Solvent Deuterium Isotope Effects of Substrate Reduction by Nitrogenase from Azotobacter vinelandii

The mechanism of nitrogenase, the enzyme responsible for biological nitrogen fixation, has been of great interest for understanding the catalytic strategy utilized to reduce dinitrogen to ammonia under ambient temperatures and pressures. The reduction mechanism of nitrogenase is generally envisioned as involving multiple cycles of electron and proton transfers, with the known substrates requiring at least two cycles. Solvent kinetic isotope effect experiments, in which changes of reaction rates or product distribution are measured upon enrichment of solvent with heavy atom isotopes, have been valuable for deciphering the mechanism of complex enzymatic reactions involving proton or hydrogen transfer. We report the distribution of ethylene, dihydrogen, and methane isotopologue products measured from nitrogenase-catalyzed reductions of acetylene, protons, and cyanide, respectively, performed in varying levels of deuterium enrichment of the solvent. As has been noted previously, the total rate of product formation by nitrogenase is largely insensitive to the presence of D2O in the solvent. Nevertheless, the incorporation of H/D into products can be measured for these substrates that reflect solvent isotope effects on hydrogen atom transfers that are faster than the overall rate-determining step for nitrogenase. From these data, a minimal isotope effect is observed for acetylene reduction (1.4 ± 0.05), while the isotope effects for hydrogen and methane evolution are significantly higher at 4.2 ± 0.1 and 4.4 ± 0.1, respectively. These results indicate that there are pronounced differences in the sensitivity to isotopic substitution of the hydrogen atom transfer steps associated with the reduction of these substrates by nitrogenase.


Nitrogenase Purification
Figure S1: Representative chromatography traces and denaturing gels from purification of nitrogenase enzyme from Azotobacter vinelandii. Av1 and Av2 purified using ion exchange chromatography (left) followed by size exclusion chromatography (middle). Details of purification described in Methods. Av1 collected from ion exchange column at 19-22.5% Elution Buffer (312-341 mM NaCl) and Av2 collected from ion exchange column at 37.5-40.8% Elution Buffer (469-497 mM NaCl). Av1 collected from size exclusion at 1671-1703 mL and Av2 collected from size exclusion column at 1940-1971 mL. To run gel (BioRad Mini-PROTEAN TGX Stain Free Gels), 15 µL sample was combined with 5 µL of loading dye and 11 µL of the mixture was loaded into each well. Gel ran at 200 V for 20 minutes then rinsed with milliQ water, stained with Coomassie blue, and destained with a 40% methanol 10% acetic acid solution.
Gas Chromatographs of Headspace of Acetylene Reduction Assays Figure S2: Representative gas chromatogram of headspace of acetylene reduction assay of nitrogenase in 100% H 2 (black) or 51% D 2 O (green). Inset is zoom of region where ethylene peak elutes (1.8 min). Assay and GC-MS quantification of ethylene isotopologues described in Methods. Headspace (50 µL) injected using a Hamilton gastight syringe. Column : GASPRO PLOT-Q (Restek), column temperature: 80°C (isothermal). Figure S3: Mass spectra of ethylene peaks in the headspace of acetylene reduction assays in 100% H 2 O, 25% D 2 O, 51% D 2 O, and 73% D 2 O. Abundance normalized to peak of greatest abundance. Mass spectra generated by integrating under ethylene peak of extracted ion chromatograms of m/z = 23-32. As the mole fraction of D 2 O in buffer increases, greater abundance seen at m/z = 29 and m/z = 30, corresponding to increase in C 2 H 3 D and C 2 H 2 D 2 in the headspace.

Isotopologues
Mass spectra of the headspace of acetylene reduction assays in deuterated solvent show abundance at m/z = 28 from C 2 H 4 , m/z = 29 from C 2 H 3 D, and m/z = 30 from C 2 H 2 D 2 .
Quantifying the relative amounts of these species is not as straightforward as simply comparing the abundance at m/z = 28 vs. m/z = 29 vs. m/z = 30 because of the fragmentation that ethylene undergoes upon mass spectrometry analysis at 70 eV. The abundance in the mass spectrum at m/z = 29 is due to the molecular ion of C 2 H 3 D and also the fragmentation of C 2 H 2 D 2 in which it has lost one atomic mass unit (amu). Similarly, the abundance at m/z = 28 is due to the molecular ion of C 2 H 4 , the fragments of C 2 H 3 D that have lost one amu, and the fragments of C 2 H 2 D 2 that have lost two amu. Therefore, these overlapping mass spectra need to be deconvoluted in order to quantify relative amounts of C 2 H 4 , C 2 H 3 D,and C 2 H 2 D 2 . The method used for this deconvolution is described here.
With each assay, a non-deuterated ethylene standard was used to measure the fragmentation of C 2 H 4 and the other ethylene isotopologues were assumed to have the same fragmentation pattern, as has been assumed in previous reports. 1 All abundance values were obtained by integrating under the extracted ion chromatogram of each ion.
First, the abundance at each m/z value for the mass spectra of C 2 H 4 was normalized to the molecular ion at m/z = 28 to generate the fraction of the abundance of the molecular ion at that particular m/z value. For example F (27, C 2 H 4 ) is equal to the abundance at m/z = 27 divided by the abundance at m/z = 28 of the same spectra: where A C 2 H 4 (m/z) is the observed abundance at that m/z value from the ethylene standard analyzed using the same method as the assay samples. For the deuterated ethylene isotopologues, F (27, C 2 H 4 ) is the probability that a molecule of ethylene will fragment with the loss of one H/D relative to the molecular ion. F (26, C 2 H 4 ) is the probability that the ethylene molecule will fragment by the loss of two H/Ds and F (25, C 2 H 4 ) is the probabililty that ethylene will fragment by losing 3 H/Ds.
Next, the abundance values from the mass spectra of the headspace samples of the acetylene reduction assays were deconvoluted by calculating the contribution at each m/z from each isotopologue (written as C(m/z, C 2 H 2 L 2 )). To do this, each isotopologue was considered independently. For the dideuterated isotopologue, the molecular ion is m/z = 30 and the only contribution to the abundance at m/z = 30 is the dideuterated ethylene. The fragments of C 2 H 2 D 2 in which one H atom is lost will contribute to the abundance at m/z = 29. Because C 2 H 2 D 2 has 2 H atoms and 2 D atoms, the probability that it will lose one H atom (when it fragments by the loss of one H or D) is 1/2. Therefore, the contribution from C 2 H 2 D 2 at m/z = 29 (C(29, C 2 H 2 D 2 )) is 1/2 multiplied by the probability of losing one H/D (F (27, C 2 H 4 )), thus C(29, C 2 H 2 D 2 ) = A(30) × 1 2 F (27, C 2 H 4 ). This analysis was repeated for each m/z from the molecular ion (m/z = 30) to m/z = 25. Then this analysis was repeated for the singly-deuterated ethylene.
Contribution from C 2 H 2 D 2 : Contribution from C 2 H 3 D: The relative ammounts of C 2 H 4 , C 2 H 3 D, and C 2 H 2 D 2 were determined by comparing the molecular ion peak of the spectra of each isotopologue after subtracting the contribution from the fragments of the other isotopologues ( C(28, C 2 H 4 ), C(29, C 2 H 3 D), and C(30, C 2 H 2 D 2 ), respectively). The results are summarized in Table S1.

FTIR Standard Curves of Deuterated Ethylenes
The FTIR spectra of the headspace of acetylene reduction assays contained sharp stretches in the 840 -1000 cm −1 region, corresponding to the wagging mode of the H-C-H or H-C-D moieties. 2 The peak heights of these stretches were used to quantify the relative amounts of ethylene isotopologue products from acetylene reduction assays. It is well reported that the molar absorptivity of the C 2 H 4 stretch (949 cm −1 ) and the C 2 H 2 D 2 stretch (843 cm −1 ) are equivalent. However, there is some discrepancy as to whether the C 2 H 3 D stretch (943 cm −1 ) has a molar absorptivity that is half of that of C 2 H 4 and C 2 H 2 D 2 or equivalent to these species. In quantifying the products of nitrogenase reduction of C 2 D 2 , Han and Newton 3 report that the molar absorptivity of the C 2 H 3 D peak is half of that of the C 2 H 4 peak so they doubled the C 2 H 3 D peak height for their analysis. In similar measurements, Benton et al. 4 and Fisher et al. 5 report that the molar absorptivity of the trans-C 2 H 2 D 2 peak is half of that of the C 2 H 4 and cis-C 2 H 2 D 2 peaks, but they don't explicitly address the molar absorptivity of the peak from the singly deuterated species, C 2 H 3 D, at 943 cm −1 . Due to this uncertainty in the literature, we performed a standard curve with C 2 H 4 and C 2 H 3 D to determine their molar absorptivities: C 2 H 3 D was quantitatively transferred to a 1 L Schlenk flask to a final pressure of 1 atm and a separate flask was purged with C 2 H 4 for 10 minutes and vented to 1 atm. FTIR spectra were obtained using the same procedure that was used when analyzing the headspace of nitrogenase acetylene reduction assays except known quantities (50 -100 µL) of 1 atm C 2 H 4 /C 2 H 3 D were transferred to the degassed sample chamber and FTIR spectra were obtained. The peak heights of the C 2 H 4 (949 cm −1 ) and C 2 H 3 D (943 cm −1 ) stretches were plotted against the moles of these species in the sample chamber to obtain a standard curve. The absorbance from these stretches was calculated from the percent transmittance and this was plotted against the known amount of moles of the ethylene isotopologue in the FTIR gas cell ( Figure S4). Trendlines were generated in Excel and the slope was the molar absorptivity of the species. In agreement with the report  Figure S4: FTIR standard curves of ethylene isotopologues C 2 H 4 and C 2 H 3 D. Peak transmittance values of the C 2 H 4 stretch at 949 cm −1 (black) and the C 2 H 3 D stretch at 943 cm −1 (red) were converted to absorbance (Abs = 2-log(% Transmittance)) and plotted against the amount of ethylene present in the FTIR gas cell (µmol). Linear least squares fit trendlines were fit to the data in Excel and the slope of the line was the molar absorptivity of the species.
FTIR Spectra of Headspace of Acetylene Reduction Assays Figure S5: FTIR spectra of the headspace of acetylene reduction assays performed in varying mole fration of deuterium in the solvent. In 100% H 2 O, the only observed stretch is at 949 cm −1 due to the presence of C 2 H 4 . Rotational stretches are observed on either side of the main C 2 H 4 peak. In 22% D 2 O, a peak emerges at 943 cm −1 due to C 2 H 3 D in the headspace. At 51% D 2 O, the C 2 H 3 D peak at 943 cm −1 increases and a peak emerges at 843 cm −1 due to the presence of cis-C 2 H 2 D 2 .  Figure S6: H 2 /HD region of 1 H NMR spectra from headspace of proton reduction assay. Without deuterium decoupling (top), HD signal is a 1:1:1 triplet. Deuterium decoupling (bottom) condenses triplet into single peak. NMR tube filled with 1 mL CDCl 3 and capped with PTFE septa. Headspace from proton reduction assay (3 mL) transferred to a NMR tube with bubbling though CDCl 3 . NMR performed on a Varian 600 MHz Spectrometer with 5 mm triple resonance inverse probe (parameters: 512 scans, 16 sec relaxation delay, pulse angle 90°C). The dashed lines are plots of Equations (left axis; light blue, orange, brown, dark blue, teal, respectively), the model expressions for the mole fraction of methane isotopologue product as a function of the isotope efect. Markers are experimental data from the reduction assays with triangles as the results of GC-MS quantification of acetylene reduction assays, squares as the FTIR quantification of acetylene reduction assays, dotted circles as the GC-MS quantification of cyanide reduction assays and circles as the 1 H NMR quantification of proton reduction assays. The black circles are the experimental data from each proton reduction assay trial. All data plotted for cyanide and proton reduction data. Mean values plotted for acetylene reduction data and error bars are smaller than marker size.

Derivation of Isotope Effect Model Curves
The reduction of acetylene to ethylene and protons to H 2 involves the incorporation of 2 H/D atoms into the product. With each H/D addition, there is a probability, p, that a hydrogen atom is added and a probability, q, that a deuterium atom is added. The probability that a hydrogen is added is a function of the rate of this transfer and the mole fraction of hydrogen in the solution: 7 where k H and k D are the rates of hydrogen and deuterium addition, respectively, and f H and f D are the mole fractions of hydrogen and deuterium in the solvent, respectively. The isotope effect (IE) is defined as k H k D and the expression for p can be rewritten as: It follows that the probability of adding a deuterium (instead of a hydrogen) can be wrtten as: To generate H 2 and C 2 H 4 , two hydrogens are added. For HD and C 2 H 3 D, one hydrogen and one deuterium are added; and for D 2 and C 2 H 2 D 2 , two deuteriums are added. Making the assumption that p and q are the same for the first and second proton addition, the mole fraction of dihydrogen and ethylene isotopologue products can be written as the product of the appropriate probabilities of adding H vs. D: For the results of the proton reduction assays, only the ratio of H 2 to HD could be obtained so the isotope effect was calculated by the following equation: In order to explicitly calculate the isotope effect from the relative amounts of isotopologue products, equations 1-4 were solved for the isotope effect: 8 Least Square Fit of Acetylene and Proton Reduction

Assay Results
A least squares fit was performed on the acetylene reduction data across all levels of deuterium enrichment for each ethylene isotopologue ( Figure S8). The relative amount of each ethylene isotopologue (percentage of total ethylene produced) was plotted against the mole fraction of deuterium in the solvent. Then, the scipy.optimize.curve fit function was used in Jupyter notebooks to perform a non-linear least squares fit of the data to Equations 1-3.
The isotope effects obtained from these fits are 1.28 ± 0.04, 1.4 ± 0.1, and 1.59 ± 0.05 for C 2 H 4 , C 2 H 3 D, and C 2 H 2 D 2 , respectively. The results of the proton reduction assay were fit to Equation 7 using the same method as described above ( Figure S9). The isotope effect value obtained from this fit was 4.2 ± 0.1.  is the reciprocal of the isotope effect. In this work, we calculate isotope effects of nitrogenase reduction assays; however, the results can also easily be expressed as fractionation factors and are relevant to studies of geological nitrogen fixation. In reports measuring deuterium fractionation during dihydrogen production by cyanobacteria 14 or purified hydrogenases 10 significant depletion of deuterium was also observed. When we convert our isotope effect values to fractionation factors (Table S3) our results are comparable to these studies on hydrogenases, which could indicate that deuterium depletion is a common feature of H 2 production. In fact, these levels of deuterium fractionation are similar to that of H 2 production by the electrolysis of water. 14 The fractionation factor (α) is defined as the ratio of deuterium to hydrogen in the products divided by the ratio of deuterium to hydrogen in the solvent: where R H 2 is the ratio of D to H in the L 2 products and R H 2 O is the ratio of D to H in the solvent (L is the nomenclature for either H or D). In order to calculate the D to H ratio of the L 2 , we need to know the relative amount of D 2 produced, which we weren't able to quantify by 1 H NMR. However, we can calculate an estimate of the relative amounts of H 2 , HD and D 2 using the isotope effect that we calculated from equation 5, by inserting the calculated isotope effect into equations 1-3. These calculations were performed with our proton reduction data and the results are shown in Table S3. The fractionation factor was estimated to be 0.24 ± 0.03, 0.25 ± 0.05, and 0.20 ± 0.02 from assays performed in 25%