Four to 5-week-old plants grown in 10 cm round pots were used for all experiments. Arabidopsis thaliana Columbia-0 wild-type and mutant plants (tpc1-2 and pop2-5) were kept at short day conditions after stratification for 2 days at 4°C. The growth chamber was adjusted to 50–60% humidity and 21°C with a 10-h-light/14-h-dark photoperiod using FLUORA^® L 36W/77 bulbs (OSRAM, Garching, Germany) with a light intensity of 100 μmol m² s^-1. Seeds of tpc1-2 were kindly provided by Prof. Edgar Peiter (University Halle) and pop2-5 seeds (GK_157D10) were purchased from GABI-Kat directly (Kleinboelting et al., 2012). To ensure the same starting conditions, all plants used for one assay were sown and germinated on the same day and were kept in the same growth chamber. For experiments investigating the systemic response and translocation of metabolites the leaves of each plant were counted according to their age (Dengler, 2006; Farmer et al., 2013; Kiep et al., 2015).

MecWorm treatment was used for mechanical wounding of the plant with punches every 5 s, totaling 12 punches per minute on treated leaf 8 (Mithöfer et al., 2005; Scholz et al., 2015). To investigate the systemic response upon treatment of leaf 8, the local and systemic leaves 5, 8, 9, 11, and/or 13 were analyzed. Untreated plants, kept exactly as the treated plants, were used as controls. The duration of the MecWorm treatment used is indicated in the respective figures. To study the wounding-induced systemic translocation of GABA in the plant, a double-deuterated GABA (Sigma–Aldrich, Munich, Germany) was used and applied to pop2-5 mutant plants. The pop2-5 mutant was used for this assay since it cannot degrade GABA due to the knock-out of the GABA-T gene (Palanivelu et al., 2003). According to (Jimenez-Aleman et al., 2015), 20 μl of 50 μM D₂-GABA or water was applied to local wounds on leaf 8, which were generated with a pattern wheel (PRYM_610940, Prym, Stolberg, Germany) without wounding the midrib. On each side of the midrib, the leaf was wounded with six vertical motions. D₂-GABA and water-treated plants were kept for 1.5 and 3 h with a cover to prevent evaporation. Samples of leaf 8 and chosen systemic leaves were harvested after the indicated time points. To avoid additional accumulation of GABA due to the cutting and harvesting process, all samples were harvested in less than 1 min and directly stored in liquid nitrogen. For all plants the leaves were harvested from oldest to youngest leaf (starting with leaf 5). All samples were kept at -80°C till further analysis.

Larvae of the generalist herbivore Spodoptera littoralis were hatched and reared on artificial diet (Bergomaz and Boppre, 1986) at 23–25°C with 10-h-light/14-h-dark cycles. For the 1 week long-term feeding assay, three larvae of first instar were placed on every wild-type and mutant plant. To achieve similar starting conditions, all larvae determined for one plant genotype were pooled and weighed prior to the experiment. The minimal starting weight of 30 larvae was set to 60 mg to ensure a survival of all larvae. After 1 week the weight of every larva was recorded separately. For the short-term feeding assay third instar larvae were used after they were kept overnight without food. This treatment ensures an immediate start of feeding after placement on the plant. The locally fed leaves were collected after the indicated time points and kept at -80°C till further analysis.

Two hundred and fifty milligram of fresh leaves were frozen in liquid nitrogen and weighed to enable a GABA determination per g fresh weight. If multiple leaves of a plant were collected, then leaves were cut according to their age starting with the oldest (leaf 5). The leaf material was homogenized in a Geno/Grinder^® 2010 (Spex Sample Prep, Stanmore, UK) equipped with aluminum racks. The racks were cooled in liquid nitrogen prior to usage to prevent a thawing of leaf material during the whole homogenization process. The amino acids (including GABA) were extracted twice with a total of 2 ml of methanol on ice. Supernatants were combined and dried using a Concentrator plus (Eppendorf, Hamburg, Germany) and re-suspended in 500 μl of methanol. The extract was diluted 1:20 (v/v) with water containing the internal standard. The algal amino acid mix ¹³C, ¹⁵N (Isotec, Miamisburg, OH, USA) was used as internal standard at a concentration of 10 μg ml^-1 in all samples. The concentration of unlabeled and D₂-GABA (M + 2) was analyzed by LC-MS/MS according to previous studies (Scholz et al., 2015). An API 5000 tandem mass spectrometer (Applied Biosystems, Darmstadt, Germany) was operated in positive ionization mode with multiple reaction monitoring (MRM) to monitor analyte parent ion → product ion: GABA (m/z 104.1 →87.1; DP 51, CE 17); this MRM is specific for GABA and does not detect any α-aminobutyric acid, D₂-GABA (m/z 106.1 →89.1+ m/z 106.1 →88.1; DP 51, CE 17).

Like the quantification of GABA, 250 mg of leaf material was used for phytohormone analysis. The extraction procedure and determination of JA and JA-Ile was carried out on ice as described before (Vadassery et al., 2012) with small changes. In this study, a different mixture of labeled jasmonates was used as internal standard. Instead of 15 ng of JA-[¹³C6]-Ile conjugate used in the previous study, 60 ng of D₆-JA-Ile (HPC Standards GmbH, Cunnersdorf, Germany) was used. Additionally, the 60 ng of 9,10-D₂-9,10-dihydrojasmonic acid was replaced by 60 ng of D₆-JA (HPC Standards GmbH, Cunnersdorf, Germany). Since it was observed that both the D₆-labeled JA and JA-Ile contained 40 % of the corresponding D₅-labeled compounds (which were not included in the analysis method), the obtained results were divided by 1.7 to exclude this inaccuracy.

All plants of different mutant lines used for one experiment were grown in the same growth chamber. Independent experiments were treated as a completely randomized design. Experiments were repeated three times to ensure reproducibility and 5–10 plants were used in each treatment for each experiment time point. Data of all independent experiments were pooled and analyzed. For comparison of two groups, the Student’s t-test or Mann Whitney U-test was applied. For statistical analyses of multiple groups, one-way analysis of variance (one-way ANOVA) or two-way analysis of variance (two-way ANOVA) were used as indicated in the figure legends. Different letters indicate significant differences between treatments. GraphPad Prism 6 and Origin Pro were used for data analysis and graph composition.

To examine if the production of GABA in untreated systemic leaves is induced in a Ca²⁺/calmodulin-dependent manner (Snedden et al., 1995), the systemic GABA accumulation was analyzed after MecWorm treatment in the tpc1-2 mutant. Recently, it was observed, that the tpc1-2 mutant did not show a systemic [Ca²⁺]_cyt elevation after mechanical wounding (Kiep et al., 2015). Thus, tpc1-2 is a good model to study whether or not the systemic GABA accumulation depends on the systemic calcium signal. As expected, after 1.5 h of MecWorm treatment on WT leaf 8, this leaf showed the highest GABA accumulation. However, a significant increase in GABA concentration was also observed in systemic leaves 5 and 11 (Figure 1). With 0.05 μmol g FW^-1 the tpc1-2 plants showed the same basic GABA level in control plants as the WT and also a comparable increase in systemic leaves 5 and 11 (Figure 1). This indicates that the insect-like wounding-induced systemic [Ca²⁺]_cyt elevation has no or limited influence on the amount of GABA produced in systemic leaves. For soybean GAD it was observed that an increase in [Ca²⁺]_cyt up to 7–11 μM is necessary to decrease the enzyme’s Km value about 55% (Snedden et al., 1995). The determined stress-induced [Ca²⁺]_cyt elevations in Arabidopsis showed an increase to a maximum of 0.6–2 μM (Knight et al., 1996, 1997; Whalley and Knight, 2013). Hence, the wounding-induced systemic [Ca²⁺]_cyt elevation appears to be insufficient to account for the increasing GAD activity observed in the present study.

Due to the results obtained with tpc1-2 plants, the possibility of direct GABA transport to systemic leaves was investigated. Therefore, we employed the GABA-transaminase knock-out mutant line, pop2-5. This mutant has not been used and characterized in the context of herbivory. Previous studies observed that increased concentrations of GABA can affect insect’s development (Ramputh and Bown, 1996; MacGregor et al., 2003; Bown et al., 2006; Scholz et al., 2015). We used the pop2-5 mutant to further evaluate the effect of high endogenous GABA concentration in planta on the herbivorous larvae of Spodoptera littoralis in a feeding assay (Figure 2). Our results demonstrate that the larvae, feeding 1 week on pop2-5 plants, gained significantly less weight than the larvae feeding on WT plants (Figure 2A). The difference could also be observed by macroscopic inspection (Figure 2B). In previous experiments with GABA-supplemented artificial diet it was shown, that larvae of C. rosaceana and S. littoralis feeding on high levels of GABA showed decreased survival, delayed development and decreased gain of weight, respectively (Ramputh and Bown, 1996; Scholz et al., 2015). This indicates that the constitutive GABA accumulation in pop2-5 plants seems to contribute to the enhanced resistance against S. littoralis larvae attack while the exact effect of GABA on the insects is not known. Enhanced GABA levels in the insect might lead to permanent activation of GABA-activated Cl^--channels resulting in hypertension or paralysis (Sattelle, 1990; Bown et al., 2006).

We further investigated the GABA response upon MecWorm wounding in well-developed leaves of pop2-5 plants (Figure 3A). With MecWorm we are able to investigate the impact of the isolated wounding process without the contribution of insect-derived compounds in oral secretions (Mithöfer et al., 2005). As shown in Figure 3B, continuous mechanical wounding of leaf 8 significantly elevated the amount of GABA in the treated leaf in both WT and pop2-5 plants. For WT plants, apart from the local GABA accumulation, we observed a significant increase of GABA in leaf 13, which is directly connected to leaf 8 and a non-significant increase in the indirectly connected leaf 5 (Dengler, 2006). For pop2-5 plants the local GABA accumulation of leaf 8 was followed by a significant increase of GABA in leaves 5 and 13. In preliminary experiments we observed that the GABA elevation in the adjacent leaves is a time-dependent response, and after 90 min of feeding-like wounding of leaf 8 a significant increase of GABA was detectable in leaf 13. As expected, there was no significant increase of GABA on leaf 9, as it is not connected to treated leaf 8. Analysis of the GABA concentration after 1.5 h of S. littoralis feeding revealed that the larvae caused a 6-fold higher GABA accumulation in pop2-5 compared to WT plants, in the fed leaves (Figure 3C). The synthesis of GABA in pop2-5 plants during treatments should be the same but in WT there is more GABA degradation by the transaminase during the assay, so the accumulation at the end was lower in the WT (Figure 3C). Here, it should be mentioned that in this set of experiments no clear difference in GABA concentrations between untreated WT and pop2-5 plants was found. This contradicts earlier findings including our own study (Palanivelu et al., 2003; Miyashita and Good, 2008; Clark et al., 2009; Scholz et al., 2015) where in pop2-5 plants a higher GABA level was detected. However, this might be explained by different growth conditions, i.e., long vs. short day. Nevertheless, taken together, the results indicate that also in pop2-5 mutant plants GABA is an efficient inducible systemic defense factor against insect feeding.

The accumulation of jasmonates was not influenced by different GABA levels in gad1/2 and gad1/2 x pop2-5 mutants. Similarly the accumulation of GABA was not changed in jar1, a JA-Ile jasmonate signaling mutant (Scholz et al., 2015). We decided to further test the involvement of jasmonates on GABA accumulation also in the non-GABA-degrading pop2-5 mutant. Therefore we measured the levels of JA and its bioactive derivative, (+)-7-iso-jasmonoyl-L-isoleucine (JA-Ile), in Arabidopsis WT and in pop2-5 plants both upon mechanical wounding and herbivore treatment. The concentration of JA and JA-Ile significantly increased after feeding assay with S. littoralis and after MecWorm treatment in both tested genotypes (Figure 4). In all cases the MecWorm treatment caused higher jasmonate accumulation but the non-treated controls showed no differences between WT and pop2-5 plants. These data support earlier results suggesting that GABA accumulation is not jasmonate-dependent (Scholz et al., 2015) and vice versa. However, we cannot completely rule out that an octadecanoic dependent, but JA/JA-Ile independent, signaling pathway might still be involved in GABA accumulation.

To further investigate if GABA is transported to or synthetized in the adjacent leaves, we treated Arabidopsis pop2-5 plants with the double-deuterated D₂-GABA (Figure 5A) on a leaf that was wounded with a pattern wheel. This approach would enable us to observe any translocation of GABA in the plant leaves, because no D₂-GABA degradation takes place in the pop2-5 mutant. The basic GABA levels in different leaves of control pop2-5 plants were comparable (Figure 5B). Leaves of the pop2-5 plants that were slightly pattern wheel-wounded and incubated with water, showed a GABA concentration similar to control plants (Figures 5B,C), indicating that the weak single mechanical wounding had no effect compared to continuous mechanical wounding achieved with MecWorm treatment. Interestingly, when GABA was measured in leaf 8-wounded pop2-5 plants that were treated with D₂-GABA, an increase of non-labeled GABA was detected, with the accumulation of GABA in directly connected (leaf 13) and indirectly connected (leaves 5 and 11) leaves. GABA accumulation was significantly higher after 1.5 h and increased more after 3 h (Figure 5C). Even leaf 9 showed, compared to wounding and water treatment (W + W), a slight elevation of unlabeled GABA after the application of D₂-GABA (W + D₂-GABA). These results suggest that GABA synthesis was stimulated by the combination of the pattern wheel wounding and D₂-GABA incubation in both the treated leaf and systemic leaves (Figure 5C). When we further analyzed the concentration of D₂-GABA on pop2-5 treated plants it was only possible to detect higher D₂-GABA level in the locally treated leaf 8 indicating that D₂-GABA was taken up. However, the fact that in the adjacent leaves D₂-GABA was detected not even nearly as in leaf 8 (Figure 5D) suggests that it is not transported from the local wounded leaf to the systemic leaves. This result, combined with the observation that also the systemic cytosolic calcium elevation after wounding is not responsible for activation of GABA synthesis in systemic unwounded leaves, demonstrates that there has to be a different signal initiating GABA synthesis and accumulation in systemic leaves. Previous studies in animals presented data concerning GABA-mediated positive feedback loop (Kamermans and Werblin, 1992; Fenalti et al., 2007) and also in plants a several-fold endogenous accumulation of GABA after external application of GABA has been found for different tissues (Shi et al., 2010; Shang et al., 2011; Malekzadeh et al., 2014; Vijayakumari and Puthur, 2016). Further studies are required to establish a role for phloem B-type cells in long-distance signaling in plants (Shelp, 2012). In a study concerning drought stress tolerance with a Poaceae species, Agrostis stolonifera, GABA was shown to enhance the accumulation of GABA and of other amino acids (glycine, valine, proline, 5-oxoproline, serine, threonine, aspartic acid and glutamic acid) (Li et al., 2017). Interestingly, in barley plants it was shown that external application of GABA was able to induce action potentials (APs) accompanied by cytoplasmatic acidification short of the position of the stimulus (Felle and Zimmermann, 2007). APs and other electrophysiological reactions represent conceivable signaling candidates for systemic leaves because they have been shown to be involved in wound-induced leaf-to-leaf signaling (Zimmermann et al., 2009, 2016; Mousavi et al., 2013).