Photosynthetic hydrogen production: Novel protocols, promising engineering approaches and application of semi-synthetic hydrogenases

Photosynthetic production of molecular hydrogen (H 2 ) by cyanobacteria and green algae is a potential source of renewable energy. These organisms are capable of water biophotolysis by taking advantage of photosynthetic apparatus that links water oxidation at Photosystem II and reduction of protons to H 2 downstream of Photosystem I. Although the process has a theoretical potential to displace fossil fuels, photosynthetic H 2 production in its current state is not yet efficient enough for industrial applications due to a number of physiological, biochemical, and engineering barriers. This article presents a short overview of the metabolic pathways and enzymes involved in H 2 photoproduction in cyanobacteria and green algae and our present understanding of the mechanisms of this process. We also summarize recent advances in engineering photosynthetic cell factories capable of overcoming the major barriers to efficient and sustainable H 2 production.


| INTRODUCTION
Molecular hydrogen (H 2 ) represents the highest energy density per mass unit among currently employed fuels and is a zero-carbon emission solution for the sustainable economy. Besides excellent properties as a fuel, H 2 is an important feedstock for many industrial processes, including methanation of CO and CO 2 , methanol synthesis from CO 2 hydrogenation, and hydrogenation of nitrogen (N 2 ) for the production of ammonia. H 2 is also used as a reducing agent in a number of different applications. Compared to conventional methods of H 2 generation, photobiological H 2 production represents an attractive renewable alternative since it utilizes water and solar light, the world's most abundant resources.
Many species of cyanobacteria and eukaryotic green algae are capable of photosynthetic water splitting to H 2 and oxygen (O 2 ) under specific conditions using dedicated enzymes. Although a theoretical maximum of light energy to H 2 energy conversion efficiency (LHCE) is as high as 10-13% for green algae and 6% for N 2fixing cyanobacteria, only a fraction of these values has been achieved (1.5-2% in algae and up to 4% in cyanobacteria) under controlled laboratory conditions typically employing low light illumination (Bolton & Hall, 1991;Kosourov et al., 2017;Sakurai et al., 2015). A recent study showed that green alga Chlorella vulgaris is capable of producing H 2 with the maximum LHCE up to 8% (3% on the average basis), though in the presence of glucose (Touloupakis et al., 2021). Importantly, volumetric H 2 yields in algal and cyanobacterial cultures are still quite low even in the controlled environment due to a number of physiological and biochemical barriers limiting H 2 production rates.
A deep understanding of the photosynthetic H 2 metabolism and its effect on cell bioenergetics opens new possibilities for the development of H 2 -producing cell factories with improved performance and enhanced H 2 production yields. This mini-review will summarize recent advances in understanding the mechanisms of light-driven H 2 production in cyanobacteria and green algae and give a short overview of strategies and approaches for engineering efficient H 2 -producing cell factories, including whole-cell applications of semi-synthetic hydrogenases.

| A SHORT OVERVIEW OF H METABOLISM IN CYANOBACTERIA
Cyanobacteria are a large and diverse group of oxygenic photoautotrophic prokaryotes. Many of them have an innate capacity to produce H 2 . Cyanobacteria have two sets of enzymes involved in H 2 metabolism: nitrogenases and hydrogenases ( Figure 1A,C). Nitrogenases (encoded by nifHDK) are found in N 2 -fixing unicellular or filamentous cyanobacteria, which produce H 2 as a byproduct of N 2 fixation and its reduction into ammonia. Cyanobacteria also possess two types of [NiFe]-hydrogenases: bidirectional hydrogenase (encoded by hoxEFUYH) and uptake hydrogenase (encoded by hupSL) enzymes (Barz et al., 2010;Lindblad, 2018;Puggioni et al., 2016).
Unlike uptake hydrogenase, which recycles H 2 produced by nitrogenase, the bidirectional hydrogenase both oxidizes and produces H 2 and can be present in diazotrophic and non-diazotrophic strains. Since O 2 is a byproduct of photosynthesis and all enzymes involved in H 2 metabolism are O 2 sensitive, cyanobacteria have developed spatial and temporal strategies for their protection: (1) heterocystous filamentous cyanobacteria spatially separate oxygenic photosynthesis (occurring in vegetative cells) and N 2 fixation (occurring in heterocysts, specialized cells that provide a low-oxygen environment) and F I G U R E 1 A schematic representation of the H 2 photoproduction (brown lines) and competing (black lines) metabolic pathways in (A) unicellular cyanobacterium Synechocystis sp. PCC 6803, (B) green alga Chlamydomonas reinhardtii, and (C) N 2 -fixing heterocystous cyanobacterium Anabaena sp. PCC 7120 (2) non-heterocystous cyanobacteria (e.g., filamentous Trichodesmium, unicellular Cyanothece) perform N 2 fixation, and thus H 2 production, typically under anoxic conditions but also under oxic conditions (e.g., due to the control of their circadian rhythm or intracellular compartmentation). The processes that support H 2 production in these organisms under oxic conditions have yet to be fully resolved. In contrast, the bidirectional hydrogenase particularly operates under microoxic or anaerobic environment and could utilize reductants produced directly by the photosynthetic electron transport chain (PETC). Cyanobacteria bearing this enzyme typically photoproduce H 2 on exposure to light, but only for a very short period followed by H 2 uptake.
The O 2 sensitivity of the bidirectional hydrogenase thus represents a major biotechnological hurdle for efficient H 2 photoproduction via this enzyme.

| A SYSTEM BIOLOGY APPROACH TO UNDERSTANDING THE MECHANISM OF H 2 PRODUCTION IN CYANOBACTERIA
To achieve a greater understanding of how cell metabolism and bioenergetics navigate during H 2 photoproduction and to identify potential targets for further metabolic engineering, a system-level omics approach, which includes genomics, transcriptomics, proteomics, and metabolomics analysis, is necessary. Most of the omics approaches have been focused on understanding the metabolism of nitrogenasebased H 2 production (see, e.g., Aryal et al., 2013;Bernstein et al., 2015;Ekman et al., 2011;Kourpa et al., 2019;Sadler et al., 2016;Stensjö et al., 2007). The differential proteomes of N 2 -fixing and non-N 2 -fixing cultures of both Nostoc sp. and Cyanothece sp. have been resolved, and the heterocyst-specific Cys-proteome was detailed (Sandh et al., 2014).
General important factors to drive nitrogenase-dependent H 2 production in both Cyanothece sp. and Nostoc sp. were (1) to provide sugars, that is, glycogen oxidation in the dark for Cyanothece and import of sucrose from the vegetative cell to heterocysts in Nostoc; (2) to provide enough reducing equivalents by an efficient oxidative pentose phosphate pathway; and (3) to provide ATP by cyclic electron flow (CEF) around Photosystem I (PSI).
For Cyanothece, this view was later challenged when it was shown that the defined key components of H 2 metabolism in Cyanothece differed substantially from that in Nostoc. For example, a positive quantitative correlation between the H 2 production profile and the abundance of proteins of Photosystem II (PSII) was demonstrated (Bernstein et al., 2015). This suggests that in contrast to heterocystous cyanobacteria, the PSII-driven linear electron flow in unicellular Cyanothece directly provides reducing equivalents and ATP for nitrogenase-dependent H 2 photoproduction. These findings bring new strategies for future metabolic engineering and optimization of growth conditions for efficient and sustainable photobiological H 2 production.
Although the usefulness of multi-omics and in silico strategies are well accepted, the attempts to use them for designing H 2 -producing cyanobacteria are still infrequent. One of the delays in the effort to use systems biology as a fundament for bioengineering is the lack of genome-scale metabolic models. Recently, two such models have been constructed to predict pathways and protein targets for improvement of H 2 photoproduction in the diazotrophic heterocystforming model organisms, Anabaena sp. PCC 7120 (Malatinszky et al., 2017) and Anabaena variabilissp. ATCC 29413 (Malek Shahkouhi & Motamedian, 2020). The in silico studies, for the most part, confirmed the targets of importance for improving heterocyst-based H 2 production emphasized in the quantitative proteomics investigations.
Thus, the initial simulations and modeling attempts display the value for further development of these methods.

| ENGINEERING CYANOBACTERIA FOR SUSTAINABLE AND EFFICIENT H 2 PRODUCTION
Since cyanobacteria are capable of catalyzing H 2 photoproduction by the direct (e.g., in unicellular cyanobacteria via bidirectional hydrogenase) and indirect water biophotolysis (e.g., in N 2 -fixing heterocystous cyanobacteria via nitrogenase localized in heterocysts) mechanisms, challenges and engineering strategies for improving H 2 photoproduction yields also differ to some extent. In general, the major metabolic bottlenecks of cyanobacterial H 2 production include (1) high O 2 sensitivity of the enzymes involved in H 2 metabolism (nitrogenases and/or hydrogenases), (2) low photosynthetic yield, (3) alternative electron transport pathways competing for photosynthetic electrons, and (4) H 2 uptake by the cells (via bidirectional or uptake hydrogenases). Therefore, the following challenges are to be addressed: (1) modulation of competing pathways for boosting the H 2 production yield without compromising cell fitness; (2) improvement of photosynthesis, including optimization of NADPH/ATP ratio, for efficient bioproduction; (3) screening available cyanobacterial culture collections; (4) mining and introducing novel, more efficient, and O 2 -tolerant enzymes; and (5) development of novel cultivation protocols for efficient H 2 production. Throughout the years, there have been numerous attempts to overcome these barriers and challenges by employing a diversity of genetic engineering approaches. Here we focus on the most recent and novel advances in engineering cyanobacteria for sustainable H 2 photoproduction.

| Deletion of the capacity to take up hydrogen
There are many examples in the literature of either disrupting or deleting a structural gene, or a gene encoding a protein of the maturation machinery, of the uptake HupSL hydrogenase in cyanobacteria (see, e.g., Lindblad, 2018). The result will be H 2 -producing cells, where H 2 evolution is catalyzed by a native nitrogenase. Using a ΔHupW strain of Anabaena sp. PCC 7120 (a strain lacking the last step in the maturation machinery) analyzed in a 5.0 L flat-panel photobioreactor system with a 3.0 L culture volume and illuminated with a mixture of red and white LED lights, Nyberg et al. (2015) showed a maximal LHCE of ca. 4.0% (in the PAR region) and a maximal hydrogen concentration of 6.89%. However, it is also important to note that inactivation of uptake hydrogenase reprograms cell metabolism (Ekman et al., 2011;Kourpa et al., 2019) and may significantly decrease the duration of the H 2 production process and cell fitness, particularly under a high C/N environment .
Another approach is to modify the flow of electrons from H 2 uptake toward H 2 evolution in the HupSL hydrogenase. The NiFecontaining active site is present in the large subunit, whereas the small subunit (HupS) harbors three FeS-clusters. HupS of the filamentous cyanobacterium Nostoc punctiforme ATCC 29133 was heterologously expressed as a fusion protein in Escherichia coli (f-HupS), purified, and characterized (Raleiras et al., 2013). Thereafter, the proximal Fe-S cluster of HupS was modified by changing the cysteine in position 12 to proline (C12P). C12P f-HupS investigated by EPR spectroscopy demonstrated a conversion from a 4Fe4S to a 3Fe4S proximal cluster.
C12P HupSL was then introduced and expressed in N. punctiforme, which resulted in cells with increased H 2 production. The modified proximal Fe-S cluster in HupS changed the flow of electrons toward hydrogen production (Raleiras et al., 2016).

| Modulation of alternative electron transport routes
Regardless of hydrogenase, a steady flow of electrons to the hydrogenase enzyme, provided by primary electron donors such as reduced ferredoxin, is needed for efficient and sustainable photobiological H 2 production (Gutekunst et al., 2014). However, there is no "bag-ofelectrons"; the available photosynthetic electrons are utilized in other assimilatory pathways, such as the respiratory electron transport system, nitrate assimilation, and carbon fixation via the Calvin-Benson-Bassham (CBB) cycle. From a biotechnological aspect, these are competing routes. Moreover, alternative electron transport pathways such as the flavodiiron proteins (FDPs)-driven water-water cycle and the cyclic electron transport around PSI can be considered as "waste" points for photosynthetic production. Therefore, a genetic engineering strategy should be applied for re-directing the electron flow away from the existing native assimilatory and auxiliary pathways toward H 2 production. This has been experimentally demonstrated in Synechocystis sp. PCC 6803 by eliminating the electron flow to (1) the respiratory terminal oxidases (Gutthann et al., 2007), (2) the nitrateassimilation pathway (Baebprasert et al., 2011), and (3) the NDH-1 complex involved in CEF, respiration, and carbon concentrating mechanism (Cournac et al., 2004). These examples all resulted in increased H 2 production yields. Specifically, removing nitrate from the growth medium, or creating strains lacking a functional pathway to assimilate nitrate, significantly increases photobiological H 2 production.
Despite the fact that filamentous heterocystous N 2 -fixing cyanobacteria are the most fascinating organisms for biotechnological applications, the electron-transport pathways inside heterocysts are largely unknown. A recent study demonstrated that the heterocyst-specific Flv3B enzyme actively eliminates O 2 thus enabling microoxic interior inside of the heterocyst, which supports diazotrophic growth and nitrogenase enzyme activity under illumination (Ermakova et al., 2014).
Accordingly, overexpression of the Flv3B protein resulted in high H 2 photoproduction, even though this phenotype was not linked to enhanced nitrogenase activity (Roumezi et al., 2020). The latter requires further investigations.

| PSI-hydrogenase fusion in vivo
Semiartificial in vitro techniques can be created in which hydrogenases are attached to isolated photosystems for H 2 production. However, generally such systems are short lived. Recently, Appel et al. (2020) reported photosynthetic H 2 production using a PSI-hydrogenase fusion in vivo. HoxYH of the NiFe-hydrogenase of Synechocystis sp. PCC 6803 was fused to its PSI subunit PsaD. The recombinant Synechocystis strain showed long-lived H 2 production in a lightdependent manner with otherwise undisturbed metabolism. with HydA from Clostridium acetobutylicum. The obtained mutant strain showed significantly higher H 2 production under an argon atmosphere, but weak H 2 photoproduction was also recorded in the presence of air (Avilan et al., 2018).

| Introduction of [FeFe]-hydrogenases in cyanobacteria
As further discussed below, another strategy is now available that circumvents the biological maturation of [FeFe]-hydrogenase by an artificial synthetic activation of the heterologously expressed HydA protein in living cells. Whereas the structural HydA apoprotein (from e.g. the green alga Chlamydomonas reinhardtii) is expressed using a designed gene fragment, a functional HydA is created by the addition of a synthetic analogue of the [2Fe] H subsite outside the cells (Wegelius et al., 2018). The experiments showed that the non-native, semi-synthetic enzyme retains its H 2 production capacity for several days after synthetic activation with the regulation of activity based on the availability of electrons.

| Whole-cell immobilization approach
The immobilization of enzymes and microbes within polymer matrices is an attractive approach to extending the lifetime of catalysts and enabling an easy and continuous production platform. Improved cell fitness and H 2 photoproduction activity have been demonstrated in cyanobacteria and green algae immobilized in thin-films (Jämsä et al., 2018;Kosourov et al., 2017;Wutthithien et al., 2019). Recently, a proof-of-concept for the sustained nitrogenase-based H 2 photoproduction was demonstrated by engineered thin-layer alginate films with entrapped heterocysts cells (Volgusheva et al., 2019).
Immobilized heterocysts were able to produce H 2 in the presence of externally added sucrose for about 20 days, whereas heterocysts in suspensions showed similar activity for only 24 h.

| RECENT ADVANCES IN UNDERSTANDING H 2 METABOLISM IN GREEN ALGAE
Hydrogen metabolism in eukaryotic green algae is mediated by [FeFe]-hydrogenase enzymes that catalyze the reversible reduction of protons to molecular hydrogen using a biologically unique diiron cofactor (the [2Fe] H subsite). Depending on the redox state of the cells and the partial pressure of H 2 in the environment, algae could either produce or oxidize H 2 . Concerning H 2 metabolism, C. reinhardtii is the most studied alga and thereafter it would be a primary focus of this review.
Similar to many other green algae, C. reinhardtii possesses two distinct hydrogenase genes in the nuclear genome that encode for two proteins, CrHYDA1 and CrHYDA2 (Forestier et al., 2003;Happe & Kaminski, 2002). Some algal species carry three hydrogenases (Skjånes et al., 2010). Until recently, the physiological basis for the presence of multiple hydrogenases in cells has not been yet determined. The electrochemical study performed by Engelbrecht et al. (2021) showed the higher catalytic preference of CrHYDA2 to H 2 oxidation, thus revealing its potential role in H 2 uptake in C. reinhardtii cells. The biosynthesis and assembly of the complete active site of [FeFe]-hydrogenases is a comparatively simple process that involves three specific maturase enzymes: HydE, HydF, and HydG (Posewitz et al., 2004). In C. reinhardtii, genes encoding HydE and HydF proteins are fused, but in the majority of other organisms hydE, hydF, and hydG exist as separately transcribed genes.
It has been widely accepted that C. reinhardtii induces [FeFe]hydrogenases and activates H 2 metabolic pathways only under strictly anoxic conditions (see, e.g., Dubini & Ghirardi, 2015). The extreme sensitivity of isolated [FeFe]-hydrogenases to atmospheric O 2 with oxidation and irreversible inactivation of their catalytic center supported the anaerobic nature of these enzymes (Stripp et al., 2009).
The first evidence on the presence of hydrogenase transcripts in aero-  (Burlacot et al., 2018). Similarly, the presence of the [FeFe]-hydrogenase in aerobic samples was proposed by Kosourov et al. (2018), but in contrast to Liran's et al.'s (2016) study, authors could not detect the production of H 2 in aerobic cultures.
Instead, they observed the appearance of the active [FeFe]-hydrogenase in cells within less than 3 min of transition from fully aerobic (21% of O 2 ) to microoxic (0.01-0.04% of O 2 ) environment (Kosourov et al., , 2020. Since [FeFe]-hydrogenases are nuclear-encoded and their transportation to the chloroplast with following maturation requires time, the authors suggested the expression of hydrogenases during aerobic cultivation and activation of these enzymes under a microoxic environment. It should be noted, however, that Kosourov and co-authors worked with synchronized cultures, which were sampled during the period of active photosynthesis. Therefore, it is possible that Liran et al. study (employing unsynchronized cultures) monitored H 2 production from a fraction of algae in a different meta- The most efficient H 2 photoproduction in C. reinhardtii cultures occurs on their exposure to light after the period of the dark anaerobic adaptation. According to Boichenko et al. (2004), some active algal producers placed under optimal conditions could reach production rates up to 300 μmol H 2 h -1 (mg Chl) À1 , which is close to the maximal steady-state rate of CO 2 -dependent O 2 evolution (Boichenko et al., 2004). Thus, H 2 photoproduction in green algae has a great potential for industrial applications, but the process proceeds for only a short time (from a few seconds to a few minutes).
Hydrogen photoproduction in algae ( Figure 1B) depends on two metabolic pathways: (1) direct water biophotolysis, which involves water oxidation at PSII, and (2) indirect process, which supplies reductants originating from the oxidation of organic substrates to the PQ-pool via the Type II NADPH dehydrogenase (NDA2) (see, e.g., Dubini & Ghirardi, 2015). The contribution of each pathway varies to a different degree depending on the metabolic state of the cell and stage of hydrogen production. While the availability of the indirect pathway in algae was shown by multiple studies with photosynthetic inhibitors, the role of the PSII-dependent process in H 2 photoproduction has been mostly probed indirectly. The reason for that was the absence of O 2 evolution during the period of efficient H 2 photoproduction and inhibition of the latter as soon as O 2 starts to accumulate in the culture. Even if the absence of net O 2 production could be explained by the removal of O 2 in respiration (especially for the sulfur-deprived algae with their residual PSII activity; Melis et al., 2000;Tsygankov et al., 2006), the other research data indicate the domination of PSII-independent pathway in algal cells. Many short-term experiments performed with dark-adapted algae showed the quantum yield of H 2 evolution significantly exceeding 0.25 (Boichenko et al., 2004), the maximum theoretical value for the process depending on two photosystems. The latter means that H 2 photoproduction in green algae occurs at least partially through the mechanism independent of water oxidation.
Moreover, PSII-deficient strains do not lose the capacity to photoproduce H 2 , at least in the short-term process (Kosourov et al., 2020).
The direct photosynthetic pathway involving both photosystems is the most promising for the development of sustainable H 2 production. It utilizes water oxidation by PSII as an electron source for the hydrogenase-driven H + reduction but also co-evolves O 2 , a by-product of water oxidation. Since O 2 inhibits the [FeFe]-hydrogenase enzyme, it is clear that the PSII activity must be contained to a level where respiration overtakes O 2 evolution thus creating an anoxic state in the cells (or at least in the pockets where the [FeFe]-hydrogenase operates) and allowing prolonged H 2 evolution. Early strategies for sustaining H 2 production in algal cultures pursued this goal. The nutrient limitation approach and, in particular, the most-studied sulfur (S)-deprivation protocol is one of them. This approach allows effective termination of the de novo PSII synthesis leading to a gradual inhibition of the PSIIdependent water oxidation activity about 20% of the original level and enabling spontaneous anoxia in algal cultures with following induction of hydrogenase enzymes in cells (Melis et al., 2000;Volgusheva et al., 2013). Another important consequence of S-deprivation is the reduced state of the PQ pool, which in some cases enables H 2 photoproduction for more than a week (Volgusheva et al., 2013). Since S-deprivation gradually degrades PSII centers in algal cells, their contribution to H 2 photoproduction declines over time and is replaced by the degradation of organic substrates through the PSII-independent pathway, primarily by the degradation of starch. The involvement of the residual PSII activity in the H 2 photoproduction yield in S-deprived algae has been proven by spectroscopic studies, fluorescence data, and inhibitory analysis (Melis et al., 2000;Volgusheva et al., 2013Volgusheva et al., , 2016.
Alternatively, the PSII activity could be restricted by specific mutations leading to similar results. However, we have to emphasize that strategies aiming at the restriction of PSII activity in cells are a temporary solution for sustaining H 2 production in algal cultures. They significantly decrease the efficiency of the process and should be replaced by the methods aiming at the natural regulation of photosynthetic electron flow from PSII to the [FeFe]-hydrogenase.

| A SYSTEM BIOLOGY APPROACH FOR UNDERSTANDING AND IMPROVING H 2 PRODUCTION IN GREEN ALGAE
Multi-omics approaches, which include data from genomics, transcriptomics, proteomics, and metabolomic analyses, have been used for identification of genes and proteins involved in H 2 production in C. reinhardtii and some other algal species (see, e.g., Xu et al., 2019).
These genes and proteins are potential new targets for metabolic engineering and improving H 2 production yields. Although genetic analysis has been plentifully employed, the analytical proteome studies of H 2 -producing algae are quite rare. As a result, the experimental data at the protein level are still limited, especially in combination with other omics approaches. Comparative proteomics has been applied to sulfur-and nitrogen-deprived algae at different stages of nutrient starvation (Chen et al., 2010;Li et al., 2021) and to anaerobically adapted C. reinhardtii cultures under continuous illumination (Terashima et al., 2010). In dark-adapted anaerobic C. reinhardtii, the proteome analysis has been combined with transcriptomics and metabolomics data (Subramanian et al., 2014). The authors of these studies identified new candidates for further investigations and targeted engineering, which could clarify peculiarities of H 2 metabolism in algae. However, no complex metabolic engineering strategies, which are based on findings from multi-omics data, have been implemented yet for improving the algal H 2 production yield.

| Balancing PSII to PSI ratio
One of the important conclusions derived from S-deprivation studies of C. reinhardtii is that the complete reduction of the PQ-pool is the initial condition for induction of H 2 photoproduction in algal cells (Volgusheva et al., 2015(Volgusheva et al., , 2016. The reduction of the PQ-pool reflects the overall reduced state in the chloroplast and also results in the decreased PSII activity. Both conditions (the reduced PQ-pool and the decrease PSII activity) facilitate H 2 evolution as a way to relieve the electron pressure in the cell.
Although the redox state of the PQ-pool in the thylakoid membrane of C. reinhardtii depends on many factors, a proper condition for the induction of H 2 photoproduction can be achieved by simply altering the balance between two photosystems, PSII and PSI. A decreased amount of PSI can potentially modulate the electron transfer from the PSII by generating the reduced PQ-pool, thus also the decreased PSII activity. Indeed, the C3 mutant strain of C. reinhardtii with an abnormally low PSI content exhibits a reduced state of the PQ-pool already at standard growth conditions (Krishna et al., 2019). Incubation of C3 mutant in closed photobioreactor resulted in the establishment of anaerobiosis within 17 h (allowing respiration to overtake O 2 evolution) and stimulated H 2 photoproduction for an unprecedented period of at least 6 weeks under the standard continuous light and growth conditions (Krishna et al., 2019). Although a low amount of PSI was pre-request for quick initiation H 2 photoproduction, after 6 days it has significantly exceeded the amount of PSII, demonstrating the importance of PSI at later stages. It is possible that by optimizing PSII/ PSI ratio in the cells even longer H 2 photoproduction can be achieved.

| Avoiding competition with CO 2 fixation and other metabolic pathways
As mentioned above, H 2 photoproduction in the nutrient-replete green algae is a transient phenomenon, which is typically observed in dark-adapted anaerobic cultures after their exposure to light. Under saturating illumination, algae evolve H 2 for only a few seconds and start accumulating O 2 . Therefore, the inactivation of H 2 photoproduction in cells was primarily attributed to the irreversible inhibition of hydrogenase enzymes by O 2 accumulated in algal cultures. For a while, the O 2 -sensitivity hypothesis became dominating despite (1) the proposed role of hydrogenases in the activation of the CBB cycle (see, e.g., Godaux et al., 2015) and (2) the available data on the direct competition of H 2 photoproduction for reducing equivalents with CO 2 fixation. The O 2 -sensitivity concept has been reconsidered after three independent studies Milrad et al., 2018;Nagy et al., 2018), which were published almost simultaneously. Since activation of the CBB cycle in dark-adapted algae requires time, Kosourov et al. (2018) proposed that a sequence of short light pulses applied to cultures instead of continuous light will prevent induction of CO 2 fixation, eliminate the competition, and thus, make a substantial number of photosynthetic electrons available to hydrogenase. The performed research corroborated the prediction. A train of 1-s light pulses interrupted by 9-s dark phases indeed induced efficient H 2 photoproduction in C. reinhardtii cultures and sustained the process for up to 3 days. The maximum rates were observed in the first 8 h. In contrast to the classical dark adaptation protocol, only a very short period (3-5 min) of purging with argon was needed for activation of the hydrogenase enzyme in cells. The cell-wall deficient strain (CC-4533) of C. reinhardtii showed H 2 evolution already on the first light pulse (Kosourov et al., 2020), indicating that hydrogenase proteins were already available in aerobic cultures. As expected, the pulse-illuminated algae could not fix CO 2 and produce biomass (Jokel et al., 2019;Kosourov et al., 2018). Instead, the PSII-dependent CO 2 release was detected by MIMS in pulse-illuminated algal cultures (Kosourov et al., 2020). The CO 2 consumption and biomass accumulation were detected when longer light pulses (5-8 s) activated the CBB cycle (Jokel et al., 2019). The latter was accompanied by the boost in the production of O 2 in the cultures and pronounced H 2 uptake.
The application of 18 O 2 -labeled water, DCMU, and PSII-and NDA2-deficient mutants showed that up to 92% of H 2 in the pulseilluminated algae originate from water oxidation activity, while the remaining H 2 is evolved by fermentation during the prolonged dark periods (Kosourov et al., 2020).
Another method for preventing competition with the CBB cycle was devised by Nagy et al. (2018). The method relies on the substrate limitation of the CBB cycle in very dense CO 2 -and acetate-limited algal cultures. In this approach, anaerobic C. reinhardtii suspensions are dark-adapted for several hours with periodic re-flushing by N 2 (or Ar) for the CO 2 removal. The dark adaptation step is followed by exposure of algae to continuous illumination, which enables H 2 photoproduction. Periodic purging of H 2 -producing algae with N 2 prevents the accumulation of CO 2 in cultures and sustains H 2 photoproduction for several days. It should be noted, however, that N 2 re-flushing removes not only CO 2 but also O 2 and H 2 from the cultures. The latter prevents inhibition of [FeFe]-hydrogenases and, in principle, should limit H 2 oxidation through the reversible process. The authors suggested that H 2 photoproduction in CO 2 -limited algae is driven by direct water biophotolysis since DCMU completely blocks the process.
Besides competition with CO 2 assimilation and O 2 photoreduction, H 2 photoproduction in C. reinhardtii cells is negatively affected by the induction of CEF around PSI. Elimination of state transition and disruption of PGR5 and PGRL1 proteins, essential components of CEF, resulted in significantly enhanced H 2 photoproduction yields in S-deprived algae (Kruse et al., 2005;Steinbeck et al., 2015;Tolleter et al., 2011). The performance of these mutants may be further improved by eliminating the competition of H 2 photoproduction with the CBB cycle, thus redirecting most of photosynthetic electrons to [FeFe]-hydrogenase.

| Engineering photosynthesis for improved electron flow toward hydrogenase
A few recent studies have shown that the efficiency of photosynthetic electron flow toward hydrogenase could be significantly improved by applying protein fusion technology. For avoiding competition with ferredoxin-NADP + oxidoreductase (FNR), ferredoxin could be fused to hydrogenase (Yacoby et al., 2011). By fusing [FeFe]hydrogenase to ferredoxin, Eilenberg et al. (2016) showed that the new synthetic enzyme supports $4.5-fold greater photosynthetic H 2 production rate in vivo and improved resistance to O 2 than the native [FeFe]-hydrogenase. Alternatively, [FeFe]-hydrogenase could be directly fused to PSI. Following this approach, Kanygin et al. (2020) designed and created the PSI-hydrogenase chimera where the HydA sequence was inserted into the PsaC subunit. The engineered strain placed under anoxic conditions demonstrated sustained H 2 photoproduction for 5 days at an average rate of 14 μmol H 2 h -1 (mg Chl) À1 .
This work further supported the hypothesis that sustained H 2 photoproduction in nutrient-replete algae could be achieved by avoiding competition with the CBB cycle.

| Eliminating the H 2 uptake component
Under an increased H 2 partial pressure, algal hydrogenases could also function in H 2 uptake. In 2012, Kosourov and co-authors evaluated the effect of the H 2 partial pressure in S-deprived cultures and found an exponential decay in the H 2 photoproduction yield with the increased H 2 level (Kosourov et al., 2012). The authors also observed a transient H 2 uptake after injecting extra H 2 in the photobioreactor headspace. The presence of active H 2 uptake in sulfur-deprived algae was later confirmed by Scoma and Hemschemeier (2017). Yet, the mechanism and metabolic relevance of this reaction have not been revealed. The possibility of CO 2 photoreduction in sulfur-deprived algae is questionable because of the noticeable degradation of the RuBisCo enzyme by the time H 2 photoproduction begins. Nevertheless, as Scoma and Hemschemeier (2017) suggested, H 2 oxidation could be linked to fermentative CO 2 reduction via PFR1, which also interacts with [FeFe]-hydrogenases. Meanwhile, the oxyhydrogen reaction should be limited by the low levels of O 2 in algae due to the degradation of PSII centers, the only source of O 2 in the sealed sulfurdeprived cultures. Recently, the presence of strong H 2 uptake was demonstrated in the pulse-illuminated algae on the shift to darkness . The reaction occurred even in the DCMUtreated cells (thus under strong anaerobiosis) and does not show a strong correlation with CO 2 consumption (Kosourov et al., 2020).

| SEMI-SYNTHETIC HYDROGENASES
Hydrogenases possess a remarkable capacity for the interconversion of H + to H 2 on par with noble metal artificial catalysts; however, the enzymes achieve this activity with first row transition metals and can thus be readily produced in abundant quantities (Kleinhaus et al., 2021;Land, Senger, et al., 2020). Therefore, they have become increasingly employed in H 2 production applications, especially [FeFe]-hydrogenases. They possess the highest catalytic activity among different hydrogenases and a comparatively simple maturation, which involves three specific enzymes denoted as HydE, -G, and -F (Posewitz et al., 2004). Through the combined activities of two radical S-adenosyl-L-methionine enzymes (HydG, HydE), a precatalyst of the [2Fe] H subsite is assembled on HydF. The pre-catalyst is subsequently transferred to the apo [FeFe]-hydrogenase, containing a pre-assembled [4Fe4S] H cluster, to yield the active enzyme (Britt et al., 2020). Despite its relative simplicity, the requirement for co-expression of the [2Fe] H maturation enzymes to obtain active [FeFe]-hydrogenases initially complicated both large-scale screening as well as detailed studies of these biotechnologically relevant enzymes, as isolation was dependent on anaerobic isolation of the enzyme following homologous expression or the use of specifically engineered E. coli strains.
In 2013, synthetic chemists reported that the complex biological H-cluster maturation machinery of [FeFe]-hydrogenases could be avoided by insertion of chemically synthesized diiron sites into the apo protein Esselborn et al., 2013).
More specifically, the synthetic [2Fe] H subsite analogue [(μ-adt) Fe 2 (CN) 2 (CO) 4 ] 2À (adt = azadithioloate) generates a fully active enzyme, but modified cofactors can also be introduced to generate The increased H 2 evolution activity of [FeFe]-hydrogenase introduced into cyanobacteria serves as an example of how an artificial metalloenzyme can tune the performance of host cell systems and outperform the native catalysis pathways. Additionally, the possibility to re-design the H-cluster with synthesized metal cofactors opens the perspective of creating enzymes that catalyzes abiotic reactions not present in natural systems. In this respect, the concept of artificial maturation/semi-synthetic hydrogenases also provides a promising perspective for the design of non-natural metabolic pathways for the production of high-value chemicals.

| CONCLUSIONS
The release of H 2 by algae and cyanobacteria in the environment is bioenergetically expensive and results in a significant loss of metabolic energy. Therefore, photosynthetic microbes typically photoproduce H 2 only "on-demand" for a very short period or recycle produced H 2 via dedicated metabolic pathways. H 2 recycling has already been known for years in N 2 -fixing cyanobacteria, where H 2 is imminently co-produced during N 2 fixation and then oxidized by the uptake hydrogenase. Elimination of HupSL and then its modification for H 2 evolution instead of H 2 oxidation, which resulted in the boost production of H 2 in N 2 -fixing heterocystous cyanobacteria, were the first and the second most noticeable achievements in this field. Recently, the presence of the active H 2 uptake component has also been confirmed in green algae, but molecular mechanisms of this process are still not clear. We expect that elimination of this pathway, if achievable, may enhance H 2 photoproduction yields in green algae as well.
Besides H 2 recycling, photosynthetic H 2 production competes for the reducing equivalents with a number of assimilatory and auxiliary pathways, from which CO 2 fixation in the CBB cycle and O 2 photoreduction by FDPs represent the strongest sinks both in cyanobacteria and green algae. For years, the decline in H 2 photoproduction via the direct water biophotolysis pathway in green algae (but not in cyanobacteria where H 2 photoproduction via the bidirectional hydrogenase is followed by noticeable H 2 uptake) has been primarily attributed to