Self-photosensitization of nonphotosynthetic bacteria for solar-to-chemical production

See allHide authors and affiliations

Science  01 Jan 2016:
Vol. 351, Issue 6268, pp. 74-77
DOI: 10.1126/science.aad3317

Using light in the darkness

Solid-state devices can efficiently capture solar energy to produce chemicals and fuels from carbon dioxide. Yet biology has already developed a high-specificity, low-cost system to do just that through photosynthesis. Sakimoto et al. developed a biological-inorganic hybrid that combines the best of both worlds (see the Perspective by Müller). They precipitated semiconductor nanoparticles on the surface of a nonphotosynthetic bacterium to serve as a light harvester. The captured energy sustained cellular metabolism, producing acetic acid: a natural waste product of respiration.

Science, this issue p. 74; see also p. 34


Improving natural photosynthesis can enable the sustainable production of chemicals. However, neither purely artificial nor purely biological approaches seem poised to realize the potential of solar-to-chemical synthesis. We developed a hybrid approach, whereby we combined the highly efficient light harvesting of inorganic semiconductors with the high specificity, low cost, and self-replication and -repair of biocatalysts. We induced the self-photosensitization of a nonphotosynthetic bacterium, Moorella thermoacetica, with cadmium sulfide nanoparticles, enabling the photosynthesis of acetic acid from carbon dioxide. Biologically precipitated cadmium sulfide nanoparticles served as the light harvester to sustain cellular metabolism. This self-augmented biological system selectively produced acetic acid continuously over several days of light-dark cycles at relatively high quantum yields, demonstrating a self-replicating route toward solar-to-chemical carbon dioxide reduction.

The necessity of improving the natural mechanisms of solar energy capture for sustainable chemical production (1) has motivated the development of photoelectrochemical devices based on inorganic solid-state materials (2). Although solid-state semiconductor light absorbers often exceed biological light harvesting in efficiency (3), the transduction of photoexcited electrons into chemical bonds (particularly toward multicarbon compounds from CO2) remains challenging with abiotic catalysts (4, 5). Such catalysts struggle to compete with the high-specificity, low-cost material requirements and the self-replicating, self-repairing properties of biological CO2 fixation (6). Thus, a viable solution must combine the best of both worlds: the light-harvesting capabilities of semiconductors with the catalytic power of biology.

Several inorganic-biological hybrid systems have been devised: semiconductor nanoparticles with hydrogenases to produce biohydrogen (7), long wavelength–absorbing nanomaterials to improve the photosynthetic efficiency of plants (8), and whole cells with photoelectrodes for CO2 fixation (9, 10). Whole-cell microorganisms are favored to facilitate the multistep process of CO2 fixation and can self-replicate and self-repair (11). Furthermore, bacteria termed “electrotrophs” can undergo direct electron transfer from an electrode (12). However, traditional chemical synthesis of the semiconductor component often requires high-purity reagents, high temperatures, and complex microfabrication techniques. Additionally, the integration of such foreign materials with biotic systems is nontrivial (13). Many reports have shown that some microorganisms induce the precipitation of nanoparticles (14), producing an inherently biocompatible nanomaterial under mild conditions.

Although photosynthetic organisms can precipitate semiconductor nanoparticles, their metabolic pathways are arguably less desirable than those of their nonphotosynthetic counterparts. Although gene modification of phototrophs has progressed (15), nonphotosynthetic bacteria remain the workhorse of synthetic biology, offering a facile way to tailor the product diversity from CO2 reduction (16). Additionally, thermodynamic comparisons reveal substantial energetic advantages to photosensitizing nonphotosynthetic CO2 reduction (17). Of particular interest is the Wood-Ljungdahl pathway, through which CO2 is reduced to acetyl coenzyme A (acetyl-CoA), a common biosynthetic intermediate, and eventually to acetic acid, both of which can be further upgraded to high-value products by wild-type and genetically engineered organisms (10, 18). This pathway is also used by CO2-fixing electrotrophs, enabling the use of semiconductor photoelectrons in this energetically efficient biosynthetic route.

We developed a hybrid system containing the nonphotosynthetic CO2-reducing bacterium Moorella thermoacetica (ATCC 39073) and its biologically precipitated CdS nanoparticles (19). CdS is a well-studied semiconductor with an appropriate band structure and is suitable for photosynthesis (20). As an acetogen and an electrotroph, M. thermoacetica serves as an ideal model organism to explore the capabilities of a hybrid system (21).

The photosynthesis of acetic acid by M. thermoacetica and CdS is a two-step, one-pot synthesis (Fig. 1). First, the precipitation of CdS by M. thermoacetica is triggered by the addition of Cd2+ and cysteine (Cys) as the sulfur source (19, 22). M. thermoacetica uses photogenerated electrons from illuminated CdS nanoparticles to carry out photosynthesis (Fig. 1B). The absorption of a photon, hν, by CdS produces an electron and hole pair, e and h+. The electron generates a reducing equivalent, [H] (see supplementary text and fig. S1 for elaboration of this process), that is passed on to the Wood-Ljungdahl pathway to synthesize acetic acid from CO2. Cysteine quenches the h+, leading to the oxidized disulfide form, cystine (CySS) (see supplementary text for the full set of reaction equations). The overall photosynthetic reaction is

Embedded Image(1)
Fig. 1

M. thermoacetica–CdS reaction schematics. (A) Depiction of the M. thermoacetica–CdS hybrid system, proceeding from the growth of the cells and bioprecipitation (loading) of the CdS nanoparticles (shown in yellow) through photosynthetic conversion of CO2 (center right) to acetic acid (right). (B) Pathway diagram for the M. thermoacetica–CdS system. Two possible routes to generate reducing equivalents, [H], exist: generation outside the cell (dashed line) or generation by direct electron transport to the cell (solid line). Hypothesized electron transfer pathways are presented in fig. S1.

The precipitation of CdS by M. thermoacetica was initiated by the addition of Cd(NO3)2 to an early exponential growth culture of glucose-grown cells supplemented with Cys (fig. S2) (19). Scanning electron microscopy (SEM), scanning transmission electron microscopy (STEM), and energy-dispersive x-ray spectroscopy (EDS) mapping revealed clusters of smaller nanoparticles (<10 nm; Fig. 2, A and B) composed of cadmium and sulfur on the cells (Fig. 2, C to E, and fig. S3). Absorption spectra and Tauc plots yielded a measured direct band gap of 2.51 ± 0.05 eV (fig. S4A). The slightly larger measured band gap relative to bulk CdS (2.42 eV) suggests the quantum confinement expected of <10-nm particles (20). Powder x-ray diffraction showed broad peaks consistent with small particles of the wurtzite phase (fig. S4B).

Fig. 2

Electron microscopy of M. thermoacetica–CdS hybrids. (A) SEM image of CdS nanoparticles on M. thermoacetica. (B) High-angle annular dark field (HAADF) STEM image of a single cell, showing clusters across the entire cell surface. (C) HAADF image and EDS mapping showing clusters mainly composed of (D) cadmium and (E) sulfur. Further elemental mapping is provided in fig. S3. Scale bars in (A) and (B), 500 nm; in (C) and (D), 50 nm.

To confirm photosynthesis, a series of deletional control experiments was carried out in which M. thermoacetica, CdS, and light were systematically removed (Fig. 3, A and B). In the absence of light (405 ± 5 nm), acetic acid concentrations decreased [from ~2 mM accumulated under initial H2:CO2 acclimation, as measured by quantitative proton nuclear magnetic resonance (1H-qNMR) spectroscopy], potentially as the result of a dark catabolic process. The viability determined by colony-forming units (CFU) assays slowly declined to ~25% after 4 days (from the initial 5.9 ± 0.4 × 104 CFU ml−1 and 1.7 ± 0.4 × 109 cells ml−1), indicating that the M. thermoacetica–CdS system requires light to maintain viability. The viability of bare M. thermoacetica without CdS dropped to 0% within the first day under light, consistent with previous observations of semiconducting and/or insulating precipitates having a photoprotective role toward bacteria (23). Only M. thermoacetica–CdS hybrids exposed to light produced acetic acid. The 1H-qNMR spectrum revealed acetic acid to be the only product of CO2 reduction, confirming the high selectivity expected of biological catalysts (fig. S5). After the first 1.5 days, the rate of production began to plateau because of the limiting amounts of the sacrificial reductant, Cys.

Fig. 3 Photosynthesis behavior of M. thermoacetica–CdS hybrids.

(A) Photosynthetic production of acetic acid by M. thermoacetica–CdS hybrids and deletional controls. The key applies to (A) and (B) only. (B) CFU viability assays for M. thermoacetica–CdS hybrids and deletional controls. (C) Rates of acetic acid production and quantum yields for increasing illumination intensities and M. thermoacetica–CdS concentrations. (D) Photosynthetic acetic acid production under low-intensity simulated sunlight with light-dark cycles. (E) Acetic acid production under dark conditions for varying illumination times. The inset shows the relation between illumination time, τ, and acetic acid yield under dark conditions at increasing multiples of τ. All points and error bars show the mean and error-propagated SD, respectively, of triplicate experiments.

We calculated a maximum yield of ~90% acetic acid (based on the initial Cys concentration), which is consistent with previous observations that in the Wood-Ljungdahl pathway, ~10% of reduced CO2 is directed toward cell biomass (24). During photosynthesis, both the viability and the cell counts of the M. thermoacetica–CdS system nearly doubled after the first day (fig. S6), on par with the doubling time of autotrophic growth (~25 hours). Although the growth was not vigorous and perhaps was limited by the total amount of CdS and other nutrients, these results suggest the possibility of a completely self-reproducing hybrid organism sustained purely through solar energy. After the third and fourth days, viability decreased in coincidence with the depletion of Cys, leading to oxidative photodamage (fig. S7).

Under increasing blue light flux (435 to 485 nm), the rate of acetic acid production increased (Fig. 3C). At 5 × 1013 photons cm−2 s−1, a quantum yield of 52 ± 17% was observed. The rate of photosynthesis increased up to 160 × 1013 photons cm−2 s−1, after which the rate dramatically decreased and the quantum yield dropped to 4 ± 1%, possibly due to photooxidative degradation under high light intensities (25). At high light fluxes, large holes formed in the cell surface and, in some cases, resulted in the complete destruction of the cell membrane (fig. S7). The high quantum yield of the M. thermoacetica–CdS system is notable, given that previous analogous systems often have had reported quantum efficiencies of ~20% (7). This result is rationalized by the low light flux of these measurements, which reduces losses from recombination (26). With four times the normal loading of M. thermoacetica–CdS hybrids, we measured a quantum yield of 85 ± 12% (Fig. 3C). Under higher concentrations, the average flux per bacterium decreased, correlating with increased quantum yield.

To further characterize their photosynthetic behavior, we illuminated M. thermoacetica–CdS hybrids under low-intensity simulated sunlight (air mass 1.5 global spectrum, 2 W m−2) with a light-dark cycle of 12 hours each to mimic day-night cycles (Fig. 3D). Unexpectedly, acetic acid concentrations not only increased under illumination but continued to increase in the dark at the same rate, through several light-dark cycles. A potential explanation lies in the accumulation of biosynthetic intermediates during the light cycle, which are then used during the dark cycle. These may include a number of reductive species [e.g., NADH (reduced nicotinamide adenine dinucleotide), NADPH (reduced NAD phosphate), or ferredoxin)] or intermediates in the Wood-Ljungdahl pathway such as acetyl-CoA (27). A proton gradient may also be storing energy for adenosine triphosphate synthesis in the dark. Further experiments that varied the duration of the light cycle (Fig. 3E) revealed a proportionality between the length of illumination, τ, and the acetic acid yield under dark conditions. Although the initial rate during and just after illumination appears to be relatively constant, consistent with a zero-order catalytic reaction, the yield begins to plateau after 2τ, exhibiting a linearity between illumination time and acetic acid yield (Fig. 3E, inset). However, at 5τ, samples that were illuminated for 24 hours break this trend. These observations suggest that during up to 12 hours of illumination, some intermediate accumulates, enabling a proportional acetic acid yield during the dark cycle. Beyond this, the intermediate may saturate, with longer illumination times yielding no further acetic acid. We measured a peak quantum yield of 2.44 ± 0.62% of total incident low-intensity simulated sunlight (Fig. 3D). These quantum yields are order-of-magnitude comparable to the year-long averages determined for plants and algae, which range from ~0.2 to 1.6% (1).

Biological routes to solid-state materials have often struggled to compete with high-quality traditionally synthesized materials. This work demonstrates not only that biomaterials can be of sufficient quality to carry out useful photochemistry, but that in some ways they may be more advantageous in biological applications. Most traditional nanoparticle syntheses require organic capping ligands to control the particle shape. These ligands present a barrier to charge transfer between the semiconductor and the catalyst, often requiring electron tunneling (13). The ligand-free approach taken here may help to establish a favorable interface between the bacteria and the semiconductor, resulting in improved efficiencies. Additionally, metal chalcogenides such as CdS have had limited application because of oxidative photodegradation; the ability of bacteria to precipitate metal chalcogenides from the products of photodissolution (Cd2+ and oxidized sulfur complex ions) suggests a potential regenerative pathway to circumvent the debilitating photoinstability through a precipitative self-regeneration.

The M. thermoacetica–CdS system displays behavior that may help it to exceed the utility of natural photosynthesis. First, the quantum yield increased with higher M. thermoacetica–CdS concentrations. The ability to tune the effective light flux per bacterium by changing the concentration of the suspension is a considerable advantage over similar light management practices in natural photosynthesis that are achieved through genetic engineering of chloroplast expression (28). Second, the catabolic energy loss observed during dark cycles in natural photosynthesis was absent in our hybrid system, which may be an innate feature of the Wood-Ljungdahl pathway, in which acetic acid is a waste product of normal respiration. Additionally, many plants and algae tend to store a large portion of their photosynthetic products as biomass, which requires extensive processing to produce useful chemicals. In contrast, the M. thermoacetica–CdS system directs ~90% of photosynthetic products toward acetic acid, reducing the cost of diversifying to other chemical products.

This system could be improved by substituting Cys oxidation with a more beneficial oxidation reaction, such as oxygen evolution, wastewater oxidation for water purification, or oxidative biomass conversion (29, 30). Expanding the material library available through biologically induced precipitation will increase the capacity for light absorption and raise the upper limit on semiconductor-bacteria photosynthetic efficiency. The availability of genetic engineering tools for M. thermoacetica (31), as well as the introduction of electrotrophic and nanoparticle precipitation behavior in model bacteria such as Escherichia coli (32, 33), suggests a potential role for synthetic biology in rationally designing such hybrid organisms.

Beyond the development of advanced solar-to-chemical synthesis platforms, this hybrid organism also has potential as a tool to study biological systems. The native integration of semiconductor nanoparticles with bacterial metabolic processes provides a distinctive optical tag for the study of microbial behavior, such as semiconductor-bacteria electron transfer (34, 35), by providing a sensitive, noninvasive, nonchemical probe.

Supplementary Materials

Materials and Methods

Supplementary Text

Figs. S1 to S9


  1. Materials and methods are available as supplementary materials on Science Online.
ACKNOWLEDGMENTS: The interface design part of this work was supported by the U.S. Department of Energy under contract no. DE-AC02-05CH11231 (PChem). Work at the Molecular Foundry was supported by the Office of Science, Office of Basic Energy Sciences, of the U.S. Department of Energy under contract no. DE-AC02-05CH11231. Solar-to-chemical production experiments were supported by NSF (grant DMR-1507914). The authors thank J. J. Gallagher and M. C. Y. Chang for the original inoculum of M. thermoacetica ATCC 39073. K.K.S acknowledges support from the NSF Graduate Research Fellowship Program under grant DGE-1106400. The authors thank the National Center for Electron Microscopy. All data are available in the body of the paper or in the supplementary materials.
View Abstract

Navigate This Article