N₂ gas is an effective fertilizer for bioethanol production by Zymomonas mobilis

Significance

Recently there has been a surge in ethanol biofuel production from cellulosic feedstocks, offering more environmental benefits than conventional ethanol production from food crops. However, cellulosic feedstocks are low in nitrogen, requiring that millions of dollars be spent on nitrogen supplements to grow the ethanol-producing microbes. Zymomonas mobilis is a bacterium that has long been viewed as a potential rival to Baker’s yeast as an ethanol producer. Contrary to published remarks, we discovered that Z. mobilis can use the most abundant gas in the atmosphere, N₂, as a nitrogen source and does so without detriment to the high ethanol yield. Using N₂-utilizing Z. mobilis could offset much of the monetary and environmental costs of current industrial nitrogen supplements.

Abstract

A nascent cellulosic ethanol industry is struggling to become cost-competitive against corn ethanol and gasoline. Millions of dollars are spent on nitrogen supplements to make up for the low nitrogen content of the cellulosic feedstock. Here we show for the first time to our knowledge that the ethanol-producing bacterium, Zymomonas mobilis, can use N₂ gas in lieu of traditional nitrogen supplements. Despite being an electron-intensive process, N₂ fixation by Z. mobilis did not divert electrons away from ethanol production, as the ethanol yield was greater than 97% of the theoretical maximum. In a defined medium, Z. mobilis produced ethanol 50% faster per cell and generated half the unwanted biomass when supplied N₂ instead of ammonium. In a cellulosic feedstock-derived medium, Z. mobilis achieved a similar cell density and a slightly higher ethanol yield when supplied N₂ instead of the industrial nitrogen supplement, corn steep liquor. We estimate that N₂-utilizing Z. mobilis could save a cellulosic ethanol production facility more than $1 million/y.

Ethanol is the most heavily used biofuel in the world and until recently has been produced almost entirely from food crops (1). Last year saw a surge in cellulosic ethanol production as several commercial facilities came online, offering more favorable land use and lower CO₂ emissions than conventional ethanol production from cornstarch and sugarcane (2–4). Even so, cellulosic ethanol is struggling to be cost-competitive against corn ethanol and gasoline (1–3). Efforts to lower the price of cellulosic ethanol have primarily focused on the largest cost contributors, such as plant feedstocks and the cellulase enzymes needed to degrade feedstocks into usable sugars (1). However, the industry is also estimated to spend millions of dollars on nitrogen supplements to compensate for the low nitrogen content of the cellulosic feedstocks and allow the ethanol-producing microbes to grow (5–9). Corn steep liquor (CSL) is an industrial supplement that serves as a source of nitrogen and vitamins. Diammonium phosphate (DAP) is another industrial nitrogen supplement. A facility that produces more than 200 million liters of ethanol per year has been estimated to incur CSL and DAP costs between $1.7 and 2.2 million/y, based on projections from bench or pilot-scale values (6, 9). Other reports estimated CSL costs between $7.7 and 18.2 million/y (5, 7, 8). Furthermore, there are projections that CSL supply cannot scale to support ethanol operations of billions of liters per year (9). Thus, a sustainable alternative nitrogen source is desired.

N₂ gas is recognized as a sustainable source of nitrogen for agriculture as leguminous crops can be exploited for the symbiotic relationships they form with N₂-fixing bacteria (10). In a similar vein, N₂ gas could serve as an economical and environmentally benign nitrogen source for industrial fermentations. For example, more than 2,800 L of pure N₂ gas could be produced onsite for as little as $0.06 (SI Appendix). However, the traditional ethanol producer, Baker’s yeast, cannot use N₂ as a nitrogen source. N₂ can only be used as a nitrogen source by select bacteria and archaea that have the enzyme nitrogenase.

Zymomonas mobilis is the most prolific bacterial ethanol producer and is an emerging competitor to Baker’s yeast, as it produces ethanol three to five times faster per cell, at a higher yield, and with less residual biomass (11). Furthermore, nearly all strains of Z. mobilis appear to encode nitrogenase (Table S1). However, Z. mobilis was previously deemed incapable of N₂ fixation by some, although this claim was not experimentally verified (12, 13).

Recognizing the potential benefit of exploiting N₂ as a nitrogen source for industrial ethanol production, we used comparative growth assays and examined the incorporation of ¹⁵N₂ into Z. mobilis protein to determine whether Z. mobilis can use N₂. Our results conclusively indicate that Z. mobilis can use N₂ as a nitrogen source. Furthermore, N₂ utilization did not affect the ethanol yield or normalized metabolic flux distributions. Rather, growth with N₂ resulted in an increased specific ethanol production rate and a lower biomass yield in a defined medium compared with growth with NH₄⁺. We also demonstrated that Z. mobilis can use N₂ in a cellulosic feedstock-derived medium, suggesting that using N₂ in lieu of industrial nitrogen supplements could save the cellulosic ethanol industry millions of dollars annually.

Results

Z. mobilis Can Use N₂ as a Nitrogen Source.

The genome sequences of most Z. mobilis isolates appear to encode nitrogenase, the only known enzyme that allows prokaryotes to convert N₂ into bioavailable NH₄⁺ in a process known as N₂ fixation (Table S1). To determine whether Z. mobilis can fix N₂ gas, we inoculated one of the most commonly used strains, ZM4, into a chemically defined medium containing the necessary metal cofactors for nitrogenase activity: Mo and Fe. ZM4 grew when either NH₄⁺ or N₂ was provided (Fig. 1) but not when they were omitted (Ar headspace control). To verify that ZM4 growth was due to N₂ and not an unknown contaminating nitrogen source, we examined whether ZM4 would assimilate N₂ enriched with the heavy isotope ¹⁵N. When cultured with ¹⁵N₂, ZM4 synthesized protein that predominantly contained ¹⁵N (Fig. 2 and Table S2). Adding NH₄⁺ prevented ¹⁵N₂ assimilation (Fig. 2 and Table S2), consistent with observations in other N₂-fixing bacteria (14). These results provide, to our knowledge, the first conclusive experimental evidence that ZM4 can use N₂ gas as a nitrogen source.

Fig. 1.

The effect of nitrogen source on ZM4 growth and metabolism. (A and B) Growth (gray squares), glucose consumption (black triangles), and ethanol production (black circles) when provided either NH₄⁺ (A) or N₂ (B) in an anaerobic defined medium. Error bars are SEM (n = 4).

Fig. 2.

Incorporation of ¹⁵N₂ into cellular protein. Relative abundances for the M-57 fragment of tert-butyl-dimethylsilyl-alanine (Inset: alanine atoms in bold) normalized to the most abundant ion. Unlabeled standard, gray; ZM4 grown with ¹⁵N₂ + NH₄⁺, black; ZM4 grown with ¹⁵N₂, white. The 10% residual abundance at m/z 260 (white) is due to unlabeled N₂ in the test tubes (9 ± 5% of the N₂) and the unlabeled inoculum (0.9 ± 0.1% of the cells). Ion species present at 261–264 m/z (black and gray) or at 262–264 m/z (white) represent natural abundances of heavy isotopes for the elements in the amino acid and derivatization agent. All 12 amino acids analyzed showed similar distributions (Table S2). Error bars are SD (n = 3).

N₂ Utilization Does Not Compete Against Ethanol Production.

Nitrogenase requires four electrons and eight ATP to convert 1/2 N₂ into NH₄⁺. We therefore examined how these electron and energy demands would impact various metabolic parameters. One might expect the electron demands of N₂ fixation to divert electrons away from ethanol production, as two electrons are required to synthesize ethanol from pyruvate. However, HPLC analysis of ZM4 supernatants showed that the ethanol yield was statistically similar (≥94% of the theoretical maximum) regardless of whether NH₄⁺ or N₂ was provided (Fig. 1 and Table 1). Instead, electrons were diverted away from the production of undesirable biomass, as the growth yield with N₂ was half of that with NH₄⁺ (Fig. 1 and Table 1). As a result, the ratio of ethanol to biomass during growth with N₂ was over twice that observed with NH₄⁺ (Table 1). Moreover, the specific ethanol production rate (i.e., the ethanol production rate normalized to the biomass) during growth with N₂ was nearly 1.5 times that with NH₄⁺ (Table 1).

Table 1.

Comparison of growth and metabolic parameters during growth with NH₄⁺ vs. N₂

Nitrogen source	Ethanol yield (mol⋅mol glucose−1)*	Growth yield (g DCW⋅mol glucose−1)	Ethanol:biomass (mmol⋅g DCW−1)	Sp. ethanol productivity (mmol⋅g DCW−1⋅h−1)	Carbon recovery (%)	Electron recovery (%)
NH4+	1.88 ± 0.08	9.5 ± 0.2	198 ± 4	69 ± 4	100 ± 4	101 ± 4
N2	1.95 ± 0.03	4.3 ± 0.1†	450 ± 17†	100 ± 2†	100 ± 2	102 ± 2

^*The theoretical maximum yield is 2 mol of ethanol/mol of glucose. Errors are SD (n = 4). DCW, dry cell weight.

^†Values are statistically different from the corresponding NH₄⁺ value (t test; P < 0.001).

We used ¹³C-labeling studies to determine whether intermediary metabolism changed in response to growth with N₂. Metabolic fluxes were estimated from measurements of glucose uptake and ethanol production, the ZM4 biomass composition (15), and amino acid labeling patterns generated by metabolic pathways that processed the ¹³C-labeled glucose (Table S3). The resulting flux distributions, normalized to the glucose uptake rate, were similar regardless of the nitrogen source, except for biosynthetic fluxes, which were twice as high during growth with NH₄⁺ compared with growth with N₂ (Fig. 3A and Table S4). However, absolute flux values for central metabolism were higher during growth with N₂ compared with growth with NH₄⁺ (Fig. 3 B and C and Table S4). Overall, growth with N₂ in a defined medium enhanced the desirable traits of a high specific ethanol production rate with a low biomass yield while exhibiting an ethanol yield of 97.5% of the theoretical maximum (Table 1).

Fig. 3.

The effect of nitrogen source on ZM4 metabolic fluxes. (A) Linear regression analysis of net metabolic fluxes normalized to the glucose uptake rate during growth with N₂ vs. NH₄⁺. Biosynthetic fluxes are shown in the inset graph. The black line is the regression line with dashed gray lines representing the 99% CI. Central metabolic fluxes, black circles; biosynthetic fluxes, white squares. (B and C) Absolute net metabolic fluxes during growth with NH₄⁺ (B) vs. N₂ (C) determined using ¹³C-labeling patterns, measured extracellular fluxes, and biomass composition (15). Arrow thickness is proportional to the flux value. The thinnest arrows represent reactions with a rate of 0.83 mmol⋅g DCW⁻¹⋅h⁻¹ or less. All flux values with SDs are in Table S4.

Z. mobilis Uses N₂ Under Simulated Industrial Conditions.

We then asked whether N₂ fixation could substitute for a typical industrial nitrogen supplement, CSL, under conditions simulating the nitrogen availability in a cellulosic medium. We chose Miscanthus grass as our model cellulosic feedstock as it offers reduced greenhouse gas emissions compared with switchgrass and corn stover (16). The Miscanthus was hydrolyzed with dilute sulfuric acid, then neutralized, diluted, and supplemented with glucose to mimic the expected sugar content that would result from cellulase treatment (Materials and Methods). Although the hydrolysate without N₂ provided some nitrogen for growth, cell densities were up to twice as high when N₂ was provided (Fig. 4A). A nominal increase in the final cell density was supported by N₂ alone, but the highest cell density was observed when the nitrogenase metal cofactors, Mo and Fe, were supplied with N₂ (Fig. 4A). We determined the minimum concentrations of Mo and Fe minerals required to reach the highest cell densities in this medium to be 21 nM and 1.3 μM, respectively. Other trace elements and supplementation with the essential vitamin, pantothenate, were unnecessary to achieve the highest cell densities (Fig. 4A). The final cell density with N₂ and metal cofactors was the same as that obtained with a typical industrial supplement of 1% CSL (5, 17) (Fig. 4A). The ethanol yield from the N₂ + Fe + Mo condition was 97 ± 2% of the theoretical maximum, which was slightly higher than the 94 ± 1% observed with 1% CSL (t test, P < 0.05; n = 3). From these data, we conclude that N₂ can effectively substitute for CSL as a growth supplement and support a high ethanol yield under these simulated industrial conditions. Although our hydrolysate medium contained sufficient pantothenate to support growth, it is possible that hydrolysates from other feedstocks may require additional pantothenate, which could be provided economically through CSL. We found that 0.01% CSL could substitute for pantothenate in supporting full growth in our defined medium (Fig. 4B). Thus, CSL could be used to provide pantothenate at 1–4% of the concentration needed to serve as a nitrogen source [0.25–1% (vol/vol) CSL].

Fig. 4.

Identification of growth-limiting nutrients. (A) ZM4 growth in Miscanthus hydrolysate. Ar (black) or N₂ (white) was provided with other supplements as indicated. Final cell density represents the final OD₆₆₀ when all glucose was consumed minus the initial OD₆₆₀. Error bars are SD (n = 4). *Statistical difference from Ar + Te, P < 0.001; ^♦statistical difference from Ar + 1% CSL, P < 0.001. (B) Identification of the CSL concentration needed to substitute for pantothenate in the defined medium with NH₄⁺. Error bars are SD (n = 3), *Statistical difference from Pan, P < 0.001. (A and B) Te, trace elements; Pan, calcium pantothenate; none, no supplements added (see Materials and Methods for concentrations). Statistical differences were assessed using one-way ANOVA with Dunnett’s multiple comparison post test.

Discussion

In this study, we provided definitive evidence that Z. mobilis ZM4 can use N₂ gas in place of NH₄⁺ and the industrial nitrogen supplement, CSL. Remarkably, ZM4 exhibited a rigid metabolism dedicated to ethanol production even during N₂ fixation (Fig. 3A and Table 1). One might expect the two processes to be at odds, as both ethanol production and N₂ fixation have large electron demands. Instead, carbon and electrons were diverted away from biosynthesis, resulting in a biomass yield during N₂ fixation that was half that during growth with NH₄⁺ (Fig. 3A and Table 1).

It is tempting to speculate that N₂ fixation resulted in a lower growth yield but not a lower ethanol yield due to ATP limitation. In addition to being an electron-intensive process, N₂ fixation also uses eight ATP to convert 1/2 N₂ into NH₄⁺. Z. mobilis catabolizes glucose exclusively via the Entner–Doudoroff pathway, which yields only one ATP per glucose consumed. Thus, the additional energetic burden from nitrogenase activity may detract from the ATP available for other biosynthetic reactions. However, other groups have noted that Z. mobilis growth yields are typically even lower than would be expected from one ATP per glucose (18, 19). From these observations of low growth yields, a model of Z. mobilis metabolism has emerged wherein the glycolytic rate exceeds that of biosynthetic reactions and that Z. mobilis actually uses ATP-dissipating futile cycles (18, 19). Our observations of N₂-fixing Z. mobilis presented herein are analogous to those observed in nitrogen- and phosphorous-limited continuous cultures in which limitation of resources other than glucose resulted in lower biomass yields but higher specific glucose consumption and ethanol production rates (18). Thus, a more likely explanation for our results is that nitrogen limitation imposed by N₂ fixation led to a lower biosynthetic rate and exaggerated the uncoupling between glycolytic and biosynthetic reactions.

We also demonstrated that ZM4 will use N₂ in lieu of CSL in a cellulosic hydrolysate medium. The costs of nitrogen supplements required to produce ethanol from cellulosic feedstocks have not received much attention, likely because they are widely considered to be inexpensive relative to feedstock and cellulase. Even so, these supplements are estimated to cost an ethanol production facility well over a million dollars per year (5–9). Based on technoeconomic analyses and publically available prices, we estimate that the combined material costs of N₂, Mo, and Fe could be 10–29% of the cost of traditional supplements (Table 2). Thus, using N₂-fixing Z. mobilis could offer nearly $2 million in prospective savings per ethanol plant per year compared with one of the more conservative estimates of CSL and DAP costs (Table 2, scenario 2). Savings could be even higher if compared against estimates that put annual CSL costs between $7.7 and 18.2 million per plant (5, 7, 8). However, such high CSL and DAP costs are unexpected as they would be on par with cellulase in one report (5). Cellulase is more commonly considered to be one of the major costs of cellulosic ethanol production (1). Nonetheless, it is evident from comparing these reports that the prices of CSL and DAP can vary widely. Using Z. mobilis with onsite N₂ generation could therefore also insulate the price of ethanol from fluctuations in CSL and DAP prices.

Table 2.

Comparison of the estimated costs of traditional nitrogen supplements vs. N₂ gas

	Scenario 1 (ref. 9)			Scenario 2 (ref. 6)
Supplement	Cost (¢ 10−3 g−1)*	Cost in medium (¢ 10−3 L−1)†	Annual cost (MM$/facility)‡	Cost (¢ 10−3 g−1)*	Cost in medium (¢ 10−3 L−1)†	Annual cost (MM$/facility)‡
CSL	5.5	13.8	0.55	17.7	44.3	1.95
DAP	96.5	31.8	1.18	15.5	5.1	0.21
Total (CSL + DAP)		45.6	1.73		49.4	2.16
Alternative scenario
N2	3.4–11.9	3.6–12.6		3.4–11.9	3.6–12.6
Sodium molybdate dihydrate	2,640.0	0.07		2,640.0	0.07
Iron sulfate heptahydrate	2,050.0	0.72		2,050.0	0.72
Total (N2 + Mo + Fe)		4.4–13.4	0.17–0.51		4.4–13.4	0.19–0.59

^*Price sources are available in SI Appendix.

^†Cost in medium is based on the amounts of supplements used in either the scenarios referenced or in this study. For detailed calculations, see SI Appendix.

^‡

Values for scenarios 1 and 2 were taken from the corresponding references for scenarios 1 and 2 (see SI Appendix for details). Values for the alternative scenario were calculated using values from the corresponding reference scenario according to the following equation:

$$\mathrm{Total}\hspace{0.17em}\mathrm{annual}\hspace{0.17em}{\mathrm{cost}}_{\mathrm{Alternative}}=\mathrm{Total}\hspace{0.17em}\mathrm{cost}\hspace{0.17em}\mathrm{in}\hspace{0.17em}{\mathrm{medium}}_{\mathrm{Alternative}}÷\mathrm{Total}\hspace{0.17em}\mathrm{cost}\hspace{0.17em}\mathrm{in}\hspace{0.17em}{\mathrm{medium}}_{\mathrm{Referenced}}\times \mathrm{Total}\hspace{0.17em}\mathrm{Annual}\hspace{0.17em}{\mathrm{cost}}_{\mathrm{Referenced}}$$

Although our cost estimates are encouraging (Table 2), they are based on material cost estimates alone and assume direct scaling of our small-scale fermentation values. Other factors contributing to the price of ethanol must also be considered. For example, it is currently unclear whether the level of fermenter agitation assumed in one report (9) would dissolve sufficient N₂ for Z. mobilis or whether higher agitation rates would be required and incur higher electricity costs. Another consideration is how a potentially longer fermentation time when using N₂ as opposed to traditional nitrogen supplements (Fig. 1) would impact ethanol costs. One technoeconomic analysis of Z. mobilis in corn stover hydrolysate estimated that doubling the fermentation time from 1.5 to 3 d would add only 1¢/gal to the minimum ethanol selling price, whereas CSL and DAP contribute 3.12¢/gal (SI Appendix) (6). Furthermore, in a simultaneous saccharification and fermentation scenario, less cellulase would be needed to match the velocity of a slower fermentation (6) (SI Appendix). Because cellulases are estimated to account for 13% of the ethanol selling price on average (1), decreasing cellulase loading could provide substantial cost savings. Further investigation into how N₂-utilizing Z. mobilis could be integrated with industrial practices is clearly required to fully appreciate any economic benefits. Such efforts could identify cost savings beyond our material estimates in Table 2.

In addition to the economic benefits, N₂-utilizing Z. mobilis offers potential environmental benefits. For example, using N₂ could eliminate the weekly transportation of tanker loads of CSL (6) and potentially eliminate residual soluble nitrogen from CSL in waste effluent. It would also be beneficial if N₂ was to replace or offset DAP supplements. DAP production relies on ammonia from the Haber-Bosch process, which is notoriously energy intensive (20).

Our discovery that Z. mobilis can use N₂ gas as a nitrogen source without sacrificing the ethanol yield recommends further investigation into how N₂-utilizing Z. mobilis can be integrated and optimized for cellulosic ethanol production on a larger scale. Genomic evidence suggests that N₂ fixation is likely a conserved trait among Z. mobilis strains (Table S1). One strain, ATCC 29192, appears to have additional N₂ fixation versatility, encoding an Fe-nitrogenase in addition to Mo-nitrogenase (21). Several other biofuel-producing bacteria, including butanol-producing Clostridia, are also capable of N₂ fixation (22). Thus, the economic and environmental benefits of N₂ fixation could extend beyond cellulosic ethanol to the production of next-generation biofuels.

Materials and Methods

Chemicals.

¹³C-Glucose and ¹⁵N₂ were purchased from Cambridge Isotope Laboratories (Tewskbury, MA). All other chemicals were purchased from Sigma-Aldrich (St. Louis, MO) or Fisher Scientific (Waltham, MA).

Cultures and Growth Conditions.

Z. mobilis ZM4 was obtained from the ARS Culture Collection (nrrl.ncaur.usda.gov/). Cultures were grown in 10-mL volumes in 27-mL anaerobic test tubes or in 60-mL volumes in 160-mL serum vials at 30 °C. Media were made anaerobic by bubbling with either N₂ or Ar and sealed as previously described (23). Test tubes were laid flat, and serum vials were left upright. All cultures were shaken at 300 rpm.

The defined medium (ZYMM) contained Na₂HPO₄ (5.75 mM), KH₂PO₄ (7 mM), NaCl (8.6 mM), and trace elements 0.1% (vol/vol). The trace elements solution contained nitrilotriacetic acid (20 g/L), MgSO₄ (28.9 g/L), CaCl₂·2 H₂O (6.67 g/L), (NH₄)₆Mo₇O₂₄·4 H₂O (18.5 mg/L), FeSO₄·7 H₂O (198 mg/L), and Metals 44 solution (24) [0.1% (vol/vol)]. After autoclaving, the following supplements were added (final concentrations): calcium pantothenate (105 nM), MgSO₄ (1 mM), and CaCl₂ (0.1 mM). When CSL was added to ZYMM calcium pantothenate was omitted (Fig. 4B). ZYMM was supplemented with either NH₄Cl (10 mM) or a 100% N₂ headspace. When NH₄Cl was omitted, NaCl (10 mM) was provided to maintain similar osmotic conditions.

Miscanthus × giganteus was grown without fertilizer at the Energy Biosciences Institute (Urbana, IL). It was harvested, dried, and chopped in December 2013. Miscanthus hydrolysate medium was prepared as previously described (25) with some modifications. Briefly, 100 g of grass was hydrolyzed in 1% H₂SO₄ for 1 h at 121 °C. Solids were removed using Whatman filter paper. The pH of the hydrolysate was adjusted to 10 with Ca(OH)₂ and then heated to 50 °C for 30 min. After cooling, the pH was adjusted to 6 using H₃PO₄. Precipitate was removed using Whatman filter paper. The filtrate was then filter sterilized using a 0.2-μm filter. The hydrolysate was diluted 1:1 with water before use, as this was necessary for consistent growth. Various supplements were added as indicated in the text or else at the following final concentrations: trace elements [0.1% (vol/vol)], (NH₄)₆Mo₇O₂₄·4 H₂O (21 nM), FeSO₄·7 H₂O (2.5 μM), and calcium pantothenate (105 nM). CSL was clarified before use by centrifugation at 16,000 × g for 5 min. The diluted hydrolysate contained 25 mM of glucose and 56 mM of xylose. Additional glucose was added to raise the concentration to 75 mM in 10 mL. Two additional glucose supplements of 500 μmol each were later added during the fermentation to simulate the total amount of sugar expected if cellulases were used to liberate additional glucose and if a strain capable of xylose utilization were used (26). These additions were made whenever the OD₆₆₀ values stopped increasing. Gas was expelled at each glucose addition to lower the pressure in the tubes as a safety precaution. Tubes were flushed with either N₂ or Ar as appropriate after each addition of glucose.

To introduce ¹⁵N₂ gas, a stir bar was inserted into the neck of the ¹⁵N₂-containing breakseal flask and then a sampling port was attached to the neck. The sampling port consisted of a 5-mL tuberculin syringe, with the plunger replaced by a rubber stopper (Geo-Microbial Technologies) that was connected to the breakseal flask neck by rubber tubing. The stir bar was then used to break the seal. Anaerobic culture tubes were evacuated with a vacuum pump. For each addition of gas to an anaerobic culture tube, the breakseal flask was overpressurized with 15 mL of Ar using a syringe to displace the ¹⁵N₂ into the sampling port. Fifteen milliliters of the resulting ¹⁵N₂/Ar mixture was then transferred to the evacuated test tube via syringe.

Analytical Procedures.

Cell density was assayed by optical density at 660 nm using a Genesys 20 visible spectrophotometer (Thermo-Fisher). Dry cell weights (DCW) were determined as previously described (27). Optical densities were converted into DCW using our experimentally determined conversion factors of 516 mg DCW/L/OD₆₆₀ for cultures grown with NH₄⁺ and 423 mg DCW/L/OD₆₆₀ for cultures grown with N₂. Electron recoveries were calculated based on available electrons as previously described (28), assuming an elemental biomass composition of CH_1.125O_0.531N_0.214 (molecular mass: 24.625 g/mol) (29). Glucose and ethanol were quantified using a Shimadzu HPLC as previously described (30). Isotopic enrichments (¹³C or ¹⁵N) in amino acids were determined by GC-MS (Agilent) as previously described (27) at the Indiana University Mass Spectrometry Facility.

¹³C-Metabolic Flux Analysis.

Metabolic fluxes were estimated from the glucose uptake rate, the ethanol production rate, the biomass composition for Z. mobilis ZM4 (15), and mass isotopomer distributions from parallel labeling experiments (i.e., using 100% [1-¹³C]glucose in one experiment and a mixture of 80% unlabeled and 20% uniformly labeled glucose in the other). Mass isotopomer distributions were corrected for natural isotopic abundances using IsoCor software (31). Flux estimates were made using 13CFLUX2 (32) software with the nag_opt_nlp optimizer found in the Numerical Algorithms Group C library (www.nag.com/numeric/CL/CLdescription.asp) as previously described (33). Data from parallel labeling experiments were fit to a single metabolic model as previously described (34).