RubisCO selection using the vigorously aerobic and metabolically versatile bacterium Ralstonia eutropha

Abstract

Recapturing atmospheric CO₂ is key to reducing global warming and increasing biological carbon availability. Ralstonia eutropha is a biotechnologically useful aerobic bacterium that uses the Calvin-Benson-Bassham (CBB) cycle and the enzyme ribulose 1,5-bisphosphate carboxylase/oxygenase (RubisCO) for CO₂ utilization, suggesting that it may be a useful host to bioselect RubisCO molecules with improved CO₂-capture capabilities. A host strain of R. eutropha was constructed for this purpose after deleting endogenous genes encoding two related RubisCOs. This strain could be complemented for CO₂-dependent growth by introducing native or heterologous RubisCO genes. Mutagenesis and suppressor-selection identified amino-acid substitutions in a hydrophobic region that specifically influences RubisCO’s interaction with its substrates, particularly O₂, which competes with CO₂ at the active site. Unlike most RubisCOs, the R. eutropha enzyme has evolved to retain optimal CO₂-fixation rates in a fast-growing host, despite the presence of high levels of competing O₂. Yet its structure-function properties resemble those of several commonly found RubisCOs, including the higher-plant enzymes, allowing strategies to engineer analogous enzymes. Because R. eutropha can be cultured rapidly under harsh environmental conditions (e.g., with toxic industrial flue gas), in the presence of near saturation levels of oxygen, artificial selection and directed evolution studies in this organism could potentially impact efforts towards improving RubisCO-dependent biological CO₂ utilization in aerobic environments.

Introduction

An increase in the number of published manuscripts and patents filed in the past decade clearly signifies R. eutropha strain H16 (Cupriavidus necator; formerly Alcaligenes eutrophus), a gram-negative β-proteobacterium, as a model organism ideally suited for both industrial applications and basic science investigation. Its genome has been sequenced and key metabolic signatures suggest flexibility in the organism’s bioenergetics [1, 2]. R. eutropha is best known for its ability to use H₂ and CO₂ as sole sources of energy and carbon under vigorous aerobic growth conditions in the absence of an organic carbon substrate [2]. It is uniquely suited to alternate between or concomitantly use heterotrophic (with sugars, organic acids, fatty acids or plant oils) or autotrophic (using CO₂ or formate) growth modes to drive cellular metabolism and the production of polyhydroxyalkanoates (PHAs). Industrial scalability of growth conditions and genetic amenability make R. eutropha an attractive host for use in the bioplastic and biofuel industries [3,4].

The CBB cycle contributes to primary production in plants, algae and various autotrophic microorganisms. RubisCO, a key enzyme of the CBB cycle, catalyzes autotrophic CO₂ fixation and often limits primary productivity. Being the world’s most abundant enzyme, RubisCO’s rate-limiting nature is primarily attributable to low k_cat values (1–13 per sec) and an alternative competing oxygenation reaction with substrate O₂, which leads to a wasteful photorespiratory pathway [5, 6]. Modeling studies have concluded that RubisCO catalytic properties are optimized for performance in specific species and environmental conditions [7–9], but several efforts are underway to re-optimize whole-organism carbon-fixation rates by heterologous expression of divergent RubisCOs (whole or parts as hybrids and chimeras) along with accessory factors and CO₂-concentrating mechanisms [10–14]. Thus, it is advantageous to “artificially” evolve RubisCO in heterologous hosts. Rhodobacter capsulatus, a photosynthetic bacterium with natural ability to support CO₂-dependent autotrophic growth, and Escherichia coli have been successfully used as hosts for evolving prokaryotic RubisCO enzymes [15, 16]. Complex assembly requirements, post-translational processing, limited availability of tools to manipulate organelle-specific DNA, and the typically long generation times preclude the use of eukaryotic models other than the green alga Chlamydomonas reinhardtii for RubisCO selection studies [17]. However, considerable progress has been made towards expressing bacterial homologs of the eukaryotic enzymes in tobacco [18,19], which holds promise for introducing artificially evolved bacterial enzymes in higher eukaryotes. This could be facilitated by recent studies in which metagenomic sampling has identified potentially novel RubisCOs from microbial “dark matter” [20–23].

In virtually all bacteria that lack mechanisms to limit O₂-exposure of RubisCO (e.g., carboxysomes), aerobic CO₂-dependent growth is compromised by the oxygenation reaction catalyzed by RubisCO [6]. However, a few bacteria like R. eutropha, which do not have any discernible means to protect RubisCO from O₂, can still fix CO₂ and grow efficiently under rigorous aerobic growth conditions [24]. The genetics and control of CO₂-assimilation has been well studied in R. eutropha. Most CBB-cycle enzymes, including RubisCO, are encoded by two homologous cbb operons, one on chromosome 2 (cbb_c) and the other on an extra-chromosomal 450-kb megaplasmid pHG1 (cbb_p) [25]. CbbR, a LysR-type regulatory protein encoded by the gene present upstream of cbb_c, tightly controls gene expression from both operons by binding to identical promoter sequences [26]. The two copies of RubisCO structural genes (cbbLS) share 97% nucleotide homology, which translates to 99% amino-acid identity among the encoded protein subunits. The x-ray crystal structure has been solved for RubisCO from a closely-related strain of R. eutropha H16; eight large (L; ~55 kDa) and eight small (S; ~15 kDa) subunits constitute a “red-like” form IC holoenzyme [27].

Prior studies have described the use of R. eutropha RubisCO-deficient strains [27–29]. However, it has not been possible to establish a functional RubisCO complementation system with these strains. Strain HF210 is a spontaneous megaplasmid pHG1-free derivative of strain H16, which is capable of organoautotrophic growth with formate as both the carbon and energy source. But the absence of hydrogenases (encoded by pHG1 in strain H16) renders the strain incapable of supporting O₂- and CO₂-dependent growth with H₂ as the sole source of energy, despite strong indications that the expression of cbb genes from cbb_c is derepressed [24, 28]. Strain HB10 was derived from strain HF210 by the insertion of transposon Tn5 within cbbL_c and thus cannot support autotrophic growth. However, the expression of phosphoribulokinase (PRK) (and possibly other CBB genes) was constitutive, suggesting that the native regulation of the cbb operon was lost in this strain [29]. Thus, a clean RubisCO-deletion strain was constructed for RubisCO selection studies. We have found that this strain, which is unable to support CO₂-dependent aerobic growth, could be complemented for growth by reintroduction of either the native or divergent heterologous RubisCO genes. Complementation occurred when these genes were either inserted into the native locus on the chromosome via homologous recombination (for maintaining native regulatory control and copy number) or introduced in trans on a broad-host range plasmid vector. Growth complementation in the R. eutropha RubisCO-deletion strain was exploited for mutagenesis and suppressor selection, which resulted in the identification of amino-acid residues that impacted RubisCO’s CO₂/O₂ selectivity.

Results

Construction of R. eutropha RubisCO-deletion strain (H16ΔLS)

A clean RubisCO-deletion strain was constructed by precisely deleting the cbbL and cbbS genes from both operons in the wild-type strain H16, along with the identical 57-bp intergenic spacer DNA present at both loci. This strain failed to support autotrophic CO₂-dependent growth. Because heterotrophic growth on fructose partially de-represses cbb gene expression [24], the deletion strain was grown on fructose medium to confirm unperturbed expression of other cbb genes. However, the cell extracts lacked PRK activity (Table 1). The gene cbbP, which encodes for PRK, is located downstream from the cbbLS genes in both operons. Hence the lack of PRK activity was an indication that the expression of other cbb genes were likely affected. This could be attributed to the presence of a stem-loop structure immediately downstream of cbbS [30] which, when present adjacent to the promoter region in the RubisCO-deletion strain, could have resulted in the complete repression of cbb gene expression. A modified RubisCO-deletion strain (H16ΔLS) was thus constructed by removing this stem-loop structure, which restored wild-type levels of PRK activity (Table 1).

Table 1.

PRK and RubisCO specific activities in RubisCO-deletion strains. Activities were measured from soluble-cell extracts of R. eutropha strains grown on heterotrophic medium with fructose as the sole carbon source. Cell extract from wild type (strain H16) was used as a positive control in these experiments.

Strain	Specific Activity (nmol/min-mg)
	PRK
H16	118
H16gmΔLS	0
H16ΔLS	196
	RubisCO
H16	73
H16gmΔLS	0
H16ΔLS	0

Complementation of H16ΔLS with native and heterologous RubisCO genes

Using a 4.5-kb homologous cbb region or a broad-host range plasmid p90, the native RubisCO sequence (cbbLS_Re) was reintroduced into R. eutropha strain H16ΔLS, resulting in CO₂-dependent autotrophic growth (Fig. 1). Using an identical strategy, the form I (cbbLS_Rp) or form II (cbbM_Rp) RubisCO sequences from Rhodopseudomonas palustris [31], or the form I (rbcLS_Se) sequence from Synechococcus elongatus strain PCC 6301 [15], or the form III (rbcL2_Af) sequence from the archaeon Archaeoglobus fulgidus [32] were each introduced into H16ΔLS (Fig. 1A). The cbbLS_Rp genes, encoding a structurally similar form I enzyme (76% identity and 87% similarity at the amino acid level), or the cbbM_Rp sequence, encoding a divergent form II enzyme (small subunits absent; large subunit is only ~32% identical), could both complement for RubisCO-dependent growth (Fig. 1C). However, the rbcLS_Se sequence encoding a divergent form I enzyme from Synechococcus (60% identity and 72% similarity) or the archaeal rbcL2_Af gene encoding an oxygen-sensitive form III enzyme, both failed to complement strain H16ΔLS for RubisCO function (Fig. 1C). The presence of substantial levels of RubisCO activities in fructose-grown cultures of all the complemented strains confirmed that the expression of heterologous genes was likely unaffected in these strains. It could be concluded from this analysis that the R. eutropha RubisCO-deletion strain is indeed useful for the expression and aerobic-selection of heterologous enzymes with functionally compatible properties.

Fig. 1.

Complementation of R. eutropha strain H16ΔLS with native and heterologous RubisCOs. (A) Schematic of a ~4.5-kb region amplified from cbb_c operon in R. eutropha strain H16, which includes the coding sequences for cbbR, cbbL, cbbS, and cbbX. This product was used as a template for replacing cbbLS_Re genes with heterologous RubisCO-encoding genes, utilizing the engineered MfeI or a ClaI, and SpeI recognition sequences. The form-I sequences from R. palustris (cbbLS_Rp) and S. elongatus strain PCC 6301 (rbcLS_Se) both included the native intergenic spacer DNA from the respective organisms. The ~1.4-kb R. palustris cbbM_Rp gene and the ~1.3-kb A. fulgidus rbcL2_Af gene both encode form II and form III enzymes, respectively. (B) Vector map of the broad-host range plasmid, p90, which was used for trans expression of RubisCO genes in R. eutropha. It features the native cbb promoter (red; between NcoI and SacI sites) to drive the expression of genes cloned into the multiple-cloning site (green; from SacI to HindIII). (C) Autotrophic growth phenotypes conferred by the RubisCO genes that had been recombined into the cbb_c locus in R. eutropha strain H16ΔLS. For all samples, data points represent average values obtained from triplicate cultures, along with the respective standard deviations. Data is representative of two independent experiments that gave similar results.

Site-directed mutagenesis of R. eutropha RubisCO

In previous studies using an R. capsulatus RubisCO-deletion strain, mutant substitutions F342V and A375V were identified in a conserved hydrophobic region adjacent to the active site of Synechococcus RubisCO, which conferred “positive” growth phenotypes [15,33]. Based on residue substitutions in at least three divergent RubisCO enzymes, it could be concluded that this region is critical for determining RubisCO’s differential interactions with the substrates CO₂, O₂ and RuBP [15, 32, 33]. Because structural alignments indicated a striking pattern of conservation in this region adjoining the RubisCO active site (Fig. 2), site-directed mutagenesis was used to create R. eutropha RubisCO mutants Y347V and A380V, analogous to the F342V and A375V mutants of Synechococcus RubisCO. Growth complementation with mutant Y347V was indistinguishable from wild type but mutant A380V was unable to complement R. eutropha H16ΔLS for aerobic CO₂-dependent growth (Fig. 3A). Both mutants could however complement the R. capsulatus RubisCO-deletion strain for anaerobic CO₂-dependent growth (Fig. 3A). Catalytic properties of recombinant wild-type and Y347V-mutant enzymes are comparable (Table 2), which is consistent with the observed growth complementation phenotypes. The carboxylation k_cat value of A380V-mutant enzyme was substantially lower than the wild-type value and the mutant enzyme was relatively oxygen-insensitive as evinced by its high K_o value (Table 2). The analogous A375V-mutant Synechococcus RubisCO has similar properties [33] and hence could not complement R. eutropha H16ΔLS for autotrophic growth, although the mutant F342V could (Fig. 3). Based on these results, it can be concluded that CO₂-dependent growth complementation in R. eutropha entails a unique interplay of assembly and structure-function properties of the introduced RubisCO.

Fig. 2.

Amino-acid conservation in a hydrophobic region near the active site of RubisCO. (A) Primary-structure alignment of amino-acid residues in a hydrophobic region adjacent to the active site in divergent form I, form II and form III RubisCO enzymes. Residues are numbered relative to the R. eutropha sequence. (B) Comparison of analogous hydrophobic regions in the crystal structures of R. eutropha (left; 1BXN) and S. elongatus (right; 1RBL) form I RubisCO enzymes. Residues targeted for site-directed mutagenesis are colored red, Thr-330 and Tyr-348 identified via suppressor-selection in R. eutropha are colored green and the invariant active-site Ser residue (381 in R. eutropha & 376 in S. elongatus) is colored blue in both (A) and (B). Dotted black line marks a hydrogen bond. Active site is marked with phosphate ions or CABP (black). All other residues in the hydrophobic region, and those within 4 Å of residues colored red (shown in stick representation), are colored yellow.

Fig. 3.

Complementation phenotypes of R. eutropha and Synechococcus RubisCO site-directed mutants. (A) Complementation of R. eutropha RubisCO (cbbLS_Re) site-directed mutants, Y348C (blue) and A380V (red), in RubisCO-deletion strains of R. eutropha (strain H16ΔLS) (left; aerobic chemoautotrophic growth) and R. capsulatus (strain SB I/II^-) (right; anaerobic photoautotrophic growth). R. eutropha complementation utilized plasmid p90 and R. capsulatus complementation utilized plasmid pRPS-MCS3 [15]. Values for R. eutropha growth complementation represent mean and standard errors of absorbance measurements from triplicate cultures. (B) Ability of various Synechococcus RubisCO mutants (rbcLS_Se) described in previous studies [15, 33] to complement for CO₂-dependent growth in R. eutropha (strain H16ΔLS). R. eutropha wild type was used as a positive control.

Table 2.

Kinetic properties of histidine-tagged recombinant site-directed mutants of R. eutropha RubisCO (encoded by the megaplasmid cbbLS_Re genes)

Enzyme	kcata (s−1)	Kca (μM)	Koa (μM)	Ko/Kcb
Wild type	3.84 ± 0.54	37 ± 4	1149 ± 56	31
Y347V	4.14 ± 0.66	35 ± 1	1139 ± 93	33
A380V	0.25 ± 0.04	34 ± 2	1435 ± 105	42

^aValues are mean ± standard deviation (n-1) from independent assays with three separate enzyme preparations

^bCalculated values

Random mutagenesis and suppressor selection with A380V-mutant R. eutropha RubisCO

The negative phenotype of mutant A380V offered an opportunity to utilize the strain H16ΔLS for suppressor selection, with an ultimate goal to identify unique structural regions and amino-acid residues that contribute to catalysis. In vitro random mutagenesis, starting with the DNA sequence encoding the negative-mutant A380V enzyme, followed by conjugation of the mutagenized DNA library into strain H16ΔLS and autotrophic-growth selection, resulted in two types of suppressor-mutant colonies. All colonies that could support wild-type levels of growth resulted from an intragenic suppressor with a tyrosine-to-cysteine substitution at position 348. All the remaining colonies, which grew consistently slower than the wild type, had an identical mutation resulting in a threonine-to-alanine substitution at position 330 of the A380V-mutant protein. These suppressors, either when present in isolation or with the A380V-negative mutant substitution, could support CO₂-dependent liquid autotrophic growth of strain H16ΔLS (Fig. 4A). SDS-PAGE and RubisCO activity assays performed with cell extracts of these liquid cultures indicated that substantial amounts of functional RubisCO were synthesized in all mutant strains (Fig. 4, B and C). Wild-type levels of all recombinant mutant proteins could be purified after expression in E. coli. It thus seems likely that the mutant substitutions did not have deleterious effects on protein assembly or stability in vivo.

Fig. 4.

Phenotypes of R. eutropha RubisCO mutants. (A) Autotrophic growth phenotypes of R. eutropha mutant A380V and its suppressors relative to wild-type cbbLS_Re genes in strain H16ΔLS. Data with mean and standard errors calculated from triplicate samples are representative of two independent experiments. (B) SDS-PAGE of soluble-crude cell extracts prepared from autotrophically grown R. eutropha cultures complemented with various mutant plasmids. Lane 1 has molecular weight standards. Samples in lanes 2–6 were prepared from strains expressing wild-type, and mutants T330A, Y348C, T330A/A380V, and Y348C/A380V, respectively. The ~55-kDa large- and the ~15-kDa small-subunits are indicated. (C) RubisCO specific activities in soluble cell-extracts prepared as described in (B). Values represent mean and standard errors calculated from triplicate samples.

Catalytic properties of recombinant wild-type and mutant R. eutropha enzymes

Wild-type and mutant RubisCO enzymes were purified as recombinant proteins from E. coli and their catalytic properties were determined as described previously [31, 33]. The CO₂/O₂ specificity factor (Ω) and the K_RuBP values obtained for the wild-type enzyme are both comparable to the values previously reported (Table 3) [34, 35]. Owing to the anaerobic assay conditions employed in this study, the wild-type enzyme’s K_m value for CO₂ (K_c; 34 μM) is significantly lower than 66 μM, which was obtained from aerobic assays (i.e., with ~21% O₂) [35]. Besides having a K_c value similar to that of plant RubisCOs, the wild-type enzyme’s high K_o value (947 μM) (Table 3), which is close to the concentration of an atmosphere saturated with O₂ (~1200 μM), likely accounts for this enzyme’s ability to support CO₂-dependent growth of R. eutropha in the presence of gas mixtures with up to 80% oxygen [24].

Table 3.

Kinetic properties of recombinant wild-type and mutant R. eutropha RubisCO enzymes (encoded by the chromosomal cbbLS_Re genes)

Enzyme	Ωa VcKo/VoKc	kcata (s−1)	Kca (μM)	Koa (μM)	Ko/Kcb	Vc/Vob	KRuBPa (μM)
Wild type	85±4	2.5±0.8	34±4	947±297	28	3.0	34±8
A380V	73±3	0.2±0.1	35±6	2276±496	65	1.1	939±161
T330A	82±2	2.8±0.3	44±4	1087±159	25	3.3	29±8
Y348C	73±7	1.6±0.1	24±2	674±117	28	2.6	36±8
T330A/A380V	78±5	0.7±0.1	28±1	1069±113	38	2.1	769±18
Y348C/A380V	61±6	0.7±0.2	20±2	652±97	33	1.8	581±42

^aValues are mean ± standard deviation (n-1) from independent assays with three separate enzyme preparations

^bCalculated values

Notably, the A380V substitution resulted in an enhanced K_o value relative to that of the wild type, similar to what was observed with the analogous A375V-mutant Synechococcus RubisCO [33] (Table 3 and Fig. 5). A ~3.5-fold higher k_cat value relative to that of the A380V negative-mutant enzyme likely explains the primary mechanism of suppression by the T330A or Y348C substitutions (Fig. 4A; Table 3). The K_c values of both the double-mutant and the Y348C single-mutant enzymes were significantly lower than the wild-type value, which indicates an improved interaction with CO₂ during catalysis. The lower K_c values also resulted in lower K_o values of the Y348C single-mutant and the double-Y348C/A380V double-mutant enzymes; the oxygen sensitivity was unperturbed by the T330A-mutant substitution. The increases in K_o/K_c ratios of both double-mutant enzymes were offset by the reduced V_c/V_o ratios, explaining why the Ω values of these enzymes were not higher than the wild-type value (Table 3). The substantially higher K_RuBP value of the A380V-mutant enzyme was barely fixed by the T330A-mutant substitution, possibly accounting for the slower growth of the T330A/A380V-double mutant; the Y348C-mutant substitution resulted in a ~40% lower K_RuBP value for the Y348C/A380V-double mutant enzyme (Table 3). It is clear from this analysis that all the mutant substitutions affect the ability of the enzyme to interact with one or more of the substrates, underpinning the utility of this system to engineer favorable substrate interactions and enhance activity after random mutagenesis under subsequent specific selection conditions (e.g., at high O₂/CO₂ ratios).

Fig. 5.

Competitive inhibition of R. eutropha RubisCO by O₂. Michaelis-Menten substrate (at varying CO₂ concentrations) vs. velocity curves obtained with R. eutropha wild-type (black) or A380V-mutant (red) enzymes in simultaneous assays performed under anaerobic (0% O₂; solid lines) or O₂-saturated (100% O₂; broken lines) conditions, showing reduced O₂-sensitivity of the mutant A380V RubisCO. Initial velocity for each data point was obtained from 250-μl assays with either 2.8 μg of the wild-type or 11 μg of A380V-mutant enzyme.

Discussion

There has been considerable interest in using directed evolution strategies to improve the expression levels, assembly and kinetic performance of RubisCO so that the rates for net carboxylation may be maximized [5, 7–12, 14–17, 36–38]. The presence of an efficient lithoautotrophic metabolism has attracted recent interest in tapping R. eutropha for biotechnological CO₂ utilization [3, 4, 39, 40]. The complementable R. eutropha strain H16ΔLS described here can now be exploited for RubisCO-selection studies, especially under robust aerobic growth conditions. The proof of concept has been demonstrated by using this strain to functionally express heterologous RubisCO enzymes, and by using it to obtain suppressors of a negative-mutant RubisCO (Figs. 1, 3, and 4). This system shares several advantages with a previously-described bioselection system utilizing a RubisCO-deletion strain of R. capsulatus. These include the ability to express diverse RubisCO genes from other bacteria and archaea [15, 31, 32] and the potential for direct functional selection of RubisCO from metagenomic DNA [21]. R. eutropha grows faster than R. capsulatus and appears to have unique requirements for aerobic CO₂-dependent growth, as demonstrated by the different phenotypes of the mutant A380V when present in either of the host strains (Fig. 3A). The two systems should thus be able to complement each other for identifying those enzymes not compatible with either of the hosts. This is advantageous for studies involving metagenomic DNA because several microbial communities, most of which are yet to be cultivated, adapt to fixing CO₂ under a variety of environmental conditions [20–22]. The presence of two similar cbb operons should allow for the convenient placement of one or more RubisCO genes at the two locations and utilize the native regulatory mechanism to facilitate a more stringent selection upon cis complementation. The broad-host range vector (p90; Fig. 1B) allows for trans complementation, which is useful when cloning and expressing mutant RubisCO-genes from large libraries.

C. reinhardtii has been a useful host for RubisCO bioselection studies that are representative of the eukaryotic form I enzymes [17, 41, 42]. Phylogenetic analysis of form I RubisCO sequences places R. eutropha RubisCO into a “red-like” group (form IC) [6] along with the red-algal enzyme from Galdieria partita, known for its high Ω (5). The CO₂/O₂ specificity factor and other kinetic properties of the R. eutropha enzyme, including its oxygen tolerance, resemble those of higher plant enzymes (Tables 2 and 3) [5, 34]. With similarities to eukaryotic form I enzymes, structure-function studies with the R. eutropha form I enzyme, as described in this study, could conceivably be extrapolated for engineering the eukaryotic form I enzymes. The shorter generation times of complemented R. eutropha cultures (Figs. 1C and 4A) will facilitate rapid sampling of several RubisCO variants. Escherichia coli, a first-choice tool for modern biotechnological engineering strategies, has been used as a modified host for artificial selection with both Synechococcus (form I) and M. burtonii (form III) RubisCOs [16, 38]. However, the primary strategy involves detoxification of RuBP in an unnatural environment and the screening procedure results in several false positives [38]. Because the performance of RubisCO may be further constrained by other functional components of the CBB cycle and regulatory inputs during autotrophic growth conditions, it may be advantageous to “evolve” RubisCO enzymes in an organism like R. eutropha with a functional autotrophic metabolism. In fact, like E. coli, R. eutropha is being actively pursued as an efficient system for protein expression studies, which has been facilitated by the development of several broad-host range vectors for expression and an electroporation procedure for transformation [43–48].

Suppressor-selection identified amino-acid substitutions that would replace bulkier groups with smaller groups in the same hydrophobic region where the smaller alanyl side chain of residue 380 was replaced with a bulkier valyl side chain (Fig. 2B). These mutant substitutions buttress the importance of this hydrophobic region adjacent to the active site, which can be potentially targeted for improvements of other related RubisCO enzymes. The intragenic-mutant substitutions improved the A380V-mutant enzyme’s k_cat and interaction with CO₂ (Table 3). Beneficial changes to the enzyme’s interaction with O₂ (mutant A380V) and CO₂ (mutants Y348C, T330A/A380V and Y348C/A380V) represent a positive trend that could be exploited for future selection of enzymes with improved carboxylation k_cat and specificity. Although naturally existing RubisCO enzymes are deemed to have been optimized for performance in their native setting [7–9], the intrinsic differences in the catalytic properties of RubisCO from various sources point to the mutable nature of these properties. Thus, it is a reasonable goal to “artificially evolve” divergent enzymes with desirable properties in a heterologous selection system like R. eutropha, which can be subjected to harsher growth conditions [1]. RubisCO sampling studies have identified an inverse correlation between the enzyme’s k_cat for carboxylation and Ω values [5, 9]. However, it has been possible to introduce rationally-designed amino-acid substitutions into native enzymes, leading to improvements in net carboxylation efficiencies via increases in one or both of these kinetic constants [49, 50]. We surmise that the R. eutropha_- selection system described here can be modified to allow for selecting such enzymes in future. These modifications would include altering the selection pressure (e.g., by changing the CO₂/O₂ gas ratio), using a controllable promoter for RubisCO expression (e.g., P_BAD, which is responsive to arabinose), and introducing introducing accessory factors (e.g., folding chaperones, modifying enzymes, carbonic anhydrase).

In conclusion, we have described a new system for artificial evolution of RubisCO utilizing the RubisCO-deletion strain of the industrially relevant bacterium R. eutropha. The utility of this strain for expressing foreign RubisCO molecules has been demonstrated by successful functional complementation with heterologous genes encoding divergent enzymes. The system has also been used for bioselecting suppressors subsequent to random mutagenesis, which shows the potential for high-throughput screening and selection methods. Much like the O₂-tolerant hydrogenases from R. eutropha [51], RubisCO enzymes with compromised oxygenation capabilities will have distinct biotechnological applications. Evolving improved RubisCO enzymes in R. eutropha should facilitate future efforts to utilize CO₂ fixation for improving primary productivity and for the production of engineered bioproducts.

Materials and Methods

Strains, plasmids and growth conditions

E. coli strains JM109 [52] or Top10 (Invitrogen) were used for routine cloning and amplification of plasmid constructs, strain BL21 (DE3) (Agilent) was used for the expression of recombinant proteins in conjunction with expression vectors pET11a or pET28a (Agilent), and strain S17-1 (ATCC 47055) was used for introducing plasmids into R. eutropha via diparental conjugation [26]. R. eutropha wild-type strain H16 (ATCC 17699), and strains HF210 and HB10 have been described elsewhere [28, 29]. Complementation in R. capsulatus strain SB I/II⁻ was done as described before, using pRPS-MCS3 as the shuttle vector [15]. All E. coli strains were grown in LB medium with shaking at 37 °C or at 25 °C post-induction with 0.5 mM IPTG for recombinant protein expression. All R. eutropha strains were grown under heterotrophic conditions on minimal medium supplemented with either 0.4% malate or 0.5% fructose at 30 °C with shaking in the dark for liquid cultures. For chemoautotrophic conditions, strains were grown in either liquid or solid (with 1.5% agar) minimal medium and bubbled with 2.5% or 10% CO₂, 50% air (10.5% O₂), balanced with H₂ [26]. Solid media plates were prepared by supplementing growth media (for both E. coli and R. eutropha) with 1.5% agar. Antibiotics were added at the following concentrations, as required for heterotrophic growth: E. coli - ampicillin at 100 μg/ml, kanamycin at 30 μg/ml, and tetracycline at 12.5 μg/ml; R. eutropha – kanamycin at 200 μg/ml, tetracycline at 25 μg/ml, and spectinomycin at 700 μg/ml. Plasmids pUC19 [52] or pCR-Blunt II-TOPO (Invitrogen) were used for intermediate cloning steps. Plasmid pJQ200mp18Km [53] was used as the suicide vector for trans conjugation into R. eutropha for homologous recombination.

PCR amplification and other molecular biology procedures

Standard PCR reactions were carried out with PrimeSTAR GXL high-fidelity DNA polymerase (Clontech) and error-prone PCR products were generated with the Taq DNA polymerase (Invitrogen) as described previously [33]. Where appropriate, site-directed mutagenesis was carried out using the QuikChange kit (Agilent). All restriction enzymes and the T4 DNA ligase used in ligations were purchased from New England Biolabs. A 4.5kb region from cbb_c with flanking XbaI restriction sites was amplified from R. eutropha strain H16 (Fig. 1A) using primers RegcbbRLSX_Xba-f and RegcbbRLSX_Xba-r (Table S1), and cloned into pUC19, generating pUC19-RegcbbRLSX4.5kb. Site-directed mutagenesis was used to introduce an MfeI or a ClaI restriction site immediately 5′ of the cbbL start codon, and a SpeI site immediately 3′ of the cbbS stop codon, to generate plasmids pUC4.5Mfe-Spe or pUC4.5Cla-Spe. These two plasmids were used for cloning mutant or heterologous RubisCO genes in place of the native cbbLS and to subsequently transfer the entire region into suicide-vector constructs for homologous recombination in R. eutropha (Fig. 1A). Plasmid p90 was constructed for the trans expression of RubisCO genes in R. eutropha (Fig. 1B). This was accomplished by cloning a 204-bp promoter region amplified from the cbb_c of R. eutropha strain H16, with primers Regcbbp_Nco-f and Regcbbp_Sac-r (Table S1), into a modified version of the broad-host range plasmid 3716 (kind gift from Oliver Lenz and Baerbel Friedrich) previously described [21]. The cbbLS genes encoding the native chromosomal or megaplasmid RubisCO were amplified from R. eutropha strain H16 or HF210 using primers RecbbLS_Sac-f and RecbbLS_Xba-r (Table S1). Primers RegcbbL_Cla-f or RegcbbL_Mfe-f and RegcbbS_Spe-r (Table S1) were used to amplify the native chromosomal cbbLS genes for cloning into suicide-vector constructs. All heterologous RubisCO genes were amplified using the respective primers from constructs previously prepared in our laboratory and described elsewhere [21, 31–33] (Table S1). DNA sequencing reactions were performed at the Plant-Microbe Genomics Facility at The Ohio State University.

Construction of the R. eutropha RubisCO-deletion strain H16ΔLS

Strain H16ΔLS was created by precisely deleting the cbbLS genes from both loci, and a stem-loop structure from cbb_c. To accomplish this construction, a 540bp 5′-untranslated region (ending the base prior to the cbbL start codon) and a 588-bp 3′-untranslated region (including the last 4 bases of the cbbS coding sequence) were amplified with primers specific to the cbb region in megaplasmid pHG1 (Table S1) and cloned separately into pCR-Blunt II-TOPO vector, generating vectors TOPO-m5UTR and TOPO-m3UTR, respectively. Using restriction sites introduced by the PCR primers and those in the multiple-cloning sites of the respective vector constructs, the 3′ UTR sequence was cloned next to the 5′UTR region in TOPO-m5UTR, generating TOPO-mUTR. An XbaI-SpeI fragment, comprising the 5′ and 3′ untranslated regions, was cloned from the plasmid TOPO-mUTR into the XbaI site of the suicide vector pJQ200mp18Km, generating plasmid pJQ-mUTR. Plasmid pJQ-mUTR was transformed into E. coli strain S17-1 and conjugated into R. eutropha strain H16. Transconjugants were selected on minimal media-plates supplemented with fructose and kanamycin, serially passaged in an identical medium without any antibiotic, and screened for double recombinants resulting in strain H16mΔLS. The deletion was confirmed by DNA sequencing of a PCR fragment obtained by using external primers (Table S1). Using a 937-bp 5′-untranslated region and a 312-bp 3′-untranslated region and an identical procedure, the chromosomal copy of the cbbLS genes were knocked out in strain H16mΔLS to generate strain H16gmΔLS. A 134-bp region between the cbbS and cbbX coding sequences, which has a potential stem-loop secondary structure, was deleted in the construct pUC19-RegcbbRLSX4.5kb, by introducing two NheI- restriction enzyme recognition sites flanking the desired region for deletion via site-directed mutagenesis and NheI-digestion of the resulting plasmid and re-ligation. This plasmid, pUC19-4.5kbΔhp, served as the template for amplifying a 937-bp 5′untranslated region and a 941-bp 3′-untranslated region (flanking cbbLS) (Table S1), which was cloned into a suicide vector and used along with strain H16gmΔLS to generate strain H16ΔLS via suicide-vector integration and homologous recombination. A spectinomycin-resistant version of strain H16ΔLS was generated by Vanessa Varaljay by placing a Spc^R-gene cassette [26] in place of the missing cbbLS genes on megaplasmid pHG1. Complementation phenotypes with this mutant (H16ΔLSmpSpc^R) were identical to those with strain H16ΔLS. This strain was used for complementation experiments with suppressors of mutant A380V.

Recombinant-protein purification and enzymatic measurements

Protein purification, assessment of purity, quantitation and assays for determining kinetic constants were done as described elsewhere [31, 33]. Tritium-labeled RuBP, which was used for measuring CO₂/O₂ specificity factors, had been synthesized in the laboratory previously [31]. PRK activities were measured in cell extracts by the addition of NaH¹⁴CO₃ (Perkin-Elmer), ribulose-5-phosphate (Sigma) and ATP (Sigma) as substrates, and coupling the reaction with excess RubisCO. All kinetic constants measured with either the histidine-tagged megaplasmid RubisCO (Table 2) or the untagged chromosomal version of the wild type and A380V single-mutant proteins (Table 3) were similar.

Supplementary Material

Supp Table S1

Table S1. List of primers used in this study, with relevant restriction sites underlined in each sequence.

Abbreviations

RubisCO

D-ribulose 1,5-bisphosphate carboxylase/oxygenase

PRK

phosphoribulokinase

RuBP

D-ribulose 1,5-bisphosphate

V_c

V_max for carboxylation

V_o

V_max for oxygenation

K_c

Michaelis constant for CO₂

K_o

Michaelis constant for O₂

K_RuBP

Michaelis constant for RuBP

Ω

CO₂/O₂ specificity factor