Genetic Engineering, Synthetic Biology and the Light Reactions of Photosynthesis^,

Abstract

Applications of synthetic biology to photosynthesis currently range from exchanging photosynthetic proteins to the utilization of photosynthesis as a source of electrons for entirely unrelated reactions.

Oxygenic photosynthesis is imperfect, and the evolutionarily conditioned patchwork nature of the light reactions in plants provides ample scope for their improvement (Leister, 2012; Blankenship and Chen, 2013). In fact, only around 40% of the incident solar energy is used for photosynthesis. Two obvious ways of reducing energy loss are to expand the spectral band used for photosynthesis and to shift saturation of the process to higher light intensities. Indeed, even minor enhancements to the efficiency or stress resistance of the light reactions of photosynthesis should have a positive impact on biomass production and yield (Leister, 2012; Blankenship and Chen, 2013; Long et al., 2015).

However, modifications to the essential structure of the light reactions of plant photosynthesis are currently limited by two main factors. One is the high degree of conservation of their structural components, which limits the efficiency gains attainable by conventional breeding approaches (Dann and Leister, 2017). The second is that the organization of these structural components into multiprotein complexes requires the simultaneous tailoring of several proteins, some of them encoded by different genetic systems in different subcellular compartments (nucleus and chloroplasts) in land plants. Therefore, the successful modification of the light reactions of plant photosynthesis has been limited to a few cases. Similarly, modification of the activity of auxiliary proteins involved in the regulation of the light reactions to enhance plant growth and yield only recently resulted in successful outcomes and is limited to a small subset of regulatory proteins (Pribil et al., 2010; Kromdijk et al., 2016; Głowacka et al., 2018).

In addition to enhancing the light reactions for improved biomass production and yield, concepts have been developed for the direct coupling of photosynthesis to other important pathways that are connected indirectly to photosynthesis in natural systems, for instance, because they reside in different subcellular compartments (Lassen et al., 2014b). Direct coupling would be expected to boost the production of rare compounds in cells and contribute to the biotechnological production of high-value compounds in vivo.

In vitro, it is possible to functionally link components of photosynthesis with entirely unrelated biotic or abiotic catalysts or with abiotic electrode materials. In fact, photosystem I (PSI) is naturally adapted for highly efficient light harvesting and charge separation (Kargul et al., 2012; Nguyen and Bruce, 2014; Martin and Frymier, 2017) and has been described as the most efficient natural nano-photochemical machine (Nelson, 2009). Upon light excitation, PSI produces the most powerful naturally occurring reducing agent, P700*, which, together with an exceptionally long-lived charge-separated state, provides sufficient driving force to reduce protons to H₂ at neutral pH. PSI operates with a quantum yield close to 1, and currently, no synthetic system has approached its remarkable efficiency. Moreover, PSI preparations generally are robust, especially those obtained from extremophilic microalgae (Kubota et al., 2010; Haniewicz et al., 2018). The superior qualities of PSI have stimulated strategies designed to generate in vitro hybrids of PSI and various types of redox-active catalysts or other materials.

In sum, the light reactions of photosynthesis are a prime target for genetic engineering and synthetic biology approaches for three major reasons. (1) Enhancement of the process in vivo to increase the efficiency of light use promises to increase biomass and crop yields. (2) Coupling of the light reactions of photosynthesis to previously unconnected pathways will enable us to utilize the reducing power of the light reactions directly to produce large amounts of high-value compounds in vivo. (3) The high efficiency and robustness of PSI should allow it to be used in hybrids with biotic or abiotic components to generate hydrogen, simple carbon-based solar fuels, or electricity in vitro. In this review, the background to and recent developments in these three strategies are discussed.

NATURAL BUILDING BLOCKS OF PHOTOSYNTHESIS

During photosynthesis, carbon dioxide (CO₂) is converted into organic compounds, principally sugars, using sunlight as energy. In plants, algae, and cyanobacteria, photosynthesis uses water as the electron donor for the chemical reduction of CO₂ and releases oxygen. Algae and plants derive from a lineage that arose from an endosymbiotic relationship between a protist and a cyanobacterium. Chloroplasts, the photosynthetic organelles in modern plants, are in fact the descendants of this ancient symbiotic cyanobacterium and possess an internal membrane system that resembles the thylakoid membranes of modern-day cyanobacteria. Indeed, during the evolutionary transition from cyanobacteria to chloroplasts, the overall organization and mode of action of the photosynthetic machinery was retained (Box 1). However, significant changes have occurred in the subunit composition of photosystems (Fig. 1), their posttranslational modification (Pesaresi et al., 2011), the harvesting of light energy, pigment composition, and the regulation of photosynthesis (Box 2; Holt et al., 2004; Mullineaux and Emlyn-Jones, 2005; Rochaix, 2007; de Bianchi et al., 2010).

Figure 1.

Subunit composition of cyanobacterial (Synechocystis) and plant (Arabidopsis) thylakoid multiprotein complexes. Subunits specific to either the cyanobacterium Synechocystis or the flowering plant Arabidopsis are indicated by black shading; subunits with domains specific to one group of organisms are shown in dark gray, and conserved subunits are shown in light gray. c6, Cytochrome c₆; Cyt b₆f, cytochrome b₆f complex; Fd, ferredoxin; FLV, flavodiiron protein; FNR, ferredoxin-NADP reductase; Fv, flavodoxin; LHCI (II), light-harvesting complex I (II). For reasons of clarity, the ATP synthase and NAD(P)H dehydrogenase complex are not shown.

Its evolutionary history makes photosynthesis well suited for synthetic biology strategies. The evolutionary diversification of the photosynthetic machinery has provided a set of building blocks, ranging from single proteins such as the soluble electron transporters (plastocyanin, cytochrome c₆, flavodoxin, and ferredoxin) to multiprotein complexes like photosystems and antenna complexes (phycobiliosomes and light-harvesting complexes [LHCs]). In principle, the building blocks should be interchangeable between cyanobacteria, algae, and plants. Instances of the swapping of homologous photosynthetic proteins between species by means of genetic engineering are discussed in the next section to highlight the complications associated with apparently straightforward approaches. The focus then shifts to genuinely synthetic approaches, in which specific photosynthetic building blocks have been introduced into photosynthetic species that lack them. In the subsequent two sections, the combination of nonphotosynthetic (bio-bio hybrids, using photosynthesis as an electron source for unrelated biological processes) and nonbiological (bio-nano hybrids, using photosynthesis as an electron source for nonbiological processes) building blocks with components of the photosynthetic machinery is described. An overview of these approaches is provided in Figure 2.

Figure 2.

Overview of genetic engineering and synthetic biology approaches related to the light reactions of photosynthesis. Different shading is used to indicate cyanobacterial (blue) and plant (green) proteins (complexes) and proteins (complexes) that are typically not associated directly with photosynthesis (red). Yellow shading indicates nonbiological materials. For designations of proteins, see Figure Box 1.

EXCHANGE OF CONSERVED PHOTOSYNTHETIC MODULES

Theoretically, it should be relatively easy to exchange individual conserved photosynthetic proteins between different species, even in such distantly related organisms as cyanobacteria and plants. However, in practice, these simple genetic engineering exercises can be problematic (Table 1). Several individual proteins from cyanobacterial PSII (D1, CP43, CP47, and PsbH) and PSI (PsaA) have indeed been genetically replaced by their plant counterparts with varying success. Synechocystis PCC6803 (Synechocystis) strains expressing the D1 protein from Poa annua or PsbH from maize (Zea mays) could still perform photosynthesis, albeit with less efficiency than the wild-type strain (Nixon et al., 1991; Chiaramonte et al., 1999), but the replacement of Synechocystis CP43 or CP47 by their homologs from spinach (Spinacia oleracea) was incompatible with photoautotrophy (Carpenter et al., 1993; Vermaas et al., 1996). Similarly, in Synechocystis strains equipped with Arabidopsis (Arabidopsis thaliana) PsaA, PSI function was severely disrupted (Viola et al., 2014).

Table 1. Replacement of conserved individual or multiple photosynthetic proteins.

Gene/Protein	Description	Protein Identity	Reference
psbA/D1	Synechocystis D1 was replaced by its homolog from Poa annua, which resulted in slower growth. This is likely to be due to increased charge recombination between the donor and acceptor sides of the reaction center.	86%	Nixon et al. (1991)
psbB/CP47	Synechocystis CP47 was replaced by its homolog from spinach and photoautotrophy was lost.	80%	Vermaas et al. (1996)
psbC/CP43	Synechocystis CP43 was replaced by its homolog from spinach and photoautotrophy was lost.	85%	Carpenter et al. (1993)
psbH/PsbH	Synechocystis PsbH was replaced by its counterpart from maize. The resulting strain displayed increased light sensitivity and lower chlorophyll content.	78%	Chiaramonte et al. (1999)
psaA/PsaA	Synechocystis PsaA was replaced by its equivalent from Arabidopsis. This resulted in reduced photoautotrophic growth and a drastically reduced chlorophyll-phycocyanin ratio.	80%	Viola et al. (2014)
psbA, -B, -C, -D, -E, -F/D1, CP47, CP43, D2, cytochrome b559-α, cytochrome b559-β	All six core subunits of C. reinhardtii were replaced by their counterparts from two different green algae (Volvox carteri and Scenedesmus obliquus). The resulting strains were photoautotrophic but showed reduced photosynthetic efficiency, and the heterologous proteins reached only between 10% and 20% of the levels of those they replaced.	82%–99%	Gimpel et al. (2016)

Such complications resulting from the replacement of core subunits of photosystems, despite their high similarity (78%–86% identity between the cyanobacterial and plant proteins; Table 1), reflect the so-called "frozen metabolic state" of the photosynthetic multiprotein complexes. This term was coined by Gimpel et al. (2016) but was introduced originally as "frozen metabolic accident" by Shi et al. (2005). The frozen metabolic accident concept refers to the observation that selection has not significantly altered biophysically and physiologically inefficient photosynthetic proteins (including the D1 protein of PSII or Rubisco of the Calvin-Benson cycle) over billions of years of evolution. In fact, bioinformatic analysis of photosynthetic cyanobacterial genes suggests that the evolution rate of proteins at the core of the photosynthetic apparatus is highly constrained by protein-protein, protein-lipid, and protein-cofactor interactions (Shi et al., 2005). This provides an internal selection pressure, conserving the sequence of proteins in photosynthetic multiprotein complexes and, in the case of prokaryotes, also the genomic organization of their genes (Shi et al., 2005). The term frozen metabolic accident was generalized to frozen metabolic state by Gimpel et al. (2016) and implies that, in each photosynthetic species, slightly distinct modules of the cores of photosynthetic multiprotein complexes have evolved that are optimized with respect to their intrinsic interactions and that rarely tolerate the alteration of single proteins by exchange or mutation. In consequence, the simultaneous exchange of entire sets of core proteins (or modules) with all their intrinsic interactions should be more feasible than substituting individual proteins that may disrupt the frozen metabolic state. Such an experiment has been performed in the green alga Chlamydomonas reinhardtii, where the six PSII core proteins D1, CP47, CP43, D2, cytochrome b₅₅₉-α, and cytochrome b₅₅₉-β were swapped for their homologs from two other green algal species (Gimpel et al., 2016). Photoautotrophy was not affected by the exchange of this synthetic biology module, although the fully altered strains performed suboptimally compared with strains in which only one to five genes were exchanged. However, in control experiments where the synthetic C. reinhardtii six-gene module was reintroduced to the C. reinhardtii deletion strain that lacked all six genes, only about 86% of wild-type PSII functionality was rescued. Tentative explanations for this decreased efficiency include the following: (1) off-target effects of the PSII gene deletions on the operon components and tRNA genes associated with these PSII loci; (2) misregulation of the transferred PSII genes because their expression cassettes lacked unidentified cis-acting DNA elements; and (3) perturbation of polycistron-dependent posttranscriptional regulation of the transformed genes, since they were no longer part of an operon (Gimpel et al., 2016).

Taken together, the outcomes of these replacement experiments indicate that exchanging conserved photosynthetic proteins from multiprotein complexes can be problematic, given the highly integrated nature of the photosynthetic machinery. Therefore, approaches designed to enhance photosynthesis, such as introducing the high light-resistant D1 protein from a green alga that lives under extreme conditions (Treves et al., 2016) into crop plants, do not appear promising. Instead, entire (sub)complexes with their internal network of evolutionarily optimized interactions should be transferable between species.

EXCHANGE OF NONCONSERVED PHOTOSYNTHETIC MODULES

Inspection of the repertoire of subunits of the photosynthetic machinery in the three model species Synechocystis PCC6803, C. reinhardtii, and Arabidopsis, which represent different stages in the evolution of oxygen-generating photosynthesis, shows that the most dramatic changes in the photosynthetic proteome occurred during the transition from the cyanobacterial endosymbiont (for which Synechocystis serves as proxy) to the chloroplast of unicellular algae (with C. reinhardtii as proxy; Fig. 3; Supplemental Table S1). In particular, phycobilisomes, flavodoxin, and several photosystem subunits were lost, while LHCs, some novel photosystem subunits, and several proteins involved in alternative electron pathways or photoprotection evolved. During the transition from algal chloroplasts to those of flowering plants, relatively few proteins were lost (e.g. flavodiiron proteins and the canonical cytochrome c₆) or acquired (Lhcb6/CP24; Fig. 3; Supplemental Table S1). The NAD(P)H dehydrogenase (NDH) complex involved in antimycin A-insensitive cyclic electron flow (Box 1) is a special case, since the chloroplast NDH from flowering plants traces back to the cyanobacterial complex, but the complex was lost during evolution in C. reinhardtii (Fig. 3; Supplemental Table S1).

Figure 3.

Evolutionary plasticity of the photosynthetic proteome. The changes in the composition of the inventory of photosynthetic proteins (top) during evolution from the cyanobacterial endosymbiont to the chloroplast in flowering plants (bottom) are shown. As proxies for the original endosymbiont and the unicellular chloroplast-containing protist that gave rise to flowering plants, the model cyanobacterium Synechocystis PCC6803 (left), the model green alga C. reinhardtii (middle), and the model flowering plant Arabidopsis (right) are used. At top left, the entire inventory of photosynthetic proteins in Synechocystis PCC6803 is listed and assigned to the five different classes PSII, PSI, other electron transport components (other ET), antenna, and photoprotection, whereby the transition from other ET to photoprotection is fluid. The proteins that have been acquired or lost during evolution in C. reinhardtii and Arabidopsis are listed in the middle and right sections, respectively, of the top part. Note that, for reasons of simplicity, we have not considered the ATP synthase complex here. The NDH complex (or Nda2 in the case of C. reinhardtii) appears only as a whole (without its individual subunits) here. NDH listed in parentheses indicates that the NDH complex is specifically lost only in C. reinhardtii and, therefore, that the plant NDH complex is not a reacquisition. Accordingly, Nda2 replaces the NDH complex in C. reinhardtii. A detailed catalog of the indicated proteins with their full names, functions, and further literature links is available in Supplemental Table S1. The bottom part provides a sketch of the evolution of flowering plants that traces back to the endosymbiosis between a unicellular eukaryote and a cyanobacterium, resulting (besides the red algal and glaucophyte lineages that are not shown) in chloroplast-containing protists that evolved further to plants.

Several attempts have been made to introduce photosynthetic proteins into species that lack the corresponding homolog (Table 2). These heterologous expression approaches have been successful for the soluble electron transporters flavodoxin, cytochrome c₆, and flavodiiron proteins. Indeed, cyanobacterial flavodoxin can at least partially replace plant ferredoxin and confer enhanced stress tolerance when expressed in addition to ferredoxin (Tognetti et al., 2006, 2007; Blanco et al., 2011). Similarly, red algal cytochrome c₆ enhances the growth and photosynthesis of Arabidopsis plants (Chida et al., 2007). Since Arabidopsis lacks a functional cytochrome c₆ that can transfer electrons from the cytochrome b₆f complex to PSI (Molina-Heredia et al., 2003; Weigel et al., 2003), it is plausible that the more oxidized plastoquinone pool in the transgenic plants is a direct consequence of additional PSI reduction mediated by the algal protein (Chida et al., 2007). Interestingly, it seems to make no difference if an endogenous soluble electron transport protein (Box 3) or its heterologous equivalent from a distant species is overexpressed. For instance, overexpression of the endogenous soluble proteins plastocyanin and ferredoxin also can enhance growth in plants (Pesaresi et al., 2009; Lin et al., 2013; Chang et al., 2017; Zhou et al., 2018). Interestingly and rather unexpectedly, this concept also can work for certain proteins that are part of multiprotein complexes (Simkin et al., 2017; see Box 3). This indicates that the quantity of such proteins is relevant for growth enhancement and not their evolutionary origin.

Table 2. Introduction of heterologous photosynthetic proteins (protein complexes).

Gene/Protein	Description	Reference
PetJ/cytochrome c6	Growth and photosynthesis of Arabidopsis plants was enhanced by the expression of a red algal (Porphyra yezoensis) cytochrome c6 gene.	Chida et al. (2007)
Fld/flavodoxin	Tobacco lines expressing a plastid-targeted cyanobacterial flavodoxin in addition to endogenous ferredoxin display increased tolerance to environmental stress. Flavodoxin can at least partially replace ferredoxin in tobacco.	Tognetti et al. (2006, 2007; Blanco et al. (2011
FlvA+B/flavodiiron protein (FLV)	FlvA and FlvB from the moss Physcomitrella patens mediate pseudocyclic electron flow in Arabidopsis.	Yamamoto et al. (2016)
Lhcb/LHCII	Pea Lhcb protein is synthesized in Synechocystis and integrated into the membrane, but it is then degraded, such that it cannot be detected by immunoblot analysis.	He et al. (1999)

More recently, the two Physcomitrella patens flavodiiron protein (FLV) genes FlvA and FlvB were introduced into Arabidopsis, which, like other angiosperms, has lost FLVs during evolution (Yamamoto et al., 2016). FLVs are the main mediators of pseudocyclic electron flow in photosynthetic organisms, but heterologous expression of FLVs in Arabidopsis had no effect on steady-state photosynthesis and growth of the transgenic plants (Yamamoto et al., 2016). However, the Arabidopsis FLV lines displayed higher photosynthetic yields just after the onset of actinic light following a long dark adaptation, suggesting that the FLVs mediated a large electron sink during the induction of photosynthesis. In fluctuating light experiments, the Arabidopsis FLV lines had much less PSI acceptor side limitation, implying that the large FLV-mediated electron sink makes photosynthesis more resistant to the fluctuating light. Consequently, this protective effect of FLVs is more pronounced in Arabidopsis lines that are more sensitive to fluctuating light, like the pgr5 mutant defective in antimycin A-sensitive cyclic electron flow (Box 1; Leister and Shikanai, 2013; Yamamoto et al., 2016). Similarly, the expression of FLVs in a rice (Oryza sativa) line with markedly reduced cyclic electron flow restored CO₂ assimilation and growth rate to wild-type levels (Wada et al., 2018). Like FLVs, LHcx/LHCSR proteins play a role in photoprotection in green algae and mosses, but they have been lost in angiosperms during evolution. The heterologous expression of P. patens LHCSR1 in Nicotiana benthamiana and Nicotiana tabacum yielded an active protein that has properties similar if not identical to those of moss LHCSR1 (Pinnola et al., 2015).

Efforts to create LHCII complexes like those in plants by heterologously expressing the membrane-spanning light-harvesting chlorophyll a/b-binding protein Lhcb from Pisum spp. (pea) in Synechocystis were unsuccessful. Although the pea Lhcb protein was synthesized in Synechocystis and integrated into the membrane, it did not accumulate to steady-state levels detectable by immunoblot analysis (He et al., 1999). Possible explanations are that Lhcb is degraded rapidly, either because its unfamiliar structure makes it a good substrate for the cyanobacterial proteolytic system or it cannot fold/assemble properly due to the lack of plant-specific pigments or assembly factors. Interestingly, chlorophyll b production (Box 2) after the introduction of plant chlorophyll a oxygenase is boosted when Lhcb is expressed, even though LHCII does not accumulate in detectable amounts (Xu et al., 2001).

Even soluble multiprotein complexes can be expressed heterologously, as has been demonstrated impressively from the carbon-fixation end of photosynthesis. For example, by coexpression of five auxiliary factors, a functional plant Rubisco complex was assembled successfully in Escherichia coli (Aigner et al., 2017). This result is in line with the frozen metabolic state concept, showing that photosynthetic proteins embedded in a network of interactions with other proteins and auxiliary factors can be introduced into distantly related species, but only with their own interaction networks. While Gimpel et al. (2016) showed that efficient gene expression needs to be accounted for when designing synthetic modules for the transfer of photosystem subunits from one species to another, Aigner et al. (2017) demonstrated that auxiliary factors required for the biogenesis of photosynthetic (sub)complexes need to be considered in such experiments, adding additional facets to the frozen metabolic state concept.

Auxiliary factors required for the accumulation of photosynthetic multiprotein complexes include the following: (1) cofactors like iron-sulfur clusters and pigments (Box 2); (2) chaperones that are required for the insertion of cofactors (e.g. pigments into LHCs; Schmid, 2008); and (3) assembly factors. Assembly factors are required to support the stepwise assembly and functionality of the multiprotein/pigment complexes, and many of these are conserved between photosynthetic species (Nixon et al., 2010; Nickelsen and Rengstl, 2013; Jensen and Leister, 2014). Therefore, the exchange of entire multiprotein complexes between distantly related species will not be simple due to the repertoire of auxiliary factors that has changed markedly during evolution. Since the species that Gimpel et al. (2016) studied were closely related, this aspect could be neglected. In consequence, the impact of these auxiliary factors on the formation of multiprotein/pigment complexes will have to be fully elucidated to understand which factors are sufficient to assemble a photosynthetic multiprotein complex; previous (genetic) approaches only identified factors required for this process. Hence, the transfer of entire photosynthetic (sub)complexes between distantly related species by a synthetic photosynthetic module must include entire sets of strongly interacting proteins (to address the frozen metabolic state), as well as all genetic elements and auxiliary factors sufficient for the efficient expression, biogenesis, and function of the proteins in the module.

BIO-BIO HYBRIDS: USING PHOTOSYNTHESIS AS AN ELECTRON SOURCE FOR UNRELATED BIOLOGICAL PROCESSES

The idea of using the reducing power of photosynthesis to drive unrelated metabolic reactions has inspired several biotechnological concepts. The photosynthesis-driven formation of secondary metabolites has been demonstrated in vivo, whereas the light-driven generation of H₂ by bio-bio hybrids so far works only in vitro and is closely related to bio-nano approaches. Therefore, the coupling of photosynthesis to previously unrelated pathways is discussed here first, followed by bio-bio systems for H₂ generation.

The redirection of PSI-reducing equivalents to drive reactions catalyzed by cytochrome P450 enzymes has been achieved in vivo by genetic modifications of plants and cyanobacteria (Lassen et al., 2014b; Nielsen et al., 2016; Mellor et al., 2017). Cytochrome P450s constitute the largest family of plant enzymes that act on various endogenous and xenobiotic molecules (Rasool and Mohamed, 2016). Their extreme versatility and irreversibility of catalyzed reactions make these enzymes very attractive for use in biotechnology, medicine, and phytoremediation. P450s are monooxygenases that insert an oxygen atom into hydrophobic molecules, which enhances their reactivity and hydrophilicity. Most eukaryotic P450s require NADPH:cytochrome P450 reductase as the electron donor.

The endoplasmic reticulum (ER) membrane generally is accepted to be the primary subcellular repository of eukaryotic P450s and their NADPH:cytochrome P450 reductase. By relocating cytochrome P450s to the chloroplasts, the reducing power of photosynthesis can be targeted directly to the reactions catalyzed by these enzymes, thus providing the basis for the large-scale production of valuable products. Initial attempts to couple P450s with photosynthesis were conducted in vitro (Table 3; Fig. 4A). Spinach chloroplasts were combined with microsomes from yeast cells that had been stably transformed with a fusion gene expressing the rat CYP1A1-NADPH:cytochrome P450 reductase fusion enzyme (Kim et al., 1996). These experiments confirmed that it is feasible to drive P450-mediated reactions using electrons derived from photosynthetically generated NADPH. Intriguingly, the NADPH:cytochrome P450 reductase is not always essential, as demonstrated by coincubating CYP79A1 from sorghum (Sorghum bicolor) with barley (Hordeum vulgare) PSI and employing ferredoxin to deliver electrons directly from PSI to the P450 (Jensen et al., 2011). This approach also was shown to be practicable in vivo. CYP79A1, either alone or in combination with CYP71E1 and the UDP-glucosyltransferase UGT85B1, was first targeted in vivo to cyanobacterial or plant thylakoid membranes, where they catalyzed the same reactions as in their original cellular compartment (the ER), utilizing photosynthetically reduced ferredoxin as the electron donor (Nielsen et al., 2013; Lassen et al., 2014a; Gnanasekaran et al., 2016; Wlodarczyk et al., 2016). Genetic fusions also have been employed for the coupling of P450s to PSI. CYP79A1 has been fused to either cyanobacterial PsaM or Arabidopsis ferredoxin, and the engineered enzyme showed light-driven activity both in vivo and in vitro (Lassen et al., 2014a; Mellor et al., 2016). In the latter case, the efficiency of the system was enhanced because the fusion could compete better with endogenous electron sinks coupled to metabolic pathways (Mellor et al., 2016).

Table 3. Combination of PSI with nonphotosynthetic proteins or complexes: bio-bio hybrids.

Gene/Protein	Description	Reference
-----------------------------------------------------------------------------PSI-P450 hybrids----------------------------------------------------------------------
Fusion enzyme of rat CYP1A1 and yeast NADPH-P450 reductase (in vitro)	Spinach chloroplasts were combined with microsomes from yeast expressing the rat CYP1A1/CPR fusion enzyme. In this system, NADP+ was photosynthetically reduced to NADPH to supply the electrons for P450, thus enabling the light-driven conversion of 7-ethoxycoumarin to 7-hydroxycoumarin.	Kim et al. (1996)
CP79A1 from sorghum (in vitro and in vivo)	The ER-derived P450 CYP79A1 was capable of converting Tyr to hydroxyphenyl-acetaldoxime, employing ferredoxin reduced by barley PSI (without the need for NADPH and CPR). In the next step, CYP79A1 was fused to cyanobacterial PsaM or Arabidopsis ferredoxin. The engineered fusions exhibited light-driven activity both in vivo and in vitro.	Jensen et al. (2011); Lassen et al. (2014a); Mellor et al. (2016)
CYP79A1, CYP71E1, and UGT85B1 from sorghum (in vivo)	The two P450s CYP79A1 and CYP71E1 and the UDP-glucosyltransferase UGT85B1 were targeted to N. benthamiana or Synechocystis thylakoid membranes, where they converted Tyr to dhurrin employing photosynthetically reduced ferredoxin.	Nielsen et al. (2013); Wlodarczyk et al. (2016)
---------------------------------------------------------------------PSI-hydrogenase hybrids---------------------------------------------------------------------
Proteobacterial [NiFe] hydrogenase genetically fused to cyanobacterial PSI (in vitro)	A fusion of cyanobacterial PsaE to an [NiFe] hydrogenase from the β-proteobacterium Ralstonia eutropha was reconstituted into a PsaE-deficient PSI from Synechocystis by self-assembly. In a modified version of this system, cytochrome c3 from Desulfovibrio vulgaris was cross-linked to the docking site of ferredoxin in PsaE, targeting electrons directly from PSI via cytochrome c3 to the hydrogenase. This gave the hydrogenase a competitive advantage over the natural acceptors of electrons from PSI. In a third variant of this system, the R. eutropha hydrogenase was genetically fused to Synechocystis PsaE and reconstituted into a PsaE-deficient PSI from Synechocystis. His tagging of PsaF enabled the assembly of the PSI-hydrogenase hybrid onto a gold electrode.	Ihara et al. (2006a, 2006b; Krassen et al. (2009
Green algal [FeFe] hydrogenase genetically fused to ferredoxin (in vitro)	The C. reinhardtii hydrogenase was fused to ferredoxin. This switched the bias of electron transfer from FNR to hydrogenase and resulted in an increased rate of hydrogen photoproduction in vitro.	Yacoby et al. (2011)
Bacterial [FeFe] hydrogenase wired to cyanobacterial PSI (in vitro)	To enable the transfer of electrons from the terminal Fe4S4 cluster in PSI of Synechococcus sp. PCC 7002 to the distal Fe4S4 cluster in the hydrogenase from Clostridium acetobutylicum, the two components were covalently coupled to each other via a molecular wire, the thiolated organic molecule octanedithiol. This allowed electrons to tunnel through the wire from PSI to the hydrogenase. Self-assembly of the PSI-wire-hydrogenase complex was obtained in vitro.	Lubner et al. (2010, 2011

Figure 4.

Design of bio-bio hybrids. A, Harnessing the reducing power of photosynthesis for cytochrome P450 enzymes (red) has been demonstrated in vitro and in vivo. In vitro approaches were based either on a CPR-CYP1A1 fusion protein that could use NADPH produced by photosynthesis or on CYP79A1 that can utilize photoreduced ferredoxin directly. The latter approach also was realized in vivo by fusing CYP79A1 to ferredoxin (not shown) or the PSI subunit PsaM. The most elaborate approach reported utilizes three enzymes (CYP79A1, CYP71E1, and UGT85B1) to couple dhurrin synthesis with photosynthesis. B, PSI-hydrogenase complexes are based either on genetic fusions of the hydrogenase (Hyd) to PsaE or ferredoxin or on covalent linking of the hydrogenase and PSI via a molecular wire. Currently, these bio-bio hybrids function only in vitro.

In contrast to PSI-P450 hybrids, PSI-hydrogenase hybrids currently function only in vitro. Unlike fossil fuels, H₂ is environmentally benign, as it produces only water when combusted. Therefore, in principle, harnessing of the reducing power of photosynthesis for the direct production of H₂ (i.e. ultimately using sunlight and water) would yield a fully sustainable system of energy generation. PSI, but not PSII, provides a standard midpoint potential that is sufficiently negative to power the reduction of protons to H₂ (Utschig et al., 2015). Hence, approaches have been developed to engineer PSI to produce H₂ either as a replacement or in addition to its natural product NADPH (see Figure Box 1) by redirecting PSI electrons to a catalytic component. This catalytic component can be abiotic or biotic (hydrogenases). The feasibility of coupling H₂ generation to photosynthesis was demonstrated 45 years ago by mixing chloroplast preparations with ferredoxin and hydrogenase (Benemann et al., 1973). Twenty-five years later, hydrogen evolution by direct electron transfer (without ferredoxin) from PSI to hydrogenases was accomplished for the first time (McTavish, 1998).

Hydrogenases catalyze the reversible interconversion of H₂ into protons and electrons and are widespread in nature; they occur in bacteria and archaea but also in some eukarya. In vivo, hydrogenases can mediate photosynthetic H₂ production, albeit mostly indirectly or under anaerobic conditions due to their oxygen sensitivity (Ghirardi, 2015; Oey et al., 2016). Hydrogenases can be classified according to their metal-ion composition (e.g. [NiFe] and [FeFe] hydrogenases; Lubitz et al., 2014; Martin and Frymier, 2017). [FeFe] hydrogenases preferentially catalyze proton reduction and can evolve H₂ at high rates but are extremely sensitive to oxygen and are the only type of hydrogenases found in eukaryotic microorganisms. [NiFe] hydrogenases are less sensitive to oxygen but preferentially oxidize H₂ under physiological conditions (Lubitz et al., 2014). In vitro, both types of hydrogenases have been linked directly to PSI (Table 3; Fig. 4B). PSI-[NiFe] hydrogenase complexes have been generated by fusing the hydrogenase genetically to the PsaE protein (Ihara et al., 2006b). PSI-[FeFe] hydrogenase complexes were obtained by genetically fusing the hydrogenase to ferredoxin (Yacoby et al., 2011) or covalently linking the FeS clusters present in the hydrogenase to PSI via a molecular wire (Lubner et al., 2010, 2011).

Two parameters characterize the efficiency of these in vitro systems: their H₂ production rate and longevity. While low rates of H₂ production (in the range from 0.1 to 10 μmol H₂ mg⁻¹ chlorophyll h⁻¹) were described for the early chloroplast extract experiments and genetic fusions of PSI to [NiFe] or [FeFe] hydrogenases (McTavish, 1998; Ihara et al., 2006b; Yacoby et al., 2011), higher rates of between 2,000 and 3,000 μmol H₂ mg⁻¹ chlorophyll h⁻¹ have been achieved with wired PSI-[FeFe] hydrogenase complexes and genetically fused PSI-[NiFe] hydrogenase complexes assembled on a gold electrode (Krassen et al., 2009; Lubner et al., 2011). Few data are available with respect to the longevity of the PSI-hydrogenase systems; however, a minimum lifetime of 64 d was reported for a wired [FeFe] hydrogenase-PSI complex at room temperature and under ambient illumination (Lubner et al., 2010).

The future use of hydrogenases in photosynthesis-driven H₂ production will depend strongly on whether it is possible to overcome the oxygen sensitivity of many hydrogenases, for instance by employing oxygen-tolerant [NiFe] hydrogenases (Burgdorf et al., 2005; Schiffels et al., 2013). If this is not possible, their efficient use in vivo in thylakoids, which inevitably generate oxygen during linear electron flow, will be impossible. However, as demonstrated with the gold surface system (Krassen et al., 2009), the design of novel matrices into which the hybrid system can be incorporated may enhance the efficiency markedly.

BIO-NANO HYBRIDS: USE OF PHOTOSYNTHESIS AS A SOURCE OF ELECTRONS FOR NONBIOLOGICAL PROCESSES

From bio-bio hybrids, it is only a small step to developing bio-nano hybrids, as demonstrated by PSI hybrids that employ abiotic catalysts for photosynthetic hydrogen production instead of hydrogenase. In fact, abiotic catalysts have the advantage of bypassing the lability of hydrogenases in the presence of oxygen. PSI-platinum hybrids have been produced by combining PSI with platinum nanoparticles (either directly or via tethering by nanowires) or platinum nanoclusters (Kargul et al., 2012; Fukuzumi, 2015; Utschig et al., 2015; Fig. 5A; Table 4). Current PSI-platinum hybrids are less efficient than the most advanced PSI-hydrogenase systems but are extremely robust (Utschig et al., 2015). However, future widespread usage of the PSI-platinum system will be limited by the high cost of platinum. A more economical alternative to precious metals are earth-abundant molecular catalysts. However, hybrids consisting of PSI and earth-abundant molecular catalysts have a much shorter working life than platinum-based configurations (Utschig et al., 2015), likely due to the instability of the molecular catalyst.

Figure 5.

Design of selected bio-nano hybrids. A, PSI-platinum hybrids are generated by combining platinum nanoparticles (dots) or nanoclusters (star) with PSI. Platinum nanoparticles also can be linked to PSI via nanowires. B, PSI-based photocurrent-generating system. A variety of such systems have been developed, which consist of PSI molecules immobilized on electrodes and implement electron transfer by means of diffusible redox mediators or nanowires. Moreover, all-solid-state PSI-based solar cells and systems in which cytochrome c was employed to interface PSI with electrode materials have been generated (Gordiichuk et al., 2014; Gizzie et al., 2015; Ciornii et al., 2017; Janna Olmos et al., 2017). The cross-containing circle indicates a current-using device, and the yellow rectangles symbolize the electrodes.

Table 4. Combination of PSI with nonbiological materials: bio-nano hybrids.

System/Nonbiological Components	Description	Reference
H2-producing PSI bio-nano hybrids	PSI-biohybrid photocatalytic systems for H2 production contain cyanobacterial PSI and precious metal catalysts, including platinum nanoparticles, platinum nanowires, and platinum nanoclusters. Examples for earth-abundant molecular catalysts in PSI-biohybrids include cobaloxime, nickel diphosphine, and nickel apoflavodoxin.	Reviewed in Kargul et al. (2012); Fukuzumi (2015); Utschig et al. (2015)
PSI-based photocurrent-generating bio-nano hybrids	PSI-based photocurrent-generating devices contain PSI from either plants or cyanobacteria immobilized on electrode materials such as gold, graphene, indium tin oxide, fluorine-doped tin oxide, TiO2, glass, ZnO, or alumina. The most commonly utilized immobilization strategies involve the usage of organothiol-based self-assembled monolayers. Electron donors to PSI include sodium ascorbate, 2,6-dichlorophenolindophenol, reduced ferricyanide, osmium complexes, ruthenium hexamine trichloride, or the immobilization wire (NTA-Ni-His6-PSI). Acceptors of PSI electrons include naphthoquinone-derivative molecular wire, methyl viologen, oxidized ferricyanide, composite Bis-aniline nanoparticle-ferredoxin, and Methylene Blue. Also, all-solid-state PSI-based solar cells that do not employ any exogenous redox mediators or buffer solutions have been generated.	Reviewed in Gordiichuk et al. (2014); Nguyen and Bruce (2014); Gizzie et al. (2015); Janna Olmos et al. (2017)

PSI-based photocurrent-producing devices constitute another class of photosynthesis-derived nano-bio systems (Table 4; Fig. 5). In such devices, PSI is immobilized onto electrodes. Many variants of this concept have been tested, such as varying the electrode materials, immobilization/orientation strategies, and/or artificial redox mediators (Nguyen and Bruce, 2014; Janna Olmos and Kargul, 2015; Plumeré and Nowaczyk, 2016; Kargul et al., 2018). PSI must be immobilized on the electrode surface in such a way that electron transfers between the electrodes and the oxidizing (P700) and reducing (F_B, the terminal [4Fe-4S] cluster, or an intermediate electron transporter) sides of PSI can proceed with the required efficiency. Electrons are transferred between the electrodes and the oxidizing or reducing sides of PSI either by a diffusible redox mediator or molecular wires. Examples of electrode materials and redox mediators are provided in Table 4. After P700 photoexcitation, electrons are transferred from P700 via several factors (P700 → A₀ → A₁ → F_X → F_A → F_B) to the iron-sulfur cluster F_B (Box 1). As in the case of PSI-hydrogenase and PSI-platinum nanoparticles, PSI can be wired to its electrodes. This has been achieved by wiring the A₁ cofactor (phylloquinone) to the substrate surface, such that electrons are transferred directly from A₁ to the electrode, thus bypassing the downstream FeS clusters (F_X, F_A, and F_B) in the stromal domain of PSI (Terasaki et al., 2007, 2009; Miyachi et al., 2009, 2010).

Modifications of PSI have been utilized to enhance the efficiency of the photocurrent-generating system (Das et al., 2004; Frolov et al., 2005; Carmeli et al., 2010). As mentioned previously, fusions to the stroma-faced PSI subunit PsaE can be used to link new components to PSI; however, in this case, PsaD fusions were used. To this end, recombinant His-tagged PsaD was immobilized on the functionalized electrode surface, which then was exposed to native PSI complexes, resulting in immobilized PSI with P700 facing away from the electrode (Das et al., 2004). Another way to control the orientation of PSI during immobilization involves introducing Cys mutations. To allow for direct thiol coupling to an gold surface, various residues on the lumen-exposed face of PSI were replaced by Cys and tested (Frolov et al., 2005). PSI attachment was achieved with all single mutants, even those placed farther from the P700 site, suggesting that a specific location is not required as long as the Cys is exposed at the luminal surface of PSI. The concept of targeted attachment via introduced Cys residues was exploited in subsequent studies to link PSI to maleimide-functionalized gallium arsenide (Frolov et al., 2008), to immobilize PSI between the substrate and a metallized scanning near-field optical microscopy tip (Gerster et al., 2012), and to bind PSI to carbon nanotubes (Kaniber et al., 2010). Another modification of PSI is represented by the attachment of plasmonic metal nanoparticles, which resulted in enhanced light absorption (Carmeli et al., 2010) even in the green part of the solar spectrum that is not absorbed normally (Szalkowski et al., 2017). This suggests that it is possible to enhance light absorption by PSI in vitro through the attachment of abiotic components that act as optical antennae to extend the spectrum of photons available for P700 activation.

Taken together, these efforts demonstrate that PSI-based photocurrent-generating systems are still in an exploratory phase, with many variants under development. In the next phase, viable building blocks and reference systems should be established that can serve as starting points for the systematic engineering of superior systems. This will require the modification of PSI for optimal effect in artificial systems and undoubtedly will differ substantially from the original environment in which PSI was molded by biological evolution.

CONCLUSION AND OUTLOOK

Enhancing photosynthesis and growth can be achieved by a simple overexpression of certain photosynthetic proteins, and PSI can be coupled to previously unconnected biotic or abiotic components to generate valuable compounds, hydrogen, or electricity. Moreover, entire photosynthetic complexes can be expressed functionally in distantly related species (as demonstrated for Rubisco; Aigner et al., 2017). Thus, what are the next goals and which challenges need to be overcome? Two challenges for the in vivo systems are obvious: (1) enhancing the efficiency of in vivo bio-bio hybrids; and (2) upscaling the size of synthetic photosynthetic modules to cover entire photosystems or complex antenna systems.

With respect to the enhancement of in vivo cyanobacterial and algal systems harboring hybrid configurations in which photosynthesis directly drives previously unconnected pathways, the use of laboratory evolution offers a unique opportunity to tailor these systems for their intended purpose. Laboratory evolution utilizes the high rate of evolution typical in microbial systems (in particular, if suitable selection conditions can be designed) to fine-tune and optimize processes through the selection of advantageous genetic variation. While this strategy has been employed in E. coli and yeast with impressive success, now photosynthetic microbial systems are emerging as attractive targets for this approach (Leister, 2018).

The exchange of entire multiprotein complexes between distantly related species will not be a simple exercise, because the frozen metabolic state of the core photosynthetic complexes will necessitate the exchange of major parts of photosystems rather than individual subunits. The Rubisco case study by Aigner et al. (2017) represents a promising proof of principle, but one needs to consider that the synthetic Rubisco module comprised only the two different subunits present in the mature complex and five auxiliary factors for its assembly. Entire photosystems will require much larger synthetic photosynthetic modules, and many of the auxiliary factors have not been identified yet. Therefore, when designing these complex synthetic photosynthetic modules, we will identify the set of components that are sufficient (and not only necessary) to drive photosynthesis, providing an unprecedentedly deep understanding of this fundamental and complex process.

Given that the problems described above can be solved, what will be the next steps in the synthetic biology of the light reactions of photosynthesis (see Outstanding Questions)? The combination of synthetic photosynthetic modules from diverse species could allow the design of novel variants of photosynthesis. Numerous instances of such recombined photosynthetic variants can be imagined, including plants that employ cyanobacteria-derived phycobilisomes for highly sufficient photosynthesis under low-light conditions, cyanobacteria that employ plant-derived LHCs as antennae to shift the light saturation of photosynthesis to higher intensities, or the integration of cyanobacterial chlorophyll d and chlorophyll f (which can absorb far-red and near infrared light) into algal or plant photosynthesis to expand the spectral region available to drive photosynthesis (Loughlin et al., 2013; Ho et al., 2016). Moreover, such variants of recombined photosynthesis could be optimized further by laboratory evolution within a suitable microbial chassis. However, such recombined photosynthetic variants cannot be considered a truly novel type of photosynthesis because they would only bring together preexisting pieces that evolution has separated. Nevertheless, they could be an important step toward the ambitious goal to design truly novel synthetic photosynthetic modules that contain more efficient substitutes of the frozen metabolic accidents discussed above.

Ample proof of functionality for in vivo hybrids (between photosynthesis and previously unconnected metabolic pathways) and in vitro hybrids (between PSI and biotic or abiotic catalysts) has been obtained, but such systems will only be commercially successful if they can compete with established nonbiological systems. Tailoring of PSI in cyanobacteria, where techniques like gene replacement and gene modification are routine, may contribute to further improving the efficiency and robustness of in vitro and in vivo hybrids. One promising route could be to identify those parts of the original PSI (which evolved under constraints imposed by the cyanobacterial cell) that are necessary for hybrid devices (a minimal PSI). Such bio-nano systems can be optimized to include novel nonbiological components that replace or complement natural pigments and/or the protein-based backbone of photosystems, going far beyond the limited toolbox of nature’s chemistry. Such novel systems, inspired by natural photosynthesis, could be used to engage biologists in the design of fundamentally different types of photosynthesis in living organisms.

Supplemental Data

The following supplemental materials are available.

• Supplemental Table S1. Absence/presence of photosynthetic proteins in the three model species Synechocystis PCC6803, C. reinhardtii, and Arabidopsis.