Hydrogen isotopes in individual amino acids reflect differentiated pools of hydrogen from food and water in <i>Escherichia coli</i>

Fogel, Marilyn L.; Griffin, Patrick L.; Newsome, Seth D.

doi:10.1073/pnas.1525703113

Hydrogen isotopes in individual amino acids reflect differentiated pools of hydrogen from food and water in Escherichia coli

Marilyn L. Fogel [email protected], Patrick L. Griffin, and Seth D. NewsomeAuthors Info & Affiliations

Edited by Thure E. Cerling, University of Utah, Salt Lake City, UT, and approved June 17, 2016 (received for review December 30, 2015)

July 21, 2016

113 (32) E4648-E4653

https://doi.org/10.1073/pnas.1525703113

PDF/EPUB

Significance

Hydrogen isotope (δ²H) values of bulk tissues have become valuable tracers for studying migration and movement patterns in animals, but the biochemical mechanism for how hydrogen is incorporated into heterotrophic organisms is not well understood. We grew Escherichia coli as a model organism on two different substrates, and then measured δ²H values of individual amino acids (AAs) in cellular material. The δ²H values of AAs were highly variable within a simple microbial culture. Using isotopic fractionation models, we show that AA δ²H provide tracers of an organism’s environmental (e.g., drinking) water, as well as its food, information of prime interest to ecologists. The work is also of significance to microbial physiologists studying metabolic pathways in microbes from extreme environments.

Abstract

Hydrogen isotope (δ²H) analysis is widely used in animal ecology to study continental-scale movement because δ²H can trace precipitation and climate. To understand the biochemical underpinnings of how hydrogen is incorporated into biomolecules, we measured the δ²H of individual amino acids (AAs) in Escherichia coli cultured in glucose-based or complex tryptone-based media in waters with δ²H values ranging from −55‰ to +1,070‰. The δ²H values of AAs in tryptone spanned a range of ∼250‰. In E. coli grown on glucose, the range of δ²H among AAs was nearly 200‰. The relative distributions of δ²H of AAs were upheld in cultures grown in enriched waters. In E. coli grown on tryptone, the δ²H of nonessential AAs varied linearly with the δ²H of media water, whereas δ²H of essential AAs was nearly identical to δ²H in diet. Model calculations determined that as much as 46% of hydrogen in some nonessential AAs originated from water, whereas no more than 12% of hydrogen in essential AAs originated from water. These findings demonstrate that δ²H can route directly at the molecular level. We conclude that the patterns and distributions in δ²H of AAs are determined through biosynthetic reactions, suggesting that δ²H could become a new biosignature for studying novel microbial pathways. Our results also show that δ²H of AAs in an organism’s tissues provides a dual tracer for food and environmental (e.g., drinking) water.

Animal movement is a defining, not to mention fascinating, characteristic of many diverse vertebrate and invertebrate groups that has been a major topic of research in animal ecology for centuries. For the large majority of animals, which are small in size, it is difficult to characterize movement, seasonal or otherwise. Over the past two decades, ecologists have turned to natural spatial gradients in stable hydrogen isotopes (δ²H) in precipitation and groundwater that intrinsically label animal tissues. Continental-scale models of precipitation δ²H patterns, or isoscapes (1), are invaluable for these efforts; however, significant intrasite variation (10–20‰) in animal tissue δ²H values impedes our ability to resolve movement patterns at subcontinental, regional scales. Approximately 15–30% of the hydrogen in animal tissues is derived from environmental water (2–4), whereas the remainder is sourced from food. Because of ecological and physiological factors, such as the type and variety of food in the diet as well as the geographic source of drinking water, the relationship between tissue δ²H values and δ²H values of local precipitation can become blurred or even nonexistent (5), even for resident species (6).

Studies of captive animals have shown that when fed diets with a constant δ²H composition, the δ²H of animal tissues are depleted in deuterium (²H) relative to water consumed by ∼20–30‰ (e.g., ref. 4). The biochemical mechanism for this isotopic discrimination has yet to be described. This gap in knowledge contrasts with our more comprehensive understanding of the carbon (δ¹³C) and nitrogen (δ¹⁵N) isotope composition of animal tissues relative to their diets, for which a biochemical framework of isotopic discrimination has been described and tested in controlled feeding experiments. Determining the magnitude and variation in diet or water to tissue discrimination factors is essential to refine the use of δ²H to characterize animal movement patterns, and to explore further the potential of δ²H as a tracer of energy within and among ecosystems (7).

Metabolic processes that fractionate hydrogen have been described primarily in microbial and plant tissues. Principal among these studies is the body of literature on δ²H fractionation during lipid synthesis (8–10). Fatty acids, hydrocarbons, and alkenones synthesized by plants, phytoplankton, and microbes are depleted in ²H (i.e., have lower δ²H values) relative to bulk tissue. Such molecules, found in geological settings such as lake sediments, are used extensively as paleoclimate proxies and can have δ²H values that are 50–90‰ lower than environmental water (10, 11).

Because proteins comprise the bulk of animal tissues of interest to ecologists, an understanding of δ²H fractionation in the amino acids (AAs) from which they are synthesized has the potential to offer significant insights into how animals assimilate resources and use them to build tissues. Stable isotope analyses of δ¹³C and δ¹⁵N in AAs have revealed that extensive fractionation occurs both between and among AAs (e.g., ref. 12). These fractionations are related to biochemical pathway in autotrophs and to a combination of biochemical pathway and direct incorporation in heterotrophs. Certain AAs (e.g., glutamate, alanine) can be synthesized by any eukaryotic organism (nonessential AAs), whereas others (e.g., valine and leucine) must originate from an animal’s diet (essential AAs) or gastrointestinal microbiota (13). As an animal incorporates an AA directly from its food, the isotopic composition of that AA influences the isotope value of the animal’s bulk tissue.

Accordingly, we designed a series of experiments with the model bacterial heterotroph Escherichia coli to measure the impact of environmental water and diet on AA δ²H. First, bacteria were grown in water of varying δ²H composition in both glucose and protein-based media. Second, we wanted to quantify the extent to which hydrogen in AAs was directly routed from diet versus synthesized from other nonprotein sources (water and glucose). We incubated E. coli in the two media described above whose waters ranged in δ²H from −55‰ to +1,070‰, which allowed us to calculate the fraction of hydrogen that originated from organic food (i.e., medium) versus media water (Fig. S1).

Fig. S1.

Relationship between the hydrogen isotopic composition of bulk *E. coli* cells and the isotopic composition of the water in the growth medium. Cells were grown on either defined glucose (○) or tryptone-based (●) media.

Results

δ²H Values of AAs in Tryptone.

To test methods for analyzing δ²H in AAs, we hydrolyzed the tryptone protein that was used as the organic source for growing E. coli by three different methods: (i) hydrolysis in 6N hydrochloric acid (HCl) for 20 h at 110 °C, (ii) vapor hydrolysis with 12N HCl for 20 h at 110 °C, and (iii) hydrolysis with ²H-spiked 6N HCl (+1,070‰ water plus 12N HCl) for 20 h at 110 °C. Following hydrolysis, AA mixtures were dried under N₂ at 110 °C and then derivatized. The derivatization steps remove exchangeable hydrogen atoms from the carboxyl and amine side groups. One hydrogen atom on the amine group is retained and likely exchanges during hydrolysis and drying. Because the derivatization reactions are carried out without any liquid water present, it is unlikely that this hydrogen atom exchanged during the derivatization process. The δ²H of AAs that we report represent what we refer to as “intrinsic” hydrogen: nonexchangeable hydrogen bonded to carbon in the AA and one hydrogen atom bonded to nitrogen that remains following derivatization.

The δ²H of individual AAs in tryptone varied by >250‰, and among the three treatments, the isotopic compositions of derivatized AAs showed the same pattern for the majority of the AAs (Table S1). Exceptions for the ²H-spiked hydrolysis treatment were aspartic and glutamic acids, in which δ²H was more positive by >100‰ relative to the other treatments. Given the nature of the derivatized AAs, hydrogen atoms at the alpha position of both carboxyl groups must have been susceptible to exchange. In the treatment hydrolyzed by vapor phase, serine was altered the most and had higher δ²H values than its counterpart in the control treatment by ∼55‰. Serine and glycine, two of the structurally simplest AAs, should be the AAs with the greatest likelihood for hydrogen exchange, because they contain only two (glycine) or three (serine) nonexchangeable hydrogen atoms. Following these results, we prepared 6N HCl by mixture of laboratory distilled water and 12N HCl and used it to hydrolyze all of the samples analyzed in our study for 20 h at 110 °C.

Table S1.

The δ²H values for AA from tryptone protein hydrolyzed by three different methods to test for isotope fractionation

AA	Regular treatment	Vapor phase	Heavy water
Ala	−94	−96	−130
Gly	−34	−21	−36
Ser	−125	−77	−133
Pro	−77	−71	−75
Asp	−209	−207	64
Glu	−178	−177	−47
Thr	−307	−276	−329
Val	−242	−229	−253
Leu	−204	−203	−207
Ileu	−294	−292	−304
Phe	−159	−163	−181

Regular treatment, 6N HCl with lab DI; Vapor phase, 6N HCl with laboratory DI not in contact with the protein; Heavy water, 6N HCl with ²H-labeled water.

The more remarkable information from these tests was the range in δ²H of the AAs from mammalian tryptone proteins regardless of hydrolysis method. Nonessential AAs had more positive δ²H values (−21 to −209‰), whereas essential AAs had more negative δ²H values (−160 to −307‰). A similar pattern in isotopic fractionation has been observed in δ¹³C values of AAs from a variety of organisms (12), and suggests that the carbon skeleton of essential AAs must have originated directly from the protein carbon in the animal’s diet. Whether or not the hydrogen in these molecules is associated directly with the carbon via direct routing from the protein fraction of diet is unknown at this time.

E. coli Grown in Glucose and Inorganic Nutrient Media.

Hydrogen sources available to E. coli included the following: glucose (−4‰), NH₄Cl (−124‰), and water (−55‰ to +1,070‰) (Fig. 1). In these experiments, E. coli needed to synthesize the full suite of AAs through biosynthetic pathways, even those AA that are considered essential for higher organisms. The range in δ²H for intrinsic hydrogen in individual AAs in cells grown in water (−55‰) on the glucose medium, which was required to synthesize all of their AAs, was almost >200‰ (Fig. 1A and Table S2). Cells were cultured in three independent experiments and then analyzed in triplicate. Proline was the most enriched AA (δ²H = −54 ± 40‰), whereas the most depleted AAs were glycine (δ²H = −268 ± 34‰) and isoleucine (δ²H = −220 ± 36‰). The range in δ²H is similar to the range measured in the tryptone digest, and is another example of the extensive range in δ²H of individual AAs from two completely different organisms, which spans almost the entire range of values found in surficial waters or organic bulk material in terrestrial and marine environments (14).

Fig. 1.

(A) δ²H of individual AAs from glucose-grown *E. coli* cells. (B) δ²H of individual AAs from tryptone-grown cells. The AAs classified as nonessential for eukaryotes are plotted on the left side of the graph, and AAs considered to be essential are plotted on the right side. Parallel lines indicate similarities in biosynthetic δ²H fractionation.

Table S2.

δ²H composition of AAs from E. coli grown in water with differing δ²H values with different organic substrates

AA	Glucose media δ²H, ‰				Tryptone media δ²H, ‰				Tryptone δ²H, ‰
δ²H H₂O	−55	225	505	1,070	−55	225	505	1,070
Ala	−141	−44	57	282	−98	6	93	305	−94
Gly	−268	−181	−117	48	−122	−127	−58	18	−34
Ser	−131	−59	−29	143	−128	−157	−58	10	−125
Pro	−54	88	225	578	126	−14	11	182	−77
Asp	−181	−115	20	200	−120	−124	−134	−55	−209
Glu	−246	−118	21	233	−131	−109	−102	10	−178
Thr	−114	33	87	304	−150	−207	−119	−88	−307
Val	−202	−109	−27	215	−260	−215	−172	−175	−242
Leu	−170	−62	71	318	−252	−189	−142	−151	−204
Ileu	−220	−103	37	279	−368	−296	−248	−283	−294
Phe	−173	8	−34	133	−216	−207	−77	−72	−159

The relative distribution of δ²H in the various AAs was largely upheld in the cultures grown in ²H-enriched media waters (Fig. 1A). Glycine had the lowest δ²H values, and proline had the most positive δ²H values. In all of the experiments, the intrinsic hydrogen δ²H values were always more negative than the δ²H value of the water in the culture medium. The contribution of glucose-derived hydrogen atoms to E. coli AAs relative to the contribution of hydrogen from water can be estimated using the slope of the relationship between the δ²H of AA hydrogen and the δ²H of water. Because we do not know the isotope fractionations associated with hydrogen incorporation from glucose or from water exactly, a model was constructed using fractionation factors for water (α_W) and food (α_F) estimated from the literature (9, 15) (Table 1). For example, the model shows that 47–64% of the hydrogen in glutamic acid originates from glucose, whereas the remainder was incorporated from water. For proline, the model shows that 16–38% originated from glucose, with the remainder derived from water (Figs. 2 A and B, 3, and 4). The mean proportion (±SD) of nonessential AA hydrogen originating from glucose varied from 27 ± 11% to 64 ± 5%, which was similar to the amount of essential AA hydrogen originating from glucose (42 ± 8% to 53 ± 7%).

Table 1.

Calculated percentages of hydrogen coming from H₂O based on model calculations using isotope fractionation (α) estimates with upper and lower potential α values

AA	Glucose, %H_H2O	Tryptone, %H_H2O
Ala	49.1 ± 7.2	46.1 ± 6.8
Gly	36.1 ± 5.3	19.8 ± 2.9
Ser	50.8 ± 7.4	17.3 ± 2.5
Pro	45.4 ± 6.6	11.6 ± 1.7
Asp	55.4 ± 8.1	18.6 ± 2.7
Glu	73.3 ± 10.7	22.2 ± 3.3
Thr	71.6 ± 10.5	12.4 ± 1.8
Val	57.0 ± 8.3	9.0 ± 1.3
Leu	48.3 ± 7.1	8.3 ± 1.2
Ileu	58.2 ± 8.5	3.8 ± 0.6

Fig. 2.

(A) Modeled contribution of hydrogen from water versus organic hydrogen source based on δ²H of individual AAs from *E. coli* grown on glucose. (B) Expanded view. Isotope fractionations for water (α_W) and food (α_F) are estimates following the method of Session and Hayes (15), using the linear regression slope associated with water- and glucose-labeling experiments.

Fig. 3.

Calculated %H (f_w) originating from water in microbes grown on either glucose or tryptone. Dashed lines represent the values of f_w that correspond to that point on the curve at α_F or α_W. Estimates of α_W and α_F follow the method of Session and Hayes (15), using the linear regression slope associated with water- and tryptone-labeling experiments.

Fig. 4.

Calculated proportion of hydrogen originating from water in microbes grown on either glucose or tryptone based on models presented in Figs. 2 and 3.

Pairs of AAs that are related biosynthetically provide additional information on hydrogen metabolism. For example, alanine and aspartate are closely related through steps that connect the glycolytic pathway with the tricarboxylic acid (TCA) cycle (Fig. 5). Their δ²H values are nearly identical in all treatments, suggesting a common hydrogen pool available for synthesis. Conversely, nonessential proline, which is synthesized from glutamate, is always more enriched than its parent glutamate by 195–355‰. Likewise, aspartate is the first AA on the pathway to isoleucine synthesis; there is a sizeable (yet variable) isotopic fractionation between these two AAs.

Fig. 5.

Conceptual dendrogram of AA relationships. AAs in italics are considered to be essential AAs in eukaryotes. The length of the horizontal lines indicates the relative complexity of steps in the AA biosynthetic pathway.

E. coli Grown with Tryptone as Sole Organic Hydrogen Source.

Hydrogen sources available to the E. coli in these experiments included tryptone (−65‰) and water (−55‰ to +1,070‰). The δ²H of intrinsic hydrogen from individual AAs from these cells ranged from +126‰ (proline) to −368 ‰ (isoleucine) in cultures grown on laboratory distilled water (n = 4 separate cultures; Fig. 1B and Table S2). These values can be compared with those AAs in the proteinaceous “food” source tryptone. Proline in the tryptone digest had a δ²H value of −77‰ compared with a δ²H value of +126‰ in E. coli, demonstrating net positive isotope fractionation between proline in the diet and microbial biosynthesis by E. coli, including the incorporation of hydrogen from water and other organic intermediates during metabolic processing. Four of the AAs have δ²H values more positive than their counterparts in tryptone, with the remainder more negative than their counterparts in tryptone. At the other end of the spectrum, low but similar δ²H values in isoleucine (δ²H = −294‰) in tryptone and in E. coli (δ²H = −346‰) indicate isotopic routing of hydrogen with little influence of fractionation during hydrogen exchange or tissue biosynthesis.

Using a similar model as explained above, our calculations show that isoleucine has the lowest percentage of 9.4 ± 1.4% of hydrogen incorporation from media water (Table 1), which makes sense because this AA is one of the most complex AAs and the majority of its hydrogen is in the intrinsic form (Figs. 3 and 4 and Table 1). Other AAs that show low hydrogen incorporation from media water include leucine (12.1 ± 1.8%) and valine (7.9 ± 1.2%). Like isoleucine, these two AAs are also branched chain AAs whose de novo biosynthesis requires catabolism of other AAs, and whose direct incorporation from media would be energetically favorable.

Alanine, unlike isoleucine, has a very simple structure and can be synthesized from pyruvate during glycolysis. Alanine has the highest proportion (46 ± 7%) of hydrogen derived from media water. Nonessential AAs generally had greater proportions of hydrogen originating from water (20 ± 14%), whereas the more complicated essential AAs contained significantly less hydrogen from water (8.7 ± 3.4%). Finally, the influence and proportion of hydrogen atoms from water in these experiments were considerably less than when E. coli was cultured on simple glucose and was required to synthesize all of its AAs. This finding demonstrates at the molecular level that hydrogen atoms can be routed directly from the specific molecules in the diet, and thus move unaltered up food chains into higher organisms.

Discussion

Our δ²H analyses of AAs from heterotrophic microbes hold information related to biosynthetic pathways and nutritional status. Our experiments with E. coli demonstrate that a majority of the hydrogen intrinsic to AAs (carbon-bonded or N-linked) is derived from the hydrogen in organic dietary substances. Not only have we shown dietary hydrogen incorporation at the cellular level but we have also demonstrated that at the molecular level, hydrogen in certain AAs is incorporated directly into protein biomass. This observation provides an explanation as to why and how much organic hydrogen derived from diet directly influences animal tissues commonly analyzed by ecologists to characterize animal movement and trace the flow of energy within and among ecosystems (1, 7, 16).

Glucose Metabolism and AA Synthesis.

Some AAs, especially nonessential ones like alanine, serine, and glycine, show strong evidence for large contributions of hydrogen from water during de novo synthesis with glucose as the sole carbon source. Glucose is transported into E. coli by the phosphotransferase system, in which carbon-bonded hydrogen and hydroxyl hydrogen atoms should enter relatively intact (17). Water molecules are brought into E. coli via aquaporins, which are enzymes that transport molecular water into and out of cells (18). In the strain of E. coli that we used, both transport mechanisms were active.

During the first reactions of glycolysis, in which glucose is phosphorylated and converted to fructose, the exiting hydrogen atom is a hydroxyl hydrogen. The conversion of glucose to fructose goes to completion, so no measurable isotopic fractionation should occur. It is not until the glyceraldehyde-3 phosphate dehydrogenase step that hydrogen atoms are removed from glucose to form NADH + H⁺. Note that the hydrogen coming from glucose is not lost or exchanged, but transferred to one of the primary activated carrier molecules for general cellular biosynthesis.

The second major alteration of hydrogen atoms during glycolysis occurs when a H₂O molecule is removed during the phosphoglycerate enolase step to form phosphoenolpyruvic acid. During the final step of glycolysis, pyruvate kinase adds one hydrogen atom to a nonexchangeable position on the methyl carbon of pyruvate. During transamination, a carbon-bonded hydrogen atom is added to the methylene carbon, in addition to the NH₃⁺ group. Our method for AA derivatization preserves one hydrogen from the amine group, which is theoretically exchangeable, and all other C-bonded hydrogen. Of the six hydrogen atoms that we analyze, only two (33%) are the original hydrogen atoms from glucose: Two come from transamination, and the remaining four come from water. Our results for both the glucose and tryptone treatments are roughly consistent with this model: Alanine, which should be the most direct AA coming from glycolysis, has 41–56% of its hydrogen originating from water in glucose cultures and 39–53% in tryptone cultures.

Glycine and serine are synthesized from 3-phosphoglycerate. Two hydrogen atoms (50%) are added during transamination from glycerate to serine, which has a total of four intrinsic hydrogen atoms, two of which should originate from glucose. Glycine, if synthesized from serine, should have only one hydrogen atom from the original glucose out of a total of three. In our glucose treatment, we measured a 43–58% hydrogen contribution from water in serine and a 31–41% contribution in glycine.

Aspartate, which is synthesized by transamination of oxaloacetate, has two carbon-bonded hydrogen atoms that should originate from water catalyzed by fumarase during repeated TCA cycling. The remaining two hydrogen atoms enter the molecule during transamination. In our glucose treatment, 39–52% of hydrogen in aspartate originated from water, similar to the 50% predicted. Glutamate is composed of two hydrogen atoms from acetyl-CoA (66% from glucose and 33% from H₂O), two from the TCA cycle (H₂O), and two from transamination. For glutamate from E. coli grown on glucose, 47–64% of the hydrogen originates from water, which also overlaps with theoretical predictions.

Isotope fractionations of hydrogen include many different components, none of which are completely constrained: synthesis reactions, routing, and exchange reactions that may or may not be in equilibrium. An example is the hydrogen atoms added during transamination, which should technically be hydrogen atoms derived from water but, instead, may be influenced by a different hydrogen pool. Hydrogen atoms in NH₄⁺/NH₃ should be in isotopic equilibrium with media water. Once in the cytosol of E. coli, however, the δ²H of these N-H atoms might be reequilibrated with metabolic water derived from diet (i.e., glucose).

Another source of hydrogen from glucose metabolism is the formation of NADH and NADPH from NAD⁺ and NADP⁺ with the added hydrogen atoms coming directly from glucose. Glucose-6-phosphate dehydrogenase, for example, is one of the primary reactions forming NADPH in bacteria (19). The hydrogen transferred to NADP⁺ comes from glucose. By the time glucose enters the TCA cycle as acetyl-CoA, only three of the original 12 hydrogen atoms remain. In the TCA cycle, an additional eight hydrogen atoms are cycled into NADH, NADPH, and FADH₂, such that by the time one full TCA cycle has been completed, none of the original hydrogen atoms are attached to any of the TCA intermediates. The hydrogen atoms in NADPH carry some of the original hydrogen from glucose and are key donors of hydrogen in many biosynthetic reactions.

Enzymatic Hydrogen Fractionation and Tunneling.

δ²H effects in enzyme reactions have traditionally been determined using ²H-substituted reactants and measuring rate constants. Many enzymatic reactions, however, manifest quantum effects in biological electron transfer. For example, l-α-AA transferase reactions occur by the ping pong bi bi mechanism (20) in which two substrates (e.g., aspartate, glutamate) are transformed into two different reactants (bi bi; e.g., oxaloacetate, α-ketoglutarate) and the substrate–enzyme complex and the activated enzyme complex take place in two distinct steps (ping pong). These enzymes are highly conserved in eubacteria, yeast, birds, and mammals, and they play central roles in catalysis and biosynthesis. Biochemists have learned about these enzymes by creating mutants, particularly with AA substitutions at the active site. Based on what is known (21), the hydrogen in the α-carbon position has a 50:50 chance of coming from intracellular water or pyridoxamine phosphate. The pyridoxamine phosphate hydrogen transferred to a newly formed AA has a 50:50 chance of originating from the AA reactant or preformed pyridoxal phosphate. We predict then that because AAs are metabolized via aminotransferase reactions, the hydrogen in them should ultimately originate from organically bonded hydrogen.

Some biological reactions exhibit nonclassical isotope effects for hydrogen, implying that hydrogen atoms can tunnel through energy barriers, particularly if the bond distance between the reactants is close enough (<2.8 Å) to permit hydrogen to overcome tunneling barriers (22). These experiments have been conducted with deuterated substrates in vitro in a growing, yet limited, number of enzymes. Tunneling at the natural abundance level has not been readily measured, although Zhang et al. (23) found possible evidence of hydrogen tunneling in lipids produced by microbial cultures grown on acetate or succinate. Reactions that transfer hydrogen to NADP⁺ in the TCA cycle (e.g., succinate dehydrogenase) could produce NADPH that is substantially ²H-enriched. The enzyme proline dehydrogenase, which catalyzes the interconversion from proline to D1-pyrroline-5-carboxylate, has a bonding distance at the active site of 2.7 Å (24), which is close enough to allow hydrogen tunneling.

New Isotopic Biosignature?

Zhang et al. (23) presented δ²Hs of lipids from four different microbes using autotrophic or alternative pathways of biosynthesis and found that the δ²H values of fatty acids vary by as much as 300‰ depending on the biosynthetic pathway and heterotrophic growth substrate, even if the microbes were cultured in media water with a constant δ²H value. Chemoautotrophic microbes can derive energy from the oxidation of H₂ to protons and electrons; for example, deep-sea microbes have unique sets of hydrogenase enzymes (25). We predict that the δ²H of AAs from organisms with different hydrogenase enzymes will be useful in elucidating which hydrogenase activities are supporting microbial growth in extreme environments.

Hydrogen cycling is inherently more complex than either carbon or nitrogen because hydrogen can enter a molecule from many different points along a metabolic pathway (26) and many organisms have two distinct sources of hydrogen (water and food) available to them, whereas carbon and nitrogen originate from a single source (food). In addition, a minor portion (∼10–15%) of hydrogen atoms in proteins and AAs are known to be exchangeable (27), yet available data show that not all δ²H exchange rapidly in biological conditions (8, 28). A deeper understanding of the cycling of water inside cell membranes (29), including mitochondria, and how this water is connected to external sources of water is needed. Compound-specific δ²H analysis of AA provides a tool to answer these questions, because these molecules are all synthesized in central metabolic processes.

Isotopic Routing and Implications for Animal Ecology.

Even though the E. coli strain we used is theoretically able to synthesize all of its required AAs (30), our data show a significant amount of isotopic routing or direct incorporation of δ²H from dietary protein (tryptone) into tissue without significant isotopic fractionation or exchange. With the exception of alanine, in which ∼39–52% of the intrinsic hydrogen is derived from water, those AAs considered nonessential in eukaryotes had only 6–22% of their hydrogen from media water, with the remainder being derived from that particular AA in the tryptone. Moreover, AAs considered to be essential in eukaryotes had only 5–12% of their intrinsic hydrogen from media water. Although δ¹³C values of essential AAs in animal tissues are nearly identical to those specific AAs in their diets (e.g., ref. 31), our AA δ²H results document, for the first time to our knowledge, that isotopic routing occurs with hydrogen as well as carbon.

We propose that the direct transfer of many of the essential AA hydrogen atoms could be used as direct tracers, not only of precipitation but also of dietary sources for animals. Thus, from one tissue sample, the isotopic composition of the drinking water can be determined by knowing the relationship between a nonessential AA (e.g., alanine) and local surficial waters (e.g., precipitation). Alternatively, the δ²H of the organic hydrogen sourced from food could be traced by understanding hydrogen fractionation patterns in essential AAs (e.g., isoleucine). Our compound-specific approach provides a powerful tool to characterize both diet and water use via analysis of individual AAs in a single tissue. Along with δ¹³C used to trace the relative inputs of primary producers with different biosynthetic pathways and δ¹⁵N commonly used to assess trophic level, δ²H of AAs provide a measurement with which to characterize location or habitat (nonessential AA δ²H) and diet source and quality (essential AA δ²H) in an individual animal.

The application of δ²H analysis in many animal movement and migration studies relies on the premise that the δ²H of precipitation is the main determinant of variation in the δ²H value of animal tissues. Presumably, precipitation δ²H is directly transferred to primary producers and then transferred up the food chain to consumers (e.g., refs. 1, 32). The assumption of direct transfer implies that there is a linear relationship between the δ²H of animal tissues and precipitation with a slope of 1, and that any isotopic offsets related to trophic discrimination are known. In reality, however, the relationship between the δ²H of animal tissues and the δ²H of precipitation varies considerably (16). This variation has been attributed to a variety of factors, including differences in laboratory protocols; variation in the δ²H of precipitation over time and space; and inherent ecological factors, such as general dietary preference (e.g., herbivore versus carnivore). Our data show that organic hydrogen from diet can route directly into tissues, supporting earlier work showing that the majority of hydrogen in animal tissues is derived from the hydrogen in organic compounds in food, with a minor component derived from preformed water, including environmental (e.g., drinking) water and water in food (3–5). δ²H analysis of individual AAs has the potential to become a more direct way of evaluating the effects of diet versus precipitation for ecologists characterizing animal movement and resource use patterns.

Materials and Methods

E. coli (MG1655) was grown in defined glucose medium with 3-(N-morpholino)propanesulfonic acid as a buffer and with NH₄Cl as the nitrogen source or in the complex tryptone salt broth, in which dietary hydrogen is derived from the pancreatic casein digest tryptone (Bacto Tryptone; Becton Dickson). The most deuterated water we used (δ²H: +1,070 ‰) was prepared by mixing 98% ²H₂O with distilled water (δ²H: −55 ‰). δ²H values of other media water treatments (δ²H: +225 ‰, +505 ‰, and +1,070‰) were achieved by mixing appropriate portions of these two waters (SI Materials and Methods).

Tissues (1–3 mg) were hydrolyzed in 6N HCl at 110 °C for 20 h. Tests with tryptone showed that, with the exception of glutamic and aspartic acid, hydrogen did not exchange during hydrolysis or derivatization. AAs were subsequently derivatized with 2-isopropanol and N-TFA (31), and then analyzed in triplicate for δ²H after separation on a 50-m DB-5 column (SGE Analytical Science) in a Thermo-Fisher Trace Gas Chromatograph. Separated AAs were thermally decomposed to H₂ in a ceramic reactor set at 1,400 °C. The δ²H values of AA were calculated from measured δ²H values on three to five separate analyses by mass balance, with adjustments being made for hydrogen removed during derivatization. Measured δ²H values include contributions from hydrogen in the isopropanol. Mass balance calculations were based on the number of hydrogen atoms in the derivatized molecule, which includes methyl-hydrogen, aliphatic-H, one nitrogen-bound H, and hydrogen from isopropanol. Extensive tests were performed with solid AA powders to determine whether or not hydrogen atoms were exchangeable with liquid water vapor at room temperature or steam at 100 °C. We used the δ²H of the native AA standards to calculate the δ²H values and the isotopic fractionation in making the derivative, similar to the method used for determining δ¹³C values of AAs (12, 13). Because a subset of the hydrogen atoms is removed during derivatization, we needed to know whether the remaining hydrogen atoms had similar δ²H values. At room temperature, about 10% of the hydrogen was exchangeable. At 100 °C, 33% of the total hydrogen was exchangeable; however, using ambient laboratory distilled, deionized water increased the δ²H value by an average of only 5‰. Changes in δ²H of serine (29‰) and proline (18‰) were significantly greater. Interpretation of exchange experiments is complicated by many factors, most of which are not relevant for the calculations we made in this study. Therefore, we used the δ²H values of the AAs measured directly from the bottles that were used to make up the standard mixtures. An analytical error of ±5‰ has little influence on the range of δ²H values we report here.

The proportion of hydrogen originating from water or diet (glucose or tryptone) for individual AAs can be determined by plotting the δ²H of medium water versus either total cellular (bulk) δ²H or specific AA δ²H. The slope of this line is roughly, but not exactly, equivalent to the proportion of hydrogen derived from water, with the remainder assumed to originate from organic hydrogen in the medium diet. This simple rendition ignores the fact that there are two fractionations involved for both water and dietary uptake, neither one of which is accurately known. We constructed a model using estimates for isotopic fractionation between media water and fatty acids for photoautotrophs (15).

SI Materials and Methods

E. coli Cultures.

E. coli (MG1655) was grown in defined glucose medium with 3-(N-morpholino)propanesulfonic acid (MOPS) as a buffer and with NH₄Cl as the nitrogen and phosphorous sources. Slightly deuterated water (δ²H = 1,070‰) was prepared by mixing 98% D₂O with laboratory distilled-deionized water (δ²H = −55 ‰). Desired media water isotopic compositions were achieved by mixing appropriate portions of these two waters and verified by thermo-chemolysis elemental analysis (TCEA) (discussed below).

Single-colony cultures of E. coli K-12 MG1655 (no. 700926; American Type Culture Collection) (33) were aerobically incubated overnight to stationary phase at 37 °C with shaking (220 rpm). Biomass was harvested via repeated rounds of centrifugation and rinsing, and then lyophilized before processing. Two bacterial culture media were used in these experiments: the chemically defined MOPS medium for enterobacteria (TekNova) (34), which contains dietary hydrogen as glucose and NH₄Cl, and the complex tryptone salt broth (35), in which dietary hydrogen is derived from the pancreatic casein digest tryptone (Bacto Tryptone; BD Biosciences). This medium was assembled from a known stock of tryptone, which was analyzed for its bulk and individual AA δD composition (discussed below).

Hydrolysis Tests.

To analyze the compound-specific δ²H in individual AAs from cellular protein, all samples needed to be hydrolyzed with 6N HCl at 110 °C to disrupt peptide bonds. During this treatment, it was possible that hydrogen could have been exchanged with the H⁺ either in the acid or from the water used to dilute the acid to the appropriate normality. In an intact protein, it is known that hydrogen exchanges very little (10); however, in its constituent AAs, exchange is highly likely. Hydrogen atoms are readily exchangeable when they are in alcohols, acids, or amine groups, which are found in all AAs. Conversely, hydrogen bonded directly to carbon in aliphatic bonds is relatively nonexchangeable under most circumstances (8, 36).

Tryptone (1–2 mg) was hydrolyzed in three different treatments: in 1 mL of 6N HCl with laboratory distilled water (δ²H = −55‰) being used to dilute concentrated HCl; in 1 mL of 6N HCl with ²H-enriched water (δ²H = +1,000‰) being used to dilute concentrated HCl; and by vapor phase hydrolysis with 1 mL of 6N HCl with laboratory-distilled water, which was not in direct contact with the tryptone. All treatments were carried out in an oven at 110 °C for ∼20 h. Following hydrolysis, all treatments were processed similarly by driving off the HCl and water through evaporation under N₂. AAs were subsequently derivatized with 2-isopropanol and N-TFA (31). Derivatized AAs were analyzed in triplicate as described below (Table S1).

δ²H Analysis.

Bulk E. coli cells were analyzed in duplicate with a TCEA system linked to a Conflo III interface and to a Thermo-Finnigan Delta XL Plus isotope ratio mass spectrometer. Standards for calibrating δ²H in bulk tissues included isotopically known gas standards and a range of organic molecules that were purchased (Isoanalytical), developed internally as standards (stearic acid and pump oil), or obtained from colleagues (chicken and turkey feathers; gifts from Chris Romanek, University of Georgia and University of Kentucky). A suite of individual AAs was prepared as isotopic standards, and its set of δD was determined singly via the TCEA. The isotopic compositions of 2-isopropanol (δ²H = −110‰) and the waters from the media were determined by injecting 1 μL into the TCEA directly. Liquid samples were calibrated against standard mean ocean water (SMOW) and Greenland ice sheet precipitation (GISP), known internationally accepted water standards, analyzed similarly.

Compound-specific AAs were analyzed for δ²H after separation on a 50-m DB-5 column (0.5-μm film thickness; Hewlett Packard) in a Thermo-Fisher Trace Gas Chromatograph (injector temperature of 220 °C). Separated AAs were thermally decomposed to H₂ in a ceramic reactor set at 1,400 °C. The reactor was conditioned every other day by injecting 1 μL of hexane. The analytical system was calibrated and checked daily with a set of three normal hydrocarbons (C14, C15, and C17) that were initially measured via TCEA and then through the GC system. SDs for these compounds were ±6.5‰ over a 3-mo time period. AA standard compounds were derivatized and measured via the GC system. These values were used to determine the reproducibility of the measurements, including potential interferences from H₂ in the ceramic reactor. These AA standards were analyzed each day before the analysis of sample AAs and also between samples. Average SDs for AA standards were ±9‰, with some AAs having less variation than others: Serine and threonine, for example, had higher SDs than alanine, leucine, proline, or glutamate.

Calculations.

The δ²H compositions of AAs were calculated from measured δ²H values on three to five separate analyses by mass balance, with adjustments being made for hydrogen lost during derivatization. Measured δ²H values include contributions from hydrogen in the isopropanol, but not in the TFA moiety, which does not contain hydrogen. We also determined the isotopic composition of the real value of the isopropanol, based on measuring known singular AAs (n = 9 measurements). Similar to the individual corrections that are used for calculating δ¹³C fractionations during derivatization (31), we found that the calculated value for the isopropanol correction (−110‰) was significantly different from the measured value for certain AAs; therefore, we used slightly different values for the δ²H values of isopropanol corrections: glycine (−120‰), valine (−88‰), proline (−96‰), aspartate (−91‰), glutamate (−78‰), and phenylalanine (−100‰). Therefore, each AA’s value was calculated using a specific value for the isopropanol correction.

Mass balance calculations were based on the number of hydrogen atoms in the derivatized molecule, which includes methyl-H, aliphatic-H, nitrogen-bound H, and hydrogen from isopropanol. The methyl-H and aliphatic-H should not exchange under the mild hydrolysis conditions, but there is a chance that hydrogen bound to nitrogen in amine groups would. A certain percentage of hydrogen in bulk proteins [e.g., feather (10%), collagen (20%)] is considered exchangeable when the AAs are in peptide linkages (8, 36). These hydrogen atoms are probably hydroxyl-H, carboxyl-H, and amine-H that are removed during derivatization reactions needed to synthesize GC-amenable products. We use the term “intrinsic hydrogen” for the hydrogen fraction that we measure in our experiments.

E. coli Grown in Glucose and Inorganic Nutrient Media.

Hydrogen sources available to E. coli included the following: glucose (−4‰), NH₄Cl₂ (−124‰), and water (variable, −55‰ to +1,070‰). Half of the hydrogen atoms in glucose should readily exchange with the glucose atoms of water at acidic pH; however, this medium was buffered at pH 8.0 with MOPS. One could assume that the hydrogen atoms in the nitrogen and phosphorous nutrient could also exchange with media water. Thus, the prediction should be that the bulk of the hydrogen in both the total cellular biomass and individual AA should originate from water. In fact, only ∼38% (±0.9) of the hydrogen in bulk cellular biomass originates from H₂O, with a net isotopic discrimination of −33‰ (Fig. S1).

E. coli Grown with Tryptone as Sole Organic Hydrogen Source.

Hydrogen sources available to the E. coli in these experiments included tryptone (−65‰) and water (variable, −55‰ to +1,070‰). Tryptone is a complex protein, and a percentage of the hydrogen atoms could possibly exchange with the media water. We checked for exchange by incubating sterile tryptone in the ²H-enriched water for several days. No change in the δ²H of tryptone was measured within analytical error. Cells grown under these conditions incorporated only 27% (±1.5) of their hydrogen from water, with a net isotopic discrimination of −50‰ (Fig. S1).

Calculation of δ²H Fractionation in AA Biosynthesis

To estimate the portion of AA hydrogen derived from organic food sources versus media water, we have modeled the biosynthesis of individual AAs as two-reagent, single-product reactions (15). This system is isotopically described as

R_{P} = R_{A} α_{A / P} f_{A} + R_{B} α_{B / P} f_{B},

[S1]

where R_X is the isotope ratio (²H/¹H) of compound X, α_X/P is the fractionation factor between reagent X and product P, and f_X is the portion of total hydrogen in P originating from reagent X, such that f_A + f_B = 1; that is, all hydrogen comes from one of the two reagents.

For the microbial culturing experiments described here using glucose as the growth substrate, hydrogen originates either from water (W) or glucose (G) and results in the production of an AA. These reactions are described by

R_{AA} = R_{W} α_{W / AA} f_{W} + R_{G} α_{G / AA} f_{G},

[S2]

or, by substituting f_G = (1 – f_w),

R_{AA} = R_{W} α_{W / AA} f_{W} + R_{G} α_{G / AA} (1 - f_{w}) .

[S3]

The reactant and product isotope ratios (R_AA, R_W, and R_G) can be measured as directed. However, three unknowns cannot be quantified: α_W/AA, α_G/AA, and f_W. Because these quantities appear as products in Eq. S3, they cannot be solved analytically. However, as prior publications have noted and used, one can constrain the values for these parameters drawing on the published literature as well as regression data gleaned from traditional substrate labeling experiments.

Estimation of f_W for Individual AAs

Two pieces of information are needed to constrain possible values of f_W. The first is the slope of the regression line describing the variations in δ_AA to changes in δ_W for each individual AA. This slope (m) is the quantity m_W = α_W/AAf_W.

In weakly fractionating systems (where α ≈ 1), the slope alone can serve as an adequate estimate of f_W. However, δ²H can be strongly fractionated during biosynthesis, so this simplification cannot be used.

Using values of α_W originally determined in the analysis of fatty acids produced by autotrophic organisms (15), we calculated values of f_W corresponding to the strongest and weakest fractionations previously measured. The data point in Fig. 4 is taken to be the median value, whereas the error bars represent extreme assumptions of α_W/AA.

Estimation of α_W/AA and α_G/AA for Individual AAs

Using these values of f_W, we can obtain reasonable estimates for α_W/AA and α_G/AA. This process of estimation is accomplished following the method of Session and Hayes (15), using the linear regression slope associated with water- and glucose-labeling experiments (m_W and m_G, respectively). Expression for m_W and m_G can be obtained by differentiating Eq. S3 with respect to R_W and R_G, respectively, and combined by solving for f_W and equating the resulting expressions:

1 - m_{G} / α_{G / A A} = m_{W} / α_{W / A A} .

The experimental results in this paper allow for the direct calculation of m_W for each AA. However, work to determine m_G remains ongoing. Barring compound-specific values for m_G, a rough estimate was obtained by regression analysis of whole-cell culture E. coli grown on glucose-based medium, where labeled glucose (δ²H values ranging from +53 to +384‰) was used as the heterotrophic growth source. Regression analysis of bulk E. coli cell material returns a slope of m_G = 0.3482, which was used to generate the curves in α_W/AA vs. α_G/AA space shown here (Figs. 2 and 3), as well as in other published analyses of δ²H fractionation factors in other biogeochemical systems (15, 23). In these plots, each ordered pair of (α_W/AA, α_G/AA) corresponds to a unique value of f_W, so tight numerical constraint on any one of these quantities allows for similarly close estimation of the remaining two.

Acknowledgments

We thank the following individuals for analytical advice, assistance, and editorial suggestions: Fred Prahl (Oregon State University), William Wurzel (Carnegie Institution of Washington), Ying Wang (Carnegie Institution of Washington), and Anne C. Jakle. This work was funded by Grant 2007-6-29 from the W. M. Keck Foundation (to M.L.F.), who was also partially funded by National Science Foundation (NSF) Grant DEB-1437845. P.G. was funded by the International Balzan Prize Foundation from a grant to R. Hemley and the W. M. Keck Foundation. S.D.N. was partially funded by the W. M. Keck Foundation and by NSF Grants DIOS-0848028 and DEB-1343015.

Supporting Information

Supporting Information (PDF)

Supporting Information

Download
185.79 KB

References

1

GJ Bowen, LI Wassenaar, KA Hobson, Global application of stable hydrogen and oxygen isotopes to wildlife forensics. Oecologia 143, 337–348 (2005).

Featured Topics

Articles By Topic

Featured Topics

Articles By Topic

Featured Topic

Articles By Topic

Significance

Abstract

Sign up for PNAS alerts.

Results

δ2H Values of AAs in Tryptone.

E. coli Grown in Glucose and Inorganic Nutrient Media.

E. coli Grown with Tryptone as Sole Organic Hydrogen Source.

Discussion

Glucose Metabolism and AA Synthesis.

Enzymatic Hydrogen Fractionation and Tunneling.

New Isotopic Biosignature?

Isotopic Routing and Implications for Animal Ecology.

Materials and Methods

SI Materials and Methods

E. coli Cultures.

Hydrolysis Tests.

δ2H Analysis.

Calculations.

E. coli Grown in Glucose and Inorganic Nutrient Media.

E. coli Grown with Tryptone as Sole Organic Hydrogen Source.

Calculation of δ2H Fractionation in AA Biosynthesis

Estimation of fW for Individual AAs

Estimation of αW/AA and αG/AA for Individual AAs

Acknowledgments

Supporting Information

References

Information

Published in

Classifications

Submission history

Keywords

Acknowledgments

Notes

Authors

Affiliations

Notes

Competing Interests

Metrics

Altmetrics

Citations

Cite this article

Cited by

View options

PDF format

Login options

Recommend to a librarian

Purchase options

Restore content access

Share

Share article link

Share on social media

Further reading in this issue

Cell size and fat content of dietary-restricted Caenorhabditis elegans are regulated by ATX-2, an mTOR repressor

Kin of coauthorship in five decades of health science literature

Unraveling the drivers of MERS-CoV transmission

Brain aging shows nonlinear transitions, suggesting a midlife “critical window” for metabolic intervention

Bodily maps of emotions

Neural network retuning and neural predictors of learning success associated with cello training

Sign up for thePNAS Highlights newsletter

δ²H Values of AAs in Tryptone.

δ²H Analysis.

Calculation of δ²H Fractionation in AA Biosynthesis

Estimation of f_W for Individual AAs

Estimation of α_W/AA and α_G/AA for Individual AAs