Synthetic biology for improved hydrogen production in Chlamydomonas reinhardtii

Summary

Hydrogen is a clean alternative to fossil fuels. It has applications for electricity generation and transportation and is used for the manufacturing of ammonia and steel. However, today, H₂ is almost exclusively produced from coal and natural gas. As such, methods to produce H₂ that do not use fossil fuels need to be developed and adopted. The biological manufacturing of H₂ may be one promising solution as this process is clean and renewable. Hydrogen is produced biologically via enzymes called hydrogenases. There are three classes of hydrogenases namely [FeFe], [NiFe] and [Fe] hydrogenases. The [FeFe] hydrogenase HydA1 from the model unicellular algae Chlamydomonas reinhardtii has been studied extensively and belongs to the A1 subclass of [FeFe] hydrogenases that have the highest turnover frequencies amongst hydrogenases (21,000 ± 12,000 H₂ s⁻¹ for CaHydA from Clostridium acetobutyliticum). Yet to date, limitations in C. reinhardtii H₂ production pathways have hampered commercial scale implementation, in part due to O₂ sensitivity of hydrogenases and competing metabolic pathways, resulting in low H₂ production efficiency. Here, we describe key processes in the biogenesis of HydA1 and H₂ production pathways in C. reinhardtii. We also summarize recent advancements of algal H₂ production using synthetic biology and describe valuable tools such as high‐throughput screening (HTS) assays to accelerate the process of engineering algae for commercial biological H₂ production.

Hydrogen as a clean energy source is the fuel of the future. In this minireview, we highlight advantages and limitations for biological hydrogen production in the unicellular algae Chlamydomonas reinhardtii. We review tools for improving this biological hydrogen production using modern Synthetic Biology.

Introduction

Hydrogen is the most abundant element in the universe, but on earth, it mostly exists as compounds with other elements. Thus, molecular H₂ is not readily available and needs to be manufactured. Hydrogen has a high gravimetric energy density (~ 120 MJ kg⁻¹), approximately triple that of gasoline (~ 46 MJ kg⁻¹) (Ley et al., 2014), and can be directly converted to electricity via a H₂ fuel cell, only emitting heat and water as local by‐products (Wu et al., 2012). Furthermore, H₂ is used for manufacturing of chemicals including the production of ammonia for fertilizer, for steel manufacturing, and also the methanol industry (Ramachandran and Menon, 1998). To date, commercially available H₂ is sourced primarily from fossil fuels with approximately 99% manufactured from coal (76%) and natural gas (23%), typically via coal gasification and steam reformation (IEA, 2019). As such, renewable and sustainable methods for H₂ production are required to decarbonize the H₂ economy.

One means to supply clean H₂ is through water electrolysis. However, the worldwide large scale electrolyser capacity is only sufficient to provide less than 0.1% of the current H₂ demand (IEA, 2019). Commercial electrolysers today also require highly purified water and typically must operate continuously, which is generally not compatible with intermittent electricity supply from solar and wind. While there is ongoing work to improve the efficiency and scale of electrolysis to be more economically viable (Burton et al., 2021), other sustainable and low‐carbon methods for producing renewable H₂ should also be examined to help accelerate the transition to achieving a renewable H₂ economy.

The field of microbial biotechnology has the potential to contribute to the production of carbon‐neutral and renewable H₂. Organisms from all 3 domains of life can produce H₂ using enzymes known as hydrogenases (Peters et al., 2015). In general, hydrogenases catalyse the reversible oxidation of protons into H₂. There are three classes of hydrogenase: [Fe], [NiFe] and [FeFe]. These three classes can be further separated into 38 subclasses according to their biological function and substrates (Søndergaard et al., 2016). The [Fe] hydrogenases are a unique case as they are only used by a select group of methanogenic archaea and function differently to [NiFe] and [FeFe] hydrogenases, splitting H₂ into H⁺ and H^‐ and delivering the hydride to methenyltetrahydromethanopterin (methenyl‐H₄MPT⁺) (Nicolet, 2019). Many similar functions are shared amongst the 29 [NiFe] subclasses and 8 [FeFe] subclasses such as H₂ evolution, H₂ oxidation, bifurcating H₂ evolution/oxidation, bidirectional H₂ evolution/oxidation and H₂/H⁺ sensory functions (Søndergaard et al., 2016). Additionally, [NiFe] and [FeFe] hydrogenases have similar structures, both contain a di‐metallic active centre bridged by sulfur (S) atoms with additional CO and CN ligands. Despite these similarities, the [NiFe] and [FeFe] hydrogenases do not share a common ancestor and seem to be a remarkable case of convergent evolution (Peters et al., 2015).

From a biotechnological standpoint, [FeFe] and [NiFe] hydrogenases each have advantages and disadvantages. [FeFe] hydrogenases have the highest rates of H₂ production reported so far, with turnover frequencies as high as 21,000 ± 12,000 H₂ s⁻¹ for CaHydA from Clostridium acetobutyliticum (Madden et al., 2012). However, the [FeFe] hydrogenase class are extremely labile in the presence of O₂ with concentrations as low as 0.01% O₂ shown to damage the enzyme in vitro (Swanson et al., 2015). In comparison, the [NiFe] hydrogenases display lower turnover frequencies, such as 1,340 ± 115 H₂ s⁻¹ for the hydrogenase produced by Synecchcoystis sp. PCC 6803 (Gutekunst et al., 2018). However, approximately 44% of the ~ 2000 identified [NiFe] hydrogenases are predicted to be O₂ tolerant, which can be advantageous as maintaining strict anaerobic conditions is difficult at scale for commercial applications (Cracknell et al., 2009; Søndergaard et al., 2016). This review will focus on FeFe hydrogenases, specifically HydA1, and developments that improve their H₂ production.

Biological H₂ production catalysed by [FeFe] hydrogenases requires a characteristic cofactor known as the H‐cluster, with a maturation system both forming and inserting the H‐cluster cofactor into the [FeFe] hydrogenase. Components of this maturation system include the S utilization factor (SUF) iron–sulfur (Fe‐S) assembly machinery in conjunction with the GTPase HydF and the radical SAM (rSAM) enzymes HydE and HydG. These components facilitate the stepwise biosynthesis and insertion of the H‐cluster into [FeFe] hydrogenases.

One of the best‐characterized [FeFe] hydrogenases is the [FeFe] hydrogenase HydA1 from C. reinhardtii. C. reinhardtii is a unicellular phototrophic green algae and a model organism, which has been studied extensively since the mid‐20th century for fundamental processes such as photosynthesis and H₂ production (Sasso et al., 2018). Phototrophic microalgae, like C. reinhardtii, are of great interest to the biotechnology industry since they can be grown to high densities in photobioreactors and harness solar energy for producing biomass and a wealth of valuable compounds, including H₂ (Fabris et al., 2020). In C. reinhardtii, there are two endogenous [FeFe] hydrogenases capable of producing H₂: HydA1 (accounting for ~ 80% of the total H₂ production) and HydA2 (accounting for ~ 20% of the total H₂ production) (Godman et al., 2010; Meuser et al., 2012). C. reinhardtii also contains a HydEF fusion protein that is analogous to the individual HydE and HydF proteins found in other organisms (Posewitz et al., 2004a). Both HydA1 and HydA2 belong to the [FeFe] A1 subclass that evolves H₂ via oxidation of ferredoxins (Søndergaard et al., 2016), which are Fe‐S proteins that function as electron carriers in several important metabolic processes including H₂ production, nitrogen fixation and photosynthesis (Zanetti and Pandini, 2013). Specifically, HydA1 and HydA2 catalyse the reversible reduction in protons into H₂ either using energy from light (biophotolysis), or by oxidizing organic compounds such as starch (dark fermentation). Biophotolysis, the process of splitting water into H₂ and O_2, is unique to green algae and cyanobacteria and is suggested to function as a transient electron sink for excess reductive energy when C. reinhardtii is transitioned from the dark to light under anoxic conditions. This rapid restoration of redox balance is essential for the onset of photosynthesis and CO₂ assimilation (Ghysels et al., 2013).

To successfully adopt C. reinhardtii H₂ production into industrial applications, several technical limitations must be overcome. A major limitation is the low efficiency of conversion of light to H₂ due to competition of the electrons with CO₂ fixation through the Calvin–Benson–Bassham (CBB) cycle. One of the highest reported efficiencies of light to H₂ conversion is only about 10% of the theoretical maximum (Kanygin et al., 2020). Second, the rapid inactivation of [FeFe] hydrogenases by O₂ is a challenge limiting H₂ production, especially when considering that O₂ is also generated during the biophotolysis process. Hence, H₂ production in C. reinhardtii is generally a short‐lived process lasting only a few minutes. The sustained production of H₂ can however be achieved over several days through the use of nutrient‐deprived media supplemented with an organic carbon source (Srirangan et al., 2011), or through implementation of tightly controlled dark‐light cycles (Kosourov et al., 2018). However, both these solutions are not ideal for commercial scale‐up processes as they are considered as non‐ideal growth conditions. To overcome these limitations, the design‐build‐test‐learn framework of synthetic biology can be applied to improving H₂ production by C. reinhardtii (Fig. 1).

Fig. 1.

Synthetic Biology Design Build Test Learn cycle for Improving H₂ production by C. reinhardtii. Knowledge of the biological systems involved in H₂ production by C. reinhardtii steers the design of synthetic modifications to H₂ production pathways. Synthetic constructs are built and screened for H₂ production, generating knowledge that can be used in future rounds of synthetic modification.

The increased availability of sequenced genomes, as well as the development of new genetic tools in Synthetic Biology, is greatly facilitating the engineering of microorganisms for robust, biotechnological applications (Salomé and Merchant, 2019). Synthetic Biology is an emergent technology that simplifies and modularizes natural systems into standardized parts with distinct functions and defined design rules. By rationally combining novel or heterologous parts, altering existing systems and undertaking iterative design‐build‐test‐learn cycles, the aim is to engineer new and better systems with specific outcomes and applications.

To date, many advances have been made in engineering microalgae such as C. reinhardtii for improved production of high‐value compounds and biofuels including H₂ (Gimpel et al., 2015; Jagadevan et al., 2018; Fabris et al., 2020; Ng et al., 2020; Wichmann et al., 2020). These advances include targeted recombinant protein expression in the chloroplast and nucleus using modularized toolkits (Lauersen et al., 2015); development of several inducible and constitutive native and synthetic promoters (Fischer and Rochaix, 2001; Fei and Deng, 2007; Scranton et al., 2016); editing tools such as precise gene editing through zinc‐finger‐nucleases technology and CRISPR/Cas9 technology (Shin et al., 2016; Greiner et al., 2017; Guzmán‐Zapata et al., 2019); and miRNA gene silencing for metabolic engineering (Crozet et al., 2018). Riboswitches, specific components of an mRNA molecule that regulate gene expression have also been used for metabolic engineering in the cyanobacterium Synechococcus elongatus and could be applied to C. reinhardtii (Nakahira et al., 2013). In addition, new HTS tools are being developed to screen large algal strain libraries for superior traits, including for H₂ production (Schrader et al., 2008; Stapleton and Swartz, 2010a, 2010b; Wecker and Ghirardi, 2014; Land et al., 2019). Typically, knowledge about the organism, its genome, gene regulations, metabolic and signalling pathways can be of great benefit for the targeted design of improved traits and phenotypes.

Therefore, this review aimed to identify Synthetic Biology targets for improving H₂ production in C. reinhardtii, by summarizing current knowledge of HydA1‐mediated H₂ production in C. reinhardtii holistically, including transcription, translation, transport, maturation and mechanistic limitations for H₂ production. We then highlight examples of how Synthetic Biology has been successfully used to alter and improve the native H₂ production system in C. reinhardtii and describe several available HTS assays that may help accelerate future screening of engineered C. reinhardtii for improved H₂ production.

Biogenesis of HydA1

Many processes are involved in the biogenesis of HydA1 in C. reinhardtii including transcription, translation, transport into the chloroplast and maturation of the active site. Understanding these processes is important to improve HydA1‐mediated H₂ production using Synthetic Biology tools and are described in the following sections. Most research studies to date have focused on HydA1 and less so on HydA2. Only where findings specifically involve HydA2, is it mentioned here.

The transcription and translation of HydA1 occur under copper or O₂ deficiency

As introduced earlier, C. reinhardtii has two hydrogenases, HydA1 and HydA2. Both belong to the [FeFe] A1 hydrogenase subclass and are localized in the chloroplast. HydA1 and HydA2 are encoded in the nucleus by HydA1 and HydA2, respectively, and are translated by cytosolic ribosomes (Happe and Naber, 1993). HydA2 shares 68% protein sequence identity to HydA1 (Forestier et al., 2003). Transcript levels of C. reinhardtii HydA1, HydEF and HydG are increased during anaerobiosis or copper deficiency and are regulated by the transcription factor copper response regulator I (CRR1) (Castruita et al., 2011), in concert with potentially other unidentified transcription factors (Pape et al., 2012). HydA1 transcripts have also been shown to increase by 20% when inducing anaerobiosis by culturing C. reinhardtii cells in S depleted media (Gonzalez‐Ballester et al., 2010). Generally, it was found that transcript levels of HydA1 reach a maximum 1.5–2 h after inducing anaerobiosis (Posewitz et al., 2004b; Mus et al., 2005), while functional HydA1 in chloroplasts was observed between 1 and 2 h post‐anaerobiosis (Happe et al., 1994).

Phototropic and autotrophic growth conditions affect the transcription of HydA1 and HydA2. When culturing C. reinhardtii autotrophically, upon inducing anaerobiosis, the transcripts of HydA1 and HydA2 reached the maximum levels, threefold slower compared with cultures grown heterotrophically (i.e. in presence of acetate) (Forestier et al., 2003). HydA1/HydA2 transcripts also degrade more rapidly in presence of O₂ when C. reinhardtii is grown autotrophically, that is without acetate supplementation (Forestier et al., 2003). These findings suggest that HydA1/HydA2 transcription is initiated more rapidly and that the transcripts remain intact longer when C. reinhardtii is cultured heterotrophically in presence of an additional carbon source such as acetate. Acetate addition has been suggested to facilitate higher starch accumulation concomitant with faster respiratory O₂ consumption rates, which results in a more rapid onset of anoxia (Fouchard et al., 2005). In support of this suggestion are the properties of a mutant of C. reinhardtii (sta7‐10 mutant) which is deficient in isoamylase activity, a starch hydrolysing enzyme. This mutant shows much lower HydA1/HydA2 transcript levels compared with wild‐type cells (Posewitz et al., 2004b). The HydA1/HydA2 transcripts’ turnover rates in this sta7‐10 mutant are also enhanced, resulting in shorter H₂ photoproduction cycles (1.5 h) than observed for wild‐type cells (> 10 h) under anoxic conditions (Posewitz et al., 2004b). The wild‐type phenotype could be restored in the sta7‐10 mutant strain through plasmid expression of the gene encoding the isoamylase enzyme. The results from this study support the hypothesis that supplementing the culture media with an additional carbon source is beneficial for extending the duration of H₂ production. The additional carbon source facilitates starch accumulation concomitant with the rapid occurrence of anoxia, and accumulation of HydA1/HydA2 transcripts with slower degradation rates.

While HydA1/HydA2 enzymes are irreversibly damaged by O₂, HydA1/HydA2 proteins are present in cells cultured aerobically under conditions where intracellular O₂ levels are quickly consumed due to high respiratory activities (Liran et al., 2016; Kosourov et al., 2018). However, HydA1/HydA2 protein levels typically increase with decreasing O₂ levels (Kosourov et al., 2018).

The import and maturation of HydA1 and HydA2

To be functionally active, HydA1 and HydA2 proteins first need to be imported into the chloroplast and matured. Following translation of HydA1 and HydA2 transcripts, the resulting pre‐proteins are imported into the chloroplast stroma using the general import machinery that spans the chloroplast outer and inner membranes. This import machinery, known as the TOC‐TIC import machinery (TOC for translocon at the outer envelope membrane of chloroplasts and TIC for translocon at the inner envelope membrane of chloroplasts), is a common feature in both plants and green algae and functions to import nuclear‐encoded plastid proteins into the chloroplast (Bölter, 2018). An N‐terminal chloroplast targeting sequence of 20–60 amino acids is required on the cytosolic protein to be imported by the TOC‐TIC complex. The import targeting peptide becomes cleaved by a peptidyl processing protein protease after import into the chloroplast. Once imported, the HydA1/HydA2 proteins require maturation in the chloroplast to become active (Sawyer et al., 2017). HydA1/HydA2 proteins are only functional when matured to their catalytically active form and under anoxic conditions (Kosourov et al., 2018).

The maturation of HydA1 occurs through a stepwise biosynthesis and insertion of a H‐cluster cofactor to form the active site of the enzyme (Britt et al., 2020). The H‐cluster is a characteristic feature of [FeFe] hydrogenases and consists of two Fe‐S clusters, a 4Fe‐4S cubane subcluster [4Fe‐4S]_H and a 2Fe subcluster [2Fe]_H (Fig. 2) (Mulder et al., 2010; Esselborn et al., 2016). A cysteine S atom links [2Fe]_H and [4Fe‐4S]_H together in the protein, constituting the full H‐cluster active site (Britt et al., 2020).

Fig. 2.

Active site H‐cluster of [FeFe] hydrogenases. The H‐cluster constitutes two subclusters connected via a cysteine S: (i) [4Fe‐4S]_H (green), a classical 4Fe‐4S cubane; and (ii) [2Fe]_H (yellow), a unique di‐iron Fe‐S cluster coordinated with CO and CN ligands, and CO and azadithiolate bridges.

The [2Fe]_H subcluster is a unique di‐iron cluster found only in [FeFe] hydrogenases and is thought to be the active site of [FeFe] hydrogenases where H⁺ or H₂ substrates bind (Mulder et al., 2017; Pelmenschikov et al., 2017; Reijerse et al., 2017; Land et al., 2020). Both Fe atoms of the [2Fe]_H subcluster have terminal CO and CN ligands, and the Fe atoms are further bridged by CO and azadithiolate ligands. The [2Fe]_H subcluster is biosynthesized and inserted into HydA1 via the enzymes HydEF and HydG (Posewitz et al., 2004a).

The [4Fe‐4S]_H subcluster balances the redox state of the enzyme and is required for hydrogenase activity (Rodríguez‐Macia et al., 2020). Four cysteine residues anchor the [4Fe‐4S]_H subcluster in [FeFe] hydrogenases (C115, C170, C362 and C366 in HydA1) (Kertess et al., 2017). The housekeeping SUF Fe‐S cluster assembly machinery synthesizes and inserts [4Fe‐4S]_H into the apo form of HydA1 followed by insertion of the [2Fe]_H subcluster (Godman and Balk, 2008; Bai et al., 2018). The order of insertion is supported by the requirement of [4Fe‐4S]_H to be present for [2Fe]_H to produce a functional enzyme (Mulder et al., 2009).

During O₂ inactivation of [FeFe] hydrogenases, the distal Fe of [2Fe]_H, Fe_d, is the first site to be ‘attacked’ and degraded by O₂. In contrast, the [4Fe‐4S]_H cluster can remain stable in the presence of up to 17% O₂. If the [4Fe‐4S]_H subcluster remains intact, the HydEF and HydG proteins can once again mature the [2Fe]_H subcluster and the hydrogenase into its active state upon returning to hypoxic conditions (Swanson et al., 2015).

Biosynthesis of [4Fe‐4S]_H

Green algae and plants have three different systems for assembly of 4Fe‐4S clusters – (i) SUF, (ii) cytosolic Fe‐S cluster assembly (CIA) and (iii) Fe‐S cluster assembly (ISC) systems (Gomez‐Casati et al., 2021). Each of these 4Fe‐4S cluster assembly systems operate in different parts of the cell. The SUF system in algae is localized in the chloroplast, the CIA system is in the cytosol, and the ISC system is in the mitochondria (Gomez‐Casati et al., 2021). Interestingly, the CIA system requires a functional ISC to work suggesting co‐dependence between these two systems (Godman and Balk, 2008; Xu and Møller, 2011).

The synthesis and insertion of the H‐cluster into HydA1 occurs after import into the chloroplast stroma, since folded or partially folded proteins cannot be transported across the chloroplast membranes by the TOC‐TIC translocon (Paila et al., 2015; Sawyer et al., 2017). As such, the SUF system, and not CIA or ISC, is responsible for the assembly and insertion of [4Fe‐4S]_H into HydA1 and functions under O₂ and iron limitation conditions (Takahashi and Tokumoto, 2002; Pérard and Choudens, 2018).

The SUF‐mediated 4Fe‐4S cluster biosynthesis is poorly characterized in C. reinhardtii; however, it is well characterized in E. coli (Fig. 3) (Blahut et al., 2020). C. reinhardtii produces 6 orthologues of the E. coli system: SufA, SufB, SufC, SufD, SufE and SufS (Godman and Balk, 2008). Sequence homology between the C. reinhardtii and E. coli SUF systems suggests they function similarly to produce 4Fe‐4S clusters, although there is little direct experimental evidence to support this (Bai et al., 2018). In E. coli, assembly of the Fe‐S complex and transfer to the target protein is initiated by the cysteine desulfurase SufS and the S shuttling protein SufE forming the SufSE complex (Fig. 3) (Bai et al., 2018; Pérard and Choudens, 2018; Blahut et al., 2020). This SufSE complex then extracts S from cysteine and transfers it to a SufBC₂D scaffold complex where the 4Fe‐4S cluster is assembled (Fig. 3) (Bai et al., 2018; Pérard and Choudens, 2018; Blahut al., 2020). Formation of the SufSE complex increases the cysteine desulfurase activity of SufS by 10‐ to 50‐fold compared with free SufS. This increased activity has also been shown to be dependent on the sulfurization state of SufSE (Dai and Outten, 2012; Selbach et al., 2013). Furthermore, SufBC₂D in the presence of SufSE further increases the desulfurase activity of SufS 20‐ to 30‐fold (Outten et al., 2003; Dai and Outten, 2012). Evidence suggests that S from SufSE, electrons from FADH₂ and Fe from an unknown source are used to build the 4Fe‐4S cluster at the SufB‐SufD interface of SufBC₂D (Pérard and Choudens, 2018; Blahut et al., 2020). The synthesized 4Fe‐4S cluster is then transferred to the carrier protein SufA which finally shuttles the assembled cluster to the target protein (Fig. 3) (Blahut et al., 2020).

Fig. 3.

Schematic Illustration of the SUF‐mediated biosynthesis of [4Fe‐4S] clusters in E. coli.

A. SufS binds SufE to form the SufSE complex, increasing the desulfurase activity of SufS up to 50‐fold.

B. Sulfur is acquired from L‐cysteine by the SufSE complex. The SufSE complex then binds with the scaffold complex SufBC₂D, further increasing desulfurase activity by another 20‐ to 30‐fold. Sulfur acquired from cysteine and iron from an unknown source is used to build the [4Fe‐4S] cluster at the interface of SufB and SufD.

C. The complete [4Fe‐4S] cluster (shown as a cubane molecule) is transferred to SufA.

D. SufA transports [4Fe‐4S] to the target protein.

Biosynthesis of [2Fe]_H

In C. reinhardtii, HydA1 containing a [4Fe‐4S]_H cluster acts as the site to which the [2Fe]_H subcluster can bind to form the full H‐cluster (Fig. 2). The formation and assembly of this [2Fe]_H subcluster in [FeFe] hydrogenases require the activity from three maturase enzymes: the two radical S‐adenosyl‐L‐methionine (rSAM) enzymes HydE and HydG, and a guanosine triphosphatase (GTPase) enzyme, HydF. The details of the different stages of the biosynthesis of the [FeFe] hydrogenase 2Fe subcluster by these three enzymes, HydE, HydF and HydG, have been described in a recent review and are summarized in Fig. 4 (Britt et al., 2020).

Fig. 4.

Biosynthesis of the [2Fe]_H subcluster.

A. HydG catalyses two reactions to form the Fe (II) intermediate species. Two tyrosine molecules are cleaved by [4Fe‐4S] into CO and CN ligands. The auxiliary Fe of [5Fe‐5S] is chelated by cysteine and coordinated by the CO and CN ligands to complete the intermediate.

B. The Fe(II) centre of the intermediate is reduced to Fe(I) by HydE. Subsequent reactions form the [2Fe]_H precursor through an unknown mechanism; however, it has been suggested that the transformation occurs through pairwise condensation of the Fe(I) intermediate.

C. The [2Fe]_H precursor is transformed into the final [2Fe]_H subcluster by forming the azadithiolate and CO bridges. The mechanisms by which the bridges form are unclear; however, two serine molecules are involved in formation of the azadithiolate bridge with HydF possibly playing a role, and it has been suggested a terminal CO ligand on the [2Fe]_H precursor is reconfigured into the CO bridge. The complete [2Fe]_H subcluster is transferred to HydA1 by HydF.

The first step in the biosynthesis of the [2Fe]_H subcluster is carried out by HydG. HydG is a free radical SAM enzyme that has two different Fe‐S clusters sitting at each end of a conserved triosephosphate isomerase (TIM) barrel channel. These two Fe‐S clusters carry out separate reactions to produce an Fe (II) intermediate species. Specifically, the Fe‐S cluster at the N‐terminus of the TIM barrel channel is a [4Fe‐4S] cluster and cleaves tyrosine to produce both the CO (Shepard et al., 2010) and CN ligands (Driesener et al., 2010). The Fe‐S cluster at the C‐terminal of the TIM barrel channel is a unique [5Fe‐5S] cluster, having a classical [4Fe‐4S] cubane structure linked to an auxiliary Fe via a bridging sulphide. Evidence points towards the auxiliary Fe of [5Fe‐5S] acting as a receiver for coordination of the CO and CN ligands derived from the cleaved tyrosine (Dinis et al., 2015). A second CO ligand, from a subsequent tyrosine cleavage, forms another bond with the auxiliary Fe, while the second CN coordinates with the cubane to form a [4Fe‐4S‐CN] complex, resulting in release of the Fe(II) intermediate species as shown in Fig. 4A (Suess et al., 2015). This Fe(II) intermediate species made by HydG then binds with 5’ deoxyadenosine of HydE, reducing the Fe(II) centre to Fe(I) as shown Fig. 4B (centre of figure) (Tao et al., 2020). The Fe(I) intermediate species undergoes further reactions to eventually yield the [2Fe]_H precursor, conceivably through pairwise condensation of the Fe(I) intermediate (Tao et al., 2020). It is currently unclear whether the Fe(I) intermediate condenses into the [2Fe]_H precursor within HydE, as shown Fig. 4B (top of figure), or if the intermediate is transferred to HydF prior to the formation of this precursor (Britt et al., 2020). Whether the formation of the CO and azadithiolate bridges are mediated by HydE or HydF remains unresolved. The addition of two serine molecules to form the azadithiolate bridge, Fig 3C (top), occurs by an unknown mechanism (Rao et al., 2020). It has been suggested that one of the terminal CO ligands of the 2Fe precursor is reconfigured into the CO bridge. Whether or not HydF is involved in these final stages of [2Fe]_H cluster biosynthesis is unclear. Evidence points towards HydF being involved with transfer of the complete [2Fe]_H to [4Fe4S]_H of HydA1 in the final stages of assembly of the H‐cluster and hydrogenase maturation (Berggren et al., 2013; Albertini et al., 2015).

Hydrogen production pathways in C. reinhardtii

There are three pathways through which HydA1 produces H₂ in C. reinhardtii. These are photosystem (PS) II‐dependent H₂ photoproduction, PS II‐independent H₂ photoproduction and H₂ production during dark anaerobic fermentation (Fig. 5). In each of these three pathways, electrons are donated to HydA1 through a reduced ferredoxin (Fd) species. There are eight Fd isoforms that have been shown or predicted to be expressed by the C. reinhardtii chloroplast (PetF, FDX2‐3 and FDX5‐9) (Sawyer and Winkler, 2017). The particular isoform that donates electrons to HydA1 is still up for debate, although PetF is hypothesized to fill this role. However, a recent transcriptome study has implicated FDX9 as the prime electron carrier as it is the only chloroplastic Fd isoform with an expression pattern that matches the expression of HydA1 and HydA2 over a 24‐hr period during natural light/dark cycles (Strenkert et al., 2019). For clarity, we will refer to the electron donor in the H₂ production pathway herein to be PetF as it is currently accepted as the isoform that donates electrons to HydA1.

Fig. 5.

H₂ production pathways in C. reinhardtii. Schematic illustration showing the photosystem (PS) II‐dependent pathway (orange arrows), the PS II‐independent pathway (yellow arrows) and the dark fermentation pathway (purple arrows). In all H₂ production pathways, reduced ferredoxin (PetF) supplies electrons to HydA1, which in turn catalyses the reduction in protons into H₂. The photosynthetic electron transport chain (PETC) occurs as linear electron flow (LEF) (orange and yellow arrows) or circular electron flow (yellow arrows, dashed). LHC, light‐harvesting complex; Cytb₆f, cytochrome b6f complex; PC, plastocyanin; FNR, ferredoxin reductase; FQR, ferredoxin‐plastoquinone reductase; NDH‐2, NAD(P)H‐PQ oxidoreductase; PFR1, Pyruvate:Ferredoxin Oxidoreductase. Further details and references are described in the main text.

Illuminating C. reinhardtii cells can activate either PS II‐dependent or PS II‐independent H₂ production (Chochois et al., 2009). Hydrogen production by the PS II‐dependent pathway occurs post‐illumination when PS II is functioning normally. However, H₂ production via this pathway is a short‐lived process, lasting only 6 min (Ben‐Zvi et al., 2019). This time frame of H₂ production is rather impractical for commercial applications. The second pathway, the PS II‐independent pathway, can be activated upon illumination by partly inhibiting PS II through nutrient deprivation, addition of the PS II specific inhibitor (DCMU), or genetically through modifying PS II gene expression. PS II‐independent H₂ photoproduction has been shown to significantly prolong H₂ production for up to 200 h. However, inducing the PS II‐independent route for H₂ production through nutrient deprivation limits cell growth (Chochois et al., 2009). The third H₂ production pathway, which is activated through dark anaerobic fermentation, can be sustained for over 6 hours (Kanygin et al., 2020). These three pathways are described in more detail in the following sections.

PS II‐dependent photoproduction of hydrogen

During PS II‐dependent photoproduction of H₂, PS II absorbs photons via the light‐harvesting complex (LHC) II antenna and splits water into protons, electrons and O₂ (Fig. 5, orange arrows). In general, the electrons are shuffled in a linear electron flow (LEF) across the photosynthetic electron transport chain (PETC) via three pigment‐proteins PS II, the b6/f cytochrome complex (Cytb₆f) and PS I. Electron transport between these three pigment‐proteins is mediated by two electron carriers plastoquinone (PQ) and plastocyanin (PC). Both these proteins can freely diffuse through the lipid layers. Electrons leaving PS I are used to reduce PetF which then distributes the electrons to either NAD(P)H ferredoxin reductase (FNR) (the primary electron acceptor) or HydA1 (the secondary electron acceptor) (Hemschemeier and Happe, 2011; Sawyer and Winkler, 2017). FNR catalyses the reduction in NADP⁺ into NADPH for use in the CBB carbon fixation cycle (Shin, 2004).

Hydrogen production via the PS II‐dependent pathway has been shown to occur upon illuminating anoxic C. reinhardtii cells after they have been placed in the dark for a minimum of 30 min (Happe et al., 1994). Hydrogen production typically begins at the onset of illumination and rates of H₂ evolution increase for ~ 2 min before steadily declining and ceasing at ~ 6‐min post‐illumination (Ben‑Zvi et al., 2019). These changes in rates of H₂ production resemble changes from LEF to cyclic electron flow (CEF) with CEF activity reaching a maximum at ~ 2 min post‐illumination, and the activation of the CBB cycle at ~ 2–4 min (Godaux et al., 2015). When CEF occurs, electrons are cycling through PS I by channelling electrons from PetF back towards the PQ pool instead to FNR or HydA1 (Fig. 5, yellow, dashed arrow). These findings suggest PS II‐dependent H₂ production is driven by PS I‐CEF until the CBB cycle becomes active and the electron flux is entirely directed towards carbon assimilation.

By producing H₂, HydA1 also removes protons from the stroma and thereby establishes a proton gradient for ATP synthesis by ATPase which is required for carbon assimilation (Fig. 5). Removing protons also results in the stroma becoming alkalized (pH 6.5 increases to pH 8.5), which is necessary for RuBisCO activity, the key enzyme of the CBB cycle. A pH change closely resembling the stromal pH change before and after illumination (pH 7 to pH 9) enhanced the HydA1‐PetF rate constant for binding threefold in silico (Diakonova et al., 2016).

One of the major factors contributing to the cessation of H₂ production 6‐min post‐illumination in the PS II‐dependent pathway has been hypothesized to be the inactivation of HydA1 by O₂ (Ghirardi, 2015). However, recent evidence supports an alternative hypothesis where HydA1 activity ceases due to electron competition with FNR. For example, it was observed that electron flow is channelled exclusively through FNR for carbon fixation before the irreversible inactivation of HydA1 by O₂ occurs (Milrad et al., 2018; Ben‑Zvi et al., 2019). This alternative hypothesis is further supported by the observation that HydA1 can remain active in aerobically grown cultures as there are high rates of local respiratory O₂ consumption, creating microoxic niches at the stromal interface of the thylakoid membrane where HydA1 is typically located (Liran et al., 2016). In these aerobic cultures, HydA1 was shown to be a minor sink for electrons leaving the PETC, particularly for reductants produced upon the transition from low to high light.

As intermittent H₂ production is problematic for scaled production, there has been ongoing efforts to prolong H₂ production via the PS II‐dependent route. By exposing cultures to a cycle of very short (1 s) pulses of light followed by dark intervals (9 s), cultures can maintain low O₂ levels and H₂ production can be prolonged for up to three days (Kosourov et al., 2018). Under such controlled light and dark cycling conditions, the maximum achieved reported rate of H₂ production was 25 μmol H₂ mg⁻¹ Chl h⁻¹. However, given that chlorophyll concentration varies with cell density and growth conditions, these units of H₂ production are not particularly useful for comparing with other biological H₂ production rates in the literature which are typically presented as mmol H₂ l⁻¹ h⁻¹ (where l represents the cell culture volume). In this review, we have therefore recalculated and presented all H₂ production units as H₂ lL⁻¹ h⁻¹ using reported cell densities (where available) or using a medium cell density of at 1 × 10⁷ cells per ml, and the reported chlorophyll amounts per cell of 2.4–3.5 × 10^‐15 mol (Polle et al., 2000). Thus, 25 μmol H₂ mg⁻¹ Chl h⁻¹ as reported above is equivalent to 0.37–0.54 mmol H₂ l⁻¹ h⁻¹.

PS II‐independent photoproduction of hydrogen

Hydrogen can also be produced via a PS II‐independent pathway through partial inhibition of PS II (Fig. 5, yellow arrows). In this pathway, reduction in PQ occurs from electrons derived through oxidation of starch reserves and residual PS II activity. The non‐photochemical reduction in the PQ pool by NAD(P)H is suggested to be mediated either through ferredoxin:plastoquinone oxidoreductase (FQR) or a type‐II NADH dehydrogenase NAD(P)H‐PQ oxidoreductase (NDH‐2) (Cournac et al., 2002; Mus et al., 2005; Takabayashi et al., 2005; Antal et al., 2009). As per the PS II‐dependent pathway, the reduced PQ pool then supplies electrons directly through Cytb₆f to PS I. PS I then reduces PetF, which can feed electrons back to the PQ pool via CEF through FQR (Fig. 5, yellow, dashed arrows), or directly to HydA1 to generate H₂ (Fig. 5, yellow arrows) (Antal et al., 2009). Competition for electrons from PetF between HydA1 and CEF was shown to slow H₂ production more than twofold (Antal et al., 2009).

PS II‐independent H₂ photoproduction in C. reinhardtii can be achieved by starving cells of S (Melis et al., 2000) or phosphorus (P) (Batyrova et al., 2012). It can also be achieved via magnesium (Mg) depletion, a key ion of chlorophyll, which is a main component of the two light‐harvesting complexes LHC II and LHC I (Volgusheva et al., 2015). All these nutrient depletion treatments result in impaired PS II assembly or repair and thus reduced PS II activity. The implications of a reduction in the PS II activity are both decreased O₂ evolution rates and carbon assimilation, which in turn leads to respiration and increased cellular stress levels.

There are four distinct cellular stages for cells that are nutrient‐deprived and continuously illuminated. These are (i) O₂ increase and starch accumulation; (ii) decrease in O₂ levels; (iii) onset of anaerobiosis accompanied by starch degradation; and (iv) H₂ production (Nagy et al., 2018). While starch accumulation during stage 1 and subsequent starch catabolism during stage 3 both contribute to H₂ production via the PS II‐independent pathway, more than 92% of the electrons for H₂ production under nutrient‐deprived conditions derive from light‐driven water splitting as a result of residual PS II activity (Kosourov et al., 2020).

Under nutrient limitation, H₂ production periods have been observed to last for up to 96 h for S‐deprived cells (Nagy et al., 2018), 150 h for P‐deprived cells (Batyrova et al., 2012) and 200 h for Mg‐deprived cells (Volgusheva et al., 2015). The maximum rates of H₂ production for these different conditions are 0.14–0.20 M h⁻¹ under S‐depravation (Kosourov et al., 2003), ~ 0.05 M⁻¹ h⁻¹ under P‐depravation (at ~ 5.7 × 10^‐6 cells) (Batyrova et al., 2012) and 0.25–0.37 M h⁻¹ under Mg‐depravation (Volgusheva et al., 2015). While these rates are lower than the 0.37–0.54 mmol H₂ l⁻¹ h⁻¹ observed in the PS II‐dependent pathway (Kosourov et al., 2018), the extended period of H₂ production increasing from a matter of minutes to over a week is an attractive quality for industrial biohydrogen production.

Dark fermentation production of hydrogen

A final pathway for H₂ production in C. reinhardtii is via dark fermentation (Fig. 5, purple arrows). Production of H₂ by dark fermentation is suggested to originate from oxidation of starch or other carbon sources and reduction in PetF by pyruvate ferredoxin oxidoreductase (PFR1) (Noth et al., 2013), or the accumulation of NADPH from other catabolic processes causing FNR to operate in reverse, reducing Fd (Petrova et al., 2020). Hydrogen production via dark fermentation is limited by the accumulation of fermentation products, notably acetic acid, that can impair H₂ production and cell growth, and compete with other fermentation pathways in the cell (Fakhimi et al., 2020). The rate of H₂ production by C. reinhardtii HydA1 during dark fermentation was reported to be nearly 50 times lower (~ 0.009–0.013 mmol H₂ l⁻¹ h⁻¹) (Kanygin et al., 2020) compared with PS II‐independent H₂ photoproduction in Mg‐deprived C. reinhardtii (~ 0.25–0.37 mmol H₂ l⁻¹ h⁻¹) (Volgusheva et al., 2015). Furthermore, H₂ production dark fermentation can be sustained for over 6 h (Kanygin et al., 2020), significantly longer than the 6‐min H₂ production period of PS II‐dependent photoproduction (Ben‑Zvi et al., 2019), although much shorter than the 200 h H₂ production period of PS II‐independent photoproduction (Volgusheva et al., 2015).

Strategies to improve H₂ production in C. reinhardtii using Synthetic Biology

A promising strategy for producing H₂ at industrial scale is by extending the C. reinhardtii H₂ production period through induction of the PS II‐independent pathway. As such, efforts to induce the PS II‐independent pathway using simpler methods than nutrient deprivation or expensive chemical inhibitors have been attempted. In addition to increasing the length of the H₂ production period, the efficiency of H₂ production needs to be improved in order to achieve the successful scaled production of H₂ in C. reinhardtii. Light‐to‐H₂ conversion efficiencies for C. reinhardtii are reported to be below 2% which is approximately 10% of the theoretical maximum (Kanygin et al., 2020). Major factors contributing to the low H₂ production efficiency in C. reinhardtii include synthesis and maturation of active HydA1, O₂ sensitivity of HydA1 as well as competition for electrons between HydA1, FNR and CEF. In this next section, we discuss these limitations and how Synthetic Biology has helped to overcome them.

Genetic Engineering to alter H₂ production pathways

Synthetic Biology has been used to improve PS II‐independent H₂ photoproduction in C. reinhardtii. In a synthetic C. reinhardtii strain, cyc6Nac2.49, expression of PS II was made inducible by fusing the Nac2 gene to the cyc6 promoter, which is induced under anaerobic conditions or copper deficiency (Surzycki et al., 2007). Nac2 is key for PS II expression as it binds to and stabilizes psbD mRNA, which translates to the PS II D2 protein. Addition of copper to cyc6Nac2.49 cultures was shown to reduce expression of Nac2 and diminished PS II D2 protein levels 8 h post‐addition. Undetectable levels of PS II D2 coincided with reduction in PS II activity and reduced O₂ evolution rates, and resulted in H₂ production rates of up to 0.18–0.27 mmol H₂ l⁻¹ h⁻¹. However, H₂ production was sustained for only a short period of time (1.5 h) as PS II protein levels eventually recovered due to the onset of anaerobiosis, which restores expression of PS II D2 protein. The duration of H₂ production achieved here is 64‐fold shorter than the 96 h reported for S‐deprived cultures (Nagy et al., 2018). By fusing other genes to the cyc6 promoter and controlling O₂ or copper levels in the media, this inducible chloroplast gene expression system might be applied for other gene targets C. reinhardtii (Surzycki et al., 2007).

In a follow‐up study, this same strain (cy6Nac2.49) was then tested for H₂ production during multiple dark‐low‐light cycles (Batyrova and Hallenbeck, 2017). The authors demonstrated that during illumination, the typical steep rise of O₂ in the culture was diminished due to O₂‐induced downregulation of the engineered Nac2‐PsbD system and diminished PS II expression. This resulted in a 10‐fold lower O₂ accumulation overall than in the parental WT strain. As a consequence, increased H₂ production rates were observed, particularly under low light (10 W m^‐2). Through cycling of dark–light conditions, H₂ production could be sustained for over 30 h. While the rates were 4.5 higher than for the parental WT strain under these conditions, the resulting rates of the mutant strain (0.038–0.056 mmol H₂ l⁻¹ h⁻¹) (Batyrova and Hallenbeck, 2017) are much lower compared with wild‐type C. reinhardtii grown in S‐deprived media (0.14–0.20 M⁻¹ h⁻¹) (Kosourov et al., 2003) and Mg‐deprived media (0.25–0.37 M ⁻¹ h⁻¹) (Volgusheva et al., 2015). In addition, the mutant strain (cy6Nac2.49) showed poor cell growth during photoheterotrophic conditions as starch did not accumulate in this anaerobically inducible PS II strain. Nevertheless, because this is a powerful example of sustaining H₂ production through an O₂ and copper‐mediated genetic circuit, this strain might be an interesting target for further strain engineering (Batyrova and Hallenbeck, 2017).

In addition to degrading or inhibiting the function of PS II, the PS II‐independent pathway can also be further improved to sustain H₂ production for longer periods of time by other methods. Approaches that do not degrade PS II are beneficial as long‐term loss of PS II function inhibits growth of cells. As mentioned previously, H₂ production via the PS II‐independent pathway competes for electrons with PS I‐mediated CEF (Cournac et al., 2002; Antal et al., 2009; Ghysels et al., 2013). Therefore, an effective strategy to increase PS II‐independent photoproduction is by targeting the disruption of CEF. For example, treating cells with inhibitors of ferredoxin‐plastoquinone reductase (FQR), a key mediator of CEF, was shown to increase PS II‐independent H₂ photoproduction in S‐starved cells, twofold, compared with untreated cells (Antal et al., 2009).

Alternatively, increased H₂ production rates were also achieved by disrupting CEF using Synthetic Biology techniques. For instance, a randomly generated mutant library of C. reinhardtii was first screened for mutants in which the LHC is locked in state 1 to favour LEF over CEF using fluorescence screening and spectroscopy (Kruse et al., 2005). Shortlisted mutants in which CEF was disrupted were then screened for enhanced H₂ production. This study has resulted in identification of a new mutant strain Stm6 with H₂ production rates 5–13 times higher than that of the wild‐type strain over a range of conditions including light intensity and culturing time (Kruse et al., 2005).

Disruption of the CEF pathway and improved H₂ production has also been achieved by generating a C. reinhardtii knockout (KO) mutant (pgrl1) incapable of producing the Proton Gradient Regulation‐Like 1 (PGRL1) (Tolleter et al., 2011). PGRL1 is required for CEF in C. reinhardtii and was suggested to act as alternative FQR, a key protein for CEF (Fig 4) (Hertle et al., 2013), and to interact with the proton gradient regulation 5 (PGR5) (Tolleter et al., 2011). In this pgrl1 KO mutant, an approximately fourfold increase in the short‐term photoproduction rate and yield of H₂ during high‐light exposure, compared with the wild‐type strain, was reported (Tolleter et al., 2011). A similar improvement in H₂ photoproduction by this mutant of nearly fourfold compared with the wild type was observed under S‐deprivation over 6 days.

The duration and thus yield of H₂ production via the PS II‐independent pathway are strongly controlled by the electrochemical gradient across the thylakoid membranes which restricts electron flow towards HydA1 (Cournac et al., 2002). This has been shown by the addition of an oxidative phosphorylation uncoupler (carbonyl cyanide‐4‐(trifluoromethoxy)phenylhydrazone (FCCP) (Cournac et al., 2002) or deletion of pgr5 (Johnson et al., 2014) and/or pgrl1 (Steinbeck et al., 2015) which were observed to diminish the thylakoid proton gradient and uncoupling of ATP synthesis from the membrane proton transport system. For example, when the chemical inhibitor FCCP is used, H₂ production yields were shown to increase 10‐fold in a PS II deficient C. reinhardtii mutant (Cournac et al., 2002), and by 37% in S‐starved C. reinhardtii (Antal et al., 2009) due to prolonged H₂ production. Similarly, the pgr5 and pgrl1 single mutant and a pgr5/ pgrl1 double C. reinhardtii mutants were reported to generate one of the highest H₂ production yields (610–850 ml H₂/L culture) (Steinbeck et al., 2015). The sustained H₂ production in these mutants was suggested to be a result of greater residual PS II activity than in WT cells, while the oxygen‐sensitive hydrogenase was protected from O₂ for extended periods through increased rates of respiration (O₂ consuming) (Steinbeck et al., 2015). Engineering cells to uncouple ATP synthesis from the membrane proton transport system therefore seems to be a promising strategy to mimic the effects of FCCP and increase H₂ production in C. reinhardtii.

Disruption of the CEF pathway to increase H₂ production has also been done by generating a PS I‐HydA2 fusion protein in a C. reinhardtii ∆HydA1, ∆HydA2 double knockout mutant strain (Kanygin et al., 2020). The authors intended to direct electrons towards HydA2 and H₂ production by eliminating competition of electrons with FNR. This was done by genetically fusing HydA2 to the PS I Fe‐S centre C (PsaC), the terminal electron acceptor subunit of PS I. As a result, HydA2 intercepts electrons directly from PS I for H₂ production, which lasted up to 5 days. This strategy was extremely successful in efficiently mediating the electron flow directly between PS I and HydA2, and as such, bypassing both the CEF and FNR pathways. It was also successful in decreasing the O₂ evolution rates from PS II below the respirational O₂ consumption rates. The PsaC‐HydA2 chimera showed a sevenfold increase in the rate of H₂ production at high‐light intensity (2000 µmol photons m^‐2 s⁻¹) compared with the WT strain which saturated at 100 µmol photons m^‐2 s⁻¹. Such high‐light intensities are usually detrimental to H₂ photoproduction as the O₂ evolution rates by PS II also increase under these conditions (Batyrova and Hallenbeck, 2017). In addition to the disruption of CEF, the CBB cycle was significantly downregulated. However, the ability of PetF to intercept electrons away from HydA2 remained. Overall, the efficiency of the conversion of light to H₂ was reported as 1.75%. This efficiency is only about 10% of the theoretical maximum (Kanygin et al., 2020) and over twofold short of the 5% efficiency target at which algal H₂ production is stated to be economically viable (Kruse et al., 2005). Further engineering the PetF binding site of HydA2 and O₂ sensitivity was suggested to be an important next step to further improve H₂ production.

Optimizing the stable expression of a functional HydA1

Controlling the synthesis and maturation of HydA1 to optimize its protein yields and stability in C. reinhardtii is another important target for improving H₂ production. A strategy to regulate the synthesis and maturation of HydA1 is to express HydA1 from the chloroplast genome under controllable conditions. The presence of homologous recombination in C. reinhardtii chloroplasts makes this strategy more feasible than genetic engineering of the nucleus, where non‐homologous DNA end ‐joining dominates. Additionally, expression of HydA1 in the chloroplast overcomes the need for protein import and the regulatory elements involved.

Using the native chloroplast homologous recombination system, HydA1 was integrated into the chloroplast genome of C. reinhardtii (Reifschneider‐Wegner et al., 2014). Integration of the synthetic gene design into the chloroplast genome required extremely tight regulation of HydA1 expression using a double repressor system induced by vitamin B12 and thiamine pyrophosphate (TPP) due to a strong selective pressure against the synthetic HydA1. As a consequence, stable integration into the chloroplast genome was only achieved when culturing the cells in the dark making H₂ photoproduction unfeasible.

Nevertheless, engineered strains were able to produce HydA1 transcripts and apoprotein under both aerobic and anaerobic conditions and increased H₂ production yields during dark fermentation twofold when compared to its parent strain (Reifschneider‐Wegner et al., 2014). Therefore, this strain might be a good starting point for further strain engineering, to allow for inducible H₂ production through the more efficient PS II pathway and PS II‐independent pathway.

Heterologous and engineered hydrogenases with improved O₂ tolerance

Native C. reinhardtii HydEF and HydG maturases in conjunction with native [4Fe‐4S] cluster machinery was demonstrated to be capable of maturing heterologous [FeFe] hydrogenases such as CpI from C. pasteurianum (Sawyer et al., 2017). The capacity of HydEF and HydG to mature heterologous [FeFe] hydrogenases in this study was suggested to be due to the homology of the H‐cluster between [FeFe] hydrogenases.

As mentioned earlier, a major limitation for photoproduction of H₂ in C. reinhardtii is the O₂ sensitivity of the [FeFe] hydrogenases (Ghirardi, 2015). To improve the tolerance of C. reinhardtii to O₂, expression of heterologous hydrogenases in C. reinhardtii has been explored to find more O₂ tolerant enzyme variants. For example, CpI expressed heterologously from the C. reinhardtii chloroplast demonstrated higher O₂ tolerance than HydA1/HydA2 (Noone et al., 2017). Additionally, H₂ production in C. reinhardtii using a double point mutated variant of CpI (CpI^T356V/S357T) (Koo and Swartz, 2018) showed improved O₂ tolerance, retaining 85% of initial activity after exposure to O₂ compared to 65% for the wild type (Elman et al., 2020).

More recently, another clostridial [FeFe] hydrogenase, CbA5H from C. beijerinckii, was shown to be O₂ tolerant (Winkler et al., 2021). CbA5h is the only [FeFe] hydrogenase shown to protect the H‐cluster from destruction by long‐term O₂ exposure. When CbA5h is exposed to air, the H‐cluster changes from an active state to an inactive O₂ protected state (H_inact). The formation of H_inact is dependent on the conserved cysteine residue C367, which is thought to act as a safety cap for the H‐cluster preventing O₂ attack. In the H_inact state, residue C367 is in close proximity (3.1 Å) to the distal Fe atom, Fe_d, of the 2Fe cluster, which is the site of degradation of the H‐cluster by O₂ in other [FeFe] hydrogenases (Swanson et al., 2015). The C367‐Fe_d distance is only slightly longer than the average distance for covalent Fe‐S bonds (2.4 Å) and is small enough to stop small molecules including H₂ and O₂ from inserting between Fe_d and C367. By comparison, the distance between Fe_d and the corresponding cysteine residue (C299) of CpI from C. pasteurianum is significantly longer (5.1 Å) accounting for the inability of CpI to enter the H_inact state. Residues L364, A561 and P386, which are in close proximity to the H‐cluster in CbA5H, also contribute to increased O₂ resistance of CbA5h. The mechanisms of O₂ tolerance in CbA5H are similar to [NiFe] hydrogenases, where two conserved cysteine motifs in close proximity to the proximal Fe‐S cluster confer O₂ tolerance to the hydrogenase (Lukey et al., 2011).

Diverting electron flow to HydA1

During illumination and photosynthesis by PS II, FNR is the primary electron acceptor of the PETC transferring electrons from reduced PetF to NADP⁺ generating NADPH for carbon fixation (Shin, 2004). While photosynthesis is concomitant with O₂ evolution, which itself inhibits HydA1 activity, it was suggested that HydA1 is outcompeted for electrons from PetF by FNR prior to the inactivation of HydA1 by O₂ (Milrad et al., 2018; Ben‑Zvi et al., 2019). As such, competition for electrons between HydA1 and FNR is hypothesized to be a major limitation during PS II‐dependent H₂ photoproduction. FNR protein levels are approximately 70‐fold higher than that of HydA1 in C. reinhardtii (Nikolova et al., 2018) and the affinity of FNR to PetF is 4‐ to 13‐fold higher than that of HydA1 (K_m (FNR) = 0.8–2.6 mM, K_m (HydA1) = 3.4–35 mM) (Kanygin et al., 2020). Thus, a strategy for improving H₂ photoproduction by C. reinhardtii has been to engineer proteins to increase the interaction of PetF with HydA1 rather than FNR (Sawyer and Winkler, 2017). A PetF‐HydA1 fusion protein was shown to successfully divert over 60% of the photosynthetic electron pool towards H₂ production in vitro compared with less than 10% observed for HydA1 (Yacoby et al., 2011). This fusion protein also improved H₂ production 4.5‐fold in vivo compared with the wild type when expressed in C. reinhardtii (Eilenberg et al., 2016). The increase in H₂ production was attributed to a tethering of PetF‐HydA1 to PS I allowing for effective interception of photosynthetic electrons. The fusion protein also showed improved O₂ resistance in vitro and in vivo, retaining 25% activity after 10‐min exposure to O₂ (1.7 µmol) compared to 7.5% activity for wild‐type HydA1 (Eilenberg et al., 2016). The PetF moiety of the HydA1‐PetF fusion was hypothesized to either reduce O₂ to O₂ ^‐, which cannot penetrate the active site of HydA1 due to its charge, or partially block O₂ to access the pathway leading to the active site of HydA1 (Eilenberg et al., 2016).

In another study, a double mutant C. reinhardtii PetF construct (D19A/D58A) produced by site‐directed mutagenesis showed a fourfold improved H₂ production in vivo (Rumpel et al., 2014). The PetF double mutant construct showed a reduced affinity for FNR, and so photosynthetic electrons were more likely to be directed towards HydA1 and H₂ production. More recently, site‐directed mutagenesis of FNR and Fd from Synechocystis PCC 6803 was undertaken to assess the protein to protein interaction with recombinant HydA1 from C. reinhardtii (Wiegand et al., 2018). Single amino acid exchange of aspartate to alanine near the C terminus (Fd‐D22A) and the N‐terminus (Fd‐D61A) of Fd increased Fd‐HydA1 interaction and H₂ photoproduction by 15 and 23% respectively (Wiegand et al., 2018). Exchange of a lysine residue to aspartate near the N‐terminus of the cyanobacterial FNR (FNR‐K78‐D), together with the Fd‐D61A mutant, significantly decreased Fd/FNR interaction and conversely increased Fd/HydA1 interaction and H₂ photoproduction in vitro by 18‐fold (Wiegand et al., 2018). Modifications to HydA1 to improve binding affinity for PetF have not yet been explored but could further improve H₂ production. This is a challenging task, suggested to likely require directed evolution rather than rational site‐directed mutagenesis (Sawyer and Winkler, 2017).

A HydA1 and superoxide dismutase (SOD) fusion protein, originally designed to increase the O₂ tolerance of HydA1, greatly enhanced H₂ production (Ben‐Zvi and Yacoby, 2016). While the HydA1‐SOD fusion protein did not protect HydA1 from O₂, the fusion protein showed H₂ production up to 10–15 mmol H₂ l⁻¹ h⁻¹ in C. reinhardtii, the highest photosynthetic rate of H₂ production reported to date. The HydA1‐SOD fusion protein was also shown to produce H₂ continuously for up to 14 days. This was hypothesized to be due to the ability of the HydA1‐SOD fusion protein to compete with the CBB cycle for photosynthetic electrons, and reduced O₂ evolution (Ben‑Zvi et al., 2019). This extended H₂ production period was also achieved with non‐limiting media rather than using nutrient‐deprived media that is typically used to activate PS II‐independent H₂ photoproduction. The majority of electrons for H₂ production using HydA1‐SOD were shown to be supplied by PS II LEF, with starch catabolism playing a secondary role. The competition with CBB does have some drawbacks, as C. reinhardtii HydA1‐SOD mutants had slower growth rates, likely due to the reduced flux through the CBB cycle and impaired photosynthetic activity. The molecular mechanism accounting for this improved H₂ production by HydA1‐SOD is unclear, although two hypotheses were put forward. First was that HydA1‐SOD may be bound to or in close proximity to PS I, hindering the access of FNR to reduced PetF. The alternate hypothesis was that HydA1‐SOD outcompetes soluble FNR for electrons in the stroma (Ben‑Zvi et al., 2019).

High‐throughput screening tools of hydrogenase activity

Screening of large libraries of synthetic, engineered or native hydrogenases and other proteins involved in biological H₂ production for improved H₂ production in C. reinhardtii ideally requires accurate, efficient and reliable HTS platforms. These HTS platforms can be automated and integrated in automated strain engineering pipelines. While several instruments and methods are available to sensitively assay biological H₂ production, such as the Clark‐type electrode (normally used for measuring dissolved O₂) in reverse polarity configuration (Mišlov et al., 2015) or gas chromatography (Weijun, 2015), these methods can be slow and cumbersome and are typically low‐throughput. In order to process large sample numbers, HTS methods (Ghirardi, 2015) have been developed which include chemochromic films (Posewitz et al., 2004b); biochemical assays using sulfonated Wilkinson’s catalyst with a tetrazolium indicator (Schrader et al., 2008); methyl‐viologen assays in combination with cell‐free synthesis using single‐molecule PCR expression (SIMPLEX) (Stapleton and Swartz, 2010a); in vitro compartmentalization using C₁₂‐resorufin fluorescence (Stapleton and Swartz, 2010b); or a Rhodobacter capsulatus H₂ biosensor (Wecker and Ghirardi, 2014). Another high‐throughput method that can be considered is the use of chemically synthesized [2Fe]_H cluster mimics which can be used to artificially mature the [FeFe] hydrogenases (Land et al., 2019). These HTS assays and their applications for characterizing hydrogenases and measuring H₂ production are briefly described in the following sections. With the exception of chemochromic films and the Rhodobacter biosensor, the HTS assays for measuring H₂ typically rely on redox reactions of the hydrogenase with a non‐natural substrate, and hence the measured rates in vitro might not always reflect H₂ production in vivo.

Pd/WO₃ chemochromic films

Chemochromic films are composed of Pd/WO₃ layers that turn blue when reduced by H₂. These films are very sensitive to H₂ and can detect amounts of H₂ as low as 4 nmol (Siebert et al., 2001). Pd/WO₃ chemochromic films have been used to measure H₂ production in a strain of C. reinhardtii containing a mutated and non‐functional isoamylase gene, STA7 (Posewitz et al., 2004b). As STA7 is involved in the accumulation of starch, the decrease in H₂ production detected by the films provided a clear link of starch degradation to H₂ production by dark fermentation (Posewitz et al., 2004b). In another study, Pd/WO₃ chemochromic films were used in a high‐throughput H₂ production assay device (H₂PAD) to screen over 10,000 C. pasteurianum CpI hydrogenase mutants and 400 Oryza sativa FNR mutants, identifying a hydrogenase with threefold higher H₂ turnover rate than wild‐type CpI, and an FNR mutant that increased NADPH‐driven H₂ production by 60% (Koo et al., 2017). The H₂PAD device consisted of a 96‐well plate at the bottom where biological H₂ production took place, a filter plate containing a desiccant (CaSO₄) and at the top the Pd/WO₃ sensor plate.

Sulfonated Wilkinson’s catalyst with tetrazolium indicator

Biological H₂ production is also effectively measured using a semi‐quantitative colorimetric assay containing a water‐soluble tetrazolium indicator (WST‐3) and soluble Wilkinson’s catalyst (Schrader et al., 2008). This method was used to measure H₂ production by Synechocystis sp. PCC 6803 encapsulated in silica sol‐gel (Dickson et al., 2009). Wilkinson’s catalyst is used to catalyse the hydrogenation of primary and secondary alkenes. In this assay, the Wilkinson’s catalyst hydrogenates the tetrazole ring of WST‐3, forming a formazan molecule which can be detected via a colour change at 433 nm. While not as sensitive as Pd/WO3 chemochromic film based assays, the colorimetric assay using Wilkinson’s catalyst has been shown to detect H₂ from whole‐cell systems in microtitre plates for amounts of H₂ as low as 20 nmol (Schrader et al., 2008).

SIMPLEX‐based methyl‐viologen HTS assay

Single‐molecule PCR‐linked expression (SIMPLEX) in conjunction with cell‐free protein synthesis (CFPS), methyl‐viologen reduction and gas chromatography (GC) has been used to screen the effects of hydrogenase mutations on H₂ production in vitro. After screening ~ 30,000 hydA1 mutants with the SIMPLEX‐based assay, a novel HydA1 mutant was identified which showed fourfold improved H₂ production compared with the wild type (Stapleton and Swartz, 2010a). SIMPLEX can amplify single molecules of DNA from a mutant library after serial dilution into microtitre plates. Subsequent expression of the mutant DNA using CFPS results in the plate wells containing on average just 2 mutant proteins which can then be characterized using high‐throughput assays. After amplification with SIMPLEX and expression with CFPS, H₂ consumption rates of hydrogenases are measured using methyl viologen as substrate. Methyl viologen can be an electron donor and acceptor for [FeFe] hydrogenases similar to PetF. Hydrogenases that oxidise H₂ in solution reduce methyl viologen which is indicated by a colour change from clear to blue, detectable at 578 nm. Following identification of high H₂ consumers, a shortlist of promising hydrogenases can then be analysed for their H₂ production rates using low‐throughput techniques such as GC. A particular advantage of the SIMPLEX‐based assay is the accelerated protein expression offered by CFPS compared with the typically slow‐growing C. reinhardtii.

In vitro compartmentalization and fluorescence‐activated cell sorting

A HTS for O₂ tolerance has been developed using a combination of in vitro compartmentalization (IVC), microbead display and fluorescence‐activated cell sorting (FACS) (Stapleton and Swartz, 2010b). The throughput of IVC‐FACS screening is significantly greater than 96‐well plate assays and is able to sort > 1000 beads per second. This HTS tool is based on beads which bind to the DNA templates and are emulsified into droplets to create individual CFPS reactors. To demonstrate the suitability of this technique, a mock library of C. pasteurianum hydrogenase CpI was successfully screened for O₂ tolerance. For this, streptavidin‐coated beads are generated that each displayed a single biotinylated DNA template and biotinylated anti‐hemagglutinin (HA) antibodies. The beads are then emulsified to form oil phase emulsion droplets to express hydrogenase proteins using emulsion CFPS (eCFPS). E. coli extracts containing the [FeFe] hydrogenase maturases HydE, HydF and HydG from Shewanella oneidnsis are included in the emulsions to mature the hydrogenases. The hydrogenases contain three HA tags which once expressed bind to the anti‐HA antibodies on the streptavidin‐coated beads. After protein synthesis, the beads are removed from the emulsion droplets, washed and exposed to O₂. New emulsion droplets containing the beads and the reporter molecule C₁₂‐resazurin are created after O₂ exposure. C₁₂‐resazurin forms a fluorescent molecule C₁₂‐resorufin when reduced by active hydrogenases in a H₂ consumption assay. Subsequently, the beads are again removed from the emulsion and FACS is used to detect and enrich fluorescent beads which contain hydrogenases that are tolerant to O₂ exposure. Since the beads still bind the respective DNA template, a direct genotype–phenotype link can be established to provide the sequence to the O₂ tolerant hydrogenase. While this particular assay was tested to screen for O₂ tolerance of hydrogenases, it is useful for screening other important hydrogenase characteristics including thermostability and tolerance to other harsh conditions. The IVC‐FACS assay was also confirmed to be compatible with HydA1 from C. reinhardtii, indicating that this screening platform may be used to screen [FeFe] hydrogenases from diverse organisms for a variety of characteristics (Stapleton and Swartz, 2010b).

Rhodobacter biosensor

A Rhodobacter capsulatus H₂ biosensor has been engineered to fluoresce upon exposure to H₂ (Wecker et al., 2011). This organism has a H₂ sensory system composed of a H₂ sensor protein (HupUV), a histidine kinase (HupT), a transcription regulator (HupR) and an uptake hydrogenase (HupSL). The fluorescent reporter emGFP was inserted downstream of HupSL on a plasmid in E. coli S17‐1 and transferred by conjugation into the recipient strain R. capsulatus JP91. After exposing the R. capsulatus biosensor to 0.1% H₂ for 4 h, the activation of emGFP by 1% O₂ exposure for 2 h was found to be sufficient to generate a fluorescent signal that was approximately eightfold higher than cells not exposed to H₂. The lower limit of detection using the biosensor was 200 pM H₂ in solution. A 1:1 ratio of C. reinhardtii and the R. capsulatus biosensor was co‐cultured in microtitre plates incubated under dark anaerobic conditions, followed by activation of emGFP by O₂ exposure. Fluorescent measurements of the co‐culture correlated with H₂ concentrations in the headspace gas measured by GC.

The R. capsulatus biosensor has since been adapted for use on solid media to rapidly test for H₂ production by overlaying the biosensor, grown on agar, over single colonies of H₂ producing strains (Wecker and Ghirardi, 2014). The biosensor was recently used to assay the O₂ tolerance of Clostridium pasteurianum CpI mutants expressed heterologously in C. reinhardtii, identifying the O₂ tolerant CpI^T356V/S357V mutant (Elman et al., 2020). The advantage of this assay is that it measures actual H₂ production of the cells and not reduction in alternative, non‐native substrates.

Artificial maturation using [2Fe]_H mimics

Recently, a HTS assay for recombinant [FeFe] hydrogenases in E. coli was described that circumvents the expression of HydEF and HydG (Land et al., 2019). When heterologously expressed in E. coli, [FeFe] hydrogenases lacking the [2Fe_H] subcluster were artificially matured with a [2Fe]_H subcluster mimic, [Fe₂(adt)(CO)₄(CN)₂]^2‐ ([2Fe]^adt) to produce the functional enzyme (Berggren et al., 2013; Esselborn et al., 2013; Birrell et al., 2016; Caserta et al., 2016; Chongdar et al., 2018). In this way, artificial maturation with ([2Fe]^adt) enables the screening of [FeFe] hydrogenases without co‐expression of HydEF and HydG. This approach simplifies any HTS assay and allows for screening of mutants that may not be able to be matured by HydEF and HydG. After artificial maturation, the hydrogenases were assayed for H₂ production using reduced methyl viologen, complemented with protein film electrochemistry, whole‐cell electron paramagnetic resonance (EPR) and Fourier transform infrared (FTIR) spectroscopy (Land et al., 2019). Together, these techniques allow a more complete characterization of hydrogenases through quantification of expression levels and turnover frequency. This study was the first to employ protein film electrochemistry to characterize hydrogenases. The combined use of these techniques helped to characterize the hydrogenase Tam‐HydA from Thermoanaerobacter mathranii, which showed a relatively low rate of H₂ production of 0.45 ± 0.18 µmol H₂ min⁻¹ l⁻¹ OD₆₀₀ ⁻¹, being eightfold lower compared with C. reinhardtii HydA1 (3.6 ± 0.30 µmol H₂ min⁻¹ l⁻¹ OD₆₀₀ ⁻¹), and which led to the discovery that this enzyme has predominantly a H₂ sensory function rather than a catalytic function. Further use of these combined techniques will be useful in future efforts to discover more efficient hydrogenases.

Conclusion and future perspectives

Many of the fundamental processes that drive H₂ production in C. reinhardtii have been described. Knowledge of these processes in conjunction with the genetic engineering tools of C. reinhardtii and target proteins have been used to improve H₂ production in C. reinhardtii. The major strategies explored were redirecting electron flow towards H₂ production, improving hydrogenase O₂ tolerance and using heterologous hydrogenases which may prove to be valuable for industrial biohydrogen production scales. Methods for assaying biological H₂ production are also now moving towards HTS which allows mutant libraries or more hydrogenase variants to be more rapidly screened, accelerating the synthetic biology ‘design‐build‐test’ cycle. Future efforts to improve H₂ production in C. reinhardtii should try to combine the proposed engineering targets mentioned above and select an appropriate HTS assay to efficiently screen the new constructs. In addition, new tools such as genetic circuits could be developed to couple H₂ production to growth or positive selection to support enrichment of mutants with superior H₂ producing traits. This use of selective pressure would allow for more randomized, evolutionary engineering approaches such as directed evolution and adaptive laboratory evolution which can rapidly increase the genetic diversity and broaden the scale of the engineering. Altogether, knowledge of the fundamental processes underlying H₂ production in C. reinhardtii, discovery of O₂ tolerant hydrogenases, ever‐evolving synthetic biology tools and development of HTS platforms has the potential to improve biological H₂ production to the point where it is economically viable to produce at an industrial scale.

Conflict of interest

The authors declare no conflict of interest.

Microbial Biotechnology (2022) 15(7), 1946–1965