Biochemical and synthetic biology approaches to improve photosynthetic CO₂-fixation

Abstract

There is an urgent need to improve agricultural productivity to secure future food and biofuel supply. Here, we summarize current approaches that aim at improving photosynthetic CO₂-fixation. We critically review, compare and comment on the four major lines of research towards this aim, which focus on (i) improving RubisCO, the CO₂-fixing enzyme in photosynthesis, (ii) implementing CO₂-concentrating mechanisms, (iii) establishing synthetic photorespiration bypasses, and (iv) engineering synthetic CO₂-fixation pathways.

Introduction

The Morrow plots are a landmark of the University of Illinois. They are an experimental corn field that is continuously farmed since 1876 [1]. During the last 140 years, and in particular since the 1950s, crop yield on the Morrow fields (and world-wide) have increased by at least a factor of three [1,2]. Yet, these past achievements in agriculture are challenged by several developments. (i) The current population increase is not matched by the current increase in agricultural productivity, (ii) there is a growing demand to use crops for biofuel and biomass production directly competing with food production, and (iii) global CO₂-emissions are continuously rising, accelerating the effects of climate change, including the loss of arable land, increased flooding and droughts. As a consequence, there is an urgent need to further improve agricultural productivity. Because in plants the conversion of light into biomass is the process with the lowest energy conservation (approx. 1%), improving photosynthetic CO₂-fixation has been identified as key to increase agricultural productivity [3].

Under optimum conditions, one limiting factor in photosynthetic CO₂-fixation is flux through the Calvin cycle, which is often restricted by the activity of the cycle’s CO₂-fixing enzyme, ribulose-1,5-bisphosphate carboxylase/oxygenase (RubisCO). The turnover number of an average RubisCO is between 1 and 10 s^-1 (http://brenda-enzymes.org). This is one to two orders of magnitude below the turnover frequencies of other enzymes in central carbon metabolism that lie on average around 50 to 100 s^-1 [4]. To allow for sufficient CO₂-fixation rates, the low activity of RubisCO is compensated by high expression levels of the enzyme. In a photosynthetic organism, RubisCO can make up to 50% of the soluble protein [5].

Besides showing low specific activity, RubisCO does not discriminate well between O₂ and CO₂, which results in an oxygenase side reaction of the enzyme. Due to the high O₂:CO₂ ratio of ambient air (approx. 500:1), an average RubisCO fixes up to two O₂ every five CO₂-fixation reactions [6]. The products of RubisCO’s side reaction are 3-phosphoglycerate (3-PG) and glycolate-2-phosphate (G2P). The latter is a toxic compound that needs to be removed or recycled. In photosynthetic organisms, this is achieved at the expense of additional energy, reducing power and fixed CO₂ in a process called photorespiration (Figure 1, 2). It is estimated that up to 30% of the photosynthetic output is lost through photorespiration [6,7].

Figure 1. Overview of photosynthetic CO₂-fixation, photorespiration and current engineering efforts.

Figure 2. Natural and synthetic photorespiration bypasses.

To increase the yield of photosynthetic CO₂-fixation, different strategies were suggested and at least partially pursued (Figure 1). These fall into one of the following four general categories: (i) improving the catalytic properties of RubisCO, (ii) improving the working conditions of RubisCO through CO₂-concentrating mechanisms (CCM), (iii) engineering synthetic photorespiration bypasses, and (iv) engineering synthetic CO₂-fixation pathways.

Improving the catalytic properties of RubisCO

Initial approaches to improve photosynthetic CO₂-fixation focused on identifying [8,9] or engineering [10,11] RubisCOs with higher CO₂-specificities and/or higher catalytic rates. These efforts have only met with limited success, because it has become apparent that RubisCO is trapped in an inherent trade-off between activity and specificity. Higher specificity for CO₂ usually results in a lower enzyme activity. Vice versa, to engineer a RuBisCO with higher activity its specificity for CO₂ has to be sacrificed, resulting in a higher oxygenation rate [12,13].

The reasons for the observed trade-off lie in the evolutionary past of RubisCO, although the emergence of the enzyme’s carboxylation and oxygenation function remain unknown. Recent investigations on the RubisCO superfamily [14,15] suggest that RubisCO was not a CO₂-fixing enzyme a priori, but rather that its carboxylation function evolved as a secondary function in the protein scaffold of primordial enolases [16,17]. These findings challenge and extend older theories according to which RubisCO evolved from a primordial carboxylating archaeal enzyme [18,19]. Independent of the true origins of the carboxylation reaction of RubisCO, it is undisputed that the evolutionary roots of RubisCO trace back to a time when the level of O₂ in the atmosphere was minimal. Thus, ancient RubisCO was primarily selected for promoting the carboxylation of ribulose-1,5-bisphosphate, but not against suppressing the oxygenation side reaction. This primordial chemistry of the enzyme caused (and still underlies) the inverse coupling of activity and selectivity in RubisCO [12,13]. With the increase of atmospheric O₂ during earth’s history, RubisCO evolved along these two parameters, which enabled adaption of a given enzyme towards specificity or activity (depending on its environmental and/or organismic context) [20], but did not allow for an uncoupling of the two opposing catalytic parameters.

Nevertheless, even only slightly improved RubisCOs could have measurable effects. It has been calculated that transplantation of a RubisCO from the red algae Griffithsia monilis into crop could increase carbon gain by 25%, because of a twofold increased CO₂-selectivity/activity ratio of the enzyme compared to plant RubisCO [21]. Screening of RubisCOs from wild wheat grasses identified enzyme variants of improved CO₂-specificity/activity ratio. These “wild enzymes” were calculated to increase carbon uptake rates by 20% upon substitution of native RubisCO in agriculturally used wheat [22].

A first step into this direction was the replacement of native RubisCO of Nicotiana tabaccum through faster homologs from the alphaproteobacterium Rhodospirillum rubrum [23] and the cyanobacterium Synechoccocus elongatus [24], although the transgenic plants were only able to grow under highly elevated CO₂-concentrations and showed severe growth deficits compared to the corresponding wild-types. Redesign of the S. elongatus transgene recently restored wild-type-like growth, still under elevated CO₂ atmosphere, but notably at tenfold lower RubisCO levels compared to the wild-type [25]. This shows that it is in principle possible to transplant exogenous RubisCOs with improved catalytic properties into plants. Yet, it remains to be demonstrated, whether improved RubisCOs alone would be actually able to substantially increase photosynthetic yield under field conditions.

Improving the working conditions of RubisCO through CCMs

Instead of improving the catalytic parameters of RubisCO, an alternative approach to increase photosynthetic productivity centers on changing the working conditions of the enzyme. By increasing the CO₂ concentration around RubisCO, the oxygenation side reaction of the enzyme can be effectively suppressed, which in turn enables faster CO₂-fixation rates, resulting in an increased CO₂-fixation efficiency.

Different CO₂-concentrating mechanisms (CCMs) emerged naturally during evolution [26]. In C4-plants, CO₂ is pre-fixed in special cells or dedicated compartments into a C4-acid, which is transported to the place of RubisCO, where it is decarboxylated again, thereby increasing the local CO₂:O₂ ratio around the enzyme [27]. Cyanobacteria on the other hand evolved several HCO₃ ^- transporters and CO₂-uptake systems that enable them to concentrate up to 40 mM HCO₃ ^- intracellularly [28]. In addition, cyanobacteria feature carboxysomes [29], proteinaceous compartments that are filled with RubisCO [30] and carbonic anhydrase [31,32]. These compartments allow the selective influx of HCO3^- and ribulose-1,5-bisphosphate through pores. HCO₃ ^- is converted into CO₂ and retained within the carboxysome, so that CO₂-fixation takes place in a dedicated compartment under a highly enriched CO₂ environment.

Current crop production relies mainly on plants that do not possess any of the known natural CCMs. More than 80% of the agricultural land used in crop production is covered by plants that lack CCMs, such as rice, potato, wheat, and barley [33]. First efforts that focused on transplanting CCMs into rice demonstrated that it is not sufficient to simply import the enzyme machinery of C4-plants [34]. Thus, current strategies that are pursued by different consortia (http://C4rice.irri.org; http://www.3to4.org/) aim at mimicking cell-specific expression patterns of C4-CCM genes [35] and introducing the structural and anatomical characteristics of C4-plants into CCM-free crops [36], which apparently is a long-term challenge.

Another line of research aims at introducing carboxysomes into chloroplasts of CCM-free crops, which is predicted to improve yield by up to 60% under hot and dry conditions [37]. Functional carboxysomes were already reconstituted in Escherichia coli [38] demonstrating the potential for robust self-assembly in foreign hosts. The transient expression of several carboxysome subunits in Nicotiana benthaniama at least resulted in the formation of organized structures that resembled empty microcompartments [39]. Together with the fact that carboxysomal RubisCO from S. elongatus is known to be functionally expressed in tobacco [24,25] (see above), these studies might pave the way to produce functional chloroplastic carboxysomes in the future. If the introduction of such complex CCM will actually be beneficial or rather a burden in planta remains to be seen, given the fact that the number of carboxysomes per chloroplast required is still unclear [40]. To reduce the genes necessary for a functional chloroplastic carboxysome, it might become necessary to streamline the assembly process by protein domain fusions [41]. Finally, the supply of these “compartments within compartments” with sufficient CO₂ will be crucial [37]. The expression of HCO₃ ^- transporters and carbonic anhydrase in chloroplasts [42,43] could provide a solution to this problem.

Engineering synthetic photorespiration bypasses

The possibility to increase photosynthetic CO₂-fixation yield by improving photorespiration has gained considerable interest in recent years. Natural photorespiration is a costly process that involves multiple enzyme reactions, which are located in different organelles in plants. Canonical photorespiration recycles two G2P molecules into one 3-PG molecule (Figure 2), while releasing one molecule of CO₂ and NH₃ during this process. The recycling of G2P and in particular the re-fixation of the lost CO₂ requires input of a considerable amount of energy and reducing power. Several alternative photorespiration bypasses, based on existing routes have been suggested that are advantageous compared to the natural process in terms of either ATP requirement, reducing potential, carbon stoichiometry, or the number of cellular compartments involved (Table 1).

Table 1.

Comparison of natural and synthetic photorespiration bypasses. Due to the different topologies, the synthetic photorespiration bypasses cannot be simply compared side-by-side. The expected advantages are of multifactorial nature and more than a simple sum of redox equivalents and ATPs comsumed. However, when normalized onto total carbon stochiometry (i.e., the total requirements to regenerate a C₃ intermediate and the net fixation of one CO₂) the individual photorespiration bypasses can be balanced as followed. Advantages compared to the canonical (i.e., natural) photorespiration bypass are highlighted in blue, disadvantages are marked in red.

	Canonical photorespiration bypass	Chloroplastic glycerate bypass	Chlorplastic glycolate oxidation bypass	Peroxisomal glycerate bypass	3-hydroxyx-propionate bypass
Enzymes required	12	5	5	7	13
ATP consumed	8	7	7	9	7
Redox power consumed	4 NAD(P)H + 2 Fd (2,200 mV)	5 NAD(P)H (1,700 mV)	4 NAD(P)H (1,360 mV)	5 NAD(P)H (1,700 mV)	3 NADPH (1,020 mV)
CO2 release (% carbon of glycolate)	Yes (25%)	Yes (25%)	Yes (100%)	Yes (25%)	No (0%)
Place of CO2 release	mitochondria (far RubisCO)	chloroplast (near RubisCO)	chloroplast (near RubisCO)	peroxisome (far RubisCO)	no CO2 released
Transport across organelle membranes	4	0	0	2	0
NH3 release	Yes	No	No	No	No
Way of CO2-re-fixation	Calvin cycle	Calvin cycle	Calvin cycle	Calvin cycle	Included in bypass
Turns of Calvin cycle required to refix CO2	2 turns	2 turns	3 turns	2 turns	none

In the chloroplastic glycerate bypass [44], two molecules of G2P are converted under loss of one CO₂ into one molecule of glycerate, which is fed back into the Calvin cycle (Figure 2). The entire process circumvents the release of NH₃, consumes less ATP, and conserves reducing power. Since the whole process was designed to take place in the chloroplast, the CO₂ is released in vicinity of RubisCO, reducing its oxygenation side reaction. When the glycerate pathway from E. coli was introduced into Arabidopsis thaliana [44] or Camelina sativa [45] chloroplasts, transgenic plants showed an enhanced photosynthesis, faster growth and higher biomass generation. However, transgenic lines only expressing glycolate dehydrogenase in the chloroplasts, showed similar results [44,45]. Thus, the role and the fate of the glyoxylate that is produced in the chloroplasts of these transgenic plants is not quite clear.

The peroxisomal glycerate bypass [46] is based on the conversion of glycolate into glycerate in peroxisomes (Figure 2). It bypasses NH₃ release, and conserves reducing power. The pathway originally from E. coli could be implemented only partially in Nicotiana tabacum. Transgenic tobacco lacked expression of one of the key enzymes, and plants stunted growth under ambient air [46].

In the chloroplastic glycolate oxidation bypass [47], G2P is converted into glycolate, which is subsequently completely oxidized into CO₂ within the chloroplast (Figure 2). This pathway bypasses the release of NH₃ and conserves reducing power. A huge disadvantage is that all of the carbon is lost instead of “only” one out of four, as in the other pathways. When experimentally realized by redirecting peroxisomal glycolate oxidase and catalase to the chloroplast, transgenic A. thaliana were demonstrated to support higher dry weight and photosynthetic rates. This effect was significant under energy-limiting growth conditions [47].

In contrast to above strategies that all release CO₂, the proposed 3-hydroxypropionate bypass [48] leads to a net fixation of CO₂ during synthetic photorespiration by converting G2P into pyruvate (Figure 2). The 3-hydroxypropionate bypass was successfully realized in the cyanobacterium S. elongatus through hetereologous expression of seven enzymes from the filamentous anoxygenic phototroph Chloroflexus aurantiacus and the betaproteobacterium Accumulibacter phosphatis. All enzyme activities were successfully demonstrated. A growth phenotype, however, was not observed, most probably because S. elongatus possesses already very efficient CCMs [48]. Transplantation of the 3-hydroxyproionate bypass into CCM-deficient strains of S. elongatus may provide a growth advantage and proof this concept.

Even though the engineering of synthetic photorespiration bypasses has already shown promising results, all projects so far were based on grafting naturally occurring pathways into photosynthetic hosts. To overcome this restriction onto natural solutions, a new European research initiative aims at systematically exploring and engineering completely artificial routes of higher efficiency in a true synthetic biology effort by combining enzyme engineering and metabolic retrosynthesis (http://FutureAgriculture.eu).

Engineering synthetic CO₂-fixation

The most ambitious approach to improve photosynthetic yield is to completely rewire CO₂-fixation in plants, algae and cyanobacteria. This research is inspired by the discovery that during the course of evolution nature itself has invented five alternative CO₂-fixation pathways to the Calvin cycle, which operate in different bacteria and archaea [49–54]. These “alternative” microbial CO₂-fixation pathways are not based on RubisCO [55] and several of them show advantages in respect to energy requirement and efficiency compared to the Calvin cycle [56]. The reconstitution of natural existing CO₂-fixation pathways in model organisms, however, has not proven successful so far [57], probably due to the complex interplay and interference with the host’s native carbon and energy metabolism.

Even more progressive are synthetic biology approaches that are based on the principle of metabolic retrosynthesis. Here, completely novel CO₂-fixation pathways of high efficiency are supposed to be designed through the free recombination of known enzyme reactions [55,58]. These efforts are further fueled by the discovery [59,60] and rational engineering [61] of highly efficient carboxylases, and the general progress in computational enzyme design [62]. The degree of freedom in these synthetic pathways allows tailoring the conversion of CO₂ into virtually any desired product, and their synthetic nature could be advantageous for in vivo transplantations due to a limited interference with natural metabolism. The realization of such synthetic CO₂-fixation pathways and their integration into living organisms still poses several challenges, but will be indispensable for freeing natural photosynthetic CO₂-fixation from its inherent disadvantages, and transforming biology from a tinkering science into a truly synthetic discipline. Compared to all other strategies discussed here, this approach holds the most promise to substantially improve photosynthetic productivity on a long-term perspective.