pubs.acs.org/NanoLett
Artificial Photosynthesis in Ranaspumin-2 Based Foam David Wendell,* Jacob Todd, and Carlo Montemagno* Biomedical Engineering Department, Engineering Research Center, 2901 Woodside Drive, University of Cincinnati, Cincinnati, Ohio 45221 ABSTRACT We present a cell-free artificial photosynthesis platform that couples the requisite enzymes of the Calvin cycle with a nanoscale photophosphorylation system engineered into a foam architecture using the Tu´ngara frog surfactant protein Ranaspumin2. This unique protein surfactant allowed lipid vesicles and coupled enzyme activity to be concentrated to the microscale Plateau channels of the foam, directing photoderived chemical energy to the singular purpose of carbon fixation and sugar synthesis, with chemical conversion efficiencies approaching 96%. KEYWORDS Artificial photosynthesis, foam, biofuel
E
properties, DMF has been synthesized from simple carbohydrates, such as glucose or fructose, harvested from biomass.6 To date, in vitro carbon fixation experiments have been limited to the examination of Calvin-Benson-Bassham (CBB) cycle intermediates and cell extracts using radiometric and spectrophotometric techniques.18-20 Photosynthesis, carbon sequestration, and carbohydrate generation involve several complex and well-studied processes; among these, a suite of eight enzymes make up the portion of the CBB cycle responsible for converting the energy of ATP into six carbon sugars such as glucose and fructose. The chief products of the light-dependent reactions of photosynthesis are NADPH and ATP. The thermophilic F0F1 ATP synthase has been purified and reconstituted in both liposomes and ABA triblock polymersomes, along with the photoactivated proton pump bacteriorhodopsin (BR) to form ATP producing vesicles.10 Recently, an improvement in the artificial synthesis of ATP was demonstrated using Tween-20 (T20)-based foam and polymersomes.11 Here we report the extension of this architecture to include lipid vesicles by using the natural surfactant protein Rsn-2, which is the major foam forming constituent of the nests of the Tu´ngara frog, Physalaemus pustulosus.21 Unlike chemical detergents, the Rsn-2 protein surfactant has evolved a unique ability to enable foam formation in low concentrations without disrupting cell membranes. Together, the vesicle infused foam represents a novel biomimetic light-driven material which converts light into ATP and then CO2 into sugar using the requisite components of the CBB cycle. For carbon sequestration and sugar synthesis we coupled our ATP producing photoconversion system with the requisite enzymes of the CBB cycle. The entire photosynthetic system was first divided into three independent reactions, including (1) an ATP production assay, (2) a RuBisCo carbonfixation assay, and (3) a glucose producing assay. After the
ngineered biological solar energy conversion has produced a variety of electrical1,2 and chemical energy3-6 storage strategies. Of the latter ATP serves as the most important natural energy molecule7 and has been formed artificially by coupling F0F1 ATP synthase to a photon-induced proton motive force.8-11 In photosynthetic organisms, long-term energy storage is accomplished through biomass synthesis, an ATP-dependent carbon fixation12-15 reaction which provides a foundation for liquid biofuel production.4,5,16 Harvesting and converting biomass to combustible fuel has been suggested as a renewable energy solution to the ongoing depletion of fossil fuels.17 Nature has acquired means for solar energy capture in the photoreactive centers of plants for the purpose of synthesizing and storing such biomass. However, the near 100% quantum efficiency of the photoreactive centers is reduced to approximately 5% of that value in usable energy due to limited wavelength sensitivities and necessary cellular processes, including growth, repair, and maintenance.3 Converting plant sugars to ethanol has been proposed as a renewable energy source, particularly from sugar cane, corn stover, and switchgrass;4-6 however, this requires that limited land and water resources be diverted, in part, to biomass production. Processes which evolve liquid fuels, such as ethanol, from biomass have been widely developed.4,6 Due to only marginal energy yield from the production of bioethanol,4 2,5dimethylfuran (DMF) is considered a prudent alternative to ethanol given its higher energy density, boiling point, and insolubility in water. In addition to attractive liquid fuel
* To whom correspondence should be addressed,
[email protected] or
[email protected]. Received for review: 02/14/2010 Published on Web: 03/05/2010 © 2010 American Chemical Society
3231
DOI: 10.1021/nl100550k | Nano Lett. 2010, 10, 3231–3236
FIGURE 1. An illustration of the BR/F0F1ATP synthase vesicle solar conversion system coupled to the RuBisCO CBB cycle and trapped within the foam channels. Sunlight is converted into ATP which is then used by the CBB enzymes to make sugar from carbon dioxide and NADH.
pared to the bulk-vesicle solution.11 For these experiments, a similar loss of ATP to that was observed in deflated T20 foam solutions (Figure 3A), which is attributed to increased detergent-mediated vesicle permeability, a noteworthy aspect as it pertains to our process both in the ATP production and in subsequent glucose production stages. We observed no interference from detergent in foam-polymer vesicle solutions which remained inflated throughout the process (Figure 3A), which revealed that rapid and sustained inflation is necessary for continual ATP and glucose production in the T20 solutions. Conversely, Rsn-2 based foams did not show this interference, keeping lipid vesicles intact within the bubble architecture (see Figure 2C). This demonstrates the utility of Rsn-2 as a superior surfactant choice to other detergent choices. Our maximum observed ATP production determined during the linear phase using regression analysis, which can be visualized in Figure 2C, for the liposomes is 130 (nmol/ min)/mg of F0F1 as compared to 110 (nmol/min)/mg of F0F1
validation of these three experiments, a single unified experiment of the whole process was performed. The production of ATP in the first reaction was determined with a luciferase bioluminescence assay. Light-driven ATP synthesis was initiated by exposing a BR/F1F0ATP synthase vesicle solution containing ADP and free phosphate to green light. This was performed in a bulk enzyme solution and inflated and deflated foam enzyme solutions, with T20 as the surfactant for the proteopolymersomes and Rsn-2 for the proteoliposomes. The amount of ATP produced was normalized to the amount of F1F0ATP synthase present in each sample and plotted as a function of the light exposure time, which can be seen in Figure 2. The rate of photoderived ATP was similar to our earlier proteopolymersome results,11 both in foam and in bulk solutions, with controls showing no photophosphorylation (Figures S1-S3 in Supporting Information). Our previous investigations with BR/ATP synthase activity in foam architecture showed increased output when com© 2010 American Chemical Society
3232
DOI: 10.1021/nl100550k | Nano Lett. 2010, 10, 3231-–3236
FIGURE 2. (A) Lipid vesicles were formed in the presence of 1 nM QD-565 (green quantum dots used to illuminate the lipid vesicles) and gel purified to separate unincorporated QDs. Eluted vesicles were resuspended with Rsn-2 and aspirated to create foam. Vesicles actively traveled through the Plateau channels and collected at the nodes. (B) Large lipid vesicles were readily observed intact within the channels and appeared relatively stationary compared to the smaller flowing vesicle assemblies. (C) Production of ATP with BR/ATP synthase lipid vesicles in Rsn-2 foam (blue up triangle), in bulk (black square), in deflated Rsn-2 foam (purple left triangle), and in T20 foam (green down triangle) and a control experiment in the dark (red circle) for comparison. (D) BR/ATP synthase function in a lipid membrane was limited to the Rsn-2 based foam since the T20 adversely affected coupled F1F0-ATPase/BR vesicle function. All error bars refer to standard deviation (n ) 3).
for the proteopolymersomes. The lipid vesicle rate is lower than our previous results,10 but as our primary interest is to highlight the relative function of the vesicles in bulk solution versus foam and their coupled function to the eight enzymes of the CBB cycle, improving this output was not necessary. Many interdependent proteoliposome formation factors influence this conversion, including vesicle size, composition, formation method, protein activity, concentration, orientation, and coupled inhibition affects. Coupled inhibition is suspected to be heavily influential when most of the other factors are accounted for.9,10 Surprisingly, we achieved a similar ATP production from the polymersomes despite a significantly decreased number of vesicles (Figures S9-S12 in Supporting Information). The photosynthetic carbon reduction cycle uses the ATP resulting from the light-dependent reactions in conjunction with the carbon-fixing enzyme ribulose-1,5-bisphosphate carboxylase/oxygenase (RuBisCO) in order to form a variety of energy storing carbon compounds.12 The first stable © 2010 American Chemical Society
derivative of the carbon fixation reaction is 3-phosphoglycerate (3-PGA), which is evolved 2-fold for each molecule of carbon dioxide fixed by RuBisCO to ribulose-1,5-bisphosphate (RuBP). This is accomplished through a well-documented catalytic pathway of defined intermediates.13 The subsequent product of the carbon-fixing reaction is glyceraldehyde-3-phosphate, which is produced from the oxidation of NADPH, though similar enzymatic activity has been shown for NADH.15 The final output from G3P assimilation is a hexose; however the catalyzing isomerase has shown variable stereospecificity for the output of glucose and fructose.22-25 In previous work, CBB cycle enzymes have been coupled to related dehydrogenase enzymes such as GAPDH in order to measure the interconversion of dinucleotides via absorbance shift, to analyze enzyme activity.19,20 In the second reaction, carbon fixation and subsequent production of G3P were monitored via the oxidation of NADH. This was first performed with a pure ATP stock 3233
DOI: 10.1021/nl100550k | Nano Lett. 2010, 10, 3231-–3236
FIGURE 3. (A) ATP synthesis using BR/ATP synthase polymersomes in T20 foam (black square), bulk (red circle), deflated T20 foam (green down triangle), and a control experiment in the dark (blue up triangle) for comparison (n ) 3 for each). Inset is the light intensity standard curve created with ATP stock dilutions. (B) The foam system containing BR/ATP synthase vesicles, RuBisCO, PGK, GAPDH, NADH, which is converting CO2 and RuBP to G3P using photoderived ATP. The RuBisCO-dependent carbon fixation reaction is fueled by lipid photophosphorylation vesicles fuel within the Rsn-2 foam (black, n ) 3) and in bulk (brown, n ) 3) and the proteopolymersomes within T20 foam (red, n ) 3) and in bulk (blue, n ) 3). Control experiments were also performed in the dark (green and purple overlapping at 0, n ) 3 for each) and with selectively removed individual components (Figures S5-S7 in Supporting Information) without detectable G3P production. (C) Surfactant properties of Rsn-2 (black, n ) 3) and a control with BSA (red, n ) 3). (D) Fluorescent image of FITC polymer vesicles within the T20 foam. All error bars are standard deviation.
solution (Figure S4 in the Supporting Information), to ensure proper multienzyme coupled functionality and then subsequently with photoderived ATP from both the bulk and foam solutions. Separate control samples were prepared with no ATP present, with the exclusion of each assay component, and with vesicle solutions incubated in the dark (Figures S5-S7 in the Supporting Information). The coupled activity of the enzymes in the presence of excess ATP can be seen in the Supporting Information. The activity for the same reaction using photoderived ATP both from bulk and foam solutions can be seen in Figure 3. As expected, the presence of higher amounts of ATP in the foam solution translated to higher output of G3P. The foams repeatedly outperformed the bulk solutions with the Rsn-2 mediated foam producing a significantly higher sugar output overall, when compared to T20 using the same starting components. The formation of glucose in the third experiment was measured by a colorimetric assay using glucose oxidase as © 2010 American Chemical Society
an end point measurement of glucose accumulated over the 4 h incubation time. Glucose was formed from G3P via the addition of triose phosphate isomerase, fructose 1,6-bisphosphate aldolase, fructose 1,6-bisphosphatase, phosphoglucose isomerase, and glucose 6-phosphatase. Like the intermediate G3P findings, glucose yield was superior in the Rsn2-based foam when compared to T20, presumably due to both the increase in available starting material and more mild surfactant. Control samples were prepared by removing each enzyme individually and by incubating the CBB enzymes without ATP present before performing the glucose assay. Glucose was observed in all samples containing G3P and all enzymes but not detectable in any control samples with one or more components missing (Tables S1-S5 in the Supporting Information). After demonstrating the system as separate experiments, the three reactions were combined in order to conduct an assay of the full process. The full process experiment, like 3234
DOI: 10.1021/nl100550k | Nano Lett. 2010, 10, 3231-–3236
TABLE 1. Energy Efficiency of Glucose Production process
array
amount of glucose (nmol)
full system
bulk dept foam Rsn-2 foam
21 ( 2.0 38 ( 1.3 60 ( 1.0
chemical conversion efficiency from ATP (%) 53 ( 5.0 96 ( 3.3 95 ( 1.3
able membrane. The pure fructose output would have a decisive advantage over cellulosic sugars, which require energy intensive extraction and conversion procedures. Our photosynthetic foam has the capability of producing 116 nmol of glucose/(mL/h), which when fully inflated must consume at least 10 times its initial volume. Biomass energy yield is traditionally examined on a per hectare basis; therefore, if we assume a foam architecture containing the glucose-producing system layered to a 1 m height and the reported conversion of sugar to DMF to be 88%,6 our system has the capability to produce 10 (kg/ha)/h of DMF. Clearly many factors would affect this yield, such as temperature, humidity, enzyme function, sunlight attenuation over the depth of the foam, and the DMF conversion process. In order to compare this to the energy yield from biomass, the length of the growing season and available light exposure must be considered. If we assume that the single hectare of foam receives sunlight 11 h/day, we find an output of 34.5 (t/ha)/a of DMF, which is ten fold greater than the 3.4 (t/ha)/a of DMF currently available from biomass on a dry matter basis.16 Given the space filling nature of the foam and that soil is not required, we imagine this production could be adapted to urban rooftop environments, nonarable land areas, and other nontraditional spaces that receive light exposure. This would maintain water and land resources critical for food production and allow a diverse range of photosynthetic architectures. It has been recently reported that the US production of biomass-based ethanol would require between 800 and 4200 L of water for each liter of ethanol produced,29 much of which is lost to plant evapotranspiration. Foam drainage and evaporation is humidity dependent30 and without the benefits of stomata regulation; however, the recovered water phase trapped in the foam during sugar extraction could be reused along with BR-ATPase vesicles and CBB enzymes assuming a well-designed filtration and recirculation system. Undoubtedly, the extraction and large scale implementation of our fully engineered multienzyme system is just beginning, but a photosynthetic foam designed for focused biologic energy conversion offers a new paradiagm for carbon fixation and biofuel generation without the biomass limitations of arable land or excess quantities of water.
the ATP production assay, was conducted in bulk as well as inflated and deflated Rsn-2/T20 foam solutions using the lipid and polymer vesicles with an end point measurement of total accumulated glucose produced. Each sample (n ) 3) was incubated for 4 h and assayed for glucose, the results are summarized in Table 1. The results for the formation of G3P in the full process assay were comparable to those obtained from the assays conducted in a stepwise fashion. The peak chemical energy conversion efficiency was 96%, calculated from the combustion energy of glucose (2800 kJ/ mol) and the Gibbs free energy of inorganic phosphate addition to ADP (35 kJ/mol).28 Of paramount interest is the stored energy and net energy conversion efficiency of our artificial photosynthesis process. The BR/ATPase lipid vesicle photoconversion system has been well characterized.8-10 Under optimized protein concentrations, F0F1 remains the conversion bottleneck with BR providing more pmf than the synthase can use. For BR concentrations close to our experimental method (0.1 mg/mL) optimized ATP synthesis and conversion stoichiometry is 4-5 H/ATP.8,9 Estimates for BR quantum efficiency range from 0.25 to 0.79, with some suggesting an equal exchange of protons for photons.27 Assuming a Gibbs free energy of ATP formation to be 35 kJ/mol,28 and 188 kJ per einstein of 570 nm light,26 we obtain a rough estimate for photoconversion efficiency of 5%. Despite an excess of protons available for transport, idealized protein concentrations still yield less than the generally accepted two to three protons8 per ATP observed in natural systems,8-10 making the reconstituted photophosphorylation method about half as efficient. The 2-fold production improvement we observe in the foam channels could point toward achieving this theoretical limit; however, it should be noted that the nonoptimized liposome ATP production rate in this study remained lower than what we have achieved in the past.10 With millions of years of evolution, natural photosynthesis serves as the metric for solar to chemical energy conversion efficiency. Thus, our goal is to compare the potential of the artificial photosynthetic system with current yield for fuels derived from biomass sources. By designing the photosynthetic foam to synthesize sugar directly, biofuels like DMF could be produced. This choice is superior to ethanol given the 40% higher energy density (37.5 MJ/kg), higher boiling point (by 20 K), and insolubility in water.6 DMF is also ideal because the recently developed catalytic synthesis procedure6 prefers fructose as the starting substrate, which would require fewer enzymes (six) and can be easily separated from these proteins using a semiperme© 2010 American Chemical Society
Acknowledgment. The authors wish to thank Jacob Schmidt for critically reading the manuscript, Cole Brokamp and Kyle Minor for proteopolymer vesicle preparations and protein purifications, and Jason Utter for surfactant measurements with Rsn-2. Supporting Information Available. A detailed methods section and supporting experiments as well as controls, supporting figures showing control experiments comparing photoderived ATP production in the bulk, foam, and deflated foam solutions, a demonstration of carbon fixation using RuBisCO RuBP and stock ATP in a bulk aqueous solution, and carbon fixation assay controls, tables summarizing the 3235
DOI: 10.1021/nl100550k | Nano Lett. 2010, 10, 3231-–3236
glucose production and glucose control assays with selected assay components removed and a comparison between bulk solution and the foam for the entire glucose production system, and figures showing glucose assay standard curve and the gene synthesis and protein purification results for Rsn-2, vesicle size information in the form of dynamic light scattering and TEM micrographs, and the fluorescent vesicles in Rsn-2 and T20 foam solutions. This material is available free of charge via the Internet at http://pubs.acs.org.
(11) Choi, H.-J.; Montemagno, C. Nanotechnology 2006, 17, 2198– 2202. (12) Bassham, J. A.; Calvin, M. Encycl. Plant Physiol. 1960, 5 (1), 884– 922. (13) Hartman, F. C.; Harpel, M. R. Annu. Rev. Biochem. 1994, 63, 197– 234. (14) Axelrod, B.; Bandurski, R. S. J. Biol. Chem. 1953, 204, 939–948. (15) Brenneman, F. N.; Volk, W. A. J. Biol. Chem. 1959, 234 (9), 2443– 2447. (16) McKendry, P. Bioresour. Technol. 2002, 83, 37–46. (17) Ward, D. A.; Keys, A. J. Photosynth. Res. 1989, 22, 167–171. (18) Racker, E. Arch. Biochem. Biophys. 1957, 69, 300–310. (19) Wu, R.; Racker, E. J. Biol. Chem. 1958, 234, 1029–1035. (20) Report on the Basic Energy Sciences Workshop on Solar Energy Utilization, United States Dept. of Energy, 2005. (21) Mackenzie, C. D.; Smith, B. O.; Meister, A.; Blume, A.; Zhao, X.; Lu, J. R.; Kennedy, M. W.; Cooper, A. Biophys. J. 2009, 96 (12), 4984–4992. (22) Meyerhof, O.; Beck, L. V. J. Biol. Chem. 1944, 156, 109–120. (23) Alefounder, P. R.; Baldwin, S. A.; Perham, R. N.; Short, N. J. Biochem. J. 1989, 257, 529–534. (24) Ramasarma, T.; Giri, K. V. Arch. Biochem. Biophys. 1956, 62, 91– 96. (25) Nordlie, R. C. In Boyer, P. D., Ed.; The Enzymes, 3rd ed.; Academic Press: New York, 1971; Vol. 4, pp 543-610. (26) Bogomolni, R. A.; Stoeckenius, W. J. Supramol. Struct. 2004, 2 (56), 775–780. (27) Govindjee, R.; Balashov, S. P.; Ebrey, T. G. Biophys. Chem. 1990, 58 (3), 597–608. (28) Alberty, R. A. J. Biol. Chem. 1969, 244, 3290–3302. (29) Dominguez-Faus, R.; Powers, S. E.; Burken, J. G.; Alvarez, P. J. Environ. Sci. Technol. 2009, 43 (9), 3005–3010. (30) Ekserova, D. R.; Krugliakov, P. M. Foam and Foam Films; Elsevier Science B.V.: Amsterdam, The Netherlands, 1998.
REFERENCES AND NOTES (1)
Choi, H.-J.; Lee, H.; Montemagno, C. D. Nanotechnology. 2005, 16, 1589–1597. (2) Ihara, M.; Nishihara, H.; Yoon, K.; Lenz, O.; Friedrich, B.; Nakamoto, H.; Kojima, K.; Honma, D.; Kamachi, T.; Okura, I. Photochem. Photobiol. 2006, 82, 676–682. (3) Zhu, X.-G.; Long, S. P.; Ort, D. R. Curr. Opin. Biotechnol. 2008, 19, 153–159. (4) Lynd, L. R.; Cushman, J. H.; Nichols, R. J.; Wyman, C. E. Science 1991, 251 (4999), 1318–1323. (5) Gordon, J. M.; Polle, J. E. W. Appl. Microbiol. Biotechnol. 2007, 76, 969–975. (6) Roma´n-Leshkov, Y.; Barrett, C. J.; Liu, Z. Y.; Dumesic, J. A. Nat. Lett. 2007, 447, 982–986. (7) Knowles, J. R. Annu. Rev. Biochem. 1980, 49, 877–919. (8) Wagner, N.; Gutweiler, M.; Pabst, R.; Dose, K. Eur. J. Biochem. 1987, 165, 177–183. (9) Pitard, B.; Richard, P.; Dunach, M.; Girault, G.; Rigaud, J.-L. Eur. J. Biochem. 1996, 235, 769–778. (10) Hazard, A.; Montemagno, C. Arch. Biochem. Biophys. 2002, 407, 117–124.
© 2010 American Chemical Society
3236
DOI: 10.1021/nl100550k | Nano Lett. 2010, 10, 3231-–3236