Aerobic growth at nanomolar oxygen concentrations
INAUGURAL ARTICLE Aerobic growth at nanomolar oxygen concentrations Daniel A. Stolpera,1, Niels Peter Revsbechb, and Donald E. Canfielda,2 a Nordic Center for Earth Evolution (NordCEE) and Institute of Biology, University of Southern Denmark, Campusvej 55, 5230 Odense M, Denmark; and Department of Biological Sciences, University of Aarhus, DK-8000 Aarhus C, Denmark b This contribution is part of the special series of Inaugural Articles by members of the National Academy of Sciences elected in 2007. aerobic respiration ∣ microbial growth ∣ precambrian ∣ nanaerobe ∣ evolution I n discussions of the history of life, aerobic respiration is often assumed to have been an unviable means of energy production until oxygen levels exceeded the “Pasteur Point” (1–3). The Pasteur Point is the partial pressure of oxygen (in equilibrium with a solution) at which facultative aerobic organisms switch to anaerobic metabolisms. It varies from organism to organism, but a value of approximately 0.01 (1%) of the present atmospheric oxygen level (PAL) is typical for modern prokaryotes and single-celled eukaryotes (4). Seawater in equilibrium with this level of atmospheric oxygen at 25 °C has a dissolved oxygen concentration of 2.2 μM, which is comparable to the K m values (Michaelis–Menten constants*) for oxygen utilization for many aerobic bacteria isolated from marine environments (5). It is known, however, that some aerobes can utilize much lower oxygen concentrations than those corresponding to the Pasteur Point. For example, a K m value as low as 31 nM has been measured for the facultative aerobe Aerobacter aerogenes (6). Additionally, oxygen concentrations down to 3 nM have been measured in soybean nodules supporting aerobic respiration (7), and the corresponding K m values for oxygen respiration of bacteria populating these nodules are in the 5- to 8-nM range (8). Likewise, K m values for the high-oxygen-affinity cytochrome bd oxidase for Escherichia coli have been measured as low as 3–8 nM, in contrast to the lower affinity E. coli cytochrome bo oxidase, which has a K m of about 200 nM (9). Despite these determinations of low-oxygen utilization by some microbes, growth data are sparse, and we are unaware of any direct experimental demonstrations of microbial growth based on aerobic respiration at oxygen levels less than 230 nM (10). We emphasize the importance of growth, because to thrive in an environment an organism must grow and reproduce. Microbes can often metabolize and respire under environmental conditions that do not allow for growth such as temperature extremes or limited substrate availability (11). Understanding the oxygen limits of aerobic microbial growth is relevant to discussions of the microbial ecology of both modern and ancient Earth systems. On the modern Earth, low-oxygen niches exist at all transitions from oxic to anoxic environments, www.pnas.org/cgi/doi/10.1073/pnas.1013435107 as found, for example, in marine and freshwater sediments (12), many soils, and in the oxygen minimum zones (OMZs) of the global ocean (13). In the late Archean Eon (from about 2.7–2.5 billion years ago, Ga), prior to the pervasive oxygenation of the Earth’s atmosphere at 2.3 to 2.4 Ga (14), elevated oxygen concentrations apparently existed locally and/or for short durations (15–18). However, atmospheric oxidant models, including those explaining the “mass-independent” fractionation of sulfur isotopes prior to 2.3 to 2.4 Ga, restrict average oxygen levels in the atmosphere to a maximum of ∼10−5 of the present level (19, 20), corresponding to a concentration of about 2.2 nM in coexisting surface waters. An unresolved question is whether such oxygen levels were sufficient to support the growth of aerobic communities. Using the Pasteur Point of most modern aerobic organisms as a reference point would lead one to conclude “no”, but biomarkers preserved in late Archean (2.67 Ga) rocks suggest the presence of aerobic communities (21). The motivation for this study is to explore whether the conventional view about the limits of aerobic growth is correct and to place quantitative limits on the lower oxygen requirements for aerobic respiration in both the past and present. To achieve this, we monitored growth of E. coli K-12 at low (nanomolar) oxygen concentrations made possible by evolving oxygen sensor technology (13). We demonstrate that aerobic growth is not constrained by the typical Pasteur Point (i.e., to > ∼ 0.01 PAL) and that aerobic growth is viable at oxygen concentrations ≤3 nM (∼10−5 PAL), at least two orders lower than previously reported. Results and Discussion Growth Under Oxygen-Limiting Conditions. A typical experiment is shown in Fig. 1A. Here, we monitored oxygen concentration and growth for 30 h. After the initial oxygen manipulations during the first 6 h, we established the final desired fixed pump rate of 36 mL h−1 corresponding to an oxygen flux of 11.5 0.3 μmol L−1 h−1 (Fig. S1). After about 20 h, oxygen concentrations reached the detection limit of the sensor (3 nM), yet growth continued to the end of the experiment 10 h later. Similarly, in a 50-h experiment with a 35-min ramp-up phase (Fig. 1B) and at a slower pump rate of 12 mL h−1 (corresponding to an oxygen flux of 5.5 0.3 μmol L−1 h−1 ; Fig. S1), oxygen concentrations decreased to the detection limit of the sensor after about 12 h. Still, growth continued for an additional 38 h (when the experiment was terminated) at oxygen concentrations at or below the detection limit of our sensors. Author contributions: D.A.S., N.P.R., and D.E.C. designed research; D.A.S. performed research; N.P.R. contributed new reagents/analytic tools; D.A.S. and D.E.C. analyzed data; and D.A.S., N.P.R., and D.E.C. wrote the paper. The authors declare no conflict of interest . *The Michaelis–Menten constant, K m , is the substrate concentration at which an enzyme or a cell’s overall respiration rate operates at half its maximum rate. 1 Present address: Division of Geological and Planetary Sciences, California Institute of Technology, MC 170-25 1200 East California Boulevard, Pasadena, CA 91125. 2 To whom correspondence should be addressed. E-mail: [email protected] This article contains supporting information online at www.pnas.org/lookup/suppl/ doi:10.1073/pnas.1013435107/-/DCSupplemental. PNAS ∣ November 2, 2010 ∣ vol. 107 ∣ no. 44 ∣ 18755–18760 ECOLOGY Molecular oxygen (O2 ) is the second most abundant gas in the Earth’s atmosphere, but in many natural environments, its concentration is reduced to low or even undetectable levels. Although low-oxygen-adapted organisms define the ecology of low-oxygen environments, their capabilities are not fully known. These capabilities also provide a framework for reconstructing a critical period in the history of life, because low, but not negligible, atmospheric oxygen levels could have persisted before the “Great Oxidation” of the Earth’s surface about 2.3 to 2.4 billion years ago. Here, we show that Escherichia coli K-12, chosen for its well-understood biochemistry, rapid growth rate, and low-oxygen-affinity terminal oxidase, grows at oxygen levels of ≤3 nM, two to three orders of magnitude lower than previously observed for aerobes. Our study expands both the environmental range and temporal history of aerobic organisms. GEOLOGY Contributed by Donald E. Canfield, September 9, 2010 (sent for review April 4, 2010) 60 12 50 10 40 8 30 6 20 4 10 2 0 0 0 5 10 15 20 25 Cells mL-1 (x10 6) O2 (nM) A 30 O2 (nM) B 60 12 50 10 40 8 30 6 20 4 10 2 0 Cells mL-1 (x10 6) Time (h) 0 0 5 10 15 20 25 30 35 40 45 50 55 Time (h) Fig. 1. Oxygen concentrations and cell counts for respiration experiments with E. coli at two different pump rates of oxygen through the silicone tubing in the reactor. (A) A terminal pump rate of 36 mL h−1 was used corresponding to an oxygen input rate into the reactor of 11.5 μmol L−1 h−1 . During the first 6 h of incubation, the pump rate was incrementally increased from zero to 36 mL h−1 in an effort to maintain oxygen concentrations at a steady value. Instead, they fluctuated between 100 and 200 nM during this time. (B) A terminal pump rate of 12 mL h−1 was used, corresponding to an oxygen flux into the reactor of 5.5 μmol L−1 h−1 . The pump rate was brought to its terminal value after 35 min. In this particular experiment, external electrical noise produced random fluctuations in the oxygen signal. These random peaks do not reflect actual oxygen concentrations, but rather experimental noise. Oxygen concentrations are given by the open symbols “o,” with uncertainties given by the error bars (see SI Text for error explanation). Cell numbers are given in the closed symbols “•.” The minimum absorbance that could be measured by the spectrophotometer used was 0.001 (corresponding to 3.47 × 105 cells mL−1 . A value of 3.47 × 108 cells mL−1 per unit absorbance was calculated with a standard error of 8.96 × 106 cells mL−1 per unit absorbance (i.e., a standard error of 2.5%). The error bars for the cells mL−1 are given as 5%, i.e., 2 standard errors. In each of our experiments, the growth rate (G, in units of Δcells mL−1 h−1 ) was constant for a given oxygen flux (or pump rate) for periods up to 49 h, despite the fact that oxygen levels decreased over this time period to below the detection limit of the sensor (e.g., Fig. 1B). A constant growth rate is expected when a limiting nutrient or metabolite is provided at a constant rate (22), which in this case was oxygen. When comparing between experiments, G was proportional to the oxygen flux into the reactor (R2 ¼ 0.96) (Fig. 2A), providing additional evidence for the control of oxygen on growth. From the data in Fig. 2, we also found a consistent growth yield between experiments of 0.51 0.06 mol of cell carbon produced per mole of O2 respired. This value compares favorably with previous experiments of aerobic growth of E. coli on glycerol (23) (see SI Text for calculations). By contrast, the specific rate of growth, μðh−1 Þ, decreased during the course of an experiment and in relation to lower oxygen concentrations (Fig. 2B). The specific growth rate is defined as ln 2∕ðρ∕GÞ, where G is growth rate (defined above), and ρ is the cellular density (cells mL−1 ); μ is related to cell doubling time, tD , by tD ¼ ln 2∕μ. The lower the specific growth rate, the longer the doubling time and the slower the growth by the organism. Under conditions of oxygen limitation, μ should depend on 18756 ∣ www.pnas.org/cgi/doi/10.1073/pnas.1013435107 the concentration of oxygen in the media (22). This dependence is expected because under oxygen-limited conditions the flux of oxygen into the cell is driven by the diffusion gradient of oxygen into the cell. The flux of oxygen into a cell and subsequent specific rates of growth should, therefore, decrease with decreasing oxygen concentration. Likewise, under oxygen-limiting conditions, the oxygen concentration in the media decreases as the population grows. Because oxygen is removed by diffusion into cells, more cells means more surface area to diffuse across, and lower oxygen concentrations provide the same oxygen removal flux (balancing the constant input rate) across the increased surface area. The Monod equation [μ ¼ μmax ð½O2 ∕ð½O2 þ K s ] (22) expresses a common relationship between μ and substrate (in our case oxygen) concentration. In this equation, μmax expresses the maximum specific growth rate, and a value of 0.4 h−1 was obtained under high-oxygen concentrations in open-air experiments (see SI Text), whereas K s is the half saturation constant† with respect to growth. We used the Monod equation to predict how μ would respond to changing oxygen concentrations with variable K s † The K s of an organism for a particular substrate represents the concentration of that substrate at which an organism grows at half its maximal rate. Stolper et al. Flux (nmol l -1 h-1) 12,000 10,000 8,000 6,000 4,000 2,000 Flux ¼ 4πDrðC∞ − Ccell Þ; 0 100,000 200,000 300,000 400,000 500,000 Growth Rate (cells mL-1 h-1) B 0.45 Specific growth rate (h -1) 0.40 0.35 0.30 0.25 0.20 0.15 Data Ks 50 nM Ks 100 nM Ks 150 nM Ks 200 nM 0.10 0.05 0.00 1 10 100 1,000 10,000 100,000 1,000,000 Oxygen (nM) Fig. 2. (A) Growth rate of cells (cells mL−1 h−1 ) compared to the flux rate of oxygen into the reactor. Each experiment (other than the zero point) was run twice at a particular pump rate and the growth rates averaged. The error bar on the y axis is the standard error of the intercept. It demonstrates that the best-fit line passes through the origin within error. (B) Specific growth rates (h−1 ) vs. oxygen with various fits to a Monod growth model. A series of different K s values were explored with a constant μmax of 0.4 h−1 . See text for details. values, and we obtained a best-fit K s value of 121 20 nM O2 (Fig. 2B). The error represents the standard deviation around the best-fit slope of predicted vs. measured values (Fig. 2B). To our knowledge, this is the lowest K s ever recorded for aerobic growth. As described above, during aerobic respiration, oxygen is used to fuel cell growth and also for cell maintenance requirements. Maintenance energy refers the energy expended for all cellular processes that do not contribute to growth including cell repair. Typically, as μ decreases, maintenance energy becomes proportionally more important resulting in a decrease in growth yield (23). Indeed, cell maintenance requirements could put lower limits on rates of cell growth. Maintenance energy requirements could have played a role in our experiments, where values of μ were as low as 0.01 h−1 (reached at the end of the experiment depicted in Fig. 1B). Yet despite these low values of μ, growth yields remained constant both within (as observed by constant growth) and among the different experiments (Fig. 2A), providing no indication of an increasing significance of maintenance requirements as specific rates of growth decreased. Exploring the Lower Limits of Oxygen on Growth. Our results demonstrate aerobic growth by E. coli at oxygen concentrations of ≤3.0 nM. However, the lower oxygen limit for aerobic growth Stolper et al.  where Flux is the flux of oxygen into the cell, D is the diffusion coefficient for oxygen, r is the cell radius, C∞ is the oxygen concentration in the bulk media, and Ccell is the oxygen concentration at the outer cell surface. At 37 °C (the temperature of our experiments) and in fresh water, D is 3 × 10−5 cm2 s−1 (25). We assume that cells have the same carbon content per volume as typical E. coli (1.85 × 10−14 mol C μm−3 ; see SI Text), that they produce cell carbon with a growth yield of 0.5 mol C∕mol O2 respired (see above), and that they double when they have accumulated twice their normal cell C content. With these values and assumptions, we can reproduce our observed specific growth rates of E. coli at low O2 if we further assume that Ccell ¼ 1∕2C∞ (Fig. 3). With these values, we calculate the oxygen levels that produce doubling times of 30 and 15 d for a variety of different cell sizes (Table 1). We chose these doubling times because in Earth– surface environments, organisms must be able to respond to changing temperatures, moisture, and energy supply, which are dynamic on time scales of days to months. Such doubling times are also typical for prokaryotes in surface marine sediments and in the marine water column (26, 27). Model results show that for a 1-μm cell, an oxygen concentration of 0.2 nM would be sufficient to allow a doubling time of 30 d, and for a cell as small as 0.5 μm [which is small for a prokaryote cell (11)], 0.050 nM O2 would suffice (Table 1). We conclude that for organisms growing with doubling times of 30 d, 0.050 nM oxygen would appear adequate to fuel an aerobic ecosystem. Slower growing or smaller organisms could, in principle, grow on even lower oxygen concentrations than this. A key point is that although we have observed aerobic growth at orders of magnitude lower oxygen concentrations than might have been expected based on the Pasteur Point argument or has been previously observed, our results make sense based on simple, plausible models. 0.25 specific growth rate (h -1) 0 data model 0.2 0.15 0.1 0.05 0 0 10 20 30 40 50 60 70 O2 nM Fig. 3. Specific growth rates from our experiments compared to results from a diffusion-limited growth model. See text for details. PNAS ∣ November 2, 2010 ∣ INAUGURAL ARTICLE y = 0.0298x + 600.68 2 R = 0.9638 vol. 107 ∣ no. 44 ∣ 18757 GEOLOGY 14,000 in E. coli, even in our experiments, is not clear given that growth continued for up to 38 h after oxygen became undetectable using our STOX (Switchable Trace OXygen) sensors. As E. coli appears capable of growing at oxygen concentrations lower than the detection limit of the STOX sensor, we will explore the lower limits for growth by modeling the fluxes of oxygen across a cell membrane. As mentioned above, when oxygen is limiting, μ is limited by the diffusional flux of oxygen across the cell membrane, which depends on oxygen concentration. With this in mind, we can model how oxygen concentration limits μ for different-sized organisms. For a spherical cell, the flux of oxygen to the cell is given by (24): ECOLOGY A 16,000 Table 1. O2 concentrations required for cell doubling times of 15 and 30 d O2 concentration, nM Cell diameter, μM 4.0 3.0 2.0 1.5 1.2 1.0 0.8 0.6 0.5 30 d 3.2 1.7 0.79 0.44 0.28 0.20 0.13 0.07 0.05 15 d 6.4 3.4 1.6 0.88 0.56 0.40 0.26 0.14 0.10 Low-Oxygen Metabolism in Other Prokaryotes. The low-oxygen metabolism in E. coli is made possible by the high oxygen affinity of the cytochrome bd oxidase (9), which is expressed when oxygen levels fall below about 10% of air saturation (28). This highaffinity oxidase is well distributed among prokaryotes (29, 30), including cyanobacteria, many aerobes, some presumably “strict” anaerobes (e.g., Methanosarcina barkeri, Desulfovibrio vulgaris, and others), and also Bacteroides fragilis, which has been termed a “nanaerobe” (29). Nanaerobes can live as strict anaerobes, cannot grow in the presence of micromolar levels of oxygen, but can metabolize at submicromolar oxygen levels; however, the extent to which oxygen fuels the growth of these organisms is unknown. Another high-affinity oxidase, the cbb3-type cytochrome oxidase, was first found in the respiratory chain of the soybean root nodule bacterium Bradyrhizobium jaonicum (31) but is also widely distributed among the Bacteria (30). Although our results apply to only one organism, from the wide distribution of low-affinity oxidase enzymes in nature, we infer that the ability to metabolize and, in particular, to grow at nanomolar (and perhaps lower) concentrations of oxygen is potentially common in nature. Geobiological Implications. Low-oxygen environments are found at the oxic-anoxic interface of aquatic sediments (11), in many soils, in stratified lakes and marine basins (11, 32), and in OMZs of the global ocean (33). Our results place quantitative constraints on where in these environments aerobic metabolisms could be supported. This issue is of particular importance in some low-oxygen environments such as OMZs, where oxygen can be reduced to submicromolar (and perhaps even nanomolar or lower) concentrations (13). OMZs are found in upwelling regions, such as those of the Eastern Tropical Pacific, the Benguela upwelling current, and the Arabian Sea, and such regions are expected to expand as a result of global warming (34). With a single exception (13), the oxygen concentrations in these regions are not known to concentrations better than about 1–2 μM (35), but with our current results, the extent of possible aerobic metabolisms can be mapped onto measurements of high-resolution oxygen distributions as made possible with the STOX sensors used here. Our results also clearly show that aerobic metabolisms are probable in the Bay of Bengal, another OMZ where oxygen concentrations reach values as low as ∼2 μM (36). Such oxygen levels have typically been viewed as supporting so-called “suboxic” metabolisms such as nitrate, manganese, and iron reduction (37), with an uncertain role for aerobic processes. Our results demonstrate that aerobic metabolisms can operate under these conditions and possibly overlap with some of these suboxic anaerobic metabolisms. It will be important to establish the balance between aerobic and anaerobic metabolisms in modern, natural environments with micromolar to subnanomolar oxygen concentrations. The balance between aerobic and anaerobic metabolisms at low-oxygen is especially important because some anaerobic metabolisms such as denitrification can also occur in the presence of oxygen (38). Combined with in situ studies of denitrification 18758 ∣ www.pnas.org/cgi/doi/10.1073/pnas.1013435107 rates in oxygen gradients, our results will help provide an understanding of the cooccurrence and competition between aerobic and anaerobic processes in nature. The nature of the coupling between aerobic and anaerobic metabolisms in nanomolar oxygen concentrations is virtually unknown and deserves further research. Low-oxygen conditions are especially important in discussions of early life on Earth. Atmospheric oxygen levels are modeled to have been ≤10−5 PAL (19) before about 2.3 to 2.4 Ga, which corresponds to about 2.2 nM dissolved oxygen when saturated with seawater at 25 °C. We have demonstrated growth to nearly this oxygen concentration (3 nM, the detection limit of our STOX sensor) and modeled it to much lower concentrations (equivalent to about 2 × 10−7 PAL). As such, 10−5 PAL should support the growth of E. coli and potentially other aerobic microbes with similar high-affinity cytochromes. Therefore, environmental conditions may have allowed for aerobic metabolism well before the large rise in atmospheric oxygen 2.3 to 2.4 Ga (39, 40). This is different from many models of Earth history that require oxygen concentrations in the atmosphere to have been at least 10−2 PAL (i.e., based on the Pasteur Point argument) before aerobic organisms could begin using oxygen for respiration (1–3) but is consistent with the geologic record, where biomarkers preserved in rocks as old as 2.67 Ga suggest the presence of eukaryotes, cyanobacteria, and oxygen-utilizing heterotrophs and autotrophs (21). This is fully 300 million years before the accepted widespread oxidation of the atmosphere. The calibration of geochemical proxies can be imprecise, but we suggest that an aerobic biosphere could have been active from at least 2.67 Ga and onward at lower levels of oxygen than inferred by occasional late Archean “whiffs” of elevated oxygen concentrations (15, 17, 18, 41). We have demonstrated that an active and growing aerobic biosphere could have been maintained with oxygen concentrations at least down to about 10−5 PAL, and perhaps as low as 2 × 10−7 PAL. Low-oxygen concentrations in the 10−7 to 10−5 PAL range would have been rapidly depleted in metabolically active areas, and a challenge to the aerobic biosphere may have been access to sufficient oxygen for respiration. One could, however, imagine aerobic communities preferentially populating a thin veneer in soils over the land surface, being supplied by low levels of atmospheric oxygen. Aerobic populations might also have been found in aquatic environments in close association with oxygenproducing microbial mats and stromatolites and in the upper photic zone where cyanobacteria were active. One might envision that aerobic metabolisms began with the evolution of oxygen-producing cyanobacteria, providing a source of oxygen for aerobic respiration. If so, the combined elemental and isotopic evidence for oxygen whiffs and eukaryotic and cyanobacterial biomarkers (see above) place the evolution of aerobic metabolism before about 2.7 Ga (though see ref. 42 for a different view). Independent of these geological arguments, the history of oxidase enzymes places other constraints on the origins of aerobiosis (30, 43, 44). Although there is active debate on the phylogenetic history of oxidase enzymes, recent models (30, 43, 44) support an ancient origin, even before the evolution of oxygen-producing cyanobacteria (45). Without a cyanobacterial source, oxygen could have come from abiotic sources, such as the rainout and disproportionation (by catalase enzymes) of atmospherically produced H2 O2 at the Earth’s surface. This process could have supported elevated oxygen concentrations in local environments before the evolution of cyanobacteria (46, 47). Our results do not prove that aerobic pathways predated cyanobacterial evolution, but they allow further development of such a hypothesis. They also provide support for an aerobic biosphere after cyanobacterial evolution—but long before the general oxidation of the Earth’s atmosphere some 2.3 to 2.4 Ga. Stolper et al. Thus far, demonstrations of aerobic respiration at low nanomolar oxygen concentrations have been based on the absorption spectrum associated with variable oxygenation of root nodule leghaemoglobin (7). We used instead the recently developed polarographic STOX sensors, which are stable and precise down to oxygen concentrations of ≤3 nM (13) and are easily adapted to growth experiments. We studied E. coli K-12 [strain Sø8HLT (48)] because of its well-characterized metabolic system, rapid doubling times, and because it possesses both high- and low-affinity cytochrome oxidases (9). E. coli is a facultative aerobe that would normally switch to anaerobic metabolisms at 0.005–0.02 PAL (1–4 μM O2 in solution) (49). To ensure aerobic growth by E. coli, we used glycerol as a carbon source, which is generally considered unfermentable (50). Although recent studies have demonstrated glycerol fermentation by E. coli, including K-12 (51), this growth was only with the addition of tryptone or proteinogenic amino acids (51), which were not included in our media. Furthermore, control experiments without oxygen addition yielded no growth of E. coli in the presence of glycerol (Fig. S2), consistent with the absence of this metabolic pathway in our experiments. We conducted growth experiments on E. coli K-12 at constant oxygen fluxes into our reactor system, making continuous measurements of oxygen concentrations down to the detection limit of the STOX sensor. We sampled the media at intervals (generally 1 h) to measure the number of cells per unit volume from which we determined growth rates. Our experiments were performed in an all-glass, well-mixed, thermostated reactor sealed with a butyl rubber stopper. Oxygen was supplied diffusively from air-saturated water pumped at controlled rates through silicone tubing looping through the reactor. Details of the experimental setup (Fig. S3), the media used, and 1. Chapman DJ, Schopf JW (1983) Biological and biochemical effects of the development of an aerobic environment. Earths Earliest Biosphere, Its Origin and Evolution, ed JW Schopf (Princeton Univ Press, Princeton, NJ), pp 302–320. 2. Towe KM (1991) Aerobic carbon cycling and cerium oxidation: Significance for archean oxygen levels and banded iron-formation deposition. Palaeogeogr Palaeocl 97:113–123. 3. Goldblatt C, Lenton TM, Watson AJ (2006) Bistability of atmospheric oxygen and the great oxidation. Nature 443:683–686. 4. Fenchel T, Finlay BF (1995) Ecology and Evolution in Anoxic Worlds (Oxford Univ Press, Oxford). 5. Devol AH (1978) Bacterial oxygen uptake kinetics as related to biological processes in oxygen deficient zones of the oceans. Deep-Sea Res 25:137–146. 6. Longmuir IS (1954) Respiration rate of bacteria as a function of oxygen concentration. Biochem J 57:81–87. 7. Kuzma MM, Hunt S, Layzell DB (1993) Role of oxygen in the limitation and inhibition of nitrogenous activity and respiration rate in individual soybean nodules. Plant Physiol 101:161–169. 8. Bergersen FJ, Turner GL (1990) Bacteroids from soybean root nodules; respiration and N2 fixation inflow-chamber reactions with oxyleghemoglobin. Proc R Soc London Ser B 238:295–320. 9. Rice CW, Hempfling WP (1978) Oxygen-limited continuous culture and respiratory energy-conservation in Escherichia coli. Journal of Bacteriology 134:115–124. 10. Johnson MJ (1967) Aerobic microbial growth at low oxygen concentrations. J Bacteriol 94:101–108. 11. Canfield DE, Kristensen E, Thamdrup B (2005) Aquatic Geomicrobiology (Springer, New York). 12. Glud RN (2008) Oxygen dynamics of marine sediments. Mar Biol Res 4:243–289. 13. Revsbech NP, et al. (2009) Determination of ultra-low oxygen concentrations in oxygen minimum zones by the STOX sensor. Limnol Oceanogr-Meth 7:371–381. 14. Guo QJ, et al. (2009) Reconstructing Earth’s surface oxidation across the ArcheanProterozoic transition. Geology 37:399–402. Stolper et al. ACKNOWLEDGMENTS. We thank Sean Crowe, Joost Hoek, Jakob MøllerJensen, Peter Søholt, Preben Sørensen, and Bo Thamdrup for helpful discussions and assistance in the lab. Tim Lyons and Dave Johnston are thanked for insightful comments on the manuscript. D.A.S. was funded through the generous support of a Fulbright Fellowship from the American and Danish governments. Research was funded through the Agouron Institute and Danmarks Grundforskningsfond. 15. Anbar AD, et al. (2007) A whiff of oxygen before the Great Oxidation Event? Science 317:1903–1906. 16. Wille M, et al. (2007) Evidence for a gradual rise of oxygen between 2.6 and 2.5 Ga from Mo isotopes and Re-PGE signatures in shales. Geochim Cosmochim Acta 71: 2417–2435. 17. Reinhard CT (2009) A late archean sulfidic sea stimulated by early oxidative weathering of the continents. Science 326:713–716. 18. Frei R, Gaucher C, Poulton SW, Canfield DE (2009) Fluctuations in Precambrian atmospheric oxygenation recorded by chromium isotopes. Nature 461:250–253. 19. Pavlov AA, Kasting JF (2002) Mass-independent fractionation of sulfur isotopes in Archean sediments: Strong evidence for an anoxic Archean atmosphere. Astrobiology 2:27–41. 20. Ono S, et al. (2003) New insights into Archean sulfur cycle from mass-independent sulfur isotope records from the Hamersley Basin, Australia. Earth Planet Sci Lett 213:15–30. 21. Waldbauer JR, Sherman LS, Sumner DY, Summons RE (2009) Late Archean molecular fossils from the Transvaal Supergroup record the antiquity of microbial diversity and aerobiosis. Precambrian Res 169:28–47. 22. Monod J (1949) The Growth of Bacterial Cultures. Annu Rev Microbiol 3:371–394. 23. Hempfling WP, Mainzer SE (1975) Effects of varying the carbon source limiting growth yield and maintenance characteristics of Escherichia coli in continuous culture. J Bacteriol 123:1076–1087. 24. Fenchel T, King GM, Blackburn TH (1998) Bacterial Biogeochemistry: The Ecophysiology of Mineral Cycling (Academic, San Diego, CA) p 307. 25. Boudreau BP (1997) Diagenetic Models and Their Implementation (Springer, New York) p 414. 26. Ducklow HW, Carlson CA (1992) Oceanic bacterial production. Advances in Microbial Ecology, ed KC Marshall (Plenum, New York), pp 113–181. 27. Karl DM, Novitsky JA (1988) Dynamics of microbial growth in surface layers of a coastal marine sediment ecosystem. Mar Eco-Prog Ser 50:169–176. PNAS ∣ November 2, 2010 ∣ vol. 107 ∣ no. 44 ∣ 18759 INAUGURAL ARTICLE GEOLOGY Materials and Methods other aspects of the experimental protocols are found in SI Text). Growth was measured by removing aliquots from the reactor, measuring the absorbance at a wavelength of 450 nm, and converting to cell numbers based on a working curve of cell concentrations determined by direct cell counts vs. absorbance. Through a series of preliminary growth experiments and negative controls, we demonstrated that growth did not occur without a flux of oxygen into the reactor (Figs. S2 and S4). Several lines of evidence support this: (i) when a degassed reactor inoculated with E. coli in the growth medium was not exposed to oxygen, no growth occurred (see SI Text for details; Fig. S2); (ii) when the flux of oxygen into the reactor was stopped, measurable growth stopped immediately (Fig. S4); and (iii) as demonstrated below, growth rates are linearly correlated with oxygen flux and have an intercept through the origin within error. Based on these observations, we conclude that growth in our experiments was due to aerobic respiration, and we rule out anaerobic respiration as a measurable source of growth. We also explored and subsequently rejected the hypothesis that our results were strongly influenced by biofilm growth on the silicone tubing transporting oxygen into the reactor or by unobserved high-oxygen microniches (see SI Text). In a typical experiment, we inoculated E. coli into our reactor containing 1–2 μM oxygen (degassed previously with nitrogen; SI Text). This oxygen was drawn down within 1.5 h to nanomolar concentrations and represents the initial decline of oxygen in an experiment (Fig. S4). We then began to circulate oxygenated water through the silicone tubing providing a flux of oxygen to the culture. In some experiments, we attempted to regulate oxygen to a steady concentration by increasing the pumping rate (and thus oxygen flux into the reactor) in concert with the growing E. coli population (Fig. S4). This regulation proved to be difficult, and our most useful results came from establishing a set pump rate, and hence a set oxygen flux into the reactor (Fig. S1). Results from two of these constant-flux experiments are shown in Fig. 1, and all calculations are done using data collected during times of constant oxygen flux into the reactors. Therefore, all growth rate calculations were performed on data generated after the final, constant pump rate was reached. At the oxygen levels of our experiment, the residence time (mass O2 in reactor/flux rate) for oxygen within the reactor was short, from less than 1 to 30 s, depending on pump rate and O2 concentration. ECOLOGY Conclusions Our experiments demonstrate that aerobic respiration is coupled to growth at oxygen levels over two orders of magnitude lower than previously demonstrated. We believe that this result encourages a reevaluation of the distribution of low-oxygen adapted organisms in modern ecosystems and offers tantalizing quantitative constraints on the potential distributions and activities of aerobes on both the modern and ancient Earth as a function of available oxygen. We also provide a potential explanation for observations that suggest that aerobic respiration evolved before the rise of oxygen ∼2.3–2.4 Ga and to the hypothesis that aerobic metabolic pathways may have evolved even before the evolution of cyanobacteria. We also demonstrate that the conventional view of the history of aerobic respiration through the lens of the Pasteur Point is not only incorrect, but incorrect by at least three orders of magnitude. 28. Tseng CP, Albrecht J, Gunsalus RP (1996) Effect of microaerophilic growth conditions on expression of the aerobic (cyABCDE and cydAB) and anaerobic (narGHJI, frdABCD, dmsABC) respiratory pathway genes in Escherichia coli. J Bacteriol 178:1094–1098. 29. Baughn AD, Malamy MH (2004) The strict anaerobe Bacteroides fragilis grows in and benefits from nanomolar concentrations of oxygen. Nature 427:441–444. 30. Ducluzeau A-L, Ouchane S, Nitscke W (2008) The cbb3 oxidase as are an ancient innovation of the Domain Bacteria. Mol Biol Evol 25:1158–1166. 31. Preisig O, Zufferey R, ThonyMeyer L, Appleby CA, Hennecke H (1996) A high-affinity cbb3 -type cytochrome oxidase terminates the symbiosis-specific respiratory chain of Bradyrhizobium japonicum. J Bacteriol 178:1532–1538. 32. Richards FA (1965) Anoxic basins and fjords. Chemical Oceanography, eds JP Riley and G Skirrow (Academic, London), 1, pp 611–645. 33. Codispoti LA (1983) Nitrogen in upwelling systems. Nitrogen in the Marine Environment, eds EJ Carpenter and DG Capone (Academic, New York). 34. Stramma L, Johnson GC, Sprintall J, Mohrholz V (2008) Expanding oxygen-minimum zones in the tropical oceans. Science 320:655–658. 35. Paulmier A, Ruiz-Pino D (2009) Oxygen minimum zones (OMZs) in the modern ocean. Prog Oceanogr 80:113–128. 36. Naqvi SWA, Narvekar PV, Desa E (2005) Coastal biogeochemical processes in the North Indian Ocean. The Sea—Ideas and Observations on the Progress in the Study of the Seas, eds AR Robinson and KH Brink (Harvard Univ Press, Cambridge, MA), Vol 14, part A, pp 723–780. 37. Karstensen J, Stramma L, Visbeck M (2008) Oxygen minimum zones in the eastern tropical Atlantic and Pacific oceans. Prog Oceanogr 77:331–350. 38. Robertson LA, Dalsgaard T, Revsbech NP, Kuenen JG (1995) Confirmation of aerobic denitrification in batch cultures, using gas chromatography and N-15 mass spectrometry. FEMS Microbiol Ecol 18:113–119. 39. Farquhar J, Bao HM, Thiemens M (2000) Atmospheric influence of Earth’s earliest sulfur cycle. Science 289:756–758. 18760 ∣ www.pnas.org/cgi/doi/10.1073/pnas.1013435107 40. Bekker A, et al. (2004) Dating the rise of atmospheric oxygen. Nature 427:117–120. 41. Kendall B, et al. (2010) Pervasive oxygenation along late Archean ocean margins. Nat Geosci 3:647–652. 42. Kirschvink JL, Kopp RE (2008) Palaeoproterozoic ice houses and the evolution of oxygen-mediating enzymes: The case for a late origin of photosystem II. Philos Trans R Soc London B 363:2755–2765. 43. Brochier-Armanet C, Talla E, Gribaldo S (2009) The multiple evolutionary histories of dioxygen reductases: Implications for the origin and evolution of aerobic respiration. Mol Biol Evol 26:285–297. 44. Pereira MM, Santana M, Teixeira M (2001) A novel scenario for the evolution of haem-copper oxygen reductases. Biochim Biophys Acta 1505:185–208. 45. Castresana J (1994) Evolution of cytochrome oxidase, an enzyme older than atmospheric oxygen. EMBO J 13:2516–2525. 46. Liang MC, Hartman H, Kopp RE, Kirschvink JL, Yung YL (2006) Production of hydrogen peroxide in the atmosphere of a Snowball Earth and the origin of oxygenic photosynthesis. Proc Natl Acad Sci USA 103:18896–18899. 47. Kasting JF, Pollack JB, Crisp D (1984) Effects of high CO2 levels on surface-temperature and atmospheric oxidation-state of the early Earth. J Atmos Chem 1:403–428. 48. Valentin-Hansen P, Svenningsen BA, Munch-Petersen A (1978) Regulation of the deo operon in Escherichia coli. Mol Gen Genet 159:191–202. 49. Becker S, Holighaus G, Gabrielczyk T, Unden G (1996) O2 as the regulatory signal for FNR-dependent gene regulation of Escherichia coli. J Bacteriol 178:4515–4521. 50. Booth IR (2005) Glycerol and methylglyoxal metabolism. EcoSal Escherichia coli and Salmonella: Cellular and Molecular Biology, ed RI Curtis (ASM Press, Washington, DC), www.ecosal.org. 51. Murarka A, Dharmadi Y, Yazdani SS, Gonzalez R (2008) Fermentative utilization of glycerol by Escherichia coli and its implications for the production of fuels and chemicals. Appl Environ Microbiol 74:1124–1135. Stolper et al.
Metagenome-Based Metabolic Reconstruction Reveals the Ecophysiological Function of Hydrocarbon-Contaminated Sulfidic