Effects of heat and drought stress on post‐illumination bursts of volatile organic compounds in isoprene‐emitting and non‐emitting poplar

Abstract Over the last decades, post‐illumination bursts (PIBs) of isoprene, acetaldehyde and green leaf volatiles (GLVs) following rapid light‐to‐dark transitions have been reported for a variety of different plant species. However, the mechanisms triggering their release still remain unclear. Here we measured PIBs of isoprene‐emitting (IE) and isoprene non‐emitting (NE) grey poplar plants grown under different climate scenarios (ambient control and three scenarios with elevated CO2 concentrations: elevated control, periodic heat and temperature stress, chronic heat and temperature stress, followed by recovery periods). PIBs of isoprene were unaffected by elevated CO2 and heat and drought stress in IE, while they were absent in NE plants. On the other hand, PIBs of acetaldehyde and also GLVs were strongly reduced in stress‐affected plants of all genotypes. After recovery from stress, distinct differences in PIB emissions in both genotypes confirmed different precursor pools for acetaldehyde and GLV emissions. Changes in PIBs of GLVs, almost absent in stressed plants and enhanced after recovery, could be mainly attributed to changes in lipoxygenase activity. Our results indicate that acetaldehyde PIBs, which recovered only partly, derive from a new mechanism in which acetaldehyde is produced from methylerythritol phosphate pathway intermediates, driven by deoxyxylulose phosphate synthase activity.


Introduction
Acetaldehyde and compounds commonly referred to as green leaf volatiles (GLVs,different C 6 compounds originating from the lipoxygenase (LOX) pathway, like E-2-hexenal, Z-3-hexenal, Z-3hexenol, etc.), are among the most prominent oxygenated volatile organic compounds (oVOC) in the earth's atmosphere (Guenther et al. 2012). Owing to indirect and direct sources from vegetation and microbial activity, they are frequently detected in different vegetative environments (see e.g. Kesselmeier & Staudt 1999, and references therein).
GLVs are often released in considerable amounts by plants under abiotic stress and herbivory or pathogen infection (Loreto & Schnitzler 2010, Holopainen & Gershenzon 2010, Scala et al. 2013. Their formation process in plant chloroplasts seems to be well understood nowadays: lipases are responsible for the release of linoleic acid (C18:2) and α-linolenic acid (C18:3) from galactolipids of the thylakoid membranes. In a second step, lipoxygenase (9-LOX and 13-LOX (linoleate:oxygen 9-or 13-oxidoreductase)) enzymes induce the degradation of these free fatty acids (Hatanaka 1993).
Subsequent action of hydroperoxide lyase (HPL), alcohol dehydrogenase (ADH) and isomerization factors are eventually responsible for a distinct temporal succession of the GLVs released (Fall et al. 1999).
It could be shown that the same GLVs are also transiently released by different plant species after suddenly terminating illumination (Holzinger et al. 2000, Graus et al. 2004, Loreto et al. 2006, Brilli et al. 2011, Ghirardo et al., 2011, Jardine et al. 2012, indicating that a similar mechanism might be involved here. It has been argued that these so called "post illumination bursts" (PIBs) might be caused by fast changes in intracellular pH values and consequently membrane stability after fast light-to-dark transitions (Brilli et al. 2011). In the algae Eremosphaera viridis a quick light-to-dark transition leads to a short term (1 -3 min) acidification of the cytosol (Bethmann et al. 1998). This effect could trigger the activation of LOX, which has its activity optimum at pH-values of 6.3 and 4.5 (Hatanaka 1993). Conversely, fast dark-to-light transitions lead to an alkalization (Bethmann et al. This article is protected by copyright. All rights reserved. 1998), which might explain why fast, repeated light-to-dark transitions don't cause PIBs, as has been shown previously using grey poplar leaves (Graus et al. 2004).
Experiments of Graus et al. (2004) and Jardine et al. (2012) revealed that GLV PIBs are strongly reduced when grey poplar leaves or mesquite (Prosopis spec.) branches, respectively, are kept under anoxic conditions. This is consistent with the supposed fatty acid degradation mechanism, since LOX activity requires oxygen.
Along with the GLVs, also acetaldehyde is emitted after leaf-cutting, during drying or leaf collapse under high temperature (Fall et al. 1999, Holzinger et al. 2000, Brilli et al. 2011, Behnke et al. 2013; transient acetaldehyde emissions have been detected after root flooding (Kreuzwieser et al. 2000(Kreuzwieser et al. , 2001 and after light-to-dark transitions (Holzinger et al. 2000, Karl et al. 2002, Brilli et al. 2011, Ghirardo et al. 2011, Jardine et al. 2012. Acetaldehyde emissions after root flooding are explained by an interconversion of ethanol formed in anoxic roots or stems. Root ethanol is transported to the leaves by the transpiration stream and then oxidized to acetaldehyde (by ADH) and eventually by aldehyde dehydrogenase (ALDH) to acetate (Kreuzwieser et al. 1999(Kreuzwieser et al. , 2000. Though there seems to be an apparent correlation between GLV and acetaldehyde emission after wounding and light-to-dark transitions, the underlying mechanisms are probably different. For acetaldehyde PIBs, some investigations were proposing a "pyruvate overflow mechanism" (Karl et al. 2002, Hayward et al. 2004. Cytosolic pyruvate was thought to increase transiently, thus lowering the cytosolic pH, due to a decreased pyruvate import into the organelles after darkening. When a threshold pyruvate concentration is reached, pyruvate decarboxylase (PDC) starts to convert pyruvate into pH neutral acetaldehyde, preventing acidification of the cytosol (Kimmerer & MacDonald 1987, Harry & Kimmerer 1991. Consistent with this pyruvate overflow theory, the inhibition of mitochondrial respiration and pyruvate transport, thus increasing pyruvate concentrations in the cytoplasm, induced acetaldehyde emissions (Karl et al. 2002).
This article is protected by copyright. All rights reserved.
With the same line of reasoning Jardine et al. (2012) explained the release of post illumination acetaldehyde, ethanol, acetic acid and acetone from mesquite branches in what they termed "pyruvate dehydrogenase (PDH) bypass pathway". In mitochondria of yeast and mammals, PDH catalyses the formation of acetyl coenzyme A (CoA) from pyruvate (see Wei et al. 2009 and references therein). In the PDH bypass mechanism, acetaldehyde is converted into acetate, which can serve as substrate for acetyl CoA synthesis in the chloroplasts. Acetyl CoA is a building block of fatty acids (Boubekeur et al. 2001, Wei et al. 2009) and together with the GLV Z-3-hexenol it can form Z-3-hexenyl acetate in a reaction catalysed by an acyl transferase (D'Auria et al. 2007).
In mesquite branches this was also the case for post illumination acetic acid, acetone, ethanol and the acetate moiety of Z-3-hexenyl acetate (Jardine et al. 2012). Graus et al. (2004) speculated that acetaldehyde emissions after fast light-to-dark transitions are not directly related to pyruvate, but rather to acetyl CoA. Similar to Jardine et al. (2012) they observed enhanced acetaldehyde PIBs when poplar leaves were kept under anoxic conditions. These enhanced bursts were interpreted as acetaldehyde released from excess acetyl CoA in the case of missing hexenal from the LOX pathway to form hexenyl acetate. In their experiments acetaldehyde PIBs were absent when the light was switched on again before the burst normally appeared.
Switching on and off light repeatedly within sub-minutes time spans, thus mimicking so-called light flecks in a natural environment, did not result in any PIBs at all.
Carbon isotope analysis of acetaldehyde emitted from leaves (of poplar, white oak, red maple and sassafras) after mechanical stress led Jardine et al. (2009) to the conclusion, that acetaldehyde is produced by fatty acid peroxidation reactions initiated by the accumulation of reactive oxygen species (ROS). Due to a kinetic isotope effect in the acetyl CoA formation from pyruvate by PDH, This article is protected by copyright. All rights reserved.
acetyl CoA-derived products like said fatty acids are depleted in 13 C (Park & Epstein 1961, DeNiro & Epstein 1977. Jardine et al. (2009) argued that acetaldehyde was as depleted in 13 C as these unsaturated lipids, while acetaldehyde derived from ethanolic fermentation, typical under anoxic conditions (e.g. after root flooding, in O 2 -free enclosure systems), would be less depleted (Hobbie & Werner 2004). Likewise, acetaldehyde released in leaf heating experiments seems to derive from a bulk biomass precursor, as carbon isotope analysis by Keppler et al. (2004) demonstrated.
Similar to acetaldehyde and the GLV, also isoprene PIBs have been reported in isoprene emitting plant species (Monson et al. 1991, Li et al. 2011. There is now evidence to suggest that the accumulation of methylerythritol phosphate (MEP) pathway intermediates, primarily methylerythritol cyclodiphosphate (MEcDP), is responsible for the post illumination isoprene burst phenomenon . The conversion of MEcDP to dimethylallyl diphosphate (DMADP) and isopentenyl diphosphate (IDP), catalysed by the iron-sulphur protein hydroxymethylbutenyl diphosphate (HMBDP) synthase, is strongly dependent on the electron transport of the photosynthetic light reaction. In darkness, HMBDP synthase switches over time to NADPH as its electron donor (Seemann & Rohmer 2007), thus relieving the intermediate pool and eventually causing the isoprene PIB in naturally isoprene emitting plants.
Aim of the present study was to elucidate the combined effects of heat and drought stress and elevated CO 2 concentration on the different post illumination emissions of volatile organic compounds (VOCs) from isoprene emitting (IE) and isoprene non-emitting (NE) poplar leaves. In NE plants, the knockdown of isoprene synthase (Behnke et al., 2007) affects early metabolites of the plastidic isoprenoid pathway  and lipid substrates of the LOX enzymes (Velikova et al., 2015). We used the combination of different biochemical and isoprene emission capabilities of IE and NE plants, in conjunction with the different stress-induced responses, to deepen our understanding of the biochemical mechanisms responsible for the post illumination burst of acetaldehyde and GLVs. Moreover, we tested the hypothesis that the amount of VOCs This article is protected by copyright. All rights reserved.

Plant material and climate change scenarios
A detailed description of the plant material and growth conditions of the plants used in this analysis is given in Vanzo et al. (2015); therefore, only a brief summary is given here. This article is protected by copyright. All rights reserved.
3. PS: periodic stress containing three cycles (each 6 days) with increased temperature and concomitant, acute drought (hereafter referred as heat and drought spell, HDS). The start of the 1 st HDS is termed as day 1 (d1) of the experiment. Between the 1 st and 2 nd and the 2 nd and 3 rd HDS a recovery time of two days was implemented, where temperature declined to control level (27 °C) and plants were irrigated to pot capacity. 4. CS: chronic stress with slowly developing drought progressing over 22 days from d0 to d22 (during these days temperature was increased as in PS).
The HDS in the CS and PS scenarios were followed by a final recovery time of 7 days (from d22 to d29) where temperature decreased to control level and pots were irrigated to saturation. These scenarios were considered separately and termed: 5. CSr: recovery from CS scenario. 6. PSr: recovery from PS scenario.
In the CS/CSr and PS/PSr scenarios the ambient CO 2 concentration was the same as in EC ([CO 2 ] = 500 μL L -1 ).
In our analysis, the EC scenario is the direct control of the CS/CSr and PS/PSr scenarios. There was no attempt to separate temperature and drought factors in this study.

Plant irrigation and simulation of water scarcity
The controlled water regime was obtained using automated drip irrigation systems placed in each pot half way between the stem and the edge of the pot. Plants were exposed to the short-term (in PS) and long-term drought (in CS) by reducing the amount of irrigation water gradually during each HDS. In the PS scenario, three drought cycles were imposed to mimic natural wet-dry cycles in the field. In the 1 st , 2 nd and 3 rd cycle the amount of water was reduced by 50%, 60% and 70% compared to AC and EC, respectively. To slow down the progression of drought in the CS scenario, in the first 5 days the irrigation amount was reduced by only 30% compared to fully watered controls in AC and EC. Every 5 days the water amount in CS was reduced by 10% reaching a reduction of 70% compared to the controls.

Leaf-level gas exchange measurements and VOC analysis
Leaf-level gas exchange measurements were made from d19-d22 (during HDS) and d25-d28 (during recovery from HDS). The measurements were performed using two Walz GFS-3000 leaf cuvette The PTR-ToF-MS was operated under standard conditions, 60 °C drift-tube temperature, 540 V drift voltage and 2.3 mbar drift pressure, corresponding to an E/N of 120 Td (E being the electric field This article is protected by copyright. All rights reserved. strength and N the gas number density; 1 Td = 10 -17 V cm 2 ). The instrument was calibrated once a week by dynamic dilution of VOCs using a standard gas mixture (Apel Riemer Environmental Inc., Broomfield (CO), USA), containing 20 compounds of different functionality distributed over the mass range of 30 -204 amu. Limits of detection were as low as 35 pptv. Full PTR-ToF-MS mass spectra were recorded up to mass to charge ratio m/z 315 with a 1 s time resolution. Raw data analysis was performed using the routines and methods described elsewhere (see Müller et al. 2013, and references therein).
In PTR-ToF-MS, acetaldehyde can be detected at m/z 45.034, ethanol at m/z 47.050 and isoprene at m/z 69.070. Certain substances like longer-chain aldehydes and alcohols are prone to fragment during the ionization in the drift tube of a PTR-MS (see e.g. Fall et al. 1999) and therefore, the GLVs can be detected at various mass-to-charge ratios, e.g.: This article is protected by copyright. All rights reserved.

Enzyme extraction and LOX activity analysis
The LOX activity (EC 1.13.11.12) was measured following the protocol described in Koch et al. (1992).
Leaf material (leaf no. 9-12 counting from the apex) of IE and NE plants was harvested at noon at the last day of stress treatment and seven days later at the end of the recovery period. Leaves were immediately frozen in liquid N 2 and stored at -80 °C until biochemical analyses. One-hundred mg of leaf material was homogenized with 50 mg Polyclar AT (Sigma Aldrich, Deisenhofen, Germany) on ice in 1 mL PEB (50 mM KP i pH 7.0; 0.1% (v/v) Triton X-100 and 20 mM DTT). After thoroughly mixing, the homogenates were centrifuged for 15 min at 16,000 g and 0.5 mL of supernatant were loaded on a pre-prepared NAP-5 column equilibrated and eluted with 1 mL of KP i (50 mM, pH 7.0). The sodium linoleate substrate was prepared by mixing 70 mg Tween-20 with 70 mg linoleic acid and 4 mL H 2 O thoroughly bubbled with argon to remove oxygen. The milky solution was cleared by adding 130 µL 2 N NaOH and filled up to 25 ml with oxygen-free H 2 O. The stock solution was stored at -20°C.
The kinetic of LOX activity was immediately measured spectrophotometrically at 234 nm with a dwell time of 10 sec. Two hundred µL of extract was mixed with 800 µL KP i (50 mM pH 7.0) and 8 µL of substrate (linoleic acid). After stabilization of enzyme activity, the reaction kinetic 234 nm was monitored for 5 min. LOX activity was measured at 25°C and is reported as µkat mg -1 total protein, assuming the molecular extinction coefficient of 2.5 x 10 4 M -1 cm -1 for the conjugated diene product (Axelrod et al. 1981). For each protein extract four technical replicates were analysed. The protein concentration was measured using the Qubit TM protein assay kit (Life Technologies GmbH, Darmstadt, Germany).

Statistical analyses
All measured data were statistically analysed in order to gain insight into the effects of plant genotype and environmental conditions on the release of PIBs. The experiment was designed in a This article is protected by copyright. All rights reserved. when the sample belonged to the corresponding class and to 0 otherwise. The X-variables were logarithmically transformed, centred and each type of data was block-wise scaled with 1 sd -1 . Each calculated significant principal component was validated using "full cross-validation", with 95% confidence level on parameters. The regression model OPLS was tested for significance using CV-ANOVA (Eriksson et al. 2008).
Moreover, two-way ANOVAs were performed in SigmaPlot (v13.0, Systat Software Inc., San Jose, CA, USA) on the means of the measured parameters using the two factors "Scenarios" (AC, EC, PS, CS, PSr, CSr) and "Genotypes" (IE, NE). Tukey post-hoc test followed the ANOVA to pairwise compare particular scenarios and the poplar genotypes. Results were considered significant at P < 0.05. compounds (Salch et al. 1995, Fall et al. 2001, Jardine et al. 2012, Shen et al. 2014. C 6 H 9 O 2 + could be a fragment of (E-2)-4-hydroxy-2-hexenal, a substance that was shown to be formed out of Z-3hexenal in soybean (Takamura & Gardner 1996). Since in our experiments the measured emissions of these unidentified compounds were low, they will not be considered further here.

Multivariate data analyses separate plant genotypes and treatment effects
Multivariate data analyses revealed a clear pattern of the samples subjected to the different climate scenarios, originating from IE and NE plant genotypes ( Fig. 2 and S2). Except for isoprene, all PIBs appeared in plants grown under unstressed conditions (AC+EC scenario), independently from plant genotypes, but they were mostly suppressed in plants grown under CS and PS scenarios. This is illustrated in Fig. 2: the first principal component (PC1) of PCA separates CS and PS samples (right This article is protected by copyright. All rights reserved. side of the loading plot in Fig. 2a) from samples grown in the AC, EC, CSr and PSr climate scenarios; VOC PIBs are located on the opposite side (negative correlation) of the loading plot (Fig. 2b). In contrast to IE plants, the NE genotypes were affected differently by the CS and PS treatments. This effect is illustrated by the clear treatment-dependent clustering within the NE samples (Fig. 2a). The PIBs originating from samples under CS and PS decreased in concert with the physiological parameters net assimilation and transpiration rates, but also with methanol emission rates (under light) and in-vitro LOX enzyme activities.
We employed OPLS analysis to further investigate which parameters correlate strongest with the plant genotypes or the different climatic conditions ( Fig. S2 and S3). The distinction between the four different plant genotypes mainly originated from their ability to either emit or not emit isoprene under constant light and was independent of treatment, as indicated by the high negative correlation coefficient (see Fig. S3a). The genotype effect was clearly revealed by the second principal component (PC2) in both PCA and OPLS ( Fig. 2 and S2, respectively) and was related to constitutive isoprene emission rates and partially to PIBs of isoprene and dark respiration (denoted as ' ' 'b69' an '-A' res ecti ely in b t Fi . 2b an 2; see also correlation coefficients in Fig. S3a).
This is the reason why we pooled together WT and PcISPS:GUS/GFP genotypes (IE) as well as RA1 and RA2 plants (NE) and considered for further analysis only the IE and NE class.
Overall, the OPLS model was very reliable as seen by the high numerical value of Q 2 (Y) = 94.4%.
This article is protected by copyright. All rights reserved.

Heat and drought spells had only little effect on post illumination bursts of isoprene
In poplar plants, isoprene is the most abundant VOC emitted from leaves; also other VOCs can be emitted significantly in a constitutive manner with leaf age dependent emission potentials (Ghirardo et al. 2011). In the transgenic poplar plants used in our experiments, knocking down of isoprene synthase (ISPS) causes the almost complete suppression of isoprene biosynthesis (Behnke et al. 2007). This had also impact on the PIBs of isoprene.
In experiments with IE plants, switching off the light caused the well-known PIBs of isoprene (cf. Fig.   1). During these bursts, approximately 0.04 to 0.06 nmol cm -2 of isoprene was emitted, independently from the plant treatment (see Fig. 3a; P > 0.05, see Supplemental table S1). Only in the CS scenario was the absolute amount of isoprene emitted as PIB slightly lower, although not statistically different. The observation that the PIB isoprene emission, like constitutive isoprene emission, is unaffected by HDS, supports the current knowledge that in both cases the emitted isoprene molecules derive from the same substrate. As mentioned earlier, the PIB of isoprene originates from a pool of MEP pathway intermediates after the light-dependent HMBDP synthase enzyme has switched to NADPH as electron donor and newly available DMADP is quickly converted into isoprene by ISPS .
Consistently, the lack of ISPS in NE plants resulted in a suppression of the typical isoprene PIBs, although in NE plants the chloroplastic DMADP pool is much higher than in IE plants ). However, from time to time we detected some tiny PIBs at the mass-to-charge ratio of isoprene (m/z 69.070) in NE plants recovering from HDS. It is unclear whether these signals were caused by isoprene or by a fragment of some interfering C 5 compound (Salch et al. 1995, Fall et al. 2001, Jardine et al. 2012, Shen et al. 2014). The latter could be supported by an apparent correlation of the appearance of these bursts with PIBs of an unidentified C 5 compound at m/z 71.086 about half a minute earlier. If these occasional early burst signals at m/z 69.070 appeared also in IE plants, they could likely not be resolved from the high isoprene signals slowly decaying after quitting the leaf This article is protected by copyright. All rights reserved.
illumination (cf. Fig. 1). While the maximum of regular isoprene PIBs appeared after a mean time span of 729 s (minimum ~ 400 s, maximum ~ 1294 s), the PIBs of NE plants appeared much faster after 102 s (minimum ~ 60 s, maximum ~ 193 s), see Fig. 4. Due to these facts, in such cases the PIB values were neglected in the actual analysis of isoprene PIBs (and Fig. 3a).

Acetaldehyde post illumination bursts as a result of DXS activity?
Acetaldehyde emissions following light-to-dark transitions have been observed in several plant species (see e.g. Holzinger et al. 2000, Karl et al. 2002, Graus et al. 2004, Ghirardo et al. 2011, Jardine et al. 2012. In our experiments the absolute amounts of C 2 compound (acetaldehyde+ethanol) emissions during PIBs were lower in NE than in IE plants (P < 0.001) and they were almost totally suppressed in the CS and PS scenarios (cf. Table 1 and Fig. 3b). This is also depicted by the negative OPLS correlation coefficients of acetaldehyde and ethanol in these scenarios (Fig. S3d+e). In plants recovering from HDS, the emissions as PIBs of the C 2 compounds were significantly lower than in the corresponding control treatments (P < 0.007 for acetaldehyde, P < 0.09 for ethanol).
Three sources of post illumination acetaldehyde emission can be imagined: (i) reaction of ADH to convert ethanol to acetaldehyde, (ii) action of pyruvate decarboxylase in the cytosol, and (iii) release during the complex kinetics of deoxyxylulose phosphate synthase.
This article is protected by copyright. All rights reserved.
The peaks of the ethanol PIBs appeared always after those of acetaldehyde (mean time difference of burst peaks: 58±18 s, see Fig. 4). The obvious correlation (R 2 =0.98) between acetaldehyde and ethanol PIBs (cf. Table 1) rather supports the reverse reaction of the ADH, in which ethanol is formed out of acetaldehyde.
When the light is abruptly turned off, CO 2 will continue to be fixed until all of the ribulose-1,5bisphosphate (RuBP) in the Calvin-Benson-Bassham cycle (CBB) is consumed (Laisk et al. 1984). ThDP. This step is the same as the first step of PDC. In the case of PDC, the hydroxyethyl ThDP is hydrolysed to acetaldehyde and ThDP. It is worth future research into whether the post illumination stromal environment causes DXS to act as a PDC for a short time.
Another possibility is that DXS could reverse part of the reaction sequence to liberate GAP and hydroxyethyl-ThDP or just acetaldehyde. Release of GAP from DMADP has been observed (A.

Banerjee, T.D. Sharkey, unpublished).
This proposed mechanism is coherent with findings of Karl et al. (2002). In 13 CO 2 labelling experiments with cottonwood leaves, after about 100 min of 13 CO 2 fumigation in light they found comparable labelling of both isoprene (78 %) and acetaldehyde (60%).
This article is protected by copyright. All rights reserved. A role for DXS in the acetaldehyde burst is also consistent with data presented in Fig. 3 and Fig. S3a, showing lower acetaldehyde PIBs in NE than in IE plants. It is known that high DMADP levels in chloroplasts act as feedback regulator of the MEP pathway, by inhibiting the activities of DXS (Wolfertz et al. 2004, Banerjee et al. 2013. In NE plants, the suppression of ISPS leads to an accumulation of chloroplastic DMADP, which causes a strong down regulation of the DXS activity ) and DXS protein levels, resulting in a reduced metabolic C-flux throughout the MEP pathway. Taken together, the lower DXS activities and protein levels of NE compared to IE plants ) and the proposed change in DXS catalytic properties might be the reason of the low PIBs of acetaldehyde from NE leaves. Moreover, the MEP pathway is ubiquitous in all plant species, which explains why the PIB emissions of acetaldehyde are detected independently from the trait to emit isoprene (Brilli et al. 2011).
Nonetheless, we cannot exclude that the differences between IE and NE plants derive from comprehensive changes in metabolite concentrations (Way et al. 2013) and metabolic fluxes , rearrangements of enzymatic and functional protein contents (Velikova et al. 2014) or changes in lipid composition (Velikova et al. 2015). Moreover, this proposed explanation for the differences between acetaldehyde emissions of IE and NE genotypes does not fully explain why the PIBs were suppressed in drought and temperature stressed plants (CS and PS scenario), although the constitutive emission of isoprene, an end product in the MEP pathway, in IE plants is unaffected by HDS. A possible explanation for this effect can be an up-regulation of ADH and ALDH genes during HDS, as has been reported for Arabidopsis (Seki et al. 2002).

Reduced LOX activity in HDS-stressed plants suppressed PIBs of green leaf volatiles
In unstressed plants (AC, EC) and plants recovering from HDS (CSr, PSr), switching off the light caused a strong transient release of different volatile C 6 compounds (GLVs). The GLVs could be associated with their precursor α-linolenic acid (C18:3), the most abundant fatty acid in lipids composing the chloroplast membranes of grey poplar (Velikova et al. 2015). This fact allowed to infer the isomeric structure of the GLVs emitted.
As examples for GLV emissions we show the amount of the isomers E-2-hexenal / Z-3-hexenal and Z-3-hexenol emitted as PIBs under the different climate scenarios (Fig. 3c+d). Generally, the PIBs of the  Fig. 3c+d). Overall, PIBs of GLVs negatively correlated to NE plants (Fig. S3a).
Apart from the EC scenario, NE genotypes generally showed lower PIBs of GLVs (cf. Table 1 and Fig. 3) compared with IE genotypes.
Additional biochemical analyses revealed that the in-vitro LOX protein activities were reduced in protein extracts from NE compared to IE (Fig. 5 and Fig. S3b) and correlated well with NE plants (Fig.   S3a). Thus, the lower amount of GLVs released from NE genotypes during light-to-dark transitions appears to be a consequence of both the reduced substrate availability -linolenic acid for the LOX enzymes, see Velikova et al. (2015)) and the lower LOX enzyme activities.
The PIBs of GLVs were almost completely suppressed during HDS (cf. Fig. 3 (Vanzo et al., 2015, supplemental data), indicating that the stress exposure was not detrimental for membranes. The applied eustress is per definition a mild, stimulating stress, strengthening plant resistance (Lichtenthaler, 1996). It can be speculated that the stress, plants experienced in the studies cited above, exceeded t e lant's li it f tolerance (= resistance minimum), leading to membrane damage and consequently GLV emissions.
In our experiments the suppressed GLV PIBs correlated with a notable reduction of in-vitro LOX activities found from leaf extracts (see Fig. 2b and Fig. 5), which was more pronounced in chronically-stressed plants than in periodically-stressed ones. However, even in stressed plants the apparent LOX activity did not drop to zero, indicating that additional mechanisms might have been involved in the suppression of in-vivo LOX activity following the light-to-dark transition. Earlier studies revealed altered portions of saturated and unsaturated fatty acids in response to drought.
For example, drought stress decreased the ratio of α-linolenic (C18:3) to linoleic acid (C18:2) in Carthamus tinctorius (Hamrouni et al. 2001) and caused a reduction in the abundance of C18:3 fatty acids as well as a complete disappearance of palmitoleic acid (C16:1) in Salvia officinalis leaves (Bettaieb et al. 2009). Also in Bermuda grass (Cynodon dactylon) the portion of unsaturated fatty acids decreased during drought stress (Zhong et al. 2011). It is likely that the reduction in unsaturated fatty acids in these examples also would cause a reduction of the PIBs of GLVs when analysed during drought stress. Another possible explanation for reduced GLV PIBs in HDS plants is the effect of elevated abscisic acid (ABA) levels. Drought stress leads to increased ABA transport from the roots to the leaves (e.g. Wilkinson & Davies 2002). It has been shown that ABA has a strong inhibitory effect on LOX, while it is increasing the antioxidant capacity in Orthosiphon stamineus (Ibrahim & Jaafar 2013). Both effects might result in a reduced formation of GLVs as we have seen herein. The strong suppression of GLV, but also acetaldehyde PIBs under drought and temperature stress may therefore be used as non-invasive marker for plant phenotyping (Niederbacher et al. 2015).
This article is protected by copyright. All rights reserved.
Apparently, in our experiments the HDS had no remarkable effect on the reaction speed during the formation of the LOX products (see Fig. 4). Only the end product Z-3-hexenyl acetate seems to have been formed with a certain delay in CS/CSr and PS/PSr plants compared to control plants.
A remarkable difference between the PIBs of GLVs and acetaldehyde/ethanol became evident in plants recovering from HDS (CSr and PSr scenarios). The PIBs of GLVs were markedly more pronounced in CSr and PSr plants compared to control plants, while the PIB emissions of C 2 compounds were reduced (see Table 1  Our results demonstrated that in grey poplar analysing PIBs of GLVs might be useful as a noninvasive phenotypic marker of stress-induced changes in free fatty acids or fatty acid composition of chloroplast membranes and in planta LOX activity under given environmental conditions. Based on the present data it might be promising to test in other plant species whether the reduction of PIBs of C 2 compounds and GLVs is a general phenomenon and is applicable as non-invasive phenotypic marker, quantifying the susceptibility of genotypes under drought and heat (eu)stress. ) and denote the area under the post illumination bursts, normalized to the leaf area. Since the PTR-ToF-MS has no isomeric separation capability and signals at several mass-to-charge ratios (m/z) correspond to fragments of more than one GLV, assigning sensitivities to certain ion signals is difficult. The emissions are therefore given in normalized counts per second per cm 2 leaf area [ncps cm -2 Leaf area ]. Refer to the main text for the assignment of molecules to m/z ratios. Statistical results are given in Supplemental table S1. Abbreviations used: IE: isoprene emitter, NE: isoprene non-emitter, AC: ambient control, EC: elevated control, CS: chronic stress, PS: periodic stress, CSr: chronic stress recovery, PSr: periodic stress recovery.

Figure 1:
Gas-exchange parameters (assimilation rate A (a), transpiration rate E (b) and water vapour conductance g H2O (c)) and signals of selected VOCs (d-f) measured in leaf cuvette experiments simulating light-to-dark transitions. The sample plants were grown under different climate scenarios: elevated control (EC, solid line), chronic stress (CS, dotted line) and chronic stress recovery (CSr, dotdashed line). The grey shaded area denotes a 10 min light period at the beginning of each experiment. After switching off the light, PIBs of several VOCs appeared such as acetaldehyde (d), isoprene (e), and the isomers Z-3-hexenal and E-2-hexenal (f). Under chronic stress PIBs were strongly suppressed. While PIBs of acetaldehyde were weaker in plants recovering from chronic heat and drought stress than in control plants, PIBs of GLVs like the hexenal isomers were enhanced. Signals are means of n=8 (EC) and n=4 (CS, CSr).
This article is protected by copyright. All rights reserved.  In the PS and CS scenario all PIBs except those of isoprene were strongly suppressed. NE species generally showed less intense PIBs of oxygenated VOC than IE species. PIBs of the C 2 compounds like acetaldehyde were less intense in the stress recovery treatments (PSr and CSr) than in the corresponding control treatment (EC). Contrary, PIB of C 6 compounds like hexenal and hexenol were enhanced in plants recovering from HDS. Letters denote statistically significant (P<0.05) different values: (s) between IE and NE plants within the climate scenario, (A) in regard to the control treatment (EC) within IE plants and (B) in regard to the control treatment within NE plants (cf. Supplemental table S1).