Report

Biological signatures in clumped isotopes of O2

See allHide authors and affiliations

Science  24 Apr 2015:
Vol. 348, Issue 6233, pp. 431-434
DOI: 10.1126/science.aaa6284

What controls clumped isotopes?

Stable isotopes of a molecule can clump together in several combinations, depending on their mass. Even for simple molecules such as O2, which can contain 16O, 17O, and 18O in various combinations, clumped isotopes can potentially reveal the temperatures at which molecules form. Away from equilibrium, however, the pattern of clumped isotopes may reflect a complex array of processes. Using high-resolution gas-phase mass spectrometry, Yeung et al. found that biological factors influence the clumped isotope signature of oxygen produced during photosynthesis (see the Perspective by Passey). Similarly, Wang et al. showed that away from equilibrium, kinetic effects causing isotope clumping can lead to overestimation of the temperature at which microbially produced methane forms.

Science, this issue p. 431; p. 428; see also p. 394

Abstract

The abundances of molecules containing more than one rare isotope have been applied broadly to determine formation temperatures of natural materials. These applications of “clumped” isotopes rely on the assumption that isotope-exchange equilibrium is reached, or at least approached, during the formation of those materials. In a closed-system terrarium experiment, we demonstrate that biological oxygen (O2) cycling drives the clumped-isotope composition of O2 away from isotopic equilibrium. Our model of the system suggests that unique biological signatures are present in clumped isotopes of O2—and not formation temperatures. Photosynthetic O2 is depleted in 18O18O and 17O18O relative to a stochastic distribution of isotopes, unlike at equilibrium, where heavy-isotope pairs are enriched. Similar signatures may be widespread in nature, offering new tracers of biological and geochemical cycling.

Statistical thermodynamics predicts that heavy isotopes will be bound together in a molecule more often than predicted by chance alone, provided the system is at isotopic equilibrium (1, 2). This preference for heavy-isotope pairing and its variation with temperature forms the basis of clumped-isotope thermometry (35), a class of approaches based on precise measurements of molecules containing more than one rare isotope. When isotope-exchange reactions facilitate the equilibration of heavy-isotope pairs, the resulting isotopic distribution has indeed been shown to achieve equilibrium across a wide range of temperatures (4, 68); however, isotopic equilibrium is the exception rather than the rule in nature. Biogenic substances, for example, are often formed through irreversible enzymatic reactions for which isotope-exchange equilibrium cannot be expected a priori. Yet, many natural materials with kinetically constrained and/or biological origins (e.g., carbonate shells) show only minor departures from equilibrium isotope fractionation (911). Large biological and physical effects on heavy-isotope pairing could complicate the interpretation of emerging clumped-isotope thermometers in methane, O2, and other candidate systems (4, 5, 12).

Here, we consider photosynthetic O2 formation from water at the oxygen-evolving complex of Photosystem II (OEC). In the OEC, O–O bond formation occurs at the end of a five-step light-dependent sequence (Fig. 1). This reaction most likely does not equilibrate O–O isotope pairs given the lack of isotopic equilibration between water and the O2 produced (1316). We argue that the tendency for two heavy oxygen isotopes to be bound together during oxygenic photosynthesis reflects primarily the isotopic preferences of water molecules binding to the OEC. These patterns of heavy-isotope pairing should be apparent in clumped isotopes of O2. Measurements of the 18O18O (mass 36) and 17O18O (mass 35) isotopologues of O2, together with bulk isotopic ratios (18O/16O and 17O/16O), characterize the number of heavy-isotope pairs in a sample relative to the number expected by chance alone (i.e., the stochastic distribution). These deviations are quantified as Δ36 and Δ35 values: Excesses of 18O18O and 17O18O relative to the stochastic distribution of isotopes in the sample results in Δ36 > 0 and Δ35 > 0, respectively. A deficit in 18O18O and 17O18O results in Δ36 < 0 and Δ35 < 0.

Fig. 1 Conceptual diagram of O2 formation at the OEC.

The five-step Kok cycle for the water-splitting reaction 2H2O + 4 → O2 + 4H+ + 4e is shown without electron flow (32). Transitions between intermediate oxidation states of the OEC (S0 to S4) occur upon absorption of visible light. The water-binding sequence is based on experimental results (19, 33, 34), which also indicate that water substrates are exchangeable at least up to state S3 on chemically distinct binding sites (18, 19). The O–O bond is formed during the S4-to-S0 transition, expressing the isotopic fractionations αA and αB from water substrate binding.

The Δ36 and Δ35 signatures of oxygenic photosynthesis can thus be estimated by assigning each water-binding site its own isotopic fractionation factor α = 18Rbound/18Rwater, where 18R is the ratio of 18O to 16O atoms in each reservoir. At natural isotopic abundances, the bulk isotopic composition of photosynthetic O2 is the weighted sum of those contributions—i.e., 18Rp ≈ ½ [(18Rwater × αA) + (18Rwater × αB)], with binding sites A and B each contributing one of two oxygen atoms in each O2 molecule. The probability of generating 18O–‒18O bonds is therefore 36Rp = (18Rwater × αA)(18Rwater × αB). The stochastic distribution of 18O atoms is calculated from the bulk 18O/16O ratio as 36Rstochastic = (18Rp)2. The expression for Δ36,p then reduces to (17)

Embedded Image(1)

Equation 1 reveals that, in all cases, Δ36,p ≤ 0; contrary to the enhanced isotope pairing that would be expected at isotopic equilibrium, there is an apparent aversion to heavy-isotope pairing associated with photosynthetic O2 production. If the isotopic preferences at each water-binding site are equal (αA = αB), then Δ36,p = 0. If the binding sites are not equivalent (αA ≠ αB), as isotope-labeling studies indicate (18, 19), then 0 ≥ Δ36,p > −0.9 per mil (‰) for plausible α-values between 0.97 and 1.03 (20, 21). A similar expression can be derived for Δ35,p values, which are predicted to be about half those of Δ36,p (see the supplementary text). These values cannot be interpreted as formation temperatures because all equilibrated samples have Δn ≥ 0 (2). Photosynthesis should therefore impart a distinct nonequilibrium clumped-isotope signature on O2.

We conducted a closed-system terrarium experiment with six water hyacinths (Eichhorniae crassipes) to explore the effects of biological oxygen cycling on five isotopologues of O2 (17). The terrarium was illuminated with fluorescent lights on a 12-hour/12-hour light-dark cycle. Headspace samples were purified and analyzed over a 1-year period for both the bulk and clumped isotopic composition of O2. We found that biological oxygen cycling altered isotopic ordering in the headspace O2, yielding apparent steady-state Δ36 and Δ35 values that are inconsistent with O2 formation temperatures and more consistent with the predicted photosynthetic endmembers (Fig. 2 and table S3). The Δ36 and Δ35 values of O2 were driven down from atmospheric values [2‰ and 1‰, respectively (4)] and down past equilibrium values at 25°C (1.5‰ and 0.8‰, respectively), finally approaching an apparent isotopic steady state at the stochastic distribution of isotopes (Δ36 = –0.01 ± 0.08‰, and Δ35 = 0.0 ± 0.1‰; 1 SEM, n = 4). The plant community shifted to an algae-dominated ecosystem during the first 6 months, altering the isotopic, chemical, and physical properties of the terrarium (fig. S1). However, the clumped-isotope composition of the headspace O2 evolved steadily toward its apparent steady state, similar to the evolution of the oxygen triple-isotope composition. Steady-state Δ′17O values were 165 parts per million (ppm), consistent with those reported in similar experiments (22, 23).

Fig. 2 Evolution of concentration and O2 isotopologue composition in the terrarium.

Observations (data points) are compared with model results (curves). Uncertainties are not shown for clarity, but long-term analytical uncertainties in O2 concentration, δ′18O, Δ′17O, Δ36, and Δ35 are 1%, 0.04‰, 5 ppm, 0.17‰, and 0.3‰, respectively. A single isotopologue discrimination factor (34εR = ‒17‰) is used here to illustrate steady-state behavior in δ′18O and Δ′17O; a more detailed model run yields better agreement for δ′18O and Δ′17O but similar results for Δ36 and Δ35. Mass-dependent exponents used in the model, β34/n, are labeled, with subscripts R and GE denoting values for respiration and gas exchange, respectively. For β34/35,GE and β34/36,GE, two model runs are shown to illustrate their effects on the Δ36 and Δ35 time traces (17).

Dark incubations of the terrarium, which consumed up to 35% of the headspace O2, caused Δ36 values to increase linearly with time up to ~1‰ (Fig. 2). The Δ35 values, in contrast, remained generally constant (means of Δ35 = 0.1 ± 0.1‰ and 0.1 ± 0.05‰; 1 SD). Returning to light-dark cycles restored the clumped-isotope composition to its apparent steady-state value after 6 months (Δ36 = –0.09 ± 0.06‰, and Δ35 = 0.0 ± 0.1‰; 1 SEM, n = 3). To test the veracity of these measurements, headspace O2 samples drawn from both light and dark incubations were photolytically equilibrated at known temperatures (4). The equilibrations yielded Δ36 and Δ35 values of O2 consistent with isotope-exchange equilibrium (table S3), suggesting that our observations are unlikely to be analytical artifacts. Atmospheric O2 leaking into the terrarium would increase δ′18O far too rapidly relative to Δ36 to explain these observations. The observed clumped-isotope variations therefore most likely arise from biological and physical processes inside the terrarium.

We constructed a two-reservoir model of O2 (i.e., in headspace and water) in the terrarium that accounts for photosynthetic O2 formation, fractionation of O2 due to respiration, and air-water gas exchange (17). We included kinetic isotope fractionation for gas transfer into and out of solution [34αGE,kinetic = 0.9972 for 18O/16O (24)]. The model was run with a range of plausible isotope fractionation factors for respiration [34αR = 0.97 – 0.99 (25, 26)] and gas-exchange rates (24, 27) to examine the sensitivity of the terrarium headspace to changes in those quantities. The oxygen triple-isotope composition of the terrarium water was measured and used as the bulk isotopic composition of photosynthetic O2 (13, 15, 17). No single set of parameters explained all of the isotopic variations during the entire experiment, likely due to the evolving biological community, so we focus on isotopic variations at steady state and during dark incubations.

The increase of headspace Δ36 and Δ35 values in the dark implies that the apparent steady-state values near zero can only be reached if light-dependent processes drive Δ36 and Δ35 values below zero. Equation 1 suggests that photosynthesis could be the relevant mechanism, because the O2 generated is likely to have Δ36,p and Δ35,p values less than zero. To estimate the composition of this source, we note that kinetic and equilibrium isotope effects for relevant photosynthetic fractionations are probably in the range 0.96 > 18α > 1.04 (20, 21), which we broaden to a more conservative plausible range of 0.9 > 18α > 1.1. This range of isotope effects gives lower limits on Δ36,p and Δ35,p of –10‰ and –5‰, respectively.

If the Δ36 increase during dark incubations were solely caused by fractionation in respiration, then large isotope effects in water-enzyme binding would be required: Δ36,p < –10‰ is needed to achieve steady-state values of Δ36 near zero (17). In addition, the associated Δ35,p < –5‰ endmember composition causes poor agreement between measured and modeled Δ35 values (fig. S4C). Furthermore, an increase in respiration rates would drive Δ36 and Δ35 values higher, whereas a decrease in respiration rates would drive the O2 toward its Δ36,p and Δ35,p photosynthetic values (17). Therefore, when the O2 cycle was out of balance in the first 6 months, Δ36 would have fluctuated inversely with O2 concentration (fig. S4, B and C). Instead, both Δ36 and Δ35 decreased nearly monotonically.

Isotopologue fractionation during nonequilibrium O2 gas exchange could explain the increases of headspace Δ36 and Δ35 values during dark incubations. The fractionation in headspace 16O18O/16O2 is closer to that for gas exchange than that for respiration (34αobserved = 0.995 versus 34αGE,kinetic = 0.9972 versus 34αR ~ 0.98), suggesting that the Δ36 and Δ35 increases are similarly dominated by gas exchange. Modeling the mass dependence of gas exchange using the dark incubation data yields Δ36,p and Δ35,p values within a plausible range (i.e., Δ36,p = –0.4‰, Δ35,p = –0.2‰) (Fig. 2). The evolution of Δ36 and Δ35 is also more robust to imbalances in the O2 cycle (17). Other oxygen-consumption mechanisms, such as sulfide oxidation, could impart additional isotopologue signatures (28), so attributing isotopologue discrimination in the dark to a single process is necessarily a simplification. Indeed, the implied mass dependence of O2 consumption in the dark terrarium is unusual, and it merits further investigation (17). A detailed understanding of isotopologue fractionation factors will require more controlled experiments of isolated biological and physical processes. Yet, the specific isotopologue discrimination during dark incubations does not affect the conclusion that photosynthesis generates O2 with an “anticlumped” isotopologue distribution (i.e., Δ36 ≤ 0 and Δ35 ≤ 0). This biological signature in O2 may be readily observed in the surface ocean, where it could be used to constrain gross primary productivity by exploiting the contrast between biological and atmospheric O2 clumped-isotope signatures (29). Isotopic ordering in atmospheric O2 is relatively unaffected by biological O2 cycling because photochemical equilibration of O2 exceeds rates of biological cycling by at least a factor of 100 (4, 30). Using a biological endmember composition of Δ36 = 0, we calculate that biological effects on the tropospheric Δ36 budget are therefore most likely on the order of 0.01‰.

Our observations indicate that variations in the isotopologue abundance of even simple molecules like O2 capture the chemistry of complex natural systems. Broader application of these techniques could yield insights into the mechanisms of biomolecule synthesis, e.g., methanogenesis, nitrogen reduction during denitrification, and molecular hydrogen release during nitrogen fixation (31). Moreover, because clumped-isotope signatures can depend only on isotope fractionation factors and not on the isotopic composition of substrates, a new class of reservoir-insensitive approaches for tracing biogeochemical cycling could emerge from these molecular-scale insights.

Supplementary Materials

www.sciencemag.org/content/348/6233/431/suppl/DC1

Materials and Methods

Supplementary Text

Figs. S1 to S5

Tables S1 to S3

References (3550)

References and Notes

  1. Materials and methods are available as supplementary materials on Science Online.
  2. Acknowledgments: We thank H. Hu and N. Levin for performing oxygen triple-isotope analyses of the terrarium water at Johns Hopkins University, and E. Schauble for helpful discussions during the course of this work. This research was supported in part by the National Science Foundation (EAR-1049655 and DGE-1144087), the National Aeronautics and Space Administration Cosmochemistry program, and the Deep Carbon Observatory. The data and model parameters used in this study are available in the supplementary materials (tables S1 to S3).
View Abstract

Navigate This Article