Probing the metabolic water contribution to intracellular water using oxygen isotope ratios of PO₄

Significance

Conventional thought holds that the stable isotopic composition (O and H) of water inside the cells of microorganisms and most aquatic macroorganisms is identical to that of the surrounding water. This assumption is widely applied in studies of environmental reconstruction. Results presented here, based on techniques of multilabeled water isotope probing and measurement of ¹⁸O/¹⁶O ratios in PO₄ moieties of intracellular biomolecules, which allows direct sampling of intracellular environments, demonstrate a significant metabolic water component in intracellular water that is consistent across multiple strains of bacteria when grown under similar conditions. The intracellular water probe presented here, based on PO₄ δ¹⁸O in DNA/biomass, may be extended to other biomolecules and organisms.

Abstract

Knowledge of the relative contributions of different water sources to intracellular fluids and body water is important for many fields of study, ranging from animal physiology to paleoclimate. The intracellular fluid environment of cells is challenging to study due to the difficulties of accessing and sampling the contents of intact cells. Previous studies of multicelled organisms, mostly mammals, have estimated body water composition—including metabolic water produced as a byproduct of metabolism—based on indirect measurements of fluids averaged over the whole organism (e.g., blood) combined with modeling calculations. In microbial cells and aquatic organisms, metabolic water is not generally considered to be a significant component of intracellular water, due to the assumed unimpeded diffusion of water across cell membranes. Here we show that the ¹⁸O/¹⁶O ratio of PO₄ in intracellular biomolecules (e.g., DNA) directly reflects the O isotopic composition of intracellular water and thus may serve as a probe allowing direct sampling of the intracellular environment. We present two independent lines of evidence showing a significant contribution of metabolic water to the intracellular water of three environmentally diverse strains of bacteria. Our results indicate that ∼30–40% of O in PO₄ comprising DNA/biomass in early stationary phase cells is derived from metabolic water, which bolsters previous results and also further suggests a constant metabolic water value for cells grown under similar conditions. These results suggest that previous studies assuming identical isotopic compositions for intracellular/extracellular water may need to be reconsidered.

Metabolic water, more precisely defined as an isotopically distinct flux of O (and H) produced during metabolism (1), has been studied extensively as an alternative water source contributing to body water in animals, such as desert mammals, insects, and migrating birds (2–6), but does not easily lend itself to direct measurement. In recent years, interest in metabolic water has been extended to its oxygen isotopic composition (¹⁸O/¹⁶O ratio or δ¹⁸O value) and contribution to body water because this information is crucial to the interpretation of biomineral [e.g., carbonate—CaCO₃—and phosphate—Ca₃(PO₄)₂] oxygen isotopic compositions used heavily in paleoclimate/paleohydrological research (7–14). This includes biomineral shells of aquatic marine organisms that are preserved in the geologic record and used to infer Earth’s climate history (15–19).

A core assumption of applications of biomineral oxygen isotopic compositions to infer environmental conditions is that the ¹⁸O:¹⁶O ratio is controlled by exchange of oxygen isotopes between oxyanions comprising the biomineral and ambient water in bodily fluids (i.e., body water) (9, 20–24). In multicellular eukaryotic organisms, body water includes all water found in various body compartments (e.g., intravascular/intercellular) and bodily fluids (e.g., blood plasma, urine, breath vapor), including water produced by metabolism, and is averaged over the entire organism (25, 26). Blood is the largest reservoir of body water in mammals. Accordingly, most previous studies have been based on measurements of total body water in blood, in urine, or in breath CO₂ that has exchanged and presumably equilibrated with body water, and, thus, not on direct measurement of intracellular water.

A strong linear relationship has been observed between δ¹⁸O values of body water determined from mammal blood and δ¹⁸O values of extracellular (ingested) water, which is equivalent to local meteoric water (1, 27–30). A contribution of metabolic water to body water can be detected using this linear relation based on deviation of the slope from a value of 1 (7, 8, 28, 30–32). Based on slopes and modeling calculations, the percentage of metabolic water has been determined to vary between 7% and 56% among different mammal species (7, 8, 30–34).

In contrast to macroorganisms and mammals, microorganisms are primarily unicellular (e.g., bacteria and archaea), and, thus, body water is equivalent to intracellular water, which has traditionally been assumed to be identical to water in the surrounding extracellular medium (35), without consideration of a metabolic water component.

Here we present two independent lines of evidence for a significant contribution of metabolic water to the intracellular water pool of bacterial cells, based on a new approach for detecting the contribution of metabolic water by direct sampling of the intracellular environment through measurement of the ¹⁸O:¹⁶O ratio of PO₄ (i.e., δ¹⁸O_PO4) in intracellular molecules (e.g., DNA) and also PO₄ in total biomass. Our findings are consistent with recent results of Kreuzer-Martin et al. (27, 36) indicating a metabolic water contribution to the intracellular water of Escherichia coli cells of as much as 70% during the log phase of growth and up to 27% for cells in the stationary phase. Knowledge of the amount and also the isotopic composition of a metabolic water component of body/intracellular water is important to studies of cell physiology and metabolism in which the ¹⁸O:¹⁶O ratio of both water and oxyanions such as CO₂/CO₃ and PO₄ are used to track metabolic processes and reaction pathways.

Intracellular Water Isotopic Composition

Water within the cytoplasm of cells is generally assumed to diffuse freely across the cell membrane and thus have an identical ¹⁸O:¹⁶O ratio to ambient water outside of cells. The same assumption holds for body water of aquatic organisms such that ambient extracellular water oxygen isotopic compositions are used to determine apparent fractionations between body water and biomineral/biomass components that contain oxygen (e.g., refs. 19 and 21–24).

For nonaquatic, terrestrial organisms (e.g., mammals) that are relatively closed systems with respect to inflow/outflow of ambient environmental water (e.g., no free diffusion of water across the outer membrane), a significant component of metabolic water has been observed in the intracellular water pool (e.g., mammals), with the metabolic water contribution increasing with increased metabolic rates and decreased body size (6, 30, 34, 37).

The ¹⁸O:¹⁶O ratio in water varies with chemical reaction rates/pathways and also physical processes such as evaporation, precipitation, adsorption, and diffusion (38). Water ¹⁸O:¹⁶O ratios (henceforth expressed as δ¹⁸O_WATER values) also vary due to the exchange or replacement of oxygen atoms (i.e., isotopes) in water with oxygen in other molecules. In the case of exchange of oxygen between water and the oxyanions that are precipitated within/by cells as biominerals (e.g., CO₃²⁻, PO₄³⁻, and SO₄²⁻), ¹⁸O:¹⁶O ratios often vary according to temperature, which has led to construction of a number a of oxyanion−water−temperature equations or “paleothermometers” that are widely used in the study of Earth’s climate (21, 23, 39–41). A primary assumption underlying the use of these oxygen isotope paleothermometers to reconstruct environmental temperatures is that oxyanion ¹⁸O:¹⁶O ratios reflect equilibrium exchange with ambient environmental water that is further assumed to be identical in oxygen isotopic composition to the intracellular water.

Approach: Oxygen Isotopic Composition of Water Imprinted onto Intracellular PO₄

Several recent studies in our group using stable isotope probing to elucidate intracellular phosphoenzyme reaction mechanisms and attendant isotope effects led to the hypothesis underlying the present work, which is as follows: PO₄ can undergo complete and rapid exchange of all four O atoms with intracellular water due to the action of the ubiquitous, intracellular enzyme inorganic pyrophosphatase (PPase) (Fig. 1 and Fig. S1) (42, 43). Previous experiments using the purified cell-free PPase enzyme, inorganic PO₄, and water firmly demonstrate that the oxygen in dissolved inorganic PO₄ (DIP) is rapidly and completely (100%) exchanged with oxygen in ambient water in the medium (Fig. S1), such that a plot of δ¹⁸O_WATER vs. δ¹⁸O_PO4 gives a slope of 1 reflecting that all four oxygens in PO₄ are derived from ambient water in the medium (42). Thus, cellular compounds and biomass that are synthesized from the intracellular dissolved PO₄ pool should also carry the isotopic signature imprinted by PPase (Fig. 1).

Fig. 1.

Schematic sketch of PPase-catalyzed O exchange between intracellular water and the intracellular DIP pool, and its transfer to DNA/biomass.

Fig. S1.

Approach of PO₄−water fractionations to steady-state, equilibrium values from opposite directions (i.e., in different isotopically labeled waters) due to PO₄−water exchange catalyzed by inorganic pyrophosphatase (PPase) at 27 °C. Reprinted from ref. 43.

In the present study, we tested the above-mentioned hypothesis, shown schematically in Fig. 1, by performing a series of controlled culture experiments to look for the imprinting of water ¹⁸O:¹⁶O ratios onto the intracellular dissolved PO₄ pool due to the action of intracellular PPase or other intracellular enzymes with similar isotope effects, and its subsequent transference into cell biomass (e.g., DNA) that is synthesized from the intracellular PO₄ pool. This should permit PO₄ in cellular biomass to serve as a proxy for intracellular water isotopic composition, assuming complete exchange of oxygen in PO₄ with oxygen in water and minimal subsequent overprinting by other cellular processes.

In contrast to previous studies reporting direct measurement of intracellular water from only one microbial strain, E. coli (27, 36), we tested several different strains of bacteria to look for species-/strain-specific effects and also used the technique of multilabeled water isotope probing (MLWIP), used in previous studies of metabolic water in rats (30) and further developed in our group to probe organophosphorus and phosphoenzyme reaction mechanisms (23, 42–44). Using this technique, multiple (three or more) labeled-water probes having different oxygen isotopic composition are used to interrogate reaction mechanisms and trace oxygen transfers between compounds in reactions. This technique contrasts with traditional isotope probing methods that typically use only a single probing compound that is usually strongly labeled with one (or more) heavy or even radioactive isotope(s). Importantly, MLWIP methods use both heavy and light isotopic probes that also have isotopic compositions close to natural abundances to provide both higher sensitivity and additional insights into the reaction mechanism, such as the fractionation involved during isotope exchange and transfer/replacement reactions (23, 29, 30, 42, 44). Here we used the MLWIP method, combined with graphical analysis, to both identify a metabolic water component in the intracellular water pool and to confirm the transfer of the oxygen isotopic signature of intracellular water to phosphate (PO₄) in DNA and cellular biomass.

Results and Discussion

Results from 14 separate 1- to 2-L microbial culture experiments are shown in Table S1 and plotted in Figs. 2–4. Oxygen isotope ratios are reported as delta (δ) values in units of per mil and referenced to the Vienna Standard Mean Ocean Water (VSMOW) international reference standard. The δ¹⁸O_PO4 values are reported with a precision of ±0.2–0.3‰. Growth experiments were carried out under optimal conditions of temperature, pH, and medium composition for each organism. All strains were grown aerobically. Each data point plotted in Figs. 3 and 4 represents a completely independent 1- to 2-L experimental growth system that underwent extensive processing to extract biomass and to extract/purify DNA [after Blake et al. (45)], which, due to the large volumes of these systems, required conventional chemical methods to extract and purify biomass/DNA with all reagents prepared from scratch and appropriately scaled (see SI Materials and Methods), versus the more typically used premade and ready-to-use small-scale DNA extraction kits used in studies involving PCR and/or sequencing. Added to this was an extensive series of steps to next extract PO₄ from biomass/DNA, purify it, and convert it to silver phosphate (Ag₃PO₄) for stable oxygen isotope analysis. Despite the large number of complex steps involved in processing/analyzing these samples, the δ¹⁸O_PO4 values of both DNA and biomass are highly consistent for a given set of conditions [e.g., δ¹⁸O_WATER, temperature (T)], even across different microbial strains (Table S1). For example, the ∼−5.0‰ water culture experiments at 23–24 °C yielded DNA δ¹⁸O_PO4 values for P.fluorescens, Acinetobacter ADP1, and M. aquaeolei of 17.2‰, 17.9‰, and 17.4‰, respectively, or 17.5 (±0.3)‰, and biomass δ¹⁸O_PO4 values of 17.2‰, 16.6‰, and 16.8‰, respectively, or 16.9 (±0.3)‰. This high level of consistency attests to the rigor and reproducibility of our experimental and analytical protocols.

Table S1.

Results of DNA and biomass PO₄ oxygen isotope analyses for 14 1-L batch culture experiments and calculation of the metabolic water component of intracellular water

Microbial strains	T, °C	δ18OH2O, ‰	δ18OPO4, ‰	δ18OH2O,* ‰	1,000 ln α (PO4−H2O)
DNA
P. fluorescens	23	−5.05	17.2	−4.6	21.6
	30	−5.01	16.8	−4.5	21.2
	37	−5.02	14.0	−4.6	18.5
Acinetobacter	23	−5.56	17.9	−4.9	22.7
	30	−5.32	16.1	−4.8	20.7
	37	−5.09	15.5	−4.6	20.0
M. aquaeolei	24	−5.57	17.4	−4.9	22.2
	37	−5.54	13.9	−4.9	18.7
	30	6.30	24.5	3.3	20.9
	37	6.28	22.0	3.3	18.5
	45	6.34	22.8	3.3	19.3
	30	28.75	39.5	18.8	20.2
	37	28.47	38.7	18.6	19.6
	45	28.20	39.3	18.4	20.3
Biomass
P. fluorescens	23	−5.05	17.2	−4.4	21.4
	30	−5.01	16.5	−4.4	20.7
	37	−5.02	15.3	−4.4	19.6
Acinetobacter	23	−5.56	16.6	−4.7	21.2
	30	−5.32	16.1	−4.6	20.6
	37	−5.09	15.9	−4.4	20.2
M. aquaeolei	24	−5.57	16.8	−4.7	21.4
	37	−5.54	15.2	−4.7	20.3
	30	6.30	23.4	2.3	20.8
	37	6.28	22.4	2.3	19.9
	45	6.34	21.1	2.3	18.6
	30	28.75	37.7	15.5	21.6
	37	28.47	35.0	15.4	19.2
	45	28.20	35.3	15.2	19.6

^*The δ¹⁸O_H2O value of intracellular water is recalculated based on the assumption that metabolic water with a δ¹⁸O_H2O value of −3.5‰ accounts for 31% (for DNA) and 41% (for biomass) of intracellular water.

Fig. 2.

The δ¹⁸O_PO4 of PO₄ in (A) DNA and (B) biomass, versus δ¹⁸O_WATER of extracellular water for cultures grown in ¹⁸O-labeled waters at different temperatures. Positive correlation between δ¹⁸O_p and δ¹⁸O_w and the slope indicates that ∼31% of the O in DNA PO₄ and ∼41% of the O in biomass PO₄ is not derived from extracellular water.

Fig. 4.

(A) Fractionation between biomass PO₄ and intracellular water before considering the contribution of metabolic water to intracellular water; (B) recalculated values, based on the assumption that metabolic water with δ¹⁸O_WATER = −3.5‰ accounts for 41% of intracellular water.

Fig. 3.

(A) Fractionation between DNA PO₄ and intracellular water before considering the contribution of metabolic water to intracellular water; (B) recalculated values, based on the assumption that metabolic water with δ¹⁸O_WATER = −3.5‰ accounts for 31% of intracellular water.

In previous studies (27), direct oxygen isotope analysis of intracellular water was achieved by extracting water from whole cells using a vacuum cryodistillation method. In the present study, we measured the oxygen isotopic composition of PO₄ in DNA and total cellular biomass, which should reflect complete exchange of all 4 O atoms in PO₄ with O in water due to the action of intracellular PPase (Fig. 1). Thus, PO₄ δ¹⁸O values serve as a proxy for intracellular water due to the 100% transfer of water O to PO₄, and also due to the consistent relationship between PO₄ and water δ¹⁸O values that has been demonstrated by controlled laboratory experiments on PPase catalyzed PO₄−water oxygen isotope exchange (42, 43).

It is also assumed that any PO₄ initially present in growth medium components is cycled through the intracellular DIP pool and reset to equilibrium by PPase, and does not bypass this route and enter directly into newly synthesized biomass (Fig. 1). In other words, all PO₄ within the intracellular pool, regardless of initial source, is reset to equilibrium with ambient water by intracellular PPase.

Two Lines of Evidence for a Metabolic Water Component in Intracellular Water.

Slope analysis of multilabeled water experiments.

We used MLWIP methods to detect oxygen transfers and probe reaction mechanisms. We first analyzed the δ¹⁸O_PO4 values of biomass and DNA from experiments with multiple labeled-water probes by evaluating the slopes of plots of δ¹⁸O_PO4 vs. the δ¹⁸O value of labeled water used in the experiment (Fig. 2 A and B). As explained above, such plots should give a slope of 1 in the case of wholesale exchange between PO₄ and ambient (i.e., extracellular) water. For this study, plots of δ¹⁸O_PO4 from DNA/biomass vs. δ¹⁸O of extracellular labeled waters (Fig. 2 A and B) yielded slopes of 0.59–0.69 (Fig. 2 A and B). This indicates that ∼30% of the O in DNA PO₄, and ∼40% of the O in total microbial biomass PO₄ is not derived from extracellular water. Note that the additional 10% nonextracellular water O in biomass compared with DNA, both from the same sample, could be an artifact of sample processing, for example, a reflection of more variable P−O bonding structures and sites for P−O hydrolysis in total biomass than in the pure diester structure of DNA. Also, possible contributions from other O species could be present, such as oxygen radicals that may have formed during the ultraviolet radiation (UVR) treatment step used to process the samples, whose oxygen isotopic compositions could not be determined.

This observation of slopes < 1 in δ¹⁸O_PO4 vs. δ¹⁸O_WATER space was the first of two lines of evidence for a metabolic water component in intracellular water based on δ¹⁸O_PO4. We further tested the hypothesis that this 31–41% nonextracellular water component imprinted onto DNA/biomass PO₄ could be metabolic water, by performing an equilibrium analysis.

Equilibrium analysis of MLWIP experimental results.

According to previously established relations based on PPase-catalyzed PO₄−water exchange (Fig. S1) (42, 43) as well as empirical studies of biological phosphates (bioapatites) (e.g., ref. 23), the difference in PO₄ and water oxygen isotopic compositions, δ¹⁸O_PO4 − δ¹⁸O_WATER (or 1,000 ln α_(p-w)) or fractionation between δ¹⁸O_PO4 and δ¹⁸O_WATER, is unchanged at equilibrium and at a given temperature, even when the isotopic composition of either PO₄ or water is changed. Thus, under the traditional assumption that intracellular water is identical in isotopic composition to extracellular water, the δ¹⁸O value of PO₄ in equilibrium with extracellular water having a known δ¹⁸O value can be predicted. In the present experiments, however, these predicted biomass/DNA equilibrium δ¹⁸O_PO4 values and invariant fractionation (1,000 ln α_(p-w)) values were not observed (Figs. 3A and 4A).

Because, according to our hypothesis, intracellular dissolved PO₄−water fractionations are expected to reflect equilibrium PO₄−water exchange catalyzed by intracellular PPase, δ¹⁸O_p and δ¹⁸O_WATER data were evaluated by comparing observed PO₄−water oxygen isotope fractionations (1,000 ln α_(p-w)) to calculated equilibrium fractionations based on our recent calibrations based on PPase-catalyzed equilibrium exchange in the more relevant, dissolved PO₄−water system (43). Again, assuming that intracellular water has the same isotopic composition as extracellular water, as shown in Figs. 3A and 4A, these calculated fractionations do not fit onto a single equilibrium line as expected.

Additionally, there is a clear trend in both DNA (Fig. 3A) and biomass (Fig. 4A) data that shows that the heavier the labeled water used to make growth media, the larger the offset of fractionations from expected equilibrium values. This second line of evidence suggests the influence of another factor in the system, a most likely candidate being some contribution to the intracellular water pool having a different isotopic composition than the extracellular water. Again, we posited that metabolic water was the most likely candidate.

Recalculation of Equilibrium PO₄−Water Fractionations with a Metabolic Water Component.

To test the hypothesis that a contribution from metabolic water to the intracellular water pool was responsible for both the deviation of slopes in δ¹⁸O_PO4 vs. δ¹⁸O_WATER space from 1 (Fig. 2 A and B) and offsets of measured PO₄−water fractionations from expected values, we recalculated these fractionations using the oxygen isotopic composition of metabolic water reported by Kreuzer-Martin et al. (27) for E. coli per below.

Assuming that the intracellular water is a mixture of extracellular water (69% for DNA and 59% for biomass) and metabolic water (31% for DNA and 41% for biomass) and using a δ¹⁸O_WATER value of −3.5‰ for metabolic water (calculated based on direct analysis of intracellular water extracted from E. coli by cryogenic distillation and modeling) (27), we calculated a new oxygen isotopic composition for intracellular water based on these mixtures and then used this “mixed water” value to recalculate the fractionations between DNA/biomass PO₄ and mixed water (Figs. 3B and 4B). Although, as stated above, ∼10% of O in biomass PO₄ could be an artifact of O species formed during sample processing that cannot be determined, we assumed, for simplicity, that this 10% O had the same isotopic composition as metabolic water. Recalculated data plotted in Figs. 3B and 4B, show close agreement with expected equilibrium fractionations based on dissolved inorganic PO₄−water systems (43). We recognize the apparent temperature dependence of PO₄−water fractionations seen in Figs. 3B and 4B, which has been thoroughly explored in studies that have been submitted separately (45).

For the three organisms studied, M. aquaeolei, P. fluorescens, and Acinetobater ADP1 harvested at their late log/early stationary phase, metabolic activities result in an ∼31–41% contribution of metabolic water to the intracellular water pool, which is very close to the percentage metabolic water contribution (30%) determined previously by Kreuzer-Martin et al. (27), for only E. coli cells in stationary phase. Close agreement between previous results (27) and our new data sets based on 14 separate culture experiments using three additional strains of microorganisms from wide-ranging, natural habitats (marine water column, soils) and having different growth conditions/behavior than E. coli—an enteric microbe found in the gut/intestines of mammals—provides strong support for the previous results.

Importantly, the constant δ¹⁸O_WATER value of −3.5‰ for metabolic water reported by Kreuzer-Martin et al. (27) is also supported by our results. Recalculation of DNA and biomass PO₄−water fractionations based on the assumption that the δ¹⁸O value of metabolic water is −3.5‰, the same as for E. coli, is justifiable because central metabolic pathways in most microorganisms are similar, and thus the metabolic water produced should also be similar, given a similar source of O for this water. Although this O source cannot be further resolved based on results of the present study, its constant value points to a source such as atmospheric O₂, which has a constant value of ∼+23.5‰ (46), or metabolic CO₂, which exchanges readily with water, and a conserved central metabolic pathway characterized by constant oxygen isotopic fractionations.

There are three O sources contributing to the metabolic flux of O: intake/drinking water, atmospheric O₂ (with a constant δ¹⁸O value of +23.5‰), and food/nutrient O. Changing the isotopic composition (i.e., food/nutrient/intake water/drinking water with various δ¹⁸O values) and/or relative contributions of any of these sources to the metabolic flux of O could potentially lead to metabolic water having variable oxygen isotopic compositions. For example, increased contribution of atmospheric O₂ by increased aeration of culture vessels could increase cellular metabolic rates compared with poorly aerated (i.e., static) systems. This also suggests possible differences in metabolic water oxygen isotopic composition for aerobes vs. anaerobes. Organisms with different food/nutrient sources, such as those inhabiting surface vs. bottom waters, or coastal vs. pelagic ocean regions, could also have metabolic water with variable oxygen isotopic compositions.

Studies of isotopic compositions of metabolic water in small mammals indicate that the metabolic flux of O atoms is 56% from ingested/drinking water, 30% from atmospheric O₂, and 15% from food (47). If the same holds true for microbes used in our study, then atmospheric O₂ would contribute much more O to metabolic water than would nutrients in the growth medium (e.g., peptone, yeast extract), such that changes in nutrient oxygen isotopic composition may not significantly impact the δ¹⁸O value of metabolic water. The δ¹⁸O value of water used to prepare growth media in our studies was similar to that used by Kreuzer-Martin et al. (27). This, combined with a minimal contribution from O in nutrients and a constant oxygen isotopic composition of atmospheric O₂, may explain the consistency in metabolic water oxygen isotopic composition (−3.5‰) observed between the two studies.

Summary and Conclusions

Results presented here from experiments using different strains of bacteria grown on media prepared with multiple ¹⁸O-labeled waters provide two independent lines of evidence for a significant contribution of metabolic water to the intracellular water pool that subsequently becomes imprinted onto PO₄ in DNA and total biomass. First is the indication from plots of δ¹⁸O_PO4 of PO₄ in DNA/biomass vs. δ¹⁸O of extracellular water that less than 100% of O in PO₄ is derived from extracellular water (i.e., slopes < 1), suggesting a possible metabolic water contribution. Second is the observation that calculated equilibrium PO₄−water fractionations match expected values when it is assumed that the nonextracellular water component indicated by the slopes < 1 has an isotopic composition of −3.5‰, determined previously for metabolic water in E. coli cells (27). The consistency of this metabolic water value across different experiments, water δ¹⁸O compositions, and microbial strains warrants further investigation.

Results presented here strongly support earlier results based on direct measurement of intracellular water in E. coli (27) and have important implications for all other studies where it is assumed that intracellular and extracellular water have identical isotopic composition. For example, if it is found that other aquatic organisms, such as invertebrates (e.g., foraminifera, bivalves, gastropods), which are widely used as paleotemperature proxies in reconstructing Earth’s climate history (22), also have a significant metabolic water component in their intracellular water, then the temperatures calculated from the ¹⁸O/¹⁶O ratio of oxyanions (e.g., phosphate, carbonate) comprising the shells of these organisms may require reevaluation.

It is generally assumed that isotopic equilibrium exists between these biominerals and intracellular/body water, which is further assumed to have the same isotopic composition as extracellular water (19, 21–24). Sensitivity analysis shows that the inclusion of 30%/50%/70% of metabolic water to intracellular water would lead to an underestimation of temperature by ∼3/5/7 °C for a system where metabolic water is −3.5‰, while the extracellular water is −6‰. This error in paleotemperature estimation is proportional to the isotopic difference between metabolic water and ambient extracellular water. In modern marine systems where the ambient extracellular water (seawater) is ∼0‰, if the oxygen isotope value of metabolic water is still assumed to be −3.5‰, then 30%/50%/70% of intracellular water from metabolism will lead to an overestimation of temperature by ∼4/7/10 °C [1 per mil difference in water equals an ∼4.6–4.8 °C difference in temperature for the carbonate system (22) and an ∼4.2 °C difference in the phosphate system (23)].

Our results provide a new probe for intracellular water based on the oxygen isotopic composition of intracellular dissolved PO₄ that is preserved in organophosphate of DNA and cell biomass/soft tissues. The intracellular water proxy presented here based on the δ¹⁸O value of PO₄ in DNA, present in all cells, can be easily applied to a wide range of studies on cell metabolism and function, water balance, and biosynthetic pathways and has potentially widespread future applications in diverse fields of biology, ecology, plant physiology, biochemistry, geobiology, and beyond.

Materials and Methods

Microbial strains were cultured aerobically with manual shaking twice daily and at large-scale (1−2 L) to provide sufficient DNA for further phosphate O isotope analysis. Biomass was harvested when cells reached late log/early stationary phase. Large-scale DNA extraction was based on the method of Ausubel et al. (48), with minor modifications after Blake et al. (45) (Fig. S2). Following extraction and purification of DNA from total biomass, PO₄ moieties in DNA and total biomass were treated using a UV digestion method developed in our laboratory (49, 50) to release free inorganic PO₄. Phosphate concentrations were determined by standard colorimetric (molybdate blue) methods (51). Using methods modified from Kolodny et al. (21) and Tudge (52), PO₄ was purified by precipitation as ammonium phosphomolybdate (APM) followed by reprecipitation as magnesium ammonium phosphate (MAP). Next, MAP crystals were dissolved and the resulting solution was treated with cation resin to remove magnesium and remaining impurities (42, 49, 50, 53). Phosphate was finally precipitated as silver phosphate for O isotope analysis. Oxygen isotope ratios of PO₄ were determined using a Thermo-Chemolysis Elemental Analyzer (TC/EA) coupled to a Delta^plusXP isotope ratio monitoring mass spectrometer operating in continuous flow mode (Thermo-Finnigan). Water δ¹⁸O values were determined by standard CO₂−water equilibration methods using a Thermo-Finnigan Gas Bench. All work (culture experiments, sample processing, and chemical analyses) was carried out at Yale University and all isotope analyses were performed at the Yale Analytical and Stable Isotope Center (YASIC) of the Yale Institute for Biospheric Studies.

Fig. S2.

Gel electrophoresis results of DNA extraction. The first column was loaded with a lambda HindIII ladder as a reference. Columns outlined in red are (from left to right) DNA extractions from Marinobacter aquaeolei grown at 23 °C in media made with three different waters (∼−5‰, 6‰, and 29‰) and Marinobacter aquaeolei grown at 37 °C in media made with three different waters (∼−5‰, 6‰, and 29‰).

SI Materials and Methods

Microbial Culture and Growth.

Bacteria representative of widely different natural environments were grown in 14 separate 1- to 2-L batch cultures, Marinobacter aquaeolei (facultative anaerobe), Pseudomonas fluorescens, and Acinetobacter ADP1 (obligate aerobe), using isotopically labeled waters ranging in δ¹⁸O value from −5.01‰ to + 28.75‰, which allowed verification of fractionations, probing of reaction mechanisms, and also detection and distinction of three different sources of water: ambient extracellular water, intracellular water, and metabolically produced water. Two groups of experiments using different microbial strains were conducted to replicate experiments and test for possible artifacts. Group 1 experiments included three strains with very different growth behavior and growth rates: Marinobacter aquaeolei, Pseudomonas fluorescens, and Acinetobacter ADP1. Group 2 experiments were conducted using Marinobacter aquaeolei only, but with different ¹⁸O-labeled waters ranging from −5.01‰ to +28.75‰. Bacteria were grown aerobically in 1- to 2-L culture flasks with manual shaking twice daily until the late log/early stationary phase was reached. Marinobacter aquaeolei cultures were grown on modified LB broth (40 g/L NaCl, 5 g/L tryptone, 5 g /L peptone, and 5 g/L yeast extract) at pH ∼7.3. Pseudomonas fluorescens and Acinetobacter ADP 1 were grown on Miller LB broth (10 g/L NaCl, 10 g/L tryptone, and 5 g/L yeast extract) at pH ∼7.0.

Microbial Biomass/DNA Collection and Processing.

For the analysis of PO₄ in both DNA and total cellular biomass, cultures were harvested at late log to early stationary phase. Following standard methods for DNA extraction (48), cells were lysed with 10% SDS, and resulting suspensions were further treated with RNase, Proteinase K, CTAB/NaCl (10% hexadecyltrimethyl ammonium bromide/0.7 M NaCl) solution and chloroform/isoamyl alcohol (24:1) to purify DNA. The DNA was precipitated using isopropanol and checked for purity/quality by electrophoresis on a 0.7% agarose gel (Fig. S2). Shearing or other degradation of extracted DNA does not compromise the subsequent PO₄ oxygen isotope analysis.

Extraction of PO₄ in Microbial DNA/Biomass.

Following biomass harvesting and DNA extraction and purification, PO₄ was extracted from DNA and biomass for δ¹⁸O analysis using methods first developed by Liang and Blake (44, 50, 54). DNA or biomass suspended in deionized water was treated with high-intensity ultraviolet radiation (UVR) to release free PO₄ from the organophosphorus compounds—primarily phosphodiesters—making up cell biomass (e.g., nucleic acids, phospholipids). Reaction temperatures were maintained below 40 °C during UVR treatment, a temperature at which no reaction of released inorganic dissolved PO₄ with water occurs based on parallel UVR treatments done on KH₂PO₄ standard solution controls (49, 50). Following UV digestion, the concentration of dissolved inorganic PO₄ was determined for each sample using the colorimetric phosphomolybdate blue method (51).

Purification of PO₄ and Precipitation as Ag₃PO₄.

Dissolved PO₄ was first precipitated as APM, followed by dissolution and reprecipitation of dissolved PO₄ as MAP after methods adapted from Tudge (52) and Kolodny et al. (21) and modified by Liang (49) and Blake et al. (55, 56). Next, MAP was dissolved, and the resulting solution was treated with cation resin to remove magnesium before final precipitation of dissolved PO₄ as silver phosphate (Ag₃PO₄) for oxygen isotope analysis. Silver phosphate samples were thoroughly homogenized and then vacuum-roasted at 550 °C to remove residual water and organic matter. Following this treatment, Ag₃PO₄ crystals were shiny and yellow.

Mass Spectrometric Analysis.

Oxygen in Ag₃PO₄ was converted to CO by reacting with graphite at 1,450 °C in a TC/EA. The CO was then introduced into a continuous flow isotope ratio-monitoring mass spectrometer (DELTA ^PLUSXP) to determine the ratio of ¹⁸O/¹⁶O, expressed as δ¹⁸O values in per mil (VSMOW). The precision of these analyses was ±0.2–0.3‰. Measured δ¹⁸O_PO4 values were calibrated against conventional fluorination values using a series of Ag₃PO₄ standards according to Vennemann et al. (24).

Water was sampled from microbial growth media at both the start and end of experiments and analyzed for δ¹⁸O using the CO₂−water equilibration method (57) and a Themo Finnigan Gas Bench connected to a Finnigan Delta^plusXP stable isotope ratio mass spectrometer. The δ¹⁸O values of water samples were normalized to the oxygen isotope compositions of three laboratory water standards (−1.40‰, −8.96‰, and −3.76‰) calibrated against VSMOW. The overall analytical precision of the water δ¹⁸O analyses, on the basis of replicate analyses of the laboratory water standards under the same analytical conditions as for the water samples, was better than ±0.22‰ (1 SD). All oxygen isotope analyses were performed at the Yale Analytical and Stable Isotope Center (YASIC) of the Yale Institute for Biospheric Studies.