, suggesting that the observed enhancement of susceptibility to S. littoralis is due to effects on plant defense chemistry rather than nutrition. These findings suggest that Pseudomonas strains that enhance plant resistance to pathogens may have neutral or negative effects on resistance to herbivores and provide insight into potential mechanisms associated with effects on different classes of plant antagonists. Improved understanding of these effects has potentially important implications for the use of rhizobacteria inoculation in agriculture.

IMPORTANCE Plant-associated microbes have significant potential to enhance agricultural production, for example, by enhancing plant resistance to pathogens and pests. Efforts to identify beneficial microbial strains typically focus on a narrow range of desirable plant traits; however, microbial symbionts can have complex effects on plant phenotypes, including susceptibility and resistance to different classes of plant antagonists. We examined the effects of 12 strains of Pseudomonas rhizobacteria on plant (Arabidopsis) resistance to a lepidopteran herbivore and a foliar pathogen. None of our strains increased plant resistance against herbivory; however, four strains enhanced pathogen resistance, and one of these made plants more susceptible to herbivory (likely via effects on plant defense chemistry). These findings indicate that microbial strains that enhance plant resistance to pathogens can have neutral or negative effects on resistance to herbivores, highlighting potential pitfalls in the application of beneficial rhizobacteria as biocontrol agents.

INTRODUCTION

Bacteria in the genus Pseudomonas are among the most abundant members of the rhizosphere and frequently exhibit plant-beneficial properties, including growth promotion, enhanced nutrient availability, and suppression of soilborne pathogens (1). These properties are highly desirable in agricultural settings, where the targeted application of rhizobacteria is increasingly common and has significant potential to reduce fertilizer and pesticide usage (2), motivating efforts to identify rhizobacterial strains with relevant effects. Beneficial Pseudomonas strains have previously been identified by screening large collections of isolates for specific traits, such as the production of growth-stimulating factors or antimicrobial activity against specific pathogens (3–5); however, such efforts have typically focused on a relatively narrow range of potential effects on host plants. It is increasingly clear that rhizobacteria can influence a broad range of plant traits, including traits that influence plant resistance to herbivores, as well as their interactions with beneficial insects (6–9), but relatively little work has explored effects of rhizobacteria on host plant traits that mediate plant-insect interactions.

In addition to demonstrating that association with rhizobacteria can alter plant traits that mediate interactions with other organisms, previous work revealed that positive effects of rhizobacteria on resistance against one attacker sometimes come at the cost of increased susceptibility to others. For instance, colonization of Arabidopsis thaliana roots by Pseudomonas simiae WCS417 increased resistance against the foliar pathogen Pseudomonas syringae and the chewing herbivores Mamestra brassicae and Spodoptera exigua but reduced resistance against the green peach aphid (Myzus persicae) (10–13). Similarly, Pseudomonas sp. strains CH267 and CH229 enhanced resistance against the chewing herbivore Trichoplusia ni but rendered plants more susceptible to Pseudomonas syringae (14). Such trade-offs in susceptibility and resistance to different antagonists may have potentially important implications for the use of rhizobacteria in agriculture. However, we currently have limited insight into the prevalence of such divergent effects or the underlying mechanisms responsible for them.

In some cases, rhizobacterial effects on different antagonists might be mediated by entirely different plant traits, while in others, effects on a single trait, such as levels of key nutrients (15), may have divergent effects on different plant antagonists. For example, Pseudomonas can affect plant nutrition by altering the uptake of key nutrients such as nitrogen (16) or the rate of photosynthesis (17), which could differentially affect the performance of both herbivores and pathogens (18, 19). There is also considerable evidence that rhizobacteria can modify plant defense traits (20), which may have divergent implications for interactions with different antagonists (14). For example, salicylic acid (SA) and jasmonic acid (JA)/ethylene (ET) are key defense-related hormones that play central roles in the induction of plant defenses against biotrophic/hemibiotrophic pathogens and herbivores, respectively; however, the JA and SA pathways engage in negative cross talk, which may play a role in tailoring defenses to specific antagonists (21) and are frequently exploited by attackers that manipulate plant defense for their own benefit (22). Effects of rhizobacteria on these pathways might plausibly cause or intensify antagonistic interactions between different defense pathways, with differential outcomes depending on the antagonist. Furthermore, previous work has established that Pseudomonas strains can prime plant defense pathways for enhanced responses by triggering induced systemic resistance (ISR), which is primarily mediated by the JA and ET phytohormone pathways (10) but also occurs via the SA pathway (23) or combinations of other pathways (24).

To gain insight into the effects of Pseudomonas on different attackers, we investigated the effects of 12 taxonomically diverse members of the Pseudomonas fluorescens group on the resistance of A. thaliana to two distinct attackers. Our primary goals were to (i) determine whether ISR is frequently induced by Pseudomonas strains, (ii) assess potential differences in ISR against pathogens (Pseudomonas syringae DC3000) and herbivores (Spodoptera littoralis), and (iii) explore the molecular mechanisms responsible for any observed variation in ISR. We found that a third of our tested Pseudomonas strains increased resistance against the pathogen P. syringae but not against the chewing herbivore S. littoralis. Moreover, we observed that plants treated with Pseudomonas sp. strain P97-38, a strain that consistently induced resistance against P. syringae, were more susceptible to S. littoralis. We further found that this susceptibility is likely due to attenuated antiherbivore defenses rather than increased host plant quality for S. littoralis. Our study demonstrates that defense priming by a rhizobacterium can have contrasting effects on the resistance of a plant to different attackers with possible implications for the application of such bacteria as biocontrol agents in agriculture.

RESULTS

Taxonomically diverse Pseudomonas strains frequently confer resistance to infection by P. syringae but not to herbivory by S. littoralis.

Root inoculation with each of our 12 strains invariably resulted in either positive or neutral effects on pathogen resistance. In two independent experiments, the strains Pseudomonas cerealis L1-45-08, Pseudomonas protegens BRIP, Pseudomonas sp. strain CMR12a, and Pseudomonas sp. P97-38 significantly reduced the intensity of P. syringae infection (bacterial cell counts via luminescence emission) by 1.8- to 16.9-fold in leaves compared to that in the noninoculated control (Fig. 1). The greatest reduction in pathogen load (16.9-fold) was observed for strain P97-38 (Fig. 1, experiment 1). To test whether this increased resistance might be due to the direct production of antimicrobials by these strains rather than plant-mediated effects, we performed an in vitro antagonism assay. We found that P. protegens BRIP and Pseudomonas sp. CMR12a, but not P. cerealis L1-45-08 and Pseudomonas sp. P97-38, were able to antagonize P. syringae in vitro (see Fig. S4 in the supplemental material). We thus cannot rule out the possibility that antibiosis contributed to the disease suppressive effect of P. protegens BRIP and Pseudomonas sp. CMR12a. However, BRIP and CMR12a were never (0 of 6 plants) and P97-38 was only occasionally (2 of 12 tested plants) found on true leaves, suggesting that the observed effects are rather due to priming of plant defense (induced systemic resistance).

FIG 1.

Various root-colonizing Pseudomonas strains increase Arabidopsis thaliana resistance against P. syringae. A. thaliana Col0 that was either root inoculated with the indicated Pseudomonas strain (black bars) or with buffer (white bar) was spray infected with P. syringae DC3000 lx, and pathogen growth was measured 4 days postinfection. Data from two independent experiments are shown. Values show the mean (± standard error [SE]) log-transformed photon counts per second normalized to the amount of leaf tissue (n = 17 to 21 biological replicates). *, P < 0.05 versus noninoculated control (generalized least-squares model).

In contrast to the results observed for P. syringae, none of our strains increased resistance against S. littoralis. Rather, root inoculation with Pseudomonas sp. P97-38 and Pseudomonas orientalis L1-3-08 had a positive effect on larval performance (Fig. 2). This effect was consistent for P97-38 but not for L1-3-08 (compare Fig. 2 and Fig. S1). Although larvae gained on average more weight on P97-38-inoculated plants (experiment 1, 43%; experiment 2, 36%), we found only a weak trend toward higher consumption of plant tissue by larvae (see Fig. S2).

FIG 2.

Effects of root-colonizing Pseudomonas strains on the performance of Spodoptera littoralis feeding on Arabidopsis thaliana Col0. Plants that were root inoculated with either the indicated Pseudomonas strain (black bars) or with buffer (white bar) were infested with one neonate S. littoralis larva. (A and B) Larvae were collected and dry weights recorded 6 days postinfestation. Two experiments were performed with different sets of strains. Bars show the mean (±SE) dry weights from 20 to 34 biological replicates. *, P < 0.05 versus noninoculated control (generalized least-squares model).

In addition to performing better on plants inoculated with P97-38, larvae also preferred to feed on these plants over noninoculated plants. In a dual-choice assay, 75% of all larvae were found on P97-38-inoculated plants 2 h after release; this proportion increased to 83.3% after 24 h (Fig. 3).

FIG 3.

Spodoptera littoralis larvae prefer Pseudomonas sp. P97-38-inoculated plants. Two 3rd-instar larvae were released into a box with a P97-38-inoculated and a noninoculated control plant in a dual-choice setup. Bars represent the total number of caterpillars found on control and P97-38-inoculated plants after 2 h and 24 h. Sixteen boxes were used for the experiment. *, P < 0.05 between treatments (chi-square test).

Pseudomonas sp. P97-38 enhances the induction of SA-dependent defenses after infection but reduces the induction of JA-dependent defenses after herbivory.

Consistent with an increased resistance toward pathogenic P. syringae, we observed higher levels of SA in both pathogen-infected and mock-infected P97-38-inoculated plants than in control plants (Fig. 4a). Bacterial root inoculation also slightly increased JA content in both treatments (mock and pathogen infected), although JA levels were consistently lower than SA levels in this experiment (Fig. 4b). In contrast, JA was strongly induced by herbivory, and P97-38-inoculated plants showed a nonsignificant trend toward smaller amounts of JA in response to S. littoralis infestation, in line with the observed increase in susceptibility of P97-38-inoculated plants to this generalist herbivore (Fig. 4d). SA levels in mock-infested and infested plants in the herbivore assay were similar and unaffected by P97-38 (Fig. 4c).

FIG 4.

Salicylic acid and jasmonic acid levels in Pseudomonas sp. P97-38-inoculated A. thaliana plants. SA (a) and JA (b) levels 24 h after P. syringae infection (gray bars) and mock infection (white bars). SA (c) and JA (d) levels in uninfested plants (white bars) and 24 h after infestation with 5 neonate S. littoralis larvae (dark gray bars). Bars show the means (±SEs) from 12 biological replicates. *, P < 0.05 for effects of the factors rhizobacterium (P97-38-inoculated versus noninoculated control) and infection/infestation (mock infected versus infected or uninfested versus infested) according to a two-way ANOVA or a generalized linear model.

We also found that P97-38 had a significant effect on the composition of glucosinolates (permutational multivariate analysis of variance [PERMANOVA], 999 permutations, P value = 0.032) and particularly attenuated the content of indolic glucosinolates (Fig. 5; Table S1). This result is consistent with the hypothesis that P97-38 inoculation increases plant susceptibility by reducing expression of JA-dependent defenses.

FIG 5.

Glucosinolate levels in Pseudomonas sp. P97-38-inoculated A. thaliana plants. Total aliphatic (a) and indole (b) glucosinolates in mock- and P97-38-inoculated plants under uninfested conditions and 6 days after infestation with 1 neonate S. littoralis larva. Each box represents 17 to 21 biological replicates. *, P < 0.05 for effects of the factors rhizobacterium (P97-38-inoculated versus noninoculated control) and infestation (uninfested versus infested) according to a two-way ANOVA.

The induced susceptibility of P97-38-inoculated plants to S. littoralis is not explained by differences in plant nutrition.

S. littoralis larvae exhibited a preference for P97-38-inoculated plants (Fig. 3) and gained more weight on these plants without consuming significantly more tissue (Fig. 2 and S2), which could indicate differences in the nutritional content of leaves. Chemical analyses of relevant plant nutrients revealed that P97-38-inoculated plants contained similar amounts of soluble sugars (sucrose, glucose, and fructose) but less protein than noninoculated control plants (Table 1).

TABLE 1.

Nutrient content in P97-38-inoculated and noninoculated A. thaliana plants that were infested with S. littoralis or uninfested

Nutrient	Nutrient content (μg or mg/g [dry weight])a
	Uninfested		S. littoralis infested
	Control	P97-38	Control	P97-38
Fructose	15.80 ± 3.82	23.09 ± 2.67	22.28 ± 2.91	18.56 ± 2.50
Glucose	93.60 ± 9.37	118.16 ± 10.22	186.31 ± 15.70	136.97 ± 13.86
Sucrose	184.73 ± 12.21	190.22 ± 15.63	177.34 ± 9.64	163.91 ± 11.40
Total sugars	294.13 ± 23.17	331.48 ± 27.19	385.93 ± 22.61	319.44 ± 27.12
Protein	24.55 ± 2.17b	17.59 ± 1.88b

^aNutrient content represents the mean (±SE) from 7 to 11 (uninfested) and 17 (S. littoralis-infested) biological replicates.

^bValues are in milligrams/gram (dry weight). There was a significant difference between noninoculated and P97-38-inoculated plants according to a Welch two-sample t test (P value < 0.05).

DISCUSSION

Our findings reveal that association with Pseudomonas rhizobacteria can have antagonist-dependent effects on plant resistance and that rhizobacteria-mediated resistance against pathogens is not a consistent predictor of herbivore resistance. Indeed, none of the 12 Pseudomonas strains we examined enhanced the resistance of Arabidopsis thaliana plants to herbivory by Spodoptera littoralis. However, several strains caused increased resistance to infection by the foliar pathogen Pseudomonas syringae, and one of these (Pseudomonas sp. P97-38) also rendered plants more susceptible to herbivory. Previous studies on ISR have primarily focused on resistance to different bacterial and fungal pathogens, and relatively few beneficial microbes have been tested for the ability to increase both pathogen and herbivore resistance, particularly under the same experimental conditions (25, 26). While some Pseudomonas and Bacillus strains were previously shown to induce resistance against pathogens and herbivores, our findings indicate that the two traits do not necessarily cooccur.

In addition to P97-38, another strain that consistently enhanced resistance against P. syringae in our experiments is Pseudomonas sp. CMR12a, which was previously shown to induce ISR against Magnaporthe oryzae in rice and Rhizoctonia solani AG2-2 in bean (27). Meanwhile, P. protegens CHA0 did not provide protection against P. syringae, which is somewhat surprising as this strain was previously shown to prime Arabidopsis against Peronospora (Hyaloperonospora) parasitica (28), grapevine against Botrytis cinerea (29), and tobacco against tobacco necrosis virus (30) but consistent with unpublished results that showed no effect on P. syringae resistance (28).

Because of its observed effects on both pathogen and herbivore resistance, we explored the effects of inoculation with Pseudomonas. sp. P97-38 on plant defense responses and other relevant biochemical traits. Consistent with its effects on resistance to both classes of antagonists, we observed stronger SA induction in plants colonized by P97-38 (versus controls) following subsequent pathogen infection (Fig. 4) but weaker induction of JA-mediated defenses following herbivory (Fig. 4 and 5). The SA-dependent defense pathway has repeatedly been shown to be essential for the defense against DC3000; for example, loss-of-function mutants impaired in SA biosynthesis are more susceptible to DC3000 (31), and plants that have higher constitutive or induced levels of SA are more resistant to infection (32, 33). JA signaling, on the other hand, is crucial for the defense against S. littoralis: an A. thaliana mutant impaired in JA signaling shows increased susceptibility against S. littoralis (34). Effects of beneficial microbes on phytohormone levels were previously reported (30, 35, 36), and microbial inoculation can alter plant resistance toward herbivores. For example, treatment of cotton plants with Bacillus pumilus INR-7 or mixes of different Bacillus strains resulted in a strong increase in JA levels, which was associated with reduced Spodoptera exigua performance through reduced larval weights, lower pupation rates, and increased mortality (37).

JA signaling was previously shown to regulate a broad range of induced plant defenses against herbivory (38). In the present study, a small reduction in expression of JA after herbivory in inoculated plants relative to that in controls may account for a significant reduction in the production of defensive secondary metabolites. P97-38 had a significant effect on the overall glucosinolate composition, which is predominantly regulated by the JA pathway (11, 39). Specifically, P97-38-inoculated plants contained smaller amounts of indolic glucosinolates, suggesting that P97-38 attenuates chemical defenses in A. thaliana (Fig. 5). The production of indolic glucosinolates was previously shown to have a negative influence on the growth rate of S. littoralis on A. thaliana (40). Glucosinolates also play an important role in the resistance mechanisms induced by other Pseudomonas strains that prime defense against chewing herbivores. For example, Pangesti et al. found that P. simiae WCS417-induced resistance against the leaf-chewing herbivore M. brassicae in A. thaliana is partly explained by higher induction of aliphatic glucosinolates rather than indolic glucosinolates (11). In contrast, the increased mortality of Spodoptera exigua observed on A. thaliana plants inoculated with P. fluorescens SS101 was absent in a mutant line deficient in indolic glucosinolates, indicating that induction of indolic glucosinolates alone can be sufficient to alter resistance toward S. exigua (23). Our no-choice performance assay did not reveal increased resistance toward S. littoralis in P. fluorescens SS101-inoculated plants, suggesting that the SS101-induced effects are less potent against this species than against S. exigua. Interestingly, van de Mortel et al. found that glucosinolates might also play a role in SS101-induced resistance against P. syringae DC3000, as the ISR phenotype was lost in mutants deficient in either aliphatic or indolic glucosinolates (23). While we did not assess changes in glucosinolates in response to P. syringae infection, we saw lower glucosinolate levels in uninfested P97-38-inoculated plants than in noninoculated control plants, suggesting that P97-38-induced resistance against P. syringae is probably not based on glucosinolates.

Despite the observed changes in JA-dependent defense, changes in plant nutrition could offer an alternative explanation for an increased herbivore performance. Herbivore performance is influenced by the nutritional quality of the leaf tissue consumed, and root-associated microbes can improve plant nutritional content by increasing the accessibility of important nutrients such as nitrogen (41), phosphorus (42), and sulfur (43). Both digestible carbohydrates and protein are highly relevant for larval development, but when feeding on plants, proteins, rather than carbohydrates, are a limiting factor (44, 45). We found that uninfested P97-38-inoculated plants had comparable levels of soluble sugars to those of noninoculated control plants but a lower protein content (Table 1). Hence, it seems unlikely that larvae feeding on P97-38-inoculated plants gained more weight due to the higher nutritional quality of the leaf material. Although larvae removed a similar percentage of leaf tissue on inoculated and uninoculated treatments (see Fig. S2 in the supplemental material), P97-38-inoculated plants tended to be larger on average than control plants (see Fig. S3), indicating that the total amount of tissue consumed by larvae might be greater in the inoculated treatment. Taken together, our results suggest that S. littoralis larvae grow better on P97-38-inoculated A. thaliana not as a result of more nutritious leaves, but rather as a result of reduced levels of JA-dependent defensive compounds in the leaves that allow higher leaf consumption.

We also observed a behavioral preference of S. littoralis larvae for P97-38-inoculated plants, with a significant majority of larvae being found on plants of the P97-38 treatment (Fig. 3). As the plant not chosen was typically left undamaged after 2 h, this preference likely reflects an initial choice based on olfactory or visual cues rather than a feeding preference based on gustatory cues. A strong initial preference could be explained by the visibly larger size of P97-38-inoculated plants during this experiment. A study investigating the effects of different mycorrhiza genotypes on S. littoralis preference for strawberry plants found larvae showed a preference for larger plants but otherwise did not distinguish between treatments (46). Choice for P97-38-inoculated plants could thus be driven by a visual cue correlating with plant size that larvae associate with a superior food source. Moreover, plant size could also be associated with olfactory cues, as larger plants likely emit higher quantities of plant volatiles, which play a role in host plant selection by lepidopteran larvae (47, 48). However, plant-associated Pseudomonas spp. were previously shown to alter plant volatile blends (49), and our results do not exclude the possibility that treatment with P97-38 modifies the plant volatile blend in a way that makes the plant more attractive to S. littoralis. Further work is necessary to disentangle plant size and a possible role for volatile compounds in driving the larval preference for inoculated plants observed in our study.

Contrasting effects by Pseudomonas on pathogen and herbivore resistance have recently been described in another study (14), which reported that treatment of A. thaliana with Pseudomonas sp. strains CH229 and CH267 led to a shift toward JA-dependent defenses, which increased resistance toward the chewing herbivore T. ni. When plants were challenged with P. syringae DC3000, however, this shift resulted in a lower induction of SA-dependent defenses, including a lower induction of the defense gene pathogenesis-related protein 2 (PR2), likely due to the antagonistic relationship between JA and SA, and increased plant susceptibility (14). While the present results do not provide direct evidence for antagonism between SA and JA, it is possible that P97-38 primes SA-dependent defenses and, this way, attenuates the activation of JA-dependent defenses via hormonal cross talk. It is notable, however, that three other strains that were found to increase resistance to P. syringae in this study did not affect herbivore resistance. Moreover, examples of other root-associated Pseudomonas strains also show that priming of either the JA or SA pathway does not necessarily result in trade-offs. Both P. simiae WCS417 and P. fluorescens SS101 increase the resistance against P. syringae and the chewing herbivore M. brassicae by priming JA/ET-dependent and SA-dependent defenses, respectively (23, 50, 51). On the other hand, P. simiae WCS417 inoculation caused increased susceptibility to the phloem-feeding herbivore Myzus persicae in A. thaliana (12) and the phloem-feeding herbivore Bemisia tabaci in Solanum lycopersicum (52). WCS417-induced susceptibility toward M. persicae could not be explained by defense gene expression, suggesting that this phenotype is caused by factors unrelated to defense such as nutrients. In the case of Bemisia tabaci, changes to plant defense, plant quality, and the composition of the rhizosphere community were offered as possible explanations for the observed susceptibility. Together with the results of our study, these studies reveal that plant-associated microbes can affect plant defenses in vastly different ways with variable outcomes depending on the ecological setting.

Researchers have long focused on the beneficial effects of rhizobacterial association on plant resistance against pathogens, but rhizobacteria can also influence plant interactions with other organisms, including insect herbivores. These broader ecological effects have potentially important implications for the targeted use of rhizobacteria in agriculture, but relatively few studies have examined effects of rhizobacterial inoculation on plant resistance against multiple antagonists, particularly under the same experimental conditions. Our findings suggest that positive effects on host-plant pathogen resistance may be common among plant-associated fluorescent Pseudomonas but that it is not necessarily accompanied by enhanced herbivore resistance. Furthermore, we identified at least one strain (Pseudomonas sp. P97-38) exhibiting divergent effects on pathogen and herbivore resistance that are potentially explained by its effects on SA- and JA-mediated plant defenses. The present findings contribute to an improved understanding of the outcomes of plant interactions with Pseudomonas and possible trade-offs in resisting attackers, which is necessary for the development of environmentally friendly biocontrol strategies.

MATERIALS AND METHODS

Bacterial strains and culture conditions.

Twelve root-colonizing Pseudomonas strains from different phylogenetic subgroups within the Pseudomonas fluorescens group (53) were used in this study (Table 2) along with the plant-pathogenic Pseudomonas syringae strain DC3000 lx. All Pseudomonas sp. strains were stored in glycerol stock solutions, with aliquots taken for use and grown on King’s B medium (54) agar plates supplemented with 25 μg ml⁻¹ kanamycin (Pseudomonas syringae DC3000 lx) or with 13 μg ml⁻¹ chloramphenicol, 40 μg ml⁻¹ ampicillin, and 100 μg ml⁻¹ cycloheximide (other Pseudomonas strains) (KB⁺⁺⁺) for 48 h at 24°C.

TABLE 2.

Pseudomonas strains used in this study

Strain	Subgroup within the Pseudomonas phylogeny according to Hesse et al. (53)	Origin	Demonstrated activitya	Reference(s)
Pseudomonas protegens CHA0	P. protegens	Tobacco, Switzerland	AF, IS, ISRP	64, 65
Pseudomonas protegens BRIP	P. protegens	Cyclops, Switzerland	AF, IS	65, 66
Pseudomonas sp. CMR5c	P. protegens	Red cocoyam, Cameroon	AF, IS	65, 67
Pseudomonas sp. CMR12a	P. protegens	Red cocoyam, Cameroon	AF, IS, ISRP	65, 67
Pseudomonas chlororaphis subsp. piscium DSM21509	P. chlororaphis	Intestine of European perch, Switzerland	AF, IS	65, 68
Pseudomonas chlororaphis subsp. aureofaciens LMG1245	P. chlororaphis	River Clay, The Netherlands	AF, IS	65, 69
Pseudomonas kilonensis P12	P. corrugata	Tobacco, Switzerland	AF	65, 70
Pseudomonas sp. P97-38	P. corrugata	Cucumber, Switzerland	AF	65, 71
Pseudomonas aridus R1-43-08r	P. fluorescens	Winter wheat, USA	AF	72, 73
Pseudomonas orientalis L1-3-08r	P. fluorescens	Winter wheat, USA	AF	72, 73
Pseudomonas cerealis L1-45-08r	P. fluorescens	Winter wheat, USA	AF	72, 73
Pseudomonas fluorescens SS101r	P. fluorescens	Wheat, The Netherlands	IS (weak), ISRP, ISRI	74
Pseudomonas syringae DC3000 lx	P. syringae		PP	56

^aAF, antifungal activity; IS, insecticidal activity; ISRP, induction of systemic resistance in plants against pathogens; ISRI, induction of systemic resistance in plants against insect herbivores; PP, plant pathogen.

For root inoculations in the pathogen and herbivore assays, Pseudomonas strains were inoculated in 40 ml King’s B liquid medium and grown for approximately 15 h at 24°C on a rotary shaker (180 rpm). Bacterial cells were washed twice by centrifuging at 1,697 rcf and subsequently resuspended in 10 mM MgSO₄, and finally adjusted to an optical density at 600 nm (OD₆₀₀) of 1 in 10 mM MgSO₄.

For root inoculation in the dual-choice assay, Pseudomonas sp. P97-38 was inoculated in 10 ml lysogeny broth (55) and grown overnight on a rotary shaker at 180 rpm and 24°C. Two hundred microliters of this culture was spread on KB (without antibiotics) and incubated for 24 h at 24°C. Bacterial cells were scraped from plates with 10 ml 10 mM MgSO₄, washed once, as described above, and finally adjusted to an OD₆₀₀ of 0.5 in sterile H₂O.

For leaf infections in the pathogen assay, P. syringae strain DC3000 lx was grown and suspensions were prepared as described above for P97-38 in the dual-choice assay with the exception that the LB was amended with 25 μg ml⁻¹ kanamycin. Suspensions used for A. thaliana infections were adjusted to an OD₆₀₀ of 0.05 (approximately 4.5 × 10⁷ cells/ml) with 10 mM MgSO₄ (0.02% Silwet L-77).

Plant growth conditions and bacterial inoculation.

Arabidopsis thaliana Col0 seeds were surface sterilized for 4 h using chlorine gas. For the pathogen and the herbivore assay, 10 seeds were placed on square plates containing half-strength Murashige & Skoog medium (including vitamins, 0.5% sucrose, pH 5.8; Duchefa Biochemie, The Netherlands) and stratified for 2 to 3 days at 3°C in the dark to synchronize seed germination. Plates were placed vertically and plants grown under controlled conditions in a climate chamber at 22°C and 60% relative humidity with a 12-h/12-h light (120 to 140 μmol s⁻¹ m⁻²)/dark period. After 9 days, seedlings were either root tip inoculated with 2 μl bacterial suspension prepared as described above or mock root tip inoculated with 2 μl 10 mM MgSO₄ (noninoculated control), as described by van de Mortel et al. (23). Fourteen-day-old seedlings were transplanted into 140 ml pots containing a sterile sand-soil (Substrat 2; Klasmann-Deilmann GmbH, Germany) mix (vol/vol = 1/1) that had been autoclaved twice (20 min at 121°C).

For the S. littoralis dual-choice assay, sterilized seeds were directly sown into pots containing the sterile sand-soil mix and grown as described previously under short-day conditions (8/16-h light/dark period). Nine-day-old seedlings were soil inoculated with 1.8 ml of a Pseudomonas sp. P97-38 suspension (OD₆₀₀ of 0.5) or mock inoculated with 1.8 ml sterile H₂O.

Pathogen infection assay.

Root-inoculated A. thaliana plants were challenged with a pathogen (P. syringae assay). For P. syringae infections, we used a transgenic strain that constitutively expresses the luxCDABE operon of Photorhabdus luminescens (56). Twenty-six-day-old A. thaliana plants root inoculated with bacterial strains or mock inoculated (noninoculated control) were spray infected with a suspension containing 4.5 × 10⁷ CFU/ml until runoff and kept under a lid for 24 h to ensure a high humidity conducive to infections. One experiment consisted of 13 treatments (12 bacterial strains and the noninoculated control) and 17 to 21 plants per treatment. Numbers per treatment varied as a result of differences in the number of available plants for each treatment. P syringae DC3000 lx cells were extracted 4 days postinfection by placing four leaves of each plant in a 2-ml Eppendorf tube containing 1 ml of 10 mM MgSO₄ (0.2% Silwet L-77) and shaking tubes for 15 min at 1,500 rpm. This method has been shown to be sufficient to extract bacterial cells from leaf tissue, and results were comparable to those obtained with grinding (57, 58). Two hundred microliters from each tube was transferred to a white 96-well plate (Greiner Bio-One, Germany), and bacterial growth was measured as a luminescent signal (photon counts per second) in a Tecan Infinite M1000 plate reader (Tecan, Switzerland). All tubes were weighed before and after the addition of leaf tissue to normalize photon counts to the amount of leaf tissue used. This experiment was performed twice.

Roots of three randomly selected plants from each treatment were collected in both pathogen infection assays and assessed for bacterial colonization. To determine root colonization, roots were weighed, placed into 30 ml of 10 mM MgSO₄, and shaken at 1,500 rpm for 5 min. Serial dilutions of the resulting suspensions were plated as 10-μl droplets in duplicates onto KB⁺⁺⁺, and colonies were counted after 24 to 48 h of incubation at 24°C depending on the growth rate of the respective strain. All strains colonized the roots at levels sufficient to trigger induced systemic resistance (>10⁵ CFU/g of root; Raaijmakers et al. [59] (data not shown). To ensure the absence of experimental Pseudomonas strains in noninoculated controls, we also assessed root colonization of three to six randomly selected noninoculated controls in several experimental assays. Pseudomonads were mostly absent in noninoculated controls; however, since we did not use a sterile system, pseudomonads were detected in some assays.

In vitro antagonism assay.

Pseudomonas strains were grown on KB⁺⁺⁺ agar plates or KB agar plates supplemented with kanamycin (P. syringae DC3000 lx) for 48 h at 24°C as described above and subsequently grown overnight in 40 ml KB liquid medium for approximately 13 h at 24°C on a rotary shaker (180 rpm). Cultures were adjusted to an OD₆₀₀ of 1 in KB liquid medium, and 50 μl of P. syringae DC3000 lx inoculum was evenly spread on half-strength KB and half-strength LB agar plates (without antibiotics). Subsequently, four 5-μl droplets of P. protegens BRIP, Pseudomonas sp. CMR12a, Pseudomonas sp. P97-38, or P. cerealis L1-45-08 were placed on each plate. Three replicates were prepared for each strain and medium. Agar plates were incubated at 24°C for 4 days, after which, the plates were photographed using a flatbed scanner. Antagonism between the leaf pathogen P. syringae DC3000 lx and other strains was defined as the formation of a visible inhibition zone.

Herbivore feeding assay.

For the S. littoralis no-choice assay, 26-day-old A. thaliana plants received one neonate S. littoralis larva (provided by Syngenta AG, Switzerland). Cellophane bags were placed over the plants to prevent larvae from leaving the plants. Larvae were collected 6 days postinfestation, freeze-dried, and weighed to the nearest 0.001 mg on a microbalance (Mettler Toledo MT5, USA). As for the pathogen infection experiment, this experiment consisted of 13 treatments (12 bacterial strains and the noninoculated control) and 20 to 34 plants per treatment. Due to space constraints, the 12 bacterial strains were divided into two sets, each consisting of 6 bacterial strains and a noninoculated control. Plant numbers per treatment varied as a result of plants that were bolting or larvae that escaped, were potentially overlooked, or died during the experiment and were excluded.

A second repetition of this experiment was performed with a subset of the best candidate strains (Pseudomonas sp. P97-38, Pseudomonas orientalis L1-3-08, and Pseudomonas aridus R1-43-08) along with additional controls that were bagged with cellophane bags but not infested with S. littoralis (uninfested). The procedure followed the first experiment, but in addition to the weight of freeze-dried larvae, the fresh shoot weights of S. littoralis-infested and uninfested plants were recorded and the percentage of plant tissue removed was estimated by taking the fresh shoot weight of each damaged plant relative to the mean fresh weight of all uninfested plants of the respective treatment.

S. littoralis preference.

A dual-choice assay with S. littoralis was performed in rectangular plastic boxes (19.5 cm by 9.5 cm by 8 cm). Pots with control plants and P97-38-inoculated plants (one each) were placed 3 cm apart in a box. Pseudomonas sp. P97-38 was selected based on the results of S. littoralis no-choice assays. Each box was filled with the sand-soil mix used to grow plants to match the level of the pots in order to create an even surface. S. littoralis was reared on general purpose lepidopteran diet (Frontiers Scientific, USA) at 27°C and an 8/16-h light/dark cycle prior to their use in the experiment. Two late 3rd-instar larvae were placed halfway between the two plants, and the top of the box was sealed with a mesh to prevent the larvae from leaving the box. Positions of larvae were recorded after 2 and 24 h. Choice of a treatment (noninoculated control or P97-38 inoculated) was defined as the presence of a larva on a plant or near its pot at the respective time point; otherwise, larvae were recorded as having made no choice. The experiment was conducted in a controlled climate chamber (22°C, 8/16-h light/dark period) and comprised 20 boxes, 40 plants (20 control, 20 treatment), and 40 S. littoralis larvae. Boxes in which larvae consumed an entire plant before the experiment ended were not included in the analysis.

Sample collection for the analysis of glucosinolates, sugars, and phytohormones.

For the analyses of glucosinolates, sugars, and proteins, whole-leaf rosettes of noninoculated and P97-38-inoculated plants that were either uninfested or infested by S. littoralis were sampled. These samples were derived from the repetition of the S. littoralis no-choice assay described above.

For the analysis of phytohormone responses to P. syringae DC3000 lx and S. littoralis, we performed a separate experiment that included noninoculated and P97-38-inoculated plants that were subject to the following four treatments: P. syringae DC3000 lx infected, mock infected, S. littoralis infested, and uninfested. For the P. syringae DC3000 lx challenge, plants were infected or mock infected by spraying them five times with a 4.5 × 10⁷ CFU/ml suspension or a mock suspension and were afterwards kept at a high humidity conducive to infection. For S. littoralis challenge, plants were infested with 5 neonate S. littoralis larvae and were enclosed in a cellophane bag or only enclosed in the bag as a mock treatment (uninfested). Twenty-four hours after the treatments, whole leaf rosettes were sampled.

In all experiments, whole leaf rosettes were frozen in liquid N₂, freeze-dried, and ground to fine powder using a bead mill (Geno/Grinder; SPEX SamplePrep, USA). Ground tissue was then weighed into tubes and used for the analysis of glucosinolates, sugars, proteins, and phytohormones as described below. Only plants in the vegetative state were considered for the analyses.

(i) Glucosinolate analysis.

Extraction of glucosinolates was according to an established procedure (60) with minor modifications. We used ∼11 to 16 mg of dry leaf tissue powder. All samples were analyzed by liquid chromatography-mass spectrometry (LC-MS) (Agilent 6550 iFunnel Q-TOF LC/MS) on a rapid-resolution high-definition (RRHD) Zorbax Eclipse Plus C₁₈ column (4.6 mm by 150 mm, 5 μm). Desulfo-glucosinolates were separated by a water (5 mM ammonium formate)-acetonitrile gradient using a mobile phase of 98% water for 0.8 min that was reduced to 65% water over 14 min and to 0.5% water over 3.2 min. Glucosinolates were tentatively identified based on mass and the loss of a hexose‐derivative from a parent aglycone (60) and quantified using the UV spectrum (229 nm) based on a standard curve of sinigrin as described by Grosser and van Dam (61).

(ii) Phytohormone analysis.

For the quantification of phytohormones, each sample (∼6 to 11 mg of dry leaf tissue powder) received 750 μl of ammonium acetate (10 mM), 650 μl of methanol, and 50 μl of an internal standard solution containing isotope-labeled jasmonic acid and salicylic acid. Samples were vortexed for 15 to 20 s, sonicated for 15 min, and centrifuged at 20,000 × g; 1 ml supernatant was then transferred to a new Eppendorf tube and dried down on a Savant SpeedVac concentrator SPP1010 (Thermo Scientific, Reinach, Switzerland). Subsequently, samples were resuspended in 100 μl of 0.1% formic acid, placed on ice for 10 min, and centrifuged again at 20,000 × g for 10 min. Ninety microliters of the supernatant was taken for analysis. Phytohormones were analyzed by LC/MS (Agilent 6550 iFunnel Q-TOF LC/MS) on an RRHD Zorbax Eclipse Plus C₁₈ column (2.1 mm by 100 mm, 1.8 μm) and separated by a water (5 mM ammonium formate)-acetonitrile gradient using a mobile phase of 99% water for 0.7 min that was reduced to 0.5% water over 5.6 min and held at 0.5% for another 1.7 min. Phytohormones were identified and quantified based on labeled phytohormone standards of known concentrations.

(iii) Analysis of sugars and protein content.

For the quantification of soluble sugars, approximately 8 to 12 mg of dry leaf tissue powder was extracted according to the procedure described by Lisec et al. (62). Samples were analyzed by gas chromatography mass spectrometry flame ionization detection (GC-MS/FID; Agilent 7890B GC, Agilent 5977A MSD) on a HP-5ms Ultra inert column (30 m by 0.25 mm, 0.25 μm) with a method that consisted of a 5-min step at 70°C, a 5°C per min ramp to 325°C, and a 2-min step of 325°C at the end. Sucrose, glucose, and fructose were identified using authentic reference standards and quantified by adding 60 μl of ribitol (0.2 μg/μl) to each sample as an internal standard.

Total protein was quantified by Bradford’s assay (63). Briefly, 500 μl of ultrapure water (containing 20 mM Tris-Cl, 50 mM NaCl) was added to approximately 8 to 12 mg of leaf tissue that had been used for sugar quantification, sonicated for 15 min, and then centrifuged for 15 min (13,000 × g, 4°C). The supernatant was transferred to a 2-ml Eppendorf tube and dried in a Savant SpeedVac concentrator SPP1010, and proteins were resuspended in 50 μl ultrapure water. After the addition of the Bradford reagent, absorbance was measured (595 nm) in a Spectra Max i3 plate reader, and total protein quantified by using a bovine serum albumin (BSA) standard curve.

Data analysis.

All data analyses were performed in R (version 3.6.3). Data were transformed throughout the analyses when necessary to conform to the assumption of normality and homogeneity of variances. Data from P. syringae infection assays (photon counts) and the S. littoralis no-choice assay (S. littoralis dry weight) were analyzed with a generalized least-squares model (explanatory factor, rhizobacterium strain) using the gls function of the nlme package with the varIdent argument to account for heterogenous variances. Data from the repetition of the S. littoralis no-choice assay (plant weight, S. littoralis dry weight, and percent leaf tissue removed) were analyzed with a one-way analysis of variance (ANOVA; explanatory factor, rhizobacterium strain) followed by a Tukey’s test. Larval choice data from the dual-choice assay were analyzed separately for each time point with a chi-square test.

For the analysis of glucosinolates, phytohormones, and sugars, data were either analyzed by a two-way ANOVA (explanatory factors, rhizobacterium strain and S. littoralis/P. syringae treatment), a two-way ANOVA with a heteroscedasticity-corrected coefficient covariance matrix when the assumption of homogeneity of variances was violated, or generalized linear models. Glucosinolate composition was analyzed by a PERMANOVA (explanatory factors, rhizobacterium strain and S. littoralis treatment) using the adonis2 function of the vegan package with Euclidean distances. Protein data were analyzed by a Welch’s two sample t test.