The membrane is gas permeable and serves as a physical barrier between the liquid phase and the ion source of the mass spectrometer connected by a vacuum line. In the setup described in Figure 1 the membrane is held at the bottom of the reaction vessel supported by a stainless-steel grid. Note that different membrane inlet setups like microprobes can be used (Kotiaho and Lauritsen, 2002; Konermann et al., 2008). The gas leak through the membrane occurs through a three-step process called permeation: adsorption at the membrane surface, diffusion through the membrane material, and desorption on the other side of the membrane. In this process, diffusion is the limiting factor (Konermann et al., 2008) and is a temperature-dependent process. Thus, membrane permeability is highly influenced by temperature ( Figure 2 ), and the reaction vessel must be precisely thermo-regulated ( Figure 1 ). The gas leak is crucial since it defines the sensitivity of the method, especially it should remain small compared to the measured reaction rates to limit noise. Gas leakage depends on the membrane properties and is linearly related to the gas concentration inside the vessel ( Methods S1 ). Various porous plastic films can be used as a membrane, although polytetrafluoroethylene (PTFE, also known as Teflon) and silicon are the most commonly used (Beckmann et al., 2009; Shevela et al., 2018). The permeability to gases depends on the nature of the membrane. Silicon membranes are more permeable to gases than PTFE membranes (Lloyd et al., 1983) (PTFE membranes can be found on www.hansatech-instruments.com/ or www.ysi.com). The gas leak depends on both membrane surface and thickness, which must be chosen depending on the required sensitivity and on the duration of the experiments to be carried out. The ion source is overly sensitive to water that interferes with ionization and therefore quantification. In this respect, a PTFE membrane has the advantage over silicone of being seven times less permeable to water relatively to N₂ (Jensen and Cox, 1988), resulting in less noise. A magnetic stirrer is used to maintain algae in suspension and homogenize the gas content of the liquid medium (Lundsgaard et al., 1978). In a case where the stirring occurs on the membrane, the stirring speed must be adjusted to ensure a proper homogenization without compromising membrane integrity. Likewise, a smooth PTFE coated stirring bar can be used to achieve optimal stirring while limiting membrane wear.

The vacuum line connecting the space below the membrane to the ion source of the mass spectrometer is, together with the membrane, a key element to consider when optimizing the response time of the experimental setup. Because of high vacuum (10⁻⁵–10⁻⁸ mbar) in the tubing, gases do not flow like a fluid but perform a random walk inside the tube towards the ion source of the mass spectrometer where gases are ionized. During their random walk, gases can adsorb and desorb at the surface of the tubing, thus slowing down the gas flow. The tubing length and cross-section should be optimized to increase the time response without compromising output signal of the mass spectrometer. Limiting length of the tubing decreases the response time of the system both because of limited inner tubing volume and decreased surface adsorption effects of the tubing. On the other hand, while small tubing cross-section favors the time response by limiting the tubing volume, this effect is counter-balanced by an increased probability of molecule absorption on the tubing surface, a too small cross-section decreasing the signal/noise ratio of the setup. We found an inner diameter of ¼ inch to be a good compromise for the type of experiments shown in this paper although smaller tubing down to 1 16 inch can be suitably used (Konermann et al., 2008).

To limit water vapor entering the mass spectrometer ion source, the vacuum line tubing passes through a cold trap where the water vapor is condensed. Since the cold trap can also condense other gases, the choice of the trap temperature depends on the gas species to be analyzed. For example, cooling the vacuum line with liquid nitrogen (77 K) allows efficiently trapping water but also CO₂ ( Figure 3 ). The condensation points of other main gases [N₂, O₂, and argon (Ar)] are much lower than that of water and are usually hard to selectively trap ( Figure 3 ). A temperature of 200 K allows selective trapping of water and can be obtained using a mix of ethanol and dry ice in the absence of a cooling unit (Bailleul et al., 2017). The cold trap should be situated as close as possible to the vacuum pump of the mass spectrometer in order to keep its internal pressure as low as possible and limit unintended trapping. For further information on optimization of the membrane inlet system, readers are referred to (Konermann et al., 2008; Beckmann et al., 2009; Shevela et al., 2018).

Mass spectrometry is a technique that allows separation and detection of charged molecules in a gas phase depending on their mass over charge (m/z) ratio. During ionization, most reactions produce a single positively charged ion. But most often, the molecule is broken down by ionization, and seldom, double ionization can occur; these effects give rise to two or more fragments. The fragmentation pattern is a signature of the molecule and depends on the ionization method used. In the case of small molecules like gases, the main ion (molecular ion) results from the loss of an electron, other fragments representing a minor part in the fragmentation pattern ( Table 1 ). The mass spectrometer, by measuring the signal intensity of detected ions for m/z of interest, enables the determination of gas amounts ( Methods S1 ). Any mass spectrometer can be used in a MIMS setup with various ion sources (e.g. pulsed ionization, glow discharged ionization) and analyzers (e.g. Time Of Flight, Ion Trap) (Johnson et al., 2000). However, magnetic sector or quadrupole mass spectrometers are the most commonly used for gas analysis in a MIMS setup, although quadrupole has the advantage of portability and low price (Ferrón et al., 2016; Bailleul et al., 2017; Chatton et al., 2017).

Because the mass spectrometer is consuming gases, measured variations of gas concentrations must be corrected from the mass spectrometer consumption to determine actual gas exchange rates between the biological sample and the extracellular medium ( Methods S2 ). Although this effect has been well described (Berlier et al., 1985; Kotiaho and Lauritsen, 2002; Konermann et al., 2008) its correction has been overlooked in the recent literature. After calculating gas concentrations inside the reaction vessel, gas exchange rates are calculated by correcting from the mass spectrometer consumption. Fluctuations of physical properties of the setup can result in variations of the flux of gas to the mass spectrometer and be an important source of noise. An additional normalization of gas exchange rates to a non-reactive gas (like N₂ or Ar) can advantageously correct such signal shifts or noise. These artifacts are therefore highly limited although biggest ones remain ( Figure S2 ). Note that prior to any experiment, one must ensure the absence of gas leakage in the system (tubing, connections, inlets), which can be done by flushing helium outside the setup and following the m/z = 4 with the mass spectrometer. Details of the calculations have been described in Methods S1 and S2 , and we have developed easy-to-run software that allows real time calculation and visualization of MIMS data (Downloadable at: https://github.com/francoisBurlacot/MIMS_Analysis).

By using ¹⁸O-enriched O₂ in illuminated microalgal suspension, Hoch and Kok (Hoch and Kok, 1963) could show that O₂ can be both produced and consumed during photosynthesis. While O₂ is produced by PSII from water splitting, O₂ is simultaneously consumed by different cellular processes ( Figures 4 and 5A ). Practically, the use of highly enriched O₂ (usually around 99% ¹⁸O) allows neglecting ¹⁸O¹⁶O species, thus measurements of ¹⁶O¹⁶O (m/z = 32) and ¹⁸O¹⁸O (m/z = 36) are used to determine rates of gross O₂ evolution (O₂ Evolution) and O₂ uptake (O₂ Uptake), net O₂ production rate (Net O₂ ) being the end result of O₂ Evolution and Uptake. Considering that water splitting only produces ¹⁶O₂ (the natural abundance of ¹⁶O being 99.8%, O₂ is produced from H₂O at 99.6% as ¹⁶O₂) and neglecting isotopic discrimination between ¹⁸O₂ and ¹⁶O₂ by uptake mechanisms, the following equations modified from (Radmer and Kok, 1976; Peltier and Thibault, 1985b) can be used:

where v O 2 18 and v O 2 16 are the gas exchange rates of ¹⁸O₂ and ¹⁶O₂ respectively; C O 2 18 ( t ) and C O 2 16 ( t ) represent the gas concentrations of ¹⁸O₂ and ¹⁶O₂ respectively (see Methods S1 and S2 ). Typical patterns of ¹⁸O₂ and ¹⁶O₂ concentration, gross O₂ evolution, O₂ uptake, and O₂ net production rates as well as cumulated O₂ exchanges measured in C. reinhardtii cells during a dark to light transient are shown in Figures 5B–D .

Measuring O₂ exchange by MIMS in microalgae and cyanobacteria allowed dissecting molecular players involved in the O₂ uptake process, initially by using various inhibitors or characterizing the dependency on O₂ or CO₂ concentrations (Badger et al., 2000). (Radmer and Kok, 1976) early proposed that a highly efficient O₂ photoreduction process was present in cyanobacteria and microalgae. The persistence of mitochondrial respiration in the light was shown to contribute (Peltier and Thibault, 1985b), and the existence of a CO₂-dependent O₂ uptake component was evidenced (Badger, 1985; Sültemeyer et al., 1987). With the development of genetic approaches, the nature and contribution of the different players were better characterized. For example, the contribution of mitochondrial respiration to O₂ uptake is enhanced in the absence of the proton gradient regulation like 1 (PGRL1)-mediated cyclic electron flow (CEF), thus showing the functional complementarity between these pathways in the production of intra-cellular ATP (Dang et al., 2014). Redox communication between chloroplast and peroxisome (Kong et al., 2018) or with mitochondria in diatoms (Bailleul et al., 2015) was also evidenced from O₂ exchange measurements. The involvement of Flavodiiron proteins (Flvs) in light-dependent O₂ uptake was further established in cyanobacteria (Helman et al., 2003; Allahverdiyeva et al., 2013), and more recently in microalgae (Chaux et al., 2017). MIMS is nowadays widely used to understand the fate of the photosynthetic electron flow in various environmental conditions and mutant of cyanobacteria (Ermakova et al., 2016; Boatman et al., 2018; Luimstra et al., 2019), microalgae (Fisher and Halsey, 2016), or coral reef symbiosis (Einbinder et al., 2016).

If laboratory studies performed on model species allowed recognizing O₂ photoreduction as a major alternative photosynthetic electron sink crucial for cell acclimation to various environmental conditions (Curien et al., 2016), the relative contribution of different O₂ uptake mechanisms in natural environments remains largely unexplored (Bailleul et al., 2017). Despite its performance in measuring gross photosynthesis, MIMS has long remained a cumbersome technique, not suitable for field experiments. On the other hand, chlorophyll fluorescence measurements using pulse-amplitude modulated (PAM) fluorimeter have been widely used for estimating PSII yield in natural environments such as in the ocean (Falkowski and Kolber, 1995). Chlorophyll fluorescence however, faces some limitations when used to determine gross O₂ production since the estimation of electron transport rates requires the measurement of cell absorbance (Genty et al., 1989; Godaux et al., 2015), which is difficult to realize in outdoor conditions. Recently, the miniaturization of mass spectrometers allowed in situ measurements of O₂ exchange in phytoplankton from the north Pacific (Ferrón et al., 2016) or in planktonic blooms from the north Atlantic (Bailleul et al., 2017), starting thus a new era for expanding research obtained in the laboratory on model species to species in their natural environments.

Indeed, fluorescence measurements have often been used together with MIMS CO₂ and O₂ exchange measurements to study the link between CO₂ fixation and non-photochemical quenching of chlorophyll fluorescence (Sültemeyer et al., 1989; Fratamico et al., 2016; Ware et al., 2020). Furthermore, in conditions where the use of ¹⁸O₂ is not possible (e.g. anaerobiosis), coupling PSII quantum yield with gas exchanges has recently allowed determining gross O₂ production and inferring the existence of an O₂ uptake process (Burlacot et al., 2018).

In microalgae and cyanobacteria, hydrogenases (H₂ases) catalyze the reversible formation of hydrogen (H₂) by direct reduction of protons (H⁺). In microalgae, the electron donor to the [Fe–Fe] H₂ase is ferredoxin (Florin et al., 2001), which can be reduced by the photosynthetic electron transport chain or by fermentative pathways (Catalanotti et al., 2013). Experimentally, H₂ production can be measured by different techniques, including modified O₂ electrodes (Wang, 1980; Godaux et al., 2015), gas chromatography (Hunt and Smith, 1961; Nagy et al., 2018), nuclear magnetic resonance spectroscopy (NMR) (Xu et al., 2016; Manz et al., 2017), or MIMS. Among these techniques, MIMS and modified O₂ electrodes allow in vivo quantitative measurement of hydrogen in a time-resolved manner. Under natural conditions, hydrogen photoproduction by microalgae is a transient phenomenon, generally considered as a safety valve avoiding over-reduction of PSI electron acceptors under anaerobiosis (Ghysels et al., 2013). Hydrogen photoproduction is limited by the O₂ sensitivity of H₂ase, O₂ being produced by PSII during illumination (Erbes et al., 1979). When using experimental conditions maintaining anaerobic conditions (thus limiting the H₂ase inhibition) like sulfur deprivation (Melis et al., 2000), low illumination (Liran et al., 2016) or O₂ quenchers like glucose and glucose oxidase/catalase (Godaux et al., 2015), a limitation of the supply of electrons to the H₂ase can be evidenced. The use of MIMS and of various C. reinhardtii mutants allowed identifying biological bottlenecks limiting the supply of electrons to the H₂ase (Tóth and Yacoby, 2019; Burlacot et al., 2020a). Lately, MIMS was used in the development of a very promising H₂ photoproducing protocol using flashing light as the light source in photoautotrophic self-anaerobic conditions (Kosourov et al., 2018).

MIMS has early been used for in vitro and in vivo measurements of H₂ase activity (Jouanneau et al., 1980; Vignais et al., 1982; Berlier et al., 1985). In the presence of H₂, H₂ase spontaneously splits H₂ (Hoberman and Rittenberg, 1943; Rittenberg and Krasna, 1955), forming one proton with the reversible reaction:

where Hyd is the binding site of H₂ase. When supplying deuterium (D₂), HD is formed during the back reaction (6) in the presence of protons in the reaction mixture:

This reaction directly depends on the turnover rate of H₂ases (i.e. H₂ase activity) (Vignais, 2005). In the absence of H₂ production or uptake, following kinetics of D₂, HD and H₂ with MIMS allows measuring the H₂ase activity (H₂ase activity) by the H/D exchange rate (V_exch):

with

where C D 2 ( t ) , C H 2 ( t ) and C H D ( t ) are the concentrations of D₂, H₂ and HD respectively and v H 2 ( t ) , v H D ( t ) being the gas exchange rates of H₂ and HD, respectively (see Methods S1 and S2 ) (Cournac et al., 2004). A few microalgal species such as C. reinhardtii harbor H₂ases (Burlacot and Peltier, 2018) under anaerobic conditions. Measuring the H/D exchange allowed monitoring of H₂ase induction or inhibition in vivo (Vignais et al., 2002; Tolleter et al., 2011; Burlacot et al., 2018). Typical patterns of D₂, H₂ and HD exchange measured upon injection of D₂ before and after induction of H₂ase in C. reinhardtii are shown on Figure 6 . Note that in conditions where H₂ase produces H₂ during illumination the H₂ase activity, needs to be corrected from the increase in total hydrogen species (H₂, HD, D₂) (Cournac et al., 2004). Although gas chromatography or NMR has also the potential to differentiate D₂, HD and H₂ (Hunt and Smith, 1961; Xu et al., 2016), MIMS allows performing such measurements in vitro and in vivo in a time resolved manner. If H/D exchange measurements using a MIMS allow determining the catalytic constant of H₂ase, it can also be used to determine the resistance of gas diffusion between the active H₂ase site and the reaction medium in vitro (Leroux et al., 2008). In vitro H/D exchange measurements have been used to study enzymatic properties of native H₂ase (Abou Hamdan et al., 2012; Gauquelin et al., 2018), including O₂-tolerant H₂ases (Liebgott et al., 2011), as well as H₂ases modified by site-directed mutagenesis in order to limit O₂ diffusion to the active site (Cano et al., 2014). H₂ photoproduction by microorganisms has recently regained huge interest for biofuel production due to recent improvements in strains and experimental protocols (Tóth and Yacoby, 2019). The use of MIMS should help in evaluating the upcoming combination of newly developed H₂ production protocols (Kosourov et al., 2018; Nagy et al., 2018) and previously characterized mutants photo-producing more H₂ (Tolleter et al., 2011; Eilenberg et al., 2016; Burlacot et al., 2018; Ben-Zvi et al., 2019). This has recently started using flv mutants (Jokel et al., 2019) and is promising for future bio H₂ production developments.

On Earth, 6% of the radiative forcing is due to N₂O (IPCC, 2013), whose greenhouse effect is 300 times that of CO₂. N₂O is produced from the reduction of nitric oxide (NO) by bacteria (Goreau et al., 1980), fungi (Maeda et al., 2015), and microalgae (Guieysse et al., 2013). The measurement of N₂O in a time resolved manner can help understand the dynamics of N₂O formation in the environment or in isolated organisms. In green microalgae, MIMS has recently been used to dissect the molecular mechanisms involved in the conversion of NO into N₂O ( Figure 7 ) (Burlacot et al., 2020b). In practical terms, the detection of N₂O requires specific calculations because its mass spectrum overlaps that of CO₂ (see Methods S1 ). However, due to a relatively high detection limit for N₂O (around 18 µM) (Chatton et al., 2017), MIMS is not suitable for detecting low N₂O amounts, such as in water in equilibrium with ambient air (4.3 µM) (IPCC, 2013).

NO is an important intracellular signaling molecule in algae like in most living organisms. While NO is crucial for growth of C. reinhardtii under anaerobiosis and for its acclimation to nitrogen, sulfur or phosphate deprivation (Hemschemeier et al., 2013; Wang and Spalding, 2014; De Mia et al., 2019; Filina et al., 2019), mechanisms of NO production remain blurry. The use of MIMS in algae fed with nitrates or nitrites, allowed to evidence the role of NO reduction mechanisms in the regulation of NO homeostasis (Burlacot et al., 2020b). Following NO production can also be based on imaging fluorescent chemical probes specifically reacting with NO (Li and Wan, 2015) or on the use of specific electrodes (Csonka et al., 2015). However, MIMS by supplying quantitative, time resolved, and stable measurements of NO (Bethke et al., 2004; Conrath et al., 2004), with further simultaneous measurement of NO reduction products such as N₂O, should help deciphering mechanisms participating in NO homeostasis.

When grown under low CO₂ concentration, many microalgae or cyanobacteria induce an active import of inorganic carbon (C_i = CO₂ and HCO₃ ⁻) ( Figure 8A ) from the extracellular medium to the active site of CO₂ fixation (Badger et al., 1980; Badger and Andrews, 1982; Sültemeyer et al, 1991; Giordano et al., 2005; Reinfelder, 2011). This mechanism, called Carbon Concentrating Mechanism (CCM), principally operates by concentrating HCO₃ ⁻ inside cells (Price et al., 2007). In microalgae, HCO₃ ⁻ is converted into CO₂ inside the pyrenoid, at the vicinity of the carbon-fixing enzyme (RuBisCO) thus increasing the local CO₂ concentration ( Figure 8A ) (Mackinder, 2018). The CCM ensures high C_i fixation rates by photosynthesis under low C_i concentration (Badger et al., 1980). When active, the CCM results in an increased apparent affinity of photosynthesis for C_i, which can be assessed by measuring O₂ production rates at various C_i concentrations, either using an O₂ electrode (Badger et al., 1980) or a MIMS (Sültemeyer et al., 1993). MIMS has the advantage of simultaneously measuring O₂ and CO₂. Since the CO₂ decreases during the time course of the experiment due to the activity of photosynthesis, it is possible to determine in one single experiment net O₂ or CO₂ exchange rates at different CO₂ concentrations ( v O 2 ( C C O 2 ( t ) ) or v O 2 ( C C O 2 ( t ) ) ) during CO₂ fixation by photosynthesis (Douchi et al., 2019). Figures 8B, C show typical experiments in which CO₂ uptake and O₂ production rates have been determined as a function of the CO₂ concentration in air-grown and CO₂-grown cells of C. reinhardtii. Note that in such experiments the cell concentration must be kept low enough to ensure that photosynthetic gas exchange kinetics are sufficiently slow as compared to the response time of the MIMS setup. Using this technique ( Figure 8 ), the apparent affinity for C_i of photosynthetic O₂ production of cells with an active CCM is about 10 times higher than in cells with no active CCM as previously reported (Badger et al., 1980; Sültemeyer et al., 1998).

During the induction of CCM, carbonic anhydrases (CAs) are also induced and catalyze the reversible hydration of CO₂ by water which is summarized in reaction (8):

In microalgae, different CA isoforms are present in the different cellular compartments, thus limiting the disequilibrium between transported and consumed C_i species ( Figure 4 ) (Sültemeyer et al., 1990; Mackinder et al., 2017). Inside the pyrenoid, CA rapidly converts HCO₃ ⁻ into CO₂, the substrate of RuBisCO, thus ensuring a high CO₂ concentration at the catalytic site of the enzyme (Mackinder, 2018).

When their CCM is active, cyanobacteria and microalgae can take both CO₂ and HCO₃ ⁻ in the medium (Sültemeyer et al., 1991; Badger et al., 1994). A way to assess the net flux of both species is to use the disequilibrium between inorganic species (see equation 8) induced by the preferential uptake of one species. Note that this approach is not possible in the presence of extracellular CA, which has limited its usage in microalgae (Badger et al., 1994; Palmqvist et al., 1994). By comparing net O₂ production (proportional to the overall net C_i uptake) to the net CO₂ uptake and knowing the uncatalyzed rate of CO₂ and HCO₃ ⁻ interconversion (8), it is possible to calculate the net HCO₃ ⁻ uptake (Badger et al., 1994). Using the disequilibrium method with MIMS has shown that HCO₃ ⁻ was preferentially taken during CCM in cyanobacteria (Sültemeyer et al., 1998).

In the presence of doubly ¹⁸O-labeled CO₂, the CA activity which catalyzes the exchange of oxygen isotopes between CO₂ and H₂O results in a progressive dilution of ¹⁸O from CO₂ (Gerster, 1971) ( Figure 9A ). To limit the background level of m/z = 44 (¹²CO₂) due to naturally present CO₂ in the algal suspension, the assay can be done using ¹⁸O-enriched ¹³CO₂ (Radmer and Kok, 1976), following the m/z = 45 (¹³C¹⁶O₂), 47 (¹³C¹⁸O¹⁶O) and 49 (¹³C¹⁸O₂). After injection of ¹³C¹⁸O₂ (supplied as H¹³C¹⁸O₃ ⁻ in a buffered reaction medium) a typical pattern of the progressive unlabeling of ¹³CO₂ in C. reinhardtii cells is shown ( Figures 9B–E ). The isotopic content of ¹⁸O in ¹³CO₂ during the progressive unlabeling is given by:

where C C 13 O 2 18 ( t ) , C C 13 O 18 O 16 ( t ) and C C 13 O 2 16 ( t ) are the concentrations of ¹³C¹⁸O₂, ¹³C¹⁸O¹⁶O and ¹³C¹⁶O₂ respectively ( Methods 1 ). On purified CA, the isotopic enrichment decays exponentially:

where θ is the rate constant of the exchange of ¹⁸O with water (Gerster, 1971; Silverman, 1982). Note that in some cases like a concomitant use of ¹⁸O-labeled O₂, the presence of ¹⁸O and C¹⁶O₂ in the ion source of the mass spectrometer can spontaneously generate C¹⁸O¹⁶O which needs to be corrected (Cournac et al., 1993). In vivo, a CA isotope exchange assay measures the global contribution of all CAs present in the biological sample with an additional effect due to diffusion/transport of C_i through the membranes (Sültemeyer and Rinast, 1996; Tolleter et al., 2017). When a periplasmic CA is present, as it is the case in C. reinhardtii cells with an active CCM, its activity dominates the exchange kinetics (Sültemeyer and Rinast, 1996). MIMS is so far the most reliable method for CA activity measurements in biological samples and is particularly suitable for in vivo measurements on algae (Dang et al., 2014; Benlloch et al., 2015; Tolleter et al., 2017), cyanobacteria (Whitehead et al., 2014), corals (Tansik et al., 2015), and plants (Peltier et al., 1995; Clausen et al., 2005; Burén et al., 2011; Benlloch et al., 2015). In vitro, the use MIMS and H¹³CO₃ ⁻ has also allowed unraveling the existence of a light-induced CO₂ production by the PSII (Koroidov et al., 2014; Shevela et al., 2020).