Overcompensation of herbivore reproduction through hyper‐suppression of plant defenses in response to competition

Summary Spider mites are destructive arthropod pests on many crops. The generalist herbivorous mite Tetranychus urticae induces defenses in tomato (Solanum lycopersicum) and this constrains its fitness. By contrast, the Solanaceae‐specialist Tetranychus evansi maintains a high reproductive performance by suppressing tomato defenses. Tetranychus evansi outcompetes T. urticae when infesting the same plant, but it is unknown whether this is facilitated by the defenses of the plant. We assessed the extent to which a secondary infestation by a competitor affects local plant defense responses (phytohormones and defense genes), mite gene expression and mite performance. We observed that T. evansi switches to hyper‐suppression of defenses after its tomato host is also invaded by its natural competitor T. urticae. Jasmonate (JA) and salicylate (SA) defenses were suppressed more strongly, albeit only locally at the feeding site of T. evansi, upon introduction of T. urticae to the infested leaflet. The hyper‐suppression of defenses coincided with increased expression of T. evansi genes coding for salivary defense‐suppressing effector proteins and was paralleled by an increased reproductive performance. Together, these observations suggest that T. evansi overcompensates its reproduction through hyper‐suppression of plant defenses in response to nearby competitors. We hypothesize that the competitor‐induced overcompensation promotes competitive population growth of T. evansi on tomato.


Fig. S1
Schematic overview of the experimental procedures of the spider mite infestation assay to assess the induction of plant defenses at a 4-d-old primary spider mite feeding site and in adjacent leaflet tissues including a 2-d-old secondary feeding site. For a detailed description we refer to the material and methods section. 1 = 4-d-old primary feeding site; 2 = 2-d-old secondary feeding site. After 2 d the tip section was subjected to a secondary infestation with either three T. evansi or three T. urticae. Again 2 d later the number of eggs produced by the mites on the leaflet tip section (indicated in green) was counted. The figure shows the average (+ SEM) number of eggs produced per female per day for (a) T. evansi and (b) T. urticae. Bars are colored according to the treatment of the middle section (primary infestation). Oviposition data was statistically evaluated per mite species by applying a linear mixed-effects model, but no significant differences were found between the treatments (P > 0.05). ns, not significant. promoter, or after agroinfiltration with the empty vector (EV) construct. To visualize suppression of jasmonic acid-regulated defense responses, agroinfiltrated leaves were wounded with a pattern wheel, after which Manduca sexta oral secretions were applied to the wounds (W+OS treatment). Untreated plants served as controls. TPI transcript abundances were normalized to Actin and then scaled to the lowest average value per treatment. Different letters above the bars indicate significant differences at a level of P ≤ 0.05, after applying a linear mixed-effects model (panels a, b) or a generalized linear model (panels c, d), followed by Tukey multiple comparisons. For a detailed description of the experimental procedures we refer to Methods S3.   Fig. 7(a). 5, T. urticae reproductive performance data (primary infestation) is presented in Fig. 7(b). 1, the RP49 amplicon is identical for both mite species

Methods S1 Isolation of phytohormones and analysis by means of LC-MS/MS
About 150-300 mg of frozen leaf material was homogenized (Precellys 24, Bertin Technologies, Aixen-Provence, France) in 1 ml of ethyl acetate. The ethyl acetate had been spiked with D6-SA and D5-JA (C/D/N Isotopes Inc, Canada) as internal standards with a final concentration of 100 ng ml -1 . Tubes were centrifuged at 13,000 rpm (15,493 g; Sigma 3-30KS; SIGMA Laborzentrifugen GmbH, Osterode am Harz, Germany) for 10 min at 4°C and the supernatant (the ethyl acetate phase) was transferred to new tubes. The pellet was re-extracted with 0.5 ml of ethyl acetate (without internal standards) and centrifuged again at 13,000 rpm for 10 min at 4°C. Both supernatants were combined and evaporated to dryness on a vacuum concentrator (CentriVap Centrifugal Concentrator, Labconco, Kansas City, MO, USA) at 30°C. The residue was re-suspended in 0.1 ml 70% methanol (v/v), centrifuged at 14,800 rpm (20,081 g) for 15 min at 4°C, and the supernatants were transferred to glass vials and then analyzed by means of LC-MS/MS. A serial dilution of pure standards of OPDA, JA, JA-Ile and SA was run separately. Measurements were conducted on a liquid chromatography tandem mass spectrometry system (Varian 320-MS LC/MS, Agilent Technologies). We injected 20 µl of each sample onto a Kinetix 5u C18 100A column (C18 phase, 5 μm particle size, 100Å pore size, 50 × 2.1 mm, Phenomenex, Torrance, CA, USA) equipped with a Phenex-RC guard cartridge (Phenomenex). The mobile phase contained solvent A (0.05% formic acid in LCMS-grade water; Sigma-Aldrich, St. Louis, MO, USA) and solvent B (0.05% formic acid in LCMS-grade methanol; Sigma-Aldrich) in varying proportions. The program, with a constant flow rate of 0.2 ml min -1 , was set as follows: (i) 95% solvent A/5% solvent B for 1 min 30 sec; (ii) followed by 6 min in which solvent B gradually increased till 98%; (iii) continuing with 98% solvent B for 5 min; (iv) then a rapid (in 1 min) but gradual decrease returning to 95% solvent A/5% solvent B until the end of the run. A negative electrospray ionization mode was used for detection. LC-MS/MS parameters, i.e., the parent ions, daughter ions, and collision energies were identical to those of Alba et al. (2015). For all oxylipins we used D5-JA to estimate the recovery rate and their in planta concentrations were subsequently quantified using the respective external standard series. For SA we used D6-SA to estimate the recovery rate and it was quantified using the external standard series.

Methods S2 Gene-expression analysis by means of quantitative reverse-transcriptase PCR (qRT-PCR)
Total RNA was isolated from tomato tissue (with or without mites) using the hot phenol method (Verwoerd et al., 1989). The NanoDrop spectrophotometer (ND-1000, Thermo Fisher Scientific, Waltham, MA, USA) was used to assess RNA purity and quantity. DNase (Ambion, Austin, TX, USA)-treated RNA was used as template for reverse transcription and first strand cDNA synthesis using RevertAid H Minus Reverse Transcriptase (Thermo Fisher Scientific). For gene expression analysis, 1 μl of diluted cDNA (i.e. the equivalent of 7.5 ng total RNA for tomato genes and 100 ng total RNA for spider mite genes) served as template in a 20 μl qRT-PCR using the 5x HOT FIREPol EvaGreen qPCR Mix Plus (ROX) kit (Solis Biodyne, Tartu, Estonia) and the ABI 7500 Real-Time PCR system (Applied Biosystems, Foster City, CA, USA), according to the instructions of the manufacturers. Using RNA from the same samples, we analyzed the transcript abundance of the tomato defense-associated marker genes OPR3, PPO-D, JIP-21, PI-IIc, PR-1a and PR-P6, as well as the spider mite effector-encoding genes Te28, Tu28, Te84 and Tu84. With the exception of OPR3, the expression patterns of the tomato genes over time in plants infested with mites from the T. urticae or T. evansi lines used here have been described in detail before (Alba et al., 2015). Tomato Actin and spider mite RP49 were used as reference genes for the respective template to normalize expression data across samples. Gene identifiers, primer sequences and references are listed in Table S2. Primer efficiency of each primer pair was calculated using standard dilution series. qRT-PCR-generated amplicons were sequenced to verify primer specificity. When gene transcripts were not detected in both technical replicates of a sample, then this sample was scored as '0' (zero) and included as such in the statistical analysis. When gene transcripts were detected in only one technical replicate of one biological replicate and not in all other replicates of the same treatment, then the normalized expression (NE) of that one technical replicate was also scored as '0'. The NE data were calculated by the ΔCt method: NE = (PEtarget Ct_target )/(PEreference Ct_reference ); in which PE is the primer efficiency and Ct the number of cycles to reach the cycle threshold value. To plot the relative expression, NE values were scaled to the treatment with the lowest average NE.

Methods S3 Suppression of JA defenses by spider mite effectors
Previously, using the Agrobacterium tumefaciens-mediated transient overexpression of spider mite effector-encoding genes in Nicotiana benthamiana leaves, Te28, Tu28, Te84 and Tu84 have been shown to suppress SA defenses (Villarroel et al., 2016). Although there were indications that (some of) these effectors also suppressed JA defenses, the agroinfiltration-induced SA response and concomitant antagonistic crosstalk with the JA pathway largely concealed the experimental outcome with respect to suppression of JA defenses (Villarroel et al., 2016). Hence, here we deliberately induced JA-regulated defense responses in agroinfiltrated N. benthamiana leaves to be able to detect suppression of these defenses by spider mite effectors. Cloning of each of the four mite effectorencoding genes (without signal peptide) into the plant expression vector pSOL2092, which contains the Cauliflower mosaic virus 35S promoter, and subsequent A. tumefaciens-mediated transient overexpression assays with mite effector or empty vector (EV) constructs in N. benthamiana were performed as described by Villarroel et al. (2016). Two days after agroinfiltration, leaves were wounded with a pattern wheel, after which 20 μl of Manduca sexta oral secretions (three times diluted) was applied to the wounds (W+OS treatment) as described by Wu et al. (2007). Four hours after the W+OS treatment the leaves were harvested for RNA isolation. Control plants were agroinfiltrated but did not receive the W+OS treatment. RNA isolation, DNAse-treatment, cDNA synthesis and qRT-PCRs to determine the transcript abundance of the JA-responsive N. benthamiana trypsin proteinase inhibitor (TPI) gene (Yoon et al., 2009) were performed as described in Methods S2. TPI transcript abundances were normalized to Actin (Table S2) and then scaled to the lowest average value per treatment. The NE of TPI was statistically evaluated per construct as described in the material and methods section. For effectors Te28 and Tu28, six plants were agroinfiltrated per treatment, while for effectors Te84 and Tu84, three plants were agroinfiltrated per treatment.

Notes S1 Within-leaflet systemic effects on induced plant responses upon the T. urticae infestation
From our set of marker genes, only the expression of JIP-21, which encodes a JA-inducible proteinase inhibitor that inhibits digestive enzymes in the gut of herbivores (Lisón et al., 2006), was up-regulated locally and systemically by T. urticae. This is in line with results from an earlier study that showed an increased proteinase inhibitor activity in T. urticae-infested leaves as well as in uninfested leaves of the same plants (Sarmento et al., 2011). In contrast to JIP-21, the expression of the JA-biosynthesis gene OPR3 was down-regulated in tissues adjacent to the 4-d-old T. urticae feeding site. Locally, OPR3 expression was not significantly altered by T. urticae, yet it was induced by T. evansi. As noted previously, not all defense-associated plant genes are (continuously) induced by T. urticae and suppressed by T. evansi, respectively, throughout the course of the infestation (Alba et al., 2015). Also the expression of OPR3 seems to depend on the duration of the infestation and/or the intensity of the infestation, as its expression was induced in the leaflet tip section after infestation with T. urticae for 2 d (Fig. 3a). The RNAi-mediated silencing of OPR3 in tomato has been shown to result in lower concentrations of jasmonates (including JA-Ile) that are produced from OPDA (Bosch et al., 2014;Scalschi et al., 2015). However, we found JA and JA-Ile concentrations not to significantly differ between tissues infested by T. urticae or T. evansi, nor in the adjacent tissues. This emphasizes that expression levels of phytohormone biosynthesis genes are not always predictive of the actual phytohormone concentrations, which in this case might be the result of feedback and feedforward mechanisms associated with JA biosynthesis (Wasternack & Hause, 2013).