pH effects on plant calcium fluxes: lessons from acidification-mediated calcium elevation induced by the γ-glutamyl-leucine dipeptide identified from Phytophthora infestans

Cytosolic Ca2+ ([Ca2+]cyt) elevation is an early signaling response upon exposure to pathogen-derived molecules (so-called microbe-associated molecular patterns, MAMPs) and has been successfully used as a quantitative read-out in genetic screens to identify MAMP receptors or their associated components. Here, we isolated and identified by mass spectrometry the dipeptide γ-Glu-Leu as a component of a Phytophthora infestans mycelium extract that induces [Ca2+]cyt elevation. Treatment of Arabidopsis seedlings with synthetic γ-Glu-Leu revealed stimulatory effects on defense signaling, including a weak enhancement of the expression of some MAMP-inducible genes or affecting the refractory period to a second MAMP elicitation. However, γ-Glu-Leu is not a classical MAMP since pH adjustment abolished these activities and importantly, the observed effects of γ-Glu-Leu could be recapitulated by mimicking extracellular acidification. Thus, although γ-Glu-Leu can act as a direct agonist of calcium sensing receptors in animal systems, the Ca2+-mobilizing activity in plants reported here is due to acidification. Low pH also shapes the Ca2+ signature of well-studied MAMPs (e.g. flg22) or excitatory amino acids such as glutamate. Overall, this work serves as a cautionary reminder that in defense signaling studies where Ca2+ flux measurements are concerned, it is important to monitor and consider the effects of pH.

(γ-Glu-Leu), because it exhibits characteristic MS fragments at m/z 244. 11 (Table 1). For further validation, Leu-Glu, Glu-Leu, and γ-Glu-Ile were either purchased (when commercially available) or synthesized in-house. Compared to γ-Glu-Leu, these isomers showed different fragmentation patterns and retention times and were therefore excluded (Fig. 2). Using calibration curves of the authentic compound, the concentration of γ-Glu-Leu in the P. infestans mycelium extract was estimated to be 110-190 nM (Fig. S2).
γ-Glu-Leu induces Ca 2+ flux and desensitizes cells for elicitation with P. infestans mycelium extract and different MAMPs. Several γ-glutamyl peptides have been discovered to be kokumi (taste-enhancing) substances and to induce a rise in intracellular [Ca 2+ ] in taste receptor cells in the lingual tissues of mice 53 . These γ-glutamyl kokumi peptides appear to directly engage calcium-sensing receptor (CaSR) in the taste buds 54 . Thus, among the identified components in the mycelium extract, γ-Glu-Leu is a candidate Ca 2+ flux-inducing compound and acts similarly in plants as an agonist of calcium channels. Alternatively, it may act as a MAMP to induce Ca 2+ elevations. To test if the dipeptide γ-Glu-Leu is indeed the active component from the mycelium extract that induces the [Ca 2+ ] cyt increase, we applied synthetic γ-Glu-Leu (ranging from 50 to 1000 µM) to aequorin-expressing Arabidopsis seedlings. Treatment with 50 µM γ-Glu-Leu had no obvious effect on the basal level of [Ca 2+ ] cyt represented by the water control, whereas all higher concentrations induced a transient elevation in Ca 2+ flux with dose-dependent magnitudes between 129 ± 7 nM [Ca 2+ ] cyt for 100 µM and 340 ± 5 nM [Ca 2+ ] cyt for 1000 µM γ-Glu-Leu (Figs 3A and S3). Based on the peak [Ca 2+ ] cyt values in these experiments, the EC 50 value of γ-Glu-Leu was estimated to be approximately 160 µM. The MAMP-induced Ca 2+ signature reported by the aequorin system is typically a transient Ca 2+ rise, followed by a return to basal resting levels. However, we observed that the rate of recovery to basal [Ca 2+ ] cyt level decelerated with increasing γ-Glu-Leu concentrations (Fig. 3A). In fact, for the 800 µM and 1000 µM γ-Glu-Leu treatments, a "second wave" of Ca 2+ rise appears to be initiated at ~12 min (Fig. 3A). Taken together, γ-Glu-Leu is likely an active component within the P. infestans mycelium extract, which elicits Ca 2+ elevations. If γ-Glu-Leu functions as a typical MAMP, the EC 50 value of the Ca 2+ induction would be indicative of a receptor-ligand interaction of low-to-moderate affinity.
Activated receptors are typically desensitized and unable to be re-activated within the so-called refractory period (for example, see Fig. 3F). This would be the scenario if γ-Glu-Leu engages the same receptor as P. infestans mycelium extract (ME). To test this, pMAQ2 seedlings were treated with a saturating concentration of γ-Glu-Leu (500 µM) for ~15 min prior to a second stimulation with ME. In comparison to the control (with water as pretreatment), γ-Glu-Leu severely reduced Ca 2+ peaks obtained with a subsequent application of ME (Fig. 3B). Similar suppression of the second Ca 2+ peak induction was seen with consecutive applications of γ-Glu-Leu ( Fig. 3C) or ME (Fig. 3D). Surprisingly, while not as strong as the γ-Glu-Leu/ME combination, pre-stimulation with γ-Glu-Leu also suppressed the Ca 2+ peaks elicited by subsequent application with flg22 (1 µM) (Fig. 3E). Similar observations were seen if the second treatment used was elf18 (1 µM), PEP (1 µM) or chitin (200 µg shrimp shell/ml) (Fig. S4A). This is reminiscent of the refractory period observed with consecutive applications of the same PAMP (e.g. flg22, Fig. 3F), which is not seen if two independent receptors are involved (e.g. elf18 treatment followed by flg22 as a second application, Fig. 3G). However, since it is unlikely that γ-Glu-Leu can act as an agonist for all these different receptors, another plausible explanation is that γ-Glu-Leu shares certain common signaling component(s) with these MAMPs/DAMPs to elicit the Ca 2+ response. For instance, γ-Glu-Leu may act similarly to the kokumi substances and engage the same plant Ca 2+ channels activated by the different MAMP/ www.nature.com/scientificreports www.nature.com/scientificreports/ DAMPs. Thus, analogous to the taste-enhancing properties in animal systems, γ-Glu-Leu may have synergistic effects on plant defense.
Simultaneous application of γ-Glu-Leu and flg22 modulates defense responses obtained with single treatments. The observations above suggest possible interplay between γ-Glu-Leu and MAMPs. As anticipated for situations resembling natural infections where there is simultaneous exposure to several MAMPs, combinatorial co-treatment with different MAMPs/DAMPs has been reported to confer additive, synergistic or reductive effects on defense responses in comparison to the single stresses [55][56][57] . We therefore tested whether a combination of γ-Glu-Leu and flg22 (as a representative MAMP) will affect Arabidopsis defense responses differently compared to single treatments. We compared the effect of single or combined stimuli on Ca 2+ flux as well as relative transcript levels of several flg22-responsive defense genes. Simultaneous treatment with non-saturating concentrations of γ-Glu-Leu (250 µM) and flg22 (10 nM) induced a similar Ca 2+ signature to that obtained with γ-Glu-Leu alone but with a significantly slightly higher amplitude of 255 ± 8 nM [Ca 2+ ] cyt (Fig. 4A).
To evaluate the effect on defense, we tested expression of twelve routinely analyzed flg22-responsive marker genes: 58,59 . When applied alone, γ-Glu-Leu did not induce the expression of any of these genes even at the high concentration of 500 µM (Figs 4B and S5). However, compared to flg22 alone, co-treatment with 500 µM γ-Glu-Leu and 10 nM flg22 resulted in a significant increase in transcript levels of ACD11-like, NHL10, MPK3, MPK11 and WRKY33 (Fig. 4B) but not of the other genes (Fig. S5).
Taken together, γ-Glu-Leu confers subtle "additive" effects on flg22-induced Ca 2+ flux and expression of a subset of defense genes. Since many of the analyzed genes are controlled by MAPK and/or CPK pathways 58,59 , the enhancement of gene expression may result from increased activities of these kinases. However, despite the enhanced [Ca 2+ ] cyt , γ-Glu-Leu treatment did not raise basal or flg22-induced activity of CPK (using in-gel phosphorylation assays with histone as substrate, data not shown) or MAPK phosphorylation (as a proxy for MAPK activity, Fig. S6A). For the latter, γ-Glu-Leu weakly induces MPK6 phosphorylation but showed no further enhancement of MAPK activation when co-treated with flg22 (Fig. S6B) www.nature.com/scientificreports www.nature.com/scientificreports/ transiently and to exclude the possibility that "the window" of any additive effect of the co-treatment was overlooked in the time points analyzed, we monitored an effect downstream of MPK6. For this, we checked the phosphorylation state of the MPK6 substrate, MVQ1, which can be visualized as a reduced electrophoretic mobility in SDS-PAGE after MAMP treatment 60 . Unlike flg22, neither γ-Glu-Leu alone, nor in combination with flg22 induced or further enhanced MVQ1 phospho-mobility shift, respectively (Fig. S6C). Hence, the boost in Ca 2+ flux or gene expression using co-treatment is unlikely to be caused by enhanced MAPK or CPK kinase activities. pH alteration may constitute the observed biological activity of γ-Glu-Leu. Cytosolic acidification is known to weakly activate MAPKs 61 . It is, thus, plausible that γ-Glu-Leu, with its additional acidic side group of the Glu moiety, promotes apoplastic and subsequently cytosolic acidification. Indeed, aqueous γ-Glu-Leu solution has an acidic pH of 3.9. When γ-Glu-Leu was dissolved in MES-buffered solution (pH 6.0), it no longer induced a rise in [Ca 2+ ] cyt and the additive effects of flg22 co-treatment on Ca 2+ flux (Fig. 4C), as well as on defense gene expression were also lost except for WRKY33 (Fig. 4D). Similarly, the apparent "refractory period" conferred by γ-Glu-Leu pretreatment to subsequent elicitation with ME or MAMPs (Fig. 3B,E) was also abolished if γ-Glu-Leu was MES-buffered at pH 6.0 (Figs 5 and S4B).
Hence, acidification by the γ-Glu-Leu solution may explain the observed induced responses. To validate this, we used a dilute acidified solution in place of γ-Glu-Leu. Here, water was acidified with acetic acid to reach the same pH as aqueous γ-Glu-Leu (i.e. ~3.9). Adding this dilute acetic acid solution to Arabidopsis seedlings indeed induced elevations in [Ca 2+ ] cyt and "suppressed" the Ca 2+ rise induced by a subsequent flg22 treatment (Fig. 6A). It also mimicked the enhancement of flg22-induced [Ca 2+ ] cyt elevation (Fig. 6B) that was seen with γ-Glu-Leu (c.f. Fig. 4A). Furthermore, co-treatment with acetic acid enhanced the flg22-induced expression levels of ACD11-like, MPK3, MPK11 and WRKY33 (Fig. 6C). However, note that unlike γ-Glu-Leu, addition of dilute acetic acid alone also weakly induced expression of most of these genes, so that the boosted gene expression are additive effects of the co-treatment. Taken together, we could recapitulate the biological effects of γ-Glu-Leu by mimicking apoplastic acidification.

Discussion
In this work, an extract prepared from P. infestans mycelium induces Ca 2+ flux, a typical early defense response, in A. thaliana seedlings. The C18-pre-fractionated extract contained relatively few compounds and mass spectrometry analysis based on authentic standards revealed one of the components as the dipeptide γ-Glu-Leu. Notably, low-abundance compounds as well as charged molecules cannot be covered by our approach (C18 cartridge and QToF), which should rather be measured with a triple quadrupole mass spectrometer using multi-targeted methods The P. infestans mycelium extract peak was annotated as gamma-glutamyl-leucine according to retention time and mass spectral features. The identity was confirmed using an authenticated reference substance.
www.nature.com/scientificreports www.nature.com/scientificreports/ (for low-abundance molecules) or be purified using an ion exchanger cartridge (for charged molecules). For the unknown compounds, CID-MS and H/D exchange chromatography experiments failed to gain further structural hints on those components. In order to elucidate these in future analyses, orthogonal analytical technologies (e.g. GC/MS) or preparative approaches (preparative LC coupled to elemental analysis or NMR analysis) will be required.
Our original goal was to isolate novel P. infestans MAMPs and use the aequorin-based Ca 2+ measurements to subsequently identify mutants of the corresponding PRRs. The finding that γ-Glu-Leu elicited Ca 2+ elevations in A. thaliana seedlings was initially promising since γ-Glu-Leu-related substances are known to also induce an intracellular [Ca 2+ ] increase in animal tissues 53 . Amplitude of [Ca 2+ ] cyt increase and the kinetics of the recovery phase were dose-dependent, which is in accordance with various other stresses such as application of hydrogen peroxide 62 , flg22 63,64 or AtPep1 63 . γ-Glu-Leu also had additive effects on induction of selected flg22-responsive  www.nature.com/scientificreports www.nature.com/scientificreports/ genes. However, γ-Glu-Leu did not (strongly) induce MAPK phosphorylation or ROS accumulation. These observations are reminiscent of recent studies involving cellulose-derived cellobiose fragments acting as a DAMP. Cellobiose weakly activated MPK6 but does not trigger production of ROS or callose deposition; however, it synergistically enhances Arabidopsis defenses in co-treatment with flg22 65 . We also tested if pre-treatment with γ-Glu-Leu would induce resistance to subsequent bacterial infection (as has been shown for other MAMPs, e.g. flg22) but this was not the case (not shown). Thus, γ-Glu-Leu is not a MAMP.
During our analysis, it became eventually clear that all the effects induced by γ-Glu-Leu can be attributed to acidification (of presumably the apoplast and eventually the cytosol). We could abrogate the biological effects of γ-Glu-Leu by buffering the pH exogenously and more importantly, mimicked these effects by applying weak acids like acetic acid (Fig. 6). We had also noticed that in treatments with high concentrations of γ-Glu-Leu, the rate of [Ca 2+ ] cyt recovery to resting level was decreased and an apparent second rise in [Ca 2+ ] cyt , resembling the beginning of some form of Ca 2+ oscillation, appeared (Figs 3A and S3.). In hindsight, both phenomena may be explainable by acidification. The post-elicitation Ca 2+ recovery phase relies on both membrane-localized Ca 2+ pumps and antiporters 17 to transport Ca 2+ out of the cytosol. ATPase activity of the pumps and antiporters would be affected by altered proton gradient caused by the acidification in a dose-dependent manner.
The additive effect of γ-Glu-Leu on flg22 induction on some of the analyzed genes was also abolished by buffering the dipeptide solution. Only WRKY33 retained the boosting effect of γ-Glu-Leu (Fig. 4D). WRKY33 is one of the upregulated genes found in transcriptome of plants exposed to acidic apoplastic pH 66 . It is possible that the buffering by MES may be incomplete and remnant regional pH changes do occur, which is sufficient to weakly boost flg22-induced expression of acidification-responsive genes such as WRKY33. Interestingly, in that report on apoplastic acidification, the transcriptome of plants exposed to low external pH was found to globally cluster with transcriptomes of plants treated with jasmonate, auxin (IAA) or salicylic acid, which are all substances that may lower external pH if not sufficiently buffered. Also clustered to the low pH dataset were transcriptomes of flg22 treatment (1 hpi) and P. infestans infection (6 and 12 hpi). For the latter, one may speculate that the pH effect of γ-Glu-Leu present on the mycelium may have shaped the P. infestans infection transcriptome. For flg22, the correlation between low pH and flg22 response is actually in agreement with our observation that the normal Ca 2+ response to flg22 is dampened when flg22 is MES-buffered (Fig. S9A). This highlights the possibility that for many of the reported studies, and particularly for flg22 treatment, acidification may also contribute to the downstream responses if non-buffered conditions were employed. MapMan analysis of the low pH-regulated genes further revealed an enrichment for "Ca 2+ regulation" elements 66 in their promoters, including the "CGCG" core, a promoter element known to be targeted by calcium-regulated CAMTA transcription factors 67 . CAMTA activity or other physiological changes induced by acidification may explain the observed enhancement of gene expression. Thus, this independent observation linking low pH and activation of calcium-mediated response corroborates our observations on the acidification-induced Ca 2+ response and gene induction through γ-Glu-Leu or weak acids like acetic acid.
While γ-Glu-Leu was originally isolated from P. infestans mycelium, its physiological function is unknown. In fact, the roles of most naturally occurring dipeptides are unknown; this includes the many dipeptides found in Arabidopsis root exudates [68][69][70] . Here, it is noteworthy that in the food industry, γ-Glu-Leu belongs to the so-called Kokumi taste-enhancing dipeptide family that is found, among others, in beans and mature cheese 71,72 . These Kokumi γ-glutamyl peptides act as agonists for the animal extracellular calcium sensing receptor, CaSR 54 and induce an increase in intracellular [Ca 2+ ] in a subset of the CaSR-expressing taste-bud cells within mouse lingual epithelia tissue 53 . Thus, one may also hypothesize that the Glu moiety of γ-Glu-Leu may directly engage www.nature.com/scientificreports www.nature.com/scientificreports/ analogous calcium-permeating channels of plants such as the ionotropic glutamate receptor-like channels (iGluR) 73,74 , which have been implicated in Ca 2+ fluxes induced by flg22, elf18 and chitin 75 . While Glu is well established as a second messenger that can mobilize Ca 2+ through iGluRs, it should be noted that most studies in plants required millimolar concentrations of Glu to elicit [Ca 2+ ] cyt changes 76,77 . The recent work on Glu as a systemic wound signal used 100 mM Glu to induce [Ca 2+ ] cyt changes and determined apoplastic Glu to reach 50 mM at the damage sites. However, such high concentrations of Glu will certainly involve cellular acidification. Indeed, comparing MES-buffered (or sodium salt of Glu) to unbuffered Glu revealed that Glu-induced Ca 2+ fluxes may be largely overestimated if pH is not taken into consideration (Fig. S7).
As mentioned above, we initially speculated that the acidic Glu moiety within γ-Glu-Leu would be responsible for the acidification-induced Ca 2+ response. However, several other dipeptides lacking acidic side chains also induced Ca 2+ elevation at high concentrations, which is eliminated with MES buffering (Fig. S8), suggesting contribution coming from the carboxyl termini of the dipeptide. Thus, this pH effect may apply to many more peptide elicitors. In this context, we also observed that MES attenuated the Ca 2+ response induced by flg22 and elf18 but not AtPep1 or chitin (Fig. S9). Compared to flg22 and elf18, AtPep1 is more basic, with one third of the sequence comprising of basic amino acids (lysine, arginine or histidine), which can titrate out the acidic properties of the C-terminus. During the manuscript review of our work, a report appeared showing similar results where flg22and elf18-induced Ca 2+ fluxes were reduced by external buffering at pH 5.5 78 . Note additionally that MES did not reduce but rather boosted the Ca 2+ response of chitin treatment (Fig. S9E). The reason for this boost is unknown at this stage but more importantly, this observation also implied that for all the other elicitors, reduction of Ca 2+ response when buffered with MES is not simply a counteraction of the plant's extracellular alkalinization response (Otherwise, the chitin response should also be dampened by MES buffering). Taken together, our current study validates a known but often overlooked fact that apoplastic acidification can induce a rise in [Ca 2+ ] cyt 79 .

Conclusions
γ-Glu-Leu is found in P. infestans mycelium extract and can induce Ca 2+ flux in Arabidopsis seedlings. Although it can act as a direct agonist of calcium sensing receptor in animal systems, the mechanistic of its Ca 2+ -mobilizing action is unknown in plants. More importantly, all of its elicitor-like properties in plants appear to be due to acidification (presumably apoplastic), and, thus, suggest that γ-Glu-Leu is not a MAMP. Nevertheless, its presence during infection may still contribute to signaling processes that are sensitive to pH alterations. Notably, pH also contributes to the Ca 2+ signature of established peptide MAMPs such as flg22 or excitatory amino acids such as glutamate. Overall, our work serves as a cautionary reminder that in studies of defense response and (especially?) where plant Ca 2+ flux measurements are concerned, control of pH is crucial but this is, unfortunately, not always true in the literature.

Materials and Methods
Plant material. The Arabidopsis line pMAQ2 was used in all experiments. pMAQ2 expresses cytosolic p35Sapoaequorin in a Col-0 background 80 . Seeds were surface-sterilized in 12-well plates (~20 seeds per well) and stratified at 4 °C for 3-5 days in 2 ml liquid MS medium per well (half strength MS, 0.25% sucrose, 1 mM MES, pH 5.7). Seedlings grew under long day conditions ( Remaining water was filtered through a nylon mesh placed in a Buchner funnel. For extraction, the protocol of Monjil et al. 52 was modified. Mycelium was ground in liquid nitrogen with a Retsch mill (Retsch GmbH, Haan, Germany) and subsequently homogenized in methanol (1 ml methanol/1 g mycelium) for 2 min using a Polytron (Kinematica AG, Luzern, Switzerland). After centrifugation (30 min, 4 °C, 3000 × g) 1 ml aliquots of the supernatant were evaporated in a SpeedVac (Thermo Fisher Scientific). Pellets were resolved in 1 ml water and combined. To remove salts, extract was loaded onto a Chromabond C 18 silica column (Chromabond Flash FM 70/10 C 18 ec, Macherey-Nagel, Düren, Germany), which was treated with methanol prior to use. The flowthrough fraction was discarded and extract components eluted with water (volume equivalent to volume of loaded extract). Thus, 1 ml of extract derived from 100 mg of mycelium. Mycelium and extract were stored at −20 °C.
Elicitors. Flg22, elf18 and AtPep1 1,2,8 were synthesized on a ResPep SL peptide synthesizer (Intavis Bioanalytical Instruments). Shrimp shells (Sigma-Aldrich) were used for chitin treatment. Stock solutions of the peptides (1 mM) and shrimp shell chitin (100 mg shrimp shell/ml) were prepared with water, dilutions of the stocks with water, 5 mM MES buffer (pH 6.0) or acetic acid (at the indicated concentrations to attain pH ~3.9, the pH of aqueous γ-Glu-Leu).
Dipeptides. The dipeptides γ-Glu-Leu and α-Leu-Glu were purchased from Bachem. Leu-Phe, Leu-Ile, Phe-Leu, Tyr-Ile and Ile-Ile were kindly provided by Christoph Böttcher (Julius-Kühn-Institut, Berlin). γ-Glu-Ile was synthesized as described in the following. For peptide coupling, protected glutamic acid (303 mg, 1.0 mmol), HOBt (149 mg, 1.1 mmol) and EDC (210 mg, 1.1 mmol) were suspended in dry CH 2 Cl 2 (10 ml) and stirred at 0 °C for 15 min. Isoleucine t-butyl ester hydrochloride (224 mg, 1.0 mmol) was added, then DIPEA (0.21 ml, 1.2 mmol) was syringed in one portion and the resulting solution was stirred at room temperature overnight (~12 h). The reaction mixture was diluted with 100 ml EtOAc, transferred to a separation funnel and sequentially washed with 0.5 M aqueous citric acid (2 × 50 mL), saturated aqueous NaHCO 3 (2 × 50 ml) and brine (1 × 30 ml). The organic phase was dried over anhydrous Na 2 SO 4 , filtered and concentrated under reduced pressure. To remove the Boc/ OtBu protection groups, the crude peptide was exposed to high vacuum for 1 h before dissolving it in a mixture 3:1 DCM/TFA (5 ml). Pressure from gas evolution generated during the dissolving process was regularly relieved by opening the reaction flask. After 3 h, no starting material was detected by thin layer chromatography and