ABSTRACT
The anoxygenic phototrophic bacterium Rhodopseudomonas palustris produces methane (CH4) from carbon dioxide (CO2) and hydrogen (H2) from protons (H+) when it expresses a variant form of molybdenum (Mo) nitrogenase that has two amino acid substitutions near its active site. We examined the influence of light energy and electron availability on in vivo production of these biofuels. Nitrogenase activity requires large amounts of ATP, and cells exposed to increasing light intensities produced increasing amounts of CH4 and H2. As expected for a phototroph, intracellular ATP increased with increasing light intensity, but there was only a loose correlation between ATP content and CH4 and H2 production. There was a much stronger correlation between decreased intracellular ADP and increased gas production with increased light intensity, suggesting that the rate-limiting step for CH4 and H2 production by R. palustris is inhibition of nitrogenase by ADP. Increasing the amounts of electrons available to nitrogenase by providing cells with organic alcohols, using nongrowing cells, blocking electrons from entering the Calvin cycle, or blocking H2 uptake resulted in higher yields of H2 and, in some cases, CH4. Our results provide a more complete understanding of the constraints on nitrogenase-based production of biofuels.
IMPORTANCE A variant form of Mo nitrogenase catalyzes the conversion of CO2 and protons to the biofuels CH4 and H2. A constant supply of electrons and ATP is needed to drive these reduction reactions. The bacterium R. palustris generates ATP from light and has a versatile metabolism that makes it ideal for manipulating electron availability intracellularly. We therefore explored its potential as a biocatalyst for CH4 and H2 production. We found that intracellular ADP had a major effect on biofuel production, more pronounced than the effect caused by ATP. This is probably due to inhibition of nitrogenase activity by ADP. In general, the amount of CH4 produced by the variant nitrogenase in vivo was affected by electron availability much less than was the amount of H2 produced. This study shows the nature of constraints on in vivo biofuel production by variant Mo nitrogenase.
INTRODUCTION
CH4 is the major component of natural gas, which is widely used as a source of energy in residential and industrial sectors, and H2 is an energy-rich gas that can substitute for gasoline as a transportation fuel. Almost 50% of the H2 produced today is made by steam methane reforming in which CH4 is heated to a high temperature. Thus, current CH4 and H2 production is dominated by fossil fuels. Mo nitrogenase, one of the most complicated enzymes known, is a catalyst for H2 production as well as for nitrogen fixation. It catalyzes the reduction of dinitrogen (N2) to ammonia (NH3), with concomitant H2 production (1, 2). Recently, a variant Mo nitrogenase with two amino acid substitutions near its active site was found to catalyze the reduction of CO2 to CH4 in vitro and in vivo in the bacterium Rhodopseudomonas palustris (3, 4). This form of the enzyme retains its ability to produce H2 but can no longer convert N2 to NH3. Mo nitrogenase has a FeMo cofactor at its active site and two catalytic components, Fe protein encoded by nifH and MoFe protein, encoded by nifDK.
The synthesis of one CH4 from one CO2 by variant nitrogenase requires eight electrons and 16 ATPs (5). H2 production by nitrogenase requires two electrons and four ATPs (equations 1 and 2 [where Pi indicates inorganic phosphate]).
R. palustris is an anoxygenic purple nonsulfur phototrophic bacterium that is an attractive host organism for in vivo studies of nitrogenase because it generates the considerable amount of ATP required for the enzyme to be active from light energy by cyclic photophosphorylation under anaerobic conditions (6, 7). Characteristics of R. palustris are that it does not produce oxygen and it typically uses organic compounds as its main source of carbon for biosynthesis (6). R. palustris can also carry out carbon dioxide fixation using the Calvin-Benson-Bassham cycle (Calvin cycle) when a source of electrons is available (Fig. 1) (8–11). R. palustris encodes three nitrogenase isozymes, Mo nitrogenase, vanadium nitrogenase, and Fe-only nitrogenase (6). The synthesis and activities of each of these enzymes are repressed and inhibited by NH3 (12, 13). However, nifA* mutants expressing Mo nitrogenase constitutively under all growth conditions tested have been developed by deleting 48 bp of the Q-linker region of the nifA gene, whose protein product is a master transcriptional activator of nitrogenase genes (10, 14). We have reported that a nifA* nifDV75AH201Q mutant strain (CGA3001) of R. palustris, expressing a Mo nitrogenase variant with two amino acid substitutions in its NifD subunit, produces CH4 and H2 (Fig. 1) (4). More recently, we found that wild-type Fe-only nitrogenase from R. palustris and other microbes produces CH4 along with H2 and NH3 as aspects of its normal catalytic cycle (15). However, here, we are focusing on a variant Mo nitrogenase and its ability to reduce CO2 to CH4.
Metabolic routes to CH4 and H2 production by R. palustris cells expressing a variant Mo nitrogenase. ATP is produced by cyclic photophosphorylation, in which electrons energized by light are cycled through a proton-pumping electron transport chain. Electrons derived from the oxidation of organic compounds are used in biosynthesis and in CO2 fixation and are also diverted to nitrogenase. CO2 and H+ are converted by the variant remodeled Mo nitrogenase to CH4 and H2, respectively. The nitrogenase variant has two amino acid substitutions in its NifD subunit, V75A and H201Q. This variant form of the enzyme does not convert N2 to NH3. CBB, Calvin-Benson-Bassham; hv, light photons.
As would be expected from the large amounts of electrons required for nitrogenase activity (equations 1 and 2), R. palustris wild-type Mo nitrogenase catalyzes the production of greater amounts of H2 when electrons are diverted away from the Calvin cycle, which is a major electron sink in R. palustris cells (10, 11). We also know that when cells are suspended in a medium that does not support growth, they divert more electrons to H2 production because nongrowing cells do not need electrons for biosynthesis (9, 16). Electron flow to nitrogenase can also be manipulated by using different carbon compounds with various oxidation-reduction values as electron donors (11). Here, we explore the influence of carbon source on H2 and CH4 production by R. palustris expressing remodeled nitrogenase in situations where the Calvin cycle is blocked, where cells are not growing, or where uptake hydrogenase is expressed. We also measured the amounts of ATP and ADP in cells incubated at different light intensities and found that in dark and at low light, intracellular ADP was sufficiently high to strongly inhibit biofuel production even though the concentration of intracellular ATP was relatively high.
RESULTS
The intracellular concentration of ADP rather than ATP determines the reaction rate of nitrogenase.Suspensions of nongrowing cells of a strain expressing variant Mo nitrogenase constitutively were incubated under different light intensities. When cells were incubated in dark (0 μmol photons/m2/s), no CH4 or H2 was detected. CH4 and H2 were detected when illumination was provided, and the yields of these gases after a 12-h incubation were positively correlated with light intensity. When light intensity was increased from 1 to 30 μmol photons/m2/s, CH4 production rose from 0.35 to 5.50 nmol/mg total protein (Fig. 2A), and H2 production rose from 0.40 to 3.25 μmol H2/mg total protein over a 12-h incubation (Fig. 2B). The ratio of CH4 to H2 increased about 2-fold (Fig. 2C). Yields of CH4 and H2 did not change significantly when the light intensity increased from 30 to 60 μmol photons/m2/s, suggesting that light intensity was no longer rate limiting for nitrogenase.
The intracellular concentration of ADP as well as the ratio of ADP to ATP are inversely correlated with CH4 and H2 production rates by R. palustris incubated at different light intensities. Nongrowing cell suspensions of an R. palustris mutant strain (CGA3001) that encodes a variant Mo nitrogenase were incubated for 12 h at 30°C with and without illumination. (A to C) Altered CH4 (A) and H2 (B) yields and ratios of CH4 to H2 (C) were observed at various light intensities. (D and E) After the measurements of CH4 and H2, cells were collected to determine intracellular ATP (D) and ADP (E) using an ATP bioluminescent assay kit. (F) The ratio of ADP to ATP was calculated from the obtained ATP and ADP values. Data are the average of three biological replicates, and the error bars represent the standard deviation (SD).
Given the large amount of ATP required for nitrogenase activity, one might expect intracellular ATP to correlate with the amounts of CH4 and H2 produced. However, as shown in Fig. 2D, nongrowing cells exposed to light at an intensity of 30 μmol photons/m2/s had ATP levels that were only about 1.6-fold higher than those of cells incubated under low light (1 μmol photons/m2/s), whereas such cells produced quantities of CH4 and H2 that were 15- and 8-fold higher, respectively. To investigate further, we measured ADP levels and found that the intracellular concentrations of ADP in nongrowing cells exposed to different light intensities were inversely correlated with the amounts of CH4 and H2 produced (Fig. 2E). The same correlation was observed for the ratio of ADP to ATP (Fig. 2F). These results suggest inhibition of nitrogenase activity in vivo by intracellular ADP is a rate-limiting step for CH4 and H2 production. ADP has been shown to inhibit nitrogenase activity, probably by competing for binding of ATP to the enzyme (17), but the literature is limited on this point. This measurement is difficult to make in vitro because the concentration of ATP changes rapidly when added to reaction mixtures.
Electron-rich alcohols support higher levels of H2 production than organic acids.To examine how electron availability affects CH4 and H2 production, the nifA*nifDV75AH201Q mutant strain was grown on alcohols and their corresponding more oxidized acids. Ethanol-acetate, butanol-butyrate, and butanediol-succinate pairs were selected for comparisons of CH4 and H2 production. The pairs of compounds take the same route to central metabolism, but the alcohols have two extra pairs of electrons to dispose of (Fig. 1). As shown in Fig. 3, cells grown on ethanol, butanol, and butanediol had final yields of H2 that were 2.4-, 3.3-, and 2.5-fold higher than those in cells grown on acetate, butyrate, and succinate, respectively. However, the amount of CH4 produced increased much less or not at all when alcohols were used as growth substrates.
An increased supply of electrons from the oxidation of alcohols boosts H2 production, and to a lesser extent, CH4 production. CH4 and H2 production by R. palustris nifA* nifDV75AH201Q (CGA3001) grown anaerobically in PM supplemented with acetate, ethanol, butyrate, butanol, succinate, or butanediol at a final concentration of 40 mM carbon is shown. NaHCO3 was added at a final concentration of 10 mM. Data are the average of three biological replicates, and the error bars represent the SD. Acet, acetate; Etha, ethanol; Buty, butyrate; Buta, butanol; Succ, succinate; Bund, butanediol; GT, generation time.
Biosynthesis competes strongly with nitrogenase for electrons.Biosynthesis is a major electron sink for R. palustris, and increased electron flow to nitrogenase can be achieved by blocking biosynthesis (16). To examine how electron oversupply influences the activity of nitrogenase, we compared CH4 and H2 production by growing cells versus nongrowing cells provided with different carbon compounds. Consistent with past work with acetate as an electron donor (4), nongrowing cells produced much more H2 and CH4 than did growing cells with all substrates tested (Fig. 4). There were up to 2.4-fold differences in the final yields of CH4 and up to 1.9-fold differences in the final yield of H2 depending on the electron donating organic compound supplied to nongrowing cells, but the differences did not correlate well with the reduction state of the electron-donating substrate. Nongrowing cells that are not carrying out biosynthesis excrete some metabolites, such as 2-oxoglutarate (16), and nitrogenase also produces formate from CO2 (5). It is likely that the kinds of compounds and amounts excreted differ depending on the starting substrate provided to cells.
CH4 and H2 production by growing and nongrowing cells. (A and B) Production of CH4 (A) and H2 (B) was compared between growing cells and nongrowing cells. The R. palustris nifA*nifDV75AH201Q (CGA3001) mutant was grown photoheterotrophically in PM until it entered stationary phase. Nongrowing cell suspensions of R. palustris nifA*nifDV75AH201Q were incubated in light for 10 days until maximal CH4 and H2 production was achieved. Organic compounds and NaHCO3 were added at a final concentrations of 40 mM carbon and 10 mM, respectively. Data are the average of three biological replicates and the error bars represent the SD. Fuma, fumarate; Succ, succinate; Acet, acetate; Buty, butyrate.
H2 production increases when electrons are blocked from entering the Calvin cycle.R. palustris strain CGA009, the parent strain of all the mutants discussed in this paper, has a mutation in its hupV regulatory gene that prevents it from expressing the NiFe uptake hydrogenase that it encodes (8). We have constructed a hupV-repaired strain that expresses and uses its uptake hydrogenase to reoxidize H2 produced by nitrogenase, thus resulting in lower H2 yields when cells are grown under nitrogen-fixing conditions (8). To test if the electrons lost as H2 can be recycled to drive CH4 production, we made a hupV-repaired nifA*nifDV75AH201Q mutant strain (CGA3002). As expected, this strain did not produce H2 when grown on fumarate, succinate, or acetate (Fig. 5), suggesting that uptake hydrogenase was indeed able to reoxidize the H2 produced by nitrogenase. However, CH4 production did not increase accordingly. One possibility is that electrons were diverted to the Calvin cycle for CO2 fixation. To test this, we blocked electrons from entering the Calvin cycle by using a ribulose 1,5-bisphosphate carboxylase (RubisCO)- and phosphoribulokinase-deficient (ΔcbbLS ΔcbbM ΔcbbP) mutant strain that expressed the variant Mo nitrogenase constitutively but lacked uptake hydrogenase activity (CGA3003). A ΔcbbLS ΔcbbM ΔcbbP mutant strain has the same growth rate as the wild type (18). This strain produced a similar amount of CH4 but 1.7-, 4.0-, and 4.6-fold more H2 when grown on fumarate, succinate, and acetate, respectively, compared to the strain with an intact Calvin cycle (Fig. 5). The further expression of an uptake hydrogenase in the Calvin cycle mutant strain resulted in decreased H2 production, but the amount of CH4 produced by cells grown on acetate or succinate did not increase. There was a dramatic decrease in CH4 production during growth on fumarate (Fig. 5).
The effect of uptake hydrogenase and Calvin cycle activities on CH4 and H2 production. (A to C) All of the four R. palustris mutant strains in this figure were expressing variant Mo nitrogenase, and their CH4 and H2 production yields from cells grown on fumarate (A), succinate (B), and acetate (C) are shown. Fumarate, succinate, and acetate were added at a final concentration of 40 mM carbon. The nifA* nifDV75AH201Q mutant strain is deficient in expressing an active-uptake hydrogenase. The nifA* nifDV75AH201Q hupV+ mutant strain has an active-uptake hydrogenase. The nifA* nifDV75AH201Q ΔcbbLSMP mutant strain is deficient in expressing both uptake hydrogenase and the Calvin cycle, and the nifA* nifDV75AH201Q hupV+ ΔcbbLSMP mutant strain is deficient in expressing the Calvin cycle. Data are the average of three biological replicates, and the error bars represent the SD.
DISCUSSION
From the data presented in this paper, we can make two major conclusions. One is that in vivo nitrogenase activity is inhibited by intracellular ADP. The other is that, in general, the amount of CH4 produced by the variant nitrogenase in an in vivo situation is affected much less by electron availability than is the amount of H2 produced.
The ATP-dependent reduction of CO2 and H+ to CH4 and H2 starts with electron transfer from reduced ferredoxin/flavodoxin to the Fe protein component of nitrogenase. Reduced Fe protein then transfers its electrons to MoFe protein, where the active center MoFe cofactor is located, and finally CO2 and H+ are reduced to CH4 and H2 at the active site (2, 19, 20). The reduction of Fe protein by reduced ferredoxin/flavodoxin and the regeneration of oxidized Fe protein, usually termed the Fe protein cycle, must repeat many times for the active site to accumulate enough electrons to reduce CO2. ATP plays a key role in the Fe protein cycle, but its binding to Fe protein is inhibited by ADP (17). ATP and ADP compete with each other for the nucleotide binding sites on the Fe protein. Because ADP exhibits a higher affinity for the nucleotide binding sites than does ATP, it can strongly inhibit ATP binding to the Fe protein (21). In our experiments, cells incubated in dark had an intracellular ADP concentration of about 7 nmol/mg total protein and a ratio of ADP to ATP of about 0.2. Under these conditions, nitrogenase activity was inhibited, as measured by H2 and CH4 production. The ADP concentration of ∼7 nmol/mg total protein is comparable to the reported concentration of ADP that showed a strong inhibition of ATP binding to Fe protein in vitro (21). As with our in vivo experiments, an ADP-to-ATP ratio of 0.2 has also been reported to completely inhibit nitrogenase activity in vitro (17). However, this measurement is difficult to make in vitro because nitrogenase activity is slow and the concentration of ATP added to reaction mixtures changes continuously. We hypothesize that in R. palustris, the ATP hydrolyzed by nitrogenase is not fully regenerated under dark or low-light conditions because of a decreasing rate of cyclic photophosphorylation. This contributed to a higher intracellular concentration of ADP, which correlated negatively with rates of CH4 and H2 production (Fig. 2).
Fumarate, succinate, acetate, and butyrate, which have different oxidation states, are all good growth substrates for R. palustris (11). Although electrons for CH4 and H2 are derived from these organic acids, we know from work with wild-type cells that the more reduced organic acids do not always support higher levels of H2 production. In addition to its oxidation state, the metabolic route for substrate use is a factor influencing product formation (11). Alcohols need to be oxidized to their corresponding organic acids by NAD-dependent alcohol/aldehyde dehydrogenases before entering central metabolism (Fig. 1). The NADH generated can be used by R. palustris cells to reduce flavodoxin or ferredoxin, the direct electron donor of nitrogenase via the FixABCX enzyme complex (22, 23). Therefore, organic acid-alcohol pairs are good candidate substrates to study how electron availability affects CH4 and H2 production. R. palustris cells provided with alcohols produced more H2 than when they were provided with the corresponding organic acids. In general, as shown in Fig. 3 and 5, CH4 production was less responsive to electron availability than was H2 production by the variant nitrogenase. One explanation for this could be that cells provided more electrons produced more formate than CH4. Besides CH4, formate is a product of CO2 reduction by variant nitrogenase (5). As with H2 production, direct hydride transfer is the major route to formate formation. However, a more energetically unfavorable “associative” pathway involving a transfer of eight electrons applies to CH4 formation (5). Given that formate formation by nitrogenase is easier to accomplish than CH4 production, increases in formate production may occur disproportionately relative to CH4 production in response to increases in electron flux. We failed to detect formate in cultures. R. palustris has an active formate dehydrogenase that reoxidizes formate with NAD+ to CO2 (24). Future studies will focus on measuring formate production by a formate dehydrogenase-deficient mutant.
In growing cells, nitrogenase competes with the Calvin cycle for electrons (10, 11). We found that disruption of the Calvin cycle by deleting cbbM, cbbLS, and cbbP encoding the two RubisCOs and phosphoribulokinase from R. palustris enhanced H2 production, especially on more reduced compounds like succinate and acetate, but it did not have a large influence on the CH4 production (Fig. 5). Previously, we reported improved CH4 production by a strain with only the RubisCO genes knocked out (CGA679 nifDV75AH201Q mutant) (4). However, this strain had an impaired growth rate because of the accumulation of toxic ribulose 1,5-biphosphate (18, 25). The further deletion of cbbP to prevent intracellular accumulation of ribulose 1,5-biphosphate restored the wild-type growth rate (18). Given that a low growth rate can shift more intracellular CO2 and electrons away from biosynthesis to CH4 and H2 production, the differences in growth rate between the two strains could explain the discrepancy in CH4 production.
Altogether, we found that ADP and electron availability were two key factors that influenced in vivo CH4 and H2 production by a variant Mo nitrogenase. An understanding of the constraints underlying these reactions should be helpful for developing an optimal biocatalyst for the future biotechnological application.
MATERIALS AND METHODS
Bacterial strains and growth conditions.For genetic manipulations, all Escherichia coli and R. palustris strains were grown as previously described (9, 10, 14). R. palustris strains were grown anaerobically in photosynthetic medium (PM) (26). Organic acids and alcohols were included at a final concentration of 40 mM carbon as the carbon sources and the medium was supplemented with 10 µM sodium molybdate (Na2MoO4). We also added 10 mM sodium bicarbonate, unless otherwise indicated. When appropriate, 10 μM nickel chloride was added for the expression of [NiFe]-uptake hydrogenase. N2 gas was provided in the headspace of sealed culture tubes. All cultures were incubated anaerobically with illumination at 30 μmol photons/m2/s from a 60-W incandescent light bulb (General Electric), unless otherwise indicated. Light intensities of 1 and 3 μmol photons/m2/s were generated by using a 15-W incandescent light bulb controlled by a dimmer switch, and illumination at 60 μmol photons/m2/s was obtained by using a 200-W incandescent light bulb. Nongrowing R. palustris cells were prepared by washing mid-exponential-phase cultures with nitrogen-free PM (NFM) and flushing the headspace with argon gas after degassing the medium, as described previously (4).
Genetic manipulation of R. palustris.All strains and plasmids used are listed in Table 1. The plasmid construction of pJQ-nifDV75AH201Q was done according to a previously published protocol (4). In-frame repair of hupV was created by PCR using Q5 high-fidelity DNA polymerase to amplify a 2,077-bp DNA fragment containing the hupV 4-bp sequence GATC (8). This fragment was then incorporated into the PstI-digested pJQ200SK suicide vector using the In-Fusion PCR cloning system (Clontech). R. palustris mutants CGA3002, CGA3003, and CGA3004 were prepared as follows. Plasmid pJQ-nifDV75AH201Q or pJQ-hupV010 was mobilized into R. palustris by conjugation with E. coli S17-1, and double-crossover events for allelic exchange were achieved using a selection and screening strategy described previously (14). Allelic exchange was verified using PCR amplification and sequencing of the resulting PCR products.
Strains and plasmids used in this study
CH4 and H2 measurements from R. palustris whole cells.When cultures of R. palustris strains reached their maximal optical density at 660 nm (OD660), gas-phase samples were withdrawn with a Hamilton sample lock syringe from the culture vial headspace. CH4 and H2 were measured with a Shimadzu GC-2014 gas chromatograph, as described previously (4). Total protein concentrations were determined using the Bio-Rad protein assay kit.
ATP and ADP measurements.Cultures incubated under different light intensities were first treated with 1% trichloroacetic acid (TCA) for 20 min. The cell lysates were then centrifuged at maximum speed in a microcentrifuge for 2 min, the supernatants containing ATP and ADP were collected, and finally, the pH of the supernatants was brought to 7.8 with Tris base. ATP was determined with the ATP bioluminescent assay kit (Sigma-Aldrich) in a 96-well white plate, following the instructions provided by the manufacturer. For the measurement of ADP content, ADP was first converted to ATP in a reaction system containing 5 mM Tris-acetate (pH 7.8), 2.5 mM MgCl2, 2 mM EDTA, 10 mM phosphoenolpyruvic acid monopotassium salt (pH 7.8), and 2 U pyruvate kinase from Bacillus stearothermophilus. The reaction system was then incubated at 30°C until all ADP was converted to ATP. The ADP content was obtained by determining the ATP content using the ATP bioluminescent assay kit.
ACKNOWLEDGMENTS
We thank James B. McKinlay for the generous gift of R. palustris strain CGA4011.
This work was supported as part of the BETCy Energy Frontier Research Center (EFRC), an EFRC funded by the U.S. Department of Energy, Office of Science grant DE-SC0012518.
FOOTNOTES
- Received 4 November 2018.
- Accepted 21 February 2019.
- Accepted manuscript posted online 1 March 2019.
- Copyright © 2019 American Society for Microbiology.
