Isoprene is the major biogenic volatile organic compound (VOC) emitted by the biosphere into the atmosphere. A wide breadth of research, and particularly the latest studies employing genetic engineering to produce transgenic plants with modified isoprene emission, have demonstrated that isoprene increases plant tolerance to oxidative stresses, caused by factors such as high temperature and ozone (Behnke et al., 2007; Loivamäki et al., 2007; Sasaki et al., 2007; Sharkey et al., 2005; Vickers et al., 2009, 2011). However, the mechanisms by which isoprene confers resistance against oxidative stresses remain unclear. A first hypothesis is that isoprene strengthens thylakoid membranes (Velikova et al., 2011). It has been shown that under physiological conditions there are less than a hundred isoprene molecules per million lipid molecules in the membranes (Harvey et al., 2015), casting doubt on this mechanism. However, there is evidence that isoprene positively influences the photosynthetic electron flow, reducing heat dissipation mechanisms and stabilizing thylakoid membranes, especially under stressful environments (Pollastri et al., 2014; Velikova et al., 2011). Recently, it has been shown that isoprene may act as a signaling molecule, inducing up-regulation of phenylpropanoid biosynthetic genes, indirectly enhancing plant resistance to heat and light stress (Harvey and Sharkey, 2016). Alternatively, isoprene may simply act as a proxy of induced activation of secondary metabolism (Tattini et al., 2015). A last hypothesis is that isoprene scavenges reactive oxygen species (ROS) by direct or indirect reaction (Loreto and Velikova, 2001). ROS exert an important signaling role in plants (Mittler et al., 2011) but can be generated in high amounts upon abiotic stresses, leading to cell damage and programmed cell death (Delledonne et al., 2001). Isoprene biosynthesis has been ascribed a ROS quenching effect, but, again, isoprene concentration in the leaf may be too low; abundant lipids and carotenoids can react with ROS much more rapidly than isoprene (Harvey et al., 2015). Nevertheless, emission of putative isoprene oxidation products that may be formed upon isoprene reaction with ROS has been reported, e.g. in a tropical rainforest mesocosm (Jardine et al., 2012), mango trees (Jardine et al., 2012, 2013), and creosote bush (Jardine et al., 2010). The putative isoprene oxidation products, the carbonyls methyl vinyl ketone (MVK) and methacrolein (MACR), are cytotoxic and must be rapidly removed from leaves once formed (Oikawa and Lerdau, 2013; Vollenweider et al., 2000). Plants also rapidly take up MACR and MVK (Andreae et al., 2002; Karl et al., 2004, 2005, 2010), suggesting that these compounds may be metabolized. Detoxification mechanisms for MACR have been reported (Muramoto et al., 2015), whereas this is not the case for MVK.

Methyl ethyl ketone (MEK) is an important oxygenated volatile organic compound (OVOC) for the atmosphere. In several studies the MEK concentration in the free troposphere has been reported to be approximately 25 % of the concentration of acetone (Moore et al., 2012; Singh et al., 2004). However, MEK has a much shorter lifetime than acetone due to a much higher reactivity (by about 1 order of magnitude) with hydroxyl radical (OH, kOH=1.15×10-12 cm3 s-1 at 296 K; Chew and Atkinson, 1996), which makes it important to total OH reactivity (Nölscher et al., 2016). Both biogenic and anthropogenic sources contribute to global atmospheric MEK (Yáñez-Serrano et al., 2016). In urban environments, MEK is typically the most abundant ketone after acetone (Feng et al., 2005; Grosjean et al., 1996), as it is widely produced in industrial processes (Legreid et al., 2007; Sin et al., 2001). The regional background of MEK was once attributed to anthropogenic sources only. More recent findings suggest that the biogenic source of MEK is comparable to the anthropogenic source (Yáñez-Serrano et al., 2015, 2016), but the origin of biogenic MEK has not been fully elucidated so far. Direct emission of MEK from vegetation seems to be the largest contribution since only a few minor biogenic VOCs (e.g, n-butane and 2-butanol) can lead to MEK via atmospheric oxidation (de Gouw et al., 2003; Jenkin et al., 1997; Singh et al., 2004; Sommariva et al., 2011). We recently found a possible relation between the detoxification by vegetation of MVK and the biogenic emission of MEK (Cappellin et al., 2017). We suggested that MVK could be efficiently detoxified by reduction reactions which lead mostly to MEK and to a minor extent to 2-butanol, though the full interconversion scheme was not studied.

In this study we explore the links between the production of isoprene from photosynthetic carbon, within-plant isoprene oxidation, and the biogenic emission of putative isoprene oxidation products MVK, MEK, and 2-butanol. We performed a series of fumigation experiments in which leaves were exposed to exogenous levels of a single carbonyl, and investigated the relationships between the uptake of each compound and the release of related products. We then studied the emission of carbonyls by leaves under progressively higher temperature conditions that likely inhibit photosynthesis and increase isoprene biosynthesis and the oxidative environment within leaves. We finally explored whether MVK, MEK, and 2-butanol produced under heat stress are products of in-plant isoprene oxidation by determining whether they share a labelling pattern with isoprene when leaves are fed 13CO2. Different labelling patterns would indicate that emitted carbonyls are not isoprene oxidation products, and would cast doubt on the hypothesis that isoprene acts as an antioxidant by reacting with ROS inside leaves.

There are consistent indications that MEK is released by vegetation at the leaf-, canopy-, and ecosystem-scale and that biogenic sources influence global atmospheric concentrations of MEK (Yáñez-Serrano et al., 2016, and references therein). However, no mechanism for biogenic MEK production has been demonstrated previously. It has been hypothesized that MEK could be generated following similar pathways to acetone, as a by-product of cyanogenesis in the transformation of a cyanohydrin lyase (Yáñez-Serrano et al., 2016). This idea has not been proven and emission would be limited to cyanogenic plants. Plant uptake of atmospheric MVK is a well-established phenomenon (Karl et al., 2010) and bidirectional exchange of MVK has been sometimes reported (Karl et al., 2005). Emissions of MVK and MEK have been rarely measured at leaf level and are challenging to measure because (i) emissions are small, especially in the case of MVK; (ii) separation of leaf and atmospheric sources in the presence of high isoprene emissions is difficult. Jardine et al. (2012, 2013) reported emissions of MACR and MVK in isoprene-emitting trees. They attributed carbonyl emissions to within-leaf isoprene oxidation, possibly by ROS induced under environmental stress conditions (e.g. heat stress).

Our results confirm the emission of carbonyls and show that the reduction of several carbonyls may occur in plants. In particular, we show that reduction reactions transform MVK into MEK, 3-buten-2-ol, and 2-butanol, and that the conversion of MEK into 2-butanol occurs in both directions (Fig. 5). This mechanism is the first possible pathway for the biogenic production and emission of MEK. The results also suggest that within-plant isoprene oxidation is not the source of these VOCs in our experiments.

Our fumigation experiments establish a causal link between-plant metabolisms of MVK and MEK production. Reaction yield calculations were performed considering the net total amount of MVK taken up during the experiment and the net total amount of MEK released. We show that 73±10 % of the MVK taken up by leaves was converted into MEK. Most of the remaining MVK was converted into 3-buten-2-ol and 2-butanol. Thus 97±6 % of MVK uptake was re-emitted as other volatile products. These data confirm previous reports at Harvard Forest, corroborated by laboratory experiments, where a correlation between uptake of MVK and emission of MEK, 2-butanol, and 3-buten-2-ol was found (Cappellin et al., 2017). The implication is that the metabolism of MVK is a likely source of MEK emissions.

Whether other mechanisms for MEK production by plants exist remains an open question. Processes on the leaf surface such as MVK adsorption or metabolization by leaf surface bacteria have been shown to be of minor importance in previous investigations of ketone and aldehyde leaf uptake (Omasa et al., 2000; Tani et al., 2010, 2013; Tani and Hewitt, 2009, p. 20). Future studies should investigate the fraction of global biogenic MEK emissions that could be accounted for by the proposed production mechanism.

Rapid transformation of MVK by secondary reactions can be interpreted as a necessary and efficient detoxification mechanism. MVK toxic effects include the accumulation of hydrogen peroxide and the activation of some stress-responsive genes (Alméras et al., 2003; Vollenweider et al., 2000). MVK metabolism has been reported in several plants (Karl et al., 2010; Tani et al., 2010). It has also been demonstrated that various plant cells are able to reduce unsaturated ketones (Kergomard et al., 1988). Long-term fumigation experiments with houseplants (Tani and Hewitt, 2009) have established that the uptake of several aldehydes and ketones exceeded by orders of magnitude the amount dissolved in leaf water, thus implying their metabolization as well. Transformation of MVK into MEK may be especially important in stressed plants as oxidative stress increases and endogenous production of MVK occurs (Jardine et al., 2012). The fumigation experiments suggest that the process can occur for exogenous MVK as well. Hence, the reduction mechanism also provides an explanation for plant uptake of MVK, a cytotoxic compound, which has been observed in many studies (e.g. Brilli et al., 2016; Karl et al., 2005, 2010).

MACR was also efficiently metabolized, consistent with other literature reports on tomato plants (Muramoto et al., 2015). In this case, however, detected volatiles produced by MACR reduction represented only about 9 % of the MACR uptake. By analogy with the case of MVK, the expected reduction products of MACR would be isobutyraldehyde, 2-methylprop-2-en-1-ol, and isobutyl alcohol. However, only a transient release of such products was detected. Isobutyraldehyde emission accounted for 6.5 % of the MACR uptake within 2 h, confirming results from a previous study on tomato plants (6.4 %; Muramoto et al., 2015). Muramoto et al. (2015) reported that most MACR in the leaf undergoes glutathionylation. Therefore, no MACR volatile products are released. Our results are in line with this interpretation, as MACR uptake implies a sink within leaves, suggesting a continuous metabolization of the compound, but the metabolic products are not released as volatiles.

The foliar uptake of MVK and MACR was rapid and sustained throughout the entire fumigation period, suggesting a large and fast metabolic sink. On the other hand, the uptake of secondary transformation products (namely MEK, 3-buten-2-ol, and 2-butanol) was not sustained, indicating that the metabolic sink was smaller and insufficient to rapidly scavenge the metabolites, leading to their temporary accumulation within leaves. Upon ending the fumigation, accumulated MEK, 3-buten-2-ol, and 2-butanol were released from the leaves back to the atmosphere (Fig. 2b, c, d).

A further aim of this study was to verify whether carbonyls emitted by plants come from isoprene oxidation in plants, as elsewhere proposed (Jardine et al., 2012). The experiments were conducted on red oak, one of the strongest isoprene-emitting plant species (Loreto and Sharkey, 1990). As isoprene biosynthesis is light- and temperature-dependent (Loreto and Sharkey, 1990), an increase in isoprene transformation products would be expected when leaves are illuminated and exposed to moderately high temperatures. This behaviour was indeed represented in the present data (Fig. 2). Under heat stress, photosynthesis and isoprene emission generally decouple and carbon sources other than direct products of photosynthetic carbon fixation are used for isoprene biosynthesis (Brilli et al., 2007). When leaves were severely heat-stressed we did not observe the expected uncoupling between photosynthesis and isoprene emission, as both parameters decreased. However, MVK + MACR + ISOPOOH, MEK, and 2-butanol increased. This increase would be predicted to occur under an increasingly oxidative environment where molecular O2 is directly photoreduced instead of being used for photosynthetic electron transport. Interestingly, the amount of MVK + MACR + ISOPOOH, MEK, and 2-butanol was similarly enhanced in illuminated and darkened leaves under the effect of high temperatures, despite the fact that no photosynthesis and minimal isoprene emission was present in darkness.

This unexpected observation was followed up with measurements of 13C labelling to further characterize the carbon source of MVK, MEK, and 2-butanol. As photosynthetic metabolites and isoprene label very quickly and almost totally when exposed to a 13CO2 atmosphere (Delwiche and Sharkey, 1993; see also Figs. 3, 4), we would expect that putative isoprene oxidation products and their further transformation products would be quickly and totally labelled as well. On the other hand, if carbon sources other than direct photosynthetic metabolites (including isoprene) are used for their biosynthesis, these compounds would not be labelled. The observation that no 13C labelling appeared in any of the emitted carbonyls suggests that isoprene is not involved in their biosynthesis. We are not aware of isoprene pools that label slowly in plants (Niinemets et al., 2004). However, sources independent of freshly assimilated carbon contribute to isoprene production when photosynthesis is stress-constrained (Brilli et al., 2007). Photorespiratory carbon may contribute to isoprene production, especially during dynamic changes of environmental conditions (Jardine et al., 2014). However, photorespiration also labels quite rapidly with 13CO2 (Delfine et al., 1999), and 13C would appear in MVK and MEK skeletons if contributed by photorespiratory carbon.

Isoprene labelling experiments using 13C-enriched glucose under CO2-free air conditions have provided a clear indication that isoprene may also incorporate carbon via the glycolytic pathway (Affek and Yakir, 2003), and possibly also by re-fixation of respiratory CO2 (Loreto, 2004; Loreto et al., 2007). This alternative carbon source can become prevalent when photosynthesis is stress-inhibited (Brilli et al., 2007), and may sustain small yet measurable emissions of isoprene in CO2-free air (Affek and Yakir, 2002) or in detached leaves with no photosynthetic activity (Brilli et al., 2011; Harrison et al., 2013; Loreto and Schnitzler, 2010). However, under the conditions used here photosynthesis was never dramatically impaired and isoprene emission was labelled completely, even under heat stress. We conclude that the emitted carbonyls are not isoprene oxidation products but derive from a different metabolism. It should be noted, however, that in nature, under high UV radiation, the emission of isoprene is generally far larger than that found in our experiment, and the oxidation potential of the atmosphere is also greater. Thus, we cannot completely rule out that isoprene also oxidizes forming carbonyls, namely MVK and MACR, in nature. If not isoprene, then what is the source of the emitted carbonyls in plants? MVK and MACR are peroxidation products of the trienoic fatty acid contained in chloroplast membranes (Alméras et al., 2003; Kai et al., 2012), and cause cytotoxicity once formed in leaves. MVK is also a component of flower scent (Knudsen et al., 1993) and may be a lipid peroxidation product, activating the expression of stress-related genes in Arabidopsis (Vollenweider et al., 2000). As we have demonstrated that isoprene is not responsible for the production of MVK and its transformation products in plants, further research should concentrate on investigating the relationships between emitted carbonyls and lipid catabolism in cellular membranes.