Water oxidation by photosystem II is the primary source of electrons for sustained H₂ photoproduction in nutrient-replete green algae

Significance

Photosynthetic H₂ production in the green alga Chlamydomonas reinhardtii is catalyzed by O₂-sensitive [FeFe]-hydrogenases, which accept electrons from photosynthetically reduced ferredoxin and reduce protons to H₂. Since the process occurs downstream of photosystem I, the contribution of photosystem II (PSII) in H₂ photoproduction has long been a subject of debate. Indeed, water oxidation by PSII results in O₂ accumulation in chloroplasts, which inhibits H₂ evolution. Therefore, clear evidence for direct water biophotolysis resulting in simultaneous H₂ and O₂ releases in algae has never been presented. This paper demonstrates that sustained H₂ photoproduction in C. reinhardtii is directly linked to PSII-dependent water oxidation and brings insights into regulation of PSII activity and H₂ production by CO₂/HCO₃^– under microoxic conditions.

Abstract

The unicellular green alga Chlamydomonas reinhardtii is capable of photosynthetic H₂ production. H₂ evolution occurs under anaerobic conditions and is difficult to sustain due to 1) competition between [FeFe]-hydrogenase (H₂ase), the key enzyme responsible for H₂ metabolism in algae, and the Calvin–Benson–Bassham (CBB) cycle for photosynthetic reductants and 2) inactivation of H₂ase by O₂ coevolved in photosynthesis. Recently, we achieved sustainable H₂ photoproduction by shifting algae from continuous illumination to a train of short (1 s) light pulses, interrupted by longer (9 s) dark periods. This illumination regime prevents activation of the CBB cycle and redirects photosynthetic electrons to H₂ase. Employing membrane-inlet mass spectrometry and ${\mathrm{H}}_2^{18}\mathrm{O}$, we now present clear evidence that efficient H₂ photoproduction in pulse-illuminated algae depends primarily on direct water biophotolysis, where water oxidation at the donor side of photosystem II (PSII) provides electrons for the reduction of protons by H₂ase downstream of photosystem I. This occurs exclusively in the absence of CO₂ fixation, while with the activation of the CBB cycle by longer (8 s) light pulses the H₂ photoproduction ceases and instead a slow overall H₂ uptake is observed. We also demonstrate that the loss of PSII activity in DCMU-treated algae or in PSII-deficient mutant cells can be partly compensated for by the indirect (PSII-independent) H₂ photoproduction pathway, but only for a short (<1 h) period. Thus, PSII activity is indispensable for a sustained process, where it is responsible for more than 92% of the final H₂ yield.

Many species of green algae have [FeFe]-hydrogenases (H₂ases) (1) that catalyze the reversible reduction of protons to molecular hydrogen:

$$2{\mathrm{H}}^{+}+2{\mathrm{e}}^{-}\rightleftarrows {\mathrm{H}}_2.$$ [1]

Since [FeFe]-H₂ases are extremely O₂-sensitive (2), reaction 1 typically proceeds under anoxic conditions. With respect to H₂ metabolism, Chlamydomonas reinhardtii is the most studied alga. This alga possesses two [FeFe]-H₂ases in the chloroplast, HYDA1 and HYDA2 (3, 4). In the light, they accept electrons from photosynthetically reduced ferredoxin (FDX1) (5), while in the dark electrons come from the activity of pyruvate ferredoxin oxidoreductase (PFR1) (6). PFR1 catalyzes the oxidation of pyruvate to acetyl-CoA, and its activity is linked to H₂ase via FDX1 (7). Since [FeFe]-H₂ases interact with the photosynthetic electron transport chain at the level of ferredoxin, they may accept electrons originating both from water oxidation via the photosystem II (PSII)-dependent pathway (“direct water biophotolysis”) and from the degradation of organic substrates via a PSII-independent mechanism (“indirect water biophotolysis” or “indirect pathway”) (8). In the latter case, the reductants are supplied to the plastoquinone (PQ) pool by the type II NADPH dehydrogenase (NDA2), thus bypassing PSII (9, 10).

The release of H₂ leads to a loss of metabolic energy. In healthy, actively growing C. reinhardtii cultures, H₂ production is therefore only a temporal phenomenon observed during dark anoxia and upon subsequent onset of illumination (11). In contrast to dark fermentation, H₂ photoproduction is a very efficient process that proceeds for only a short period of time (from a few seconds to a few minutes). Two theories have been developed to explain the short duration. The first is based on the oxygen sensitivity of H₂ases (12, 13). In the light, algae accumulate O₂ that is produced by water oxidation at PSII (14). As a result, H₂ photoproduction may cease over time (14, 15), and the duration of this process is reported to shorten with increased light intensity (16). Because of the negative correlation between the rates of H₂ photoproduction and O₂ evolution, the inhibition of H₂ases by O₂ is frequently quoted as the primary reason for the rapid loss in H₂ photoproduction after the onset of illumination (17).

Alternatively, the loss in the H₂ photoproduction efficiency during illumination could be explained by the light-induced induction of competitive pathways, which may drain reducing equivalents away from the [FeFe]-H₂ase enzyme (18, 19). Candidates for this role are the Mehler-like reaction driven by flavodiiron proteins (FDPs) (15, 20, 21) and the Calvin–Benson–Bassham (CBB) cycle (22). Compelling evidence for the competition between these two pathways and H₂ production has been accumulated in recent studies (23–25). As CO₂ fixation provides the strongest sink for photosynthetic reductants, it should play a major role in the cessation of H₂ photoproduction in algae when the CBB cycle is active (19, 22).

For preventing competition between the [FeFe]-H₂ases and the CBB cycle, we recently devised a pulse-illumination protocol that allows H₂ production in nutrient-replete algal cultures for up to 3 d (23). To achieve this, we specifically selected the duration of light pulses in the light/dark sequence to avoid activation of the CBB cycle, thus allowing for the redirection of photosynthetic electrons toward the [FeFe]-H₂ases. Typically, a train of 1- to 6-s light pulses interrupted by 9-s dark periods is sufficient for sustained H₂ photoproduction in C. reinhardtii cultures (23, 25). Our protocol thus differs from earlier pulse-illumination approaches that aimed at preventing the accumulation of O₂ in the cultures (26).

While we could demonstrate competition of [FeFe]-H₂ase with FDPs (25), the origin of reductants for H₂ photoproduction in the pulse-illuminated algae remained unclear. The relatively high efficiency of the process suggests the involvement of water oxidation by PSII, and consequently the simultaneous production of H₂ and O₂. Although widely proposed in the current literature (8, 24), the presence of the direct water biophotolysis in H₂-producing green algae has not yet been proven by direct experimental data.

In the present study, we provide clear evidence for the presence of PSII-dependent oxidation of ¹⁸O-labeled water $\left({\mathrm{H}}_2^{18}\mathrm{O}\right)$ with concomitant evolution of ¹⁶O₂ and ^16,18O₂ during H₂ photoproduction in the pulse-illuminated green alga C. reinhardtii under anoxic conditions. O₂ evolution is balanced by light-dependent and light-independent respiration that sustains the anoxic condition. We also demonstrate that the loss of PSII activity in algae can be partly compensated by the PSII-independent H₂ photoproduction pathway. Nevertheless, the activity of PSII is indispensable for the sustained process, where it contributes to more than 92% of the final H₂ yield.

Results

Net H₂ Photoproduction, Water Oxidation, and O₂ Exchange.

Employing membrane-inlet mass spectrometry (MIMS), we confirmed the induction of sustained H₂ photoproduction in anoxic C. reinhardtii cultures by a train of 1-s light pulses interrupted by 9-s dark periods (Fig. 1A). The signal exhibited a typical sawtooth wave with the H₂ production transients during pulse illumination and a strong H₂ uptake on the shift to darkness, which is in line with previous data (23). The H₂ uptake is the result of H₂ consumption by the MIMS setup and the reversible action of the H₂ase (as demonstrated in SI Appendix, Fig. S1). The response of anoxic algae to the train of light pulses was strain-specific. While in the cell wall-deficient mutant the first H₂ peak appeared already on the first 1-s flash (Fig. 1C, CC-4533), the wild-type CC-124 strain required some time before demonstrating the pronounced H₂ photoproduction yield (Fig. 1A).

Fig. 1.

H₂ photoproduction (A) and O₂ exchange (B) in pulse-illuminated algae. Anoxia was achieved within 3 min after placement of aerobic algae in the O₂-consuming MIMS chamber under complete darkness. The total (a + b) chlorophyll (Chl) content in the samples was around 9 mg⋅L⁻¹. The photochemical reactions were initiated by a train of 1-s light pulses interrupted by 9-s dark periods. (C) The behavior of the m/z 2 and m/z 34 signals at better time resolution for the cell wall-deficient strain (CC-4533) in the beginning of pulse illumination and for the wild-type strain (CC-124) in the middle of H₂ photoproduction. One representative result out of three repeats is presented. Additional experimental data can be found in SI Appendix, Fig. S9.

The application of ¹⁸O-labeled water $\left({\mathrm{H}}_2^{18}\mathrm{O}\right)$ to the algae allowed simultaneous monitoring of PSII activity via both nonlabeled and ¹⁸O-labeled O₂ isotopologues (27). Thus, as shown in Fig. 1B, the injection of ${\mathrm{H}}_2^{18}\mathrm{O}$ into the MIMS cell filled with the algal suspension led to the detection of O₂ evolution at m/z 32 (for ambient ¹⁶O₂) and m/z 34 (for singly labeled ^16,18O₂) signals as a result of the water-oxidizing activity of PSII centers during pulse illumination:

$$2{\mathrm{H}}_2^{16}\mathrm{O}\to {\mathrm{O}}_2^{16}+4{\mathrm{H}}^{+},$$ [2]

$${\mathrm{H}}_2^{16}\mathrm{O}+{\mathrm{H}}_2^{18}\mathrm{O}\to {\mathrm{O}}_2^{16,18}+4{\mathrm{H}}^{+}.$$ [3]

The m/z 34 signal showed O₂ evolution at the first 1-s light pulse, and algal cultures reached maximum O₂ production yields after 10 to 60 pulses (Fig. 1B). Then, the signal slowly declined but kept the sawtooth shape until the end of the pulse-illumination period. The decline in the net O₂ evolution coincided well with the simultaneous increase in the H₂ photoproduction yield (Fig. 1A). Surprisingly, ¹⁶O₂ production (m/z 32) of midlog-phase C. reinhardtii cultures was significantly delayed by strong light-dependent uptake at the beginning of pulse illumination (Fig. 1B). Since ambient ¹⁶O₂ dominates in the MIMS chamber and the level of ^16,18O₂ is negligible in the beginning of pulse illumination, the pronounced uptake of ¹⁶O₂ in the first few cycles of pulse illumination indicates the domination of light-dependent respiration over water oxidation during this period. On the contrary, the mature (close to the stationary phase) algae did not show any signs of the light-dependent ¹⁶O₂ uptake (SI Appendix, Fig. S2). In the latter case, the m/z 32 signal simply repeated the shape of the m/z 34 curve. Thus, light-dependent O₂ uptake is not caused by the self-shading effect and activation of respiration in dense cultures. This phenomenon is most likely linked to O₂ photoreduction by FDPs that operate during the dark-to-light (or low-light to high-light) transients under oxic and microoxic conditions (25, 28).

The rise of the H₂ (m/z 2) signal occurred almost immediately upon the firing of each light pulse. This was followed by a rise in the O₂ (m/z 34) signal with ∼3-s delay (Fig. 1C). As a result, the H₂ production wave showed maxima at almost the minimums of the m/z 34 signal. Since the diffusivity of H₂ in the polydimethylsiloxane membrane used in the MIMS cell (applied to separate the liquid phase of the sample from the high vacuum of the mass spectrometer) is approximately four times faster than for O₂ (29)_, this behavior of H₂ and O₂ signals is not a physiological response but rather a reflection of the detection method. Therefore, we suggest simultaneous photoproduction of these two gases in C. reinhardtii cells.

The Role of PSII-Dependent and PSII-Independent Pathways.

The addition of 30 μM of DCMU [3-(3,4-dichlorophenyl)-1,1-dimethylurea], a specific inhibitor that binds to the Q_B-pocket of PSII and blocks electron transport from PSII to the PQ pool, did not affect H₂ photoproduction much during the first 20 min of pulse illumination (Fig. 2A). The m/z 2 signal of the DCMU-treated algae (CC-124 + DCMU) was only slightly reduced as compared to the control (CC-124). This response of pulse-illuminated algae to DCMU was confirmed using an H₂ electrode (SI Appendix, Fig. S3). In contrast, the presence of DCMU significantly reduced the ^16,18O₂-evolving activity in algae at the start of pulse illumination (Fig. 2B), but inhibition was not complete. Thus, as seen in Fig. 2B, Inset a residual water-oxidizing activity (<1.5% of the control) was observed in the DCMU-treated samples. At the same time, DCMU-treated algae demonstrated no light-induced evolution of nonlabeled ¹⁶O₂ (m/z 32) but a strong light-dependent ¹⁶O₂ uptake (Fig. 2C).

Fig. 2.

The effect of DCMU and the psbA deletion (FuD7) on H₂ (A), ^16,18O₂ (B), and ¹⁶O₂ (C) yields in pulse-illuminated algae. The experimental conditions were the same as in Fig. 1, except that the total Chl content was increased to ∼20 mg⋅L⁻¹ (to distinguish the difference between O₂ production and O₂ consumption in the control and DCMU-treated samples). DCMU was introduced to aerobic algae at a final concentration of 30 μM. Curves are the mean of two to four independent experiments. The H₂ photoproduction of the FuD7 mutant is a result of one measurement with MIMS but was repeated nine times with a H₂ sensor, producing similar results (SI Appendix, Fig. S3). Additional experimental data can also be found in SI Appendix, Fig. S10.

Since the m/z 34 signal represents the net ^16,18O₂ exchange with a significant expected share of light-induced O₂ uptake (as seen from the m/z 32 signal in Fig. 2C), the contribution of the residual PSII activity in the H₂ photoproduction yield of the DCMU-treated samples remained unclear. Therefore, in the next experiment, a train of light pulses was applied to the PSII-deficient (FuD7) mutant. Due the absence of fully assembled PSII centers in the thylakoid membrane (30), this strain completely lacks water-oxidizing activity (SI Appendix, Fig. S4A). Nevertheless, FuD7 still produced H₂, albeit at the decreased rate (Fig. 2A and SI Appendix, Fig. S3). Similar to DCMU-treated algae, the PSII-deficient mutant showed a strong light-dependent ¹⁶O₂ uptake (SI Appendix, Fig. S4B).

The short-term experiments undertaken with the DCMU-treated wild-type strain and the FuD7 mutant revealed functioning of the PSII-independent H₂ photoproduction pathway when PSII was not available. Therefore, it was worth checking whether this pathway could sustain H₂ photoproduction in the long-term process. It was subsequently observed that the PSII-independent pathway could not substitute the PSII-dependent pathway in the long-term process (Fig. 3A). The H₂ photoproduction yields both in the FuD7 mutant and in the DCMU-treated wild-type strain began decreasing during the first hour (Fig. 3 A, Inset). This result is in good agreement with those of the short-term experiments (SI Appendix, Fig. S3). In the CC-124 and CC-5325 strains (Fig. 3 A and B), which possess full PSII activity, DCMU addition stopped H₂ photoproduction after 6 h, while H₂ production of the FuD7 mutant continued, but at a greatly reduced rate. In all cases, final H₂ yields were only slightly higher than in the dark samples.

Fig. 3.

Long-term H₂ photoproduction by pulse-illuminated algae. The cultures of the wild-type (CC-124) strain and the PSII-deficient (FuD7) mutant (A), and NDA2-deficient (NDA2) mutant and its wild-type progenitor (CC-5325) (B) were exposed to a train of 1-s light/9-s dark pulses in the absence and in the presence of 30 μM DCMU under an Ar atmosphere. The H₂ production activity of the same strains was also checked under complete darkness. (Insets) The same samples in the first 4 h of pulse illumination. Values are mean of 6 to 13 independent replicates (± SD).

In contrast to PSII deficiency, the absence of NDA2 (responsible for nonphotochemical reduction of the PQ pool) (9, 10) in C. reinhardtii cells was almost completely compensated for by the PSII-dependent pathway. This was particularly evident in the first 4 h of the experiment (Fig. 3 B, Inset). Here, the input of the PSII-independent pathway was noticeable only by the end of the experiment, but even in that case the difference was not significant.

CO₂ Exchange of Algal Cultures Correlates with O₂ Evolution and H₂ Photoproduction.

Previously, we demonstrated that C. reinhardtii cells exposed to a train of 1-s light pulses interrupted by 9-s dark periods do not fix CO₂ and as a result do not accumulate biomass (23, 25). These previous experiments were performed at low cell densities (<10 mg total Chl per L) to ensure optimal photosynthetic performance. In the current work, we employed sensitive MIMS methodology at a higher cell density of around 20 mg Chl⋅L⁻¹ to improve the resolution of CO₂ analysis. Thus, the culture was observed to release CO₂ during the 1-s light/9-s dark pulse-illumination train (Fig. 4). The ambient (m/z 44) CO₂ signal rose within 5 min of engaging the pulse-illumination train and then slowly declined until train cycling was concluded. At this point, the m/z 44 signal decayed exponentially to reach the initial state. Along with ambient CO₂ release, C. reinhardtii cells also produced ¹⁸O-labeled CO₂ isotopologues: C^16,18O₂ (m/z 46) and C^18,18O₂ (m/z 48) (Fig. 4).

Fig. 4.

The release of ambient (m/z 44) CO₂ and ¹⁸O-labeled (m/z 46, m/z 48) CO₂ isotopologues during pulse illumination. The experimental conditions were similar to those presented in Fig. 1. The total Chl content was around 20 mg⋅L⁻¹. A representative experiment is presented. Additional experimental data can be found in SI Appendix, Fig. S11.

Obviously, the release of all CO₂ isotopologues was light-dependent, but the signals did not fluctuate during pulse illumination (Fig. 4). The typical sawtooth shape appeared only when the duration of the light pulse in the light/dark sequence was extended to 8 s. In this regime, a clear sign of CO₂ fixation appeared 5 min after the start of pulse illumination (Fig. 5A). All three CO₂ signals, m/z 44, m/z 46, and m/z 48, were affected (Fig. 5A and SI Appendix, Fig. S5). A similar trend in behavior of m/z 44, m/z 46, and m/z 48 on activation of the CBB cycle indicates that these three signals primarily represent C¹⁶O₂, C^16,18O₂, and C^18,18O₂, but not other possible gases with a similar molecular weight like N₂¹⁶O and N¹⁶O₂. For example, the latter two may appear in C. reinhardtii as a result of NO detoxification (31).

Fig. 5.

The effect of light pulse duration in the pulse-illumination sequence on CO₂ exchange (A), O₂ evolution (B), and H₂ photoproduction (C) of algal cultures. In A, only ambient (m/z 44) CO₂ signals are presented. For behavior of the m/z 46 and m/z 48 CO₂ signals see SI Appendix, Fig. S5. In B, m/z 34 O₂ signal for the 8-s/9-s illumination regime is also shown at low resolution (V) scale. Additional experimental data can be found in SI Appendix, Fig. S12.

As expected, prolonging the light pulse in the light/dark sequence led to enhanced O₂ release, but the burst amounts of O₂ were produced only in the 8-s/9-s regime where the CBB cycle was activated (Fig. 5B; notice the different scale for the 8-s/9-s pulse protocol). By contrast, C. reinhardtii cells exposed to the 5-s light/9-s dark protocol yielded almost the same amount of O₂ as the algae exposed to 3-s/9-s pulse illumination, presumably due to a more pronounced O₂ consumption during the dark phases of the same duration. The enhanced O₂ consumption, however, did not correlate with CO₂ release (Fig. 5A). Interestingly, 3-s/9-s and 5-s/9-s pulse-illumination protocols yielded almost the same amounts of H₂ (Fig. 5C). By contrast, the 8-s/9-s protocol released H₂ only in the beginning of pulse illumination. Thus, the activation of the CBB cycle immediately led to H₂ uptake (Fig. 5C). As seen in Fig. 5 A and C, the H₂ trace (m/z 2) starts to decline at exactly the time when CO₂ signals establish a pronounced sawtooth wave.

Discussion

Clear Evidence for Simultaneous Water Oxidation and H₂ Photoproduction in Green Algae.

The role of PSII in H₂ photoproduction in green algae has long been a subject of debate (32). The PSII-dependent net O₂ release of dark-adapted anoxic algae, when exposed to light, appears only after a period of efficient H₂ production (14). Therefore, direct involvement of PSII in H₂ photoproduction is not obvious, especially because DCMU does not completely inhibit H₂ production (33, 34). In sulfur (S)-deprived C. reinhardtii cells, where the recovery of PSII reaction centers is affected by stress (35), the involvement of residual PSII activity in the H₂ photoproduction yield has been proposed based on inhibitory analysis, electron paramagnetic resonance spectroscopy, and fluorescence data (36–38). Nevertheless, Clark-type O₂ electrodes used directly in algal suspensions have not indicated water-splitting activity in S-deprived cells (36), either due to low sensitivity of the technique (39) or due to the absence of net O₂ release in the actively respiring cells (40). The situation improved with application of a high-sensitive MIMS approach (15, 22). The publications showed that the release of O₂ in dark-adapted algae occurs a few seconds after appearance of H₂ in the system, suggesting direct involvement of PSII in H₂ photoproduction.

As shown in Fig. 1, the application of the 1-s light/9-s dark protocol to anoxic algae results in sustained H₂ photoproduction and also leads to the simultaneous evolution of O₂ (detected at m/z 32 and m/z 34). The appearance of both nonlabeled and ¹⁸O-labeled isotopologues of O₂ upon enrichment with H₂¹⁸O is well known for monitoring PSII-driven water oxidation reaction (41). The appearance of photosynthetically produced O₂ in algal cultures should indicate a shift in the environment from anoxic to microoxic. However, the situation is not so simple. This is for two reasons: First, light activates strong O₂ uptake, which is observed in midlog-phase cultures (Fig. 1B, m/z 32 signal) in the FuD7 mutant (SI Appendix, Fig. S4B) and in the DCMU-treated algae (Fig. 2C, +DCMU trace); second, the release of photosynthetically produced O₂ occurs simultaneously with mitochondrial respiration (which was subtracted from all signals during data processing; SI Appendix, Fig. S6 A and B). These respiratory processes under normal atmospheric pressure balance photosynthetically produced O₂ to a level undetectable by polarographic techniques (Clark-type O₂ electrode) (23). Nevertheless, the competition of H₂ photoproduction with FDP-driven O₂ photoreduction under such low O₂ levels is still highly possible (25), especially in the beginning of the experiment where extracellular O₂ concentrations are at around 0.1 to 0.5 μM. In this context, the gradually increasing H₂ photoproduction rate (Fig. 1A) may reflect a release of FDPs of such competition. Indeed, the level of extracellular O₂ does decrease in the course of the experiment due to its consumption by the instrument and the culture itself (SI Appendix, Fig. S6A).

Surprisingly, the cell wall-deficient CC-4533 strain showed the production of H₂ on the first light pulse (Fig. 1C, CC-4533). The rapid response of cells to the establishment of anoxic conditions (∼3 min from fully aerobic environment) suggests the expression of the O₂-sensitive [FeFe]-H₂ase in C. reinhardtii cells at the time of aerobic cultivation, as has been previously proposed (42). In agreement with this suggestion, transcripts of hydA1 and hydA2 hydrogenases and the hydEF and hydG maturation factors have been found in aerobic algae (43, 44). Since the HYDA1/HYDA2 proteins are also already available in aerobic C. reinhardtii (23), the delay in net H₂ evolution of the CC-124 strain (Fig. 1A) is most likely caused by the limitation of H₂ diffusion through the cell wall. Assuming the reversible nature of the H₂ase-driven reaction (45, 46), enhanced intracellular levels of H₂ may increase H₂ consumption during dark periods and decrease the final H₂ yield. In accordance with this suggestion, prolongation of the light pulse duration to above 3 s in the pulse-illumination sequence demonstrates the early appearance of the H₂ signal in the same strain (Fig. 5C). Similar to the CC-4533 strain, the periods of H₂ photoproduction in the CC-124 algae coincide well with O₂ evolution (Fig. 1C, CC-124). Simultaneous production of H₂ and O₂ in the light indicates the direct involvement of electrons from water oxidation by PSII in H₂ photoproduction.

PSII Plays a Major Role in Sustained H₂ Photoproduction.

Reductants for the H₂ photoproduction of C. reinhardtii are suggested to be supplied not only by PSII but also by the degradation of stored organic substrates, primarily starch (47). In a similar manner to the PSII-dependent process, the indirect pathway requires active PSI for the donation of electrons to H₂ase. However, by contrast, the indirect process results in CO₂ release with a maximum molar stoichiometry of 1 CO₂ per 2 H₂. It has previously been shown that contribution of the indirect pathway to the H₂ photoproduction yield might be significant in S-deprived cells, where the PSII activity is substantially affected by the stress (48).

The results obtained with DCMU-treated algae and the PSII-deficient mutant (Fig. 2A and SI Appendix, Fig. S3) clearly show that the PSII-independent H₂ production pathway can partly compensate for the loss of the PSII-dependent H₂ production pathway toward H₂ photoproduction of pulse-illuminated cells. The indirect process, however, could not sustain H₂ production for longer than 6 h (Fig. 3A), and the specific rate of H₂ production had already started to decline during the first hour (Fig. 3 A, Inset). On the contrary, the elimination of NDA2 in C. reinhardtii, which is the main player of the PSII-independent pathway (10), shows almost no effect on H₂ photoproduction under pulse-illumination conditions (Fig. 3B). It has been previously demonstrated that NDA2 deficiency decreases the H₂ photoproduction activity of S-deprived algae (9, 10). However, S deprivation leads to a significant accumulation of starch reserves in cells during the photosynthetic stage (36). Therefore, it is not surprising that S deprivation increases the contribution of the NDA2-dependent pathway in the H₂ photoproduction yield, especially when the activity of PSII is limited or fully absent (49). Thus, S-deprivation data do not contradict our conclusions, since in actively growing algae, which are used in this study, the starch reserves are limited. In the presence of DCMU, which blocks the electron flow from PSII, the nda2 deletion mutant produces almost the same amount of H₂ as algae placed in complete darkness. The slightly higher H₂ yield of the DCMU-treated nda2 deletion mutant might be attributed to 1) the light activation of fermentation, 2) incomplete inhibition of electron flow by DCMU (Fig. 2B), or 3) the presence of other player(s) in the PSII-independent pathway. If any, their input in the total H₂ yield is negligible and limited to the beginning of pulse-illumination (Fig. 3 B, Inset). A full compensation of algal NDA2 elimination by PSII confirms that H₂ photoproduction during 1-s illumination periods proceeds via the most efficient mechanism of direct water biophotolysis. This conclusion is in agreement with fluorescence data showing that PSII activity in dark-adapted C. reinhardtii cells is linearly related to the hydrogenase capacity observed during the first seconds of illumination (50).

In the 1-s/9-s pulse-illumination regime, algae spend most of the time in darkness, where fermentation plays a major role. According to the data presented in Fig. 3A, up to 4% of the H₂ yield in the CC-124 strain is supported by dark fermentation, while 96% is supported by PSII. In the presence of DCMU, the PSII-independent pathway doubles the H₂ yield in pulse-illuminated algae as compared to the dark samples (Fig. 3A). Thus, even if both pathways (PSII-independent and dark) operate in the CC-124 cells during pulse illumination, PSII is still responsible for about 92% of the final H₂ yield. Very similar results (93% of the PSII input) were obtained for the nda2 deletion mutant, where the contribution of fermentation was close to 7% (Fig. 3B). However, the NDA2 deficiency may also enhance fermentative H₂ production in the mutant. Taking into account all data, we conclude that H₂ production during 1-s light pulses is driven primarily or even exclusively by the PSII-dependent pathway.

Efficient H₂ Photoproduction in C. reinhardtii Occurs before Activation of the CBB Cycle.

The light-dependent water oxidation performed by PSII results in O₂ evolution with a simultaneous release of protons (H⁺) into the thylakoid lumen. In addition, protons are pumped into the lumen by the PQ/plastohydroquinone cycle driven by PSII and the cytochrome b₆f complex. The accumulation of H⁺ in the lumen builds up the ΔpH across the thylakoid membrane and ensures adenosine 5′-triphosphate (ATP) biosynthesis. Simultaneously, a relatively small release of CO₂, which was first detected by MIMS in PSII membranes isolated from spinach (51), could be observed. This CO₂ is known to be released by PSII and is the result of $\mathrm{HC}{\mathrm{O}}_3^{-}$ reaction with H⁺ (produced during water splitting) followed by the subsequent bicarbonate dehydration:

$${\mathrm{H}}^{+}+\mathrm{HC}{\mathrm{O}}_3^{-}\to {\mathrm{H}}_2\mathrm{O}+\mathrm{C}{\mathrm{O}}_2.$$ [4]

This reaction facilitates the removal of H⁺ from PSII. Moreover, the electron-acceptor side of PSII is known to be another source for CO₂ evolution under intensive light illumination (52, 53).

In C. reinhardtii, the bicarbonate dehydration reaction (and in general interconversions of inorganic carbon) can be accelerated by a luminal carbonic anhydrase, CrCAH3 (54–56). However, in whole cells under normal conditions, the light-dependent CO₂ release is barely detectable due to the presence of two processes: 1) the consumption of CO₂ by Rubisco (i.e., by CBB cycle) and 2) mitochondrial respiration. Therefore, it is not surprising that we observed CO₂ release in the pulse-illuminated algae, when the CBB cycle is not active (Fig. 4) and mitochondrial respiration is restricted by an extremely low level of O₂ in the microoxic environment (because of the efficient O₂ consumption by the MIMS setup; SI Appendix, Fig. S6A). As shown in Fig. 4, the light-induced rise of m/z 44, m/z 46, and m/z 48 signals occurs simultaneously with O₂ evolution and H₂ photoproduction (Fig. 1). The appearance of m/z 46 and m/z 48 signals together with ambient CO₂ (m/z 44) indicate a CO₂–water interexchange:

$${\mathrm{H}}_2^{18}\text{O}+{\mathrm{C}}^{16}{\mathrm{O}}_2\rightleftarrows {\mathrm{H}}_2{\mathrm{C}}^{\mathrm{16,16,18}}{\mathrm{O}}_3\rightleftarrows {\mathrm{H}}_2^{16}\text{O}+{\mathrm{C}}^{\mathrm{16,18}}{\mathrm{O}}_2,$$ [5]

$${\mathrm{H}}_2^{18}\mathrm{O}+{\mathrm{C}}^{\mathrm{16,18}}{\mathrm{O}}_2\rightleftarrows {\mathrm{H}}_2{\mathrm{C}}^{\mathrm{16,18,18}}{\mathrm{O}}_3\rightleftarrows {\mathrm{H}}_2^{16}\mathrm{O}+{\mathrm{C}}^{\mathrm{18,18}}{\mathrm{O}}_2.$$ [6]

Alternatively, the inclusion of labeled O₂, which appears as a product of water oxidation (reaction 3), in the organic substrate and its immediate degradation should be expected:

$$n^{\mathrm{16,18}}{\mathrm{O}}_2+{\left[\mathrm{C}{\mathrm{H}}_2^{16}\mathrm{O}\right]}_n\to n\hspace{0.17em}{\mathrm{H}}_2^{16}\text{O}+n\hspace{0.17em}{\mathrm{C}}^{\mathrm{16,18}}{\mathrm{O}}_2,$$ [7]

$$n^{\mathrm{16,18}}{\mathrm{O}}_2+{\left[\mathrm{C}{\mathrm{H}}_2^{16}\mathrm{O}\right]}_n\to n\hspace{0.17em}{\mathrm{H}}_2^{16}\text{O}+n\hspace{0.17em}{\mathrm{C}}^{\mathrm{16,18}}{\mathrm{O}}_2.$$ [8]

The probability of reaction 7, and especially reaction 8, is much lower than reactions 5 and 6, unless they proceed at the same site as reaction 3. Since the reactions 7 and 8 are expected to occur in the stroma of chloroplasts, they are unlikely to satisfy kinetics of light-dependent release of CO₂ isotopologues during pulse illumination (Fig. 4).

Although the observed CO₂ evolution could be potentially caused by activation of substrate degradation in the light (photofermentation), the positive correlation between O₂ evolution (SI Appendix, Fig. S2) and CO₂ release (Fig. 4), and relatively fast recovery of all three CO₂ signals after conclusion of pulse illumination (Fig. 4) strongly suggest direct involvement of PSII in CO₂ evolution. Since the equilibration rate between labeled water and CO₂ occurs with half-times of t_1/2 ≈ 30 s or less (55, 57), any donor-side CO₂ formation or other acidification of the lumen would produce CO₂ isotopolouges reflecting the ${\mathrm{H}}_2^{18}\mathrm{O}$ enrichment of the water (given that measurements were initiated about 5 min after ${\mathrm{H}}_2^{18}\mathrm{O}$ addition). By contrast, CO₂ release at the acceptor side of PSII, where bicarbonate is bound to the nonheme iron and equilibrates slowly with bulk water, would be predominantly C¹⁶O₂. Thus, the different time dependence of C¹⁶O₂ versus C^16,18O₂ and C¹⁸O₂ (Fig. 4, m/z 46 and m/z 48 signals) suggests that light-induced CO₂ release may originate from both the acceptor and donor sides of PSII, as well as the acidification of the lumen. This is further supported by estimating the total amount of released CO₂: If just the acceptor side would contribute, we should expect around 2 nmol of CO₂ released per mg Chl [assuming 1.83 mmol of Q_A per mol of total Chl in photomixotrophic C. reinhardtii cells (58)]. Instead, formation of around 4 to 8 nmol CO₂ per mg Chl is typically observed in algal cultures (Figs. 4 and 5 and SI Appendix, Fig. S11). These data show PSII-associated CO₂ formation in intact cells, which supports the idea that it has an important regulatory and protective function for PSII in vivo.

The prolongation of light pulses to 8 s in the pulse-illumination sequence gradually activates the CBB cycle, leading to a pronounced CO₂ uptake after 5 min of pulse illumination (Fig. 5A and SI Appendix, Fig. S5). As shown in Fig. 5, the activation of the CBB cycle immediately results in the burst release of O₂ (Fig. 5B) and inhibition of H₂ evolution followed by H₂ uptake (Fig. 5C). These data support the previously proposed hypothesis that the primary loss of H₂ evolution activity in C. reinhardtii cells is caused by competition between ferredoxin-NADP⁺ oxidoreductase and H₂ases for reduced ferredoxin, rather than by the sensitivity of hydrogenase to oxygen (19, 22). Besides the activation of the CBB cycle, increasing O₂ levels inside the chloroplast can also enhance the flow of photosynthetic electrons toward FDPs. Such competition indeed becomes more pronounced with prolongation of the light pulse in the sequence (25). In long-term experiments, prolongation of the light pulse results in biomass accumulation (25). These experiments show that the pulse-illumination approach allows a fine tuning of algal metabolism between microoxic H₂ photoproduction and aerobic CO₂ fixation, thus enabling PSII-dependent water-oxidation activity to be maintained at the desired level.

Summary

The illumination of anoxic C. reinhardtii cultures with a train of short light pulses interrupted by longer dark periods demonstrates that efficient H₂ photoproduction in algal cells occurs exclusively in the absence of photosynthetic CO₂ fixation, thus when the CBB cycle is not active. The cells produce H₂ via the most efficient mechanism of direct water biophotolysis, where water oxidation at the donor side of PSII provides the electrons for reduction of protons by the [FeFe]-H₂ase enzyme(s) downstream of PSI. Thus, the two reactions occur simultaneously. However, under normal conditions O₂ is not released by the cells due to its consumption by respiration. During short periods of light illumination, the H₂ase activity supports a linear photosynthetic electron flow from PSII to PSI, promoting proton translocation across the thylakoid membrane and ensuring efficient water oxidation by PSII. Thus, [FeFe]-H₂ase, in concert with PSII, creates favorable conditions for O₂ accumulation and ATP biosynthesis to levels sufficient for the activation of mitochondrial respiration and the CBB cycle. If the duration of light pulses is not sufficient to fulfill the above condition, the algae continue to produce H₂. The long-term production of H₂ in the absence of CO₂ fixation suggests the central role of [FeFe]-H₂ase in supporting algal photosynthesis and cell fitness under anoxic conditions.

Materials and Methods

The wild-type C. reinhardtii strain CC-124, cell wall-deficient strains CC-4533 and CC-5325, psbA deletion (FuD7) mutant CC-4147, and the nda2 deletion mutant (LMJ.RY0402.257129) were obtained from the Chlamydomonas Resource Center at the University of Minnesota, St. Paul, MN. The nda2 deletion is characterized as described in SI Appendix, Fig. S7. All cultures were maintained, grown, and checked for H₂ photoproduction activity under pulse illumination with H₂ and O₂ microsensors (H2-NP and OX-NP; Unisense A/S) as described in detail by Kosourov et al. (23). Conditions for the long-term experiments were reported in the same publication. All experiments were performed in TAP medium with unstressed, actively growing photomixotrophic algae, which were pipetted into the MIMS chamber or vials just before measurements.

The gas exchange in the suspension of algal cells was studied by time-resolved MIMS setup (SI Appendix, Fig. S8) as described previously (59, 60). Briefly, the setup consisted of an isotope ratio mass spectrometer (Delta V Plus; Thermo Fischer Scientific), a cooling trap (−78 °C; dry ice + EtOH), and an in-house-built gas-tight membrane-inlet chamber with 200-µL working volume. Before the measurements, ${\mathrm{H}}_2^{18}\mathrm{O}$ (97%; Larodan Fine Chemicals AB) was added to the MIMS chamber to a final enrichment of 4%. Analysis of O₂ reactions (evolution/consumption) was based on the m/z 32 (¹⁶O₂) and m/z 34 (^16,18O₂) signals, with Faraday cup amplification of 3 × 10⁸ and 1 × 10¹¹, respectively. The signal m/z 36 (^18,18O₂) was not considered due to its low evolution at quite low ${\mathrm{H}}_2^{18}\mathrm{O}$-enrichement and contamination of this signal by the presence of ³⁶Ar. Analysis of CO₂ reactions was based on simultaneous monitoring of m/z 44 (¹²C¹⁶O₂), m/z 46 (¹²C^16,18O₂), and m/z 48 (¹²C¹⁸O₂) signals with cup amplification of 1 × 10⁹, 1 × 10¹¹, and 1 × 10¹², accordingly. The argon signal (m/z 40) with cup amplification of 1 × 10¹¹ was used as a control. H₂ photoproduction activity was studied in separate runs by monitoring the m/z 2 (¹H₂) signal with an amplification of 1 × 10¹². No ${\mathrm{H}}_2^{18}\mathrm{O}$ was added for H₂ assays. The white light-emitting diode light pulses (∼1,000 μmol photons⋅m⁻²⋅s⁻¹) were applied using the STM32F103 microcontroller board controlled by the OxyHydrogen software. Before each measurement, microoxic environment inside the MIMS chamber was achieved within ∼2 min after sealing by degassing the sample with the vacuum pump of the mass spectrometer. The initial levels of oxygen slightly varied from experiment to experiment from 0.1 to 0.5 μM. The final gas exchange curves were obtained after correction for the gas consumption by the mass spectrometer during the dark periods in the beginning and in the end of each experiment as demonstrated in SI Appendix, Fig. S6. This correction also included the gas diffusion component and consumption/production of the gas by the culture during darkness. The analysis of the collected MIMS signals was done using OriginPro 2019 software. The H₂, ¹⁶O₂, and ¹²C¹⁶O₂ yields were determined in moles per milligram of total Chl according to the standard calibration. All other signals were normalized to Chl.

Data Availability.

All data needed to evaluate conclusions of this paper are present in the paper and in SI Appendix.