Exposure to different light intensities affects emission of volatiles and accumulations of both pigments and phenolics in Azolla filiculoides.

Many agronomic trials demonstrated the nitrogen-fixing ability of the ferns Azolla spp. and its obligate cyanobiont Trichormus azollae. In this study, we have screened the emission of volatile organic compounds (VOCs) and analyzed pigments (chlorophylls, carotenoids) as well as phenolic compounds in Azolla filiculoides-T. azollae symbionts exposed to different light intensities. Our results revealed VOC emission mainly comprising isoprene and methanol (~82% and ~13% of the overall blend, respectively). In particular, by dissecting VOC emission from A. filiculoides and T. azollae, we found that the cyanobacterium does not emit isoprene, whereas it relevantly contributes to the methanol flux. Enhanced isoprene emission capacity (15.95 ± 2.95 nmol m-2  s-1 ), along with increased content of both phenolic compounds and carotenoids, was measured in A. filiculoides grown for long-term under high (700 μmol m-2  s-1 ) rather than medium (400 μmol m-2  s-1 ) and low (100 μmol m-2  s-1 ) light intensity. Moreover, light-responses of chlorophyll fluorescence demonstrated that A. filiculoides was able to acclimate to high growth light. However, exposure of A. filiculoides from low (100 μmol m-2  s-1 ) to very high light (1000 μmol m-2  s-1 ) did not affect, in the short term, photosynthesis, but slightly decreased isoprene emission and leaf pigment content whereas, at the same time, dramatically raised the accumulation of phenolic compounds (i.e. deoxyanthocyanidins and phlobaphenes). Our results highlight a coordinated photoprotection mechanism consisting of isoprene emission and phenolic compounds accumulation employed by A. filiculoides to cope with increasing light intensities.

INTRODUCTION

Species belonging to the genus Azolla are globally widespread small free‐floating ferns (Lumpkin & Plucknett, 1980). All the Azolla spp. harbor in their leaflet cavities a complex microbiome (Banach et al., 2019; Nierzwicki‐Bauer & Aulfinger, 1991) mainly represented by the filamentous heterocystous cyanobacteria known as Anabaena azollae, recently revised as Trichormus azollae (Dijkhuizen et al., 2017; Kumar et al., 2019; Pereira & Vasconcelos, 2014). The association of Trichormus azollae with Azolla spp. is an obligatory and always mutual beneficial (symbiotic) relationship: on the one hand, the cyanobacterium fixes the atmospheric nitrogen (N2) and provides it mainly as ammonia to the fern for the biosynthesis of amino acids; on the other, the fern gives to the cyanobacterium a sheltered habitat which guarantees the supply of carbon and other nutrients (Peters & Meeks, 1989).

The efficient capacity of T. azollae to fix nitrogen from the atmosphere (Meeks et al., 1985) has been exploited in agricultural practices (Watanabe & Liu, 1992). Azolla is cultivated either as a monoculture or intercropped with other species (Ito & Watanabe, 1985; Roy et al., 2016; Wagner, 1997) because its high nitrogen content is slowly released into the soil following plant death and decomposition (Mahanty et al., 2017). Such biofertilizer capacity allows Azolla to improve the physical and nutrient soil properties (Pabby et al., 2004). The use of Azolla has been recently revived as an eco‐friendly solution to replace the chemical nitrogen fertilizers (Yo et al., 2016) and sustainably improve the rice yield (Lumpkin & Plucknett, 1980; Singh & Singh, 1987), even under sub‐optimal environmental conditions (Khumairoh et al., 2018). At the same time, co‐cultivation of crops with Azolla enhance atmospheric CO2 sequestration from agricultural ecosystems (Kollah et al., 2016). Indeed, both efficient capacity for N2 fixation (Silver & Schröder, 1984) and high photosynthetic rate (Allen et al., 1988) enable Azolla to spread fast and colonize freshwater habitats (Park & Song, 2017; Pereira, 2017; Speelman et al., 2009). Being perennial plants, Azolla spp. are also resistant to variations of environmental conditions over the seasons. In particular, these ferns display leaf reddening upon the occurrence of stress (i.e. high light, low temperature) due to the accumulation of polyphenolic compounds (Costarelli et al., 2021; Kösesakal, 2014; Pieterse et al., 1977). Anthocyanins and other classes of flavonoids have been recognized to play multiple roles in photoprotection: they alleviate the photo‐oxidative damage to the chloroplasts by absorbing the excess of incident light and contribute to scavenging of oxidants and free radicals (Agati et al., 2020; Landi et al., 2015), thus maintaining reactive oxygen species (ROS) to the level required to exert a signaling function (Mittler, 2002).

Despite many studies, fundamental aspects of Azolla physiology in response to abiotic stress remain poorly understood. It has been shown that the light intensity has an impact on the optimal growth of Azolla plants (Moretti & Gigliano, 1988; Peters et al., 1980) and the synthesis of both anthocyanins and chlorophylls (Zimmerman, 1985). However, light also affects the plants' emission of volatile organic compounds (VOCs) (Loreto et al., 2006; Loreto & Schnitzler, 2010). VOC are often associated with the production of phenolic compounds (Behnke et al., 2010a). In fact, volatile isoprenoids can exert an antioxidant activity per se (Vickers et al., 2009) and interact with other non‐volatiles isoprenoids (Beckett et al., 2012; Brunetti et al., 2015; Tattini et al., 2014), as well as with phenolic compounds (Liu et al., 2021; Tattini et al., 2015) to protect the photosynthetic membranes against oxidative damage occurring under abiotic stress conditions.

Fern species have already been reported to produce a wide variety of VOC from different biochemical pathways (Fons et al., 2018; Froissard et al., 2011; Kessler et al., 2014; Radhika et al., 2012). Nevertheless, emission of either acetylene (Zimmerman, 1985) or ethylene (Wong Fong San et al., 1987) from Azolla spp. have only been reported to date, and only one study measured the mixture of VOC released from the extract of Azolla plants (Pereira et al., 2009). Emissions of VOC from cyanobacteria have also been documented (Achyuthan et al., 2017; Milovanović et al., 2015), although the VOC profile emitted by T. azollae has never been analyzed.

In this study, we aimed to investigate the following aspects in Azolla filiculoides: (a) the whole blend of VOC in vivo, also dissecting the possible contribution of its cyanobiont T. azollae; (b) how growth under different light intensities affects, in the long‐term (1 year), the production of VOC, plant pigments (i.e. chlorophylls, carotenoids) and phenolic compounds; (c) to what extent either the emission of VOC and/or the synthesis of plant pigments and phenolic compounds contribute to the tolerance of A. filiculoides following a short‐term (4 weeks) exposure from low to very high light intensity.

MATERIALS AND METHODS

Experimental material and growing conditions

Azolla filiculoides

Plants of Azolla were bought from a local nursery in Pistoia (Italy) and further identified as Azolla filiculoides (Lam.) through molecular markers (Costarelli et al., 2021). The plants were grown in batch cultures by providing a liquid nutrient solution (as defined in Watanabe & Liu, 1992), replaced once a week to avoid nutrient limitation. Before they were subjected to any experimental tests, starting from the same initial plant material, the Azolla plants were cultured in seven large batches, which were left to acclimate for 1 year to different light intensities; in particular: (a) two batches were grown under low light (100 μmol of photons m−2 s−1), (b) four batches were grown under medium light (400 μmol of photons m−2 s−1) and (c) one batch was grown at high light (700 μmol of photons m−2 s−1). Illumination was provided by a LED system (C‐led) with a 12/12 h light/dark diurnal cycle in a growth room maintained at 23°C ± 2°C (Figure S1). Further experimental tests were: (d) one batch of plants acclimated to low light (100 μmol m−2 s−1) was exposed to a 10‐fold higher intensity (1000 μmol m−2 s−1, referred to very high light) for 4 weeks before starting measurements (Figure S1); (e) two batches of plants grown under 400 μmol m−2 s−1 were supplemented with antibiotics (one with erythromycin and one with novobiocin) at a concentration of 60 μg ml−1 to suppress the cyanobacterial symbiont, as well with 2 mM of NH4NO3 to compensate for the inhibition of the symbiont N2‐fixing activity and thus avoid nitrogen starvation (Forni et al., 1991). Two of the four batches of Azolla plants acclimated to 400 μmol m−2 s−1, with unaltered nutrient solution (Watanabe & Liu, 1992), were considered as controls to be compared with the antibiotic treated plants.

After VOC and chlorophyll fluorescence measurements, samples of A. filiculoides were collected and either frozen in liquid N2 and stored at −80°C for further analysis or placed in an oven at 60°C until a constant dry weight was reached. The dry weight was used to calculate the biomass density as dry weight/surface area of the Azolla crop before sampling.

Trichormus azollae

The cyanobacterium T. azollae used in this study was obtained from the Culture Collection belonging to the Institute of Bio Economy (IBE), National Research Council of Italy (CNR; Sesto Fiorentino, Firenze, Italy). The cyanobacterium was grown in batch mode using vertical glass columns (5 cm light path, 600 ml working volume) and BG110 (nitrogen‐free) as culture medium (Rippka et al., 1979). Continuous illumination of 40 μmol m−2 s−1 was provided by means of cool white lamps (Dulux L, 55W/840, Osram), and a culture temperature of 22°C ± 2°C was maintained by thermostat‐cultivation room. Cultures were bubbled with a sterile air/CO2 mixture (98/2, v/v) to ensure continuous mixing, remove dissolved oxygen, and maintain pH within the desired range (7.5–8.0).

Cultures reaching a biomass dry weight of 1–2 g L−1 were collected from three vertical glass columns and pooled together, centrifuged, and the resulting pellet dissolved in sterilized fresh BG110 medium to reach the desired concentrations before a subsample of 100 ml was used for VOC measurements. Dry weight (DW) determination was performed in triplicate using 5 ml culture samples filtered with pre‐weighted 47 mm diameter glass microfiber filter membranes (Whatman GF/C), washed twice with deionized water, oven‐dried afterwards at 105°C until constant weight, and weighed on an analytical balance (AT460 Delta Range Mettler‐Toledo).

Analysis of volatile organic compounds

Proton transfer reaction‐quadrupole mass spectrometer

Before starting the measurements of VOC emitted from A. filiculoides, plants were transferred to a plexiglass cuvette internally coated with Teflon foil, filled with 4 L of nutrient solution enclosing a headspace air volume of 22 L (Figure S2A). After closing, the cuvette was continuously flushed with 1 L min−1 of VOC‐free air obtained through a custom‐made catalytic converter that did not alter the air CO2 concentration nor the relative humidity (RH%) while purifying air from undesired VOC and contaminants. To improve the sampling of well‐mixed air, 10 cm long inlet‐ and outlet‐PFA Teflon tubes were inserted into the core of the cuvette. Part of the air (100 ml min−1) exiting the cuvette was diverted by mean of a PFA Teflon sampling line heated at 50°C, to a Proton Transfer Reaction‐Quadrupole Mass Spectrometer (PTR‐QMS) (Ionicon Analytic). Analysis by PTR‐QMS was performed through chemical ionization between high density H3O+ (produced in an ion source) and all the VOC present in the sampled air (having a proton affinity >165 kcal mol−1) in a drift tube under constant conditions of pressure (=2.2 mbar), temperature (=50°C), and electrical field (600 V cm−2) which result in an ionization energy E/N = 130 Td (Lindinger et al., 1998). In our analysis, we screened in real‐time all the protonated ions related to VOC and/or fragment of VOC having a mass unit (m/z) ranging between 20 and 220, with a dwell time of 1 s for each m/z, resulting in a single duty cycle of 200 s. Data acquisition by PTR‐QMS began immediately after the closure of the cuvette.

We applied the following equation reported by Marynick and Marynick (1975) to foresee when steady‐state conditions would have been reached within our cuvette:

where % (G) is the percentage of air originally present into the cuvette and that remains after a certain time since its closure (t); V is the volume of the cuvette (=22 L), and FR is the flow rate of air circulation inside the cuvette (=1 L min−1).

Our simulation indicates that only 20% of the air originally present within the cuvette would be present 35 min after closing it, and this fraction further decreases to 10% after 50 min (inset of Figure S2B). This theoretical calculation was further validated by the measured decay of background acetone concentration (detected as the protonated ion m/z = 59) present in the indoor air initially entrapped within the headspace of the cuvette following its closure (Figure S2B), leading to a steady‐state concentration already after 35 min. Only data recorded by the PTR‐QMS during five consecutive duty cycles (15 min long), following the 35 min wash‐out time were considered and averaged to calculate the VOC flux rates accordingly to this formula:

where F is the VOC flux rate (nmol m−2 s−1); [VOC] is the mixing ratio of VOC (nmol mol−1, or ppbv) in the air exits the cuvette; FR is the flow rate of air circulation inside the cuvette (=1 L min−1); A is the VOC emitting surface equal to the cuvette area (=504 cm−2) which was always fully covered by Azolla plants; DW is the Azolla plants dry weight (mg).

The PTR‐QMS background was recorded by measuring the VOC‐free air flowing out of the headspace of an empty cuvette (filled with only 4 L of nutrient solution) and then subtracted to the measurement of the VOC signals collected from the same cuvette containing Azolla plants. All the measurements were run initially under a light intensity of 400 μmol m−2 s−1 before switching the light off and recording under dark conditions for 35 min. The temperature of both the nutrient solution and the room air were always measured before and after each measurement and ranged between 22°C ± 1°C and 24°C ± 1°C, respectively. When measuring the response of isoprene emission to increasing measuring light intensities, the isoprene emission rates were normalized for consequent co‐variation of air temperature according to the model G93 of Guenther et al. (1993).

The same PTR‐QMS setting was used to screen the VOC emitted from T. azollae. An aliquot of T. azollae culture was transferred into a 1 L air‐tight glass flask fitted with Tefon valves and filled with 100 ml nutrient solution, continuously stirred and flushed with 500 ml min−1 of VOC‐free air under a light intensity of 400 μmol m−2 s−1 (Figure S3). Measurements were run at room temperature (24°C ± 1°C). Data were recorded by PTR‐QMS continuously for 50 min, but only the last 15 min (when the signal was stable for five consecutive duty cycles) were averaged to calculate the flux rate of VOC emission from T. azollae. The PTR‐QMS background of the empty flask (filled with only 100 ml of nutrient solution) was measured by flowing it with 500 ml min−1 VOC‐free air and then subtracted from the VOC signals recorded in the air exiting the flask containing T. azollae.

In the measurements of VOC emitted from both A. filiculoides and T. azollae, only the protonated ions with positive values after background subtraction were reported. For methanol and isoprene, conversion of PTR‐QMS signals (counts per second, cps) into concentrations (ppbv) was performed by measuring a multicomponent gas standard (Apel Riemer) after diluting it with VOC‐free air to achieve a concentration level within the ppbv range. The limit of detection (LoD) of methanol and isoprene were calculated as 3σ (sd) of the background and resulted in 0.09 ± 0.02 and 0.44 ± 0.03 ppbv, respectively.

Gas chromatography–mass spectrometry

Simultaneously to the PTR‐MS analysis, 1 L of air exiting the cuvette headspace was concentrated in a cartridge filled with 200 mg of Tenax GC (Markes International, Ltd) by mean of an external pump (Pocket Pump SKC Inc.) at a flow rate of 200 ml min−1. The cartridge was thermally desorbed for 15 min at 280°C with a helium flow rate of 50 ml min−1 (Markes International, Series 2 Unity), and the sampled air was transferred into a cold trap which was quickly heated from 10°C to 280°C. Isoprene was then separated and further identified with a 7890A gas chromatograph coupled with a 5975C mass detector (gas chromatography–mass spectrometry [GC–MS], Agilent Technologies) through fast injection onto a capillary column (ZB‐1, 60 m × 0.25 mm I.D. × 0.25 μm film of polymethylsiloxane; Phenomenex, Inc.) via a transfer line heated at 200°C. Unambiguous isoprene identification was achieved through software comparison of retention times and fragmentation patterns with the Wiley 275 database of mass spectra.

Chlorophyll fluorescence measurements

The chlorophyll fluorescence parameters were measured using an Imaging Pam M‐series fluorimeter (Heinz Walz GmbH). Before starting the measurements, plants were adapted to dark conditions for 30 min, and maximal PSII quantum yield (Fv/Fm) was calculated after applying a saturating pulse, according to Genty et al. (1989). Then, actinic light was stepwise increased to 11, 21, 26, 56, 81, 111, 146, 186, 231, 281, 336, 461, 531, 611, and 700 μmol m−2 s−1 and values of electron transport rate (ETR), effective PSII quantum yield (ΦPSII), non‐photochemical quenching (NPQ), and the coefficient of non‐photochemical quenching (qL, indicating the fraction of open PSII reaction centers) were calculated according to Genty et al. (1989), Maxwell and Johnson (2000), and Baker (2008). All chlorophyll fluorescence parameters were measured in different points (10–62) simultaneously in a single A. filiculoides sample and pooled with those collected in different samples (3–5) of A. filiculoides grown within the same batch under the same light intensity to have mean representative values ± se for each light condition.

Analysis of chlorophylls, carotenoids, and phenolic compounds

In order to determine the chlorophyll and carotenoid contents of the Azolla leaves 100 mg of fresh material were used. Leaves were mechanically broken with quartz until a fine powder was obtained and extracted twice in 80% acetone and centrifuged at 3000g (4°C) for 15 min. Supernatants were collected and read at the spectrophotometer for chlorophylls and carotenoids.

Chlorophylls were measured according to Ritchie (2006) by using the following equations;

The pigments concentration, expressed as μg g−1 fresh weight, were determined by using the following equations of Wellburn (1994):

All spectrophotometric readings were corrected by subtracting the absorbance at 750 nm to exclude the contribution of the residual scattering of the acetone solution (Wellburn, 1994).

In T. azollae, chlorophyll a content was determined according to Bennet and Bogorad (1973). Briefly, 5 ml of cyanobacterial suspension was centrifuged at 4000g for 15 min, extraction was performed in the resulting pellet with 5 ml of methanol at 70°C for 2 min, and the supernatant was recovered. The absorbance of the supernatant was measured with a Varian Cary 50 UV/Visible spectrophotometer (Varian, Mulgrave) at 665 and 750 nm. In order to exclude the contribution of the residual scattering of the methanol solution, the corrected absorbance A (665 nm) for each sample was obtained by subtracting the absorbance at 750 nm from the value of absorbance measured at 665 nm. The chlorophyll a concentration was calculated by using the following equation:

where, 12.65 was the extinction coefficient of chlorophyll a at 665 nm in methanol for T. azollae.

Anthocyanidins and deoxyanthocyanidins were extracted from Azolla plants as described by Dong et al. (2001) and Cohen et al. (2002), with minor modifications. Extraction with liquid nitrogen in 1.5 ml of 1% (v/v) HCl in methanol was performed on 100 mg of Azolla leaves left at 4°C in an overnight agitation. The extract was then centrifuged at 13,000g at 4°C for 10 min, and 1 ml of distilled water was added to the supernatant. Anthocyanidins were quantified using A530 for cyaniding 3‐glucoside (ɛ = 26,900 L m−1 mol−1, MW = 484.82 Da). Moreover, the extract obtained above was read at 479 nm for apigeninidin (ɛ = 38,000 L m−1 mol−1, MW = 255.24 Da) and 496 nm for luteolinidin‐5‐O‐glucoside quantification, as reported in Cohen et al. (2002). Soluble and insoluble protoantocyanidins were extracted from 100 mg of Azolla leaves and quantified as reported in Li et al. (1996).

Spectrophotometric determination of phenolic acids and flavonols was carried out according to Cassani et al. (2017), with some modifications. Samples of 15 mg of Azolla leaves were first boiled with 200 μl of distilled water for 30 min and then left at 4°C in agitation overnight with 1.5 ml of solution (1% HCl, 95% ethanol). This extract was then centrifuged at 13,000g at 4°C for 10 min, and both flavonols and phenolic acids were quantified from the supernatant through the absorbance at 350 and 280 nm, respectively. In particular, the amount of flavonols was calculated as quercetin 3‐glucoside equivalents (ɛ = 21,877 L m−1 mol−1, MW = 464.82 Da) and that of phenolics as ferulic acids equivalents (ɛ = 14,700 L m−1 mol−1, MW = 194.18 Da).

Phlobaphenes were extracted from 100 mg of Azolla leaves by adding sequentially 200 μl of concentrated HCl and 800 μl of dimethyl sulfoxide (DMSO) and applying vigorous vortexing after each of these additions. Extracts were then centrifuged at 14,000 rpm at 4°C for 45 min, and the supernatants were diluted with methanol (20% final concentration). The concentration of phlobaphenes was expressed as the value of absorbance recorded at their λ max (510 nm) per g of fresh weight, as reported by Landoni et al. (2020).

DNA analysis to quantify the relative abundance of Trichormus azollae in the fern

Before and after antibiotics treatment, DNA from A. filiculoides leaves was isolated, as reported in Doyle and Doyle (1987). The relative abundance of T. azollae DNA was calculated by using the RNase P RNA (RnpB) and Ribosomal protein L25 (RPL25) loci as well as the A. filiculoides ITS DNA locus in quantitative PCR (qPCR). The primer pairs employed in these amplifications are listed in Table S1. The qPCRs were performed using the BlasTaq 2× qPCR Mater Mix (abm), according to the supplier's instructions, into an ABI PRISM 7300 SDS apparatus (Applied Biosystems, Thermo Fisher Scientific, Massachusetts) and following these cycling parameters: an initial step at 95°C for 3 min, then 40 cycles each including a step at 95°C for 15 s and a step at 60°C for 1 min. A melting curve was added after each run. The efficiency of each primer pair was verified through the amplification of a serially diluted mixture of DNA samples, and the relative abundance of T. azollae DNA was calculated by employing the 2(−ΔCt) algorithm as reported in Schmittgen and Livak (2008), and considering the A. filiculoides ITS locus as reference. Four replicates for each gene tested were analyzed in three biological replicates.

Statistical analysis

Bonferroni t‐test was applied to perform all pairwise multiple comparisons between the mean values shown within a figure or table by always keeping a constant value of α = 0.05 without applying any further correction using the statistical package implemented in Sigma Plot version 12.0 (Systat Software Inc.).

Multiple correlations were tested among the content of phenolic compounds, leaf pigments and isoprene emission rates measured in Azolla plants exposed to different light conditions (n = 5–7) by using Minitab statistical package v20.2 (Minitab Inc.). In particular, a covariance‐based principal component analysis (PCA) was performed. In order to run this test, out of 204 data points collected in all the measurements, three data points were added by calculating the mean values from the data recorded at the same light intensity. To avoid dominance of high magnitude variables, the data were statistically standardized to get the unit variance for all variables.

In addition, differences in mean values of phenolic compounds, leaf pigments, and isoprene emission rates measured in Azolla grown at low light after the transition to very high light were specifically assessed by using one‐way ANOVA followed by a post hoc Tukey's test (α = 0.05).

RESULTS

Volatile organic compounds emission from symbionts

Emission of VOC from Azolla filiculoides grown under different light intensities

Our real‐time and in vivo wide‐range screening of the VOC emitted from A. filiculoides (including its cyanobiont T. azollae) highlighted that ~95% of the VOC blend is made up of isoprene and methanol (Tables 1 and S2). While the protonated ion detected at m/z = 33 (13.24% ± 0.72%) univocally identifies methanol, m/z = 69 (61.02% ± 14.20%) was assigned to isoprene after further separation and identification by gas chromatography–mass spectrometry analysis. Moreover, a ratio = 0.055 resulted between the protonated ions detected by PTR‐QMS at m/z 70 (=3.38% ± 0.76%) and that at m/z 69 (Table 1), as expected from the natural abundance of 13C/12C (=0.011) in the C5H8 isoprene molecule (one out of the five C being 13C). Finally, a small part of the isoprene molecule breaks down into a methylvinyl (C3) fragment with m/z = 41 (Brilli et al., 2007) in the PTR‐QMS, which in our analysis, accounts for ~16.87% ± 3.86% of the whole blend (Table 1). Since the ratio between m/z 42/41 = 0.035 calculated from the results of our analysis (Table 1) approached the theoretical one (=0.033), we can assign the protonated ion detected at m/z = 42 to the isotope of the methylvinyl fragment of isoprene.

TABLE 1

Mean abundances (±sd) of single protonated ions related to VOC and/or fragment of VOC on the overall blend emitted form Azolla filiculoides (n = 7)

Protonated ion (m/z)	Abundance (%)	Assignment to specific VOC and/or fragment of VOC
33	13.24 ± 0.72	Methanol (CH3OH–H+)
41	16.87 ± 3.86	Fragment of isoprene (C3H5+)
42	0.59 ± 0.13	Fragment of isoprene isotope (13C3H5+)
69	61.02 ± 14.20	Isoprene (C5H8+)
70	3.38 ± 0.76	Isoprene isotope (13C5H8+)
Total	~95%

The isoprene emission rate increased in A. filiculoides grown under higher light intensities (from 2.26 ± 0.25 in low light to 15.95 ± 2.95 nmol m−2 s−1 at high light; Figure 1A). However, the isoprene flux rates measured from plants grown under medium and high light were not significantly different, especially calculated per biomass area (Figure 1A).

FIGURE 1

Emissions of (A) isoprene and (B) methanol from A. filiculoides grown under a: (low) light intensity = 100 μmol m−2 s−1; (medium) light intensity = 400 μmol m−2 s−1; and (high) light intensity = 700 μmol m−2 s−1. On the right side of the dotted line, emissions from Azolla plants grown under a (low) light intensity = 100 μmol m−2 s−1 after exposure to a (very high) light intensity = 1000 μmol m−2 s−1. White bars are mean emission flux rates (+se) calculated on an area basis (left Y‐axis), whereas black bars are mean emission flux rates (+se) calculated on a weight basis (right Y‐axis). Uppercase letters on white bars indicate statistically significant differences between flux emission rates calculated on an area basis, and lowercase letters on black bars indicate statistically significant differences between flux emission rates measured on a weight basis from Azolla plants exposed to different light intensities (P < 0.05; n = 3–7)

The response of isoprene emission to increasing measuring light was similar in A. filiculoides grown under low and medium light intensities; in particular, the flux of isoprene increased exponentially by increasing measuring light (Figure S4A), but this was mainly an effect of a co‐occurring increase in air temperature. After normalization for the temperature dependency (Figure S4B), isoprene emission was again similar in response to increasing light to that measured when comparing plants grown under different light intensities, i.e. reaching saturation at medium light. In addition, following darkening, the post illumination decay of isoprene emission in A. filiculoides grown under low light was faster than that of plants grown under medium and high light (Figure 2).

FIGURE 2

Normalized isoprene emission following darkening. White circles are mean ± se of isoprene emission from A. filiculoides grown under a (low) light intensity = 100 μmol m−2 s−1 (n = 5); gray circles are mean ± se of isoprene emission from A. filiculoides grown under a (medium) light intensity = 400 μmol m−2 s−1 (n = 7); dark gray circles are mean ± se of isoprene emission from A. filiculoides grown under a (high) light intensity = 700 μmol m−2 s−1 (n = 3); black circles are mean ± se of isoprene emission from A. filiculoides grown under a (low) light intensity = 100 μmol m−2 s−1 after exposure to a (very high) light intensity = 1000 μmol m−2 s−1 (n = 3)

The emission rate of methanol from A. filiculoides was enhanced from 0.20 ± 0.08 in plants grown at low light to 1.67 ± 0.18 nmol m−2 s−1 in those grown at high light. The effect was clear when calculated on an area basis, whereas there was no significant difference between growth at low and high light when the flux of methanol was calculated on a weight basis (Figure 1B).

Emission of VOC from Trichormus azollae grown in vitro

Methanol represented ~50% of the overall blend emitted by the pure culture of T. azollae (Table 2). In addition to methanol, other protonated ions were detected and tentatively assigned to VOC and/or VOC fragments based on previous studies (Achyuthan et al., 2017; Ma et al., 2013; Milovanović et al., 2015; Yue et al., 2016) (Tables 2 and S3). In particular, when estimating the flux on a dry weight basis, methanol emission rate from pure cultures of T. azollae was of the same order of magnitude (=11.00 ± 5.00 nmol mg−1 DW s−1 × 10−4; Table S5) to that of A. filiculoides (Figure 1B). Since we did not detect isoprene emission from pure cultures of T. azollae, the entire isoprene emission from Azolla–Trichormus symbionts can be attributed to A. filiculoides.

TABLE 2

Mean abundance (±sd) of single protonated ions related to VOC and/or fragment of VOC on the overall blend emitted from Trichormus azollae pure culture (n = 6)

Protonated ion (m/z)	Abundance (%)	Assignment to specific VOC and/or fragment of VOC
33	46.89 ± 27.23	Methanol (CH3OH–H+)
42	13.22 ± 0.44	For example acetonitril
47	5.66 ± 27.16	Ethanol (C2H5OH–H+)
49	5.37 ± 0.90	For example methanethiol a (CH4S–H+)
51	1.00 ± 0.66	Unknown (?)
75	2.33 ± 1.09	Methyl Ethyl Ketone (MEK)
91	5.74 ± 2.93	Unknown (?)
109	2.59 ± 0.79	For example hydrocarbon fragment
127	2.21 ± 0.52	For example cyclohexane b (C8H14O–H+)
129	5.16 ± 2.45	For example nonane b (C9H20–H+)
Total	~90

Achyuthan et al., 2017.

Milovanović et al., 2015.

Inhibition of Trichormus azollae through antibiotics treatment of Azolla filiculoides

The addition of either erythromycin or novobiocin to the medium of A. filiculoides partially suppressed the cyanobiont T. azollae. Indeed, estimation by qPCR under a light intensity of 400 μmol m−2 s−1 showed a large reduction of the expression of genes of two cyanobacterial loci (RPL25 and RPNB) after antibiotic addition (Figure S5). Moreover, since T. azollae possesses only chlorophyll a (Table S4), the lower ratio of chlorophylls a/b measured in A. filiculoides treated with either erythromycin or novobiocin (Table 4) indirectly confirmed that antibiotics inhibited only T. azollae growth.

TABLE 4

Mean value ± se of chlorophylls and carotenoids contents in A. filiculoides grown under a (low) 100, (medium) 400, (high) 700 μmol m−2 s−1 light intensity, as well as in plants grown under a low light and then exposed to a very high (1000 μmol m−2 s−1) light intensity

Growth light intensity (μmol m−2 s−1)	Chlorophyll a (μg g−1 fresh weight)	Chlorophyll b (μg g−1 fresh weight)	Chlorophyll a/chlorophyll b	Carotenoids (μg g−1 fresh weight)
100 (n = 5)	1.50 ± 0.07c	0.30 ± 0.01b	4.99 ± 0.11b	0.59 ± 0.04b
400 (n = 7)	3.11 ± 0.14a	0.41 ± 0.05a	6.89 ± 0.19a	0.95 ± 0.11a
700 (n = 5)	3.05 ± 0.10ab	0.48 ± 0.02a	6.38 ± 0.10a	1.04 ± 0.04a
From 100 to 1000 (n = 5)	0.90 ± 0.17d	0.19 ± 0.03c	4.60 ± 0.42b	0.44 ± 0.04c
400 (+erythromycin) (n = 5)	2.65 ± 0.21b	0.53 ± 0.04a	5.00 ± 0.07b	0.97 ± 0.06a
400 (+novobiocin) (n = 5)	2.44 ± 0.10b	0.51 ± 0.02a	4.78 ± 0.09b	0.88 ± 0.05a

Note: Lowercase letters indicate statistically significant differences (P < 0.05).

The addition of either erythromycin or novobiocin to the medium of A. filiculoides did not impair the maximal yield of chlorophyll fluorescence, and thus the photosynthetic performances of the fern (Table 3). These antibiotics did not affect the content of carotenoids (Table 4) and phenolic compounds (Table 5), and the emission of isoprene (Figure 3). However, the emission of methanol decreased following antibiotic treatments (Figure 3). This further confirmed that the only source of isoprene from Azolla–Trichormus symbionts was A. filiculoides, whereas methanol was mainly produced by T. azollae.

TABLE 3

Values are mean ± se of biomass density (mg cm−2), maximal PSII quantum yield (Fv/Fm) measured after dark adaptation, electron transport rate (ETR) measured at 400 μmol m−2 s−1 of light intensity, estimated photosynthetic rate (μmol CO2 m−2 s−1) on the basis of the measured ETR according to Haimeirong et al. (2002), the amount of photosynthetically assimilated carbon (%) remitted as isoprene on the basis of the estimated photosynthetic rate and the measured isoprene fluxes (shown in Figures 1 and 3) in A. filiculoides grown under a (low) 100, (medium) 400, (high) 700 μmol m−2 s−1 light intensity, as well as in plants grown under a low light and then exposed to very high (1000 μmol m−2 s−1) light intensity

Growth light intensity (μmol m−2 s−1)	Biomass density (mg cm−2)	Fv/Fm	ETR	Estimated photosynthetic rate (μmol CO2 m−2 s−1)	Photosynthetically assimilated carbon remitted as isoprene (%)
100	0.78 ± 0.03b (n = 4)	0.803 ± 0.010a (n = 38)	63.5 ± 1.3b (n = 34)	5.25 ± 0.44	0.21
400	1.61 ± 0.23c (n = 6)	0.729 ± 0.006c (n = 57)	81.1 ± 0.8a (n = 58)	6.75 ± 0.07	0.81
700	3.21 ± 0.13a (n = 4)	0.714 ± 0.005c (n = 38)	74.7 ± 1.4c (n = 28)	6.17 ± 0.12	1.29
From 100 to 1000	1.49 ± 0.17bc (n = 4)	0.583 ± 0.018b (n = 38)	55.3 ± 2.7b (n = 12)	4.58 ± 0.22	0.17
400 (+erythromycin)	1.32 ± 0.15bc (n = 3)	0.753 ± 0.005ac (n = 9)	77.5 ± 1.5ac (n = 9)	6.46 ± 011	0.98
400 (+novobiocin)	1.37 ± 0.13bc (n = 3)	0.675 ± 0.019c (n = 10)	64.0 ± 2.7b (n = 10)	5.33 ± 0.33	0.96

Note: Lowercase letters indicate statistically significant differences (P < 0.05).

TABLE 5

Mean value ± SE of phenolic compounds in A. filiculoides grown under a (low) 100, (medium) 400, (high) 700 μmol m−2 s−1 light intensity, as well as in plants grown under a low light and then exposed to very high (1000 μmol m−2 s−1) light intensity

Growth light intensity (μmol m−2 s−1)	Anthocyanidins (nmol/gDW)	Phlobaphenes (Abs510/gDW)	Flavonols (μmol/gDW)	Phenolic acids (μmol/gDW)	Apigeninidin (nmol/gDW)	Luteolinidin‐5‐O glucoside (Abs496/gDW)	Soluble Proanthocyanidins (mg catechin/gDW)	Insoluble Proanthocyanidins (μmol/gDW)
100 (n = 5)	306.71 ± 55.49a	47.94 ± 4.63a	8.23 ± 0.88b	24.05 ± 0.78b	488.19 ± 107.58a	7.71 ± 1.74a	4038.79 ± 173.84ce	18.93 ± 1.05ab
400 (n = 7)	500.22 ± 59.79a	50.37 ± 4.08a	16.35 ± 1.26a	56.10 ± 3.46a	797.44 ± 70.06a	15.62 ± 1.34a	3532.16 ± 345.02abc	17.62 ± 0.88a
700 (n = 5)	883.41 ± 78.11b	85.76 ± 5.62b	12.64 ± 0.64a	46.43 ± 6.55a	1317.94 ± 98.18b	25.42 ± 1.81b	4869.94 ± 445.69de	18.59 ± 2.38ab
From 100 to 1000 (n = 5)	1068.43 ± 112.48b	90.11 ± 3.63b	15.54 ± 2.27a	17.97 ± 1.49c	2020.66 ± 234.96c	39.15 ± 4.97c	6063.94 ± 288.70d	25.77 ± 2.98b
400 (+erythromycin) (n = 5)	424.18 ± 37.53a	77.16 ± 3.73b	15.12 ± 0.94a	47.87 ± 2.56a	684.14 ± 35.79a	13.21 ± 0.74a	3052.28 ± 249.69ab	18.70 ± 0.96ab
400 (+novobiocin) (n = 5)	401.99 ± 24.19a	64.37 ± 4.53a	14.93 ± 1.47a	42.61 ± 3.92a	609.51 ± 30.18a	11.27 ± 0.52a	3210.18 ± 304.50ab	16.09 ± 1.43a

Note: Lowercase letters indicate statistically significant differences (P < 0.05).

FIGURE 3

Emissions flux rates of (A) isoprene and B) methanol from A. filiculoides before and after treatment with either erythromycin (white bars) or novobiocin (black bars); bars are mean (+se). Statistically significant differences between (isoprene and methanol) flux emission rates before and after the antibiotics treatment are indicated by asterisk (*) (n = 5; P < 0.05)

Long‐term growth of Azolla filiculoides under different light intensities

Growth of A. filiculoides was stimulated, in the long‐term, by increasing light intensities as shown by the significantly enhanced biomass density of plants under high light (3.21 ± 0.13 mg cm−2) compared with those grown under medium (1.61 ± 0.23 mg cm−2) and low light (0.78 ± 0.03 mg cm−2) (Table 3). Higher light‐dependent growth of A. filiculoides was likely associated with a higher ETR measured in plants acclimated to medium and high light (81.1 ± 0.8 and 74.7 ± 1.4, respectively) compared with plants grown at low light (63.5 ± 1.3; Table 3). Consequently, the estimated amount of carbon re‐emitted as isoprene increased from 0.17% to 1.29% with increasing growth light intensity (Table 3).

Light‐response curves of chlorophyll fluorescence also showed that ETR saturated at a higher intensity in A. filiculoides grown under a medium and high light than in plants grown at low light (Figure 4A). The PSII quantum yield (ΦPSII) was similar in plants grown at low, medium and high light when measured at light intensities <350 μmol m−2 s−1. However, when measured at light intensities >350 μmol m−2 s−1, ΦPSII of plants grown at medium or high light increased more than in plants grown at low light (Figure 4B). The level of the non‐photochemical quenching (NPQ) was similar when measured at light intensities <150 μmol m−2 s−1 irrespective of A. filiculoides growth light (Figure 4C). However, when measured at light intensities >150 μmol m−2 s−1, NPQ increased much faster in A. filiculoides grown under low light than in all other conditions (Figure 4C). Measurements of the open PSII reaction centers (qL) mirrored NPQ results, with plants grown in low light showing significantly lower qL than plants grown at higher light intensities (Figure 4D).

FIGURE 4

Chlorophyll fluorescence parameters in responses to increasing light intensities: (A) electron transport rate (ETR); (B) effective PSII quantum yield (ΦPSII); (C) non‐photochemical quenching (NPQ); (D) coefficient of photochemical quenching (qL). White circles are mean ± se of data collected from A. filiculoides grown under a (low) light intensity = 100 μmol m−2 s−1 (n = 27–38); gray circles are mean ± se of data collected from A. filiculoides grown under a (medium) light intensity = 400 μmol m−2 s−1 (n = 41–62); dark gray circles are mean ± se of data collected from A. filiculoides grown under a (high) light intensity = 700 μmol m−2 s−1 (n = 25–28); black circles are mean ± se of data collected from A. filiculoides grown under a (low) light intensity = 100 μmol m−2 s−1 after exposure to a (very high) light intensity = 1000 μmol m−2 s−1 (n = 10–15)

A. filiculoides grown under low light intensity showed the lowest concentration of chlorophylls (both type a and b) and carotenoids (Table 4), as well as flavonols and phenolic acids (Table 5). Higher light intensities stimulated the production of chlorophylls and carotenoids, but these were not significantly different in A. filiculoides grown under medium and high light (Table 4).

Moreover, A. filiculoides grown under high light showed an enhanced concentration of anthocyanidins phlobaphenes, 3‐deoxyanthocyanidins, apigeninidin luteolinidin‐5‐O glucoside, and soluble proanthocyanidins than plants grown under medium and low light intensity (Table 5). However, the concentration of flavonols and phenolic acids were not significantly different in plants grown under high and medium lights (Table 5).

When evaluated together, the content of chlorophylls, carotenoids, phenolic compounds and the emission of isoprene measured in A. filiculoides grown under different light intensities formed distinct clusters on the PCA score plot (Figure 5A). While the increasing amounts of anthocyanidins, 3‐deoxyanthocyanidins apigeninidin, luteolinidin‐5‐O glucoside, soluble‐, and insoluble‐proanthocyanidins all positively correlated with medium and high growth light, a positive correlation between enhanced content of flavonols, carotenoids, chlorophylls and isoprene emission was found only with medium growth light intensity (Figure 5A). A different trend characterized phenolic acids, whose content correlated positively with medium light but negatively with high light. In addition, only phenolic acids showed negative relationships with all the other compounds across different growing light intensities (Figure 5B). Under low growing light conditions, the content of chlorophylls, carotenoids, isoprene emission and many phenolic compounds (except for phlobaphenes) positively correlated to each other (Figure S6A), whereas negative correlations started to occur under medium (Figure S6B) and high light (Figure S6C).

FIGURE 5

(A) Principal component analysis (PCA) biplot which combines all 17 measurements collected in A. filiculoides grown under A: (low) light intensity = 100 μmol m−2 s−1 (square symbols; n = 5); (medium) light intensity = 400 μmol m−2 s−1 (circles symbols; n = 7); (high) light intensity = 700 μmol m−2 s−1 (triangle symbols; n = 5). The first principal component (PC) had the largest variance (=58%) and, together with second PC (=24%) captured 82% of total variance. Remaining PCs were excluded. The components loading vectors were proportionally superimposed to their contribution. (B) Correlogram showing the degree of correlation between the content of chlorophylls, carotenoids, phenolic compounds, and isoprene emission rates measured in A. filiculoides plants grown under all the different light intensities of: 100 (low), 400 (medium), and 700 (high) μmol m−2 s−1

Response of Azolla filiculoides grown at low light after short‐term exposure to very high light

Exposure of A. filiculoides plants grown under low light (100 μmol m−2 s−1) to a 10‐fold higher (1000 μmol m−2 s−1) intensity for 4 weeks did not induce a significant increase in the biomass density and negatively affected the maximal yield of chlorophyll fluorescence (Fv/Fm; Table 3).

As for VOC, this short‐term exposure did not affect methanol and only slightly decreased isoprene emission (Figures 1 and S7). Moreover, following this change in light intensity, the post‐illumination decay of isoprene emission slowed down and became similar to the kinetic of plants grown under medium and high light (Figure 2).

After exposure to very high light, chlorophyll fluorescence light‐response curves still showed a lower saturation level of ETR than that of A. filiculoides grown under a medium and high light (Figure 4A), although the ETR began to plateau at a higher light intensity than 400 μmol m−2 s−1. ΦPSII dropped at very low measuring light in plants previously exposed to very high light, reaching values much lower than in plants grown under the other light conditions, but similar to those seen in plants grown in low light and measured at the highest light intensities (Figure 4B). Upon exposure to very high light, the value of NPQ and qL followed the same trend as those measured in A. filiculoides grown under medium and high light intensity (Figure 4C).

The shift from low light to very high light reduced the concentration of chlorophylls and carotenoids and dramatically altered the color of A. filiculoides (Figure S1D) by significantly increasing the level of anthocyanidins (~348%), apigeninidin (~495%), luteolinidin‐5‐O‐glucoside (~330%), phlobaphenes (~187%), flavonols (~189%), and soluble proanthocyanidins (~150%); only the concentration of phenolic acids decreased (~15%) (Table 5; Figure S7).

DISCUSSION

Emission of isoprene from Azolla filliculoides is comparable as flux rate and light‐dependence to those of other vascular plant species

Only one study has, so far, investigated the VOC mixture released from A. filiculoides after distillation of plant material (Pereira et al., 2009). In this work, we detected in vivo VOC emitted from A. filiculoides. Our screening revealed that such emission is mainly constituted of isoprene and methanol, which together accounted for ~95% of the overall blend (Tables 1 and S2). Isoprene emission is widespread in hygrophytic species (Loreto et al., 2014), and terrestrial ferns have been already documented to emit isoprene (Hanson et al., 1999; Monson et al., 2013; Tingey et al., 1987). Our real‐time analysis showed that the flux rate of isoprene emission from A. filiculoides, calculated on an area basis, is as high as that measured, under similar light and temperature conditions, in high‐emitting plants (Geron et al., 2001; Kesselmeier & Staudt, 1999), either herbaceous (e.g. Phragmites spp.; Fares et al., 2008) or tree species (e.g. Poplar spp.; Brilli et al., 2007). Due to the vast area where Azolla is cultivated in rice paddies, the emission rate we have measured makes Azolla spp. a potentially important isoprene source to be accounted for by global models of biogenic VOC (Guenther et al., 2006). Moreover, the photosynthetic rate of A. filiculoides we estimated based on the ETR (Haimeirong et al., 2002) is consistent with values directly measured in Azolla pinnata (Allen et al., 1988). The amount of photosynthetically assimilated carbon used by A. filiculoides to produce isoprene is thus ~1%, as typically found in non‐stressed emitting plants (Sharkey & Yeh, 2001).

Long‐term (1 year) growth under higher light conditions enhanced the isoprene emission capacity of A. filiculoides (Figure 1). This is consistent with previous studies on terrestrial plants showing increasing steady‐state isoprene emission rates at higher growth lights (Hanson & Sharkey, 2001; Harley et al., 1997; Sharkey et al., 1991), driven by either larger availability of the photosynthetic intermediate substrate dimethylallyl diphosphate (DMADP) or greater isoprene synthase activity (Niinemets & Sun, 2014). The integrated area of post illumination isoprene emission provides an in vivo estimate of the DMADP pool employed for isoprene production before darkening (Rasulov et al., 2009a). The rapid decay of isoprene after darkening in A. filiculoides plants acclimated to the lowest light intensity for a year suggests that DMADP availability might have indeed limited isoprene emission in this case (Niinemets & Sun, 2014). Past studies have already reported how variations in DMADP pool size are key in controlling isoprene emission during transient changes of light intensity (Rasulov et al., 2009b). A similar response of isoprene emission to fast changes in light and temperature (Figure S4) in A. filiculoides grown at low and medium light most likely indicates that energy intermediates (i.e. ferredoxin, NADPH, ATP, and CTP) and isoprene synthase activity (Lantz et al., 2019) are similarly available in plants grown under different light intensities.

Exposure to very high light of Azolla filiculoides grown at low light does not induce an increase in isoprene emission capacity

The transition of A. filliculoides from low to a 10‐fold higher light intensity for 4 weeks slightly decreased isoprene emission capacity (Figures 1 and S7). This result was unexpected due to the light‐dependency of isoprene emission that we have confirmed in Azolla plants grown at different light intensities (Figure S4B). Enhanced isoprene emission could have been prevented by the diversion of substrate DMADP for the synthesis of other photoprotective molecules. Indeed, production of photosynthetic pigments through the 2‐C‐methylerythritol 5‐phosphate (MEP) pathway has been shown to limit isoprene emission by competition for DMADP during the early stage of leaf development (Rasulov et al., 2014), and suppression of isoprene emission shunts carbon to carotenoid synthesis (Ghirardo et al., 2014). However, we measured a decrease in carotenoids content when A. filiculoides acclimated to low light was exposed to a 10‐fold higher light intensity for 4 weeks (Table 4; Figure S7A). Therefore, the whole MEP pathway was likely downregulated following the transition from low to very high light. As an alternative hypothesis, isoprene synthase activity, which has often been suggested to control isoprene upon changing environmental conditions (Fortunati et al., 2008), might have limited isoprene emission when A. filiculoides plants were exposed to very high light. In fact, the integrated area of post illumination isoprene emission (and thus DMADP availability) was higher in A. filiculoides plants after exposure to very high light intensity (Figure 2), but this did not result in a higher isoprene emission rate (Figure 1A). On the other hand, the similar dependence to both light and temperature of isoprene emission (Figure S4) imply that isoprene synthase was similarly active in Azolla grown under different light conditions and rather suggests that a reduced level of isoprene synthase protein could have prevented, in the short‐term, an increase in isoprene emission following exposure to very high light (Wiberley et al., 2005).

Transition from low to very high light reduces carotenoids and isoprene production but greatly enhance the levels of phenolic compounds in Azolla filiculoides

Long‐term growth under higher growing light conditions enhanced isoprene emission as well as the accumulation of carotenoids and phenolic compounds without affecting the chlorophyll content in A. filiculoides. However, in the short time, exposure of A. filiculoides grown at low light to very high light reduced carotenoids production (Table 4) and isoprene emission capacity (Figure 1A). These results differ from what is often observed in higher plants, where carotenoids have shown protective roles against light stress (Harvaux & Kloppstech, 2001), and an enhanced isoprene emission by increasing light intensity exerts a protective function on the photochemistry of photosynthesis (Behnke et al., 2010b; Sharkey & Singsaas, 1995).

On the other hand, multiple correlations were found among leaf pigments, phenolic compounds, and isoprene emission, suggesting that an efficient combination of photoprotective metabolites operates in Azolla grown under increasing light intensity (Figure 5). In particular, since many of the mutually positive correlations found in Azolla grown under low light became negative under medium and high light (Figure S5), trade‐offs and thus preferences between the production of some flavonoids (e.g. apigeninidins, anthocyanidins) and phenolic acids occurred under increasing growing light intensities. Under medium growing lights, chlorophyll a, and carotenoids were still strongly positively correlated, while both were negatively correlated with isoprene emission rate (Figure S6B). The latter was consistent with the proposed and observed constraints on photosynthetic carbon distribution among volatile and other non‐volatile isoprenoids such as carotenoids and phytol chain of chlorophylls from the MEP pathway under unstressed conditions (Dani et al., 2016, 2021). In addition, the concentration of apigeninidins, luteolinidin, anthocyanidins, phlobaphenes positively correlated both within and across different growing light conditions (Figures 5 and S6). This indicates that, on the one hand, phenylpropanoid synthesis seemed to be always stimulated in Azolla by increasing light intensity (Cronin & Lodge, 2003) and, on the other, the production of flavonoids through the shikimate/phenylpropanoids pathways rather than carotenoids and isoprene through the MEP pathway was favored by high light conditions.

Indeed, the synthesis of all phenolic compounds originating from the phenylpropanoid pathway, including anthocyanins, deoxyanthocyanidins (apigeninidin and luteolinidin), phlobaphenes, and soluble proanthocyanidins was particularly stimulated when A. filiculoides grown under low light was exposed to a 10‐fold higher light (Table 5; Figure S7C–J). These phenolic compounds are responsible for the reddish leaf color observed in Azolla spp., especially in wintertime (Costarelli et al., 2021; Pieterse et al., 1977), being the red pigment luteolinidin 5‐O glucoside recognized as a peculiar taxonomic marker of the genus Azolla (Iwashina et al., 2010). Leaf reddening is also common across higher plant species where it occurs as an adaptive mechanism to tolerate multiple environmental stresses (Hughes, 2011). The ability of Azolla spp. to rapidly accumulate phenolic compounds in response to very high light could result in an effective photoprotection mechanism, especially in wintertime, when isoprene emission is inhibited by the low temperatures (Monson et al., 1992). Indeed, the synthesis of flavonoids was triggered in Azolla spp. after exposure to higher UV radiation (Jayakumar et al., 1999), and their accumulation has been already demonstrated in Azolla plants to scavenge an excessive ROS production induced by cadmium stress (Dai et al., 2012). Among the phenolic compounds we analyzed, phenolic acids were the only ones that were not stimulated by increasing light conditions (Figure S7E,F). Thus, we hypnotize that the synthesis of phenolic acids may have been downregulated in Azolla to generate more flavonoids under very high light. Phenolic acids are among the first products made by the phenyl propanoids pathway, and their synthesis can be less sensitive to light than that of flavonoids (Lillo et al., 2008).

Photochemistry measurements suggest adaptation to higher light intensity but photo‐inhibition after exposure from low‐ to very high light in Azolla filiculoides

Consistent with the literature, the saturating ETR values we have measured in A. filiculoides confirmed 400 μmol m−2 s−1 to be the optimal light intensity that maximize the growth of Azolla spp. (Peters et al., 1980; Zimmerman, 1985). A. filiculoides plants grown under 400 (medium) and 700 (high) μmol m−2 s−1 of light intensity (Figure 4) showed similar photosynthetic performances (similar light‐curve responses of ΦPSII, NPQ and qL), indicating that these plants were able to adapt to high growth lights and this reflected in higher biomass density and thus better productivity (Table 3). When grown under low light, A. filiculoides displayed the lowest ETR, ΦPSII, qL and the highest NPQ in response to increasing measuring light intensities, demonstrating a down‐regulation of the photochemical apparatus but also an efficient capacity to dissipate the excess of light as heat. However, A. filiculoides grown at low light did not immediately acclimate after exposure to very high light, despite the observed increase in photo‐protective phenolic compounds (Table 5). Indeed, in these plants, a stressful condition of the photosynthetic apparatus was revealed by the reduction of PSII quantum yield (both in dark and light conditions) and of ETR (Table 3; Figure 4A,B), which was not accompanied by the expected increase of NPQ (Figure 4C). The latter, in particular, likely indicates that excess light might not have been efficiently dissipated as heat after the transition from low to very high light. As ETR decreased only by ~13% after transition from low to very high light (Table 3), A. filiculoides may have diverted the surplus of light energy to an upregulated photorespiratory sink (Dani et al., 2014; van Kempen et al., 2016). However, it is also possible that non‐photosynthetic pigments (e.g. anthocyanins), which accumulated after exposure to very high light, might have interfered with measurements of photosynthetic light use, leading to a less reliable estimation of chlorophyll fluorescence parameters in Azolla plants under stress (MacLane & Sharkey, 2019).

Dissecting VOC emissions from symbionts: Isoprene is emitted only by the fern and methanol mainly by the cyanobacterium

We have dissected the VOC emission from A. filiculoides and T. azollae by analyzing changes in VOC emission before and after treatment with antibiotics inhibiting the cyanobiont. In particular, the addition of erythromycin or novobiocin to the nutrient solution decreased the chlorophyll a content of A. filiculoides by 15% and 21%, respectively (Table 4), without inducing stressful conditions (Table 3). Since T. azollae can contribute 10%–20% of the chlorophyll a of Azolla plants (Peters & Mayne, 1974), the lower amount of chlorophyll a after the antibiotic treatments indicates an almost complete inhibition of the symbiont cyanobacteria. The antibiotic efficacy in inhibiting T. azollae was further confirmed by DNA quantification through qPCR analysis (Figure S3).

Our results highlight that isoprene emission was not reduced following the antibiotics treatment inhibiting T. azollae (Figure 3A) and that isoprene was not detected from pure cultures of T. azollae. Therefore, we conclude that A. filiculoides is the only isoprene‐emitter in the Azolla–Trichormus symbiotic system. Moreover, we showed that T. azollae mostly contributed to methanol flux released from the Azolla–Trichormus symbiotic system, as indicated by the following three results: first, methanol was the most abundant VOC released from pure cultures of the cyanobacterium (Tables 2 and S3), at a rate similar to that measured from the A. filiculoides plants (Figure 1B; Table S4); second, the antibiotic treatments inhibiting T. azollae caused methanol emission to drop (Figure 3B); third, increasing light intensity enhanced biomass density, and thus stimulated A. filiculoides growth (Table 3), without significantly affecting the flux of methanol calculated on a weight basis (Figure 1B). A previous study did not report methanol among the mixture of VOC emitted from Trichormus spp. and other cyanobacterial strains (Milovanović et al., 2015). This is possibly due to analytical conditions not suitable to retain such a low boiling VOC, whereas the PTR‐QMS technique used in our experiment, due to the high proton affinity of alcohols, is particularly sensitive to methanol (Brilli et al., 2011).

Methanol is one of the most abundant VOC exchanged among ecosystems (Wohlfahrt et al., 2015). In leaves, methanol is formed through the pectin methyl esterase (PME) activity degrading the pectin polysaccharides of cell walls (Nemecek‐Marshall et al., 1995), and its emission is particularly high during plant growth (Brilli et al., 2016; Manco et al., 2021). However, methanol emission in T. azollae might originate from a different mechanism because cyanobacterial cell envelops, consisting of a peptidoglycan layer and an outer lipopolysaccharides membrane, frequently covered by an external surface embedding carbohydrate structures (Hoiczyk & Hansel, 2000), do not include pectin. Therefore, methanol might be produced by T. azollae through demethoxylation of cellular polysaccharides, as already observed in microalgae, cyanobacteria (Achyuthan et al., 2017), and marine phytoplankton (Mincer & Aicher, 2016). Alternatively, methanol may be a by‐product released during the biosynthesis of indole alkaloids (e.g. ajmaline), which are involved in heterocyst formation and nitrogen exchange between cyanobacteria and their host plant (Gorelova & Kleimenov, 2003). Further investigations are needed to unambiguously clarify the biosynthetic origin of methanol emission from T. azollae.

In conclusion, a coordinated photoprotection system consisting of isoprene emission and phenolic compounds synthesis is employed by Azolla plants to cope with a wide range of light intensities. Indeed, accumulation of red‐pigments, mainly 3‐deoxyanthocyanidins and phlobaphenes, and other phenolic compounds more readily triggered by exposure from low to very high light might prevent photooxidative damage in the short‐term. At the same time, the antioxidant activity exerted by an enhanced isoprene capacity and carotenoids production might contribute to optimize the photosynthetic performances of Azolla plants to increasingly growing light conditions in the longer term.

AUTHOR CONTRIBUTION

Federico Brilli designed the research, run measurements, analyzed the data and wrote the article with contributions of all the authors; Stefania Pasqualini, Alma Costarelli, Sara Cannavò, Francesco Paolocci, Graziella Chini Zittelli, Gianmarco Mugnai, and K.G. Srikanta Dani analyzed the data and performed research; K.G. Srikanta Dani, Rita Baraldi, and Francesco Loreto supervised and contributed to the revision of the manuscript.

Appendix S1 Supporting Information.

Click here for additional data file.